TECHNICAL REPORT 2024 © 2023 The World Bank Group 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org This work is a product of the staff of The World Bank Group with external contributions. “The World Bank Group” refers to the legally separate organizations of the International Bank for Reconstruction and Development (IBRD), the International Devel- opment Association (IDA), the International Finance Corporation (IFC), and the Multilateral Investment Guarantee Agency (MIGA). The World Bank Group does not guarantee the accuracy, reliability or completeness of the content included in this work, or the conclusions or judgments described herein, and accepts no responsibility or liability for any omissions or errors (including, without limitation, typographical errors and technical errors) in the content whatsoever or for reliance thereon. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank Group concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the organizations of the World Bank Group, their respective Boards of Executive Directors, and the governments they represent. The contents of this work are intended for general informational purposes only and are not intended to constitute legal, securities, or investment advice, an opinion regarding the appropriateness of any investment, or a solicitation of any type. Some of the orga- nizations of the World Bank Group or their affiliates may have an investment in, provide other advice or services to, or otherwise have a financial interest in, certain of the companies and parties named herein. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of any of the organizations of The World Bank Group, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank Group encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncom- mercial purposes as long as full attribution to this work is given and all further permissions that may be required for such use (as noted herein) are acquired. The World Bank Group does not warrant that the content contained in this work will not infringe TECHNICAL REPORT on the rights of third parties, and accepts no responsibility or liability in this regard. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Acknowledgments Leonardo Iacovone, Matias Belacin, Xavier Cirera, and Santiago Reyes Ortega authored this report. Chapter 1 is based on an early note by Giorgi Mukhilishvili. The report was prepared under the supervision of Ilias Skamnelos (Practice Manager) and the overall guidance of Lalita Moorty (Regional Director) and Sebastian Mo- lineus (Country Director). The authors are thankful for comments received during the review meeting from Kevin Carey, Penelope Ann Mealy, Marcio Cruz, Mona Prasad, Florian Kitt, Miguel Eduar- do Sanchez Martin, and Teona Elizarashvili. The authors are grate- ful to the companies that agreed to participate in the case studies, our counterparts from the Government of Georgia, in particular the Ministry of Economy and Sustainable Development and the Min- istry of Finance (MoF), as well as various stakeholders, including representatives of the private sector and the European Commis- sion (EC) delegation in Tbilisi with whom the preliminary findings of this report were shared and discussed. The authors are especial- ly thankful to the National Statistics Office of Georgia (GEOSTAT) for an extremely effective and productive collaboration on the data collection for the Technology Adoption Survey and their willingness to share anonymized firm-level data, which allowed the authors to perform most of the analysis presented in this report. I Greening Firms in Georgia Abbreviations BAT Best available technology CAP Climate Action Plan CCC Climate Change Council CFL Compact fluorescent lamps CIP Conservation Improvement Program CO2 Carbon dioxide CPI Consumer Price Index CRM Customer Relationship Management DOPD Dynamic Olley-Pakes Decomposition EBRD European Bank for Reconstruction and Development EC European Commission ECA Europe and Central Asia EE Energy efficiency EED Energy Efficiency Directive ERP Enterprise Resource Planning ESCO Energy service companies ESI Energy savings insurance ETL Energy Technology List EUREM European Energy Managers FAT Firm-level Adoption of Technologies FE Fixed effects FEMP Federal Energy Management Program FIRA Fideicomisos Instituidos en Relación con la Agricultura GBF General Business Functions GCF Green Climate Fund GEFF Green Economy Financing Facility GEL Georgian Lari (currency) GEOSTAT National Statistics Office of Georgia GGF Green for Growth Fund GITA Georgia’s Innovation and Technology Agency GNERC Georgian National Energy and Water Supply Regulatory Commission II Technical Report HELP Home Energy Lending Program IPCC Intergovernmental Panel on Climate Change IPF International patent family kWh kilowatt-hour LCU Local Currency Units LED Light-Emitting Diode MEPA Ministry of Environment Protection and Agriculture MoESD Ministry of Economy and Sustainable Development MoF Ministry of Finance MRV Monitoring, Reporting, and Verification NBG National Bank of Georgia NDC Nationally Determined Contribution NEA National Environmental Agency OLS Ordinary least squares PATSTAT Patent Statistical Database RED Renewable Energy Directive RISE Regulatory Indicators for Sustainable Energy SF Sustainable Finance SME Small and medium-sized enterprises SRM Supplier Relation Management SSBF Sector-Specific Business Functions TFP Total factor productivity UNIDO United Nations Industrial Development Organization VAV Variable Air Volume WBCSD World Business Council for Sustainable Development WBES World Bank Enterprise Surveys WRI World Resources Institute III Greening Firms in Georgia Figures Figure 1 Countries are Cutting CO2 Emissions although They Keep Growing 4 Figure 2 Total CO2 Emissions and CO2 Intensity 5 Figure 3 Georgia’s Performance on Energy Efficiency and Renewable Energy Regulatory Framework Compared with Regional Peers and Aspirational Countries 7 Figure 4 The Elements of the Energy Policy and Climate Change Strategy of Georgia 11 Figure 5 Electricity and Gas Prices by Country in 2021 18 Figure 6 Explicit Energy Subsidies 20 Figure 7 Energy Efficiency over Time and across Sectors 23 Figure 8 GDP Growth in Georgia, Regional Peers and Selected EU Countries (Index, 2005 = 100) 24 Figure 9 Decomposition of Energy Consumption 26 Figure 10 Dynamic Olley-Pakes Energy Efficiency Decomposition 29 Figure 11 Change in Aggregate Energy Consumption by Industry, 2018–21 30 Figure 12 Firm-Level Energy Intensity and Change in Energy Consumption by Sectoral Energy Intensity, 2018–21 31 Figure 13 CO2 Emissions and Energy Consumption in Georgia 33 Figure 14 CO2 Emissions by Sector Energy Intensity 34 Figure 15 Energy Efficiency Dispersion across Georgian Firms 37 Figure 16 Average Energy Intensity and Dispersion across Sectors 38 Figure 17 Change in Sectoral Energy Efficiency over Time 39 Figure 18 Energy Efficiency over Time at the Firm Level 40 Figure 19 Simulation: Energy Efficiency Distribution before and after Efficiency Improvements 41 Figure 20 Relationship between the Efficiency Gap and the Capital Gap 42 Figure 21 Capital Requirements to Improve Efficiency? 43 Figure 22 Investment Amounts Required for Upgrading Capital among Inefficient Firms 44 Figure 23 Aggregate Energy Consumption before and after the Intervention by Type of Firm (mln. GEL) 44 Figure 24 Simulated Impact of Improving Energy Efficiency on Energy Savings, Costs, and Profits (2021, million GEL) 45 Figure 25 CO2 Emissions before and after the Energy Efficiency Improvement 45 Figure 26 Contribution to CO2 Savings 46 Figure 27 Mapping of Technologies in General Business Functions 50 Figure 28 Share of Firms Facing Power Outages by Country 52 Figure 29 Outages, Use of Generators, and Production of Renewable Energy 53 IV Technical Report Figure 30 Green Technology Adoption 55 Figure 31 Technology Adoption in Georgian Firms 56 Figure 32 The Correlation between Green and General Technology Adoption 57 Figure 33 Change in Energy Efficiency and GBF Technologies by Firm Size 61 Figure 34 New Green Patents in Georgia and across Regional Peers 64 Figure 35 The Relationship between Energy Efficiency and Investment, Innovation, ICT, and Exporting 71 Figure 36 Energy Efficiency Change within and across Sectors 72 Figure 37 Assessing Energy Efficiency Convergence in the Private Sector of Georgia 73 Figure 38 Distribution of Green Management Practices in Georgian Firms 81 Figure 39 Type of Green Management Practices by Sector 81 Figure 40 Green Management Practices across Size and Age Classes, Ownership, and Exporting Status 82 Figure 41 The Relationship between Green Management, Overall management, and Labor Productivity 83 Figure 42 The Relationship between Energy Efficiency and Green Management Practices 83 Figure 43 Non-Household Energy Tariffs Adjusted for Electricity Only (2021) 88 Figure 44 Changes in Electricity Prices Over Time (Relative to 2014 Changes) 90 Figure 45 Electricity Elasticity in the Manufacturing and Construction Industries 93 Figure 46 Electricity Elasticity Over Time 94 Figure 47 Structure of Policy Recommendations 96 Figure 48 Solutions for Access to Finance 98 Figure 49A Firm Energy Consumption and CO2 Emissions versus Changes in the Supply of Energy 113 Figure 50A Share of Each Energy Source in Firms’ Total Energy Costs in 2021 116 Figure 51A Olley-Pakes Static Decomposition of the TFP Growth 117 Figure 52A Distribution of Electricity and Gas Prices Paid by Georgian Firms 117 Figure 53A Change in Carbon Efficiency Due to Changes in Energy Efficiency 118 Figure 54A The Relationship between Energy Efficiency and Investment, Innovation, ICT, and Exporting 119 Figure 55A Green Management Score in Georgia and Peer Countries 125 Figure 56B Management Practices and Green Technology Investment 129 Figure 57C Green Patents and Economic Development 131 Contents 01 PAGE INTRODUCTION 03 PAGE Chapter 1 BENCHMARKING AGGREGATE GREEN OUTCOMES, INSTITUTIONS, AND BARRIERS TO GREENING THE PRIVATE SECTOR IN GEORGIA 1.1 Aggregate Green Outcomes 3 1.2 Regulatory Framework for Energy Efficiency 5 20 1.3 Stakeholders, Policies, Strategies, and Legal Framework 8 PAGE Chapter 2 DECOMPOSING THE DRIVERS OF ENERGY CONSUMPTION 2.1 Descriptive Statistics 22 35 2.2 Energy Consumption and Emission Decompositions 23 2.3 Energy Consumption Decomposition Results 26 PAGE Chapter 3 ENERGY EFFICIENCY IN GEORGIA: A FIRM-LEVEL ANALYSIS 3.1 Heterogeneity in Energy Efficiency between Firms 36 3.2 Evolution of Energy Intensity over Time 39 3.3 A Thought Experiment to Improve the Targeting and Focus of Green Policies 40 48 PAGE Chapter 4 TECHNOLOGY ADOPTION, INNOVATION AND ENERGY EFFICIENCY 4.1 The Data 49 4.2 Green Energy Sources 51 4.3 Green and General Technology Adoption 53 4.4 What is the Role of Technology in Explaining Energy Efficiency? 58 66 4.5 Summing up: The Extent of Green Technology Adoption 65 PAGE Chapter 5 EXPLAINING ENERGY EFFICIENCY AT THE FIRM-LEVEL 5.1 The Correlates of Energy Efficiency 66 5.2 Energy Efficiency Convergence 72 86 5.3 The Importance of Green Management Practices for Energy Efficiency 77 PAGE Chapter 6 ASSESSING THE ROLE OF PRICES IN ENERGY CONSUMPTION 6.1 The Data 87 6.2 Identification Strategy 90 96 6.3 Preliminary Results 91 6.4 Preliminary Conclusions 94 PAGE Chapter 7 POLICY RECOMMENDATIONS 7.1 Completing and Implementing an Enabling Institutional and Legal Framework 91 7.2 Supporting Firms to Invest in Energy Efficiency and Upgrade Technologies and Practices 94 References 107 Appendix A Additional Analysis 112 Appendix B Data Quality Checks and Investing in Green Technology 126 Appendix C Green Innovation: An Analysis of New Green Patents 130 1 Introduction Greening Firms in Georgia Introduction 1—2 PAGE C limate change will likely be the most critical threat greenhouse gas (GHG) emissions through improving en- to private sector resilience in the coming years. ergy efficiency and upgrading technology. Greenhouse gas Minimizing the potential adverse shocks from climate emissions (GHG) resulting from energy use by the private sec- and the increasing energy shocks associated with geo- tor (agriculture, industry, services, transport, and commercial political tensions demand swift action to decarbonize buildings) account for nearly 60 percent of total GHG emissions the private sector and make it more energy efficient. in Georgia,1 whereas those produced as a byproduct from in- Greening the private sector can also create new sources dustrial processes represent almost 9 percent (Our World in of competitive advantage, such as preferential access to devel- Data, 2023). The fact that the lion’s share of GHG emissions is oped country markets. Decarbonizing the most polluting ener- related to energy demands of enterprises across a wide range of gy and carbon-intensive sectors are important. But there is also activities shows that there is considerable potential to reduce a need to improve energy efficiency more generally across other them by using energy more efficiently, that is, reducing the sectors, which together account for a non-negligible share of energy requirements without reducing output. In this regard, energy consumption. Reducing energy consumption without energy efficiency improvements and technology adoption are affecting output is critical for improving overall efficiency, critical areas of opportunity because they can reduce GHG increasing profits, and cultivating participation in global val- emissions and simultaneously increase overall firm compet- ue chains in the context of surging energy prices. Therefore, itiveness and profitability. Cutting-edge machinery and en- greening the private sector can boost the Georgian economy’s ergy-saving managerial practices consume less energy and 1 Greenhouse gas (GHG) productivity and competitiveness while developing more sus- lower GHG emissions. They also enhance the energy efficiency emissions refer to gases tainable production processes. of firms, which, under current global energy conditions, can released into the atmosphere significantly affect costs and profits. that trap heat and contribute Despite Georgia’s relatively small contribution to global to global warming. These gases include carbon dioxide carbon emissions, the country has experienced signifi- The green transition can help Georgia access global (CO2), methane (CH4), nitrous cant growth in emissions over the past 15 years, showing markets and have a more prominent role in global value oxide (N2O), and fluorinated no signs of decoupling emissions from gross domestic chains, fueling export growth. Beyond the overall efficiency gases. CO2 enters the product (GDP) growth. In 2021, CO2 emissions in Georgia and environmental benefits of reducing GHG emissions and atmosphere through burning accounted for a mere 0.3 percent of global carbon emissions, enhancing energy efficiency, the green transition can ease the fossil fuels (coal, natural gas, and per capita emissions were below the EU average. However, access of Georgian firms to international markets, especially and oil), solid waste, trees recent trends are concerning—if the prevailing trend of the past those of developed countries. For instance, the EU’s move to- and other biological materials, and also as a result of certain decade continues, Georgia’s per capita CO2 emissions are pro- ward net zero (for example, the Carbon Border Adjustment chemical reactions (for jected to surpass those of the European Union by 2035. In con- Mechanism or CBAM) will have trade and certification impli- example, cement production; trast to many high- and middle-income countries, Georgia still cations outside the EU. Also, US efforts to go green will further United States Environmental has not decoupled carbon emissions from economic growth. affect import and export flows. Therefore, for Georgia to com- Protection Agency, 2023). pete in global markets, which is particularly important given Usually, CO2 accounts for the The private sector (agriculture, industry, services, trans- the size of its domestic market, its firms will need to become lion’s share of GHG emissions port and commercial buildings) is a major contributor more efficient, increase their efforts to go green, and prepare (in the US in 2021, CO2 to the growth of these emissions, primarily due to ener- to follow new regulations and carbon pricing mechanisms. accounted for 79.2 percent of total GHG emissions). gy consumption, leaving significant room for reducing 2 Introduction Technical Report In addition to boosting efficiency and access to global Nationally Determined Contribution (NDC) but, more broad- markets, Georgia needs to make efforts to transition to- ly, supply lessons for other middle-income-countries (MICs) ward more sustainable growth processes, given its inter- facing similar challenges. est in becoming an EU member. Besides short-term potential trade and efficiency gains of making firms more energy efficient The report is organized as follows. Chapter 1 discusses and greener, Georgia’s candidacy to join the EU further sup- the current institutional and regulatory framework in Georgia, ports the importance of a swift green transition. Joining the providing a benchmark and discussing the main challenges EU implies moving toward net zero and adapting energy and for the institutional development agenda. This chapter also environmental standards in line with EU practices. discusses economy-wide challenges linked to energy prices and subsidies. Relying on detailed firm-level data, Chapter 2 But despite the urgency of making the private sector more presents a novel decomposition of the drivers of Georgia’s en- energy efficient, little is known about current energy effi- ergy consumption and carbon emissions. Leveraging the same ciency levels within and across industries and the sources microdata with detailed information on energy consumption and barriers that affect firms’ decisions to become green- and firm characteristics, in Chapters 3 and 4, we comprehen- er. Due to the lack of disaggregated data, little is known about sively assess the characteristics and recent evolution of energy the dispersion in energy efficiency and GHG emissions within efficiency in the private sector in Georgia and identify the un- and between sectors that could guide policies and regulations derlying firm-level factors that determine how energy-efficient in dealing with the likely heterogeneity in green outcomes. businesses are over time. The diagnostic contributes to prior- Important questions (such as where most of the gaps in en- itizing energy and industrial policy as we detect the sectors ergy efficiency are concentrated—types of firms, sectors, re- with the widest margins for improvement and those that can gions—or how large efficiency differences between firms are) contribute the most to energy savings. Additionally, Chapter remain unanswered, making effective policy interventions and 4 contributes to a better understanding of the link between targeting difficult. management practices (generic and environment-related) and energy efficiency by analyzing the relationship between ge- This report uses very detailed firm-level data collected by neric and green management, green technology investment, GEOSTAT that links information on energy consumption efficiency, and firm performance. Relying on novel technology with technology adoption data to shed some light on these adoption data, Chapter 5 dissects the relationship between questions. Georgia provides a window into the challenges of technology adoption (both green and generic technologies) a middle-income country growing significantly in the last 15 and firms’ energy efficiency. Chapter 6 presents a preliminary years without decoupling its emissions from economic growth. analysis of the impact of prices on energy consumption and At the same time, Georgia is in an interesting and complex provides some early evidence of the initial short term (that is, institutional transition. It is adopting some of the EU green during the first 12 months) impact of the energy price increase legislation while reducing distortions on energy prices that that occurred in January 2021.2 Finally, the report is closed 2 Data availability limits reduced the profitability of investing in green technologies with a concluding chapter with detailed policy recommenda- us to be able to perform only and processes. Thus, the evidence presented here will not only tions and a discussion of relevant examples from international preliminary analysis for 2021. help Georgia design policies for achieving the targets in its good practices • 3 Chapter 1 Greening Firms in Georgia Chapter 1 3—19 PAGE Benchmarking Aggregate Green Outcomes, Institutions, and Barriers to Greening the Private Sector in Georgia A CHAPTER 1.1 key element of climate mitigation is decoupling high-income countries, upper-middle-income countries, and GDP from carbon emissions growth. Recent trends lower-middle-income countries, CO2 emissions and GDP trends Aggregate show that countries, particularly higher-income ones, are steadily cutting emissions alongside GDP growth (Burn-Murdoch, 2022). This decoupling has acceler- are diverging, meaning they are successfully addressing cli- mate change concerns while they keep growing. They are be- coming less carbon intensive. For instance, over 2000–22, GDP Green ated with technological change, the development of renewable, non-fossil fuels, and the economic struc- tural transformation of production from more carbon-intensive in high-income countries grew by 49 percent, but CO2 emis- sions declined by 7 percent, while in upper-middle-income and lower-middle-income countries, GDP expanded by 202 percent Outcomes manufacturing to less carbon-intensive services. Despite substantial progress even among less advanced and 185 percent, but carbon emissions grew significantly less (118 percent and 133 percent respectively). However, carbon emissions in Georgia are increasing vis-à-vis output expansion. European countries, Georgia has not yet converged to this From the beginning of the century until 2022, GDP rose by trend, suggesting it is not reducing carbon intensity. Fig- 184 percent and CO2 emissions by 140 percent, but if a shorter ure 1 plots the cumulative GDP and CO2 growth from 2000 period is considered, 2010–22, emissions growth outpaced the onwards in Georgia, regional peers and selected developed rise of output (77 percent vs. 58 percent). So, while many coun- countries. In Western and Eastern Europe and the US, carbon tries—irrespective of their income levels—have sustainedly dioxide emissions have fallen or remained unchanged in the reduced CO2 emissions per unit of output, Georgia still lags in last 20 years, despite sustained economic growth, especially this greening process. in less advanced countries of the European Union. Among 4 Chapter 1 Technical Report FIGURE 1 Countries are Cutting CO2 Emissions although They Keep Growing CO emissions GDP (PPP, $) Cumulative growth since 2000 (%) Bulgaria Czech Rep. Romania Poland 100 100 100 100 50 50 50 50 0 0 0 0 -50 -50 -50 -50 -100 -100 -100 -100 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 Germany France United Kingdom United States 100 100 100 100 50 50 50 50 0 0 0 0 -50 -50 -50 -50 -100 -100 -100 -100 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 Georgia Kazakhstan Türkiye North Macedonia 300 300 300 300 200 200 200 200 100 100 100 100 0 0 0 0 -100 -100 -100 -100 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 World High-income countries Upper-middle income Lower-middle income 300 300 300 300 200 200 200 200 100 100 100 100 0 0 0 0 -100 -100 -100 -100 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 ‘00 ‘04 ‘08 ‘12 ‘16 ‘20 Source: Global Carbon Budget (2022) Several reasons motivate the need to improve the energy industry and service activities account for 60 percent of total use of the Georgian private sector. First, while worldwide emissions, so energy efficiency through better-managed energy trends show that countries are successfully lowering carbon use and technology adoption has the potential to lower these emissions per unit of output without jeopardizing economic emissions. At the same time, it can increase Georgian firms’ growth, Georgia still fails to converge to this global pattern with overall efficiency and profitability, especially in the context of the risk of becoming even more carbon pollutant -in per capita higher energy prices. Moreover, transitioning toward a greener terms or per unit of output- than the EU by 2035. Measured economy is likely to facilitate access to global markets -especially per US$ 1,000 of GDP, Georgia has become more carbon-in- to those already implementing carbon tariffs for domestic and tensive than EU countries and regional peers in the last 35 foreign goods- and speed up joining the EU, which can result in years. Second, the private sector is a central actor in the green a substantial welfare increase. Therefore, the private sector of transition of Georgia. Carbon dioxide, methane, and nitrous Georgia has the potential to play a crucial role in decarbonizing oxide emissions related to energy consumption in agriculture, Georgia and making the economy more competitive. 5 Chapter 1 Greening Firms in Georgia FIGURE 2 Total CO2 Emissions and CO2 Intensity CHAPTER 1.2 PANEL A CO₂ emissions Regulatory Framework CO₂ tonnes (5-year moving average index, 1975 = 1) 1.4 for Energy Efficiency 1.2 1 EU27 GEO ARM .8 SRB HUN .6 BGR MKD .4 ROU .2 0 O 1990 1995 2000 2005 2010 2015 2020 2025 ne key aspect to explain this disap- PANEL B CO₂ emissions intensity pointing performance is the institu- tional setting and regulatory frame- CO₂ tonnes per US$ 1,000 (5-year moving average index, 1975 = 1) work determining the incentives of energy use and efficiency. Steering 1.6 investment toward cleaner activities, technological change, and human cap- GEO 1.4 ital are the drivers for greening the economy without hampering economic growth. When gov- 1.2 ernments seek to increase investments, particu- larly in renewable energies and energy efficiency, 1 the set of incentives, determined by prices and .8 regulations are critical. For example, if energy is subsidized, investors may find energy-intensive .6 industries more attractive than in neighboring EU27 countries, allocating resources to firms where .4 they can maximize their profits. However, if pol- MKD icies regulate energy investments and tax carbon .2 BGR emissions, investors may find those investments ARM 0 less attractive. Also, investors may look at best ROU practices, access to technology, the stringency of 1990 1995 2000 2005 2010 2015 2020 2025 government regulations, and the government’s capacity to enforce and audit (RISE, 2023). Notes: CO2 intensity is the total CO2 emissions per US$ 1,000 of GDP. Source: Global Carbon Budget (2022) The Regulatory Indicators for Sustainable Energy (RISE) data set measures a set of coun- try-comparable indicators for sustainable energy that includes the policy setting, the regulatory framework and the scheme of in- centives related to energy efficiency and re- newable energy. RISE ranks countries’ energy efficiency (EE) and renewable energy based on their policy and regulatory support. Regarding EE, the score summarizes the countries’ performance among comprehensive aspects related to the regu- latory framework and energy policies. In renewable energy, the index measures the legal framework and incentives for investing in renewables. 6 Chapter 1 Technical Report BOX 1 RISE indicators for energy efficiency and renewable energy RISE assesses countries’ policy and regulatory ance); viii) energy labeling systems (either man- support and summarizes them in a set of indi- datory or voluntary); ix) building energy codes cators that allow comparing national policy and (strategy, regulations and required standards Georgia underperforms in terms of the policy regulatory frameworks for sustainable energy. for buildings); x) transport sector (tracking and and regulatory framework for EE and devel- The data collected includes 30 indicators cov- report of EE metrics in transport, initiatives to oping renewable energy sources across all re- ering 140 countries and representing over 98 reduce demand or shift to more efficient trans- gional peers and EU aspirational countries. percent of the world population. port modes); xi) carbon pricing and monitoring Figure 3 illustrates the performance of Georgia (e.g., the existence of a carbon monitoring or in terms of EE—incentives (A) and regulations (B)—and renewable energy (C) based on the lat- In the case of energy efficiency, the indicator pricing mechanism that covers GHG emissions). est available data as of 2019–21. The charts show averages the score of 11 sub-pillars: i) nation- that Georgia ranked at the bottom of the EE and al energy efficiency planning (whether there is The renewable energy score includes seven in- renewable energy scores among nine selected an action plan, legal framework and targets to dicators that jointly assess the environment for regional and aspirational peers -e.g., Bulgaria, increase EE); ii) energy efficiency entities (in- renewable energy generation. The renewable Croatia, Kazakhstan, Romania, Poland, Türkiye, dependent bodies that set EE strategies, stan- energy analysis includes: i) the existence of a Germany, the United Kingdom, and the United States. As for EE, Georgia shows a poor perfor- dards, certifications); iii) incentives and man- legal framework for renewable energy; ii) plan- mance. Concerning the incentives and mandates dates for the business (whether large energy ning for renewable energy expansion (targets to enhance EE, the performance of Georgia is well users have mandates on targets, audits, mon- and plans for electricity, heating and cooling, below benchmark countries in terms of incentives itoring and control, energy managers for EE transport, institutions, and renewable energy and mandates to the industry and commerce end and receive penalties for non-compliances and generation); iii) incentives and regulatory sup- users, financing mechanisms, energy utility pro- if there are incentive programs for industrial port for renewable energy (financial and regula- grams, and carbon pricing and monitoring mech- anisms. As for regulations, Georgia also ranks low consumers and SMEs); iv) incentives and man- tory support for electricity, heating and cooling, in minimum EE performance standards, labeling dates for the public sector (EE obligations, man- and transport, grid access and dispatch); iv) systems and building energy codes. Overall, there dates, guidelines for energy-efficient options for attributes of financial and regulatory incentives are several margins to improve EE in Georgia procurement); v) energy utility programs (e.g., (competition environments, tariffs changes, ex- regarding incentives and regulations. Further- mandates to carry out EE initiatives in the area istence of auctions); v) network connection and more, as for renewable energy, Georgia ranks at the bottom among comparable countries and across producers and consumers); vi) financ- use (connection procedures, allocation of costs, shows systematic underperformance in renew- ing mechanisms for energy efficiency (whether network using and pricing systems, and integra- able expansion planning, incentives and regu- there is a “national financial coverage” mecha- tion of renewable grid integration); vi) counter- latory support, and the financial and regulatory nisms in place for energy efficiency activities in party risk (creditworthiness and payment risk support characteristics. In sum, there is ample residential, commercial and industrial sector); mitigation); vii) carbon pricing and monitoring room for improving the scheme of incentives and vii) minimum energy efficiency performance (GHG emissions coverage under a carbon pric- regulations nationwide. standards (adoption of minimum EE standards ing mechanism). and verification and penalties for non-compli- Source: https://rise.esmap.org/ 7 Chapter 1 Greening Firms in Georgia FIGURE 3 Georgia’s Performance on Energy Efficiency and Renewable Energy Regulatory Framework Compared with Regional Peers and Aspirational Countries PANEL PANEL Energy AA e Energy e ciency: ciency: PANEL B Energy e ciency: PANEL C Renewable Energy Incentives Incentives Regulations Planning for I&M: Industry I&M: and Industry and Building energy renewable expansion commerce end commerce end users users codes Incentives and I&M: Energy I&M: utility Energy utility Energy labeling regulatory support programs programs systems Attributes of financial Financing mechanisms Financing mechanisms Minimum energy and regulatory incentives e ciency performance Carbon pricing Carbon and pricing and standards monitoring monitoring Germany Germany Germany United Kingdom ma ia ia Po nia Ge y Poand Ro man an Ka an l or m lan zak rm Ro Ro gia d hs Ge tan GeG or omom Kaza es Geo eg oir gd khst Stat rgia ania agia KinK ingd an ed Rom 10 0 10 0 80 80 i te d Un Unite d 100 80 Unit 1 00 80 60 6 0 60 60 40 4 0 40 40 UniU te nd Bulg ite SdtaS aria Türk teta stes s iye n n tas tate akzhsk h ta rkiye ed S Kaz K a a Tü Unit United Kingdom Bu Bu Po Bu lga tia lga la Tü kiye nd lga ye nd oa ria Croatia ria la Croatia rki ria r Croatia Cr Po Tü Note: I&M = incentives and mandates. Source: Regulatory Indicators for Sustainable Energy (RISE, 2019-2021). 8 Chapter 1 Technical Report CHAPTER 1.3 Stakeholders, Policies, Strategies, and Legal Framework O ver the last years Georgia has been making prog- Under this new regulatory framework and with further ress in improving the legal and regulatory frame- efforts and resources devoted to improving EE perfor- work, although there is still ample space to en- mance, Georgia has the potential to mitigate GHG emis- hance it. As of November 2022, the implementation sions by at least 8.1 MtCO2e by 2030. Cutting carbon emis- of the required legal and regulatory framework in the sions demands a national strategy that seeks to improve the EE sector of Georgia is moderately advanced, accord- coordination of stakeholders, sets an appropriate regulatory ing to the Annual Implementation Report of Georgia environment and scheme of incentives, outlines programs and (Energy Community Secretariat, 2022) prepared by the Energy interventions, disseminates technology and knowledge, en- 3 There are four Community. The Laws on Energy Efficiency and Energy Perfor- courages information sharing and aligns incentives, taking into market forces of change: mance of Buildings are in place. Moreover, in 2021 the govern- account sector characteristics. Under a successful strategy, the sustainability, digitalization, decentralization, and ment adopted the minimum energy performance requirements Georgian industry sector has the potential for decarbonization business models for industrial for buildings, building units and building elements and the by setting up energy audits and the best available technology transformation in Georgia. national calculation methodology for buildings performance. (BAT) system to assess the cost-effectiveness of EE projects. The study also provides ten In addition, Georgia adopted several bylaws to implement the This ambitious strategy still requires significant efforts and ways forward for private Energy Efficiency Law, including procedures and guidelines for resources to be focused on lowering barriers that impede mit- sector engagement in Georgia: implementing EE criteria in public procurement procedures. igating carbon emissions. For example, upgrading regulations 1. Renewable Decentralized The online platform for monitoring and verifying savings was on emissions, enhancing domestic capacities and technology Energy and Mini-Grids, 2. Energy Efficiency and Energy completed with EU4Energy support and is under the agen- access across businesses and organizations, expanding financ- Management 3. Smart cy of the Ministry of Economy and Sustainable Development ing opportunities, and developing infrastructure are part of the Powering & Smart Lighting 4. (MoESD) of Georgia. The adoption of other key bylaws, such as action plan to reduce barriers for greening the economy. So, E-mobility 5. Smart Farming certification rules and the regulation on inspection of heating the next subsection looks at the role of stakeholders, policies, and Precision Agriculture 6. and air-conditioning systems, is pending. Also, the Energy regulations, and incentives for green growth in Georgia. As for Digitalization and Automation Efficiency Law promotes the use of energy service companies the private sector, it is important to consider and strategize their 7. Waste Recycling and (ESCOs) but does not include a framework to support public fi- participation in voluntary carbon markets through investment Circular Economy Market, nancing (i.e., an EE fund). Still, various international technical in mitigation (energy, industry, waste) and adaptation (agricul- 8. Ensure access to finance 9. Carbon Market and credit assistance projects and financing programs support EE mea- ture, forestry) projects.3 mechanisms 10. Research and sures, especially in the building sector. However, the study also systemic innovation (UNDP, finds private sector engagement requires further development 2022). of the ESCO market (Energy Community Secretariat, 2022). A Stakeholders There are several stakeholders involved in the green transi- sustainable development practices. This lack of knowledge and tion. The actor map includes over 13 organizations from public resources limits the ability of public institutions to effectively and private entities, non-profit organizations, businesses, and address complex issues related to climate change and the use technological and financial associations. Among the most rel- of energy, but no further actions to increase climate change evant actors are the MoESD, the MoF, the Ministry of Environ- education are ongoing to deal with these problems. ment Protection and Agriculture (MEPA), the Climate Change Council (CCC), Enterprise Georgia, Georgia’s Innovation and Among public stakeholders, agencies need infrastructure Technology Agency (GITA), the Georgian National Energy and and capacities to deliver effective EE regulations. As for the Water Supply Regulatory Commission (GNERC), donors, NGOs, MoESD, establishing a dedicated entity responsible for imple- the National Bank of Georgia (NBG), International Financial In- menting and coordinating programs and activities embedded in stitutions (IFIs), local banks, and business associations support EE laws is essential for the new regulatory framework’s success. the EE of the Georgian private sector. Box 2 maps public and As for the Ministry of Environment Protection, strengthening private stakeholders and examines their missions, scope and internal capacities through new technical staff and coordinating agency for addressing climate change challenges. the donor platform more actively is fundamental. Developing a sustainable Measurement-Monitoring, Reporting, and Verifica- Private and public stakeholders face various challenges tion (MRV) system requires upgrading hardware and software. in enhancing EE. One of the crucial challenges is increasing While the latter includes policies, procedures, and protocols the level of coordination and cooperation between them. These that govern data collection, analysis and reporting, personnel problems are further exacerbated by limited resources for fi- training, and capacity building, the former involves the equip- nancing EE projects, the lack of technical capacities among ment and tools necessary for gathering and processing data. public institutions, and the low awareness of the importance of 9 Chapter 1 Greening Firms in Georgia BOX 2 Mapping Public Sector Stakeholders On a policy level, the primary decision maker is the challenging mandate of encouraging EE investments. En- MoESD. The Energy Efficiency and Renewable Energy Poli- terprise Georgia is responsible for business support, export cy and Sustainable Development Department of the MoESD promotion and investment in Georgia. The business division is responsible for developing the national EE policy and implements firm support mechanisms through several instru- strategy and for setting long-term EE targets. Among its ments such as grants, credit lines and guarantees. Although main responsibilities, it can adopt bylaws and technical reg- businesses can use programs for EE improvements and ulations on EE, design measures to improve EE and promote boosting competitiveness, up to date, there are no specific their implementation, and promote EE-related investments. programs focused on EE and the environmental footprint of firms. So, there is space for designing and delivering pro- The Ministry of Finance (MoF) allocates public resourc- grams in coordination with public and private bodies. es to national programs, including those aiming to en- hance EE. Beyond the MoESD, the MoF is a strategic stake- A dynamic innovation system is crucial for public and pri- holder in the carbon mitigation strategy. The MoF assigns vate organizations to invest in physical and human capital, public funds to national public policies, but its scope extends improve processes and products, and reduce the envi- to international collaboration for strategic projects (i.e., en- ronmental impact of production, so Georgia’s Innovation gage Georgia with international partners and donor organi- and Technology Agency (GITA) has a critical role in green- zations). So, the coordination between the MoESD and the ing the economy. The GITA’s mission is to create a dynamic MoF is strategic to maximize the cost-effectiveness of public innovation ecosystem driven by technology adoption, infor- interventions related to EE and GHG mitigation initiatives. mation sharing, building knowledge and investment. For this mission, GITA promotes the exchange of innovative knowl- The Ministry of Environment Protection and Agriculture edge and accessibility to frontier technology and equipment (MEPA) is a relevant policy actor for the country’s en- within and across industries. The agency endeavors to use its ergy policy and green strategy. The MEPA has the crucial infrastructure to develop innovations and technologies, pri- mandate of promoting sustainable development principles oritize the commercialization of innovations and machinery, and integrating a green economy within more traditional facilitate capital investment and encourage the translation of industries. It is also responsible for the formulation and im- basic research into product and processes innovations. As plementation of national policies related to climate change. with Enterprise Georgia, specific EE programs are still miss- Working closely with the MoESD and other government ing and need to be a priority in order to untap the potential of agencies, the MEPA, through its dedicated the Department firms in cutting GHG emissions. of Environment and Climate Change, spearheads the devel- opment of the Nationally Determined Contributions and the Energy prices are an essential part of household and in- Climate Change Strategy, which set the GHG emissions tar- dustrial consumers’ incentives for energy consumption gets for 2030 and include EE interventions as pivotal for the decisions, so the Georgian National Energy and Water climate change mitigation strategy. Supply Regulatory Commission (GNERC) play a key role on GHG mitigation. The GNERC sets energy tariffs for reg- The Climate Change Council (CCC) has the crucial ulated companies and monitors electricity and gas markets. mission of coordinating climate change policy across Also, although energy prices for big consumers in the com- stakeholders. The CCC was established in 2020 and con- mercial sector are not regulated, the GNERC still monitors sists of a council -formed by ministers, the National Bank prices and energy consumption. Furthermore, the GNERC of Georgia, the Director of GEOSTAT, and the head of Ab- seeks to balance companies’ and customers’ interests and khazia and Adjara AR-, a coordination group of the signa- promote a competitive environment. tory municipalities of the Covenant of Mayors, and working groups -including public servants, experts, and academia, Greening firms requires a variety of well-functioning fi- among others-. One of its main roles is coordinating cli- nancing mechanisms, so the National Bank of Georgia mate change mitigation and adaptation policies among (NBG) is a relevant actor in promoting green invest- ministries and relevant environment-related actors. ments. The NBG ensures the stability and transparency of the financial system and promotes sustainable economic Businesses are central to pursuing a more environ- growth while ensuring price stability. In 2022, the NBG ap- ment-friendly economy, so Enterprise Georgia has a proved the Sustainable Finance Taxonomy and the Regula- 10 Chapter 1 Technical Report BOX 2 Mapping Public Sector Stakeholders (cont.) tion on Loan Classification and Reporting according to the About five NGOs are actively working on EE and GHG emis- Sustainable Finance Taxonomy. The Sustainable Finance Tax- sion topics.4 onomy classifies economic activities into green, social, and sustainable based on specific technical criteria. The Regu- International Financial Institutions also support the gov- lation on Loan Classification and Reporting also applies this ernment in policy design and institutional development. classification to loans. This is central for green investments IFIs such as the European Bank for Reconstruction and De- because the regulation establishes that commercial banks velopment (EBRD) and the KfW provide financial and techni- shall comply with the taxonomy criteria when classifying cer- cal assistance and consultancy to national and local govern- tain loans. In addition, the regulation imposes reporting re- ments. Some also provide low-interest-rate credit to local quirements on banks according to the taxonomy from 2023. commercial banks for Georgia’s green and EE projects.5 They provide technical support to the Government of Geor- 4 Energy Efficiency Center Other non-public stakeholders gia in implementing EE laws.6 (EEC), World Experience for Georgia (WEG), Caucasus Several bilateral donor organizations support Georgian Local commercial banks finance the private sector in Environmental NGO Network (CENN), Greens Movement of institutions at various levels, especially when it comes to developing EE measures. Local banks, such as Procredit Georgia, REMISSIA. building capacity for making evidence-based policy choic- Bank, TBC, BOG, and Basis Bank, promote energy-efficient 5 the EBRD’s Green es. Donor organizations provide technical, financial and policy technology adoption and penetration in the food sector using Economy Financing Facility support, which ultimately affects the performance of the pri- various financing schemes. Procredit Bank and TBC are the (GEFF) provides financing vate sector. For example, the MoESD has developed EE laws most active banks in EE support. for energy efficiency and renewable energy projects in and bylaws relying on donor support (i.e., KfW, USAID, WFD, Georgia, as well as technical UNDP, GIZ, etc.). Finally, business associations (BAG, SME) support assistance and capacity private businesses in developing EE measures by un- building for businesses and Donor support also helps NGOs to raise their voice. Us- dertaking various initiatives. They promote networking, financial institutions. https:// ebrdgeff.com/georgia_ ing donor support, NGOs provide policy analysis, iden- increase awareness, share information and best practic- facilities/ tify gaps and needs and make policy recommendations es, lobby their interest, and help engage firms, the public 6 KfW supports MoESD for different stakeholders on EE. They are instrumental sector and academia. on EE law and bylaw in awareness-raising campaigns and information sharing. development through GOPA. 11 Chapter 1 Greening Firms in Georgia B Strategies & Plans The Energy Policy of Georgia defines EE as one of the key 2030 and sets targets for EE and renewable energy sources de- priorities and sets strategies, actions and targets. The En- velopment. Recently, the MoESD released the National Energy ergy Policy provides a roadmap for private and public organi- Policy of Georgia, a milestone for the national energy policy of zations by establishing strategies, actions and targets around the country, primarily because it is the first detailed document EE and GHG mitigation. The government of Georgia considers in Georgia that reflects the current development, challenges energy policy, in general, and improving EE, in particular, a and relevant long-term policy directions in a context in which priority, given the benefits a resilient and efficient energy sys- a large number of countries are multiplying efforts to cope tem has. Enhancing EE helps improve energy security, boost with climate change threats. Furthermore, the National Energy economic growth and employment, enhance technology and Policy considers various visions and perspectives, including innovation and reduce social conflicts. Georgia has developed public entities, private organizations, NGOs and independent various strategies and action plans to improve EE and set long- experts. It took more than two years to reach a consensus and term target indicators. Yet, some strategic documents are still include different stakeholders’ viewpoints. The result is an in development and have not been officially adopted. energy policy that considers all the energy resources used in the country, and it includes a national integrated energy and The National Energy Policy of Georgia, drafted in Septem- climate plan aiming to provide energy security and solidari- ber 2022 by the MoESD, is a milestone for climate change ty, the management of energy markets, the reduction of GHG mitigation as it sets the direction of the energy policy for emissions, and support innovation and competition. One of the next ten years. The National Energy Policy of Georgia the salient objectives of the National Energy Policy is a 24 per- establishes the vision, priorities, and policy measures until cent cut in final energy consumption by 2030 –compared to a FIGURE 4 The Elements of the Energy Policy and Climate Change Strategy of Georgia NDC and CSAP UNTIL 2030 LT–LEDs UNTIL 2050 Policies and strategies relevant to energy efficacy in private sector Energy Policy SME and NECP Development Strategy 2021–2030 2021–2025 12 Chapter 1 Technical Report scenario with no active measures - and a 27.4 percent share of Part of the required coordination is aimed to be achieved renewable in final energy consumption. through the Georgia’s 2030 Climate Change Strategy (CSAP) and the Climate Action Plan (CAP). The CSAP The National Energy Policy of Georgia is part of a broader and CAP coordinate efforts and planning toward meeting strategy developed by the MoESD, including the National nationally determined targets for climate change mitigation Integrated Energy and Climate Plan of Georgia (NECP) for by identifying specific directions and actions for GHG reduc- 2021-2030. The National Energy Policy of Georgia includes an tion in a way that sets a pathway to meet international and annex that provides a detailed overview of the current energy national targets for tackling climate change (Government of system and the energy and climate policy. The NECP envisages Georgia, 2021). The CSAP includes the sectoral dimension by strategic actions to encourage savings of industrial consumers, providing detailed sector-specific targets and relevant actions which include (a) the development of a system of incentives across energy- and carbon-intensive activities such as energy to implement cost-effective EE measures; (b) the development generation and transmission, transport, buildings, industry, of energy-saving model agreements for high-energy-intensive agriculture, waste management and forestry. Measures in- companies; (c) setting the requirements for energy audits for clude (i) increasing the share of low- and zero-emission and non-small and medium enterprises (SMEs) and enhancing roadworthy private vehicles; (ii) switching from fossil fuels support mechanisms for SMEs; (d) the implementation of EE to biofuels; and promoting non-motorized means of mobility certifications and minimum EE standards for buildings; and and public transport; (iii) developing a system for EE building (e) the definition of standards, norms and labeling schemes for certifications; (iv) raising consumer awareness about EE; (iii) energy-consuming equipment. To date, the NECP includes 25 encouraging energy-efficient approaches and installing more measures, and technical teams are elaborating specific indica- efficient lighting systems in commercial buildings; (vi) sup- tors for monitoring the implementation of the action plan. In porting the use of solar energy for water heating and using of order for the NECP to be effective, it needs to create effective energy-efficient stoves; training high -professional standard mechanisms for achieving energy policy goals and a good co- personnel in EE; (vii) reducing the level of GHG emissions operation environment, as its implementation is monitored from industrial processes and introducing modern technolo- by the Energy Community as well. gy to save; (viii) developing a system for studying the emission factors in the industry sector and for data management; (ix) In parallel, the MEPA is concerned with EE, underscoring building capacities to generate scientific evidence for devel- the importance of coordination across stakeholders. In oping climate-smart approaches in the agriculture sector. April 2021, the Government of Georgia adopted GHG emissions Based on this action plan, the CSAP and CAP steer efforts and targets by 2030.7 The government aims for GHG emissions to be resources devoted to coping with climate change. 7 Included in the (unconditionally) 35 percent below those of 1990 by 2030. If the Georgia’s Updated Nationally global average temperature increased by 1.5-2 °C by 2030, the The government of Georgia has also developed the Long- Determined Contribution government would commit to set GHG emissions 57-50 percent Term Low Emission Development Concept (LT-LED),8 (NDC) document, developed below 1990. The NDC encourages the development of low-car- which addresses how Georgia plans to boost green eco- by the MEPA. bon technologies, processes and methods in the building sector, nomic growth. The LT-LED is a visionary policy document 8 Georgia’s Low Emission including public buildings and tourist landmarks. Since many with long-term sectoral priorities (until 2050) aligned with Development Strategy (LEDS) actors are carrying out actions to mitigate carbon emissions, the NDC and Georgia’s 2030 Climate Change Strategy doc- will be adopted in 2023. cooperation is crucial for the effectiveness of the whole strategy. uments. The LT-LED states that Georgia intends to go fully 13 Chapter 1 Greening Firms in Georgia ‘green’ by 2050 through energy-efficient technologies and trepreneurship and helping SMEs insert in the green economy. renewable energies. Georgia plans to combine greening and The SME Development Strategy aims to achieve a 20 percent growth leveraging transformative innovations, thereby reduc- growth in SMEs’ value added and productivity, as well as a 10 ing GHG emissions. In this regard, technological reequipment percent increase in employment by 2025, compared to the 2019 and modernization are crucial for economic development and baseline. The strategy aims to boost inclusive and sustainable decarbonization via increased efficiency, waste minimization growth based on the main principles of the Small Business Act and low-emission technologies. for Europe -including “Think Small First”- and on SMEs EU best policy practices. SME development is one pillar for ensuring green eco- nomic growth. The MoESD developed the SME Development Delivering the green strategy and policy plans effectively Strategy 2021-20259, emphasizing the leading role of small and requires high level coordination and cooperation, finan- medium-sized enterprises in the green transition. The strategy cial resources, and building capacities. The effectiveness of sets the priorities and initiatives for boosting the performance the climate change strategy and green-growth policy objectives of SMEs and entrepreneurship. First, it stresses the impor- relies on public and private entities working together efficiently tance of refining legislation and gradually converging to the and effectively toward the reduction of the carbon footprint of EU legislation, strengthening institutions and improving the Georgia. Absent aligned efforts, there is a risk of underwhelm- competition and operational environment for SMEs. Second, it ing results, wastage of resources, and losing momentum for promotes the development of entrepreneurship and the entre- the overall strategy. Financial resources are also crucial be- preneurial culture. Furthermore, it underscores the importance cause transitioning to a greener economy demands small and 9 https://www.economy. ge/uploads/files/2017/ek__ of easing access to finance, promoting export growth, inserting large-scale investments. The strategy’s success also depends politika/sme_strategy/2022/ Georgian SMEs into global value chains, and encouraging SMEs on building technical capacities such as relevant educational sme_strategy_2021_2025_ to go digital and innovative. Moreover, the strategy seeks to programs, training of qualified personnel, and monitoring and eng_2.pdf contribute to narrowing the gender gap by boosting women en- evaluating EE improvements through appropriate data. C Legislation and Regulation As Georgia strives to integrate with the European Union, it The Energy Efficiency Law requires large companies10 to has been diligently transposing EU energy directives and conduct an energy audit every four years. The energy audit regulations into national legislation and progressively aims to identify sources for reducing energy consumption and implementing them. In 2019, the Parliament of Georgia passed provide tailor-made recommendations to firms for becoming the Law on Energy and Water Supply, aligned to key EU energy more energy efficient. The EE by-law also mandates regula- acquis, providing a general legal framework for the production, tions about the structure of the measurement and verification transmission, distribution, supply and trade of electricity and platform and the inclusion of information. It enforces actions gas. The law promotes developing and integrating well-func- to enhance EE, specifies the reporting method for achieved tioning, transparent, and competitive electricity and natural energy savings, and establishes the requirement for adminis- gas markets to enhance EE and renewable energy production. trative bodies and enterprises to register and publish data on the platform. Beneficiary enterprises must report EE improve- Moreover, the Law of Georgia on Energy Efficiency seeks ments to the platform. to increase energy savings and define obligations for the private sector’s EE. The law, adopted in 2020 and developed The current regulation (i.e., the EE by-law) obliges the in compliance with the EU Energy Efficiency Directive (EED), MoESD to report yearly the performance of the EE national aims to increase energy savings and determine obligations and target indicator to the Secretariat of the Energy Communi- responsibilities for the EE of the private sector. The legislation ty and assess the factors impeding the reduction of energy mandates the MoESD to develop SMEs-targeted programs that consumption at the sector level.11 The annual “On Energy Effi- 10 Companies that generate encourage small and medium enterprises to undertake energy ciency” document reports energy consumption metrics at the more than 100 million GEL audits and adopt recommendations from such assessments. aggregate and sector level. Remarkably, it carefully analyzes in revenue or have assets of Moreover, it envisages the MoESD sharing information on the energy consumption across sectors. When energy consump- more than 50 million GEL or benefits and experiences of upgrading energy management tion remains stable or increases, the MoESD must analyze the more than 250 employees. systems. The MoESD is improving the legislation and policy causes of this performance. 11 Adopted by the programs to ensure the effective implementation of these pro- Minister’s Order (Ministry grams. Furthermore, the law established the legal framework The Law on Energy Efficiency of Buildings adopted in of Economy and Sustainable for the energy services market. Related bylaws and model con- 202012 promotes the rational use of energy resources and Development of Georgia, tracts for energy performance contracting are in the drafting more energy-efficient buildings. Since 2020, permanent MoESD) of May 20, 2022. phase and expected to be adopted by 2023-2024. commercial and residential are subject to regulations that aim 12 On energy efficiency to measure and certify their EE based on minimum energy buildings, see matsne.gov.ge performance requirements, inspections of heating and cooling 14 Chapter 1 Technical Report systems and the assessment of independent experts certifying EU energy legislation. Although the MoESD is responsible the energy performance.13 The law addresses the importance for developing EE policies and laws, no institutional body is of information and financial incentives to raise the EE perfor- seeking to implement this new regulatory framework. In this mance of buildings by encouraging the implementation of dis- regard, there is still much work to be done. For example, qual- semination programs targeted to owners and tenants, delivery ification, accreditation, and certification schemes for energy of training to energy experts, and financial instruments, which service providers, auditors, and managers must be further de- are still in consideration. veloped. Moreover, developing sustainable financing mecha- nisms for energy service providers/companies (ESCOs) remains Establishing an effective institutional framework remains challenging. challenging despite the harmonization process with the BOX 3 Drivers of successful Energy Service 13 The law does not apply to the following buildings: temporary buildings intended for use for a period not Companies (ESCO) exceeding 2 years, industrial facilities, workshops, non-residential, agricultural, low-energy buildings; Energy service companies (ESCOs) have suc- 3 United Kingdom: The United Kingdom has a free-standing buildings, the useful floor area of which ceeded in various countries worldwide, but growing ESCO market, driven by policies such is less than 50 m2. some countries where they have been particu- as the Climate Change 14 Energy Service Companies (ESCOs) by the larly successful are14: International Energy Agency (IEA). ESCO Market Report 2019 - Energy Efficiency Trends and Policies in Act of 2008 and the Energy Savings Opportu- Europe, by the European Commission. ESCO Market 1 United States: The US has a mature ESCO nity Scheme. These policies have created a sup- Report 2020 - Global Trends and Developments by the market driven by policies such as the Energy portive regulatory environment for ESCOs and Energy Efficiency Financial Institutions Group (EEFIG). https://bfu-ag.de/en/activities/energy Policy Act of 1992 and the Energy Indepen- incentivized the implementation of EE projects. dence and Security Act of 2007. These policies 15 Best practice examples for ESCOs in the EU are the EPC model and PPPs, to deliver energy efficiency created a favorable regulatory environment for The main drivers of successful ESCO devel- projects that meet the needs of their clients and ESCOs and incentivized the implementation of opment in these countries are a supportive contribute to the EU’s energy efficiency targets. The EE projects. regulatory environment, favorable policies and Energy Performance Contracting (EPC) model: This model is a type of ESCO service that enables building regulations, financial incentives, and a strong owners to finance energy efficiency projects through 2 Germany: Germany has a well-developed commitment to improving EE and reducing the guaranteed energy savings achieved over time. ESCO market, supported by favorable policies greenhouse gas emissions. Additionally, the The EPC model is widely used in the EU and has been successful in reducing energy consumption and and regulations, such as the Energy Services availability of skilled professionals, innovative greenhouse gas emissions. Public-Private Partnerships Act and the Renewable Energy Act. The country technologies, and effective project management (PPPs): PPPs are a collaborative approach between has also implemented various financial instru- strategies have also contributed to the success public and private sector entities to finance and deliver energy efficiency projects. PPPs are particularly ments, such as loans and subsidies, to promote of the ESCO market in these countries.15 useful for large-scale projects that require significant the growth of the ESCO market. investment and expertise. 15 Chapter 1 Greening Firms in Georgia In addition to EE legislation, the legislative framework expected to be adopted by 2023 and enter into force in 2026, on renewable energy sources and industrial emissions but the obligations for industrial firms will be required only is also essential for the private sector of Georgia. The Law from 2028 onwards16. The law applies to industrial activities, of Georgia on Promoting the Generation and Consumption of combustion plants, and waste incineration and co-incineration Energy from Renewable Sources, developed in compliance with enterprises that cause pollution (energy production, production the EU’s Renewable Energy Directive (RED), was adopted in and processing of metals, processing of mineral raw materials, 2019. Its objective is to encourage generating and consuming chemical industry, waste management). The National Envi- 16 According to the energy from various renewable sources. The law determines the ronmental Agency (NEA) will identify the relevant industrial article 29 of the draft law required national common target indicators of the total share firms and issue the integrated permits, while the Department until January 1, 2026, the of energy received from renewable sources in the total final en- of Environmental Supervision will monitor implementation Government of Georgia ergy consumption and energy consumption by transport. The of the law. should adopt the following law envisages that the Government of Georgia shall facilitate resolutions: a) “About the usage of renewable sources when designing, constructing, Implementing and developing a monitoring and informa- combustion devices”; b) and renovating industrial territories. It also shall ensure the tion system with appropriate incentives and penalties is “About those activities that implementation of measures which, in the case of biomass, an open agenda, especially for SMEs, mostly not covered use organic solvents”; c) “On waste incineration and promote methods that enables at least 85 percent processing ef- by requirements. The effective implementation and enforce- co-incineration enterprises”. ficiency when used for household or commercial purposes and ment of the legal and regulatory pieces are crucial. Further- d) “Conclusions on the best at least 70 percent efficiency when used for industrial purposes. more, it should be noted that these legal provisions mainly available techniques” and provide recommendations and create no obligations for SMEs. further ensure their periodic The draft law “On Industrial Emissions” was developed In this context, it is necessary to establish a sustainable mea- updating. by the MEPA and submitted to the Parliament for discus- surement/monitoring and reporting system for the effective 17 Reactive power is the sion in December 2022. The law aligns with the DIRECTIVE development of EE policy in the private sector. Going forward, power needed to establish and 2010/75/EU on industrial emissions (integrated pollution pre- EE programs should be developed based on market analyses maintain electric and magnetic vention and control). The purpose of the draft law is to pre- with plans open to public consultation and periodic evaluation. fields in alternating current vent emissions into the ambient air, water, and land because Similarly, penalties for non-compliance with EE programs for circuits. Industrial consumers of industrial activities or, where this is practically impossible, large energy users should be in place. In addition, creating may produce excessive to reduce and control emissions and prevent the generation of a program to publicly recognize commercial end users who reactive power, which can waste. This law defines procedures and conditions for issuing have achieved significant energy savings could also be valuable cause several problems on the grid, including increased integrated permits for industrial activities causing pollution, incentive measure. Similarly, setting charges for industrial con- losses, voltage instability and the control of the fulfillment of these conditions, and the rights sumers that generate reactive power is another good practice capacity limitations. and duties of state bodies and natural/legal entities. The law is that should be considered.17 D Programs and Interventions Several programs spurring innovation, energy savings ducing the SF taxonomy, commercial banks and microfinance and renewable energies in the private sector are underway organizations used their definitions of green credit and iden- in parallel with developing the EE policy and legislative tified activities considered eligible for green credit. The NBG framework. Through budgetary support, the MoESD estab- adopted the Regulation on Loan Classification and Reporting lished two sister agencies, Georgia’s Innovation and Technol- according to the Sustainable Finance Taxonomy (Taxonomy ogy Agency (GITA) and Enterprise Georgia, to promote SME Regulation) for commercial banks in July 202218. The SF taxon- development, strengthen their competitiveness and spur in- omy broadens existing definitions, increasing the number of novation. Both agencies provide financial support to SMEs projects defined as green and sustainable. It officially defines and entrepreneurs and a broader range of services, includ- green, social, and sustainable loans and sets reporting require- ing access to special infrastructure, mentoring, training, and ments for taxonomy-aligned loans for commercial banks. This various advisory services. Despite recognizing that Georgia new taxonomy is expected to attract more green investments is progressing in the supporting innovation and technology and increase financing possibilities for SMEs carrying out re- adoption, enhancing wide-ranging adoption of energy-efficient source-efficient and clean production measures. technologies and energy-saving measures remains a challenge. In Georgia, the Green Economy Financing Facility (GEFF) 18 The Taxonomy The Sustainable Finance Taxonomy adopted by the NBG provides finance and advice to help businesses become Regulation entered into force in 2022 provides a classification system for identifying ac- more competitive to reduce energy costs by investing in on January 1, 2023 (see tivities consistent with the country’s objectives regarding high-performance technologies and adopting EE practic- First Green and Resilient Georgia Development Policy climate change, green growth and sustainable develop- es. The GEFF has been implemented since 2018, succeeding Operation). ment. The primary purpose of the Sustainable Finance Tax- the 2012 Energocredit program.19 The GEFF is a product of the onomy is to support the development of a sustainable finance European Bank for Reconstruction and Development (EBRD), 19 See EBRD and partners market and contribute to the country’s sustainable develop- supported by the Green Climate Fund (GCF) and the Austrian deepen green finance in ment. It consists of Social and Green taxonomies. Before intro- Federal Ministry of Finance (BMF). Currently, the Facility sup- Georgia. 16 Chapter 1 Technical Report 20 In the Green Technology ports Georgia’s green economy transition with US$57.75 mil- With Net Metering, companies can satisfy their consumption Selector, items must meet lion through local financial institutions. There are two basic needs, save on utility bills, and even get reimbursements from or exceed high technical approaches: pre-approved investments available on the EBRD the distribution company. Moreover, they increase clean ener- performance standards GEFF Green Technology Selector20 and assessed investments gy generation, reducing the environmental impact. Remark- defined by GEFF engineers and set by the EBRD as with free technical consultancy.21 ably, the micro-generating power plant may not necessarily technical eligibility criteria for be owned by energy consumers; it can be acquired through a GEFF. Eligible technologies Net Metering is a regulatory mechanism of microgenera- leasing contract, rented or other types of financial agreements. include refrigerators and tors with a capacity below 500 kW active since 2016. The The Net Metering mechanism has become popular among freezers; air conditioners; LED integration of variable renewable energy sources (e.g., solar, medium-large commercial firms in Georgia. Due to the latest lighting; gas boilers; biomass wind) into the system has been limited to avoid a deteriora- electricity price increases in 2021, the installed capacity of mi- boilers; hot water cylinders; tion of the quality of the electricity supply in the country. Net crogenerators has increased significantly in the last years and thermopane windows and Metering lets businesses and residential consumers offset their reached over 30 MW. However, for many companies the level doors; building insulation materials; heat pumps; solar electricity costs by exporting excess power back to the grid of 500 kW is relatively low and limits their capacity to expand collectors; solar PV; energy generated by microgenerators if, in the case of Georgia, they their use of renewable sources (mainly solar). metering and control systems; are connected to the same distribution company as consumers. and water-saving systems and technologies. 21 The local GEFF team, particularly the local experts of the Internationale Projekt Consult (IPC) GmbH (Frankfurt am Main, Germany), provide technical support services out under contract with the EBRD. E Prices and Subsidies Electricity and gas tariff regulations in Georgia distin- level. While in the former case, tariffs are set during the bilat- guish between ‘small’ and ‘large’ non-household energy eral agreement, the most important adjustment to electricity consumers, ultimately affecting charges paid and subsi- tariffs for regulated consumers occurred in January 2021, when dies received. Tariffs applied to non-household direct cus- they increased by 52-90 percent, bringing prices more in line tomers with the largest electricity (gas) consumption are not with marginal costs (Table 1). In 2022, non-household gas prices 22 Other non-household regulated, meaning the price depends on a bilateral agreement rose about 20 percent due to the substantial increase in the na- customers (including public between the customer and the independent producer. The state tional currency (GEL) exchange rate to the US dollar (Table 2).22 sector) consumed 12 percent does not subsidize electricity (gas) tariffs for this segment of of gas and 51 percent of electricity in the same period consumers. But regulated non-household consumers -typically Despite recent price adjustments, energy charges in Geor- GEOSTAT energy balance of small consumers- buy electricity from distribution companies gia are still below those of aspirational EU peers. Electricity Georgia, 2022 at a subsidized tariff, with the price varying with the voltage and gas prices vary substantially across countries due to dif- 17 Chapter 1 Greening Firms in Georgia ferent national energy policies, the availability TABLE 1 Non-Household Regulated Electricity Prices of fossil and non-fossil resources, transportation (GEL/100 kWh, Excluding VAT) costs, regulations and market functioning. Elec- tricity and gas prices in Georgia in 2021 were like those in Türkiye, the Slovak Republic, Armenia Distribution Voltage level 2017 2018–2020 2021– Change and Latvia. Yet, prices are almost between one- and supply present 2021–2020 third and half of those charged in Bulgaria, Croa- companies (percent) tia, Romania or the Czech Republic. Relatively low energy prices in Georgia re- Telasi / Telmico High Voltage 12.324 13.653 23.195 70 spond to a history of subsidizing energy pric- (35–110 kV) es, particularly for natural gas and electricity. The government has subsidized households, SMEs, Medium Voltage 12.981 14.307 25.124 76 and socially vulnerable groups to help offset the (3.3–6.10 kV) cost of energy. Government assistance has been provided as tax and non-tax support. Tax support Low Voltage 16.740 18.067 27.908 54 has mainly consisted of VAT and other tax exemp- (220–380 V) tions, but non-tax measures have been gas and electricity direct transfers to targeted beneficia- Energo-Pro High Voltage 11.502 12.593 23.955 90 ries. Table 3 presents the energy subsidies during Georgia / EPG (35–110 kV) the last five years by program, type of support and Supply energy source based on the World Experience for Medium Voltage 12.681 13.772 26.169 90 Georgia project for the OECD.23 (3.3–6.10 kV) In addition to transfers to consumers, Geor- Low Voltage 16.776 17.867 27.085 52 gia allocates significant resources through (220–380 V) subsidies to keep energy prices low compared Note: kWh = kilowatt-hour; kV = kilovolt; V = volt. Source: Georgian National Energy and Water Supply to regional and aspirational countries. Explic- Regulatory Commission (GNERC). it subsidies—subsidies due to supply costs great- er than retail prices—for all energy sources24 as a percentage of the GDP are relatively high (Figure 6). According to the IMF (2022), explicit energy TABLE 2 Non-Household Regulated Electricity Prices subsidies in Georgia are above 4 percent of the (GEL/100 kWh, Excluding VAT) GDP, while countries such as the Czech Repub- lic, Italy, Germany, and the Slovak Republic are Distribution Nov 2018 Feb 2019 Feb 2021 Jul 2022 Change 2022 below 0.25 percent. and supply –Jan 2019 –Jan 2021 –Jun 2022 –present –2021 However, energy subsidies can create disin- companies centives for the green transition in Georgia Tbilisi Energy 68.22 72.88 93.22 114.41 23 percent in several ways. Since prices reveal information to producers and consumers to allocate resources Jan 2018– Jan 2019 Dec 2021 Aug 2022 Change efficiently, cheap energy may discourage efforts to Dec 2018 –Nov 2021 –Jul 2022 –present 2022–2021 improve EE and adopt greener technology. First, artificially low energy prices can distort consum- SOCAR Georgia 72.88 80.51 97.46 116.95 20 percent ers’ behaviors, leading to overconsumption and Gas waste. Excessive consumption can also threaten energy infrastructure by putting more strain on Source: Georgian National Energy and Water Supply Regulatory Commission (GNERC). the power grid and making energy supply less re- liable and stable. This can be particularly prob- lematic in the case of Georgia, where the energy sector is undergoing substantial reforms and in- frastructure investments. Second, cheap energy reduces the financial returns of EE investments, so many projects may turn out unprofitable. For example, firms with outdated, energy-consuming machinery may have low incentives to upgrade 23 National Integrated Energy and Climate Plan of Georgia, Drafted in September 2022. 24 Energy sources include gasoline, diesel, kerosene, liquid petroleum gas, oil, natural gas, coal and electricity. 18 Chapter 1 Technical Report FIGURE 5 Electricity and Gas Prices by Country in 2021 PANEL A Electricity Prices in Industry PANEL B Gas Prices in Industry –2021 IEA –2021 IEA USD/WMH USD/WMH 0 40 80 120 160 200 0 10 20 30 40 50 60 70 Bulgaria Bulgaria North Macedonia Hungary Romania Romania Croatia Croatia Czech Republic Bosnia and Herzegovina Bosnia and Herzegovina Czech Republic Hungary North Macedonia Moldova Georgia Ukraine Slovak Republic Lithuania Türkiye Georgia Moldova Slovenia Ukraine Latvia Latvia Türkiye Lithuania Slovak Republic Armenia Armenia Slovenia Source: International Energy Agency it, invest in on-site green energy generation or improve ener- Reforms to change energy prices are complex and require gy management. The return of such investments under high an adequate planning and communication. To overcome price distortions would be relatively low compared to other these challenges and promote EE in Georgia, policymakers portfolio options. Last but not least, energy subsidies can divert may need to consider a range of strategies, including phas- 25 This report does not address the issue of the resources away from more sustainable and socially beneficial ing out subsidies gradually, providing targeted support for EE political economy of energy uses, such as developing renewable and cleaner energy sources measures, and implementing policies that promote more sus- price reforms. or supporting vulnerable populations. tainable energy use and development.25 19 Chapter 1 Greening Firms in Georgia TABLE 3 Energy Subsidies in Georgia (Million GEL) Program Support Indicator Fuel 2015 2016 2017 2018 2019 scheme VAT exemption for imported natural gas tax expenditure PSE natural gas 37.9 31.8 33.6 31.8 44.6 supplied to thermal power plants Tax exemptions to oil and gas producing tax expenditure PSE natural gas, 8.8 9.1 9.5 9.5 10.4 companies for certain operations crude oil Provision of gas to households in the direct transfer CSE natural gas 4.3 4.1 6.7 7.4 8.0 Kazbegi and Dusheti municipalities for free Utility subsidy for socially vulnerable direct transfer CSE electricity 47.0 25.6 7.3 7.3 8.3 households in Tbilisi municipality Gas subsidy for households living near direct transfer CSE natural gas n.a. n.a. 2.0 2.4 2.6 dividing line of occupied Abkhazia and the Tskhinvali Region Electricity subsidies for households in high direct transfer CSE electricity n.a. n.a. 6.6 9.3 9.8 mountainous areas Electricity subsidy for socially vulnerable direct transfer CSE electricity 1.5 3.8 3.0 2.7 2.9 consumers Electricity subsidy for families with four and direct transfer CSE electricity n.a. n.a. n.a. n.a. 0.1 more children Total (million GEL) 99.5 74.4 68.7 70.6 86.6 Notes: n.a. not applicable. Source: International Energy Agency 20 Chapter 2 Technical Report Chapter 2 FIGURE 6 Explicit Energy Subsidies PAGE 20–34 Percentage of GDP (%) 0.0 1.0 2.0 3.0 4.0 5.0 Estonia Georgia Moldova Bulgaria North Macedonia Decomposing Serbia Hungary the Drivers of Energy Italy Romania Poland Consumption I Lithuania n this section, we calculate new EE mea- may demand more energy even when the average Bosnia & Herzegovina sures and decompose changes in overall energy intensity of firms does not vary or even private sector energy consumption and decreases. Moreover, the emergence of new, more France carbon emissions to understand their energy-efficient sectors—or the expansion of ex- Croatia drivers and identify policy priorities isting ones—can push average energy consump- to improve EE. In the first part, we rely on tion downward, despite an increase in aggregate Czechia firm-level reported energy consumption ex- business energy demand. Further, the reallocation Germany penses (see Box 4) to examine what drives aggre- of market shares between firms with different lev- gate energy consumption (see Box 5 below for els of EE, or entrants and exiters, can also influence Slovakia the method used). Several elements could con- the overall level of energy consumption and emis- tribute to increasing energy consumption. First, sions. Identifying and measuring these margins is an expanding economy where output is growing critical to the understanding of the drivers of EE. Notes: Explicit subsidies reflect subsidies due to supply costs greater than retail prices. Source: Eurostat and IMF. 21 Chapter 2 Greening Firms in Georgia BOX 4 Data sources Throughout this section, we rely mainly on number of firms in specific sectors while a suffi- firm-level data collected by the National cient number in others. To deal with sample size Statistics Office of Georgia (GEOSTAT). The concerns, we set the level of sector aggregation National Statistics Office of Georgia conducts (from lowest to highest) based on a minimum systematic statistical surveys of enterprises number of firms within each sector. Following for individuals and legal persons. Our database a bottom-up approach, the minimum required is a panel of firms with at least one employed number of firms in this analysis is 280 busi- person (including the owner) over 2007-2021, nesses in 2007-2021 (20 firms per year). Put so we exploit variation across and within firms. differently, we count the number of firms at the The survey consistently gathers information on 4-digit level of NACE Rev. 1 and those indus- annual sales, intermediate consumption, num- tries with at least 280 observations are defined ber of employees, assets, location and type of at the 4-digit level. Those sectors that cannot ownership. Also, it collects data on aggregate be defined at that level are considered at the 3-, and source-level energy expenses and con- 2- and 1-digit level following the same criteria sumed quantities of electricity (kilowatt-hours, except for the latter, which is determined based kWh) and gas (cubic meters, m3). This infor- on a minimum of 100 observations rather than mation allows for computing producer-level EE 280. and carbon emissions metrics. Despite having granular data and high cover- The analysis concentrates on individual or age rates of micro, small, medium and large legal persons with a commercial purpose. producers, some given limitations introduce We include firm-level data of businesses in Ag- some caveats to this analysis. First, energy riculture and fishing (activities 1 to 5 of NACE and productivity measures are computed for Rev. 1), Mining and quarrying (10 to 14), Man- firms that report all necessary variables (i.e., ufacturing (activities 15 to 37), Utilities (40 to sales, labor, inputs, and capital). As larger firms 41), Construction (45), Sale and repair of ve- are more likely to report this information, sam- hicles (50), Wholesale and retail trade (51 to ple coverage rates are higher for larger firms, 52), Accommodation and restaurants (55), and limiting the scope of conclusions to firms that Transport and auxiliary activities (60 to 63), are, on average, larger than the typical Geor- Other services (65 to 74) and Education, health gian firm. Second, data on energy expenses by and community services (80 to 93). Public ad- source are available from 2010 onwards, while ministration and Defense activities (75), Private data on consumed quantities of electricity and households with employed people (95), and Ex- gas from 2013 onwards. Finally, firm-level in- tra-territorial organizations and bodies (99) are formation on oil and fossil fuels quantity con- discarded due to the nature of their activity. sumption is not available, so CO2 emissions are estimates based on publicly available data The level of sector aggregation used de- on prices and net caloric values. A detailed de- pends on the number of firm counts. One lim- scription of quantity consumption and emis- itation of the GEOSTAT business data set is that sions estimates is provided in the Appendix, using a narrowly defined industry classification with reported net calorific value factors and (e.g., four digits of NACE) could result in a small emission conversion factors (Table 25A). 22 Chapter 2 Technical Report R elying on over 14,000 firms for 2021, descriptive EE over time for the 25th, 50th, and 75th percentiles of the EE CHAPTER 2.1 statistics show that SMEs (1-99 employees) and distribution and the unweighted mean. There are remarkable Descriptive large (100+) enterprises display very different pro- duction and consumption patterns in Georgia. Table 4 summarizes descriptive statistics of the main differences in energy use across firms that rank higher and lower in the EE distribution. For example, the top 25 percent of firms—p(75)—produce almost 12 times as much output as the Statistics variables used, broken down by size class. There are large (unconditional) differences between SMEs and large firms. Before controlling for sectoral, geographic and bottom 25—p(25)—with the same value unit of energy. But the average and median EE display similar patterns. Consistent- ly across efficiency groups, EE increased between 2007-2011, time factors, firms with at least 100 employees are 8-10 times remained unchanged between 2011 and 2013 and displayed a as large as those with 1-99 employees in terms of sales, inter- negative trend afterward. One source of these differences in mediate consumption and fixed assets and 29 times as large EE levels is related to sector-level (i.e., specific technologies when it comes to employment. Furthermore, energy require- and production methods) and firm-level idiosyncratic char- ments vary remarkably. Large firms consume 15-20 times as acteristics. Figure 7B reveals that EE varies remarkably across much energy as SMEs, both measured in value and quantities and within sectors. Some sectors, such as wholesale and retail consumed. Given the output produced and energy consumed, and other services, tend to use energy more efficiently, that is, SMEs seem more energy efficient than large firms. The average generate more output with the same energy consumption than EE -the logarithm of the output value per value unit of energy others, such as transport and mining. This could be explained consumed- is 3.99 among SMEs (GEL 54 per 1 GEL of energy) by the intensity with which these sectors require energy to carry and 3.63 among large firms (GEL 38 per 1 GEL of energy). out their business activities. For example, transport companies are likely to be more intensive on energy than retail trade, which EE varies across and within sectors due to industry and may require more labor than energy. However, dispersion with- firm characteristics, but overall, EE has evolved similarly in sectors also varies considerably, suggesting that even within across firm groups with large efficiency disparities per- similar industries, firms use energy differently. sisting through time. Figure 7A presents the evolution of TABLE 4 Descriptive Statistics Size class Mean (million US$, constant prices) Mean quantity cons. Mean Counts Sales Raw inputs Fixed Energy bill Electricity Gas Energy Employ- Firms assets (MWh) (‘000 m3) efficiency ment SMEs 0.93 0.77 0.51 0.04 204.5 41.7 3.99 12.07 13,186 Large 8.39 6.51 4.96 0.55 3,284.5 817.8 3.63 353.46 877 Total 1.40 1.15 1.05 0.09 432.5 155.1 3.96 33.36 14,063 Notes: For presentation purposes, financial variables (sales, raw inputs, fixed assets and energy bills) are reported in US$ million of 2018. Electricity (MWh) and gas (thousands of cubic meters) are the average reported quantities of those firms that report energy consumption. Firm counts are all firms reporting employment in the database. Source: World Bank elaboration based on GEOSTAT. 23 Chapter 2 Greening Firms in Georgia FIGURE 7 Energy Efficiency over Time and across Sectors CHAPTER 2.2 PANEL A Energy e ciency over time Energy p(25) p(50) p(75) Unweighted mean Consumption and Emission Decompositions 6.0 5.5 5.0 Energy e ciency 4.5 4.0 G 3.5 eorgia has experienced significant growth in the 3.0 last 17 years (2005-2021), except for 2020 due to the COVID-19 pandemic. During the last 15 years, eco- 2.5 nomic growth in Georgia outperformed most regional peers and Europe and Central Asia (ECA) countries 2.0 (Figure 8). Between 2005 and 2021, the economic ac- 2007 2009 2011 2013 2015 2017 2019 2021 tivity grew at an average annual rate of 4.5 percent, the highest rate after Azerbaijan and 1 percentage point below its long-term growth rate.26 Despite the GDP growth rate slow- PANEL B Energy e ciency across sectors down in 2014-2016 due to the slump in global oil prices, which negatively affected the country’s trading partners and thus ex- ports and remittance inflows, Georgia still outperformed most 25th-75th percentiles Median Mean developing ECA countries. In fact, Georgia is one of the few ECA countries that achieved some convergence in GDP per capita Agriculture with its aspirational peers, transition economies that are now EU member states. Georgia’s GDP per capita (PPP at current Mining US$) in 2021 was 45 percent of the average GDP per capita in the 11 transition economies now EU members (EU-11)27, a 9 percent point improvement compared to the start of the decade. Due to Manufacturing its economic structure relying more on services and tourism, the economic shock in 2020 triggered by the COVID-19 pan- Utilities demic had one of the largest economic effects in the country compared to peers. Construction However, economic growth is not enough to explain the rise Wholesale and retail in energy consumption. Expanding output typically demands more energy but, at the aggregate level, there are factors beyond Hotels and restaurants the pace of economic growth, such as the economy’s structural transformation and the evolution of energy intensity at firm Transport level that also influences energy demand. A decomposition anal- ysis helps quantify the contribution of each different factors Other services that explain the growth of energy consumption in the business sector, including the role of economic growth and structural Education and health change coupled with global trends toward using more efficient production processes. 0 1 2 3 4 5 6 7 Energy e ciency 26 We define the long-term GDP growth trend as the average annual growth rate between 1995 and 2021. 27 These are Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Notes: Figure B reports energy efficiency metrics for 2021. Source: World Bank elaboration based on GEOSTAT. Lithuania, Poland, Slovak Republic, Slovenia and Romania. 24 Chapter 2 Technical Report FIGURE 8 GDP Growth in Georgia, Regional Peers and Selected EU Countries (Index, 2005 = 100) 300 Armenia Azerbaijan Bulgaria Georgia Hungary North Macedonia Poland Romania Türkiye Ukraine 250 200 150 100 50 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: GDP at constant local currency unit (LCU). Source: World Bank Data. BOX 5 Decomposing aggregate energy consumption and CO2 emissions in the business economy The decomposition28 analyses leverage poses changes in aggregate energy consump- firm-level energy consumption and carbon tion (CO2 emissions) into changes due to output dioxide emissions estimates to isolate the ef- growth (‘scale effect’), shifts in the sectoral fects of economic growth, structural change composition of the economy (‘structural trans- and sector and firm intensity. Aggregate en- formation effect’) and changes in the (weight- ergy consumption is defined as the annual ex- ed) energy intensity of sectors, which ultimately 28 A growing body of literature uses penses paid for electricity, natural gas, fossil depends on the changes in the (unweighted) energy and GHG emission decompositions fuels, oil products, and renewables reported average energy intensity of firms (‘firm intensi- to assess what drives energy demand and by the firm from 2007 to 2021. We use total ty’) and the reallocation of output across firms air pollution. Energy and carbon emission growth decompositions have been widely energy costs because most businesses re- within the same industry (‘market reallocation used in energy and environmental economics port aggregate consumption but only a smaller effect’). The intensity measure closely relates to understand how countries deal with number report energy consumption by source to the EE metric used in this report. Energy climate change issues. They help quantify in what magnitude energy consumption and -electricity, gas, oil, and fossil fuel products. intensity is defined as the quantity of energy GHG emission changes are due to economic As for the carbon emissions growth decom- required -deflated energy value consumption- growth, structural change, and sector and position, we estimate firm-level CO2 emissions to produce one unit of output -deflated sales- firm-level behaviors. For example, Rotten & von Graevenitz (2021) use this approach to examine based on reported and estimated energy quan- while EE refers to the output produced per the drivers of the growth of carbon emissions tity consumption and carbon conversion factors unit of energy input (US Department of Energy, in the German manufacturing industry in the (see Box 7 for further details on the estimation 2023). Below, we provide the conceptual and last decade. In the same vein, Levinson (2015) and Shapiro & Walker (2018) decompose 1990- of carbon emissions). formal definition of each term and an illustra- 2008 air pollution emissions changes from tion (Diagram 1) for further clarification. the US manufacturing using firm-level data. The method used throughout this report dis- The latter authors also explore the economic forces and environmental regulations driving entangles the factors that affect aggregate Scale effect (production scale): changes in emissions reduction. Also, Ang (2015), Ang & energy consumption or carbon emissions aggregate energy consumption (CO2 emissions) Zhang (2000), Liu & Ang, (2007), and Xu & Ang into those related to economic expansion, attributable to output growth or change in the (2013) decompose aggregate energy demand shifts using the Logarithmic Mean Divisia Index structural change, reallocation of resources overall size of the economy. (LMDI), a very similar technique to the one used and firm performance. This document decom- in this report. 25 Chapter 2 Greening Firms in Georgia BOX 5 Decomposing aggregate energy consumption and CO2 emissions in the business economy (cont.) Overall intensity: changes resulting from the economy’s structural transformation DIAGRAM 1 (‘structural effect,’ i.e., changes in output shares across sectors and firms) and chang- es from variations in firms’ average energy (CO2) intensity. Overall efficiency can be Decomposing aggregate energy further decomposed into the structural transformation effect and the sector intensity consumption and CO2 emissions in effect: the business economy • Structural effect (composition): changes in energy consumption (CO2 emis- A sions) due to changes in the overall composition of the economy (expansion and Energy consumption contraction of sectors with different energy requirements). (CO2 emission) change • Average sector intensity effect (technique effect): changes in the (weighted) B.1 B.2 sector-level average energy consumption (CO2 emissions), which can be fur- Overall efficiency Scale effect ther decomposed into: → Sector-level (unweighted) average intensity component: the arith- C.2 metic unweighted mean of the energy use across firms within the same Sector intensity Structural sector. transformation → Market reallocation component: the reallocation of output across D.1 D.2 firms within the same sector. Market Unweighted reallocation average firm Formally, equation (1) presents the decomposition of energy consumption (CO2 emis- intensity sions) between t and t₀: ∆Et,t₀=scalet,t₀+ structural changet,t₀+market reallocationt,t₀ + firm energy intensityt,t₀ (1) ∆Et,t₀= ∆Yt,t₀ t₀ Scale effect 26 Chapter 2 Technical Report CHAPTER 2.3 Energy Consumption Decomposition Results A The Scale, Composition, and Intensity Effects The rise in energy consumption throughout 2007-2021 energy consumption was mainly explained by output growth was mainly driven by output growth (scale effect) and the and output reallocation to firms that used energy more inten- relative expansion of more energy-intensive sectors (com- sively within industries. A considerable reduction in the energy position effect). In the last 14 years, the energy consumption intensity of plants partially offset this growing energy demand, of the private sector grew by 155 percent. The larger energy de- but these improvements were reversed in the subsequent years. mand was mainly explained by output expansion (scale effect) The steady increase in firm intensity between 2017-2021 im- and the structural change toward sectors that require more plied larger average energy consumption at firm-level, driv- energy to produce one unit of output (Figure 9). In contrast, ing the rise in energy consumption. In this latter period the lower energy requirements at the level of firms (‘firm intensity’) production scale and structural change played a less relevant and the reallocation of output in less energy-intensive firms role while still contributing to expanding aggregate energy within industries helped curb the growth in the energy demand requirements. The contribution of market reallocation also considerably. For instance, had there been no firm energy in- changed. While from 2007 to 2016, high-energy-intensive firms tensity changes and market reallocation across industries, the gained market share within their sectors, since 2017, output private sector energy demand would have increased by 222 rose relatively more in low energy-intensive firms, contributing percent between 2007 and 2021 due to economic growth and to moderate growing energy demand. structural change factors. Put differently, businesses would have required 70 percentage points more energy than the ac- Consistent with the economic cycle of booms and down- tual growth without sector and firms’ intensity improvements, turns, the scale effect is positively associated with the pace of a noticeable result from the economic perspective. GDP growth. Businesses require more energy when they increase production. During the first half of the 2010s, the production scale The contribution of each driver varied substantially. was one of the main drivers of energy consumption growth, con- sistent with higher GDP growth rates. Nonetheless, losing output Figure 9 decomposes energy requirements by computing the dynamism from 2014 onwards curbed energy growth. 3-year energy consumption growth. During 2007–16, the rise in FIGURE 9 Decomposition of Energy Consumption 3-year cumulative growth at the end of the period relative to baseline, percentage 250% Scale e ect Firm intensity Structural transformation Market reallocation 70% Change in agg. energy consumption GDP growth (right) 60% 200% 50% 150% 40% 100% 30% 20% 50% 10% 0% 0% -10% -50% -20% -100% -30% -150% -40% 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2007 -2021 Source: World Bank elaboration based on GEOSTAT 27 Chapter 2 Greening Firms in Georgia The persistent contribution of the structural change to footprint of production. While we further discuss what could energy consumption growth since 2016 indicates that be driving this lack of efficiency improvement in the following high-energy-intensive industries grew faster than low sections of this report, for now, it is important to stress that energy-intensive ones, possibly due to incentives in the underinvestment in green and energy-saving technologies and institutional environment and relatively low energy poor improvements in management quality are key drivers of price. Throughout the period, structural transformation (i.e., these results. Ultimately, this is driven by a lack of incentives, production shifts across sectors and changes in patterns of an underdeveloped green innovation ecosystem and poten- sectoral specialization) pushed energy consumption upward, tial low investment returns, and problems in financing green suggesting that energy-intensive industries gained market projects. share, especially from 2016 onwards. The fact that resources flew to more energy-intensive sectors could be due to different Finally, market reallocation counterbalanced firm inten- reasons. A plausible explanation is that low, subsidized ener- sity, moderating energy consumption growth in the last gy prices can enhance the performance of industries with a years. Market reallocation was negatively associated with firm higher-than-average energy intensity to the detriment of less energy intensity. For example, when the energy intensity of energy-intensive activities. Because energy is one of the es- firms decreased (2010-2016), market reallocation was positive sential inputs for production, cheap energy prices may steer in nearly all years, meaning that output shares increased in investments in high-energy-intensive sectors, explaining the higher energy-intensive firms. Conversely, market reallocation positive contribution of structural transformation to energy was negative when firm intensity increased (2017-2021). Over consumption growth. Alternatively, the positive structural 2017-2021, output went to low energy-intensive firms, curbing change term could result from higher credit barriers for na- energy consumption growth. The reallocation of output in scent activities, which are more likely to be greener than tra- firms that use energy less intensively could suggest that more ditional ones. Although bank credit to the private sector in efficient (less intensive) firms are gaining market share within Georgia is above the regional average and improved in the last sectors. If this were the case, market reallocation should also years, it is mainly focused on traditional instruments, which contribute to productivity growth positively. Indeed, this is might particularly affect new industries and startups that are likely to be the case. The results of the Olley-Pakes static de- typically greener. Additionally, business angel and seed-stage composition (see Figure 51A in Appendix), which decomposes investment activities that would fuel innovative growth and are the aggregate productivity (TFP) change into the unweighted likely to have more impact on less energy-intensive sectors are TFP and the market reallocation terms, show that the latter not yet mature, hindering their performance. term was positive during 2016-2019, consistent with results in Figure 9. The increase in firm energy intensity in the last six years coincides with global technology advances toward more Overall performance of aggregate energy consumption in efficient and greener machinery and equipment and the Georgia over 2007-2012 was similar to that of the Turkish rising awareness of the carbon footprint of economic ac- manufacturing sector in 2005-2012. Research on energy tivity. Despite global trends toward saving energy and reducing consumption in Türkiye (Sahin, 2017) shows that the rise in waste, Georgian firms have exhibited a disappointing perfor- energy demand and the factors driving this increase were like mance. The increase in firm energy intensity in the last years those of Georgia. Manufacturing firms in Türkiye increased suggests serious difficulties in using energy more efficiently, energy consumption by 27 percent between 2005 and 2012, that is, being less intensive. On the one hand, higher intensity with output expansion being the primary factor driving energy could result from the lack of innovation and underinvestment demand. Furthermore, firms decreased their energy intensity in energy-saving technologies such as on-site green energy in both countries. However, the effect of composition shifts generation, energy management systems, cutting-edge tech- on energy consumption differed between countries. In Geor- nology, LEED certifications and LED lighting. On the other, gia, the structural change tended to push energy consumption it could also point to poor quality management. As Levinson upward (Figure 9), while in Türkiye, Sahin (2017) reports that (2015) and Rotten & von Graevenitz (2022) emphasize, tech- within the manufacturing sector, output shifted to activities nology adoption and improvements in energy use at the firm with lower energy requirements per unit of output. level show the highest potential to reduce the environmental 28 Chapter 2 Technical Report B Decomposing the The Dynamic Olley-Pakes Decomposition ergy-intensive entrants and exiters are relative to (DOPD) provides a complementary analysis incumbents). So, the DOPD examines the evolu- Energy Efficiency of to understand what drives changes in EE. tion of aggregate EE calculated as the logarithm Firms Dynamically Based on the energy consumption decomposition of the revenue-to-energy ratio29 (Melitz & Polanec, results, the DOPD complements this analysis by 2015). The main advantage of this decomposition inspecting the drivers of EE shifts. In the previous is that it splits aggregate EE changes into compo- decomposition, the intensity term -the unweight- nents related to the changes in the efficiency of in- ed average energy intensity between firms of the cumbents, how resources are reallocated among same sector- does not provide information on these firms, and the effects that firm entry and whether changes are explained by within-firm exit may have on aggregate EE (Box 6). Results improvements or selection effects (i.e., how en- are presented in Figure 10. BOX 6 Decomposing Energy Efficiency Using the Dynamic Olley-Pakes Decomposition (DOPD) We also decompose the dynamics of higher (lower) EE than incumbents, then aver- EE, defined here as the deflated (logged) age EE will increase (reduce), ceteris paribus. sales-to-energy expenses ratio, into the Hence, the entry component measures the ef- within-firm term, the market reallocation fect of entry on EE changes. effects of surviving firms and the changes in EE caused by firm entry and exit. Following Firm exit: Similarly, the exit component mea- Melitz & Polanec (2015), we perform a dynamic sures the change in EE caused by firms quitting Olley-Pakes decomposition (DOPD), noting that the market. If exiting firms display higher (low- previously explained criteria for sector catego- er) EE than incumbents, EE will fall (increase) ries apply. due to firm exit, ceteris paribus. Within-firm intensity (within term): Accounts Net entry (entry minus exit): Since our data- for firm-level changes in EE due to technology base restricts the analysis to all businesses re- adoption, innovation, and upgrade of organi- porting energy consumption and sales, it may zational capabilities among incumbents (firms be the case that for some firms, entry and exit active in t and t – 1). could be related to an administrative rather than an economic event (i.e., entering or exit- Market reallocation (between): Accounts for ing the database due to reporting rather than 29 Box 8 provides further EE variation due to changes in the reallocation an actual opening or closure). As this is likely to details on the computation of market shares among incumbents (firms in- affect both entry and exit, reporting the net en- and advantages of the energy efficiency calculation. In a creasing or reducing their market shares over try component would (at least partially) account nutshell, this metric provides time). for churning due to purely administrative events intuitive results as it is and thus provide a more accurate measure of constructed as the (logged) output generated per unit of Firm entry: Changes in aggregate EE may be the contribution of actual firm demography energy, following the same explained by new firms displaying higher or low- events to aggregate EE. rationale as for firm or labor er EE relative to incumbents. If new firms have efficiency indicators. 29 Chapter 2 Greening Firms in Georgia FIGURE 10 Dynamic Olley-Pakes Energy Efficiency Decomposition 3-year aggregate energy efficiency change 40% Within-firm Between-firm Net entry Agg. change in energy e ciency 30% 20% 10% 0% -10% -20% Source: World Bank -30% elaboration based on 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 GEOSTAT. The recent decline in aggregate EE is linked to poor with- consumption. However, the recent decline in the selection in-firm performance: established firms are failing to effect contribution raises concerns about whether this trend increase EE, while both entrants and exiters contribute weakens. The COVID-19 pandemic put many firms under finan- to enhancing aggregate efficiency. The EE of the Georgian cial strain and led many to go bankrupt when no government business sector shows a disappointing performance. Aggregate or financial assistance were available. EE only grew between 2011-2013, remained stable in 2014-2017 and declined substantially between 2018 and 2021. From 2010 Lack of investment, difficulties in adjusting energy con- until 2017, the selection component (i.e., net entry) played a sumption and poor managerial capabilities could be be- crucial role in explaining overall efficiency changes improve- hind the recent decline in firm-level EE. Various reasons ments, meaning that entrants tended to be more efficient than may explain the disappointing EE performance in Georgia. For incumbents and exiters less efficient. Since then, within-firm example, due to inherent characteristics of the production pro- performance has been the main driver for explaining the de- cess, firms may face difficulties adjusting energy consumption cline in EE. The within-firm term has been negative and large. to a decline in output. Other explanations may be the firms’ In other words, the average efficiency of continuing firms (i.e., low awareness of their environmental footprint or a scheme those active in t and t-3) declined considerably. Notably, the of incentives that distorts energy consumption. We observe between-firm component30 was negative in all years until 2019 among incumbents that average energy requirements grew (except for 2016 and 2018), but since the COVID-19 pandemic, it disproportionately compared to sales, particularly in mining, has turned positive. This suggests a potential change in output manufacturing, hotels and restaurants, and transport. Howev- allocation patterns. Indeed, firms with higher digital capacities er, as sales grow, this decline in efficiency is more likely to come (e.g., online sales channels and customer relationship man- from a weakening scheme of incentives rather than rigidities agement -CRM-, cloud and network technologies), which are in the production process. more likely to be energy efficient, are expected to have gained market share during the last pandemic. Investing in outdated technologies and poor managerial quality could also explain efficiency underperformance. The positive contribution of the selection effect suggests In particular, technology and the vintage of capital used by the that startups are greener and have replaced exiters with firms are key for determining energy consumption. Investing outdated processes and technologies. That the net entry in modern, greener technologies can substantially improve term remained positive nearly in all years indicates that new- efficiency, so a potential explanation for not seeing this is that ly born businesses are more efficient than incumbents, and firms are investing in outdated machinery. Beyond the vintage closing plants are less efficient than incumbents, on average. of capital, there are also factors within the firms that may affect Entrants are likely to be greener for a variety of reasons. First, efficiency. A growing body of economic literature shows (posi- they are likely to invest in new technology, go digital, and in- tive) links between EE and the quality of management (Bloom 30 Positive when incumbent novate. Second, startups are more likely to enter low-pollutant et al., 2010; Martin et al., 2012). Good quality management, in firms with higher-than- activities due to digital trends, and customers also demand particular green managerial practices, can boost overall firm average efficiency gain market share (among incumbent products with a lower environmental footprint. Finally, new- efficiency -through improving input choices- and be a strate- firms) and negative when they ly born firms also address environmental aspects internally, gic complementarity for unlocking innovation and greener lose market share. monitoring and evaluating their carbon footprint and energy technology adoption (Grover, Iacovone & Chakraborty, 2019). 30 Chapter 2 Technical Report FIGURE 11 Change in Aggregate Energy Consumption by Industry, 2018–21 Scale e ect Firm intensity Structural transformation Market reallocation Change in agg. energy consumption Share of energy consumption (baseline) -150% -100% -50% 0% 50% 100% 150% 200% Agriculture and fishing Mining and quarrying Manufacturing Utilities Construction Wholesale and retail Accommodation and restaurants Source: World Bank Transport act. elaboration based on GEOSTAT. Other services C Which Businesses Have Been Driving Increasing Energy Intensity? Firm energy intensity has driven 2018-2021 energy con- Structural transformation and market reallocation in- sumption growth across sectors. This subsection concen- creased energy consumption in nearly all sectors. In the trates on the latest developments in energy consumption sectoral analysis, the structural transformation effect measures growth. Despite some heterogeneity, the general increase in the reallocation of output between low and high-energy-inten- aggregate energy consumption across sectors was driven main- sive industries within each activity and market reallocation ly by the intensity with which firms use energy and, in specific gauges how output flows between low- and high-energy-inten- industries, due to output growth and market reallocation ef- sive firms within narrowly defined industries. For example, the fects. Across most of the biggest energy consumers, manufac- structural transformation term in the manufacturing activity turing (28 percent of total energy consumption), construction captures how output is allocated between manufacturing ac- (16 percent), and transport activities (12 percent), the rise in tivities (e.g., processing of food and processing of beverages), the average energy requirements per unit of output was, by and the market reallocation term shows how output is allocated large, the main contributor to energy consumption growth, across businesses in each industry of the manufacturing sec- especially after comparing the magnitude of this effect to the tor (e.g., within the food industry). Results plotted in Figure 11 scale, structural change and reallocation terms. Nonetheless, confirm the conclusions derived from Figure 10—beyond out- in a few activities that also account for a large share of ener- put growth, wholesale and retail and transport activities have gy consumption, such as utilities (17 percent) but also among been driving the increase in 2018–21 energy consumption. After smaller consumers, such as agriculture (2 percent) and services conditioning on activity, during the last three years, industries (5 percent), the reduction in firms’ energy requirements helped with higher energy intensity gained output share, and so did moderate the increase in energy consumption considerably. high-energy-intensive firms within sectors. Only in agriculture and mining activities most of the The rise of firm energy intensity in lower-middle, up- increase in energy consumption could be attributed to per-middle and high-energy-intensive sectors largely output growth. The scale effect played a meaningful role contributed to the increase in energy consumption. Fig- among primary activities, was less relevant in manufacturing, ure 12 performs the same decomposition grouping sectors into 31 To avoid industries switching across quartiles utilities, commerce and transport, and even pushed energy quartiles according to the intensity level of energy consump- over time, the classification is consumption downward in industries such as construction, tion.31 The categorization is based on industries rather than based on the average energy accommodation and restaurants, and services. Figure 11 shows firms, so plants with heterogeneous energy intensity levels intensity over time. Energy that, other factors constant, output growth in agriculture and can coexist within a certain quartile. After splitting the sample intensity is computed at the mining would have required doubling energy consumption. into industry-level quartiles, the aggregate change in energy sector level for each year and However, the scale effect was very modest or even negative in consumption is decomposed into the scale, structural trans- then averaged over the period the remaining industries, which means that economic growth formation, firm intensity and market reallocation terms. The 2007-2021. Then, sectors could not principally explain larger energy consumption. structural transformation terms are related to the change in are divided into quartiles according to their energy output shares across sectors within each quartile category, intensity value. while market reallocation refers to shifts in market shares of 31 Chapter 2 Greening Firms in Georgia firms within industries conditional on a given quartile. The But the reallocation of resources within sectors toward less figure shows substantial variation in the firm intensity term, energy-intensive firms partially countered the role of firm in- even after considering the magnitude of the energy consump- tensity. From a policy perspective, this is an important insight tion change, meaning that firm-level energy intensity drove the because it reveals that firm-level dynamics in energy require- increase of energy consumption in low and high-energy-in- ments during the last years have explained energy consump- tensive sectors over 2018–21. The rise in energy consumption tion growth. Put differently, the decompositions highlight the of lower-middle-, upper-middle- and high-energy-intensive need to address the green transition putting firms’ capacities at industries has been almost entirely driven by increasing energy the center and encourage investments in industries and firms requirements at the level of firms to produce the same output. with lower energy requirements. FIGURE 12 Firm-Level Energy Intensity and Change in Energy Consumption by Sectoral Energy Intensity, 2018–21 Scale e ect Firm intensity Structural transformation Market reallocation Change in agg, energy consumption Share of energy consumption (2017) -75% -50% -25% 0% 25% 50% 75% 100% Low intensive Lower-middle intensive Upper-middle intensive High intensive Source: World Bank elaboration based on GEOSTAT. D From Energy Consumption to CO2 Emissions BOX 7 Measuring energy quantity consumption and CO2 emissions The previous analysis raises concerns about the accura- and fossil fuel sources. Nearly 12 percent and 4 percent of cy of the energy intensity measure if the case that energy firms report electricity (kWh) and gas (m3) consumed quan- tariffs (price per unit of energy kilowatt-hour, kWh) are tities (25 percent of businesses reporting electricity and gas firm-specific or vary. For example, energy regulators may costs). For the remaining 75 percent of firms reporting elec- set non-residential energy prices according to businesses’ tricity and gas bills but not quantities, we use the average characteristics such as the consumption or voltage level, lo- price at the sector-size-year level to estimate electricity and cation or sector of activity or suddenly adjust energy tariffs. gas quantity consumption (Figure 52A in Appendix A). Next, In such cases, EE variation would be driven by price discrim- we estimate fossil fuel quantity consumption by dividing con- ination or tariff adjustment factors rather than by the actual sumption value by the average gasoline price, diesel and ker- output produced per unit of energy unit. osene fuels at the year level and oil quantity consumption by dividing oil value consumption by the average oil price. Last, Conclusions remain consistent after using energy quan- we convert energy quantity consumption into Terajoules (TJ) tity consumption instead or energy bills. We aggregate using net calorific value factors reported in Table 25A (see unit-equivalent energy consumption from electricity, gas, Appendix A). Remarkably, we find very similar conclusions. 32 Chapter 2 Technical Report BOX 7 Measuring energy quantity consumption and CO2 emissions (cont.) Energy quantity consumption provides a unique oppor- Last, technology adoption could reduce emission inten- tunity to understand carbon (CO2) emissions drivers. sity levels beyond changing energy sources. The third Because of climate change, measuring the environmental channel implies changing production techniques by using footprint of economic activity is vital to assess whether pol- machinery and equipment that emits less carbon while using icy actions are effective in transitioning toward a greener energy more efficiently. Technology-led changes are expect- economy. Although the correlation between energy con- ed to yield higher benefits and offer the largest potential for sumption efficiency and CO2 emissions efficiency is expected global pollution reduction (Rottner & von Graevenitz, 2021; to be high, it is not necessarily one because of factors with Levinson, 2015). varying potential affecting how energy quantity consumption is transformed into emissions. The estimated elasticity between energy and CO2 emis- sion efficiency ranges between 1.05-1.06 when quantities The first channel is related to industry composition are used and 0.92-0.94 if bills are used. We assess the shifts. Sectoral composition changes toward less emis- relationship between energy and carbon emission efficien- sion-intensive activities can potentially reduce the aggregate cy -log of the sales-to-energy and sales-to-emission ratios- carbon intensity. However, while it could be regarded as a by regressing CO2 emission efficiency on EE and account- cleanup, from the environmental perspective benefits may ing for industry, geographic and firm fixed effects. Results be limited (Levinson, 2015). For example, if these shifts are (Table 27A, Appendix A) show that a 1 percent increase in driven by outsourcing goods and services requiring higher energy (expenses) efficiency is associated with an average CO2 content, they will reduce emissions locally but not re- 0.92-0.94 rise in carbon emission efficiency, implying some gionally. imperfect pass-through of EE on carbon emission changes. However, once differences in energy prices are cleaned up, Switching energy sources can affect the aggregate level a 1 percent rise in energy (quantity) efficiency is associated of emissions intensity, but this potential can be exhaust- with a 1.05-1.06 increase in CO2 emission efficiency, suggest- ed in the short- or mid-term (Rottner & von Graevenitz, ing a larger-than-1 pass-through, consistent with firms not 2021). Changing from fossil fuels to electricity could sub- investing in greener, less pollutant technology and process- stantially reduce CO2 emissions intensity due to differences es and switching toward more pollutant energy fuels. More- in the CO2 energy content of each source. Although high-in- over, energy carbon content factors can change over time, come countries have made a lot of progress moving from affecting the pollution content of energy use (see Appendix A fossil fuels to renewable energy, there is still an open margin for a detailed methodological description). for moving forward in this direction in less advanced ones. Carbon (CO2) emissions grew over 2015-2021 mainly be- cation helped moderate the growth rate of carbon emissions, cause of production scale and sector composition effects especially during the first year of the pandemic. but also due to higher firm carbon emission intensity. Alongside rising energy consumption, CO2 emissions grew Firms have struggled to reduce their dependency on fossil steadily from 2015 to 2021, only curbed by the COVID-19 pan- fuels, which remains a main obstacle to the green transi- demic in 2020. Despite recent trends showing countries’ re- tion. Beyond investing in EE systems, upgrading machinery markable progress in cutting GHG emissions alongside GDP and equipment, going digital, and enhancing internal capaci- growth and declining carbon intensity alongside higher GDP ties, firms can reduce CO2 emissions by switching to cleaner en- per capita (Burn-Murdoch, 2022), the CO2 emissions of the ergy sources by generating on-site renewable energy or moving Georgian business sector keep growing. But as with energy toward renewable sources. Although this is not a permanent consumption, this rise cannot only be attributed to economic mechanism for lowering carbon intensity, it remains untapped growth. The contribution of output expansion to CO2 emissions in the case of Georgia and exhibits promising short- and mid- was significant in 2018-2019 but less relevant afterward. In term potential. Data shows that although the share of electricity contrast, carbon-intensive activities gaining market shares consumption among businesses grew from 16 percent to 23 increased the contribution of the structural transformation to percent over 2015-2021, the participation of fossil fuels also rose CO2 emissions growth. Figure 13 also shows that businesses in- from 38 percent to 50 percent (see Figure 49A in Appendix A), creased the intensity of emissions (carbon emissions per unit hence perpetuating the dependence on fossil fuels and damp- of output) until 2020. CO2 emissions keep growing in Georgia ening efforts to reduce carbon dioxide emissions. but below energy consumption. Remarkably, market reallo- 33 Chapter 2 Greening Firms in Georgia FIGURE 13 CO2 Emissions and Energy Consumption in Georgia Scale e ect Firm intensity Structural transformation Market reallocation Change in energy consumption Change in emissions 80% 60% 40% 20% 0% -20% -40% -60% -80% 2018 2019 2020 2021 2015-2021 Source: World Bank elaboration Despite considerable resources devoted to clean up pro- and 2008 can be attributed to the reduction in each industry’s duction, international evidence stresses the importance emission intensity rather than shifts in the allocation of out- of coordinated actions to reduce firm-level carbon emis- put between manufacturing activities. The magnitude of the sions to achieve long-lasting results. Empirical evidence decrease in emission intensity was so large that manufactur- from carbon emissions decompositions in Germany over 2005- ing pollution declined by 52 percent to 69 percent, although 2017 reveals that it is not enough to encourage low-pollutant production rose by 35 percent. Shapiro & Walker (2018) also activities or the generation of renewable energy. Rotten & von examine air pollution emissions from US manufacturers over Graevenitz (2021) find that even though the German manufac- 1990-2008. Consistent with earlier conclusions, they show that turing industry has shifted to less carbon-intensive activities the fall in emissions was primarily driven by within-product and increased renewable electricity generation, emissions grew changes in CO2 emission intensity rather than by the scale or due to production scale and businesses consuming more en- composition effects. Remarkably, carbon pollution reductions ergy. Levinson (2015) shows that the fall in air pollution emit- can be mainly attributed to environmental regulations (i.e., ted by manufacturing establishments in the US between 1990 pollution taxes) rather than changes in productivity and trade. 34 Chapter 2 Technical Report E Carbon Pollution Emitted by Low- and High-Energy-Intensive Sectors 32 Technologies such as energy-efficient lighting, variable air/air- High-energy-intensive sectors have driven recent growth between 5 percent to 11 percent. Part of the carbon pollution conditioned volume systems, programmable thermostats, in carbon emissions (2018-2021), although contributing growth is a byproduct of structural change since, in the case of photovoltaic panels and factors are beyond the firm-level performance. The en- Georgia, production is expanding relatively faster in high-pol- renewable or non-fuel vironmental performance of high-energy-intensive sectors is lutant industries relative to low-pollutant ones. However, the generators, robots, online crucial for halting CO2 emissions growth. Energy-intensive decomposition also shows that firms have consistently failed to sales channels, building industries account for eight out of 10 emitted CO2 metric tons reduce carbon intensity. Only in low and high-intensive sectors insulation, efficient motor and have incremented CO2 emissions by 26 percent between have they reduced the amount of CO2 per unit of output. Still, vehicles and engines can be 2018 and 2021 -compared to an economy-wide growth of 19 in lower-middle and upper-middle energy-intensive activities, important for saving energy and cutting carbon emissions, percent. In fact, during the last three years, CO2 emissions only they did not reduce intensity levels. Promoting firm-level effi- in addition to cutting-edge grew considerably (26 percent) in high-energy-intensive sec- ciency32 in these mid-energy-intensive industries becomes an industry-specific technologies tors, while in low, lower-middle and upper-middle energy-in- important policy priority going forward • and production methods. tensive sectors they remained unchanged or even decreased FIGURE 14 CO2 Emissions by Sector Energy Intensity Scale e ect Firm intensity Structural transformation Market reallocation Change in agg. CO2 emissions Share of CO2 emissions (2017) -75% -50% -25% 0% 25% 50% 75% 100% Low intensive Lower-middle intensive Upper-middle intensive High intensive Source: World Bank elaboration 35 Chapter 3 Greening Firms in Georgia Chapter 3 35–47 PAGE Energy Efficiency in Georgia: A Firm-Level Analysis R elying on detailed firm-level information, this interventions. We use the dispersion in efficiency levels as the section focuses on the sectoral dimensions of EE main metric to indicate room for improvement. Next, we eval- to understand the sectoral priorities for policy- uate the potential impact of enhancing EE on aggregate energy makers to maximize energy savings and reduce consumption, production costs, and profits. Before delving CO2 emissions. We start by analyzing efficiency dis- into the analysis, Box 8 introduces the main definitions and persion within and across sectors and the evolution of some of the technical concepts used throughout the chapter. firms’ EE over time to guide the prioritization of public BOX 8 Estimating Firm Energy Efficiency in Georgia EE could be regarded as the output produced per unit of firms, we use energy expenses to report the main results of energy consumed. Measuring efficiency is complex because the efficiency analysis. We also perform robustness checks it requires detailed information on the quantity of energy using deflated sales per unit of TJ consumed. consumed and an aggregate measure of goods and services produced at the firm level. After computing real energy costs and revenues, EE is cal- culated as in (2): This document defines EE as the amount of deflated sales produced per unit of energy expenses consumed. Energy consumption is the sum of electricity, natural gas, fossil fuels, oil products, and renewable costs. Energy costs and revenue are deflated using price deflators to account for price chang- es. We use the consumer price index (CPI) for deflating ener- gy costs and the two-digit producer price index. Because en- where subscript i denotes the unit of analysis (firm), j the ergy quantity consumption is available for a smaller subset of sector, r the region of location, and t the unit of time (year). 36 Chapter 3 Technical Report BOX 8 Estimating Firm Energy Efficiency in Georgia Also, the change in EE is defined as the first difference in the level of efficiency: The Local, Sectoral, and Regional Frontiers of EE The chapter also explores underlying factors that influence the converge process of firm-level EE to the local efficiency frontier since, although energy prices are national, access to technology, knowledge, and capital, know-how, and proxim- where the subscripts “—r” and “—j” denote that firms within ity to innovation hubs can vary across regions and sectors. the same region (5) or sector (6) of the firm are excluded Equation (4) refers to the calculation of the local frontier, from the calculation of each indicator. defined as the efficiency of top firms within the sector and region where the firm operates. The local efficiency frontier Finally, the distance of firm i to the local frontier is cal- is the average efficiency level of the top quartile most efficient culated as the difference between the local frontier firms within their sector and region of operations (4). The efficiency and the firm efficiency . subscript definitions apply as above. Because we do not have energy firm-level data for countries with the highest efficiency levels, the con- vergence analysis only includes domestic frontiers. Similarly, equations (5) and (6) define the sectoral and re- gional frontiers, respectively: CHAPTER 3.1 Heterogeneity in Energy Efficiency between Firms 33 To be even more precise we group firms not only by industry but by industry and size group to narrow the possibility that differences E are not driven by efficiency differences but by differences in the nature of their E efficiency between firms within same industry33 at the 75th percentile of their sector-by-size group produce 6.5 businesses. are even larger than between industries, implying as much output with the same energy inputs as firms at the room for improving EE without affecting output 25th percentile. This suggests that EE firm dispersion is driven 34 Dispersion within and the economic structure. It is not surprising to find more by firm-level energy use than by inherent technological industries-by-size categories EE differences across sectors and size categories as in- and productive differences that typically arise between sector- is a key measure to quantify dustries and business sizes are characterized by specific by-size groups. This is a key finding from a policy perspective the potential gains from reallocation of resources. scales, technologies, and production processes. Howev- because it means there is significant scope to reduce energy For example, Hsieh & Klenow er, in Georgia, we find even larger differences in EE between consumption and improve efficiency levels without affecting (2009) look at within-industry firms within sector-by-size groups than across groups. Figure 15 overall output and the sectoral structure of the economy.34 dispersion to estimate TFP depicts the distribution of EE within and across sector-by-size gains in the United States, groups. As the figure illustrates, EE dispersion within sector- Within-industry EE heterogeneity appears particularly China and India. by-size categories (i.e., differences in EE between firms within a large among big energy consumers, suggesting potential- 35 These includes size and sector) is larger than variation across industry-by-size ly large energy savings from well-targeted interventions. passenger and non-passenger categories (i.e., differences between the average efficiency of a Sectors-size groups with a higher-than-average efficiency dis- land, water and air transport. size and sector group and the economy-wide efficiency). Firms persion, which include transport and storage activities35, con- 37 Chapter 3 Greening Firms in Georgia FIGURE 15 Energy Efficiency Dispersion struction36, agriculture activities, and certain manufacturing across Georgian Firms industries such as basic metals, production of meat, footwear, and treatment and coating of metals, accounted only for 25 0.55 Density percent of firms but 55 percent of energy value consumption in 2021. So, interventions that target these groups and facili- tate the catch-up of ‘laggards’ to the EE frontier (e.g., through 0.50 technology diffusion and adoption, information exchange, and firm capabilities upgrade) could significantly impact on energy 0.45 Within-sector TFP savings at the macro level. 0.40 The EE level and dispersion within homogeneous groups of firms (that is same sector and size) can guide policy Across priorities for better targeting. Figure 16 displays different 0.35 Sector-by-size EE sectors-size groups along two dimensions: levels of average efficiency and within group dispersion. First, those that are 0.30 more energy efficient and for which we see a large dispersion between firms (top-right quadrant). Such groups, which ac- 0.25 counted for nearly 9 percent of energy consumption in 2021, include general construction of buildings (both SMEs and large firms), SMEs in transport agency activities, large firms in rec- 0.20 reational and leisure activities, and small and medium whole- salers of construction materials and manufacturers of builders’ 0.15 carpentry.37 Second, those characterized by higher-than-aver- age EE and lower dispersion (bottom-right quadrant), which 0.10 accounted for 13 percent of energy consumption, including Within large firms of electricity, gas, steam and hot water supply, large 0.05 Sector-by-size EE wholesalers and retailers -especially of food and beverages, household goods, vehicles and other specialized stores such as clothing and pharmaceutical goods- or large educational estab- 0 lishments. On the other hand, groups with lower-than-average -4 -3 -2 -1 0 1 2 3 4 efficiency but higher-than-average dispersion (top-left quad- rant) include some of the largest energy consumers (together, Energy E ciency they account for 47 percent of energy expenses), such as large and SMEs manufacturers of basic metals, non-metallic miner- Note: Within sector-by-size distributions are relative to sector-by-size averages; across sector-by-size als, food and beverages, and textiles, together with transport distributions are computed as differences between sector-size and economy-wide averages. Source: and storage activities, SMEs constructors, health services and World Bank elaboration based on GEOSTAT. specific primary activities such as SMEs in mining, forestry, fishing, and growing of cereals and vegetables. Finally, the bot- tom-left quadrant includes activities with lower-than-average efficiency and dispersion that together explain one-third of en- ergy consumption. Large construction firms, retail stores, large electricity and gas suppliers, large manufacturers of food and beverages, together with telecommunication companies and freight activities are within this quadrant. Sectors with higher-than-average efficiency dispersion show the highest potential for low-efficiency firms to catch up, hence high returns for interventions. Higher dispersion suggests that it is possible to improve energy con- sumption for a given level of output precisely because the ex- istence of efficient firms indicates that the current technology availability and best practices in their sectors have already 36 This includes general been adopted. Moreover, improving EE for groups with large construction of buildings and energy consumption and low-efficiency levels (i.e., big bubbles civil engineering works and in the top- and bottom-left quadrants) could result in signifi- construction of motorways, cant energy savings and considerable CO2 reductions as they roads, and facilities. use energy in large quantities. However, for firms in the bot- 37 Activity selection based tom-left quadrant, enabling technology diffusion, knowledge on total energy consumption sharing, and a proper business environment that unleashes share. innovation and investment in energy-saving machinery may require a more holistic approach. 38 Chapter 3 Technical Report FIGURE 16 Average Energy Intensity and Dispersion across Sectors Within sector-by-size dispersion (p75-p25) 4.5 3.5 2.5 1.5 0.5 -0.5 -1.5 0 1 2 3 4 5 6 7 Unweighted sector-by-size e ciency Notes: Bubble size quantifies the sector-by-size energy consumption. Source: World Bank elaboration based on GEOSTAT. 39 Chapter 3 Greening Firms in Georgia FIGURE 17 Change in Sectoral Energy Efficiency over Time CHAPTER 3.2 Agg. energy efficiency over time Evolution of Energy Intensity over Time PANEL A Fitted values I Avg. energy e ciency in 2017-2021 n the last four years, two-thirds of sector-size groups e 8 5˚ lin improved their EE compared to 2007-2011, and these 4 are not mainly large consumers. Figure 17(A) plots each 7 sector-size group’s average aggregate EE in 2007-2011 and 2017-2021 of each sector-size combination. Dots over the red 6 line reflect no efficiency changes between the two periods, while points above (below) show a rise (fall) in the aggregate 5 efficiency of each group over time. Across the 226 sector-size combinations observed in both periods, 62 percent increased 4 their aggregate efficiency, and the remaining 38 decreased it. However, sectors that become more efficient are not necessarily 3 large energy consumers compared to those that became less ef- ficient. These sectors account for 63 percent of firms, 65 percent 2 of employment, and 58 percent of aggregate sales, but taken together they consume nearly the same energy (in aggregate 1 terms) as industries that did not improve efficiency. Moreover, 1 2 3 4 5 6 7 8 among the 129 sector-size groups that improved their efficiency levels, big energy consumers, namely sector-size groups that Avg. energy e ciency in 2007-2011 account for at least 1 percent of total energy consumption, represented less than 10 percent of total groups. PANEL B Initial efficiency and change over time Figure 17 also suggests some efficiency convergence across sectors except in those with large employment shares like Fitted values retail. Specifically, we find that sectors that were less energy efficient initially (2007-2011; left of Figure 17A), such as elec- Energy e ciency change tricity, gas and water supply, transport, mining and quarrying 4 and specific construction activities, became more efficient and displayed larger jumps (moved above the 45° line). Instead, 3 sector-size groups that were more efficient in 2007-2011 (at the top-right of Figure 17A) became less efficient over time, lying below the 45° line. The latter mainly include most wholesale 2 and retail activities run by SMEs, such as wholesale activities related to the commerce of grains, seeds, animal feeds, meat, 1 motorcycles, wood, metal structures, and fuels, among many others. 0 Firm-level efficiency dynamics show a similar conver- -1 gence pattern by which energy (in)efficiency tends to per- sist over time, which may lead to slow convergence. While -2 we previously focused on sector-size changes, we concentrate now on analyzing firms. Similarly to Figure 17, Figure 18 plots -3 the average EE of incumbent firms38 over the first and last four 0 1 2 3 4 5 6 7 8 years of the sample (2007-2011 and 2017-2021). The 45° red line distinguishes firms that enhanced their EE (dots above the red Avg. firm energy e ciency 2007-2011 line) from those whose efficiency worsened (dots below the red Note: The bubble size denotes sector-by-size market share. line). For example, if the efficiency level of each firm remained Average EE is the weighted efficiency calculated at the sector-by- constant over time, firms (dots) would lie along (or be close to) size level. Source: World Bank calculations based on GEOSTAT. the 45° red line. In contrast, large efficiency changes over time would push firms further away from the red line. According to the graph, EE persists over time; on average, high-efficiency in- cumbents in 2007-2011 still displayed the highest efficiency lev- 38 Firms active in both els in 2017-2019. However, energy improvements were poor for periods: 2007-2011 and most firms and mainly driven by low-efficiency plants, which 2017-2019. 40 Chapter 3 Technical Report enhanced efficiency, in contrast to top-efficiency firms, which did not improve it. Hence, although firms show persistent EE levels, the largest improvements came from low-efficiency firms. The plot also shows that the number of firms below the red line is larger than that above. In other words, the majority of incumbents (53 percent of total firms in both periods) did not improve EE.39 39 In Section 5 we present a more detailed and exhaustive analysis on energy efficiency convergence. CHAPTER 3.3 A Thought Experiment to Improve the Targeting and Focus of Green Policies S imulating the impact of improving the EE of firms below the median helps quantify the potential gains of optimizing energy consumption among low-efficiency plants. A key question for policymak- ers is how large the potential gains for EE improve- ments are when supporting laggards in achieving the efficiency of the median firm in their sector-size group. This benchmark has the advantage that these are realistic levels known in the sector-size groups. To this end, this section pres- ents a simple thought experiment, an accounting simulation, to provide an order of magnitude of the potential impact of im- FIGURE 18 Energy Efficiency over Time at the Firm Level proving EE on overall energy consumption and other firm vari- ables such as costs -raw materials plus energy expenses- and Fitted values (unweighted) Fitted values (weighted) profits -proxied by revenue net of costs and wages. To estimate Avg. firm energy e ciency in 2017-2021 this impact, we simulate a policy intervention that targets low ne 10 45 ˚ li EE companies and reduces energy consumption keeping the 9 output level unchanged. To this aim, we simulate what would happen if firms whose efficiency level is below the median of 8 their sector-by-size group improve their efficiency and bring it 7 to the median firm in their same sector and size class. 6 The simulation focuses on improving the EE of business- 5 es, referring to firm-level actions that do not necessarily require large-scale investments and that can be adopted 4 without extensive technological knowledge. The Ministry 3 of Environmental Protection and Agriculture of Georgia (2021) identifies EE, switching fuels, and technology for capturing 2 and storing carbon dioxide as key actions to mitigate GHG 1 emissions. However, there are substantial differences in the cost of the actions that policy interventions must consider. For 0 example, firms can replace outdated technology and processes 0 1 2 3 4 5 6 7 8 9 10 with greener machinery and methods, high-carbon intensity energy sources with low-carbon ones, or invest in equipment Avg. firm energy e ciency 2007-2011 to capture and store carbon dioxide. Still, costs and required knowledge of such actions can deter firms from investing. So, it is likely that in the context of financial constraints and low Note: Dots represent incumbents’ average energy efficiency; Source: World Bank calculations incentives for large-scale green projects, businesses focus on bubble size depicts incumbents’ market share. Blue lines are based on GEOSTAT. those not requiring significant capital investments, particularly fitted values both weighted (dashed) and unweighted (solid). small and accessible EE measures. 41 Chapter 3 Greening Firms in Georgia FIGURE 19 Simulation: Energy Efficiency Efficiency measures should be distinguished according to their ability to be identified and adopted. Some measures Distribution before and after are production process-specific, which requires the opportu- Efficiency Improvements nity to be identified in energy audits and a cost-benefit assess- ment to determine its economic viability. Other measures are Actual Simulated relatively general, off-the-shelf, and can be applied to different areas of the firm and have widely-known benefits even without Density prior auditing procedures (i.e., energy-efficient engines like elec- 0.30 tric motors with frequency regulators, efficient heating/cooling systems, on-site green energy generation, energy management 0.25 systems, and LED lighting). The simulation analysis assumes these EE actions are known or can be acquired by firms, given 0.20 firms in same sector and of similar size are already using them.40 By increasing the EE of low-efficiency firms, the simulated 0.15 intervention increases the average EE level and reduces the dispersion within sector-size groups. Figure 19 depicts 0.10 the economy-wide distribution of the observed (blue line) and simulated (red line) EE with their corresponding medians (verti- cal lines). After the simulated intervention, firms become more 0.05 efficient on average because the left-hand side part of the dis- tribution shifts to the right. Mechanically, the shift reduces ef- 0.00 ficiency dispersion because the left tail, populated by relatively inefficient firms, shrinks as the efficiency of lower-than-medi- an efficiency firms converges to the median and that of high- 0 1 2 3 4 5 6 7 8 9 10 er-than-median efficiency firms remains unchanged. Energy e ciency Notes: The figure estimates the energy efficiency distribution in 2021, before and after a hypothetical improvement in energy efficiency. Blue and red dashed vertical lines depict the economy-wide median before and after the simulation for illustration purposes, although the imputation is performed at the sector-by-size A Does Improving Energy level. Source: World Bank calculations based on GEOSTAT. Efficiency Require Large Investments? Improving the EE of laggards can generate large aggregate benefits, so policy interventions should consider firms’ needs but also assess their costs, such as the required in- vestments needed. Improving EE is not necessarily an easy process. Some firms may require significant investments in technology, equipment and other forms of capital, especially 40 As a reference firm we those with outdated assets, while others may need to optimize take the median in order to energy use given their current equipment. One way to under- make this simulation realistic. stand how large potential new capital requirements are is to It is possible to extend compare the capital stock of firms below the EE median of the simulation using as a their sector-size category with those ‘very close’ to the median. reference the more efficient firms such as the top 25 percent in terms of energy While not all, a significant share of firms that have low EE efficiency. levels also display lower levels of capital stock, empha- sizing the need for capital, equipment and technology 41 To avoid misleading upgrading. Firms further away from the median efficiency conclusions from comparing tend to display lower capital levels. This suggests that more the capital endowment of or upgraded machinery and equipment would be necessary inefficient firms with those of the median efficiency firm to close the gap. In other words, for catching up to the medi- (at most two companies), an efficiency, some inefficient firms may require investing in the distance is calculated as modern vintages of capital. However, capital is not the only the difference between the factor determining EE, as efficiency also critically depends on capital of a firm below the organization and green managerial practices. We also find that efficiency median and the returns to capital decrease as the capital gap narrows and Fig- average capital of those firms ure 20 illustrates this point, depicting the relationship between ‘close’ to the median efficiency, that is with an efficiency level the efficiency gap (that is, the difference in efficiency between equal to the median efficiency inefficient firms and the median efficiency) and the capital gap ± 15 percent of a standard (that is, the differences between inefficient firms and those deviation of energy efficiency. closer the median level of efficiency)41, after controlling for 42 Chapter 3 Technical Report location, industry, age of the company42 and time fixed effects. determining EE, in addition to capital upgrading. The graph shows a positive, non-linear relation between the efficiency and capital gaps. Firms with larger efficiency gaps Among firms requiring capital, investment needs can be tend to have larger capital gaps too. However, as the capital large, so interventions need to develop financial instru- 42 Whether the firm is a gap decreases (capital stock increases) the benefit of further ments and align the scheme of incentives so such invest- startup (0-4 years), young narrowing the capital gap (that is, more investment in capital) ments take place. Inefficient firms requiring capital would (5-10) or established (11+) generates less benefit in terms of closing the efficiency gap. In need as much investments as four times their current assets on business. this context, the biggest returns to capital upgrade in terms of average (Figure 21B), highly considerable from the enterprise 43 It should be noted that EE are for those companies that appear to have bigger capital perspective. These large capital gaps may reveal why firms for firms that require capital gaps. In other words, firms further away from the median effi- still underinvest in energy-saving technologies—despite their this is not necessarily the ciency require more capital to close the gap. potential, these investments appear to have a potential positive only constraints as they may impact on their profitability. To put this in perspective, we also need to improve their Nearly 80 percent of inefficient firms would require up- find that the average required capital investment is 24 times organization and management in order to fully reap the grading or augmenting capital to converge to the median the expected profit change. Hence, once the expected change benefits from the upgraded efficiency, while 20 percent do not require capital upgrad- in annual profits is rescaled by the necessary investments, capital and machineries. ing. Considering the positive relation between EE and capital investment returns of energy-saving projects appear to be low. requirements, the simulation allows to estimate the share of This has important implications for policy. First, to succeed, inefficient plants whose capital is below that of the median effi- the scheme of incentives, namely the regulation framework ciency firm within their sector and size. For 80 percent of firms, and energy price signals, should encourage investment-led improving EE would likely require as a necessary condition efficiency improvements. Second, public interventions need replacing outdated machinery, equipment and infrastructure to develop credit instruments that ease credit access for finan- for new vintages of assets (Figure 21A), which goes from im- cially excluded firms and lowers finance costs and increase proving lighting, heating and ventilation systems to investing maturity to compensate the expected low returns. in energy-saving technology and building insulation.43 For the remaining 20 percent of firms, capital does to be needed there- Around three-quarters of inefficient firms require capital fore efficiency improvements depends on improving man- investment for an overall total amount of up to 2 million agement and organization, and increasing awareness about GEL in terms of total capital needs. Capital requirements benefits from improving EE, as well as provide information vary across sector-by-size groups as industries use specific on what solutions are available could be sufficient to improve technologies and have different production methods. However, their efficiency levels. This result points to the importance of all assets value also varies within sector-size groups, which ulti- aspects related to management and organization as key factors mately depends on the production scale, access to finance, and FIGURE 20 Relationship between the Efficiency Gap and the Capital Gap 1.1 Distance to the median e ciency 1.0 0.9 0.8 0.7 6 8 10 12 14 16 18 Log distance to the capital of the median e ciency firm Notes: Binned scatter using age class, two-digit sector of NACE Rev. 1 and region of location as controls (70 bins). Regression outcome is d(eff)itit=0.44+0.0687 d(cap)it—0.0024 d(cap)it2 2+ . Source: World Bank elaboration based on GEOSTAT. 43 Chapter 3 Greening Firms in Georgia FIGURE 21 Capital Requirements to Improve Efficiency? A How PANEL PANEL A How firms many many firmsrequire capital require capital to to to to converge converge the median the firm? median firm? firms require capital to converge to the median firm? PANEL A How many firms firm? Firms Firms require that that capital require capital Firms Firms do not that that not require dorequire capital capital Firms that require capital Firms that do not require capital 8% 80 8 0% 0% 7% 7979% 9% 20 2 119 91% 0 2% % % 9% 0% Improvement Improvement Improvement Improvement Improvement Improvement to the toto median the the median median toto the to the top-25% the top-25% top-25% efficiency efficiency efficiency efficiency efficiency efficiency PANEL B Capital requirements for improving e ciency (Mln. GEL) Capital of ine cient firms 3.62 3.26 requiring investment 0.90 Capital of median e ciency firms Capital of top-25% e ciency firms Source: World Bank calculation based on GEOSTAT. firms’ production function. To quantify the potential demand energy consumption is decomposed into capital-led and organi- for machinery and equipment among inefficient firms, Figure zation-led savings to assess the potential effect of policy support 22 plots the distribution of investment amounts required for to expanding capital versus improving organization to close the upgrading capital. Nearly two-thirds of plants would need less efficiency gap. Figure 23 shows that upgrading firm capabilities than 1 million GEL (~US$ 300,000 in 2021)44, equivalent to dou- and capital are both relevant for reducing energy consumption. bling their current average capital, and three-quarters would Capital-led energy consumption reductions would explain 46 demand funds for less than 2 million GEL (~US$600,000). The percent of total savings, while 54 percent would come from required capital for 11 percent of firms would be at least 6 mil- organization-led reductions (businesses that do not require lion GEL (~US$1,800,000). capital upgrading). This implies that the contribution of the 20 percent of firms that do not need capital upgrading to gain In addition to investment in fixed capital, policy interven- efficiency could account for a larger share of total savings, sug- 44 tions should also focus on upgrading firm organization gesting that there may be larger returns for policies that try to Expressed in and management because returns could be higher among improve information, organization and management, including GEL of 2021. firms that do not lack capital. The reduction in aggregate integrated energy management systems. 44 Chapter 3 Technical Report FIGURE 22 Investment Amounts Required for Upgrading Capital among Inefficient Firms 62.6 13.8 5.0 2.4 2.5 0.8 1.9 3.1 3.9 3.3 Source: World Bank 0.7 calculation based on GEOSTAT. < 1M 1-2 2-3 3-4 4-5 5-6 6-8 8-10 10-15 15-20 20-30 Capital requirement (mln. GEL) FIGURE 23 Aggregate Energy Consumption before and after the Intervention by Type of Firm (mln. GEL) 1,270 Baseline Ine . firms that do not require 527 capital Ine . firms that require capital 439 Efficiency improvement to the median 631 182 146 E cient firms 304 304 Source: World Bank calculation based on GEOSTAT. 45 Chapter 3 Greening Firms in Georgia B The Impact of Improving Energy Efficiency The simulation shows significant energy savings and large from 3.9 percent to 10.9 percent. These results, coupled with increases in profits but much smaller savings in terms of large capital requirements can help us to understand why firms overall costs. Our back-of-the-envelope estimate, which as- face limited incentives to become more energy efficient in the sumes a scenario where all inefficient firms reach level of effi- absence of appropriate nudges or energy policies that strongly ciency of the median firm in their same sector and size group encourage investing in green technology or introducing more (Figure 24), shows that aggregate energy consumption would efficient production and organization processes. halve.45 This is an economically significant reduction entirely driven by firms below the median efficiency threshold. How- Enhancing EE not only helps cut CO2 emissions but also ever, cost reductions (raw materials plus energy bills) would be creates a positive externality that, although firms can- much more modest (middle panel in Figure 24). For inefficient not benefit directly from, can significantly improve air firms, the average cost would reduce by 7 percent as energy quality and health outcomes, bolster trade performance, bills only represent a small fraction of total costs (8.6 percent in and foster the integration of Georgia into EU. EE policies 45 Although not strictly 2021 among inefficient firms and 4.4 percent among all firms). focused on technology adoption and capability upgrading can comparable, a similar exercise Therefore, the average profits of inefficient firms would increase have positive externalities beyond private profit increase and conducted in Poland that looks at total emissions finds that by 114 percent or 0.12 million GEL (right chart). The profit in- cost reduction. In addition to the effects on efficiency and prof- improving emission efficiency crease seems considerable in relative terms, but this is driven itability, they could also lower aggregate carbon emissions by would reduce aggregate by small baseline benefits. In fact, the profit rate, measured as 51 percent, in line with total energy savings (Figure 25). Energy emissions by 46 percent. the average profit-to-cost ratio, would rise by 7 percent points, policies could therefore play a dual role. They could boost firm FIGURE 24 Simulated Impact of Improving Energy Efficiency on Energy Savings, Costs, and Profits (2021, million GEL) Aggregate energy expenses Average costs of ine cient firms Average profits of ine cient firms (Mln. GEL) (Mln. GEL) (Mln. GEL) Aggregate energy expenses Average costs of ine cient firms Average profits of ine cient firms (Mln. GEL) (Mln. GEL) ▼ 7% (Mln. GEL) 1,270 2.97 0.22 ▼ 50% 2.75 ▼ 7% 1,270 2.97 0.22 ▲ 114% ▼ 50% 2.75 631 0.10 ▲ 114% 631 E ciency E ciency E ciency 0.10 improvement to the improvement to improvement to the Baseline median Baseline the median Baseline median ciency E of Note: profits are revenues net E ciency total costs and wages. Source: World Bank calculations based on GEOSTAT and simulations. E ciency improvement to the improvement to improvement to the Baseline median Baseline the median Baseline median FIGURE of Thousands CO2 Emissions before and after the Energy 25CO metric tons Efficiency Improvement 805 Thousands of CO metric tons ▼ 51% 805 392 ▼ 51% Baseline E ciency improvement to the median 392 Baseline E ciency improvement to the median Source: World Bank calculations based on GEOSTAT. 46 Chapter 3 Technical Report performance by improving efficiency (energy is part of total efforts and guide prioritization. For descriptive purposes, costs and profits), but also cut CO2 emissions with the asso- we aggregate carbon emission savings into 52 sectors (2-digit ciated positive externalities. Furthermore, mitigation of CO2 of NACE Rev. 1), illustrated in Figure 26. Slightly more than 75 emissions can positively affect Georgia’s participation in global percent of carbon emission reductions would be concentrated value chains and export growth, not to mention the benefit it in construction, non-metallic minerals, mining of metal ores, can yield for Georgia’s interest in becoming an EU member. water and auxiliary transport activities, manufacturing of food However, firms lack incentives to act since it would only reduce and beverages, wholesale trade and basic metals. The magnitude costs and raise profits marginally and for a significant share of of the contribution ultimately depends on the importance of the firms would require important investments. sector in terms of the energy sources used (‘sources’), the sector output size and the intensity of energy in the production process Almost 75 percent of carbon emissions savings would be (‘scale’ and ‘intensity’) and the size of the tail of inefficiency concentrated in a few sectors, which helps to focus policy firms (‘dispersion’). FIGURE 25 Contribution to CO2 Savings Share of total, percentage Construction 24.8 Mining of metal ores 8.3 Non - metallic minerals 12.6 6.1 Food and beverages Water transport 7.9 4.3 Utilities Basic metals 4.7 Other sectors 6.0 2.4 0.9 1.5 Air transport Agriculture 1.8 4.0 Aux. transport act. 5.8 0.88 Paper 1.1 Retail trade 1.6 Other mining 5.3 business Source: World Bank Wholesale trade Other calculation based on act. Chemicals Land transport GEOSTAT. 47 Chapter 3 Greening Firms in Georgia Improving EE in non-energy-intensive sectors can also policy interventions can take the form of nudges such as low- yield considerable reductions in CO2 emissions. While it cost informational campaigns and feedback provision about is clear that the largest impact on CO2 reductions would come energy consumption (e.g., how to optimize it and a ranking of from high-energy-intensive sectors such as construction or consumption considering similar firms). Also, demonstration transport, other sectors such as wholesale and retail or food and voucher programs to try out EE practices can greatly af- and beverages would explain up to 13 percent of CO2 emissions fect some firms. In other sectors with higher energy intensity cut. This suggests that targeting of priorities sector should not and large capital requirements (e.g., construction, transport, just be based on the level of sectoral energy intensity but also chemicals, metals and non-metal mineral products), the cost consider the number of enterprises and the existing levels of of enhancing more energy-efficient practices may involve shift- inefficiency among less efficient firms. ing toward greener production processes, investing in new, energy-saving technologies, adapting goods and services, ad- Public interventions for enhancing EE and curbing CO2 hering to environmental regulations, and setting a system of emissions may vary according to the sector’s energy in- EE monitoring and targeting. Such changes could entail sig- tensity and the inherent nature of the productive process. nificant investments and demand adequate access to finance For some sectors, mainly those less intensive in energy and and technical knowledge.46 capital (e.g., wholesale and retail trade, hotels and restaurants), C Financial Barriers to Energy Efficiency Firms confirm the importance of easing credit conditions, projects. Second, bank collateral requirements –nearly 220 pointing to financial constraints as one of the highest bar- percent of the loan’s value– add further difficulties for Geor- riers impeding EE investments. A recent study, relying on a gian companies to take loans, especially SMEs. Commercial small survey with few private SMEs conducted in 2022 (MoESD banks usually do not reach the threshold of uncollateralized & UNESCAP, 2022), found that the main barriers to EE faced by loans stipulated by law (25 percent of total portfolio), suggest- firms are high investment costs, capital expenditure with long ing that they may perceive greater risks than the regulations payback periods (more than three years), low access to ‘green’ require (EIB, 2016). Moreover, attractive short-term lending finance and limited internal resources hinder investments in opportunities in Georgia, such as retail banking (rather than EE projects. Given the estimated capital requirements and the corporate banking), often acerbate a shortage of long-term cap- limited availability of internal funds, easing credit access is ital that could be mobilized to finance climate action (OECD, crucial for unleashing green investment in Georgia. 2018). There is a need for better internal systems for green fi- nance, including ESG. The issue of collateralization, especially Improving credit access requires increasing available concerning more extensive budget financing, loan sizing and amounts and improving conditions simultaneously. Ac- segmentation, was brought up by some surveyed FIs, especially cess to finance is a top priority to encourage firms to invest in as they relate to the cost of borrowing. Cost pressures for mi- energy-saving technologies, digitalization and building in- cro-mobility and personal EE, however, borrowing costs start at sulation. However, according to the IFC (2021), the supply of two times the NBG rate; construction market segmentation and credit instruments does not match the business demand for viewing real estate through the larger finance value chain. Most several reasons. First, firms require loan amounts below the FIs, other than those with strong agricultural loan portfolios, minimum requested amount offered by commercial banks. were not particularly interested in agri-finance. In addition, Although commercial banks offer some low-interest-rate credit the studies suggest that current regulations requiring all loans instruments, the minimum requested amount -GEL 100,000 smaller than 200,000 GEL to be issued in national currency or nearly US$ 38,600– is above the amount that SMEs typically create a constraint for smaller institutions due to the high costs SMEs need for green projects -US$ 9,000 to 40,000- for green of foreign currency risk hedging. D Strategic Complements for Credit Access Technology diffusion, bureaucracy simplification and ments uptake (i.e., art sensors and meters) to monitor energy regulations are strategic complements for credit access. consumption. Most firms claim they cannot separately meter In addition to financial barriers, low awareness about green energy consumed for lighting, heating, cooling, and venti- technologies also inhibits energy-efficient technology adoption lation. Also, they are unaware of the latest green technology (MoESD & UNESCAP 2022). Information and the availability for their production process and cannot quantify its benefits. and diffusion of green technologies are key enablers for (and Information barriers such as low awareness about energy-effi- complements of) EE investments. Even if policy actions toward cient and renewable-energy technologies (i.e., benefits, costs, easing access to green credit instruments succeed, businesses financing) emerge as relevant factors that hinder EE among the need to be aware of technological advances in their industry surveyed businesses. Finally, firms also indicate that, on the 46 See the last chapter and detect specific areas for improving energy management. regulatory side, the low number of energy audits undertaken for more discussion The case studies show that firms acknowledge internal limita- and the absence of certified energy auditors limit a more effi- about specific policy tions to adopting sophisticated technology or simply improving cient energy use. recommendations. energy monitoring systems. For example, there is low instru- 48 Chapter 4 Technical Report Chapter 4 48—4 PAGE Technology Adoption, Innovation and Energy Efficiency A key channel for improving EE and reducing emis- economic activity on the environment at a lower cost. There are sions is the adoption green technologies and several environment-related areas where innovations can arise. practices. Technology can help make different stag- Firms, universities and public organizations can develop and es of production more energy efficient. Starting with patent green technologies focused on energy substitution and inputs, resources and energy, firms can reduce the car- efficiency (e.g., biofuels, fuels from waste, technology for im- bon footprint by investing in the generation of renew- proved input/output efficiency), cross-cutting enabling technol- ables on-site—wind, solar or geothermal—non-fuel ogies (e.g., technologies related to transport, buildings, energy generators or energy storage technologies. Information tech- sector, chemicals & oil), developing low-carbon energy supply nologies can help account for the inputs’ carbon footprint (e.g., hydro, geothermal, solar, wind) and those that mitigate or and search for new, less carbon-intensive suppliers. Technol- adapt to climate change (e.g., solid waste recycling, information ogies can contribute to reducing power consumption in the and communications technology (ICT)). The dynamism of the production process by using—for example—energy-efficient innovation ecosystem and the number of sustainable technol- motors and transforming premises that use Lighting-Emiting ogy patents is likely to reflect the interaction and coordination Diode (LED) lighting; heating, ventilation, and air conditioning between public and private actors, the resources devoted to (HVAC) systems; Internet of Things (IoT) systems; or intelligent developing more environment-friendly technologies and the thermostats in smart buildings. Upgrading carbon-fueled to influence of government policy to guide innovation to more electrical technologies and adopting carbon capture or seques- ecologically favorable ends (Haščič & Migotto, 2015). tration technologies also cut CO2 emissions during production. This section assesses technology adoption and green pat- Beyond technologies directly affecting green outcomes, enting among Georgian firms. It uses new data collected in other general47 technologies, especially information-re- Georgia, the Firm-Level Adoption of Technology survey (FAT), lated ones, can also help support EE and greening of busi- merged with GEOSTAT administrative data to (i) measure the nesses. “General” technologies are those that do not have as extent of adoption of green technologies; and, (ii) the associa- a direct objective reducing energy consumption or the level tion between technology adoption and EE and leverages the of GHG emissions. For example, improving the information Patent Statistical Database (PATSTAT) data of global interna- management of suppliers, delivery and production processes tional patent families in sustainable technologies to examine is critical for optimizing energy use even though it may not be the dynamics of green innovation. The section is organized as their primary purpose. Thus, we also should expect that the follows. First, the section presents a summary of the data used. 47 We define “general” use of these technologies also affects EE. Second, it focuses on the efficiency level of energy inputs used. technology as a technology Third, it describes the adoption of green technologies in Georgia that does not have as a direct objective of reducing energy Also, innovation in sustainable technologies plays a key and measures the relationship between technology adoption consumption or the level of role in EE and environmental-friendly production prac- and EE. Fourth, it looks at innovation and local patent dynam- GHG emissions. tices. Green innovations can reduce the negative impact of ics. The last part of the section presents the main conclusions. 49 Chapter 4 Greening Firms in Georgia T he FAT is a nationally representative online survey group relates to general business functions (GBFs), mainly im- CHAPTER 4.1 to 1,500 firms, with additional 300 firms surveyed plemented in management-, administration- and sales-related The Data by telephone in 2021. The survey includes firms in tasks (see Box 9). The second group relates to sector-specific agriculture, manufacturing and services and is nation- business functions (SSBFs), business functions that are applied ally representative. The survey measures the adoption to production processes, and thus are sector-specific. Firms and use of over 300 technologies, ranging from manual are also asked about adopting other technologies and prac- processes to frontier technologies. These technolo- tices that can reduce energy consumption or emissions, such gies are structured into two groups of technologies based on as energy-efficient lighting, Energy Star equipment, and ISO the business function where they perform the task. The first 14000s certifications. BOX 9 Measuring technology adoption The World Bank has been measuring the level of technol- tion and productive use of technologies for both general and ogy adoption by firms in over 15 countries and 18,000 sector-specific business functions. For each business func- establishments through the Firm-Level Adoption of Tech- tion, the survey goes from the most basic technology to the nology (FAT) Survey. This survey captures the firm’s adop- most advanced one (Figure 27). 50 Chapter 4 Technical Report BOX 9 Measuring technology adoption (cont.) FIGURE 27 Mapping of Technologies in General Business Functions Administration (HR Handwritten processes Computers with standard Mobile apps or digital Computers with Enterprise resource planning 1 processes, Finance, software (e.g., Excel) platforms specialized installed (ERP) or equivalent software Accounting) software integrated with other back o ce functions Production or service Handwritten processes Computers with standard Mobile apps or digital Specialized software for Enterprise resource planning 2 operations planning software platforms demand planning, demand (ERP) or equivalent software forecast integrated with other back o ce functions Sourcing and Manual search of Computers with standard Online social media, Supplier relation Supplier relation 3 Procurement suppliers, without software specialized apps or digital management (SRM) not management (SRM) centralized database. platforms integrated with production integrated with production planning planning Marketing/ Costumer Informal chat Online chat (e.g., Structured costumer Costumer relationship Big data analytics / Artificial 4 information (face-to-face) WhatsApp or Internet) surveys management (CRM) intelligence software Fabrication technology Manual Machines controlled by Machines Robots Additive Other advanced manufacturing 5 processes (e.g. laser, plasma and automation processes operators controlled by manufacturing computers including rapid sputtering, high speed machine, prototyping E-beam, micromachining) and 3D printers Sales Direct sales Direct sales by Sales through Online sales using Online sales Electronic orders 6 at the phone or e-mail social media external digital (e-commerce) using its integrated to specialized establish- platforms or platforms (e.g., Amazon, own website supply chain ment apps eBay, Alibaba) management systems Payment methods Cash Check, Prepaid, credit Online or electronic Online through Virtual or cryptocurrency 7 payment voucher or or debit card platform bank wire Quality control Manual, visual or written processes without Manual, visual or written Statistical process Automated systems for 8 the support of digital technologies processes with the support control inspection of digital technologies Source: World Bank elaboration 51 Chapter 4 Greening Firms in Georgia We analyze the relationship between technology, green and general, and EE by matching FAT and GEOSTAT TABLE 5 Matching Results firm-level records and exploiting energy consumption data. Given the probabilistic sample nature of the GEOSTAT Category FAT Matched Share in small firm strata, we match 1,151 out of 1,793 survey respondents sample sample matched (Table 5). Medium and large firms in the final matched data sample set are likely to be overrepresented due to the probabilistic (percent) sampling of small firms, but the resulting sample is nationally representative.48 A. Size class Small 734 307 26.7 Medium 632 460 40.0 Large 427 384 33.3 B. Sector Agriculture 194 105 9.1 Manufacturing 583 405 35.2 Services 1,016 641 55.7 Total 1,793 1,151 48 Appendix B performs Source: World Bank elaboration based on FAT surveys and GEOSTAT. some statistical tests to gauge the possibility of bias in using the matched sample compared to the original nationally representative FAT sample. Table 32B exhibits some differences between the matched and original sample. However, Table 33B also shows that after controlling for size and sector, there are no statistical differences between the two samples regarding technology adoption, management quality, or adoption of green technologies. T he type of energy used is key for understanding electricity access. Nearly all firms (96 percent) are affected by CHAPTER 4.2 the carbon footprint of firms, and in Georgia only power outages, which occur twice per month, on average. The Green a small share of firms produce renewable energy share of firms experiencing power outages is comparable to on-site, and many firms rely on generators that India and African countries such as Bangladesh, Kenya, Ma- use fossil fuels. While all firms in the survey have an lawi and Ghana (Figure 28). To deal with power outages, 30 Energy electricity connection, a large fraction of firms also rely on fossil fuels. Table 6 reports the quality of electricity percent of Georgian firms—50 percent among large firms— own electricity generators, a rational response to production Sources access and the firm’s energy sources. Regarding energy sourc- problems arising from low-quality electricity supply. Data show es, 46 percent of firms use fuel oil, and 56 percent use natural a positive correlation between the number of power outag- gas. Producing renewables on-site can significantly reduce es and the probability of owning a generator. Put differently, their carbon footprint, but the uptake is limited to less than 1 firms that experience more problems with electricity access are percent of firms. Among firms that generate renewable energy more likely to invest in energy generators to minimize energy on-site, renewable-generated electricity represents 7 percent bottlenecks that can stop production (Figure 28). However, 49 In the US, the U.S. of total electricity (more than 20 percent among large firms). generators tend to be more carbon-intensive than grid elec- Energy Information tricity,49 which create additional environmental problems. But Administration using EPA Firms respond to low-quality electricity access by own- the correlation between the number of monthly power outages national datasets for 2020 ing fossil-fuel energy generators instead of investing in and producing renewables on-site is negative. Hence, while estimates that a typical diesel generator emits 2.15 on-site renewables, confirming the weak incentives to firms respond to energy grid reliability problems by investing times more CO2 than the US adopt cleaner energy technologies described in Section in energy generators, such problems do not incentivize on-site electricity grid. 1.2. Georgian firms experience problems with the quality of renewable energy production. 52 Chapter 4 Technical Report TABLE 6 Quality of Electricity and Energy Sources All Small Medium Large Quality of access to electricity percent firms with power outages 96.4 95.3 98.0 99.4 Avg. power outages per month 1.9 1.9 1.7 2.2 percent firms with generator 30.4 25.1 37.5 54.7 Energy sources percent of firms using fuel oil 46.0 41.0 53.1 50.6 percent of firms using natural gas 56.0 52.0 54.9 78.8 percent of firms using LPG 5.4 1.6 10.3 10.2 percent of firms using coal or wood fuels 1.4 0.7 1.5 4.3 percent of firms that produce renewable 0.4 0.4 0.3 0.6 energy on-site percent of electricity generated by renew- 7.0 4.9 10.3 21.4 ables (conditional on producing renewables) Source: World Bank elaboration based on FAT surveys and GEOSTAT. FIGURE 28 Share of Firms Facing Power Outages by Country India Bangladesh Kenya Malawi Georgia Ghana Ethiopia BurkinaFaso Senegal Vietnam Brazil Poland Chile Korea 0 20 40 60 80 100 Share of firms that faced power outages Source: World Bank elaboration based on FAT surveys. 53 Chapter 4 Greening Firms in Georgia FIGURE 29 Outages, Use of Generators, and Production of Renewable Energy Probablity of owning a generator Probablity of producing renewable energy 50 50 40 40 30 30 20 20 10 10 0 0 0 5 10 15 0 5 10 15 Avg. number of power outages in a month Avg. number of power outages in a month KER N EL = EPAN ECHN I KOV K ERNEL = EPA NEC H NIKOV D EG R E E = 0 DEGREE = 0 BAN DWI DTH = 3 BA NDW IDTH = 3 PWI DTH = 1. 42 PW IDTH = 1.42 Note: the figure shows a local polynomial smoothing between the average number of outages per month and the probability of a firm owning a generator (left) and the likelihood of producing renewable energy (right). Pooled data from five countries with available information: Brazil, Cambodia, Chile, Ethiopia, and Georgia. Source: World Bank elaboration based on FAT surveys. T he adoption of technologies in production pro- green management practices and EE technologies (Figure 30A), CHAPTER 4.3 cesses is a key area for improving EE and reducing almost half of the firms have not adopted any green sustain- Green carbon emissions. First, firms can adopt a set of green able practice or EE technology, and 45 percent have adopted technologies and practices with the aim of directly between one and three of the eight mapped practices. The level boosting EE. For example, using “smart” buildings with of sophistication of the adopted green technologies is also low. and General LEED certification, LED lighting and other features, or Energy Star-rated machinery can lead to significant The green technologies with higher adoption are efficient light- ing (LED/CFL lamps instead of incandescent) and VAV (variable Technology energy savings. Second, firms can also adopt technologies to air volume) HVAC systems. Less than 20 percent of the firms manage production and information more efficiently, for ex- have ISO 14000s certifications, are LEED certified, use Energy ample, by optimizing the level of energy used for any given Star equipment, have programmable thermostats, or monitor Adoption output level. For example, technologies like enterprise resource planning (ERP) are key in implementing green and sustainable plans since they allow leaner production, waste and overpro- CO2 emissions. Low green technology take-up suggests there is room for introducing sustainable practices and technologies that do not necessarily require significant investments but duction reduction, and better coordination. Jointly with oth- can foster EE. er technologies, such as supplier relation management (SRM), they allow more streamlined logistics and route optimization, General technologies, like information management and improving resource efficiency and reducing transportation and production planning, are essential for implementing sus- distribution emissions. Table 7 briefly describes these off-the- tainability plans and increasing EE. However, a key ques- shelf technologies and processes and how they reduce energy tion is why, if the adoption of these general technologies is very consumption and GHG emissions. heterogeneous, as described in Cirera et al. (2022), adopting “green” technologies should be more uniform. Table 8 maps “Green” technologies and practices adoption is still very general technologies with each GBF ordered from the least incipient in Georgia, showing high potential for boost- advanced to the most advanced. Cirera et al. (2022) show that ing energy and overall efficiency by adopting these off- most firms do not always make productive use of such tech- the-shelf technologies and without requiring large-scale nologies either by not adopting them or when adopting them, capital investments. Starting with the number of sustainable by not using them intensively. 54 Chapter 4 Technical Report TABLE 7 Green Technologies: What Are They and How Do They Work? Technology Description How do they reduce energy consumption and GHG emissions? LEED certification Rating system for green buildings and sustainable design that It can improve energy efficiency by optimizing evaluates buildings based on energy efficiency, water conservation, natural light, using renewable energy, and and sustainable materials. optimizing the premises’ performance to control temperature. Energy Star Program that identifies energy-efficient products such as Based on the program results, managers can appliances, lighting, and electronics and provides technical detect opportunities to optimize energy use. assistance and benchmarking. Energy-efficient Efficient lighting systems like LED or Compact fluorescent lamps As it converts a higher percentage of electrical lighting (CFLs) use less energy to produce the same amount of light as energy into visible light, energy-efficient lighting traditional lighting systems, reducing firms’ costs. minimizes wasted energy in the form of heat and consumes less energy. Variable Air Volume Heating, ventilation, and air-conditioning VAV/HVAC systems use By adjusting the airflow based on specific zone (VAV) HVAC systems dampers and sensors to adjust the amount of air delivered to each demands, these systems optimize energy and zone within a building. allow for energy savings compared to constant air volume systems. Programmable Heating and cooling control devices that can adjust HVAC systems Setting temperature setbacks during off hours or thermostats to have a pre-set schedule based on desired comfort levels and periods of lower occupancy conserves energy and occupancy patterns. thus improves energy efficiency. IoT-enabled Advanced HVAC systems that incorporate IoT technology (e.g., IoT-enabled temperature systems use advanced temperature systems sensors) to provide enhanced control, monitoring, and automation algorithms and data analytics to optimize energy of temperature and humidity settings in commercial and industrial usage based on occupancy patterns, weather buildings. conditions, and other relevant variables. ISO 14000s Family of certifications designed to help organizations manage their These certifications demonstrate the firm’s certifications environmental impact and performance. They cover multiple topics, commitment to environmental management and including the analysis and assessment of energy consumption and sustainability and provide a framework to address GHG emissions, and set targets to improve the firm’s environmental environmental aspects systematically. performance. Monitoring CO2 Good environmental practice that lets firms identify their impact on By monitoring emissions, managers can better emissions carbon dioxide (and GHG) emissions. understand the fluctuations and source of emissions and undertake initiatives to reduce the enterprise’s carbon footprint. Source: World Bank elaboration 55 Chapter 4 Greening Firms in Georgia FIGURE 30 Green Technology Adoption TABLE 8 Mapped Technologies into Each General Busi- PANEL A Number of green technologies ness Function and practices adopted General Mapped general technology O 1 2 3 4 5 6 7 8 business function 47 Business Administration Handwritten processes, computers with standard software (such as Excel), mobile apps or digital platforms, computers with specialized software, enterprise resource planning (ERP) Production/ Handwritten processes, computers Service with standard software (such 19 16 Planning as Excel), mobile apps or digital platforms, computers with specialized software, enterprise 10 3 3 1 0 0 Sourcing and resource planning (ERP) Manual search of suppliers without Procurement centralized database, computers Number of green technologies (out of 8) with standard software (such as Excel), online social media or specialized apps, SRM not integrated with production planning, PANEL B Radar of adopted green technologies SRM integrated with production and practices planning Marketing/ Informal chat, online chat, Customer structured customer surveys, information customer relationship management (CRM), big data analytics or AI LEED certified Sales Direct sales at the establishment, M channels direct sales by phone or email, sales on ito through social media, online sales rC using external digital platforms, r O a St Em online sales using its own website, gy electronic orders integrated er iss En ion into specialized supply chain s management Payment Exchange of goods, cash, check ISO 14000 methods or bank wire, credit or debit 10 0 80 60 40 20 cards, online electronic payment, Energy e cient lighting online through platform, virtual or cryptocurrencies m Quality Manual, visual or written processes, te Programmable thermostats sys control manual, visual or written processes VA p. tem supported by digital technologies, V HV led statistical process control, AC ab automated system for inspection en sy IoT st em s Source: World Bank elaboration Source: World Bank elaboration based on FAT surveys. 56 Chapter 4 Technical Report The sophistication of general business technologies in adoption of top technologies in all GBFs, such as ERPs, SRMs, Georgia is low suggesting there are considerable oppor- or CRMs, is very low. Overall, this represents a 2.8 average score tunities for improvement. Georgia’s Technology Adoption (out of a maximum of 5) in the technology adoption index in the Survey results show similar patterns than in other countries. extensive margin and a 2.1 in the intensive margin. These re- First, while there is more sophistication in the technologies sults suggest significant barriers to adoption in Georgia, which used for business administration, production planning (i.e., are likely to also affect the adoption of green technologies. specialized software or ERP systems), and payment methods (i.e., digital payments), the sophistication of other GBFs is very The adoption of general and “green” technologies is pos- low (Figure 31A). For example, only 15 percent of the firms have itively correlated, suggesting the existence of common online sales channels, and 25 percent use statistical processes barriers and drivers and potential complementarities or automated systems for quality control. Furthermore, the that policy design should consider. As expected, there is a FIGURE 31 Technology Adoption in Georgian Firms PANEL A Adoption of advanced PANEL B Adoption of advanced technologies by GBF technologies by GBF Advanced Technology Top Technology Tech adoption index GBF Density of Firms (intensive) Business administration 75.0 0.5 15.7 Tech adoption index Payment methods GBF (extensive) 70.6 0.4 6.2 Production or service operation 55.5 0.3 13.3 Sourcing and procurement 36.7 0.2 10.6 Marketing and product development 36.3 0.1 7.6 Quality control 25.3 0.0 7.0 1 2 3 4 5 Sales channels Index 15.1 13.8 Fabrication (manufacturing) 4.6 4.6 0 20 40 60 80 100 Share of Firms Notes: Advanced technologies by GBF: Business administration: Specialized software, ERP (top); Payment methods: debit/credit card, e-payments through bank wire or platform (top); Production or service operation: specialized software, ERP (top); Sourcing and procurement: social media or specialized apps/platforms, SRM (top); Marketing and product development: structured surveys, CRM (top), big data analytics (top); Quality control: statistical process with monitoring software, automated systems (top); Sales channels: social media, external platforms (top), own website (top), e-orders (top); Fabrication: robots, additive manufacturing, advanced manufacturing. Source: World Bank elaboration based on FAT surveys. 57 Chapter 4 Greening Firms in Georgia strongly positive correlation between the adoption of general Technology adoption needs also to consider demand fac- and “green” technologies. Figure 32 shows the relationship be- tors as well as the capabilities and preparedness of firms tween the GBF and SSBF adoption indexes and the adoption of to adopt and productively use technologies. Table 9 shows green technologies. While firms with low technology adoption the results of linearly regressing the number of adopted “green” levels tend to adopt less than one green technology, those with technologies and practices on the GBF technology adoption, higher adoption levels tend to adopt between 1 and 2 green managerial quality and firm human capital indexes plus ad- technologies or practices. This positive correlation suggests ditional firm characteristics. The results highlight that even that some barriers and drivers between the types of technol- after controlling for size, sector, ownership characteristics, ogies are common and that there could be potential comple- and skills, firms that adopt more general technologies also mentarities between both types of technologies. For example, adopt more “green” technologies. This suggests the importance barriers to adoption like lack of finance, information or skills of “supply” side factors such as stronger firm capabilities (as may limit the uptake of both types of technologies. Similarly, proxied by the GBF tech adoption index) in driving demand for incentives due to increased competition can incentivize the green tech and green managerial practices. In addition, Table adoption of both sets of technologies to increase competitive- 9 also highlights the importance of understanding demand ness. To deliver cost-effective interventions, policy actions factors driving the decision of adopting “green” technologies. need to consider how barriers and drivers to technology uptake For example, exporting firms tend to adopt more green tech- work, the strategic complementarities between technologies nologies, which could mean demand from exports markets and how regulation and price incentives operate in undermin- and participation in global value chains (GVCs) appear to be ing or fostering technology adoption. another very important factor in providing incentives for firms to adopt green technologies and managerial practices. FIGURE 32 The Correlation between Green and General Technology Adoption PANEL A GBF tech. adoption PANEL B Sector tech. adop- and green tech. adoption tion and green tech. adoption 2.5 Predicted number of green tech. & practices 50 Predicted number of green tech. & practices 2 40 1.5 30 1 20 0.5 10 0 0 1 2 3 4 1 2 3 4 GBF technology index – Extensive Sector specific tech adoption index – Extensive Source: World Bank Sector specific tech adoption index – Extensive elaboration based on FAT LOCAL POLYNOMIAL SMOOTHING ESTIMATION surveys LOCAL POLYNOMIAL SMOOTHING ESTIMATION 58 Chapter 4 Technical Report TABLE 9 Explaining the Adoption of Green Technologies # of green tech. & practices (1) (2) Tech. adoption index GBF (extensive) 0.253*** 0.229** (0.093) (0.096) Exporter 0.278** 0.280** (0.131) (0.130) Managerial quality index 0.103 (0.064) Firm human capital index –0.044 (0.065) Size and log assets controls Yes Yes Sector; region & ownership FE Yes Yes Observations 849 849 Notes: Robust standard errors in parentheses. (*), (**) and Source: World Bank calculations based (***) indicate significance at (10), (5) and (1) percent level. on FAT surveys. CHAPTER 4.4 What is the Role of Technology in Explaining Energy Efficiency? G reen technologies and practices are likely to of magnitudes, general technology adoption might explain a boost EE by saving energy in tasks related to larger share of within-sector EE variation. For example, so- the production process, for example, by main- phistication, measured through the GBF technology adoption taining optimal temperatures in food process- index, explains around 16 percent of the gap in EE between the ing firms, or by monitoring energy consump- 10th percentile and the 90th percentile. A similar measure for tion in high-energy-intensive industries. Given the number of green technologies is associated with closing the the strong relationship between the adoption of gap by around 7 percent. general and green technology, we explore the role of technology adoption in improving EE. Based on GBF technologies can have a bigger impact on EE than the earlier results, both types of technologies are likely adoption of some narrow green technology. There are vari- to impact EE but through different channels. On the one hand, ous reasons behind the importance of GBF technologies for EE. green technologies and practices can directly raise EE by saving First, sophistication in these general technologies could also be energy in activities complementary to production. Because capturing the effects of other technologies and more efficient most of the green technologies measured in the survey do not processes. Second, since green technologies are, by nature, affect the production process directly -except thermostats- more relevant only in certain industries and relate to using energy savings are related mainly to more efficient premises, smart building technologies primarily, it is plausible that gen- namely “smart buildings.” On the other hand, ICT technologies eral GBFs may be displaying a more prominent role. To better reduce energy consumption indirectly. Specifically, they work understand the heterogeneous impact, Table 11 explores the re- through improved information management that makes pro- lationship between individual technologies and EE within sec- duction more efficient and the implementation of sustainability tors. The results show that firms that, on average, adopt or have plans. Table 10 explores the relationship between EE, calculated ERP systems, online sales or automated inspection systems for as the logarithm of the ratio between annual sales and energy quality control tend to be systematically more energy efficient. consumption, and the general technology use. Columns 1 and Remarkably, quality control technologies explain 18 percent of 2 show that the adoption of technologies in GBFs is positive- the within-sector EE gap between (10th-90th percentile gap). So ly correlated with improvements in EE, and columns 3 and 4 GBF technologies that reduce energy consumption or improve show the same relationship for green technologies. In terms the selling process efficiency potentially enhance EE. 59 Chapter 4 Greening Firms in Georgia TABLE 10 Energy Efficiency and Technology Adoption Energy efficiency (De-meaned by sector) (1) (2) (3) (4) Tech. adoption index GBF (extensive) 0.170** 0.158** (0.070) (0.077) Number of green technologies & practices 0.056* 0.053* (0.031) (0.029) Managerial quality index 0.079 0.087* (0.051) (0.052) Firm human capital index -0.039 -0.017 (0.049) (0.047) Size & log assets controls Yes Yes Yes Yes Sector, region, ownership & exporting status FE Yes Yes Yes Yes Observations 885 885 885 885 R-squared 0.116 0.129 0.112 0.127 Share of 90-10 explained by tech. adoption | green tech. & 17.1 15.9 7.5 7.1 practices (percent) Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance at (10), (5) and (1) percent Source: World Bank calculations based level. on FAT surveys. 60 Chapter 4 Technical Report TABLE 11 Energy Efficiency and Specific GBF Technologies Energy efficiency (De-meaned by sector) (1) (2) (3) (4) (5) (6) (7) ERP 0.155* (0.090) SRM 0.134 (0.093) CRM or customer-data AI 0.235 (0.224) Advanced tech. Fabrication 0.134 (0.114) Has online sales/orders channels 0.219* (0.119) Accepts online/card payments -0.010 (0.135) Automated inspection systems 0.405*** for quality control (0.139) Size & log assets controls Yes Yes Yes Yes Yes Yes Yes Sector, region, ownership & exporting Yes Yes Yes Yes Yes Yes Yes status FE Observations 885 851 867 178 858 883 641 R-squared 0.105 0.097 0.099 0.240 0.106 0.099 0.148 Share of 90-10 explained by adopting 6.9 - - - 9.7 - 18.0 each technology (percent) Notes: Robust standard errors in parentheses. (*), (**) and (***) Source: World Bank calculations based indicate significance at (10), (5) and (1) percent level. on FAT surveys. Because the effects of ERP, online sales and automated for quality control technologies. These technologies usually quality control systems technologies on EE vary with affect the production efficiency of the firm and can reduce the firm size, policy programs should use a tailor-made energy consumption by avoiding reprocesses and waste. For approach based on specific firms’ characteristics. Figure smaller firms, the returns of these technologies are low but in- 33 sheds some light on the potential mechanisms underlying crease with the production scale and are considerable for large how particular technologies can improve EE. On the one hand, companies. Last, ERP systems let managers allocate resourc- online sales channels can improve firms’ EE through the sales es efficiently and have similar returns across size categories, channel. Since the returns of online sales channels decrease meaning they can use them to optimize resources irrespective with firm growth, this family of technologies could yield a of the firm size. greater impact among small enterprises. The opposite is true 61 Chapter 4 Greening Firms in Georgia FIGURE 33 Change in Energy Efficiency and GBF Technologies by Firm Size E ects on mean function of adopting each technology 0.2 Online Sales Channels Adv. tech. for quality control 0.15 0.1 ERP 0.05 0.0 Source: World Bank 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 elaboration based on FAT surveys. Log fixed assets Among green technologies, automated thermostats and correlates with EE, explaining around 11 percent of the EE gap. ISO 14000 certifications show the highest potential for in- In contrast, according to the data, there is no statistical cor- creasing EE. Table 12 shows the correlation between individual relation between lighting or energy-efficient rated equipment green technologies and within-sector EE. Automated thermo- and EE, in the latter case likely because, in many sectors, these stats have the highest (positive) correlation with EE, explaining ratings do not necessarily apply to production equipment. The the 18 percent of the EE gap (the 10th-90th difference) within most surprising result, however, is the negative correlation be- sectors. This result is consistent with some industry claims that tween monitoring emissions and EE. A potential explanation electronic and, especially, smart thermostats can have large EE can be that only highly energy-intensive and low-energy-ef- gains. In addition, being ISO 14000 certified -having a set of ficient firms within each sector monitor emissions to comply environmental processes in place like environmental manage- with energy audits and avoid penalties. ment systems or ecological performance evaluation- positively 62 Chapter 4 Technical Report TABLE 12 Energy Efficiency and Green Technologies and Practices Energy efficiency (De-meaned by sector) (1) (2) (3) (4) (5) Energy eff. tech. (LEED, Energy Star, lighting) 0.081 (0.109) VAV HVAC systems -0.043 (0.106) Automated thermostat (programmable or IoT) 0.437*** (0.103) Has ISO 14000 0.254* (0.142) Monitors emissions -0.293* (0.170) Size & log assets controls Yes Yes Yes Yes Yes Sector, region, ownership & exporting status FE Yes Yes Yes Yes Yes Observations 885 885 885 885 885 R-squared 0.101 0.100 0.155 0.112 0.117 Share of 90-10 explained by adopting each - - 19.4 11.3 -13.0 technology (percent) percent percent percent Notes: Robust standard errors in parentheses. (*), (**) and (***) Source: World Bank calculations indicate significance at (10), (5) and (1) percent level. based on FAT surveys. There is less clear evidence on the relationship between nologies is correlated with EE by allowing the optimization of sector-specific technologies and EE. Sector-specific tech- routes and fleets. For all the remaining selected technologies, nologies like automated irrigation systems, machines with the correlation is not statistically significant. On the one hand, automated processes, or pricing and inventory systems aim it is worth mentioning that the small sample size limits robust to make processes more efficient and reduce waste. Still, they inference and could therefore drive (some of ) these findings. are powered by electricity and, to some extent, replace man- However, outcomes could also indicate no efficiency gains ual labor, potentially increasing energy consumption. Table from adopting these technologies and, sometimes, that they 13 explores the correlation between selected sector-specific increase energy consumption because the alternative action technologies and EE. The only significant result is for the is to use manual processes. transportation sector, where monitoring performance tech- 63 Chapter 4 Greening Firms in Georgia TABLE 13 Energy Efficiency and Green Technologies and Practices Energy efficiency (De-meaned by sector) (1) (2) (3) (4) (5) Measure performance based on real-time data and 0.475* integration with other software (Transportation) (0.231) Monitors load and delivery with software and -0.042 integration with other software in firm (Transp.) (0.447) Power equipment with minimal human interaction 0.075 for mixing/cooking (Food preparation) (0.090) Fully automated climate and security-controlled -0.010 storage (Food preparation) (0.117) Has Warehouse Management System with 0.056 specialized software (Retail) (0.092) Size & log assets controls Yes Yes Yes Yes Yes Sector; region; ownership & exporting status FE Yes Yes Yes Yes Yes Observations 34 34 126 121 128 R-squared 0.485 0.397 0.263 0.297 0.367 Notes: Robust standard errors in parentheses. (*), (**) and (***) Source: World Bank calculations indicate significance at (10), (5) and (1) percent level. based on FAT surveys. BOX 10 The innovation ecosystem in Georgia Patents are a measure of the innovation ecosystem dy- The evolution of green patents over recent years in Geor- namism and the potential for the generation of green gia show a lack of innovation growth in sustainable technol- solutions. Patents protect and incentivize innovations by ogies. Over the last years, the number of new green patents granting exclusive rights to inventors. These protections has not shown signals of improvement despite the sustained encourage investment in R&D, promote competition, and growth worldwide (Figure 34). Even though the number of new facilitate technology transfer. Green patent analysis pro- green patents increased worldwide, especially between 2002 vides valuable information on the Georgian innovation eco- and 2018, in Georgia, it has fluctuated between 0 and 2 since system related to the green economy. Based on the data- 2010. Green patent applications are mostly concentrated in base compiled from PATSTAT of global international patent enabling technologies, EE and the energy transition (low-car- families in sustainable technologies for 2000-2019, we an- bon energy supply), like applications in comparable countries. alyze the innovation in sustainable technologies in Georgia In contrast, patent applications related to climate change (mit- and compare them with those in regional peers and aspira- igation and adaptation) are rare across the region (Figure 34). tional countries. Box 16 in Appendix C provides a detailed explanation of the methodology used. 64 Chapter 4 Technical Report FIGURE A 34 PANEL New Green Patents in Georgia and across Regional Peers NewPatents Green PANEL A New Green Patents 4 35,000 World (right) 4 35,000 30,000 World (right) 3 30,000 25,000 3 25,000 20,000 2 20,000 15,000 2 15,000 10,000 1 10,000 5,000 1 Georgia 0 5,000 0 Georgia 0 0 2002 2004 2006 2008 2010 2012 2014 2016 2018 2002 2004 2006 2008 2010 2012 2014 2016 2018 PANEL B New green patents by category Climate change adaptation Climate change mitigation Enabling technologies Energy substitution and e ciency Low-carbon energy supply Ukraine (N = 72) Turkey (N = 95) Russia (N = 548) Serbia (N = 12) Kazakhstan (N = 8) Georgia (N = 18) Armenia (N = 3) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Note: (A) reports the 3-year moving average. Source: World Bank calculations based on PATSTAT data and World Bank Data. 65 Chapter 4 Greening Firms in Georgia T his section uses a unique data set that merges Second, the take-up of standard green technologies to CHAPTER 4.5 firm-level energy consumption with very granu- make buildings and premises for greener goods and ser- Summing lar information on the adoption and use of tech- nologies. This lets us explore the role of green and general technologies in explaining green outcomes vices production is still very incipient in Georgia. Only LED lighting (40 percent of firms) and VAC/HVAC systems (30 percent of firms) are more commonly adopted. Other rel- up: The across different production sectors. Both EE and GHG emissions are affected by the technologies used during evant technologies, such as smart thermostats, IoT systems or energy-efficient machinery, are rarely used but show the Extent the production process. The analysis shows that some of these highest potential for improving EE. Not surprisingly, these general technologies not necessarily associated with greener rarely adopted technologies explain a larger share (around 18 outcomes have a strong impact on EE differences, underscor- percent) of the EE gap between the 10th and 90th percentile than of Green ing the potential of broader technology adoption agenda to improve EE. An agenda where improvement in EE and broader the most common ones. Technology productivity improvements go hand in hand. Furthermore, the analysis highlights the relevance of gen- eral technologies in explaining EE differences. We find that The first result is that greening the private sector starts simple indices of technology sophistication explain around 16 Adoption at the energy supply level. To deal with power outages and a low-quality electricity grid, one-third of firms—50 percent among large firms—own a generator, but less than 1 percent produce renewables on-site. This translates into higher GHG percent-17 percent of the 10th to 90th percentile differences in EE, while individual technologies like ERP and automated quality control processes explain 7 percent and 18 percent, re- spectively. Hence, technologies supporting information man- emissions. Evidence from the US shows that diesel generators agement can be vital when implementing sustainability and generate twice as much carbon emissions as grid electricity for EE plans. This suggests that there is a strong complementarity producing the same energy amount. between the agenda to promote digitalization of businesses and the agenda of promoting EE.50 50 Although we acknowledge that this section reports correlations from a cross-section dataset of firms, which flags the need for better data for causal identification, the findings suggest there is an important role that technology plays in greening industries and firms, including green but also general technologies. 66 Chapter 5 Technical Report Chapter 5 66—85 PAGE I n this section, we move into a more detailed micro-lev- Explaining Energy el analysis to identify the firm-level drivers that would help tackle the challenges of mitigating emissions. Specifically, the section assesses the factors underlying EE Efficiency at the and the drivers of efficiency convergence toward the most efficient firms. Due to the impact of enhancing EE on ener- gy consumption and profits, it examines the drivers of EE Firm-Level in the business sector, looking at firm attributes and spillovers from other firms. Understanding such factors provides valuable information for designing and targeting public interventions at the firm level. P roductivity drives EE improvement in Georgia, percent, on average (columns 6 and 5). Alternatively, an in- CHAPTER 5.1 even among small and large firms. Besides factors crease in productivity sufficient to move the firm from the The affecting energy demand, such as prices and subsi- dies, firm-level characteristics can provide insights to bottom quartile to the top quartile would double its EE, which for the median firm would imply a reduction in energy costs Correlates drive business EE policy in Georgia. Table 14 shows equal to 51 percent.52 Hence, firm upgrading (innovation, ICT the determinants of firm EE in Georgia. As the table adoption, managerial practices) could have a dual effect on shows, total factor productivity (TFP)51 positively cor- EE. A direct effect is from changing production techniques of Energy relates with EE, even after controlling for many firm charac- that reduce energy consumption per output, and an indirect teristics, including time-invariant unobservable factors. A 10 effect is from TFP efficiency gains. A rise in EE due to TFP percent rise in firm TFP would enhance EE by 4.9 percent-8.1 improvements could benefit firms of all sizes.53 Efficiency 51 The TFP is estimated at the firm level using a structural production function estimation approach for each industry group (two-digit level) separately. The analysis of total factor productivity is based on an estimation of Cobb-Douglas type production function in the following from: , where , , and represent the output, intermediate inputs, labor and capital of firm i in time t respectively. Production functions are estimated using Ackerberg et al. (2015). The estimation sample covers the period from 2007 to 2019. 52 Increasing TFP in a magnitude equal to moving a firm from the bottom to the top quartile of the distribution implies boosting TFP by 1.701, equal to an efficiency growth of . For the median firm of the energy efficiency distribution , this would imply a reduction in energy costs by 51 percent. 53 In other specifications we consider the interaction between size class and productivity (size ×TFP) and the type of ownership and productivity (ownership ×TFP). However, results reveal very small or no differential relationship between energy efficiency and TFP across firm size class. 67 Chapter 5 Greening Firms in Georgia TABLE 14 Correlates of Energy Efficiency OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) Market concentration (sector x -0.005*** -0.003*** municipality) (0.001) (0.001) Local efficiency frontier (sector x 0.473*** 0.234*** municipality) (0.023) (0.008) Sector efficiency frontier -0.042* 0.008 (0.024) (0.008) Regional efficiency frontier -0.019 0.092* (0.059) (0.046) TFP (log) 0.838*** 0.533*** 0.820*** 0.500*** (0.047) (0.025) (0.052) (0.033) Log capital per worker -0.093*** -0.053*** -0.088*** -0.039*** -0.091*** -0.038*** (0.009) (0.010) (0.009) (0.012) (0.009) (0.012) Large (+100) -0.404*** -0.149*** -0.255*** -0.119*** 0.063** 0.055 (0.056) (0.033) (0.031) (0.041) (0.025) (0.046) Start-up (0-4) 0.218*** -0.148** 0.065 -0.249*** 0.025 -0.273*** (0.051) (0.059) (0.052) (0.060) (0.049) (0.043) Foreign-owned private 0.114* 0.055 0.123** 0.038* 0.107** 0.032** (0.066) (0.049) (0.059) (0.023) (0.045) (0.015) State-owned -0.450*** 0.062 -0.237*** 0.113** -0.213** -0.119** (0.107) (0.059) (0.062) (0.056) (0.081) (0.048) Self-governed -0.523*** -0.033 -0.311*** 0.013 -0.569*** -0.209*** (0.146) (0.093) (0.084) (0.097) (0.119) (0.069) Start-up x log capital per worker -0.006 0.020*** 0.011* 0.032*** 0.012* 0.033*** (0.006) (0.006) (0.006) (0.007) (0.006) (0.005) Constant 4.099*** 4.630*** 1.790*** 3.166*** -0.136 1.537*** (0.190) (0.094) (0.329) (0.180) (0.480) (0.338) Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No Yes No Yes No Industry FE (2-digit) Yes No Yes No Yes No Firm FE No Yes No Yes No Yes R-squared 0.368 0.015 0.423 0.056 0.425 0.064 Observations 67,888 67,888 61,474 61,474 47,638 47,638 No. firms 20,764 19,599 15,946 No. clusters 67 67 66 66 47 47 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance at (10), (5) and (1) percent level. The left-hand side variable is energy efficiency, defined as Yit in equation (1). Baseline categories are micro (0-9 employees), established (5+ years), and domestic- owned firms. Source World Bank elaboration based on GEOSTAT.Source: World Bank calculations based on FAT surveys. 68 Chapter 5 Technical Report 54 Due to sample size limitations, firms are grouped at the 2-digit sector by size class level in this section. Size class classifies firms according to the number of employees -SMEs (1-99) and large (100+). 55 Unobservable individual effects that do not change over time. Market concentration, although small in magnitude, ap- tween 2.34 percent and 4.73 percent in the firm’s efficiency.57,58 56 Indeed, there is at least anecdotal evidence from pears to have some (negative) impact on EE, which could other countries that where point to the effect of lack of competition on innovation Once local frontier spillovers are accounted for, sectoral incentives are created – e.g., and investment incentives. Market concentration, measured frontier spillovers are small and insignificant. Sectoral subsidies – for carrying out as the average markup of top-25 percent firms with the highest spillovers are related to industry-specific ways technology and energy efficiency audits, the markups within their sector-size group54 and municipality, knowledge are shared. Although cross-section analysis shows emergence of these local correlates negatively with EE. After accounting for firm fixed negative point estimates, the sign of the estimated coefficients business services plays an effects55, the negative association persists, although it is close to turns positive after accounting for firm FE. These sectoral spill- important role in driving efficiency gains. zero and significant only at the 10 percent level -including firm overs are statistically insignificant and economically small in FE could obfuscate its impact given that market concentration magnitude. In fact, a 10 percent growth in the sector frontier 57 We distinguish the local, changes slowly. A plausible explanation for this effect is that efficiency would increase firm-level efficiency by 0.8 percent, sectoral and regional frontiers market concentration may hinder innovation and investment on average (ceteris paribus). for better understanding (De Loecker, Eeckhout & Unger, 2020), which ultimately could factors that may be behind affect the efficiency at which firms use energy. Regional efficiency spillovers can contribute to enhancing these effects. The local frontier points to factors EE but moderately. Spillovers from the regional efficiency within the same sector and There are positive spillovers from the presence of more frontier -those purely attributed to the region, beyond the local region of location of the efficient firms in the same industry or region pointing (sector-region) frontier- on EE are close to zero in the pooled firm (as a reminder, it is the to the importance of market failures and underinvest- regression (OLS), but when firm FE are included, they are sig- average efficiency of the top ment in EE at the firm-level. The analysis includes the local nificant: a 10 percent increase in the region average efficiency 25 percent most efficient (sector × municipality), sectoral (2-digit-by-size) and munici- of top firms is associated with a rise of 0.92 percent in the ef- firms -top 25 percent - pality efficiency frontiers (columns 5 to 6) to detect spillovers ficiency of the firm. operating in a given sector and region). The sectoral frontier from other firms. The local frontier -the average efficiency of relates to the same sector of top-25 percent efficient firms within each sector-region com- Overall, spillover effects underscore the importance of the firm (across all regions; bination- can only affect the firm efficiency level if there are collaboration and synergies between firms and the role it is the average efficiency technological spillovers (i.e., technology and knowledge that of dynamic, innovative local ecosystems. Transfer of tech- level of the top 25 percent spread from more to less efficient firms). Additionally, spatial nology and knowledge is relevant for EE. Low-efficiency firms excluding companies in the spillovers could also happen through the labor market (i.e., could benefit and learn about best practices, technology and same region of location of the workers moving between firms and sharing know-how, knowl- know-how of firms at the frontier of EE. Also, frontier firms firm), and the regional frontier edge and best practices), supply chains relationships (i.e., sup- can take advantage of more dynamic and vibrant economic relates to the same region of the firm (across all sectors; pliers and customers offering and demanding products with environments. Thus, policy actions should leverage spatial and it is the average efficiency specific energy standards, and exchanging information about sectoral spillovers in their green transition strategy. level of the top 20 percent frontier technology), or interactions between managers and excluding those companies in technical staff. An alternative way through which spatial spill- Controlling for basic firm characteristics, more capital-in- the same sector of activity of overs occur could also be the existence of localized business tensive firms are less efficient, meaning capital demands the firm). service markets (consultants in EE).56 more energy than manual labor. Machinery and equipment 58 The average require energy to produce goods and services. So, it is expected efficiency of the local Firms physically surrounded by high-EE firms operating that conditional on firm characteristics and overall efficiency, frontier is and in the same sector tend to display higher levels EE. Estimat- firms using capital more intensively (i.e., having larger amounts . A 10 percent ed coefficients for the local efficiency frontier are positive and of fixed assets per worker) demand more energy to produce one increase is less than one-tenth statistically significant. Columns (5) and (6) report cross-sec- unit of output than those that rely more intensively on labor. of a standard deviation. tion and fixed effect point estimates that include FE at the firm According to the estimates, a 10 percent increase in capital per 59 Private firms with at level. In both cases, estimates are positive and statistically worker is associated with 3.9 percent lower EE. least 50 percent of total significant, although FE estimates are lower in magnitude than shares owned by foreign cross-section ones. An increase of 10 percent in the efficiency Domestic-private businesses are less energy efficient than shareholders. level of the local frontier is associated with an average rise be- foreign59 firms, but more efficient than state-owned and 69 Chapter 5 Greening Firms in Georgia self-governed firms, partly due to productivity levels.60 perspective. At the same time, startups are likely to face capital Evidence from the pooled regression (columns 1, 3 and 5) shows constraints and may have to adopt outdated machineries be- that foreign-private firms are between 11-12 percent more effi- cause of lack of credit. cient than domestic ones (baseline) and 3.2-3.8 percent more efficient if firm-level TFP and unobservable time-invariant One concern of the results is related to the impact of effects are accounted for (columns 4 and 6).61 Differences in firm outsourcing on firm differences in EE. For example, efficiency levels between foreign- and domestic-private could high-productivity or more sophisticated companies may out- point to standardized production methods, quality and effi- source transport or high-energy-intensive activities, which ciency standards, and management practices that multina- could lead to higher efficiency levels. If the outsourcing lev- tional or international companies tend to adopt but that tend el were positively correlated with productivity, this would be to be less common among domestic firms. Moreover, even for (at least partially) captured by including the TFP as a control the same TFP levels, public (those with at least 50 percent of variable. Unfortunately, since we do not have a variable that state-owned shares) and self-governed companies62 are con- allows the identification of the degree of outsourcing and we sistently less efficient than domestic-private businesses (12 acknowledge that contracting external suppliers may be rel- to 21 percent in the former and 21 to 57 percent in the latter evant and related in several ways to EE, this could be part of in the most preferred specifications -columns 5 and 6). This future research on EE once data more data become available. result stresses that improving EE and cutting carbon emissions need to be part of the productivity agenda among public and A second concern may point to the fact that the energy private businesses. source consumed by firms could drive the relationship between firm characteristics and EE. We replicate the ap- Large firms (100+ employees) appear to be less energy ef- proach undertaken in Table 14 but slightly change the EE defi- ficient than SMEs (1-99), but this result is not robust to in- nition. Since energy sources vary across sectors, efficiency is cluding more controls. SMEs (1–99 employees) tend to use en- defined in terms of the main energy input used by sectors. For ergy more efficiently than large firms according to the estimates example, if electricity in a sector is the largest energy input, EE 60 We consider alternative specifications to check that do not consider the company’s market share (columns 1-4) for every firm within that sector is computed using electrici- whether other factors are but that account for firm FE. However, once the market concen- ty costs rather than total energy costs.64 Logically, as energy driving these results. For tration and firm FE are included (i.e., the share of sales within the inputs vary across sectors, so do the main energy sources. We instance, we omit the size and sector and municipality), efficiency differences between SMEs perform this analysis for electricity, and oil and oil products, age dummies but conclusions and large companies shrink and are no longer significant (col- the two largest sources of energy and the ones with a larger about the relationship umn 6). Large firms appear as efficient as SMEs when comparing number of firm records. between type of ownership firms with similar capital intensity and market concentration and energy efficiency hold. and controlling for unobservable time-invariant firm attributes. TFP is positively and significantly correlated with elec- 61 It is worth noting that When we add these controls, larger companies are marginally tricity and oil EE. TFP matters for energy enhancement and firm ownership does not more efficient, which could ultimately point to factors such as remains a key energy improvement driver. While positive and change substantially over differential access to finance and greener technology and larger statistically significant -irrespective of the energy source-, the time as businesses tend to room to develop environmental-oriented managerial actions. correlation between TFP and efficiency is stronger among those keep their ownership status. sectors more intensive in oil energy after controlling for firm FE. Hence, within-firm variation is low, consequently limiting Supporting access to capital and management training the interpretation of the FE of young firms and startups could help improve the EE of The local frontier is key in electricity, gas and oil-depen- approach. startups. Sometimes it is assumed that startups are more energy dent activities. Spillovers from the local EE frontier are rele- efficient than mature firms as they may enter in less pollut- vant for the firm EE, irrespective of whether businesses con- 62 We observe an average ant activities with modern technology and greener production sume electricity, gas or oil and fossil fuels more intensively. The of 164 firms with foreign shareholders per in 2007- methods. However, this is not always the case. The EE of Geor- assessment of the correlates of EE, which groups activities by 2021. gian startups (firms established since less than 4-year) varies the intensity with which they use energy sources, shows that considerably and is positively associated with capital intensity estimated coefficients are positive and statistically significant 63 The relationship between (i.e., capital per worker). High capital-intensive startups are more both with and without firm FE for all power sources (Table 15). age class and capital intensity energy efficient than low capital-intensive ones comparing busi- Remarkably, spillovers from the local frontier across electric- is gauged through the nesses with similar size and productivity (column 6).63 While on ity-intensive industries double in magnitude those of sectors interaction between a startup dummy variable and the average startups are less energy efficient than established firms that demand oil and fossil fuel products as their main power logged capital per worker. (i.e., firms aged 5 or more), these results are mainly driven by low source. Beyond the local frontier, evidence shows that condi- capital-intensive startups. EE is substantially lower among low tional on the power source, sectoral and regional spillovers 64 Firms have only one capital-intensive startups while differences between established appear not to have differential effects since, even though they indicator of energy efficiency, firms and high capital-intensive startups are more nuanced. This tend to be positive, estimates are not statistically significant in either based on electricity, result could be driven by two factors. First, high capital-intensive the OLS and FE regressions. At face value, this points out the gas, or oil and oil products. Hence, those with electricity startups are likely to adopt modern, cutting-edge technology relevance of physical proximity for technology and knowledge as main source only enters in and have younger managers that understand the importance sharing within sectors and that linkages across sectors or re- the electricity equation. of optimizing energy use with an efficiency and environmental gions tend to be weaker. 70 Chapter 5 Technical Report TABLE 15 Robustness check: determinants of energy efficiency by main source Electricity Gas Oil and fossil fuels OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) Market concentration -0.006*** -0.002** -0.006 -0.005** -0.003* -0.003 (0.002) (0.001) (0.004) (0.002) (0.002) (0.002) Local efficiency frontier 0.572*** 0.194*** 0.454*** 0.134*** 0.253*** 0.092*** (0.039) (0.010) (0.053) (0.029) (0.018) (0.016) Sector efficiency frontier -0.023 0.025 0.021 -0.029 -0.014 0.037 (0.027) (0.016) (0.127) (0.044) (0.016) (0.023) Regional efficiency frontier -0.189* -0.003 0.027 0.029 0.051 0.090** (0.107) (0.041) (0.137) (0.088) (0.055) (0.044) TFP 0.682*** 0.411*** 0.890*** 0.437*** 0.882*** 0.550*** (0.079) (0.032) (0.075) (0.059) (0.037) (0.048) Log capital per worker -0.133*** -0.041*** -0.065*** -0.045*** -0.061*** -0.018 (0.006) (0.013) (0.013) (0.015) (0.018) (0.011) Start-up x log capital per worker 0.047*** 0.009 0.050*** 0.043*** -0.006 0.014* (0.017) (0.009) (0.018) (0.014) (0.008) (0.008) Constant 1.623** 2.882*** -1.477 2.461*** -0.147 1.612*** (0.648) (0.288) (1.041) (0.536) (0.405) (0.404) Size, ownership and age controls Yes Yes Yes Yes Yes Yes Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No Yes No Yes No Industry FE (two-digit) Yes No Yes No Yes No Firm FE No Yes No Yes No Yes R-squared 0.492 0.113 0.639 0.113 0.430 0.078 Observations 19,847 19,847 4,298 4,298 16,501 16,501 No. firms 8,764 2,304 6,180 No. clusters 44 44 33 33 43 43 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Energy efficiency is based on the main energy input at the sector level. Definitions and note Table 14 apply. Source: World Bank elaboration based on GEOSTAT. 71 Chapter 5 Greening Firms in Georgia Georgian firms that invest in physical or intangible assets, that these activities could be relevant for greening. Furthermore, R&D, or ICT are greener. Technology adoption, innovation, outcomes differences are small (except for innovation) when digitalization and trading with foreign firms are part of a broader comparing low- and high-EE activities (Figure 54A, Appendix set of instruments that can help upgrade firm capabilities and A). Still, estimated coefficients associated with investment in efficiency. However, less is known about how they relate to green fixed assets, R&D, and innovation are positive and statistically firms. Figure 35 asses the relationship between these character- significant for explaining EE while statistically equal to zero for istics after accounting for productivity level, size, age, sector, high-efficiency groups. Firm digitalization can therefore play a region, and time factors to address this question.65,66 Firms that key role in reducing energy consumption in Georgia while also either invest in tangible or intangible capital are more efficient boosting business performance through TFP growth. In the next than firms that report no investments. Also, the scale of invest- section, we will build on this and provide a deeper dive into the ments is positively associated with the level of EE. Similar con- effect of technology on EE. clusions can be drawn from firms investing in R&D, suggesting 65 Estimated coefficients for investment, innovation and ICT adoption firm-level characteristics result from a two-step regression. The first step regresses firm-level energy efficiency on the TFP and a set of covariates including size and age class, industry (two-digit level), region, and year of reference ( , where j denotes firm sector, r the region of location, and t the year of reference; is the error term). We generate regression residuals ( , and then residuals are regressed on each variable. 66 We also distinguish between high and low and high energy efficiency sectors (sectors above or below the median of energy efficiency). Estimated coefficients and confidence intervals are plotted in Figure 54A. We also added the type of innovation (product or process) specification, but it did not change previous results. FIGURE 35 The Relationship between Energy Efficiency and Investment, Innovation, ICT, and Exporting Invests Growth in capital stock Invests in R&D Innovates ICT adoption (z-score) Firm exports -0.10 0.00 0.10 0.20 0.30 0.40 0.50 Gap relative to baseline Notes: Dots depict point estimates and brackets confidence intervals (10 percent level of significance). All variables are binary (No = 0, Yes = 1) except for log investment and ICT adoption z-score. Growth in capital stock is the first difference of the logged capital stock (conditional on being positive). Source: World Bank elaboration based on GEOSTAT and the ICT Usage in Enterprises and Innovative Activity of Enterprises Surveys. 72 Chapter 5 Technical Report W hile in the previous section we concentrat- Acknowledging this heterogeneity in the convergence CHAPTER 5.2 ed on correlates of firm efficiency levels, in growth rates of EE across and within sectors, and the Energy this part, we focus on the firm characteris- tics associated with changes in efficiency, firm-level factors underlying such differences, this sec- tion focuses on the firm convergence to the EE domestic Efficiency and specifically with convergence toward frontier following a similar approach as in Brown et al. the efficiency “frontier.” 67 (2016). Firms may absorb knowledge, make strategic decisions about technology and ICT adoption and production methods, Conver- There are striking differences in efficiency improvements across firms within the same sector that motivate the convergence analysis. As Figure 36 illustrates, as well as implement general and green managerial practic- es. So, they differ in their capacity to improve EE and, thus, in the convergence speed to the efficiency frontier. Based gence there is a huge variation in within-sector efficiency changes. So firm decisions appear to be relevant for overall efficiency convergence rather than sector composition of the Georgian on the main findings from Table 14, the forthcoming analy- sis explores the firm-level EE convergence to the efficiency frontier. The efficiency frontiers are defined at the level of economy. In this regard, firm-level upgrading can be crucial sector-by-region, sector, and region68 to capture different for promoting a greener business economy in Georgia through convergence drivers and potential spillover effects. Box 11 the optimization of resource usage but also by upgrading presents the baseline specification, estimation methods and firm capabilities in terms of clean technology, green ener- methodological notes used throughout this section. Defini- gy generation and management, control of pollution, and tions introduced in Box 8 also apply. improvements of infrastructure, machinery, and equipment related to energy consumption. 67 We define the local efficiency frontier as the average efficiency level of the 25 percent most efficient firms within their sector-size group and municipality of operations, as described in equation (3). Additionally, we introduce the sector and regional frontiers in the analysis as defined previously. The former is defined as the average efficiency level of the 25 percent most efficient firms within the sector-size group excluding firms in the same municipality of location as the firm of reference, while the latter is the average efficiency of the 25 percent most efficient firms within the municipality of location excluding firms within the same sector-size group of operations as the firm of reference. 68 Definitions in Box 8 apply. FIGURE 36 Energy Efficiency Change within and across Sectors 0.8 Density 0.7 Across sectors-by-size 0.6 0.5 0.4 0.3 0.2 Within 0.1 sector-by-size 0 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 Source: World Bank elaboration based on GEOSTAT. Energy e ciency change 73 Chapter 5 Greening Firms in Georgia BOX 11 Assessing Energy Efficiency Convergence in the Private Sector of Georgia The firm’s EE level could be regarded as a function of tance to the sector-region, sector, and region efficiency fron- the EE in the previous year, the sector efficiency frontier tier, respectively, and Xisrt-1 is a vector of the firm’s lagged within the sector of production, the domestic sector effi- time-variant characteristics (including capital intensity, av- ciency frontier, and the overall regional efficiency frontier erage wage, investment, size, age, and ownership type). The in addition to a set of firm features such as productivi- vector ∆ t denotes the change in the local, sector, and re- ty growth, size, age, type of ownership, exporting sta- gion efficiency frontier. The baseline equation also considers tus, wages paid (a proxy for the workers’ level of skills), time-invariant industry (πs) and geographic effects ( r) capital intensity, and investments. Hence, taking the first that could affect the changes in EE, technology and econom- differences, equation (8) presents the baseline regression ic trends ( ), and firm-level time-invariant unobservable for the EE converge assessment. The terms characteristics ( ). denote the firm’s dis- Different specifications of equation (8) are estimated both firm’s distance to the efficiency frontier is a (reverse) function using ordinary least squared (OLS) and fixed effects (FE) of the firm’s efficiency in the previous year. According to Bond methods. Besides the latter providing estimated coefficients (2002), the OLS approach would yield downward-biased esti- net of firm-level unobservable time-invariant effects that mates, while FE estimates would be upward-biased. As point- may otherwise bias results, there is an additional reason for ed out by Brown et al. (2016), absent instrumental variables presenting both results for each specification. A lagged de- that could eliminate the endogeneity problem, reporting both pendent variable on the right-hand side of the equation can OLS and FE results could shed light on the range in which the retrieve inconsistent estimates under both OLS and FE ap- ‘true’ estimate lies, as the former could be interpreted as the proaches (Brown et al., 2016). The term that measures the lower bound and the latter as the upper bound estimate. Firms further away from the ‘local’ efficiency frontier (the and sectoral (two-digit level) efficiency frontiers rather than average efficiency of the top-25 efficiency firms within the ‘local’ frontier only. Equations (columns 5 and 6) include their sector and municipality) make greater EE improve- the sectoral and regional frontiers. Overall, evidence shows ments, suggesting the existence of a convergence process. systematic convergence toward the regional frontier and po- Considering a broad number of specifications and even after tential convergence to the sector frontier. Firms appear to adding the distance to sectoral and regional frontiers and firm converge toward the most efficient firms in their municipal- characteristics, regression results provide consistent evidence ity, even when they do not belong to the same industry. This of EE convergence across Georgian firms toward the ‘local’ could suggest that in contrast to the factors associated with frontier.69 Firms further away from the sector frontier within EE levels, knowledge and best practices are shared between their municipality display the largest changes in EE. Also, con- firms within municipality boundaries, underscoring the rel- clusions hold if we restrict the analysis to ‘stayers’ (i.e., active evance of physical proximity for knowledge and technology firms in all years of 2007-2021), ruling out potential concerns transfer. Although evidence is limited, geographic proximity of observing an allocative effect instead of real convergence.70 could enhance knowledge transfer. For example, Abramovsky & To quantify the time that firms may spend converging to the Simpson (2011) find that firms of specific industries (i.e., phar- frontier, back-of-the-envelope estimations suggest that, by maceutical) that locate R&D near frontier research departments 69 Conclusions hold if energy efficiency is defined 2029, one-quarter of laggard firms (those at the bottom-75 per- are more likely to engage with universities. Moreover, Helm- in terms of energy quantity cent of the efficiency distribution) would catch up to the local ers (2019) finds evidence of positive inter- and intra-industry consumption instead of energy frontier if they managed to keep their current catch-up speed. knowledge spillovers among firms close to a science park. In the value consumption for all same vein, Siegel, Westhead & Wright (2003) provide evidence robustness checks presented The regional efficiency frontier also drives convergence that firms within university science parks have higher research in this section. dynamics in EE, underscoring the relevance of geographic productivity than equivalent firms outside. Unfortunately, with 70 We provide further proximity for knowledge sharing. To further understand the the available data, we cannot test the potential mechanisms proof of these results in the efficiency catch-up dynamics, the analysis assesses whether that are the source of these spatial spillovers, and it is a margin Appendix A (see Table 27A). the convergence is also toward the regional (municipality-level) for improving the current analysis. 74 Chapter 5 Technical Report Our results indicate knowledge flowing from high- to technological shifters can only affect the firm’s efficiency levels low-efficiency firms within sectors. Inter-firm relationships if there are technological spillovers (i.e., EE improvements are crucial for knowledge and best practices sharing. There are due to technology changes that spread across firms; Brown et several channels through which knowledge flows across firms al., 2016). Spillovers are estimated by assessing the relation- within the same industry. For example, industry associations ship between the change in the different definitions of the (i.e., round tables, networking events, sharing of best practices efficiency frontier (local, sector and region) and the outcome via collaboration workshops) and supply-chain relationships variable (efficiency change). Columns (1) to (4) include local (i.e., supplier-customer linkages) may be two relevant chan- frontier changes, whereas columns (5) and (6) add the changes nels. Moreover, labor mobility (i.e., employees reallocating in both the sector and region frontier. The results in Table 16 between firms and sharing their know-how) and training pro- suggest positive local and regional spillovers, consistent with grams could also improve energy use among low-efficiency findings in Table 14. According to estimates from columns firms. Also, the competition environment and the innovative (5) and (6), local and regional technological shifters are rele- performance of high-efficiency firms within sectors could po- vant for explaining the changes in efficiency improvements, tentiate the knowledge of low-efficiency firms through learning whereas sector technological shifters are positive but weak. by observing or by appropriating (at least partially) part of the To provide an order of magnitude, a one standard deviation knowledge generated by frontier firms. Evidence suggests that increase (sd=0.836) in the change of the local efficiency fron- convergence to the sector frontier is weak once the distance to tier would speed up efficiency improvements in one-tenth of the local and regional frontiers is controlled (columns 5 and 6). a standard deviation of firm-level efficiency changes. Results Some factors might be hindering knowledge and technology for the sector and regional frontier changes are half and up to flows within sectors that, if correctly addressed, can potentiate one-fourth in magnitude compared to the local frontier due the EE of firms. Unfortunately, such factors cannot be tested to the size of the respective standard deviations.71 The three due to lack of data. sources of spillovers reinforce the process because they are all positive and, in the case of the local and regional frontiers, also Technological shifters also affect the energy efficiency of statistically significant for both OLS and FE regressions. From firms, so access to cutting-edge technology and produc- a policy perspective, the spillovers point to the relevance of tion methods is crucial for greening firms in Georgia. The information sharing, collaboration, imitation, and technology analysis also accounts for the potential presence of techno- transfer as channels that enhance EE in Georgia. logical shifters that affect the efficiency level of firms. These 71 The mean change in the local efficiency frontier is ( = 0.035 and the standard deviation . For firm-level energy efficiency changes, =–0.039 and =0.982. Given that =0.196 (mean of columns 5-6), a one-standard deviation increase in the change of the local frontier would result results on an average acceleration of efficiency improvements of 0.164= × , equal to one-tenth of efficiency change standard deviation 0.167= . Similarly, for the sector frontier change, =0.024 and =0.361, so 0.009= . For the regional frontier change, =0.216 and =0.461, so 0.101= . 75 Chapter 5 Greening Firms in Georgia TABLE 16 Estimation Results: Energy Efficiency Convergence OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) D(local frontier) 0.249*** 0.733*** 0.240*** 0.717*** 0.138*** 0.240*** (0.005) (0.011) (0.006) (0.016) (0.009) (0.015) D(sector frontier) -0.003 0.065* (0.009) (0.038) D(regional frontier) 0.099*** 0.500*** (0.009) (0.031) Delta local frontier 0.276*** 0.464*** 0.258*** 0.445*** 0.209*** 0.183*** (0.011) (0.011) (0.011) (0.014) (0.012) (0.015) Delta sector frontier 0.001 0.046* (0.015) (0.025) Delta regional frontier 0.107*** 0.325*** (0.037) (0.039) Delta TFP 0.448*** 0.330*** 0.454*** 0.305*** (0.019) (0.016) (0.017) (0.013) Large (t-1) 0.067*** 0.575*** 0.085*** 0.550*** -0.033** 0.169*** (0.012) (0.052) (0.012) (0.053) (0.015) (0.052) Start-up (t-1) 0.015* 0.002 0.004 -0.020** -0.063*** -0.031* (0.009) (0.014) (0.007) (0.010) (0.016) (0.017) Foreign-owned private (t-1) 0.021 0.039*** 0.022** 0.058** 0.035 0.064 (0.013) (0.014) (0.010) (0.022) (0.024) (0.046) State-owned (t-1) -0.131*** -0.146* -0.049* -0.007 0.037 -0.086 (0.038) (0.083) (0.028) (0.034) (0.042) (0.101) Self-governed (t-1) -0.002 0.023 0.129 0.068 0.344*** 0.171 (0.073) (0.161) (0.103) (0.228) (0.116) (0.226) Start-up (t-1) x D(local frontier) 0.042*** 0.022** (0.011) (0.011) Large (t-1) x D(local frontier) 0.015** -0.047*** (0.006) (0.009) Foreign-owned private (t-1) x D(local -0.005 -0.005 frontier) (0.010) (0.026) 76 Chapter 5 Technical Report TABLE 16 Estimation Results: Energy Efficiency Convergence (cont.) OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) State-owned (t-1) x D(local frontier) -0.051*** 0.040 (0.016) (0.036) Self-governed (t-1) x D(local frontier) -0.092** -0.058 (0.036) (0.075) Constant -0.284*** -1.049*** -0.313*** -1.039*** -0.435*** -1.174*** (0.044) (0.026) (0.051) (0.025) (0.062) (0.026) Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No Yes No Yes No Industry FE (two-digit) Yes No Yes No Yes No Firm FE No Yes No Yes No Yes R-squared 0.143 0.383 0.157 0.390 0.162 0.433 Observations 35,405 35,405 26,862 26,862 25,951 25,951 No. firms 10,869 8,050 7,951 No. clusters 60 60 55 55 55 55 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. The left-hand side variable is efficiency change (ΔY_isrt). Right-hand side variables are lagged once except those expressed as first differences (changes). Each specification includes industry (two-digit level of NACE Rev. 1) and geographic (municipality) fixed effects (FE), and year effects. Columns (2), (4) and (6) add firm FE. Baseline categories are SMEs (1-99), established (5+ years old), and domestically owned firms. Source: World Bank elaboration based on GEOSTAT Tbilisi does not drive efficiency convergence; it is a coun- compared to the rest of the country. At some point, this can be try-wide process. Since the dominance of Tbilisi in the econ- expected. Tbilisi concentrates one-third of the population of omy could be driving the convergence results, we include the Georgia and is around six times as big as the second largest city interaction of Tbilisi with key explanatory variables such as of Georgia. So, although they exist across the country, agglom- distance to frontiers, change in the frontiers and TFP growth. eration forces are expected to be stronger in Tbilisi. Second, Were Tbilisi running convergence outcomes, assessing the in- firms converge to the sectoral EE frontier, although they appear teraction between Tbilisi and key variables of interest would to be close to zero in the case of Tbilisi. OLS and FE estimates help to address this concern. Table 29A (Appendix A) reports associated with D(sector frontier) range between 0.013-0.178 regression outcomes. Overall, we do not find evidence of Tbilisi but those associated with the differential effect of Tbilisi - D(- driving convergence results. In some cases, Tbilisi potentiates sector frontier)×Tbilisi- turn negative, ranging between -0.033 a country-wide catch-up process, while in others it makes it and -0.166. Finally, as for regional frontier - D(regional fron- slower. Estimated coefficients associated with the interaction tier)-, estimated coefficients are between 0.098-0.430, while between Tbilisi and the distance to the local frontier -D(local evidence shows no differential spillovers for Tbilisi – the esti- frontier)×Tbilisi- reported in columns (5) and (6) are positive mated parameter associated with D(regional frontier)×Tbilisi is and significant (both in the OLS and FE specifications). This not statistically significant. In sum, according to the evidence, means that convergence speed is faster among Tbilisi firms but spatial concentration around Tbilisi is not driving efficiency does not imply that there is no effect beyond the capital city catch-up. Still, it is worth mentioning that this finding does not of Georgia. Indeed, estimated coefficients associated with the rule out that, in some cases, spillover effects can be stronger distance to the local frontier -D(local frontier)- are economi- around Tbilisi, given the characteristics of the municipality. cally and statistically positive, implying that the convergence to the local frontier is country-wide rather than Tbilisi-driv- The analysis also attempts to identify firm character- en. However, the estimate associated with the interaction be- istics that may systematically explain the variation of tween Tbilisi and the distance to the local frontier - D(local efficiency improvements and the speed of efficiency con- frontier)×Tbilisi- is also positive and significant, suggesting vergence. Table 16 reports the estimated coefficients for the that within-sector agglomeration effects are stronger in Tbilisi firm’s ownership type, size and age class (columns 1 to 6) and 77 Chapter 5 Greening Firms in Georgia the TFP change (3 to 6). To shed light on the factors affecting are lower, even though they are not statistically significant the speed of convergence, Table 16 also includes the interac- when controlling for firm fixed effects. tion between these variables and the firm’s distance to the region-sector-by-size efficiency frontier (columns 5 to 6). Start-up firms converge faster than older firms to the local frontier, while large companies make it at a slower pace. Productivity growth is key for greening firms in Georgia. Convergence analysis also considers firm characteristics that Productivity performance appears to drive EE improvements may affect the efficiency catch-up dynamics across Georgian at the firm level in Georgia. Table 16 (columns 3 to 6) reports a firms by interacting those attributes with the distance to the lo- positive relation between changes in TFP and efficiency. These cal efficiency frontier (columns 5 and 6 in Table 16). According conclusions hold after accounting for firm-level time-invariant to regression results, the convergence speed to the local frontier effects, albeit FE point estimates are lower than OLS. This find- among large companies is slower than SMEs (column 6). As of ing is crucial for a greening strategy from the policy viewpoint. age, the pace of convergence of startups is faster than that of If higher TFP growth is associated with greater energy improve- established firms. According to these results, startups appear to ment, then boosting within-firm productivity and upgrading be more likely to acquire greener technologies and managerial firms’ capabilities through promoting technology adoption, practices in addition to addressing market demands on the use encouraging innovation and investment in ICT, improving of energy and the emissions of greenhouse gases. management and organization practices, and easing access to finance can make firms greener. Finally, firm-level efficiency convergence also occurs in electricity- and oil-intensive sectors. Since efficiency catch- Evidence, although weak, shows that efficiency improve- up may be related to the type of energy consumed, we perform ments could be higher among startups and large firms. more robustness checks distinguishing between activities with Regression outcomes show evidence that efficiency improve- different energy input requirements (electricity, gas, and oil and ments among foreign-owned private enterprises do not differ fossil fuels; Table 30A in Appendix A). Results show efficiency from domestic-private firms. In the preferred specification convergence to the local, sector, and regional frontiers across all (6), the point estimate associated with foreign-owned firms is energy sources (except for the case of gas and the sector frontier). positive but not statistically significant, meaning no difference The size of the spillovers varies with the energy source. In elec- in EE improvements between domestic and foreign-private tricity-intensive industries, local spillovers are greater than sec- firms. Looking at size and age class characteristics, there is tor and regional spillovers but smaller than local frontier effects evidence of the differential performance of large firms (at least in gas-intensive industries. Furthermore, the regional frontier is 100 employees) relative to SMEs (1-99 employees). As for age, more relevant than the local frontier in oil-intensive activities. estimates show that efficiency improvements among startups CHAPTER 5.3 The Importance of Green Management Practices for Energy Efficiency A The Drivers of Green Management T his subsection addresses the importance of man- tices undertaken by Georgian firms. We consider it a key input agement and green management quality as a po- for business-targeted interventions. tential instrument for improving EE and reducing greenhouse emissions in the business sector of Management practices are one of the key drivers—or Georgia. Benefiting from the World Bank Enterprise “levers,” as explained by Syverson (2011)—for upgrading Surveys (WBES), which include a module on the green firms’ internal capabilities to operate efficiently and im- economy, this section tries to answer what drives green proving productivity. Bloom et al. (2013) explain that man- management practices. The analysis provides insights that can agement practices can be regarded as technology—manage- help design and implement policy interventions seeking the ment is intangible capital positively correlated with output. green transition in Georgia. However, given that we are using Management practices can help explain a relevant share of a different data source, results may not be fully consistent with productivity differences across regions and firms (Bloom et the previous section, although our findings are similar or do al., 2012, 2017, 2020; Bloom & Van Reenen, 2010). not necessarily contradict previous conclusions. It is worth mentioning that firm-level records from the GEOSTAT Enter- Managerial practices are not limited to boosting firm prise Survey are the most reliable data, as the statistical agency productivity directly through improving organizational systematically collects information on thousands of enterprises capacity and efficiency. Management practices can green over time, while the sample size of the WBES is significantly firms by unlocking innovation and technology adoption. For smaller and has information on the green economy for 2019 example, Grover, Iacovone & Chakraborty (2019) find that high- only. With that said, this section sheds light on the green prac- er quality management is associated with a higher probabil- 78 Chapter 5 Technical Report ity of adopting sophisticated technology, underscoring their establishments, mainly focused on energy consumption potential effects on environmental-related outcomes besides only. There are different green management practices at the firm performance. firm level, such as setting green strategic objectives, including a manager with a mandate to deal with environment or climate Because energy consumption and waste treatment are part change issues, monitoring and setting targets for consumption of the firm’s costs, green management practices can be re- of energy or greenhouse gas emission, and adopting measures garded as an extension of general management focused to enhance EE. Overall adoption is low among Georgian plants, on improving input choices and the firm’s environmental despite the several areas where green management practices impact. Good management practices primarily seek to improve can focus. Even though nearly half of the surveyed firms report firm performance through monitoring and organizing produc- monitoring energy consumption, only a few adopt other green tion processes and improving working conditions. Similarly, practices, revealing a relatively passive behavior for being more green management practices can boost performance because environmentally friendly. For example, only 12 percent of firms energy inputs and waste management may affect efficiency and, adopt measures to enhance EE, albeit 44 percent monitor con- thus, productivity. Even without considering the negative exter- sumption and one-tenth have green objectives or hire a man- nalities of direct emissions, environmentally friendly practices ager responsible for green issues. may also be part of the business strategy for improving brand awareness and customer attraction. Reducing GHG emissions Georgian establishments are less likely to adopt green and increasing the use of less-polluting inputs, either directly or management practices when compared to those in ECA indirectly, may affect sales given the increasing pressure from countries, consistent with earlier findings about low tech- customers -firms and consumers- to firms for meeting environ- nology adoption. Table 17 reports the fraction of firms carry- mental standards for doing business (EBRD, 2019). Also, in line ing out green management practices in Georgia and selected with the rising awareness of environmental challenges and the benchmark ECA countries. Among comparable peers, Georgia foreseen consequences of climate change, firms are increas- ranks at the bottom-3 of the countries across the different di- ingly looking to reduce the environmental impact by adopting mensions of green management, such as strategy, responsi- greener technologies, reducing pollution and using resources bility, monitoring and targeting. For example, only 10 percent more efficiently. In this regard, managerial practices focused on of firms have strategic objectives related to the environment, greening firms, cutting GHG emissions, and optimizing the use the lowest among all countries except Azerbaijan. Also, in most of energy-related resources can contribute to this aim. countries, establishments are more likely to have a manager responsible for environmental issues and monitoring and tar- Good quality green management is rare among Georgian geting energy consumption. TABLE 17 Share of Establishments Adopting Green Management Practices in Georgia and ECA Countries Establishment has: GEO MKD SVN HUN CZE BGR POL SRB AZE Strategic objectives related to 0.10 0.30 0.29 0.28 0.26 0.23 0.19 0.17 0.09 the environment Manager responsible for 0.07 0.19 0.13 0.14 0.22 0.12 0.15 0.17 0.04 environmental issues Monitors Energy consumption 0.44 0.75 0.81 0.54 0.85 0.34 0.37 0.74 0.61 and emissions ↓ Water usage* 0.37 0.59 0.69 0.46 0.77 0.26 0.24 0.59 0.36 CO2 emissions 0.05 0.04 0.08 0.07 0.10 0.04 0.06 0.07 0.00 CO2 emissions along its supply chain 0.04 0.06 0.06 0.02 0.02 0.08 0.04 0.03 0.01 Emissions of pollutants other than CO2 0.03 0.02 0.08 0.05 0.15 0.07 0.05 0.06 0.00 Targets for Energy consumption 0.12 0.42 0.35 0.40 0.40 0.30 0.17 0.48 0.08 and emissions ↓ CO 2 emissions 0.02 0.06 0.07 0.06 0.06 0.08 0.05 0.06 0.00 Pollution emissions other than CO 2 0.02 0.03 0.16 0.06 0.13 0.13 0.05 0.06 0.00 Notes: a. manufacturing establishments only; Each variable adopts values 0 or 1; Environment includes climate change issues. Column names refer to country codes: GEO: Georgia; MKD: North Macedonia; SVN: Slovenia; HUN: Hungary; CZE: Czech Republic; BGR: Bulgaria; POL: Poland; SRB: Serbia; AZE: Azerbaijan. Source: World Bank elaboration based on WBES. 79 Chapter 5 Greening Firms in Georgia FIGURE 37 Differences in Average Green Management Compared to ECA countries, green management quality is systematically lower among Georgian establishments, z-Score Relative to ECA Countries letting public and private initiatives focus on managerial actions to enhance EE. Since some of these differences may re- Basic Controls sult from different economic structures and firm attributes, we look at overall green management quality, controlling for poten- tial factors underlying such results. Figure 37 depicts the abso- lute difference in average green management z-scores between Georgia and ECA countries, defined as the baseline category by the vertical red line. First, we compute the average difference controlling for firm size and industry—‘basic controls’—and All controls find that the average z-score of green management in Georgia is 0.43 lower than in ECA countries, equal to one-quarter of a standard deviation. However, since establishments may differ in their productivity, overall management quality and other relevant attributes, we include the management z-score (a proxy for overall management quality) and sales per worker as covari- ates in addition to foreign, exporting, and legal (domestic- or -0.6 -0.4 -0.2 -0.0 0.1 foreign-owned) status and age class controls—‘all controls.’ The quality of green management across Georgian establishments Average green management z-score in remains 0.23 lower than ECA peers, equal to one-seventh of a Georgia relative to benchmark countries standard deviation of the green management z-score. So, there is ample space for public and private organizations to enhance their green management quality and various actions that can Notes: ‘Basic controls’ include firm size class and industry. ‘All controls’ include sales per worker, the management be undertaken. z-score, legal status, foreign status (domestically- or foreign-owned company), exporting status, and age class, in addition to basic controls. The outcome variable is green management z-score at the firm level and the right-hand side variable is a dummy that equals 1 if the establishment is in Georgia and 0 if located either in Albania, Azerbaijan, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Kyrgyz Republic, Lithuania, North Macedonia, Poland, Serbia, Slovenia, Türkiye, and Ukraine. Source: World Bank elaboration based on WBES. B Understanding Green Management Practices in Georgia BOX 12 Measuring green management practices To measure the quality of green management practices, this Did this establishment have a manager responsible for envi- chapter benefits from the 2019 round of the Enterprise Sur- ronmental and climate change issues? veys conducted by the World Bank, which includes a special module on the green economy. The module collects estab- Green monitoring lishment-level records on environmental, energy use, and The survey asks the establishment questions about moni- pollution-related practices undertaken by the establishment. toring (4), frequency of monitoring (4), and completion of Also, establishments are asked questions about green strate- external audits (4) of its… gic objectives; responsibilities for environmental and climate change issues; monitoring of energy and water consumption, 1. Energy consumption? and of greenhouse gases and other pollutant emissions; tar- gets related to energy consumption and emissions. The ques- 2. Water usage? (manufacturing firms only) tions considered for constructing the green management z-score are: 3. CO2 emissions? Green strategic objectives 4. Emissions of pollutants other than CO2? Did this firm have strategic objectives that mention environ- mental or climate change issues? Green targets Over the last three years, did the establishment monitor Responsibilities of environmental have targets for… and climate change issues 1. Energy consumption? 80 Chapter 5 Technical Report BOX 12 Measuring green management practices (cont.) 2. CO2 emissions? responsibilities, monitoring, targets), and the overall green management z-score is calculated as the unweighted average 3. Pollution emissions other than CO2? score of the four types of practices. For cross-country comparison, the approach undertaken in Results throughout this section are robust to alternative this chapter follows the Transition Report 2019-2020 (EBRD, green management scores. We run the same analysis con- 2020), which measures the green management practices structing an alternative metric, an energy-centric green and green investment in nearly 40 countries of Central and management z-score following Grover & Karplus (2020). Eastern Europe, Baltic countries, and the Caucasus. The authors emphasize energy consumption and efficiency, focusing on targets for energy consumption, whether the The green management z-score summarizes firm-level re- manager responsible for green-related issues is evaluated or cords about environmental and green practices. The score not, the adoption of measures to enhance EE, and the pro- for each question is normalized so each has zero mean and a cess undertaken by the establishment to develop such mea- standard deviation of one (z-score). After variable normaliza- sures. Despite a different approach, the findings are consis- tion, z-scores are aggregated into four categories (strategy, tent with those presented in this section. There is substantial variation in green management qual- help explain the variation in green management scores. ity, suggesting that a small group of firms are concerned As energy intensity and pollution vary across economic activi- with the environmental and cost-saving benefits of re- ties, the firm’s sector could be a relevant factor in determining ducing energy consumption. Figure 38 plots the distribution the firm’s willingness to adopt green management practices. It of the overall and green management z-scores to understand also could reflect environmental regulations into force at the how managerial practices vary across firms. As depicted, a national or local level. Following the EBRD (2020), economic large mass of firms (around 60 percent of surveyed companies) activities are classified according to their level of emission in- scores lower than average (set to zero) in green management tensity. Those with a lower-than-median emission intensity are quality. In contrast, only a few (long thin tail on the right-hand classified as “clean” activities, whereas those above are “emis- side of the graph) adopts several, complementary practices, sion-intensive”. Figure 39 shows that emission-intensive activi- suggesting that a small share of firms are concerned about ties display higher quality of green management than clean ac- environmental sustainability issues or can adopt green man- tivities. Although clean and emission-intensive establishments agement within the establishment. The variation in green tend to monitor energy consumption, water usage, CO2, and management adoption is considerable compared to general other pollutant emissions similarly, differences arise when it management practices. Also, green management quality is less comes to complementary practices such as strategic planning, evenly distributed than overall managerial practices. In this re- hierarchical responsibility structure, and setting consumption gard, there is a wide margin to encourage the adoption of green and pollution-emission targets. For example, emission-inten- management practices among Georgian firms. For some busi- sive industries are more likely than clean industries to employ nesses, this requires introducing the importance, scope, and managers responsible for green issues, set green objectives, instruments related to green management, while for others, and implement targets for energy consumption, water usage, policy goals should be set at expanding green practices (e.g., and CO2 and other emissions (Figure 39). Nevertheless, part including objectives and responsibilities in the formal struc- of this higher take-up could be driven by emission-intensive ture of the organization, monitoring and targeting greenhouse activities being more prone to generate pollution and regu- gas emissions, and improving EE through active measures). latory bound. Regulations and customer pressures for more sustainable production processes can push green management Sectoral differences in green management adoption could adoption among higher GHG-intensive firms. 81 Chapter 5 Greening Firms in Georgia FIGURE 38 Distribution of Green Management A second set of underlying factors that could explain green management quality across firms are firm size, age, Practices in Georgian Firms ownership, and exporting status. As firms grow and become larger, they might be more visible to third-party auditors and Density have more incentives (or be obliged) to comply with national 0.5 laws and regulations. Young firms may invest in less-pollut- ing technology and adopt practices to make the firm greener 0.45 because they face demand pressures for environmental cer- 0.4 tifications or due to self-awareness on sustainability issues. Results from the WBES, which show that large and older firms 0.35 tend to have higher quality green management than small and young firms (Figure 40A), are consistent with this explanation, 0.3 although the available information is insufficient to confirm 0.25 these hypotheses. 0.2 General Management Similarly, exporters may face pressure from foreign cus- 0.15 tomers to use low-pollution inputs or meet environmen- tal regulations, having to adapt products and services to 0.1 international standards. Also, foreign-owned firms could be more prone to introduce green management practices since 0.05 Green Management they adopt management practices from headquarters, absorb 0 technology according to green standards and are more likely to be scrutinized by public agencies. Figure 40B shows that -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 firms with a significant fraction of shares owned by foreign z-score shareholders have better green management practices than domestically owned companies. In the same vein, exporters Source: World Bank elaboration based on EBRD (2020) and WBES. outperform non-exporters in the green management score. FIGURE 39 Type of Green Management Practices by Sector Domestically owned 25% or more foreign-owned Overall Green Green strategy Green management Green targets Green monitoring Management responsibilities 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 Source: World Bank elaboration based on EBRD (2020) and WBES. 82 Chapter 5 Technical Report Because of rising energy costs and increasing awareness management quality and productivity in Georgia. Figure of the environmental impact of economic activity, less 41, Panel A shows that the higher the quality of general man- is known about the link between management practic- agement, the higher the quality of green management. In this es, green management, and energy use. Grover & Karplus regard, management practices can be a first step to promoting (2020) explore the relationship between firms’ management green management among firms. Businesses may need to build practices and energy consumption and find that management on existing capabilities to perform more sophisticated or com- actions are related to cost reduction but not necessarily to the plex practices that could be perceived as less impactful on their firm’s environmental impact. Remarkably, the quality of overall daily performance. Hence, extending managerial practices management practices is positively associated with a greater dis- to green issues could therefore come after structuring basic cipline of energy management, suggesting that general manage- tasks. Only after achieving this would firms understand the ment matters for improving the quality of green management. importance of addressing green issues. Additionally, green Furthermore, the authors point out that specific practices, like management is positively and strongly associated with val- target-setting, are more correlated to lower energy intensity. ue added per worker, similar to the results found in Grover & Karplus (2020). The evidence reported in Figure 41, Panel B is Green management is positively correlated with overall consistent with the findings reported in Table 14. FIGURE 40 Green Management Practices across Size and Age Classes, Ownership, and Exporting Status PANEL A Size and age class PANEL B Ownership and exporting status 1.4 Average quality of green management (z-score) 0.6 Average quality of green management (z-score) 1.2 Large 0.5 >25% 1.0 young shares 0.4 0.8 firms foreign-o 0.3 wned Exporter 0.6 0.2 0.4 Old 0.1 0.2 SME 0.0 0.0 Note: Young firms are defined Large Large as firms aged less than 6; -0.2 old old -0.1 Domestically SMEs are firms with no more firms firms owned Non- than 100 employees. Source: -0.4 Young -0.2 SME exporter World Bank elaboration based on EBRD (2020) and WBES. 83 Chapter 5 Greening Firms in Georgia FIGURE 41 The Relationship between Green Management, Overall management, and Labor Productivity PANEL A Green management PANEL B Green management and general management and labor productivity 2.0 General management z-score 11.5 Log value added per worker 1.5 11.0 1.0 10.5 0.5 10.0 0.0 -0.5 9.5 -1.0 9.0 -1.5 8.5 -2.0 8.0 -3 -2 -1 0 1 2 3 4 5 6 7 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Green management z-score Green management z-score Notes: Binned scatterplot (25 bins). Estimated equations for (A) and (B) are: (A) zmgmti =-0.034 + 0.114*** × zgreen mgmti; (B) labprodi=9.708*** + 0.137***×zgreen mgmti. Both specifications control for size and age class, industry, and region FE. Value added per worker is sales net of raw costs divided by the number of permanent full-time employees. Source: World Bank elaboration based on EBRD (2020) and WBES. FIGURE 42 The Relationship between Energy Efficiency and Green Management Practices -4.0 Energy e ciency -4.5 -5.0 -5.5 -6.0 -6.5 -7.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 Green management z-score Notes: Estimated equation is Yi=-5.677***-0.059×z-green mgmti. Both specifications control for size and age classes, industry, and region FE; number of bins is 25. Source: World Bank elaboration based on WBES. Data suggests no relationship exists between EE and green as a low-impact strategy for reducing costs and improving effi- management in Georgia. The adoption of green management ciency. Also, it can affect EE because distorted energy prices can seems uncorrelated with EE, according to Figure 42.72 One of the lead to overconsumption. For example, Schweiger & Stepanov 72 Conclusions hold main factors limiting the impact (or hindering the adoption) of (2022) find that the (inverse) relationship between management even after separating regression analysis between green management could be the fuel subsidies spent by the gov- practices and energy intensity weakens in countries with high manufacturing and service ernment. Low electricity prices and high-energy subsidies may fuel subsidies. Energy subsidies drive prices down and thus sectors. discourage environmental practices because firms can see them affect the relationship between fuel intensity and management. 84 Chapter 5 Technical Report Besides external factors (e.g., prices) that could influence Large firms show higher quality of green management EE and green management practices, other elements may also practices, regardless of age. High-quality management prac- affect green behavior. A relevant (and open) question is what tices are associated with firm size. Large firms (at least 100 em- drives green management practices across organizations. Due ployees) display higher green management z-scores than SMEs to the rising awareness of environmental protection among (0-99). These findings are consistent with EBRD (2020) research citizens and firms and the recent hikes in energy prices, the results covering a broader set of ECA countries. One plausible answer to this question is far more critical for policymakers explanation is that firms are more prone to adopt management around the world, who pursue EE and environmentally friendly and green management practices as they grow and develop activities for a variety of reasons like achieving sound (and new or better capabilities. Alternatively, firms become more sustainable) economic growth, keeping public and external visible and are forced to comply with regulations as they reach accounts balanced, and preserving public health. Green man- a size threshold. For example, public agencies could heavily agement can have positive externalities beyond enhancing scrutinize large firms regarding GHG emissions and pollution firm-level performance. Table 18 shows the results of regressing and oblige them to monitor emissions and set targets accord- the green management z-score (an indicator of green manage- ing to national environmental regulations. This is consistent ment quality) on the general management z-score (i.e., quality with the OECD (2018) report and the results from the United of overall managerial practices), firm characteristics (e.g., size, Nations Industrial Development Organization (UNIDO) sur- age, export, foreign and legal status). Each specification also vey conducted in Georgia (Chorgolashvili, 2017). Respondents includes industry and geographic fixed effects, whether the stated that stricter regulations influence their green technology firm is financially audited, and if it operates in an emission-in- investment decisions. Last but not least, customers’ visibility, tensive industry. Column (1) reports the results for the baseline especially relevant for large firms, could be driving these results. specification and columns (2) and (3) add external factors such Besides size, a remarkable finding is that young firms do not as customer pressure to meet environmental standards, envi- systematically have higher green management quality than ronmental regulations, and monetary losses due to extreme established plants. Results show no relationship between green weather conditions or pollution caused by others. management quality and age. Still, the sample size limits the statistical power of these results. TABLE 18 Determinants of the Quality of Firms’ Green Management Practices OLS (1) OLS (2) OLS (3) General management z-score 0.336*** 0.319*** (0.099) (0.107) Large firm (at least 100 employees) 1.199*** 0.999*** 0.865*** (0.245) (0.291) (0.285) Mature firm (aged 6+) 0.040 0.006 0.007 (0.097) (0.180) (0.176) Foreign investors own >25 percent shares 0.186 0.213 0.215 (0.238) (0.314) (0.293) Firm exports 0.559*** 0.658** 0.562* (0.188) (0.323) (0.332) Sole proprietorship 0.076 0.169 0.142 (0.052) (0.135) (0.133) Financial reports audited (Yes = 1) 0.360*** 0.193 0.179 (0.099) (0.107) Customer pressure (Yes = 1) 1.361** (0.649) Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Left-hand side variable is green management z-score. Source: World Bank calculations based on WBES. 85 Chapter 5 Greening Firms in Georgia TABLE 18 Determinants of the Quality of Firms’ Green Management Practices OLS (1) OLS (2) OLS (3) Monetary losses due to extreme weather 0.177 (Yes = 1) (0.278) Monetary losses due to pollution caused by others (Yes = 1) 1.497*** (0.471) Energy tax/levy -0.123 (0.317) Emission-intensive 0.520 0.613 0.610 (0.363) (0.560) (0.554) Other sectors -0.060 -0.399* -0.261 (0.144) (0.228) (0.235) Constant -0.333* -0.174 -0.062 (0.185) (0.291) (0.342) Industry FE Yes Yes Yes Geographic FE Yes Yes Yes R-squared 0.202 0.231 0.276 Observations 566 268 268 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Left-hand side variable is green management z-score. Source: World Bank calculations based on WBES. The quality of general management practices is key for display higher green management quality, suggesting that green management. The general management z-score is domestic and foreign demand can be an important driver for positively and strongly correlated with the quality of green greening firms. Alternatively, meeting environmental stan- management. An improvement of the overall management dards could unlock new markets for Georgian firms, so encour- score of one standard deviation (sd = 1) is associated with an aging green practices could improve their performance even average increase of one-fifth of a standard deviation (sd =1.39) when some establishments do not report customer pressures. in the green management score. For the average firm at the Second, there is evidence that monetary losses due to pollution fifth decile of the green management score distribution, this or climate change effects could influence green management would imply moving to the sixth decile. adoption. Albeit small in number, firms that report losses due to pollution have higher quality green management relative External factors also determine green management adop- to firms that do not report such losses. Similarly, the estimate tion. The green management analysis also considers the influ- associated with losses due to extreme weather is also positive 73 It measures the quality of ence of external factors such as customer pressure, negative but not statistically significant. In contrast, energy taxes are not management practices related to resolution of production consequences of extreme weather or pollution, and energy correlated with green management. Unfortunately, with the problems, monitoring and taxes for understanding within-firm green practices (column information available, we cannot dive deeper into the reasons targeting performance, and 3). Firms whose customers require environmental certifications underlying such results. rewarding performance. or adherence to specific green standards for doing business 86 Chapter 6 Technical Report Chapter 6 86—95 PAGE Assessing the Role of Prices in Energy Consumption T his section examines the channels of firms’ ad- efficiency and be more innovative, even though this is not the justments to changes in energy costs, particular- most typical response. Grubb et al. (2021), Hassler et al. (2021), ly electricity and gas prices, exploiting a natural Popp et al. (2010) and Popp (2002) document an increase in the experiment due to a large and rather unexpected number of EE innovations due to price shocks. In response to increase in energy prices on January 1, 2021. higher energy prices, firms can also switch industries, change product portfolios (Abeberese, 2017) or reallocate production Energy prices are crucial for consumers and sup- between plants. For example, Elliot et al. (2019) find that busi- pliers to allocate resources efficiently. On the one nesses are more likely to switch the industry of their main hand, distorted prices, typically artificially low energy prices product to less energy-intensive industries to reduce their due to subsidies, can lead to overconsumption and waste, which energy dependency when energy prices surge. Fontagne et al. not only affect the level of consumption and GHG emissions but (2023) show that energy demand and production increase in also may demand more investments in energy infrastructure to establishments with lower energy prices. However, the extent cope with higher demand. For example, excess consumption to which plants can respond in these ways depends on the can put more strain on the power grid, increase the number of costs and expected revenue of such shifts. In the same vein, power outages and reduce the security of the energy system. some firms respond by outsourcing production, substituting In certain contexts, this could lead to an increase in fossil fuel in-house production of energy-intensive inputs for foreign dependency. But cheap energy prices discourage investments ones, particularly from markets with regulated energy prices in greener technology and renewable energy sources, such as (Fontagne et al., 2023; Rentschler & Kornejew, 2017). Although energy-saving machinery and equipment, on-site electricity more difficult to assess, if firms could not substitute product generation, insulation of buildings, and improving lighting, inputs and were constrained by market competition, they could cooling, heating and ventilation systems. Under lower energy respond by downgrading product quality as well. prices, green investment returns are lower, steering funds to other projects. Finally, subsidizing energy could also distort Last but not least, firms may adjust production factors, business decisions, which would not be optimal under efficient including energy and output. Energy consumption appears prices. At some point, businesses decide which sector entry, to be the most direct way businesses can adjust to an energy minimum profits, production factor intensity, the product mix price shock -even when they do not enhance efficiency-, either and what goods and services are produced in-house based on by reducing the overall energy consumption or by switching energy prices. These decisions would be considerably different energy sources. Recent studies that leverage firm-level data in the case energy subsidies were removed. from Germany, France and Italy estimate the price elasticity of electricity and gas. Rottner & von Graevenitz (2022) estimate According to the empirical evidence, firms may adapt in the average elasticity of electricity usage in manufacturing multiple ways, although some responses might be stron- firms in Germany between -0.4 and -0.6 in the short term. How- ger than others. For example, Fotagne et al. (2023) and Jous- ever, longer-term elasticity is much smaller and non-signifi- sier et al. (2023) show that during the recent energy price shock cant, suggesting only temporary responses of manufacturing (2021–22), French firms passed on energy costs to production plants to changes in electricity prices. Evidence shows that the prices. Price adjustments are also documented by Ganapati et energy elasticity can vary over time and that, in fact, could be al. (2020) and Sadath & Acharya (2015). Firms can also enhance declining in some countries. In Germany, a 1 percent increase 87 Chapter 6 Greening Firms in Georgia in network charges was associated with a reduction in energy firms in 1992-2015, they estimate an average electricity elastic- consumption of about 0.7-0.9 in 2010-2011 but to only 0.3-0.5 ity of -0.7 (-0.8 in energy-intensive trade-exposed industries). in 2016–17. Fontagne et al. (2023) estimate the energy elasticity Such responses can also be country specific. For example, Al- of French manufacturing firms over the period 1996-2019 and pino et al. (2023) find that for a sample of medium and large find that businesses adjust energy consumption strongly and firms in Italy (at least 50 employees), the electricity and gas rapidly to higher energy prices. The estimated demand elas- elasticities during the recent energy crisis were small (-0.2) and ticity is around -0.4 for electricity and -0.9 for gas. As Rottner not statistically different from zero. However, they document & von Graevenitz (2022), the authors also report that electricity that for plants subject to the European Emissions Trading Sys- elasticity decreases with time, even for large price hikes. An tem (EU ETS), the gas elasticity is much larger (around -0.8). interpretation for this is that firms have mostly adapted to price Differences in estimated elasticities could be explained by the shocks in the past and now have less space for adjustments. fact that the gas intensity of EU ETS firms is much higher than These results are aligned with those reported by Wolverton, non-EU ETS firms and that natural gas price changes in 2021 Shadbegian & Wayne (2022). For a sample of US manufacturing were larger for the former group of enterprises. T CHAPTER 6.1 he analysis relies on firm-level data on energy Electricity charges in Georgia expenses and consumption as described in Box tend to increase smoothly. The Data 4 and Box 5. GEOSTAT provides information on elec- tricity and gas quantity consumption for a subset of Table 19 reports basic descriptive statistics on electricity around 2,500 firms per year for 2013-2021 (around and gas prices, the average dependency on electricity and 20,000 firms). We use reported electricity and gas con- gas, the number of firms and the average firm size in 2013- sumption both in value and quantity (kWh and m3) to 2021. Only in 2021, electricity prices rose more than 34 percent, estimate prices paid, calculated as electricity (gas) expenses the largest increase registered since. To put it in perspective, divided by electricity (gas) quantity consumption expressed the electricity price change between 2013-2020 (31.6 percent) GELs per 100 kWh (m3). We estimate the average charge paid was smaller than the 2020-2021 change. Despite the electricity by the firm as we do not have plant-level data, so we cannot price adjustment, gas charges did not change substantially account for energy price variation across establishments. We over 2013-2021: they rose by 12.2 percent in 2013-2020 and 1.7 complement energy price and consumption information with percent in 2020-2021. relevant firms’ characteristics such as the region of location, industry of operations, output, employment and efficiency time-variant metrics. TABLE 19 Determinants of the Quality of Firms’ Green Management Practices Year Elec. price Gas price Elec.dep. Gas dep. Firm Employ- (GEL per 100 kWh) (GEL per 100 m3) ment Mean Median Mean Median Mean Mean Count Mean 2013 14.38 13.67 68.30 65.05 2.19 1.15 1,769 53.54 2014 13.94 13.05 67.13 65.01 2.16 1.24 1,993 48.75 2015 15.33 16.00 68.69 67.20 2.15 1.22 2,290 44.70 2016 16.42 16.77 68.36 65.53 2.29 0.98 2,251 46.55 2017 17.41 17.53 71.30 69.89 2.09 1.18 1,119 73.92 2018 18.56 18.57 74.92 73.53 2.65 1.28 2,508 43.56 2019 18.64 18.13 76.48 75.00 2.52 1.00 2,689 41.14 2020 18.92 18.87 76.66 75.03 2.38 1.12 2,753 41.39 2021 25.44 25.77 77.93 75.00 3.55 1.05 2,644 43.44 Notes: Electricity and gas dependency is the electricity (gas) Source: World Bank calculations based on GEOSTAT. expenses-to-sales rescaled to 100. 88 Chapter 6 Technical Report FIGURE 43 Non-Household Energy Tariffs Adjusted for Electricity Only (2021) PANEL A Gas PANEL B Electricity .2 Fraction of firms .1 Fraction of firms .18 .09 .16 .08 .14 .07 .12 .06 .1 .05 .08 .04 .06 .03 .04 .02 .02 .01 0 0 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6 .7 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6 .7 Source: World Bank calculations based on Price change (%) in 2021 Price change (%) in 2021 GEOSTAT. In 2021, electricity prices rose in a context of surging electricity consumption and price changes. Here, a negative global energy prices and a centralized decision to adjust relation would suggest that the error measure decrease with electricity tariffs for regulated (small) consumers. On the the electricity consumption scale because fixed charges would one hand, the rise of global energy prices due to the COVID-19 account for a lower share of total electricity expenses. recovery and the geopolitical tensions between Russia and Ukraine before the beginning of the invasion is likely to have We explore whether price changes are correlated with en- pressured local energy prices upward in line with most EU ergy consumption and other production scale metrics. We countries.74 On the other, the price adjustment applied to reg- regress the firm-level first difference of the logged electricity ulated non-household consumers appears to be a centralized prices on electricity consumption, deflated sales and employ- decision according to the market functioning and the mag- ment (all in logs). In columns 1-3 of Table 20, we use no addi- nitude of charge adjustments. New, adjusted prices for 2021 tional controls, columns 4-6 include industry and geographic were announced within the first days of January 2021. This FE plus year effects, and columns 7-9 add firm FE. Since firms would rule out the possibility that the price paid results from may relocate due to geographic variations in electricity costs, the interaction of demand and supply for small electricity con- we allow the region of the firm to vary over time.75 By including sumers. In contrast, for large consumers, electricity charges firm FE controls, the specification controls for time-invariant are negotiated under a bilateral agreement with the supplier. factors at the firm level that could influence electricity price changes and be correlated with our variables of interest. For 74 According to Eurostat, Even when electricity price changes could be considered example, firm FE can absorb the overall bargaining power of the the kilowatt-hour (kWh) net-of-tax electricity price exogenous to the firm for most businesses, two concerns firm and other fixed attributes that may be relevant at the time for medium-sized industrial could go against the exogeneity assumption: measure- of contract negotiation. We acknowledge that the bargaining consumers increased by 90 ment error and large consumers’ bargaining power when power can vary over time, but we consider that this can deal percent for the average EU negotiating the contracts. Electricity (gas) tariffs include well with the endogeneity problem. country between the first fixed and variable components, which affect the estimation of semester of 2019 and 2022 prices. Second, among large non-household consumers, elec- Firm-level price changes are negatively correlated with while the equivalent natural tricity (gas) prices are negotiated between the customer and the electricity consumption but not with other measures of gas price per gigajoule (GJ) rose by 108 percent over the supplier, so their bargaining power is likely to affect observed production scale. Table 20 reports regression outcomes. In same period. prices. Although these two factors can occur simultaneously, if the baseline specification (no firm, industry, region and time the bargaining power affects observed prices, we should be able controls), pooled regression results show no statistical cor- 75 As noted in previous to see a correlation between price changes and different pro- relation between the firm scale metrics and electricity price analysis of this report, we duction scale metrics, namely electricity consumption, sales changes (columns 1–3). However, when industry, region and computed both the region and employment. In this regard, we assume that businesses time controls are included, there is a negative relation between and municipality where the firm located most of the time with larger output and employment can negotiate more conve- price changes and consumption but no relationship with sales and the area of location in nient contracts. However, if only the measurement error holds, or employment. The result between energy consumption and each year. we would expect a (presumably negative) correlation between price changes could be driven by either a measurement error 89 Chapter 6 Greening Firms in Georgia TABLE 20 Associations between the Magnitude of Electricity Price Changes, Energy Consumption, and Other Firm Scale Metrics Dep. var.: ∆Elec. OLS (1) OLS (2) OLS (3) OLS (4) OLS (5) OLS (6) FE (7) FE (8) FE (9) price Elect. cons. -0.001 -0.003*** -0.030*** (0.001) (0.001) (0.003) Sales 0.001 0.001 -0.003 (0.001) (0.001) (0.003) Employment -0.000 0.002 0.003 (0.001) (0.001) (0.005) Firm FE No No No No No No Yes Yes Yes Industry FE No No No Yes Yes Yes No No No Region FE No No No Yes Yes Yes Yes Yes Yes Year effects No No No Yes Yes Yes Yes Yes Yes Observations 9,238 8,984 9,238 9,238 8,984 9,238 9,238 8,984 9,238 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance at (10), (5) and (1) percent level. The left-hand side variable is the first difference of (logged) electricity prices (GEL units per 100 kWh) at the firm level; electricity consumption, sales and employment are in logs. Industry FE control for 3-digit sector shocks. Source: World Bank calculations based on GEOSTAT. or the firm’s bargaining power (or both), so we include fixed effects at the firm level. We expect that under the bargaining power hypothesis, the relationship between price shifts and consumption under time-invariant firm-level effects would decline or even disappear because even though the bargaining power can vary, it is likely to be part of a permanent character- istic of the firm. Under the measurement cost hypothesis, the relationship would still hold since it is a problem related direct- ly to the amount of electricity consumed. After including firm FE, we still observe a negative relation between consumption and price changes. At the same time, the signs of the estimat- ed coefficients for sales and employment are mixed but still non-significant. So, we find no direct evidence to support the bargaining power hypothesis but more space to think of the measurement error problem. The 2021 electricity price adjustment was significantly higher than earlier adjustments over 2014–20. The average electricity price change was around 35 percent after consid- ering energy consumption, time-invariant factors at the firm level, and region and industry FE. Figure 44 plots the average year-on-year (YoY) change in electricity charges resulting from a different set of controls. The average firm-level price change does not vary significantly with the controls used, suggesting that price adjustments are mostly exogenous to the firm. 90 Chapter 6 Technical Report FIGURE 44 Changes in Electricity Prices Over Time (Relative to 2014 Changes) .4 Electricity price change .35 Elec. cons. + firm and region FE Elec. cons. + region and industry FE .3 Elec. cons. control No controls .25 .2 .15 .1 .05 0 Source: World Bank calculations based on 2015 2016 2017 2018 2019 2020 2021 GEOSTAT. CHAPTER 6.2 Identification Strategy T he baseline identification strategy exploits shocks could influence the estimated prices when there is a fixed within-firm price variation to evaluate the dif- charge for electricity supply. We adopt an instrumental variable ferent margins of adjustments of price shocks. (IV) approach to address the endogeneity concerns following The basic specification is expressed in equation (9): Carli et al. (2022) and Fotagne, Martin & Orefice (2023). Where . We also allow the average Where yisrt is the outcome of interest of firm i operating in sec- price, , to vary across regions by calculating and . 0 tor s and in region r in year t, is the logged electricity price This approach includes electricity charges of nearby firms in and Ωisrt a vector of firm-level time-varying covariates. All the spirit of Cali et al. (2022). The authors instrument energy specifications control for firm-level time-invariant unobserv- price changes using prices of neighboring plants in Indonesia able covariates ( ), spatial shocks ( ), industry-year effects or the average charge of plants from the same sector and state ( ), such as industry-specific technological, business cycle in Mexico, excluding the firm in each case. We combine the and energy-related shocks, and economy-wide business cycle approaches in Fotagne, Martin & Orefice (2023) and Cali et al. and technological shocks ( ). (2022). However, we include an alternative instrument more in line with Cali et al. (2022), leaving out the initial bargaining Based on the previous discussion, price changes are likely power of the firm (eq. 11). If the measurement error is the en- to have a strong exogenous component due to a global sup- dogeneity source and the initial bargaining power is irrelevant, ply shock and the centralized decision of the Georgian gov- instrumenting price by calculating the industry-region price ernment to align electricity charges to marginal costs for average could make the measurement error uncorrelated with most non-residential consumers. Still, firm-level time-vary- firm characteristics. We check the two approaches’ validity and ing unobservable shocks could affect negotiated electricity test the results’ robustness. charges. Fotagne et al. (2023) relate those shocks to expected demand shocks that may help negotiate contracts differently. Moreover, firm-level shocks like temporary blackouts or demand 91 Chapter 6 Greening Firms in Georgia T able 21 presents the results of regressing the elec- to changes in production requirements rather than price shifts. CHAPTER 6.3 tricity consumption on electricity and gas prices. Columns 1-3 include firm-level electricity and gas prices, es- Preliminary Every specification controls for firm FE, industry-year FE and time effects, so the results should be interpret- ed as changes in within-firm electricity consumption timated by dividing energy expenses and consumption. Col- umns 4-6 use instrumented prices. Columns 4 and 5 instru- ment electricity prices as in (eq. 10), being the main difference Results due to within-firm price variation. Also, columns 2-6 include scale and efficiency time-variant metrics to account for the fact that electricity consumption may respond that (4) is calculated using 3-digit sector average prices and in (5) is calculated using 2-digit sector-by-region average prices. Finally, column 6 instruments using (eq. 11). TABLE 21 The Effects of an Energy Price Shock on Electricity Consumption Dep. var.: OLS (1) OLS (2) OLS (3) 2SLS (4) 2SLS (5) 2SLS (6) Elec. cons. (log) Elec. price (log) -1.036*** -1.128*** -1.341*** -1.130*** -1.231*** -0.913** (0.077) (0.076) (0.106) (0.081) (0.308) (0.394) Elec. price x elec. dep. 0.044*** (0.005) Gas price (log) 0.156 (0.109) Gas price x gas dep. 0.154 (0.157) Sales (log) 0.491*** 0.594*** 0.492*** 0.490*** 0.496*** (0.018) (0.034) (0.019) (0.020) (0.019) Labor productivity (log) -0.068*** -0.029* -0.066*** -0.069*** -0.069*** (0.012) (0.016) (0.012) (0.013) (0.012) Firm FE Yes Yes Yes Yes Yes Yes Geographic FE Yes Yes Yes Yes Yes Yes Industry x Year FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes R-squared 0.876 0.902 0.934 0.132 0.131 0.133 Observations 15,464 12,254 4,240 11,307 10,252 11,364 1st-stage results IV coefficient 0.987*** 0.085*** 0.566*** (0.002) (0.004) (0.035) 1st stage R-squared 0.985 0.161 0.676 C-D Wald F 268,033 636 317 S-W S stat. 251.916 21.935 7.186 K-P Wald F-stat 191,289 512 259 A-R Wald test (p-value) 0.000 0.000 0.023 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Left-hand side variable is the (logged) electricity consumption (kWh); electricity dependency is the electricity-to-sales ratio. Geographic FE are at the region level; industry × year FE capture time-varying 2-digit industry effects. Source: World Bank calculations based on GEOSTAT. 92 Chapter 6 Technical Report The elasticity of the electricity demand is estimated be- these concerns by instrumenting firm-level prices with the tween -1.3 and -0.9, suggesting that firms respond to price average price of firms within their sector or their sector-region, incentives by adjusting the quantity of electricity con- excluding the analyzed firm in all cases. The exogeneity of our sumed. In all cases, the estimated coefficients are statistical- instrument is justified by the fact that the prices of other firms ly significant and around 1. When we include the interaction in the same sector or same sector-region should not influence between the electricity price and the electricity dependency of the decision of a specific firms except through the correlation the firm, calculated as the share of the electricity cost in total that these prices have with the price faced by the firm.76 revenues (scaled to 0-100), we observe that firms with a higher electricity dependency are associated with an elasticity closer Estimation results from the instrumental variable strat- to zero, suggesting that they are less responsive to energy price egy (columns 4-6) are similar to those from the OLS ap- shocks. A plausible interpretation of this result is that plants proach. The estimated elasticity variables are between -0.9 more dependent on electricity are less able to fuel switching and -1.3. If is constructed as in equation 10 (using sector and hence cannot respond by reducing electricity consumption 3-digit average prices, 0 and , the estimated electricity for a given amount of output. Also, we find some evidence that elasticity is -1.2. If the 2-digit sector-by-region average prices, this elasticity decreases with time in absolute terms, suggesting 0 and , are used, the estimated elasticity is -1.3. Finally, that firms could be becoming slightly less responsive to energy column 6 presents an estimated elasticity of -0.9 using the price shocks over time (Figure 46). 2-digit sector average price instrument, . Due to price endogeneity concerns and reverse causality, We acknowledge that firm responses to changes in elec- our identification strategy relies on an instrumental vari- tricity prices could be heterogeneous across sectors as able approach. Two factors can explain to some extent why industries have different technologies, energy dependen- estimated electricity (gas) price changes decrease with quan- cy and capacity to adjust consumption (e.g., save energy 76 Cali et al. (2022) tities consumed. First, electricity (gas) expenses include fixed without altering output). Figure 45 plots estimated electricity and Fotagne, Martin & and variable components, which affect the estimation of pric- elasticities for manufacturing and construction, the two sectors Orefice (2023) use a similar es, calculated as the electricity (gas) value consumption over for which we have enough observations. Figure 45 compares instrumental variable electricity (gas) quantity consumption. Second, non-household the OLS and 2SLS estimates, controlling for firm time-variant approach to deal with very similar endogeneity electricity (gas) prices are negotiated between the customer characteristics such as sales and labor productivity (value added concerns as explained in the and the supplier, so bargaining power is likely to affect ob- per worker), and adding geographic, industry-by-year, and time identification strategy section. served prices. Our instrumental variable strategy deals with effects, following the approach in columns 2 and 4 in Table 21). 93 Chapter 6 Greening Firms in Georgia FIGURE 45 Electricity Elasticity in the Manufacturing and Construction Industries OLS OLS 2SLS (column 2SLS 5) 5) (column 2SLS (column 2SLS 6) 6) (column PANEL A Manufacturing PANEL A Manufacturing PANEL B Construction PANEL B Construction P elecelec itP it -3.5-3.5 -3 -3 -2.5-2.5 -2 -2 -1.5 -1.5 -1 -1 -.5 -.5 0 0 .5 .5 -3.5-3.5 -3 -3 -2.5-2.5 -2 -2 -1.5 -1.5 -1 -1 -.5 -.5 0 0 .5 .5 Notes: Confidence intervals displayed at the 90 percent level of confidence. Source: World Bank elaboration based on GEOSTAT. We find that manufacturing and construction activities same region and 2-digit sector excluding the evaluated firm. respond to rising electricity prices by reducing electricity Again, we control for firm-level unobserved time-invariant consumption. The estimated elasticities are negative and characteristics to deal with endogeneity concerns, geographic statistically significant. In manufacturing firms, the average (regional) FE, industry (2-digit) specific shocks within each year elasticity is -1.45 in the basic setting but almost a half, -0.74, -industry × year FE- and economy-wide year shocks. under the instrumental variable strategy. In construction ac- tivities, the estimated elasticity is -1.01 in the OLS setting and Firms respond to higher electricity prices by improving doubles (-1.93) in the IV setting. Although estimated standard EE, underscoring the relevance of price incentives to op- errors increase substantially in the IV approach compared to timize energy use. Table 22 shows interesting conclusions the OLS setting, results are statistically significant at the 90 about how firms respond to electricity price shocks. Columns percent confidence level. 1 to 4 report the effect on energy and electricity efficiency. The estimated effect is positive and statistically significant for both Finally, we examine the effects of an increase in electricity aggregate and electricity efficiency. According to the results prices on different margin responses reported in Table 22. from the IV strategy -columns 2 and 4-, a 1 percent increase in Due to data availability constraints, we are limited to evaluating electricity prices is associated with a rise of 1.25 percent in EE firms’ responses on aggregate and electricity efficiency -the log and 1.47 percent in electricity efficiency. Also, OLS estimates of produced output per unit of TJ or kWh consumed-, employ- lead to similar conclusions. With higher prices businesses are ment, sales, costs -intermediate consumption plus wage bill, expected to make more efficient use of the input for which the profits -sales net of costs-, investments -the first difference of relative price increases. Since electricity accounts for at least the stock of fixed assets-, and the consumption of intermediate 20 percent of total energy quantity consumption, overall EE is goods. Outcome variables are expressed logs, so results should expected to increase as well. be interpreted as elasticities. Table 22 reports OLS and 2SLS regression results. In the latter case, electricity prices are in- Preliminary findings suggest that employment does strumented as the average electricity price of firms within the not respond significantly to rising electricity charges in 94 Chapter 6 Technical Report Georgia, a highly relevant conclusion for the energy policy no evidence of such effects.77 debate. Besides improving efficiency, firms do not appear to adjust employment while costs and sales increase, with the Firms seem to be responding by passing on the increase former being larger than the latter. As for employment effects, in prices to final consumers. We observe a positive increase we find positive but not statistically significant effects of ener- in overall costs and sales. For example, a 1 percent increase in 77 These results are gy price changes after controlling for output effects, both for energy prices increases costs by 2.48 percent and sales by 1.20 preliminary because our OLS and 2SLS estimates. In Georgia, this is a relevant finding percent. Although we cannot observe product price respons- data only cover 2021 (that because employment does not appear to respond significantly es, one interpretation of the positive effects on both costs and is, 12 months after the to increasing energy costs (or reducing energy subsidies). For sales is that firms pass higher energy costs on prices rather price increase). With 2022 example, during the 2021–22 energy crisis after the COVID-19 than they can increase production and therefore augment microdata, we will be able recovery and the Russian invasion of Ukraine, many EU policy intermediate consumption. Although we cannot rule out the to evaluate the impacts leaders advocated for business support policies in the form of latter explanation, the former hypothesis is consistent with of electricity and gas price caps and subsidies, claiming that thousands of jobs were a well-developed body of empirical research (Fotagne et al., adjustments over a longer horizon (24 months) and on the brink due to surging energy prices. However, according 2023; Ganapati et al., 2020; Joussier et al., 2023; Sadath & confirm or refute these to the data available and acknowledging we have only one year Acharya, 2015), which underscores the pass-through mecha- preliminary findings. in the sample after the January 2021 price adjustment, there is nism as one of the most relevant. FIGURE 46 Electricity Elasticity Over Time .25 0 -.25 -.5 -.75 -1 -1.25 -1.5 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: Confidence intervals displayed at the 90 percent level of confidence. Source: World Bank elaboration based on GEOSTAT. T his section assessed how firms respond to changes decisions. Nevertheless, we address potential endogeneity CHAPTER 6.4 in electricity prices. The analysis provides important concerns due to price measurement errors and the firms’ bar- Preliminary preliminary findings that help answer the question of gaining power when they negotiate contracts by instrumenting what works for improving EE in Georgia, especially re- firm-level prices with the average electricity price of businesses garding the role of price incentives. Based on firm-level within the same sector and sector-by-region. Conclusions data on electricity and gas consumption and prices, we estimate the price elasticity of electricity demand and examine the margins of adjustment that firms have to rising The first set of preliminary results reveals that firms re- spond to electricity price changes by adjusting electricity charges. The identification strategy relies on the exogenous consumption. A 1 percent price increase is associated with an changes in electricity prices due to disruptive events in glob- electricity reduction of 0.9 percent-1.3 percent. We also find al energy markets and the centralization of tariff adjustment that firms with greater electricity dependency are less respon- 95 Chapter 6 Greening Firms in Georgia TABLE 22 Firm-Level Outcomes Dep. Energy quanti- Electricity Log employ- Log profit Log investment Log costs Log sales variable: ty efficiency quantity effi- ment ciency OLS 2SLS OLS 2SLS OLS 2SLS OLS 2SLS OLS 2SLS OLS 2SLS OLS 2SLS (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Elec. price 0.741*** 1.247*** 1.081*** 1.471*** 0.017 0.199 0.130 0.932 -0.372 -1.281 0.100 2.482*** 0.125** 1.204*** (log) (0.084) (0.475) (0.077) (0.412) (0.035) (0.197) (0.132) (0.712) (0.365) (1.693) (0.080) (0.449) (0.059) (0.346) Sales (log) 0.331*** 0.336*** 0.415*** (0.010) (0.011) (0.108) Labor prod. 0.178*** 0.174*** 0.190*** 0.186*** 0.077 0.008 (0.012) (0.013) (0.012) (0.012) (0.049) (0.057) Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Geographic Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes FE Industry x Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observa- 12,099 11,221 12,098 11,226 14,925 13,741 7,829 7,345 2,270 2,147 15,464 14,197 14,925 13,741 tions Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Left-hand side variable is the (logged) electricity consumption (kWh); electricity dependency is the electricity-to-sales ratio. Geographic FE are at the region level; industry × year FE capture time-varying 2-digit industry effects. Source: World Bank calculations based on GEOSTAT. sive to price changes, which suggests that price instruments Finally, despite examining many margins of adjustment could be less effective for encouraging energy saving among like employment, profits, investments, costs and sales, we energy-intensive businesses. Furthermore, when conditioning only find an increase in costs and sales in response to high- on manufacturing and construction, two activities for which er electricity prices in the IV setting. That the magnitude of we have enough observations, we observe that consumption the cost response is larger than the sales response suggests that adjustments due to price variation are significant in both sec- businesses are passing on energy costs to production prices. tors. At face value, firms appear to respond to energy prices However, as we do not currently have production price data, rationally, supporting the hypothesis that subsidized, cheap we cannot confirm this is the mechanism driving this effect. energy could lead to overconsumption and energy waste. To sum up, we find evidence that price incentives effec- The second set of results further reinforces the idea of tively reduce energy consumption and, more important, rational energy consumption. Energy price changes are increase energy and electricity efficiency. We acknowledge positively associated with energy and electricity efficiency. A that further research needs to be done to assess the robustness 1 percent increase in electricity prices is associated with a 0.74 of these results. Still, we consider this first set of preliminary percent-1.25 percent rise in energy (quantity) efficiency and findings could contribute to the discussion on the role of price with a 1.08 percent-1.47 percent rise in electricity (quantity) incentives in production and Georgia’s green transition. efficiency after considering firm fixed effects, industry-specific trends, geographic, and economy-wide shocks. To our under- standing, this is a strong preliminary finding since it highlights the role of price incentives in enhancing EE in Georgia. 96 Chapter 6 Technical Report Chapter 7 96—106 PAGE Policy Recommendations A critical finding of this report is that there is scope To implement this greening process, this report distin- for improving EE without significant changes in guishes two levels of policy recommendations: recom- output or the industry composition of the econ- mendations at the systemic level, focusing on institutions omy. Given the large differences in EE across firms and the legal framework, and recommendations that tar- within sectors, our analysis suggests that greening the get firms and sectors. Based on the findings discussed in the Georgian economy can be largely done by investing in previous sections, the policy recommendations can be divided upgrading firm-level efficiency rather than changing into two groups. First, we discuss country-wide cross-cutting the patterns of economic specialization. A cost-effective ap- recommendations focused on developing a conducive enabling proach would be guided by focusing at least initially on key pri- business environment centered on the appropriate price sig- ority sectors characterized by: (i) high level of EE dispersion, (ii) nals and regulations. Second, we concentrate on firm-level high level of consumption due to sector intensity and size (in recommendations and group them around three pillars: (1) in- terms of output and employment). For example, construction, formation, (2) capabilities -management, technology and skills non-metallic products, and transport services would account - and (3) finance (see Figure 47 for a schematic view of this). for about two-thirds of total energy savings if efficiency in firms below the median was improved to the median efficiency level in their sector. FIGURE 47 1 2 3 Information Capabilities Finance Structure of Policy Recom- mendations Awareness Management and organization Guarantees Information about suppliers of Technology adoption Blended finance and credit solutions lines Skills for workers, and managers Energy audits Grants and tax incentives Understanding about solutions ESCOs Utility on-bill financing Leasing Energy savings insurance Conducive enabling business environment Source: World Bank I N ST I T U T I O N S + R E G U L AT I O N S + P R I C E S elaboration. 97 Chapter 6 Greening Firms in Georgia CHAPTER 7.1 Completing and Implementing an Enabling Institutional and Legal Framework E xisting green strategies and plans need to be and certification schemes for energy service providers, auditors aligned, approved and translated into concrete and and managers.81 Second, to address the current shortage of applied regulatory measures. Georgia already has var- qualified energy auditors, it is essential to deliver accredita- ious ambitious strategic documents that define targets tion programs that certify core competencies and training in and objectives until 2030 and longer-term ones with the consultation with industry experts. Accredited experts can be Long-Term Low Emission Development Concept. At this listed in a directory such as the one managed by the Federal stage, two strategic documents have been drafted and Ministry for Economic Affairs and Energy of Germany.82 The require approval: the National Energy Policy of Georgia and third relates to resources and financial incentives. Sustainable the National Integrated Energy and Climate Plan of Georgia. financing mechanisms for ESCOs and SMEs should be further Beyond aligning these strategic documents to have a shared developed. An example of this type of support is the “Germany vision and targets and approving drafted ones, the most crucial makes it efficient” initiative. The support grant covers up to 80 challenge today is implementing these visions and strategies. It percent of SMEs’ energy audit costs (up to €6,000).83 requires work along three dimensions: institutions, regulations and resources. From the macro level perspective, it is critical to make sure prices do not distort business incentives to invest in Strengthening institutional governance and improving EE and sustainable technologies. Non-household electricity coordination of green policies and regulations are at the prices in Georgia are among the lowest compared to benchmark 78 Specifically, out of 19 heart of the greening process. One of the main challenges to countries, while it is one of the countries that subsidize energy bylaws required by the Energy Efficiency Law only 4 have improving governance is defining or establishing a dedicated the most. The results from simulating an improvement of EE been adopted as of May 30, entity responsible for implementing and coordinating activ- in the business sector show small savings in production costs 2023. ities required by the legal framework for enhancing EE and and an even smaller profit increase compared to the aggregate reducing industrial emissions. This entity should effectively reduction in energy consumption. In 2021, Georgia experienced 79 Specifically, out of 13 drive EE improvements, coordinate and strategically orient do- a significant increase in industrial electricity prices and a sub- bylaws required by this Law only 3 have been adopted as nor support, facilitate sustainable development, and contrib- sequent rise in gas prices in 2020, whose impact needs to be bet- of May 30, 2023. ute to decarbonizing the economy. One of its primary tasks is ter analyzed. However, it is important to consider establishing developing an effective and sustainable system for measuring, carbon pricing mechanisms to provide the appropriate pricing 80 The Unified National monitoring, reporting and verifying energy savings. signals as an option, at least for specific sectors and consumers. Body of Accreditation In the European context, an example of carbon pricing is the that currently performs The regulatory framework needs to be completed. While EU ETS system, which relies on the “cap and trade” principle. accreditation activities for Conformity Assessment significant progress has been made in approving several bylaws Within a cap, companies receive or buy emission allowances, Bodies could take this role. required by the Energy Efficiency Law, key bylaws still require which they can trade. This total cap is then reduced over time to approval (e.g., certification rules and inspection regulations of lower emission levels and make the value of carbon allowances 81 Typically, this body heating and air-conditioning systems).78 Similarly, the Energy go up, providing a direct price signal to improve efficiency and would be responsible for Efficiency Law (art. 19) requires creating programs to inform reduce emissions. White certificate schemes like emissions standardization, quality SMEs about the benefits of introducing energy management trading systems are alternative market-based instruments assurance and market transparency. Additionally, it systems and encourage them to undertake audits. However, wherein businesses earn certificates for efficiency improve- could also be responsible for the regulatory framework for these programs is yet to be de- ment. The advantage for SMEs is they can acquire such certif- professional development and veloped. Bylaws required by the Energy Efficiency in Buildings icates by implementing an EE project, reducing project costs. offer training and professional law also need to be adopted.79 Another fundamental piece of Also, administration costs can be reduced when equipment or development opportunities the regulatory framework is the Law on Industrial Emissions, service providers manage the administration of certificates. in partnership with existing submitted to the Parliament in December 2022. training institutions. Addressing subsidies and cross-subsidies can make fiscal 82 https://www.energie- Implementing these ambitious strategies and regulatory space for promoting investments in EE. A detailed review effizienz-experten.de/ packages is the most pressing challenge to create a con- of existing energy subsidies should be undertaken to phase ducing environment for unleashing EE investments. Three out inefficient and distortionary ones. In return, it can expand 83 For smaller companies areas are critical for the implementation stage. First, to ensure fiscal space for allocating more resources toward investments that spend less than €10,000 on their energy bills, the the effective execution of EE measures and to build trust in in EE. Similarly, limiting cross-subsidies between commercial maximum amount is €1,200 the services energy professionals provide, it is important to and residential users is important to make EE investments (BMWK, 2023). consider the definition80 of a body responsible for accreditation more attractive for businesses. 98 Chapter 6 Technical Report CHAPTER 7.2 Supporting Firms to Invest in Energy Efficiency and Upgrade Technologies and Practices A ccelerating the green transition of Georgian firms can be summarized as (1) ensuring flows of information; (2) im- requires a set of policies that consider the hetero- proving capabilities in terms of management, technology, and geneity found in the report and build on existing skills; and (3) easing access to green finance. The cross-cutting programs and agencies’ capabilities to design and area focuses on a conducive enabling environment for greening implement green policies. Based on the findings the private sector, centered on appropriate price signals and of this report, we group firm-level recommendations regulations. This enabling environment for EE will dictate the around three vertical pillars with specific solutions speed of the green transition of Georgian businesses. and a cross-cutting area, as Figure 47 depicts. The three pillars FIGURE 48 Solutions for Access to Finance EXAMPLE: EXAMPLE: • Commercial loans • Private equity • Concessional loans • Venture capital • Micro-credits • Mezzanine capital Debt financing Equity financing EXAMPLE: • Receivables management to improve short-term cashflow EXAMPLE: • Contracts between • Public grants Institutional Financing solu- Leaser two parties where • Tax incentives funding tions factoring the lessor provides • Loan guarantees an asset for usage to another party for a period of time, in return for specified payments Supporting Tailored non-financial financial EXAMPLE: services instruments • Arrangement with energy and EXAMPLE: service companies (ESCOs) • Capacity building (from its internal funds or by on financial literacy the customer, or by a third party funding) • Carbon finance (CDM) Source: Mueller & Tuncer (2013) 99 Chapter 6 Greening Firms in Georgia A Information and Awareness The importance of information and awareness emerges nate information about the potential impact of EE on climate from the analysis as a critical constraint for firms in Geor- change and sustainability through workshops, webinars, and gia. Our results suggest that firms are often unaware or ignore educational events. Likewise, campaigns could disseminate in- the potential savings of becoming more energy efficient and, formation to firms on the EE potential and their role in climate more important, that they are also unaware of what EE tech- change global challenges. Central to these efforts is to provide nologies are available for their needs. In this context, low-cost tools for businesses with benchmarks of their green technology information campaigns can be relevant for raising awareness adoption and energy-saving calculators.84 Targeting events about firms’ environmental impact and providing feedback for specific sectors or firm types can address unique needs. and benchmarking about EE savings (e.g., giving examples of Following Enterprise Georgia’s lead in digitalization programs practices for optimizing energy consumption or ranking sim- that include diagnostics and benchmarking tools, this can be ilar firms according to energy consumption). These initiatives easily extended to green initiatives. can address information gaps and minimize existing behavior- al biases that limit the adoption of green practices. Similarly, Additionally, agencies partnering with industry associ- our results point to the role of information spillovers as firms ations should make sure SMEs access information about seem to be learning or imitating good efficiency practices of tailored EE solutions and who supplies these solutions. peers in nearby locations and the same sector. These spillovers Even when firms are aware of the relevance of EE, the lack of -knowledge sharing, demonstration effects or hiring key work- information and knowledge about the availability85 of ener- ers from greener firms- play a critical role in the diffusion of gy solutions and suppliers may prevent them from investing. green practices and technologies. One approach to overcome this problem is informing SMEs through a centralized online platform that lists green technol- To encourage firms’ investments in EE, the initial step is ogy solutions and their suppliers, organized by type of solution to enhance awareness of its importance and benefits. This and sector. The platform can include other firms’ reviews as a can contribute to reducing behavioral biases that make firms mechanism for vetting providers and ensuring service quali- perceive the EE investment returns as “too” low—the “energy ty, as well as the cost of different solutions for helping firms efficiency paradox.” A comprehensive information campaign make more informed decisions. In addition, the platform can showcase the economic and environmental advantages could include resources to help firms evaluate suppliers and of EE. These initiatives can illustrate successful cases of firms their solutions, such as guides to conducting due diligence implementing EE measures and their cost-saving and benefits or questions to ask potential suppliers. achievements. Similarly, such campaigns could also dissemi- BOX 13 Examples of campaigns and programs to raise awareness 84 Allcott & Mullainathan (2010) discuss the potential The Energy Efficiency Network Initiative in Germany is an ex- mation on EE potential actions and the application process for cost-effective non-price ample of a public initiative that brings together companies to for government support.87 The campaign covers short-term behavioral interventions form EE networks where they can exchange information and and simple measures (“quick and easy no-cost actions”), in nudging beneficiaries to become more energy efficient. learn from each other’s experiences. Each network consists mid-term, low-cost, energy-saving actions (e.g., installing of 8-15 companies assisted by an energy consultant who smart meters or conducting an energy usage assessment) 85 An example is the Tech Selector platform which is helps them identify ways to improve EE and enter a moder- and information about longer-term investments typically in- supported by the Austrian ated dialogue through which the companies exchange expe- volving more energy-efficient technologies. Government and provides rience and ideas. The goal is to implement EE measures that a global country-specific platform to connect vendors would otherwise not have been realized (BMWK, 2020).86 In Sweden, the ENIG is an EE network focused on SMEs in of green technologies to the manufacturing sector. ENIG is run by Swedish Research business customers. A more recent example is the UK Government’s recent cam- Institute for Industrial Renewal and Sustainable Growth 86 For more information paign supporting businesses to increase their EE. The cam- (Swerea); the Swedish Energy Agency is a partner and see the Energy Efficiency paign targets SMEs and offers guidance on how to reduce funder. ENIG creates, collects and disseminates information Networks Initiative and energy efficiency campaigns networks energy consumption and cut emissions, from installing light on EE technologies, practices and methods. The network en- (DENA). and heating timers to turning down boiler flow temperature ables cross-industry collaboration in areas of common inter- and changing light bulbs. The initiative also provides specific est, such as ventilation, compressed air and lighting. 87 See the UK Energy Efficiency for Business web case studies and examples in addition to guidance and infor- site. 100 Chapter 6 Technical Report Expanding the market of energy audits. Energy audits can the expansion of the energy audits market, the Government play a crucial role in helping firms to identify and eliminate of Georgia could develop an online marketplace where SMEs energy waste while raising awareness about the opportuni- access to a list of certified energy auditors and their previous ties to upgrade EE. The market for energy audits in Georgia engagements, including feedback from earlier clients. is incipient, requiring both supply- and demand-side public interventions to develop it further. On the demand side, a first A key issue for the energy audits market to work effec- step was taken with the introduction of compulsory audits for tively is the level of understanding of EE solutions and the 300 largest commercial energy consumers. However, this their impact on businesses. Besides understanding who requirement does not apply to SMEs; therefore, to increase the suppliers of solutions are, firms need to also be better in- audit take-up, firms need incentives. For example, following formed about the portfolio solutions and their impact on the the example of Germany,88 the government could provide fi- consumption of energy. Educational resources explaining how nancial support through partial grants or small vouchers for solutions work, their potential benefits, and how to implement SMEs implementing an energy audit with a certified supplier. them can address these information problems. Moreover, they Similarly, delivering financial support through blended finance can also include guides, tutorials, case studies and interactive for implementing energy-saving measures identified in the au- tools to simulate potential cost savings and other benefits of dits would also increase the demand. On the supply side, train- different solutions. Complementary information about best ing and certification programs for can increase the number of practices for implementing EE measures, helping firms avoid qualified and accredited energy auditors. Finally, to build trust common pitfalls and maximize their investment returns can in the system and reduce information asymmetries that limit be helpful as well. 88 The Federal Ministry for Economic Affairs and Energy provides funding for energy audits that cover 80 BOX 14 Examples of information about solutions and suppliers of solutions percent of the energy audit costs (up to € 6,000) with a There are many examples in the US about delivering EE The UK Energy Technology List (ETL) is a government list list of experts accessible at the Energie Effizienz Experten solutions and providing information to firms. The Feder- of energy-saving technologies businesses can choose to website. al Energy Management Program (FEMP) provides a list reduce energy consumption and claim an enhanced capital 89 For more details, see of qualified ESCO approved by the Department of Energy allowance. This list informs on energy-efficient equipment Federal Energy Management Qualification Review Board.89 Also, the Green Suppliers (for example, boilers, electric motors, air-conditioning and Program. Network is a partnership between the US Environmental refrigeration equipment) and energy evaluation services, 90 For further discussion, Protection Agency, the Department of Commerce, and in- helping firms in finding trusted suppliers. Buyers can com- check the Energy Technology dustry to improve the performance of manufacturers and pare products listed by technology category and get details Factsheet elaborated by the Department for Business, their suppliers. The network helps businesses connect with about manufacturers or suppliers.90 Energy & Industrial Strategy suppliers of energy-efficient solutions. of the UK Government. B Improving Capabilities: Management and Organization, Technology, and Skills As shown in Section 3, achieving the median firm EE may programs vary from general to customized services that de- require large investments in some cases but in others is velop specific business skills (e.g., hard and soft) with wider or more linked to upgrading management and organization- narrower perspectives according to the organization’s needs. al capabilities. Public policies incentivizing the upgrade of Fostering the organizations’ capacities can raise EE awareness managerial skills may encourage green management adop- among managers and workers because of cost-saving decisions tion across firms. Because green management quality tends and environmental-related parts. to be higher among firms with more sophisticated managerial skills, developing firms’ organizational capabilities could lead Management training should be streamlined in existing to greater adoption of green management, while also becoming Enterprise Georgia or GITA support when possible. These more productive. Implementing sustainability plans requires agencies, EG for SMEs and GITA focusing more on startups, setting and monitoring energy consumption targets, inputs already offer services to SMEs through knowledge providers, and production processes. Providing incentives and perfor- which should include courses on sustainable management and mance-based pay can also increase incentives for workers to training with industry associations. implement sustainability plans and energy savings. Existing market failures associated with behavioral biases and informa- Like management support, programs supporting technol- tion asymmetries with knowledge providers (Cirera & Maloney, ogy adoption can complement green practices. Digital ICT 2017) often result in low management quality. Public interven- technologies like ERP can enhance EE by helping firms better tions that include but are not limited to training, consulting, plan the production process and achieve energy consumption and outsourcing can help improve managerial practices and targets. These technologies complement management quality introduce more sustainable green practices in the firm. These and support firms in better planning production and manage- 101 Chapter 6 Greening Firms in Georgia ment systems. The green dimension can be integrated into the formance contracts (ESPCs) to fund firms’ improvements with Enterprise Georgia policy toolbox. For example, EG already sup- generated savings. These contracts also ensure high-quality ports digitalization by providing additional training on using technical assistance due to incentives for the service provid- these digital tools for EE. In addition, encouraging the adoption er. However, this type of contract is more suitable for large of off-the-shelf green technologies, such as thermostats, led firms, which can generate larger EE gains than small firms. lighting, and VAC/HVAC systems, by providing information A second model to complement ESCO is the more traditional and technical assistance that can accelerate the adoption of technology extension services, which should be established as such technologies. public-private partnerships, jointly with industry associations. They can provide services that demonstrate and diffuse these Increasing EE and reducing carbon emissions in certain green technologies in specific sectors. sectors is likely to require adopting more complex and expensive technologies. Changing production techniques in These technology services can also play an important specific energy-intensive sectors requires specialized technical role in re-skilling the labor force. Lack of adequate skills assistance, skills and finance. In some industries, energy-effi- is ranked by Georgian firms as an important barrier to tech cient firms can be much more capital-intensive than inefficient adoption, especially among smaller firms. In addition, Bastos ones, meaning that inefficient firms may require either a larg- et al. (forthcoming) show how the demand for green jobs has er scale or investments. In these cases, specialized technical increased exponentially since 2021. Using technology services assistance can be offered via two types of support programs. infrastructure to deliver training in partnership with industry The first enables market-based instruments, such as energy associations can address part of the skill gap and train workers savings contracting (ESCO), which uses energy savings per- in using these more complex green technologies. BOX 15 Examples of programs to improve manage- rial capacities, organization and skills The European Energy Managers (EUREM) is a standardized ment malfunctions. The process took three months (four energy management training that provides graduate courses days a week of staff time), and it was possible because many for becoming accredited experts for energy audits or energy procedures were already in place and needed only minor managers, joining a large pan-European network. The initia- changes to make it applicable for ISO 50001. A key success tive is carried out by the Chamber of Commerce/Industry factor was providing training tailored to the skills and needs of Nuremberg and funded through EU-Horizon 2020. As of of non-technical staff. Energy savings have been 5 percent to today, the network consists of a large pool of alumni (over 10 percent on average (with savings on store level up to 30 4,000 energy managers) and offers an online platform for percent), largely due to continual attention to the functioning them to interact, exchange information and act as social of the system and rapid response to problems. In the future, communities. Events and awards for energy management energy management-related activities could be expanded to excellence are regularly held to keep the community active. the supply chain. The Swedish National Energy Efficiency Network is a pro- Another example of building skills for EE and greening busi- gram established in 2015 and run by the Swedish Energy nesses is a program implemented jointly by the Hong Kong Agency to set up networks of SMEs (6-16) to exchange best Electronic Industries Association and Hong Kong Productiv- practices, individual counseling and group consultancy from ity Council, with funding from the Singapore government’s an external energy expert. SME Development Fund. The project was developed to show that energy management systems are relevant for SMEs. It An interesting example is provided by a program of energy has adapted training designed for large organizations so that management in the retail sector in the Netherlands, where SMEs can use the ISO 50001 standard for energy manage- the company Lidl has ISO 50001 certified almost 400 of ment. The support initiative introduces the benefits of energy its branch stores, with about 28 employees per store. The management and the skills needed for successful implemen- most important motivations were cost reduction and energy tation. The program includes a telephone hotline, subsidized awareness within the organization. A key aim was to enhance energy reviews of SME operations and short seminars ex- the company’s reputation. The investments required were plaining the main requirements of the ISO 50001 standard. moderate: €12,000 for certification and €4,000 for staff For SMEs that want to go further, the initiative offers a refer- training focused on understanding where and how energy is ence guide and training sessions to assist companies through used and quickly finding and addressing problems or equip- the implementation process. 102 Chapter 6 Technical Report C Finance Providing financial incentives for firms to invest in up- with lending for small, non-traditional projects. Development grading and EE should consider the vast portfolio of in- banks provide special credit line initiatives directly or through struments available.91 The financial requirements and needs local commercial banks. One example of the former is the KfW are likely to vary across different types of firms. To effectively bank in Germany, which finances investments that aim to im- address them, the Government of Georgia can develop a com- prove the environmental situation, such as the development of prehensive portfolio of well-designed, targeted instruments, renewable energies –e.g., electricity and heat from the ground, as shown in Figure 48. sun, wind and water. The KfW finances small enterprises up to €25 million per project at favorable interest rates and up to Loan guarantees to help reduce risks and encourage three repayment-free start-up years. Another example is the banks lending. Lack of access to capital for EE investments EBRD through the Sustainable Energy Financing Facilities can be addressed by providing more incentives for financial (SEFFs). The EBRD extends credit lines to local financial in- institutions to lend to businesses for EE investments. Building stitutions seeking to develop sustainable energy financing as on the recently approved Sustainable Finance Taxonomy, the a permanent business area. There are two key areas for project government could consider setting up specific support of loan financing: EE and the development of small-scale renewable guarantees for green investments. There are many examples energy. Local financial institutions on-lend the funds they have of such types of programs. In the US, the Loan Guarantee Pro- received from the EBRD to their clients, among which are small gram of the Department of Energy provides loan guarantees and medium-sized businesses, corporate and residential bor- for EE projects. Another example in the European context is rowers, and renewable energy project developers. the Smart Finance for Smart Buildings initiative developed by the European Commission with EIB, which provides a flexible Grants and tax incentives are needed to encourage in- 91 A useful reference guarantee facility model that helps de-risk investments in EE vestments at early stages and to address specific market for instruments, programs for buildings. The Swedish Credit Guarantees gives another failures. In the context of high levels of uncertainty, such as in- and experience to finance example of green investments. It offers state loan guarantees vesting in early stages to improve EE (e.g., contracting an audit), greening businesses is the that credit institutions provide for large green industrial in- grants and tax incentives may be the only solution to unlock OECD Clean Energy Finance vestments.92 Such solutions are likely to be more important for SMEs’ interest and investments. Grants can show additionality and Investment Mobilization those investments with higher levels of risk or for supporting and be prioritized for projects that businesses would not other- (CEFIM) Program: a Forum access to finance for riskier firms that would be otherwise un- wise finance. However, an important factor to consider for these that provides guidance on policy design and facilitates likely to access traditional banking finance. This type of loan schemes’ success is ensuring a streamlined application process dialogue with investors guarantee program could be especially important at the initial and adequate information and guidance for applicants. Lim- to attract private capital phase of the development of the market for EE investments, as iting administrative costs is key for demand, especially when for investment in green local banks are usually less familiar with EE business concepts grants are small. Grants and tax relief schemes are common. technologies / energy or have limited capacity to mitigate associated risks.93 For example, the UK Energy Technology List (ETL) scheme efficiency. Similarly for offers tax relief for businesses that invest in energy-efficient more details and country Blended finance and concessional loans can be relevant in equipment.96 The Federal Office for Economic Affairs and Ex- experiences about financing energy efficiency see the increasing incentives and unleashing investment in SMEs’ port Control (Germany), which provide grants for investments report by Taylor et al (2008). EE. Blended finance for EE investments involves a mix of debt in energy-efficient technologies, are another example of orga- financing with a grant component usually triggered by reaching nizations delivering these initiatives. These grants distinguish 92 For more details, see the specific milestones, measured in savings or achievements of different groups of technologies and provide different levels Credit guarantees for green a certain certification class. Another solution is concessional of subsidies for each of them: (1) cross-cutting technologies investments project on the loans. Based on grants and public subsidies, concessional loans that typically are tried-and-trusted energy-efficient alternatives Swedish National Debt Office Website. have lower interest rates and longer repayment methods for spe- (such as ventilation systems and pumps) are supported with cific investments in EE. These solutions can be helpful in con- a grant covering 30-40 percent of the costs; (2) technologies 93 For further discussion texts of high capital costs, especially for local SMEs. For example, for process heat from renewable energy (for example, efficient see Taylor et al. (2008). the KfW in Germany offers promotional low-interest rate loans heat pumps from biomass installations) are supported with a 94 See the KfW website. to companies for EE upgrades in production facilities.94 Another grant covering 40-55 percent of costs; (3) metering and control example is the PF4EE (Private Finance for Energy Efficiency), technology for sensors to integrate into an energy manage- 95 See the Private Finance a joint initiative between the EIB and European Commission.95 ment systems (including energy management software and for Energy Efficiency website. training of employees) receive a subsidy of 30-40 percent of 96 See the UK Government Special credit lines to overcome collateral requirements costs; (4) investments for energy-related optimization of instal- ETL Scheme website. and information asymmetry simultaneously. Govern- lations and processes can also receive grants to fund up to 30-40 ments and development banks can extend special credit lines percent of eligible costs under the condition that companies 97 For further information, when businesses do not meet collateral requirements and com- demonstrate there are specific energy savings resulting from visit the Federal Ministry of mercial banks lack the screening technology and familiarity to these investments.97 Another example of tax incentives is the Economic Affairs and Climate Action of Germany website. underwrite small-scale projects. Special credit lines function Energy Efficiency Tax Incentive in South Africa. The scheme as financial instruments that governments and development provides registered companies tax relief of US$0.45 for each 98 Therefore, a key banks can use to increase the credit volume under improved kilowatt-hour (kWh) of energy saved. The key condition for pre-condition for success conditions. Governments and development banks allocate joining the program was that projects must establish a 12-month would be to first establish funds with discretion to specific projects through local com- energy usage baseline before claiming the incentive, which communication and mercial banks. By easing credit requirements and reducing the requires a certain level of awareness and capacity to monitor awareness campaign and develop sufficient capacities to financial risks of commercial banks, these preferential loans are energy usage, often rare among Georgian SMEs.98 monitor energy consumption attractive for businesses that otherwise would not qualify for among businesses. credit. At the same time, they also help local banks get familiar 103 Chapter 6 Greening Firms in Georgia Developing a market for ESCOs with shared savings can Equipment leasing can help SMEs with limited own be decisive for undertaking large-scale EE projects. ESCOs capital and no access to finance to enhance EE. Leasing provide EE solutions and finance the upfront costs. Specifical- energy-efficient machinery and equipment can be useful for ly, they design, construct, operate and finance EE equipment small businesses with limited capital that do not yet meet the and upgrading, and then the customer pays for energy savings requirements for accessing commercial credit lines to over- through an agreed rate conditional on the level of energy sav- come the upfront cost of improving EE. In a leasing agree- ings99 or pays a fee for a guaranteed level of service. Typically, ment, the customer pays for the right to use the equipment to the market failure addressed by this funding mechanism is the the financier, who owns the asset, instead of buying it. In cer- existence of high upfront costs and the risk of EE investments. tain cases, the customer may be rewarded for reduced energy Under this funding model with shared savings, the customer costs. This instrument substantially lowers technology access does not invest in the project but receives a fraction of the en- requirements, especially among entrepreneurs, startups and ergy savings (World Bank, 2016). ESCOs usually function best small-scale businesses. There are many examples of leasing in with large-scale projects and prefer large companies to avoid green projects. The Energy Leasing Program (administered by risks during the project (Akman et al., 2013) —some examples Virginia Department of the Treasure, US) leases to agencies, of ESCOs function under a funding model with shared savings institutions, and authorities for EE projects. Also, the Stra- across the globe. For example, 45 registered ESCOs in Thailand tegic Bank Corporation of Ireland offers leasing to SMEs to function using the guaranteed and shared savings model and finance assets, including plant and machinery and transport focus on industry, hospital and government buildings. Similar equipment. Some initiatives are already in place in Georgia. examples exist in the United States, Canada, Japan, Korea, and For example, the Green for Growth Fund (GGF) approved a loan China (World Bank, 2016). ESCOs, usually work well -especially to the Tbilisi Business Center Leasing in 2020 to bolster EE. in the context of larger projects- and prefer to work with larger businesses that are less likely to fail during the funding period. Energy savings insurance (ESI) overcomes the lack of local technical capacities for evaluating the viability of Utility on-bill financing and on-bill repayment have the green investments. Local banks and energy-saving insurance potential to overcome credit market constraints and boost SMEs may not have the technical capability to evaluate the EE. Capital requirements and credit market conditions are two energy-saving potential of green investments and estimate the relevant barriers faced by firms, especially SMEs, in Georgia. project’s return. Energy-saving insurance offers an instrument On-bill financing and repayment programs offer firms to pay that mitigates project risks and generates investor confidence for clean energy upgrades through their utility. The energy re- through a scheme of guaranteed energy performance. The ESI tailer or a third-party finance the EE project, and the custom- model builds trust and credibility to reduce SMEs risks by pro- er repays the investment through an additional charge on the viding a financing structure, a standardized contract, energy monthly bill. This program can lower credit risks considerably savings insurance and a validation entity. If the energy savings since the financier can proxy bill repayment using past bills, and value is not met, the insurance pays out. Successful examples failure to pay can be tied to disconnection. On-bill financing in Latin America demonstrate the potential of this instrument. has increased its popularity in some countries. In the US, for Mexico and Colombia are implementing the ESI model through example, the Minnesota Conservation Improvement Program development banks (FIRA —Fideicomisos Instituidos en Rel- (CIP) authorizes on-bill and similar instruments. In Arkansas in ación con la Agricultura in Spanish- in the former case and 2013, the Ouachita Electric Cooperative piloted the first on-bill BancoIdex in the latter). The ESI finances mid- and long-term 99 In this case ESCOs and customers need to agree on financing program, the Home Energy Lending Program (HELP). loans and mitigation instruments (contracts, energy-saving how cost savings are shared, Other state cooperatives have adopted the scheme, approving insurance and technical validation). measured and verified. investments above US$ 1.6 million within three years. D Prioritization of Policies Policy recommendations could be grouped in terms of oped, require a competitive and transparent functioning of the their ‘priority’ and ‘complexity’ to guide the design and electricity and gas markets, so albeit desirable, they cannot implementation stage. Policy recommendations developed be expected in the short term. For similar reasons, developing throughout this section can be classified in terms of their pri- and implementing such instruments is highly complex. In the ority and complexity. While priority involves an assessment of same way, consulting services first require technical capacities the urgency with which these interventions should be designed and awareness so firms can identify the need and willingness and the time they would take for an effective implementation, to pay for environmental-related consulting. In contrast, infor- complexity refers to the technical and political economy feasi- mation campaigns can be launched swiftly and involve little bility of these actions. For example, developing specific types coordination and regulatory procedures. Table 23 summarizes of credit instruments (e.g., utility on-bill financing) may take previous policy recommendations and classifies them in terms considerable time since it requires first developing financial of priority (short-, mid-, or mid-long- term) and complexity markets, changing the regulatory framework and making the (low, medium-low, medium, and medium-high), with the in- retailer or credit company invest in screening and risk assess- tention to guide prioritization and implementation in Georgia. ment technologies. Usually, these initiatives, still underdevel- 104 Chapter 6 Technical Report TABLE 23 Prioritizing Policy Recommendations and Assessing Their Complexity Information ▶ Raise awareness ▶ Provide feedback and ▶ Ensure firms have ▶ Extend market through low-cost benchmarking about access to green solutions energy audits information campaigns energy savings and suppliers SHORT LOW SHORT MED-LOW SHORT LOW MID MED ▶ Develop interactive ▶ Develop educational tools for information resources for energy dissemination efficiency SHORT MED-LOW MID MED-LOW Firms' and Workers' Capabilities ▶ Provide consulting and ▶ Train managers and ▶ Provide specialized outsourcing services workers on energy technical assistance for efficiency production techniques MID MED MID MED-HI MID MED ▶ Support firm digitalization ▶ Boost technology adoption MID MED MID MED-HI Finance ▶ Encourage banks to ▶ Expand blended finance ▶ Extend special ▶ Provide grants provide loan guarantees and concessional loans credit lines through and tax incentives through incentives development banks MID MED MID MED SHORT MED-LOW SHORT MED-LOW ▶ Enable energy savings ▶ Develop utility on-bill ▶ Enhance equipment ▶ Develop energy contracting (ESCO market) financing and repayment leasing savings insurance MID MED-HI LONG HI LONG MED-HI LONG MED-HI Source: World Bank elaboration. E Beyond Specific Solutions: The Importance of Targeting As mentioned throughout this report, businesses face is the Energy Star Portfolio Manager, which lets businesses different needs and barriers to becoming greener, so a measure and track energy and water consumption and GHG one-size-fits-all solution is unlikely to accelerate the emissions. It can benchmark the performance of one building green transition effectively. This reinforces the need to tar- or a whole portfolio of buildings. The Carbon Trust’s Carbon get and effectively tailor specific policy solutions instead of Footprint Calculator Tool or its eAunAunergy management implementing a blank approach that would result in low EE self-assessment tool are also examples of evaluation initia- gains or lack of effectiveness. To inform this targeting, the first tives helping businesses measure their corporate emissions, step is to develop appropriate diagnostic and benchmark tools environmental impact, energy management performance100 that provide more information on both needs and capacities and detect areas where they can improve EE. Another set of 100 Energy management of individual firms. Such diagnostic tools could help assess diagnostics is the Greenhouse Gas Protocol tools developed by and policy, organization, the “readiness” of the firms for different types of interven- the World Resources Institute (WRI) and the World Business training, performance tions and channel the firm investments and support toward Council for Sustainable Development (WBCSD). These tools101 measurement, communication, the more pressing needs and binding constraints. For exam- provide standards and guidance for businesses to measure and and investment. ple, the IACs (Industrial Assessment Centers) in the US, spon- manage GHG emissions and contribute to identifying areas for 101 The tools include cross- sored by the Department of Energy, are located at universities optimizing energy consumption. This diagnostic can also help sector tools as well as sector- across the country and provide free energy, productivity, and other support policies assess potential beneficiaries’ readiness specific tools. waste assessments to SMEs. Another example of online tools and improve targeting. 105 Chapter 6 Greening Firms in Georgia One justification for targeting is the need to treat sectors -having better regulation and less distorted energy prices- can differently because of the differences in the potential significantly affect EE. In others, the priority should be adopt- sources for EE gains. The decomposition analysis imple- ing and diffusing green technologies that can change efficiency mented in Section 2 can help identify the priorities that should and techniques. Finally, other industries also need to change guide policy in improving EE. Table 24 shows three key policy specialization patterns within sectors. Priorities can sometimes areas for greening sectors based on the potential sources for EE vary between EE and GHG emission efficiency policies. gains. First, for some industries, better functioning of markets TABLE 24 What Should Policy Actions Address? Agriculture and fishing Mining and quarrying Energy Consumption C02 Emissions Energy Consumption C02 Emissions Patterns of activity Functioning Functioning – specialization of markets of markets Manufacturing Utilities Energy Consumption C02 Emissions Energy Consumption C02 Emissions Technology adoption Technology adoption Functioning Patterns of activity of markets specialization Construction Wholesale and retail Energy Consumption C02 Emissions Energy Consumption C02 Emissions Technology adoption Technology adoption Functioning Functioning of markets of markets Accommodation Transport activities and restaurants Energy Consumption C02 Emissions Energy Consumption C02 Emissions Functioning Functioning Technology adoption Functioning of markets of markets of markets Other services Energy Consumption C02 Emissions Patterns of activity Patterns of activity Source: World Bank elaboration based on GEOSTAT. specialization specialization 106 Chapter 6 Technical Report Approach the problem of promoting EE with a modular eliminate import tariffs); iii) assess the readiness of potential portfolio approach. In this perspective, any support program beneficiaries with diagnostics and benchmark tools and iden- could be considered a set of modular tools and a portfolio of in- tify relevant actors through which portfolio interventions can terventions that should be put forward based on international be delivered and businesses can be reached; and iv) minimize good practices but also grounded on a sound understanding moral hazard problems by including commitment premiums in of the distinctive needs and situations of businesses in Geor- support policies when possible (i.e., interest rate reductions or gia. Some key principles to deploy this modular approach are cash-back grants based on achieving EE targets). When possi- to i) ensure the enabling regulatory environment and reduce ble, support solutions that can be applied to clusters or groups price distortions; ii) minimize the information gaps and make of firms to promote further the emergence of positive spillovers sure technologies are available at the lowest cost possible (i.e., in terms of information and knowledge sharing. 107 References Greening Firms in Georgia References Abeberese, A. (2017). Electricity Cost and Firm Performance: Evidence from India. The Review of Economics and Sta- tistics, 99(5), 839–852. https://doi.org/10.1162/REST_a_00641 Abramovsky, L., & Simpson, H. (2011). Geographic proximity and firm—university innovation linkages: evidence from Great Britain. Journal of Economic Geography, 11(6), 949–977. Allcott H, Mullainathan S. (2010) “Energy. Behavior and energy policy”. Science. Mar 5;327(5970):1204-5 Alpino, M., Citino, L. and Frigo, A. (2023 ). The effects of the 2021 energy crisis on medium-sized and large industrial firms: Evidence from Italy. Questioni di Economia e Finanza (Occasional Papers) 776, Bank of Italy, Economic Research and International Relations Area. Ang, B. & Zhang, F. (2000). A survey index of index decomposition analysis in energy and environmental studies. En- ergy, 25, 1149-1176. Ang, B. (2015). LMDI decomposition approach: A guide for implementation. Energy Policy, 86, 233-238. Bloom, N. & Van Reenen, J. (2010). Why do management practices differ across firms and countries?. Journal of Eco- nomic Perspectives, 24(1), 203-224. Bloom, N., Brynjolfsson, E., Foster, L., Jarmin, R., Patnaik, M., Saporta-Eksten, I. and Van Reenen, J. (2019). What drives differences in management practices?. American Economic Review 109(5), 1648-1683. Bloom, N., Eifert, B., Mahajan, A., McKenzie, D. and Roberts, J. (2013). Does management matter? Evidence from India. The Quarterly Journal of Economics, 128(1), 1-51. Bloom, N., Sadun, R., and Van Reenen, J. (2012). Americans do IT Better: US multinationals and the Productivity Mir- acle. American Economic Review, 102(1), 167-201. Bloom, N., Sadun, R., and Van Reenen, J. (2017). Management as a technology? Harvard Business School Working Paper 16-133. Bond, S. (2002). Dynamic panel data models: a guide to microdata methods and practice. CeMMAP working papers CWP09/02, Centre for Microdata Methods and Practice, Institute for Fiscal Studies. Brown, J., Crespi, G., Iacovone, L., and Marcolin, L. (2016). Productivity convergence at the firm level: New evidence from the Americas. Understanding the income and efficiency gap in Latin America and the Caribbean, 117-186. https:// doi.org/10.1596/978-1-4648-0450-2_ch5 Burke, M. and Emerick, K. (2016). Adaptation to climate change: Evidence from US agriculture. American Economic Journal, 8(3), 106-140. http://dx.doi.org/10.1257/pol.20130025 Burn-Murdoch, J. (2023, September 23). Economics may take us to net zero all on its own. https://www.ft.com/con- tent/967e1d77-8d3c-4256-9339-6ea7025cd5d3 108 References Technical Report Cali, M., Cantore, N., Iacovone, L., Pereira-López, M. and Presidente, G. (2022). Too much energy: The perverse effect of low fuel prices on firms. Journal of Environmental Economics and Management, 111, 102587. https://doi. org/10.1016/j.jeem.2021.102587 Chorgolashvili, A. (2017). Research about opportunities and barriers to promote RECP financing in Georgia. Presen- tation to OECD workshop on green finance mobilisation in Georgia, Tbilisi, 22-23 June 2017, https://www.slideshare. net/OECD_ENV/session- 6-presentation-by-unido (accessed 29 May 2022). Cirera, X. and Maloney, W. (2017). The innovation paradox: Developing-country capabilities and the unrealized promise of technological catch-up. Washington, DC: World Bank. Cirera, X., Comin, D., Cruz, M. (2022). Bridging the Technological Divide: Technology Adoption by Firms in Developing Countries. The World Bank Productivity Project, Washington, DC. http://hdl.handle.net/10986/37527 De Loecker, J., Eeckhout, J. and Unger, G. (2020). The Rise of Market Power and the Macroeconomic Implications. The Quarterly Journal of Economics, 135(2), 561-644. http://hdl.handle.net/10.1093/qje/qjz041. Elliott, R, Sun, P. and Zhu, T. (2019). Electricity prices and industry switching: Evidence from Chinese manufacturing firms. Energy Economics 78(C), 567–588. 10.1016/j.eneco.2018.11.029 Energy Community Secretariat (2022). Annual Implementation Report. Energy Community, Vienna. https://www.en- ergy-community.org/dam/jcr:3706068e-9a3d-47d5-ae22-f7af0bbd1097/EnC_IR2022.pdf European Bank for Reconstruction and Development (2018). Credit lines – Lending through financial intermediar- ies. EBRD Evaluation Department. https://www.ebrd.com/documents/evaluation/credit-lines-lending-through-finan- cial-intermediaries.pdf European Bank for Reconstruction and Development (2020). Transition report 2019-20. European Bank for Recon- struction and Development, London. https://www.ebrd.com/documents/oce/transition-report-201920-better-gov- ernance.pdf European Patent Office and International Energy Agency (2021). Patents and the Energy Transition – Global Trends in Clean Energy Technology Innovation. Fontagne, L., Martin, P. and Orefice, G. (2023). The many channels of firm’s adjustment to energy shocks: Evidence from France. 77th Economic Policy Panel Meeting, Economic Policy. Ganapati, S., Shapiro, J. and Walker, R. (2020). Energy Cost Pass-Through in U.S. Manufacturing: Estimates and Im- plications for Carbon Taxes. American Economic Journal: Applied Economics, 12(2), 303-42. 10.1257/app.20180474 Government of Georgia (2016). First biennial update report on climate change. Government of Georgia, Tbilisi. Re- trieved from https://unfccc.int/files/national_reports/non-annex_i_parties/ica/application/pdf/first_bur_-_georgia. pdf Government of Georgia. (2021). Georgia’s 2030 Climate Change Strategy and Action Plan. https://policy. asiapacificenergy.org/sites/default/files/Georgia% percent27s% percent202030%I percent20Climate% percen- t20Change% percent20Strategy% percent20% percent28Mitigation% percent29.pdf Greve, H., Lay, J. and Negrete, A. (2023). How vulnerable are small firms to energy price increases? Evidence from Mexico. Environment and Development Economics, 28, 89-109. 10.1017/S1355770X22000080 109 References Greening Firms in Georgia Grover, A. and Karplus, V. (2020). The energy-management nexus in firms. World Bank Policy Research Working Paper 9367. Grover, A., Iacovone, L. and Chakraborty, P. (2019). Management practices in Croatia: Drivers and consequences for firm performance. World Bank Policy Research Working Paper 9067. Grubb et al. (2021). Induced innovation in energy technologies and systems: a review of evidence and potential impli- cations for CO2 mitigation. Environmental Research Letters, 16(4). https://doi.org/10.1088/1748-9326/abde07 Haščič, I. and M. Migotto (2015). Measuring environmental innovation using patent data. OECD Environment Working Papers, 89, OECD Publishing, Paris, https://doi.org/10.1787/5js009kf48xw-en Hassler, J., Krusell, P. and Olovsson, C. (2021). Directed technical change as a response to natural-resource scarcity. Journal of Political Economy, 129(11). 3039-72. https://doi.org/10.1086/715849 Helmers, C. (2019). Choose the Neighbor before the House: Agglomeration Externalities in a UK Science Park. Journal of Economic Geography, 19(1), 31–55, https://doi.org/10.1093/jeg/lbx042 Hsieh, C., and Klenow, P. (2009). Misallocation and manufacturing TFP in China and India. The Quarterly Journal of Economics, 124(4), 1403-1448. Iacovone, L., Maloney, W., McKenzie, D. (2021). Improving management with individual and group-based consulting: Results from a randomized experiment in Colombia. The Review of Economic Studies, 89(1), 346-371. https://doi. org/10.1093/restud/rdab005 International Financial Corporation and Federal Ministry of the Republic of Austria (2021). Climate Finance Mapping, Georgia Country Report. Joussier, R., Martin, J. and Mejean, I. (2023). Cost pass-through and the rise of inflation. Institut national de la statis- que et des études économiques, 2023/13. Levinson, A. (2015). A direct estimate of the technique effect: Changes in the pollution intensity of US manufactur- ing, 1990-2008. Journal of the Association of Environmental and Resource Economists, 2(1), 43-56. https://doi. org/10.1086/680039 Melitz, M. and Polanec, S. (2015). Dynamic Olley-Pakes productivity decomposition with entry and exit. RAND Journal of Economics, 46(2), 362–375. https://doi.org/10.1111/1756-2171.12088 Ministry of Economy and Sustainable Development of Georgia and United Nations Economic and Social Commission for Asia and the Pacific (2022). Penetration of EE and RE technologies in Georgian SMEs. http://www.economy.ge/ uploads/files/2017/energy/2022/case_study_final_4_12_12_2022.pdf Ministry of Environment Protection and Agriculture of Georgia (2021). Fourth National Communication of Georgia to UNFCCC on climate change. https://www.undp.org/georgia/publications/fourth-national-communication-geor- gia-under-united-nations-framework-convention-climate-change# Ministry of Environment Protection and Agriculture of Georgia (2021). National Greenhouse gas inventory. Report of Georgia 1990-2017. Mueller, S. and B. Tuncer (2013), Greening SMEs by Enabling Access to Finance, SWITCH-Asia Network Facility 110 References Technical Report Najdov, E.; Dolidze, M. and Nair, A. (2022). Country Economic Memorandum: Charting Georgia’s Future (English). Washington, D.C.: World Bank Group. OECD (2018). Mobilising Finance for Climate Action in Georgia. Green Finance and Investment, OECD Publishing, Paris. http://dx.doi.org/10.1787/9789264289727-en OECD (2019). Access to Green Finance for SMEs in Georgia. Green Finance and Investment, OECD Publishing, Paris. https://doi.org/10.1787/dc98f97b-en Our World in Data (n.d.). Georgia: CO2 Country Profile. Retrieved April 15th, 2023 from https://ourworldindata.org/ co2/country/georgia# Popp, D. (2002). Induced innovation and energy prices. The American Economic Review, 92(1), 160-180. https://www. jstor.org/stable/3083326 Popp, D. Newell, R. and Adam, J. (2010). Energy, the environment, and technological change. Handbook of the Eco- nomics of innovation, in: Bronwyn H. Hall and Nathan Rosenberg (eds.), Handbook of the Economics of innovation, edition 1, volume 2, chapter 0, 873-937. Regulatory Indicators for Sustainable Energy (2023). Retrieved April 15th, 2023 from https://rise.esmap.org/indica- tors Rentschler, J., M. Kornejew (2017). Energy price variation and competitiveness: Firm level evidence from Indonesia. Energy Economics, 67, 242-254. 10.1016/j.eneco.2017.08.015 Rottner and von Graevenitz. (2021). What drives carbon emissions in German manufacturing: Scale, technique or composition? ZEW Discussion Papers, No. 21-027, ZEW - Leibniz-Zentrum für Europäische Wirtschaftsforschung, Mannheim. http://hdl.handle.net/10419/264291 Rottner and von Graevenitz. ________________________ (2022). Do manufacturing plants respond to exogeneous changes in electricity prices? Evidence from administrative micro-data. ZEW – Centre for European Economic Research Dis- cussion Paper No. 22-038. http://dx.doi.org/10.2139/ssrn.4251564 Sadath, A. and Acharya, R. (2015). Effects of energy price rise on investment: Firm level evidence from Indian manu- facturing sector. Energy Economics, 49(C), 516-522. 10.1016/j.eneco.2015.03.011 Sahin, S. (2017). Firm-level decomposition of energy consumption in Turkish manufacturing industry. Policy Research Working Paper No. 8211. World Bank, Washington, DC. Schweiger, H. and Stepanov, A. (2022). When good managers face bad incentives: Management quality and energy intensity in the presence of price distortions. Energy Policy, 164(2022). https://doi.org/10.1016/j.enpol.2022.112827 Shapiro, J. and Walker, R. (2018). Why is pollution from US manufacturing declining? The role of environmental regu- lation, productivity and trade. American Economic Review, 108(12), 3814-54. DOI: 10.1257/aer.20151272 Siegel, D., Westhead, P., Wright, M. (2003). Assessing the impact of university science parks on research productivity: Exploratory firm-level evidence from the United Kingdom. International Journal of Industrial Organization, 21(9), 1357- 1369. 10.1016/S0167-7187(03)00086-9. 111 References Greening Firms in Georgia Syverson, C. (2011). What determines productivity? Journal of Economic Literature, 49(2), 326-365. 10.1257/ jel.49.2.326 Taylor, R., Govindarajalu, C., Levin, J., Meyer, A. and Ward, W. (2008). Financing energy efficiency: Lessons from Brazil, China, India and beyond. Energy Sector Management Assistance Program (ESMAP), Washington, DC: World Bank. http://hdl.handle.net/10986/6349 United States Environmental Protection Agency (2023). Overview of Greenhouse Gases. https://www.epa.gov/gh- gemissions/overview-greenhouse-gases Wolverton, A., Shadbegian, R. and Wayne, G. (2022). The US manufacturing sector’s response to higher electricity prices: Evidence from state-level renewable portfolio standards. Environmental Economics Working Paper Series, 2022-03. https://www.epa.gov/environmental-economics/us-manufacturing-sectors-response-higher-electrici- ty-prices-evidence-state Xu, X. & Ang, B. (2013). Index decomposition analysis applied to CO2 emission studies. Ecological Economics, 93, 313- 329. 112 Appendix Technical Report Appendix A Additional Analysis This appendix provides more information on the analysis performed. A.1 Energy Quantity Consumption and Carbon Emissions Estimates Table 25A reports energy and CO2 emission conversion ness energy consumption since firms may differ in the price factors for Georgia. Firms report energy expenses broken they face and the energy product requirements, even within down by energy source -electricity, gas, oil and oil products narrowly defined power sources. Still, to our knowledge, there and fossil fuels-. Moreover, they provide electricity and gas is no information on fossil fuels and oil quantity consumption consumption, measured in kilowatt-hours and cubic meters, at the business level and available reported prices are at the respectively. We estimated consumed units using average country level. Overcoming these data limitations will be deci- prices over time for the remaining energy sources for which sive in making a more accurate assessment of energy consump- firms do not report quantity consumption. As for oil prices, tion and emissions in the business sector of Georgia. Second, we use the average annual price of coal in Georgia, reported due to energy consumption misreporting (both expenses and by the IMF (2022) in US dollars and converted into GEL using electricity and gas consumption), the dynamics of aggregate spot exchange rates. Similarly, the fuel price is computed as consumption (emissions) may differ from the estimated one. the average gasoline, kerosene and diesel price reported. We Typically, smaller firms tend to misreport energy consumption then aggregate firm-level energy demand into Terajoule (TJ) information, meaning that although their contribution to ag- unit-equivalent quantity consumption. First, we convert orig- gregate consumption (emissions) is likely small, it could affect inal quantities into ‘converted’ quantities (column 4, Table country-level figures’ dynamics. Finally, only a subset of firms 25A) to use net calorific value factors that enables the direct that use electricity and gas report consumption quantities. For conversion of different energy units into TJs as reported by the firms that report electricity and gas expenses but do not pro- Intergovernmental Panel on Climate Change (IPCC) Guidelines vide quantity consumption, we use average electricity and gas for National Greenhouse Gas Inventories for Georgia (2006). prices at the sector-by-size level to estimate their consumption Finally, we translate TJ consumption into (metric tons of) car- based on reported expenses. bon (CO2) emissions per Terajoule. TJ to CO2 conversion factors are reported in the last column of Table 25A. Aggregate energy value and quantity consumption and CO2 emissions, indexed to 2015, are plotted in Figure 49A. Energy consumption and emissions estimates are subject Energy quantity consumption and CO2 emissions performed to some caveats. First, we acknowledge that using average fos- similarly over 2016-2021, while energy value consumption grew sil fuels and coal prices in Georgia may not reflect actual busi- above due to price changes in energy bills. TABLE 25A Energy and CO2 Emission Conversion Factors Energy source Value Quantity Quantity Net calorific Emissions (original) (converted) value factor (CO2 metric (TJ) tons per TJ) Electricity GEL kWh mWh 0.0036 100.55 Gas GEL m3 000 m 3 0.0035 56.10 Oil and oil products GEL liters 000 kg 0.0440 73.64 Fossil fuels GEL 000 kg 000 kg 0.0189 96.10 Notes: CO2 = carbon dioxide; kg = kilogram; kWh = kilowatt-hour; m3 = cubic meters; mWh = megawatt-hour; TJ = terajoule. Source: World Bank elaboration based on GEOSTAT and IPCC Guidelines for National Greenhouse Gas Inventories (2006). 113 Appendix Greening Firms in Georgia FIGURE 49A Firm Energy Consumption and CO2 Emissions versus Changes in the Supply of Energy CO2 emissions Energy consumption (quantity) Energy consumption (value) Index, 2015 = 1 Energy source share 2.0 100% 1.8 Share of oil and fossil fuels 1.6 80% 1.4 1.2 60% 1.0 0.8 40% 0.6 0.4 Share of gas 20% 0.2 Share of electricity 0.0 0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Source: World Bank elaboration based on GEOSTAT. A.2 Descriptive Statistics of Business in Georgia Table 26A presents a snapshot of the Georgian busi- Figure 50A presents the share of each energy source in ness sector covered by GEOSTAT by size class and sector total energy costs at the sector level based on firm-level (two-digit of NACE Rev. 1). Columns 2 to 7 provide key descrip- reported energy bills. Oil, oil products and fossil fuels appear tive metrics such as firm counts (‘firms’), the average size as the to be the Georgian private sector’s main energy source. As of mean number of employees (‘employment’), the average sales 2021, oil and oil-related products accounted for 42 percent of in GEL of 2021 (‘sales’), the average and coefficient of variation total energy expenses, whereas fossil fuels accounted for 8 of the sales-to-energy ratio (the key input for measuring EE), percent. Transport activities, construction, mining, utilities, the unweighted average TFP, and the number firms reporting wholesale trade, and manufacturing industries such as textiles energy expenses as a fraction of total firms. There are large dif- and production of coke and wood heavily rely on oil and fossil ferences in the reported metrics across size and sector groups. fuel energy sources. But electricity represents 38 percent of The average large firm is 29 times larger and generates revenues energy consumption, being particularly relevant for specific nine times higher than the average SME. Also, large firms are, manufacturing industries like chemicals, plastics, machinery, on average, less energy efficient than SMEs. Their revenue is and furniture but also for retail trade, the hospitality indus- 46 percent lower per unit of energy consumed. Finally, they try and services (e.g., research and development, health, and appear to be slightly more efficient overall. education). Finally, gas energy, which represents 11 percent of total energy bills. Gas is particularly relevant for utilities, paper products, education, specific mining activities, health and hotel and restaurants. 114 Appendix Technical Report TABLE 26A Descriptive Statistics of Georgian Firms in 2021 Firms Employment Sales (‘000) Sales-to- CV Sales-to- TFP Share firms energy ratio energy ratio reporting energy Panel A. Size class SME (1-99) 13,186 12.07 2,943 257.38 2.54 2.63 0.55 Large (+100) 877 353.46 27,247 138.28 2.76 2.66 0.89 Panel B. Sector Agriculture, hunting, rel. act. 407 29.92 2,224 136.00 2.26 2.87 0.72 Forestry, logging, rel. act. 59 5.05 390 65.93 1.22 3.69 0.75 Fishing 87 20.79 1,379 217.59 3.94 3.43 0.46 Mining of coal and lignite 8 14.88 5,050 102.30 1.41 3.82 0.75 Extraction of petroleum and gas 28 15.75 2,284 41.90 1.52 3.30 0.61 Mining of metal ores 66 45.61 4,127 11.76 1.19 3.77 0.91 Other mining and quarrying 214 13.09 1,069 47.99 4.37 3.08 0.81 Food products and beverages 819 33.81 5,098 81.09 3.64 1.74 0.84 Tobacco products 7 45.29 7,581 249.10 1.53 1.80 0.71 Textiles 67 19.46 1,345 55.53 1.23 2.35 0.63 Wearing apparel 77 116.68 3,568 54.52 1.45 2.51 0.71 Tanning and dressing of leather 50 8.82 1,954 206.23 3.75 2.07 0.76 Wood and wood products 353 18.38 4,062 378.91 1.81 1.79 0.59 Pulp, paper and paper products 92 20.12 2,315 292.21 2.55 2.79 0.79 Publishing and printing 121 20.82 1,488 172.07 2.60 2.63 0.59 Coke and refined petroleum 12 12.25 6,625 41.16 0.97 1.44 0.92 Chemicals 106 54.44 5,438 98.74 2.07 2.76 0.84 Rubber and plastic products 220 12.31 2,402 97.88 1.82 1.80 0.75 Non-metallic minerals 367 19.45 3,248 94.76 4.64 1.78 0.82 Basic metals 60 86.05 14,204 102.03 4.02 2.35 0.77 Fabricated metal products 180 23.66 2,790 146.37 1.45 2.35 0.69 Machinery and equipment 97 15.73 2,562 78.31 2.20 3.44 0.54 Office machinery and computers 38 22.45 2,730 83.07 1.12 n.a. 0.24 Electrical machinery 23 20.26 5,084 127.16 0.84 3.64 0.65 Radio, TV and comms equip. 10 8.40 304 62.30 0.59 n.a. 0.50 Medical, precision and optical inst. 11 11.45 970 292.24 1.61 5.37 0.55 Motor vehicles 4 4.50 174 92.51 1.29 3.27 0.75 115 Appendix Greening Firms in Georgia TABLE 26A Descriptive Statistics of Georgian Firms in 2021 (cont.) Firms Employment Sales (‘000) Sales-to- CV Sales-to- TFP Share firms energy ratio energy ratio reporting energy Transport equipment 39 29.64 1,484 81.52 1.65 4.41 0.72 Furniture 143 15.23 1,308 109.51 1.19 2.25 0.71 Recycling 9 5.78 572 60.72 1.17 n.a. 0.33 Electricity, gas, steam, hot water 81 171.68 21,694 480.07 2.03 4.33 0.74 Collection, purification and dist. 12 366.25 8,922 4.16 0.48 2.87 1.00 Construction 890 39.74 5,501 209.18 2.78 3.49 0.68 Sale and repair of vehicles 554 23.45 6,347 391.28 1.76 2.35 0.44 Wholesale trade 2,711 22.60 7,554 587.70 1.73 1.65 0.42 Retail trade 2,434 22.80 2,695 314.26 2.01 1.65 0.45 Hotel and restaurants 449 39.74 2,548 63.67 4.31 2.13 0.71 Land transport 511 10.57 1,168 76.14 6.16 2.89 0.77 Water transport 4 5.50 42,868 3.38 0.08 n.a. 0.75 Air transport 16 61.38 29,575 143.79 1.76 n.a. 0.69 Auxiliary transport act. 329 37.97 10,294 752.41 1.56 3.89 0.53 Post and telecommunications 153 88.94 3,863 169.48 2.14 2.87 0.51 Real estate 286 14.57 2,715 171.07 2.21 4.68 0.48 Renting of machinery 87 10.67 1,656 76.17 1.53 3.94 0.44 Computer activities 185 23.13 2,391 290.53 1.65 4.20 0.35 Research and development 24 19.92 1,826 654.20 2.08 4.64 0.42 Other business activities 581 45.82 2,415 268.99 2.38 4.41 0.44 Education 197 65.55 2,029 127.89 4.91 3.81 0.66 Health and social work 414 161.91 3,402 81.73 2.42 3.08 0.76 Sewage and refuse disposal 39 60.67 1,569 16.22 1.30 2.97 0.69 Leisure act. 237 55.71 7,897 112.36 1.71 3.76 0.49 Other services 95 15.17 571 57.44 3.47 3.09 0.61 Notes: ‘firms’ is firm counts; ‘employment’ is the average number of employed persons; ‘sales’ is the average sales value per firm (‘000 GEL); ‘sales-to-energy ratio’ is the average energy efficiency (before taking logs) and the CV ‘sales-to-energy ratio’ is the coefficient of variation; TFP is the (logged) total factor productivity; ‘share of firms reporting energy’ is the fraction of firms that report energy expenses over the total number of firms. Group of activities are displayed at the two-digit of NACE for presentation purposes. Source: World Bank elaboration based on GEOSTAT. 116 Appendix Technical Report FIGURE 50A Share of Each Energy Source in Firms’ Total Energy Costs in 2021 Electricity Gas Oil and oil products Solid fossil fuels Renewables and waste 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% AGRICULTURE, HUNTING, REL. ACT. FORESTRY, LOGGING, REL. ACT. FISHING MINING OF COAL AND LIGNITE EXTRACTION OF PETROLEUM AND GAS MINING OF METAL ORES OTHER MINING AND QUARRYING FOOD PRODUCTS AND BEVERAGES TOBACCO PRODUCTS TEXTILES WEARING APPAREL TANNING AND DRESSING OF LEATHER WOOD AND WOOD PRODUCTS PULP, PAPER AND PAPER PRODUCTS PUBLISHING AND PRINTING COKE AND REFINED PETROLEUM CHEMICALS RUBBER AND PLASTIC PRODUCTS NON METALLIC MINERALS BASIC METALS FABRICATED METAL PRODUCTS MACHINERY AND EQUIPMENT OFFICE MACHINERY AND COMPUTERS ELECTRICAL MACHINERY RADIO, TV AND COMMS EQUIP. MEDICAL, PRECISION AND OPTICAL INST. MOTOR VEHICLES TRANSPORT EQUIPMENT FURNITURE RECYCLING ELECTRICITY, GAS, STEAM, HOT WATER COLLECTION, PURICATION AND DIST. CONSTRUCTION SALE AND REPAIR OF VEHICLES WHOLESALE TRADE RETAIL TRADE HOTEL AND RESTAURANTS LAND TRANSPORT WATER TRANSPORT AIR TRANSPORT AUXILIARY TRANSPORT ACT. POST AND TELECOMMUNICATIONS REAL ESTATE RENTING OF MACHINERY COMPUTER ACTIVITIES RESEARCH AND DEVELOPMENT OTHER BUSINESS ACT. EDUCATION HEALTH AND SOCIAL WORK SEWAGE AND REFUSE DISPOSAL LEISURE ACT. OTHER SERVICES Notes: energy costs for heating, fuel or receive energy in 2021. Natural gas includes liquefied gas; Oil and oil products comprises kerosene, gasoline, aviation gasoline, diesel, bitumen and similar products; Solid fossil fuels include coal, lignite, peat, coke, and coke-oven gas; Renewables and waste include energy produced from renewable and waste (wood, charcoal, industrial and municipal waste, biomass, solar, wind, etcetera). 117 Appendix Greening Firms in Georgia A.3 Extended Analysis FIGURE 51A Olley-Pakes Static Decomposition of the TFP Growth 15% Unweighted TFP Market reallocation Agg. change in TFP 10% 5% 0% -5% -10% -15% Source: World Bank -20% elaboration based on 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 GEOSTAT. Figure 52A plots the distribution of electricity and gas prices paid by Georgian firms. Prices are expressed in GEL per 100 kWh and 100 m3 consumed. Both charts show that mar- ket operators adjusted electricity and gas prices over time, but only in 2021 was there a discrete jump in the price of electricity, in contrast to the gas charge. FIGURE 52A Distribution of Electricity and Gas Prices Paid by Georgian Firms PANEL A Electricity price PANEL B Gas price 2013 2014 2015 2017 2016 2018 2019 2021 2020 2013 2014 2015 2017 2016 2018 2019 2021 2020 .6 Density .12 .5 .1 .4 .08 .3 .06 .04 .2 .02 .1 0 10 15 20 25 30 50 55 60 65 70 75 80 85 90 95 100 105 GEL per 100 KWh GEL per 100 m3 Notes: Predicted electricity and gas prices paid after accounting for municipality, industry (4-digit of NACE) and firm size class effects. Source: World Bank elaboration based on GEOSTAT. 118 Appendix Technical Report TABLE 27A Carbon and Energy Efficiency Dep: CO2 efficiency Energy value efficiency Energy quantity efficiency (1) (2) (3) (5) (6) (7) Energy efficiency 1.020*** 0.941*** 0.929*** 1.049*** 1.048*** 1.067*** (0.002) (0.005) (0.006) (0.001) (0.001) (0.003) Size class controls No Yes Yes No Yes Yes Firm FE No No Yes No No Yes Region FE No Yes No No Yes No Industry FE No Yes No No Yes No Year effects No Yes Yes No Yes Yes R-squared 0.829 0.819 0.612 0.958 0.977 0.957 Observations 60,377 24,366 60,377 60,570 40,641 60,570 No. firms 28,017 28,160 Source: World Bank elaboration based on GEOSTAT. FIGURE 53A Change in Carbon Efficiency Due to Changes in Energy Efficiency PANEL A Energy value e ciency PANEL B Energy quantity e ciency 1.1 1.1 1.05 1.08 1 1.06 .95 1.04 .9 .85 1.02 Notes: point estimates from FE regressions. confidence .8 intervals displayed at the 1 95 percent. Source: World 2014 2015 2016 2017 2018 2019 2020 2021 Bank elaboration based on 2014 2015 2016 2017 2018 2019 2020 2021 GEOSTAT. 119 Appendix Greening Firms in Georgia A.4 Correlates of Energy Efficiency and Convergence a. Correlates of Energy Efficiency in High- and in fixed assets and R&D (whether the firm invests or not, 1 and Low-Efficiency Sectors 0 respectively) are positive and statistically correlated to EE, while (linearly) uncorrelated as for low-efficiency groups. As Following the analysis of the correlates of EE, Figure 54A for investment amounts, the ICT adoption score and the ex- plots the coefficient estimates of the two-step procedure porting status are all positive and statistically correlated to EE, introduced in Section 5. The specification conditions on irrespective of the baseline level of efficiency of the sector (al- whether the EE of the sector-by-size group is above (‘high-ef- though estimates tend to be larger for high-efficiency groups). ficiency’) or below (‘low-efficiency’) the median. Results show In this regard, firm digitalization and lower trade barriers could that average outcome differences are small across groups (ex- contribute to reducing energy consumption and boosting firm cept for innovation and exporters). Estimated coefficients asso- performance through TFP growth. ciated with high-efficiency groups for the status of investment FIGURE 54A The Relationship between Energy Efficiency and Investment, Innovation, ICT, and Exporting High-e ciency Low-e ciency Invests Growth in capital stock Invests in R&D Innovates ICT z-score Exporter -.2 -.1 0 .1 .2 .3 .4 .5 .6 Gap relative to baseline Notes: Dots depict point estimates and brackets confidence intervals (5 percent level of significance). All variables are binary (No = 0, Yes = 1) except for log investment and ICT adoption z-score. Growth in capital stock is the first difference of the logged capital stock (conditional on being positive). Low-intensive and high-intensive sectors are classified according to the energy-to-revenue ratio below or above the median. Source: World Bank calculations based on GEOSTAT. b. Allocative Efficiency and the Tbilisi Effect Convergence results seen in Table 16 could be due to al- exists among stayers and that the change in the frontier also locative effects, reflecting improvements in EE due to the contributes positively to enhancing efficiency. In columns 5 least efficient firms dropping out of the market. Firms and 6, the estimated coefficients associated with the sectoral could converge to the frontier due to firm entry and exit rath- frontier are non-significant. Still, those related to the local and er than incumbents converge to the frontier due to EE gains. regional frontier are, and the change in the local and region- To address this potential concern, Table 28A looks at ‘stayers’ al frontiers also contributes to improving efficiency. Also, all only (i.e., firms active in all years of 2007-2019). Results re- specifications show a robust positive correlation between TFP main similar to the ones in Table 16, despite having a smaller and EE changes. sample of firms. Columns 1 to 4 suggest that convergence still 120 Appendix Technical Report TABLE 28A Robustness Check: Energy Efficiency Convergence in ‘Stayers’ OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) D(local frontier) 0.226*** 0.579*** 0.131*** 0.193*** 0.131*** 0.176*** (0.009) (0.020) (0.016) (0.012) (0.017) (0.011) D(sector frontier) -0.005 0.033 -0.010 0.030 (0.013) (0.027) (0.012) (0.022) D(regional frontier) 0.104*** 0.462*** 0.110*** 0.489*** (0.016) (0.035) (0.014) (0.022) Delta local frontier 0.216*** 0.362*** 0.171*** 0.166*** 0.165*** 0.132*** (0.014) (0.012) (0.017) (0.015) (0.021) (0.017) Delta sector frontier -0.004 0.037* (0.015) (0.019) Delta regional frontier 0.119* 0.326*** (0.065) (0.051) Delta TFP 0.402*** 0.322*** 0.410*** 0.304*** 0.409*** 0.298*** (0.034) (0.028) (0.034) (0.025) (0.034) (0.025) Large (t-1) 0.091*** 0.545*** -0.022 0.165*** -0.026 0.142*** (0.015) (0.063) (0.030) (0.060) (0.027) (0.045) Start-up (t-1) -0.018* -0.045*** -0.179*** -0.065** -0.177*** -0.061* (0.011) (0.013) (0.027) (0.031) (0.027) (0.031) Foreign-owned private (t-1) 0.014 0.017 0.130** 0.055 0.131** 0.067 (0.016) (0.053) (0.053) (0.060) (0.053) (0.056) State-owned (t-1) -0.047 -0.021 -0.147 -0.268** -0.164 -0.254** (0.037) (0.050) (0.163) (0.105) (0.168) (0.109) Self-governed (t-1) 0.020 0.186 0.116 0.289*** 0.142 0.340** (0.077) (0.231) (0.083) (0.097) (0.086) (0.127) Start-up (t-1) x D(local frontier) 0.104*** 0.034** 0.104*** 0.034*** (0.014) (0.013) (0.014) (0.012) Large (t-1) x D(local frontier) 0.003 -0.044*** 0.002 -0.046*** (0.007) (0.013) (0.007) (0.015) Foreign-owned private (t-1) x -0.060*** -0.002 -0.061*** -0.007 D(local frontier) (0.020) (0.038) (0.020) (0.039) 121 Appendix Greening Firms in Georgia TABLE 28A Robustness Check: Energy Efficiency Convergence in ‘Stayers’ (cont.) State-owned (t-1) x D(local frontier) 0.055 0.146*** 0.058 0.141*** (0.056) (0.028) (0.058) (0.031) Self-governed (t-1) x D(local -0.065** -0.069 -0.072** -0.076 frontier) (0.030) (0.043) (0.030) (0.047) Constant -0.422*** -0.959*** -0.531*** -1.078*** -0.534*** -1.105*** (0.108) (0.021) (0.099) (0.020) (0.098) (0.025) Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No Yes No Yes No Industry FE (two-digit) Yes No Yes No Yes No Firm FE No Yes No Yes No Yes R-squared 0.134 0.315 0.151 0.327 0.161 0.383 Observations 10,236 10,236 8,870 8,870 8,370 8,370 No. firms 1,028 1,008 992 No. clusters 39 39 37 37 37 37 Because of the dominance of Tbilisi in the economy, it is im- local frontier is country-wide instead driven by a specific area. portant to assess whether convergence and spillover effects However, estimates associated with the interaction between are nationwide or circumscribed to Tbilisi from a public Tbilisi and the distance to the local frontier – × policy viewpoint. Hence, Table 29A addresses this potential - are also positive and significant, suggesting that with- concern by interacting the Tbilisi region with the three defini- in-sector agglomeration effects are stronger in Tbilisi compared tions of frontiers. If Tbilisi were the only driver of convergence, to the rest of the country. In particular, interaction estimates as- we would expect positive and statistically significant estimates sociated with the regional and sector frontiers - only for Tbilisi interacted variables (weak or no significance × and × - show interesting conclu- among non-interacted variables). In contrast, the lack of sig- sions. According to the evidence, spatial concentration around nificance of the Tbilisi interacted estimates would point to no Tbilisi is not driving efficiency catch-up. Estimated coefficients differential effects. Overall, there is no evidence that Tbilisi is the associated with the regional frontier – - are only driver of EE convergence. However, it can change the speed between 0.098-0.430, while those related to the Tbilisi interac- of the process (either by speeding it up or slowing it down). The tion are small and not statistically significant. So, we reject the coefficient estimates associated with the interaction between hypothesis of differential effects. In terms of sector spillovers Tbilisi and the distance to the local frontier - - , OLS and FE estimates are positive (0.013- × - reported in columns (5) and (6) are positive and sig- 0.178) but capturing the differential Tbilisi - × nificant (both in the OLS and FE specifications). This means a - are negative, ranging between -0.033 and -0.166. In other higher convergence speed among Tbilisi-based firms but does words, convergence to the sector efficiency in Tbilisi is close to not imply the absence of convergence beyond the capital city zero , slightly of Georgia. Indeed, estimated coefficients associated with the different to what happens in the rest of the country. In sum, distance to the local frontier - - are economically according to the evidence available, spatial concentration and and statistically positive, meaning that the convergence to the sector linkages around Tbilisi do not drive efficiency catch-up. 122 Appendix Technical Report TABLE 29A Robustness Check: Energy Efficiency Con- vergence and the Tbilisi Effect OLS (1) FE (2) OLS (3) FE (4) OLS (5) FE (6) Tbilisi -0.101*** -0.087** -0.037 D(local frontier) 0.251*** 0.715*** 0.234*** 0.682*** 0.123*** 0.199*** D(local frontier) x Tbilisi -0.002 0.027 0.008 0.051 0.041** 0.071** D(sector frontier) 0.013 0.178** D(sector frontier) x Tbilisi -0.033* -0.166** D(regional frontier) 0.098*** 0.430*** D(regional frontier) x Tbilisi -0.009* 0.092 Delta local frontier 0.285*** 0.466*** 0.263*** 0.437*** 0.200*** 0.165*** Delta local frontier x Tbilisi -0.028* -0.015 -0.013 0.006 0.027 0.035 Delta sector frontier 0.065* 0.117* Delta sector frontier x Tbilisi -0.086** -0.105 Delta region frontier 0.096** 0.278*** Delta region frontier x Tbilisi 0.084 0.122** Delta TFP 0.490*** 0.353*** 0.488*** 0.316*** Delta TFP x Tbilisi -0.060* -0.035 -0.051 -0.016 Large (t-1) 0.068*** 0.577*** 0.085*** 0.554*** -0.018 0.181*** Start-up (t-1) 0.015* 0.001 0.003 -0.020** -0.062*** -0.030* Foreign-owned private (t-1) 0.022* 0.039*** 0.021** 0.058*** 0.033 0.066 State-owned (t-1) -0.131*** -0.145* -0.050* -0.005 0.035 -0.085 Self-governed (t-1) -0.001 0.020 0.119 0.068 0.340*** 0.171 Start-up (t-1) x D(local frontier) 0.042*** 0.022** Large (t-1) x D(local frontier) 0.016*** -0.047*** Foreign-owned private (t-1) x D(local frontier) -0.005 0.008 State (t-1) x D(local frontier) -0.051*** 0.039 Self-governed (t-1) x D(local frontier) -0.091** -0.052 Constant -0.285*** -1.048*** -0.248*** -0.985*** -0.449*** -1.197*** Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No No No Yes No Industry FE (two-digit) Yes No No No Yes No Firm FE No Yes No Yes No Yes R-squared 0.143 0.383 0.155 0.391 0.163 0.434 Observations 35,405 35,405 26,862 26,862 25,951 25,951 No. firms 10,869 8,050 7,951 No. clusters 60 60 55 55 55 55 Notes: Due to presentation purposes, we only report estimated coefficients. (*), (**) and (***) indicate significance at (10), (5) and (1) percent level. Tbilisi is a dummy variable that equals 1 if the region is Tbilisi and 0 otherwise. All notes in Table 28 apply. Source: World Bank elaboration based on GEOSTAT. 123 Appendix Greening Firms in Georgia c. Convergence by Main Energy Source An additional robustness check performs the same anal- ysis defining EE as the cost of the most important energy input at the sector level relative to sales. Table 30A reports regression results for electricity- and oil and oil products-in- tensive sectors. Conclusions are like those reported in Table 16 and consistent with several robustness checks performed in this report. TABLE 30A Robustness Check: Energy Efficiency Convergence by Main Energy Source Electricity Gas Oil and oil products OLS (1) FE (1) OLS (1) FE (1) OLS (2) FE (2) D(local frontier) 0.191*** 0.179*** 0.232*** 0.273*** 0.067*** 0.145*** (0.018) (0.038) (0.040) (0.043) (0.012) (0.028) D(sector frontier) 0.026** 0.086** -0.008 0.040 0.033 0.148** (0.011) (0.040) (0.085) (0.132) (0.025) (0.061) D(regional frontier) 0.002 0.576*** 0.068 0.500*** 0.231*** 0.572*** (0.020) (0.038) (0.099) (0.101) (0.015) (0.036) Delta local frontier 0.254*** 0.117*** 0.230*** 0.180*** 0.136*** 0.077*** (0.022) (0.026) (0.046) (0.043) (0.015) (0.018) Delta sector frontier 0.037*** 0.067* 0.043 0.028 0.019 0.066*** (0.012) (0.039) (0.065) (0.089) (0.017) (0.022) Delta region frontier 0.049 0.352*** 0.037 0.241*** 0.176** 0.317*** (0.034) (0.041) (0.085) (0.080) (0.070) (0.065) Delta TFP 0.382*** 0.259*** 0.410*** 0.237*** 0.476*** 0.323*** (0.064) (0.041) (0.055) (0.044) (0.030) (0.013) Constant -0.116 -0.821*** -1.185*** -1.408*** -0.741*** – (0.100) (0.056) (0.211) (0.043) (0.088) (0.031) Size, ownership and age class Yes Yes Yes Yes Yes Yes Ownership × D(frontier) controls Yes Yes Yes Yes Yes Yes Size × D(frontier) controls Yes Yes Yes Yes Yes Yes Age × D(frontier) controls Yes Yes Yes Yes Yes Yes Year effects Yes Yes Yes Yes Yes Yes Geographic FE (municipality) Yes No Yes No Yes No Industry FE (two-digit) Yes No Yes No Yes No 124 Appendix Technical Report TABLE 30A Robustness Check: Energy Efficiency Convergence by Main Energy Source (cont.) Firm FE No Yes No Yes No Yes R-squared 0.184 0.525 0.257 0.587 0.268 0.628 Observations 11,004 11,004 2,473 2,473 10,213 10,213 No. firms 4,380 1,233 3,625 No. clusters 49 49 35 35 50 50 Notes: Due to presentation purposes, we only report estimated coefficients. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Energy efficiency is calculated at the firm level using the main energy input at the sector level. All notes in Table 28 apply. Source: World Bank elaboration based on GEOSTAT. 102 Comparable countries are Albania, Azerbaijan, Bulgaria, Croatia, Czech A.5 Green Management: Additional Analysis Republic, Estonia, Hungary, Georgian establishments display lower quality of green of comparable countries102 and the vertical red line shows the Kyrgyz Republic, Lithuania, North Macedonia, Serbia, management. Figure 55A estimates the distribution of the average of the green management z-score. A larger number of Slovenia, Türkiye, Ukraine, and green management z-score, which indicates the quality of firms in Georgia is at the left of the red line compared to peers, Poland. Source: Word Bank green management practices across Georgian establishments. so green management quality is lower in Georgia relative to elaboration based on WBES. The distribution is compared against the average distribution comparable countries. FIGURE 55A Green Management Score in Georgia and Peer Countries .6 Georgia .55 .5 .45 .4 .35 .3 .25 .2 .15 Comparable countries .1 .05 0 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 Notes: Differences in Georgia Green Management scores emerge from pooling cross-country data and standardizing the green management z-score at the cross-country Green Management Z-score level. Comparable countries are Albania, Azerbaijan, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Kyrgyz Republic, Lithuania, North Macedonia, Serbia, Slovenia, Türkiye, Ukraine, and Poland. The vertical red line is the green management z-score average. Source: Word Bank elaboration based on WBES. Finally, Table 31A shows the positive relation between EE and labor productivity (proxied as sales per worker). 125 Appendix Greening Firms in Georgia TABLE 31A Firm Energy Efficiency, Green Management, and Labor Productivity OLS (1) OLS (2) OLS (3) OLS (4) OLS (5) Log sales per worker 0.515*** 0.504*** 0.455*** 0.657*** (0.049) (0.051) (0.073) (0.069) Green Management z-score -0.061 -0.108*** -0.133*** -0.136*** -0.171*** (0.039) (0.041) (0.038) (0.045) (0.056) Large firms (>99 emp.) -0.067 -0.114 0.355 -0.427* (0.189) (0.185) (0.281) (0.232) Mature (>5 years) -0.038 -0.054 0.010 -0.008 (0.137) (0.137) (0.219) (0.180) Foreign investors own >25 percent 0.109 -0.270 0.107 shares (0.152) (0.253) (0.197) Firm exports 0.139 0.444** -0.264 (0.129) (0.221) (0.185) Customer pressure (Yes = 1) 0.188 -0.000 0.518 (0.212) (0.221) (0.578) Energy tax/levy 0.281** 0.084 0.387** (0.126) (0.191) (0.169) Constant -5.396*** -10.985*** -11.102*** -10.454*** -12.463*** (0.199) (0.582) (0.599) (0.866) (0.787) Geographic FE Yes Yes Yes Yes Yes Industry FE Yes Yes Yes No No Industries All All All Manufacturing Services R-squared 0.210 0.380 0.387 0.295 0.396 Observations 435 431 430 180 250 Notes: Robust standard errors in parentheses. (*), (**) and (***) indicate significance level at (10), (5) and (1) percent level. Left-hand side variable is energy efficiency. Source: World Bank calculations based on Enterprise Surveys. 126 Appendix Technical Report Appendix B Data Quality Checks and Investing in Green Technology B.1 Data Quality Checks Table 32B shows large differences in the distribution of tween groups. This means that using the matched database for firms between the group that matched and those that did analysis can provide unbiased results that are not biased due to not. However, Table 33B shows that after controlling for size the construction of the sample. Still, any extrapolation to the and sector, there are no differences in technology adoption, entire population of firms needs to consider the differences in management quality, or adoption of green technologies be- distribution in Table 5. 127 Appendix Greening Firms in Georgia TABLE 32B Differences in Matching Survey and Administrative Data Mean non-matched Mean matched Diff p-value A. Size class Small (5-19) 0.665 0.267 0.398 0.000 Medium (20-99) 0.268 0.400 -0.132 0.000 Large (100+) 0.067 0.334 -0.267 0.000 B. Sector Agriculture 0.098 0.061 0.037 0.004 Livestock 0.039 0.030 0.009 0.285 Food processing 0.098 0.125 -0.027 0.087 Apparel 0.073 0.034 0.039 0.000 Man. of motor vehicles 0.002 0.001 0.001 0.675 Pharmaceuticals 0.012 0.022 -0.009 0.162 Wholesale and retail 0.076 0.164 -0.088 0.000 Financial services 0.023 0.000 0.023 0.000 Land transport 0.090 0.066 0.024 0.061 Health services 0.061 0.116 -0.055 0.000 Leather products 0.008 0.007 0.001 0.841 Accommodations 0.157 0.069 0.089 0.000 Other manufacturing 0.083 0.168 -0.085 0.000 Other services 0.179 0.139 0.040 0.024 Source: World Bank elaboration based on FAT surveys. TABLE 33B Biases in Technology Adoption (1) (2) (3) (4) Variables Tech. adoption Tech. adoption Managerial Number of index GBF index GBF quality index green tech. (extensive) (intensive) adopted Matched firms FAT-GEOSTAT 0.003 0.035 0.000 0.052 (0.038) (0.028) (0.013) (0.074) Size and sector controls Yes Yes Yes Yes Observations 1,793 1,793 1,793 1,793 R-squared 0.175 0.139 0.112 0.069 Notes: Robust standard errors in parentheses. (*), (**) and (***) Source: World Bank calculations based on indicate significance level at (10), (5) and (1) percent level. FAT surveys. 128 Appendix Technical Report B.2 Investing in Green Technology Green management is part of the wider toolbox for firms Table 34B also reports the fraction of establishments that to improve EE and reduce their environmental impact. invest in different types of pure and mixed green technol- Firms may invest in measures that favor greening. Sometimes, ogies in benchmark countries. While the share of firms that reductions in environmental impact are by-products of achiev- invest in mixed technologies is average in Georgia, it ranks at ing other goals (EBRD, 2020). For example, new machinery and the bottom of the distribution regarding pure technologies. vehicle engines consume less energy and reduce gas emissions The fraction of firms investing in on-site green energy gen- compared to old technology. Also, new technologies tend to eration, energy management, measures to enhance EE, and use greener power sources such as electricity instead of fos- waste minimization is particularly low across Georgian estab- sil fuels. In other cases, firms explicitly target environmental lishments when compared to regional and aspirational Eastern impact and invest in pure green technologies to generate their EU countries. own green energy, improve energy and water management, minimize waste, or increase recycling. In addition to price and regulation incentives, access to green credit among SMEs could also explain lower invest- Consistent with earlier results on low energy prices and ment levels. For example, the OECD (2019) points to the exist- green management adoption, investment efforts are de- ing gap in the green credit market that excludes SMEs from ac- voted toward mixed rather gran pure green technologies. cessing loans at competitive conditions. According to the OECD Between 40 and 50 percent of the surveyed firms report in- (2019), financial institutions tend to offer green credit lines that vesting in mixed green technologies, such as improvements on are larger than might be required for a typical investment in heating, cooling or lighting systems or machinery and vehicle EE investment. On the other side of the market, organizations upgrades. In contrast, nearly 20 percent of the firms invest in granting micro-credits provide credit lines that are too small for the management of energy and water each, but other invest- the typical SME and at high interest rates. Hence, this gap in the ments that may require larger funding or reveal a strong com- green credit market might hinder energy-saving investments mitment of the firm to environmental impact (e.g., on-site green for many enterprises. energy generation) have a lower take-up. TABLE 34B Fraction of Firms Investing in Green Technologies across Countries GEO BGR SVN MKD HUN POL SRB CZE AZE Panel A. Pure green technology On-site green energy generation 0.08 0.23 0.20 0.16 0.12 0.11 0.09 0.08 0.05 Energy management 0.21 0.33 0.36 0.31 0.39 0.30 0.28 0.27 0.18 Waste min., recycling & waste 0.27 0.44 0.70 0.35 0.49 0.44 0.48 0.63 0.33 mgmt. Air pollution control measures 0.19 0.24 0.20 0.21 0.11 0.12 0.15 0.16 0.11 Water management 0.23 0.19 0.25 0.21 0.13 0.14 0.11 0.21 0.13 Other pollution control measures 0.11 0.16 0.13 0.16 0.11 0.09 0.10 0.08 0.12 Adopts measures to enhance 0.18 0.17 0.46 0.41 0.42 0.22 0.40 0.33 0.00 energy efficiency Panel B. Mixed green technology Heating & cooling system 0.42 0.43 0.56 0.40 0.30 0.29 0.42 0.55 0.36 improvement Machinery upgrades 0.45 0.47 0.54 0.40 0.62 0.48 0.40 0.67 0.32 Vehicle upgrades 0.42 0.36 0.28 0.28 0.44 0.29 0.37 0.68 0.18 Lighting system improvement 0.52 0.34 0.53 0.42 0.49 0.36 0.57 0.61 0.40 Notes: Column names refer to country codes: Georgia; POL: Poland; SRB: Serbia; AZE: Azerbaijan; MKD: North Macedonia; CZE: Czech Republic; HUN: Hungary; SVN: Slovenia; BGR: Bulgaria. Source: World Bank elaboration based on WBES. 129 Appendix Greening Firms in Georgia Green technology adoption is positively correlated to the between the management z-score and the green technology quality of management. To account for all possible invest- investment score after controlling for firm size, age, industry, ments in green technology, the green investment score is the and region. Besides the importance of managerial practices on firm-level average of 0 and 1. If the firm invests in all technol- green procedures, better-managed firms also are more likely ogies, the green management score equals 1; if it reports no in- to invest in green technology, underscoring the importance vestments, it equals 0. Figure 56B depicts the positive relation of improving management across organizations in Georgia. FIGURE 56B Management Practices and Green Technology Investment 0.7 Green technology investment score 0.6 0.5 0.4 0.3 0.2 0.1 Notes: Controls include firm size class, age class, industry, and region of location. Source World Bank elaboration based on WBES. 0.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 Management z-score 130 Appendix Technical Report Appendix C Green Innovation: An Analysis of New Green Patents A pplicants from Georgia were granted 18 green in- as only 18 IPFs were granted in the last 20 years, which equals ternational patents in 2000-2019, accounting for about 4.7 patents per million inhabitants. This is far below 0.1 percent of total IPFs granted to countries other the average of EU members, even compared to those below than China, EU members, USA, Japan, Korea, UK the median of the GDP per capita. Compared to this subset of and China. Green international patents are highly EU members, the number of patents per million inhabitants concentrated in developed countries. EU members, is around two-thirds of the average country (7.8). However, USA, Japan, Korea, and UK account for nearly 90 per- compared to countries other than green patent leaders (RW), cent of new green patents granted in the last two decades. The the number of patents produced was similar on average (the number of green patents in Georgia appears to be relatively low, average country produced 4.6 patents). BOX 16 Introduction to green patent analysis Green patent analysis provides useful information on the by multiple patent offices. The analysis reports IPF counts, Georgian innovation ecosystem related to the green econo- a proxy for high-quality innovation output. Only high-value my. This section leverages a custom database compiled from innovations are patented in more than one country since it PATSTAT of global international patent families in sustainable involves monetary and time costs that work as barriers to technologies for 2000-2019. One advantage of using this patenting lower-value innovations. data set is that it allows for cross-country comparisons of innovation and specialization areas. The analysis relies on the Y-section code of the Internation- al Patent Classification to distinguish patents in sustainable The patent analysis looks at family-level international patents technologies from patents related to other types of technol- of high market value. Patents with low or zero commercial ogies. Technologies were grouped following guidance from value (e.g., obtained for compliance or administrative pur- the European Patent Office and International Energy Agency poses) are excluded. An international patent family (IPF) re- (2021) and classified into types of green technologies (Dia- fers to a collection of patents covering the same technical gram 2). content, defined as DOCDB simple patent family granted DIAGRAM 2 Types of green technology Green technologies Energy substitu- Climate change tion and energy Low-carbon mitigation efficiency energy supply Cross-cutting enabling technol- Climate change ogies adaptation 131 Appendix Greening Firms in Georgia Adjusted by the income level, Georgia has a high- average compared to countries with similar income per capita. er-than-average number of new green patents per million According to the available information, Georgia outperforms inhabitants. Figure 57C plots the GDP per capita and the num- its peers (Russian Federation, Türkiye, Ukraine, Kazakhstan, ber of new green patents per million inhabitants, where the Serbia, Armenia). However, all EU countries (except Romania) red dotted line is the linear trend. Georgia is slightly above the have more patents per capita. FIGURE 57C Green Patents and Economic Development China EU Japan Korea Peers RW Georgia UK USA 7 6 5 4 Log Green patents per million pop. 3 2 1 0 -1 -2 -3 -4 -5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0 Log GDP per capita (US$, PPP) Source: World Bank calculations based on PATSTAT data and World Bank Data.