WORLD BANK GROUP Raising Firm Productivity November 2021 Kosovo Country Economic Memorandum  1 Kosovo Country Economic Memorandum Raising Firm Productivity Table of Contents Executive summary 7 Firm Characteristics and Dynamics 11 1.1. Firm Characteristics 12 1.2. Firm Dynamics  22 Firm Productivity and Access to Credit 35 1.3. TFPR and Labor Productivity  36 1.4. Access to Credit and Productivity 43 1.5. Barriers to Productivity  47 The Impact of COVID-19 on Firms 48 Conclusions and Tentative Policy Implications 51 References 55 Technical Appendix 57 Additional Tables and Figures 61 2  Figures Fig 1. Active Firms by Size and Total, Already Active, 2012–18, Fig 31. Labor, Capital, and Average, 2010–18 p.12 Percent p.24 Intermediate Inputs, 2012–17, Fig 2. Average Firm Size and Density, Fig 17. Drivers of Entry and Exit Rates, Index 100=2012 p.37 2012–18 p.13 Percentage Points p.24 Fig 32. TFPR and value added per Fig 3. Turnover Comparisons by Firm Fig 18. Firm Size at Entry p.25 worker by Sector and Average Size, Relative to EU Averages, p.38 Fig 19. Firm age and Employment 2018. Index EU=100 p.14 Growth p.26 Fig 33. Decomposition of TFPR Fig 4. Active Enterprises by Main Growth, 2014–17, Percent Fig 20. Employees by Firm Age and Activity Compared to the EU, Change p.39 Size over Five Years, Average Percent of Total p.14 2012–18, Index, Entry Year = 1 Fig 34. Characteristics of Entrants Fig 5. Employment Share by Sector, p.27 Relative to the Median Firm. Percent of Total p.15 Percent p.40 Fig 21. Firm Survival in Kosovo, Fig 6. Firms, Employment, and Sales Average 2012–18, Percent p.27 Fig 35. TFPR, Employment, Revenue, by Municipality, Percent p.15 and Value Added per Worker, Fig 22. Thousands of Employees in Logarithms, 2012–18 p.41 Fig 7. Exporters and Importers p.17 Active Firms, 2010–18 p.28 Fig 36. Total Factor Productivity Fig 8. Exports by Sector p.18 Fig 23. Firms Reporting Employees, Revenue (TFPR) by Sector, Fig 9. Firms with Foreign Percent of Total Firms, 2010– 2013–17 p.42 Shareholders, percent of firms, 2018 p.29 Fig 37. Domestic Credit to the Private 2018 p.19 Fig 24. Contribution to Employment Sector, Percent of GDP p.43 Fig 10. Gender of Firm Decisionmakers Growth by Sector, 2010–14 and 2015–18, Percent p.30 Fig 38. Credit Disbursed and by Sector, Percent p.20 Employment by Sector, Fig 11. Digital Firms, 2010–10, Percent Fig 25. Employment Growth, Firms 2010–18, Percent of Total p.44 of Total Firms p.21 by Sector Surviving 5 Years, Average Number of Employees, Fig 39. Credit to the Private Sector by Fig 12. Concentration in the Kosovar Type, Average 2010–18, Euros 2017–18 p.30 Economy p.21 and Percent of Total p.45 Fig 26. Net Job Creation, 2011–18 p.31 Fig 13. Net Firm Turnover, Entry. and Fig 40. Loans to the Private Sector by Exit Rates, Percent of Firms, Fig 27. Job Creation, Destruction, and Term (Years), Average 2015–18, Average 2012–18 p.22 Employment by Firm Age and Percent of Total p.45 Size, Average 2016–18 p.32 Fig 14. New Business Density and GDP Fig 41. Productivity and Access to per Capita in Constant US$ PPP, Fig 28. Employment by Firm Age, Credit p.47 2018 p.23 Percent of Total, Average 2016–18 p.32 Fig 42. Barriers to Investment and Fig 15. Economic and Administrative Exports. Percent of firms p.47 Firm Entry and Exit, Number of Fig 29. TFPR, Sales, and Value Added per Worker, 2013–17 p.36 Fig 43. Digitalization and Innovation Firms, 2012–18 p.23 in Response to the COVID-19 Fig 16. Size of Firms Entering a Market Fig 30. Labor Productivity Relative to Outbreak Compared, Percent Relative to the Average Firm the EU, 2018, Percent p.37 of Firms p.49 Tables Table 1.  Selected Characteristics of Table 4.  TFPR in Micro-Enterprises Table 8.  Main Drivers of Lower TFPR Firms in Kosovo, 2017, Percent p.19 Relative to Other Firm Classes p.41 Growth and Recommended Policy Table 2.  Five-Year Transition Matrix Table 5.  Credit in Kosovo, Summary Changes p.53 by Firm Class, 2012 and 2013 Firm Statistics, 2018 p.44 Boxes Cohorts, Average p.26 Table 6.  Access to Credit by Firm Class, Table 3.  Ratio of Productivity in Average, 2015–18 p.46 Box 1. Measuring Firm Dynamics and “Laggard” (90th percentile) to “Frontier” Productivity in Kosovo p.8 Table 7.  Potential Correlates with (10th percentile) Firms by Sector and Productivity in Kosovo p.48 Box 2. Decomposing the Sources of Size, Average 2012-18 p.40 TFPR Growth p.39  © 2021 International Bank for Reconstruction and Development / The World Bank 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 with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. 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 concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. 4  Acknowledgments This note was prepared by Lucio Castro (Senior Economist), Leonardo Iacovone (Lead Economist), and Aslı Şenkal (Senior Economist) and with contributions from Blerta Qerimi (Senior Private Sector Specialist), Besart Myderrizi (Research Analyst), and Temel Taskin (Economist). The note benefitted from the comments of Bledi Çeliku (Senior Economist), Enrique Blanco Armas (Practice Manager), and Donato De Rosa (Lead Economist). This note has been prepared as part of the Kosovo Country Economic Memorandum led by Aslı Şenkal (Senior Economist, Task Team Leader). The work was overseen by Linda Van Gelder (Country Director, Western Balkans), Massimiliano Paolucci (Country Manager, Kosovo and North Macedonia), Jasmin Chakeri (Practice Manager, Macro, Trade, and Investment Global Practice), and Denis Mehmedev (Practice Manager, Finance, Competitiveness, and Innovation Global Practice). The team is grateful to Kosovo Agency of Statistics and Central Bank of Kosovo for their support in providing data.  5 List of Acronyms BPS World Bank business pulse survey CEM Country Economic Memorandum EBRD European Bank for Reconstruction and Development ES Enterprise survey EU European Union FDI Foreign direct investment GDP Gross domestic product HHI Herfindahl index KAS Kosovo Agency of Statistics KCR Kosovo credit registry dataset KPS Kosovo productivity survey KTR Kosovo tax registry dataset LP Labor productivity ICT Information and communications technology MSME Micro, small, and medium sized enterprises NIS National innovation system OECD Organisation for Economic Co-operation and Development PPP Purchasing power parity R&D Research and development SME Small and medium sized enterprises TFP Total factor productivity TFPR Total revenue factor productivity VA Value added WB6 Western Balkans WDI World Development Indicators W&R Wholesale and retail trade 6  Executive summary To boost economic growth and foster sustained formal job creation in Kosovo, igniting firm productivity is crucial. Both demographically and historically, Kosovo is a small, young transition economy. Although in the decade leading up to the COVID-19 crisis, the country’s economic growth outperformed the other Western Balkans (WB6) peers,1 it was not able to provide enough jobs, especially for women and the young, or to reduce high unemployment.2 Sustained job creation can only be driven by private companies that are productive enough to grow and expand. That is why understanding firm dynamics, particularly what drives productivity, the engine of economic development, is crucial to identify barriers to growth and job creation, and to formulate policy reforms that will accelerate firm growth and formal job creation. Based on detailed micro-data, this note examines the characteristics and recent evolution of firms in Kosovo, with particular attention to firm productivity. It benefits from a dataset compiled from anonymized tax records complemented by data from the Kosovo Productivity Survey (KPS). To explore the links between productivity and credit, it also draws on a unique dataset on access to credit from the Kosovo Credit Registry (KCR). Box 1 describes the data and methodology used. Annex 1 provides further technical details. The note also provides a snapshot of the impact of COVID-19 on firms, based on the Business Pulse Survey (BPS)3 and administrative data. For the last decade, the landscape of firms in Kosovo has been dominated by microenterprises with low productivity, employment and survival rates. According to the administrative data, in 2010–18 micro companies (firms averaging fewer than 10 employees) and small enterprises (10–49 employees) accounted for more than 98 percent of firms. Micro enterprises are less productive and have higher net firm turnover,4 lower survival rates,5 and much lower employment growth. Moreover, the average annual net rate of firm creation in Kosovo—the difference between firm entry and exit rates—was close to zero, an indication of a lack of dynamism in the last decade, especially for a young economy. Firm creation and growth, firm density, average size, and the likelihood of survival are all low, which implies that there are severe constraints on private sector development. Firm density, the number of firms per million inhabitants, is considerably lower than in the European Union (EU) and also than what would be predicted by Kosovo’s income per capita. Partly reflecting the small size of the economy, Kosovar firms are also much smaller than their peers in the EU and in the aspirational economies.6 Firm survival is also much lower than in comparator countries, especially among micro enterprises. Unlike firms in aspirational peers, Kosovo seems not to have many middle and large companies. Only about 2 percent of the micro enterprises became small firms, 5.5 percent of small enterprises become medium, and 4.6 percent of medium firms became large. Medium companies report about 37 percent less in sales than the EU mean. Large firms in Kosovo report the lowest proportion of sales, just 15 percent of the EU average, than large firms in aspirational peers. Kosovo’s firms are only tenuously linked to global markets and the country is lagging in the share of female-run companies. The share of exporter enterprises (0.3 percent per the Kosovo Statistics Authority (KAS) and 3.4 percent per the Kosovo Productivity Survey [KPS]) is far below the averages for the EU (7.4 percent); aspirational peers (e.g., 7.4 percent in Croatia); and the Western Balkans; only 6 percent have any foreign shareholders; and just 9 percent are female-run. Impediments to firm growth and exit are numerous. It appears that constraints on firms are sizable from the beginning and become more pervasive the larger the firm; companies increase the number of their employees until they are mid-sized, then seem to practically stop growing. Administrative costs to exit the market are also high in terms of liquidating assets and legally closing a company. Positive and rising net job creation in 2015–18 was driven by higher formalization of jobs and the increasing size of incumbent firms, especially young SMEs. The number of net jobs created annually— the difference between gross job creation and gross job destruction—practically doubled in 2015–18 compared to 2011–14. Net job creation was driven by the increasing size of incumbent firms and higher job formalization: the share of firms with employees increased from 40 percent in 2010 to 70 percent in 2018. 1 In addition to Kosovo, the WB6 countries are Albania, North Macedonia, Serbia, Montenegro and Bosnia and Herzegovina. 2 World Bank 2018. 3 See Section 3 for further details on the BPS. 4 Firm turnover is the difference between firm entry and firm exit. 5 Share of active firms that continue reporting sales per year in total firm population. 6 In this note, aspirational peers are Croatia, Czechia, Latvia, Lithuania, Estonia, and Slovakia—all EU members.  7 In Kosovo labor productivity (LP) is far below LP in the EU and aspirational countries; annual average LP growth was negative and average yearly total revenue factor productivity, a revenue- based measure of productivity growth (TFPR),7 was low in 2012–17, the period for which data are available. Clearly, there is room for further economic convergence with the more developed European countries: LP in Kosovo is only a third of the EU average and is much lower than in the aspirational economies. Moreover, LP, proxied by sales per worker, fell on average by –0.25 percent annually between 2012 and 2017, while TFPR increased by just 0.5 percent. Declining productivity of incumbent firms explains Kosovo’s disappointing TFPR growth rate; firm dynamism plays only a minor, and somewhat negative, role. Kosovo’s low TFP growth is explained by the declining market share and productivity of incumbent firms, only partly compensated for by rising average firm productivity. The latter component accounted for just a third of TFPR growth in 2012–17, far below the average of developing countries for which there are comparable data. Exit of less productive firms was only a minor factor in driving aggregate TFPR growth, and entry of new firms contributed negatively to productivity growth. Linked to Kosovo’s meager TFPR performance is the fact that, relative to the EU and other peers, private investment in research and development (R&D) and management techniques is also low. In spite of the strong empirical evidence of the positive impact of investment in R&D on firm productivity (Cirera and Maloney 2017), only 1 percent of businesses in Kosovo report investing in R&D; the average for the EU is 14 percent and among some aspirational economies like Croatia it is 7.4 percent.8 Private investment in R&D), according to the KPS, is even lower among micro enterprises (0.1 percent) and small companies (0.2 percent. Similarly, only about 6 percent of the plants report investing in management training, despite the evidence of its beneficial effects on firm productivity (Bloom et al. 2010). More productive firms create more jobs and sell more; exporters and foreign-controlled companies tend to be larger, more productive, and more capital-intensive than firms oriented to the domestic market and pay higher wages. Higher firm TFPR is associated with higher average revenues and more workers per company, which underscores the importance of fostering firm productivity for economic growth and job creation. Furthermore, the track record of exporters and foreign-owned firms highlights the relevance of integration into global markets, especially for a small economy like Kosovo. Access to credit is low and biased toward short-term financing; and financing is not allocated to the most productive firms. Only one-third of firms have access to formal credit. About 70 percent of loans granted have terms no longer than three years, suggesting that the local financial system is not providing enough capital for long-term investment. More productive and larger firms in more concentrated sectors are more likely to get access to credit; firms based in non-majority municipalities9 have less access. Micro and small enterprises and startups are particularly deterred by the lack of access to formal financing. The COVID-19 shock affected MSMEs the most, especially in sectors under lockdown. While most of the firms adjusted by granting leaves and adjusting wages and worker hours, MSMEs and female- run companies experienced the largest net job losses. Despite layoffs were largest in sectors under lockdown, 2020 employment did pick up in most sectors, including hospitality. However, more than half of companies reported they expected to fall into arrears end 2020; MSMEs in particular were financially stressed. Kosovo needs a multidimensional policy strategy to foster growth in firm productivity. Based on the study findings and the results of other notes prepared for Kosovo’s Country Economic Memorandum (CEM), this note proposes a policy strategy that targets the three main sources of firm productivity growth: (1) firm productivity (the “within” component); (2) market reallocation (the “between” component); and (c) firm dynamics (entry and exit). In what follows, Section 1 examines the characteristics and recent dynamics of Kosovar firms. Section 2 analyzes the drivers and evolution of productivity, with emphasis on the links between productivity and access to credit. It also assesses the main barriers to productivity growth. Section 3 sheds light on how COVID-19 has affected Kosovar firms. Section 4 concludes by discussing tentative policy implications of the analysis. 7 In the absence of firm level prices, this note uses a revenue-based measure of TFP (see Box 1). Please see Cusolito and Maloney (2016) for a more detailed description of advantages and disadvantages of using TFPR measures. 8 See World Bank (2018). 9 See Box 1 for the definition of non-majority municipalities. 8  Box 1. Measuring Firm Dynamics and Productivity in Kosovo Total factor productivity (TFP) and labor productivity (LP) are two ways of computing productivity, the measure of firm efficiency at combining or converting inputs (capital, labor, et al.) to produce goods or services (output). More precisely, productivity can be defined as the variation in output that cannot be explained by changes in factors of production. While LP measures output per worker, and therefore looks only at employment, TFP incorporates not only adjustments in other production factors (e.g. capital stock and intermediate inputs) but also, and more importantly, technology, innovation, management practices, and other ways of augmenting firm output with the same input endowment (Syverson 2011). As no data on firm-level prices are available for Kosovo, this note calculates revenue TFP (TFPR) instead of physical TFP. TFPR reflects not only firm physical efficiency at production but also prices that reflect product quality and markups in addition to input costs (Cusolito and Maloney 2018). This note is unfortunately unable to capture directly those price-related dimensions. This chapter relies on two datasets containing detailed information on variables required to analyze firm dynamics and estimate TFPR and LP. Administrative tax records from the Kosovo Statistics Authority (KAS) is the main source of data. The KTR dataset contains records for about 30,000 formal establishments with annual information on company characteristics (location by municipality, activity at four digits of NACE , year of starting and closing operations), employment, revenues, exports, imports, inputs, costs, and investment in durable goods. A panel dataset for 3,600 establishments reporting the information required to estimate our preferred measure of productivity for TFPR in 2012–17: investment, revenues, intermediate consumption, wages and employment was built. (Annex 2 provides summary statistics.) The KTR panel data set was complemented with information from the 2017 World Bank Kosovo Productivity Survey (KPS). The KPS covered a representative sample of 3,083 firms. In addition to the variables contained in the KTR, the KPS dataset provides among other variables detailed information on input purchases, worker skills, firm location (region), and shareholding (foreign and domestic). We can match 3,314 firms from the KPS with the KTR dataset. (See Annex 2 for summary statistics on the KPS dataset.) Capital stock is estimated using investment data by establishment in durable goods from the KTR dataset and combining it, where possible, with the estimates of firm total assets provided by the KPS dataset. LP is calculated as the ratio of sales to the number of employees (except when we state that it refers to value-added per employee); TFP is estimated by applying semi-parametric methods following Levinsohn and Petrin (2003), given the small sample of firms reporting investment in the KTA dataset. (See Annex 1.A for further details.) We also control for the potential effects of firm entry and exit in TFP growth by applying the decomposition proposed by Meliz and Polanec (2015). As firms reporting the data available to estimate TFPR are mainly medium and large firms, the panel dataset is reweighted using information on firm employment shares by NACE 2-digit sector from the entire KAS dataset, to give more weight to smaller companies to increase representativeness of the dataset, where small and micro productive units are more dominant. To assess the links between access to credit and productivity, a panel dataset with information by firm from the Kosovo Credit Registry (KCR) with information on loan amount disbursed, loan term, lending by financial institutions (banking and micro-finance), nominal interest rate per financial instrument, and loan payments, among other variables, for the universe of firms for 2010–18 is used. The KCR and KTR datasets are matched by firm, producing a panel with complete information for about 6,000 firms. (See Annex 2 for summary statistics on the KCR panel dataset.) Sales, wages, exports, imports, and other variables measured in euros were deflated by the Producer Price Index (PPI). Investment was deflated using the corresponding investment price deflators. All monetary variables were converted to 2018 euros. Observations reporting negative turnover, sales, or investment were identified as missing. Finally, using KAS data, “non-majority municipalities” were identified as those subnational units with a share of population of Serbian origin of 40 percent or more of the of total population. The other municipalities are predominantly Albanian/Turkish majority.  9 01 Firm Characteristics and Dynamics This section presents an overview of the main characteristics of firms in Kosovo. It describes the number and evolution of active firms, average firm size, number of firms per capita, and firm and employment distribution by sector. It also assesses the relative importance and features of firms engaged in international trade and benefiting from foreign direct investment (FDI). Moreover, it examines the gender composition of firm decisionmakers. Finally, it analyzes the economic concentration of the Kosovar economy. (Annex 1 provides technical details on how firm characteristics and dynamics in Kosovo were analyzed.) 1.1. Firm Characteristics More than 98 percent of Kosovar firms are micro and small enterprises10 and the population of active firms has been relatively stable for the last decade. Between 2010 and 2018, every year an average of about 30,000 enterprises reported sales for at least two years (Figure 1A). While the proportion of micro enterprises is marginally lower than in some aspirational peers, the country has more small firms, except for Croatia. The share of medium and large enterprises is similar to Croatia and Lithuania (Figure 1B). Figure 1. Active Firms by Size and Total, Average, 2010–18 a) Number of active firms b) Active firms by size, percent of total 34,000 Micro . 32,000 , ( -) . , , . , . 30,000 , . , , , , Small . 28,000 ( - ) . . . 26,000 . Medium . 24,000 ( - ) . . . 22,000 . Kosovo Large . Croatia 20,000 ( +) . Lithuania . . Czechia . Slovakia Source: KTR, World Bank staff calculations. Notes: Active units are those with positive turnover. Activities included: B-N (excluding K64.2) according to NACE Rev. 2. Micro firms average 0–9 employees, small firms 10–49, medium firms 50–249, and large firms 250 and above. In the last decade, the share of medium and large companies was practically unchanged, possibly implying barriers to firm growth. There were no major changes in the share of medium enterprises (50–249 employees) and large firms (250 and more).11 The share of micro enterprises fell from 95 percent in 2010 to 90 percent in 2018, as the share of small firms rose from 4 to 8.3 percent. These figures suggest that there are barriers to growth beyond small size in Kosovo. Average firm size is smaller than in many of the aspirational peers though similar to the EU average, but firm density is below even what can be expected by the country´s income per capita. The average firm in Kosovo employs 5 workers, far below most aspirational peers except Czechia (4.5) and Slovakia (3.5), and close to the EU mean (5.1). Firm density reaches only 16,300 firms per million inhabitants, just over a third of the EU average (54,600), and lower than in any of the aspirational economies (Figure 2B). Moreover, Kosovo has less firm density than its income per capita would suggest, in contrast with aspirational peers and Western Balkans comparator countries, except for Serbia (Figure 2C). 10 Micro firms average 0–9 employees and small companies 10–49. 11 KTR, World Bank staff calculations. 12 Firm Characteristics and Dynamics Figure 2. Average Firm Size and Density, 2012–18 a) Size: average number of workers b) Density: active firms, thousand per million inhabitants 8 Czechia . . 7 Slovakia . . 6 Lithuania . . . . . 5 Estonia . 4 . . EU (2020) . 3 Latvia . 2 Croatia . 1 Kosovo . 0 Croatia Latvia Lithuania Estonia EU( ) Kosovo Czechia Slovakia c) Active firms per million inhabitants. and GDP per capita (log, current US$, PPP), 2018 . Firms per million CZE Kosovo people (log) SVK R² = . Comparators . MLT LVA PRT SVN SWE Aspirational peers GRC LTU ITA Other EU countries MNE ESP BEL NLD . HRV LUX HUNEST FIN AUT IRL BGR ALB POL EU FRA DNK . ROU CYP DEU MKD . KOS SRB . . . . . . . . . GDP per capita (PPP, in logs) Source: KAS and Eurostat data, World Bank staff calculations. Notes: Active firms are those with positive employment or turnover on each country adjusted by population size. Activities: B-N (excluding K64.2) according to NACE Rev. 2. Firm size as total persons employed in the population of active enterprises / total active enterprises. Employed persons are those aged 15 and over who perform part-time or full-time work for pay, profit, or family gain, according to Eurostat. Measured by sales, firms in Kosovo are also smaller than companies in aspirational peers and the EU, especially the medium and large enterprises. Micro and small companies report sales that are half the revenues reported by average EU peers. Sales reported by small Kosovar companies are about 44 percent of the EU mean and large company sales are 15 percent smaller. While these differences are not so large for Kosovo’s MSMEs, except relative to Slovenian peers, its large firms, companies with more than 250 employees, report the lowest proportion of sales relative to the EU average compared to aspirational peers (Figure 3). Firm Characteristics and Dynamics 13 Figure 3. Turnover Comparisons by Firm Size, Relative to EU Averages, 2018. Index EU=100 Kosovo Latvia Lithuania Czechia Slovenia Micro Small Medium Large ( - employees) ( - employees) ( - employees) ( + employees) Source: KTR and Eurostat data, World Bank staff calculations. Notes: Total turnover in 2018 in euros of active businesses (except finance and insurance) divided by total active enterprises. Baseline is the EU average (indexed to 100) for countries that report aggregate turnover by firm size. Activities: B-N (excluding K64.2) according to NACE Rev. 2. Micro are firms with an average of 0–9 employees, small 10–49 employees, medium 50–249 employees, and large 250 and more. A distinctive feature of Kosovo’s economy is the relative importance of wholesale and retail (W&R) firms. More than 50 percent of firms are in W&R; 14 percent are in manufacturing, 11.2 percent in accommodation and food services, about 7 percent in construction, and the rest in other services activities. W&R is much more important in Kosovo than in aspirational peers like Croatia and Slovakia, where it only accounts for about 30 percent of firms (Figure 4). Figure 4. Active Enterprises by Main Activity Compared to the EU, Percent of Total a) Kosovo Wholesale & Retail trade Manufacturing Construction 48.4 14.7 8.5 b) Croatia Transportation & Storage Administrative & Manufacturing Wholesale & Retail 25.3 support services 13.7 trade 16.2 12.6 c) Slovakia Accommodation & Food Manufacturing Information & services 16.2 Communication 25.3 12.6 Wholesale & Retail trade 13.1 Utilities Source: KTR and Eurostat data, 2020, World Bank staff calculations. Notes: Active units are those with positive turnover. Activities included: B-N (excluding K –Financial and insurance activities) according to NACE Rev. 2. See Annex 1 for more detail on sector classification. 14 Firm Characteristics and Dynamics W&R is also one of the largest employers, along with manufacturing and other service activities. It employs about 33 percent of workers in Kosovo, followed by manufacturing, 15 percent; construction, 10.7 percent; accommodation and food services, 7.3 percent; and financial services, about 7 percent (Figure 5). Figure 5. Employment Share by Sector, Percent of Total Wholesale and retail trade . Manufacturing . Construction . Accommodation and food services . Financial services . Information and communication . Administrative and support services . Electricity, gas, steam and air-conditioning supply . Transportation and storage . Professional, scientific and technical act. . Water supply; sewerage, waste mgmt. and remediation act. . Mining and quarrying . Real Estate . Source: KTR and Eurostat data, World Bank staff calculations. Notes: Share of employed persons in active firms. Activities: B-N (excluding K64.2) of NACE Rev. 2. Micro firms have an average of 0–9 employees, small firms 10–49, medium firms 50–249, and large firms 250 and more. Employment and sales are heavily concentrated in Pristina. Although only 24.1 percent of firms are in Pristina, the municipality accounts for about 43 percent of Kosovar employment and about 44 percent of sales, similar to comparator countries like Moldova (Cojocaru 2016). Ferizaj is second in importance with 6.6 of employment and about 7 percent of sales (Figure 6). Figure 6. Firms, Employment, and Sales by Municipality, Percent a) Firms Leposaviq 0.2 Mitrovicë e Veriut Zveçan 0.3 0.2 Zubin Podujevë Potok 3.0 Istog 0.1 Mitrovicë Prishtinë 1.4 3.6 24.1 Pejë Skënderaj Vushtrri Novobërdë 5.5 1.3 2.5 0.1 Kamenicë Klinë Drenas Obiliq 1.4 2.1 0.8 1.4 Deçan Malishevë Fushë Gjilan Kosovë 1.2 1.6 6.4 Suharekë 2.9 Graçanicë Junik Ranillug 2.8 0.0 Rahovec Lipjan Kllokot 2.5 2.4 Gjakovë Shtime Ferizaj Partesh 5.6 1.1 8.5 Shtërpcë Viti Mamushë 0.4 1.9 Prizren Kaçanik 10.5 Hani i 1.6 Elezit 0.2 Dragash 1.2 Firm Characteristics and Dynamics 15 Figure 6. Firms, Employment, and Sales by Municipality, Percent b) Employment Leposaviq 0.1 Mitrovicë e Veriut Zveçan 0.1 0.0 Zubin Podujevë Potok 1.5 Istog 0.0 Mitrovicë Prishtinë 1.1 3.7 42.7 Pejë Skënderaj Vushtrri Novobërdë 5.6 0.9 1.7 0.0 Kamenicë Klinë Drenas Obiliq 0.9 2.0 0.9 0.6 Deçan Malishevë Fushë Gjilan Kosovë 0.7 1.0 4.5 Suharekë 3.7 Graçanicë Junik Ranillug 2.2 0.0 Rahovec Lipjan Kllokot 1.3 1.4 Gjakovë Shtime Ferizaj Partesh 3.4 0.6 6.6 Shtërpcë Viti Mamushë 0.1 1.1 Prizren Kaçanik 6.2 Hani i 0.8 Elezit 0.4 Dragash 0.4 c) Sales Leposaviq 0.1 Mitrovicë e Veriut Zveçan 0.1 0.1 Zubin Podujevë Potok 1.5 Istog 0.0 Mitrovicë Prishtinë 0.9 1.9 43.9 Pejë Skënderaj Vushtrri Novobërdë 5.9 0.9 1.6 0.0 Kamenicë Klinë Drenas Obiliq 0.9 1.7 1.0 0.3 Deçan Malishevë Fushë Gjilan Kosovë 0.7 1.1 2.9 Suharekë 4.8 Graçanicë Junik Ranillug 2.0 0.1 Rahovec Lipjan Kllokot 1.3 1.7 Gjakovë Shtime Ferizaj Partesh 2.1 0.7 7.3 Shtërpcë Viti Mamushë 0.1 1.8 Prizren Kaçanik 5.3 Hani i 0.7 Elezit 0.5 Dragash 0.2 Source: KTR data, World Bank staff calculations. Notes: Activities: B-N (excluding K64.2) according to NACE Rev. 2. Firms not reporting location were excluded. 16 Firm Characteristics and Dynamics Kosovo has the lowest share of exporters among Western Balkans countries and a much lower proportion of manufacturing exporters than aspirational peers except Czechia. Less than 4 percent of firms export and import; about 0.3 percent are solely exporters, and about 25 percent are exclusively importers (Figure 7A). Self-reported KPS data indicates that 1.4 percent of firms identified themselves as exporters and 34 percent as importers.12 Exporters and importers tend to be medium and large firms (Figure 7B). Focusing on manufacturing, Figure 7C reveals that the fraction of exporters is lower (6.4 percent) than in any aspirational country except Czechia (4.3 percent), and less than a fourth of the EU average (21.7 percent). According to the 2020 Business Environment and Enterprise Performance Survey (ES), Kosovo also has the lowest proportion of exporters among Western Balkans economies.13 Figure 7. Exporters and Importers a) Percent of total firms b) Percent of total firms by sector and size Non-exporter & Importer and Only Only Non-importer exporter Exporter Importer Exporters . % . % . % . % Importers Micro Small Medium Large ( - ) ( - ) ( - ) (+ ) c) Manufacturing exporters, percent of manufacturing firms Estonia 46.0 Latvia 27.5 EU (2020) 21.7 Lithuania 17.8 Croatia 14.2 Slovakia 9.5 Kosovo . % of manufacturing firms Czechia 4.3 Source: KTR and Eurostat data, 2020, World Bank staff calculations. Note: Exporters or importers are those companies that report €10,000 or more in exports or imports to the KAS in the relevant year. “Importer and exporter” classification refers to a company that both imports and exports in the same period. “Only exporters” or importers) are firms that export or import but do not import or export. Figure 7 C depicts the number of manufacturing exporters relative to the total manufacturing active firms as classified in Eurostat. Figures are average values for 2012–18. Activities: B-N (excluding K64.2) according to NACE Rev. 2. 12 See Table 2 in Annex 2. 13 See https://www.beeps-ebrd.com/indicators/ Firm Characteristics and Dynamics 17 Manufacturing accounts for about 45 percent of exports. Mining, manufacturing, and some service activities have the largest export-to-sales ratios. W&R is the second largest exporter at 24.6 percent, followed by ICT at 7.4 percent (Figure 8A). According to KPS, furniture is the main manufactured export, accounting for about 38 percent. Exports account for about 40 percent of sales in mining and about 14 percent of sales in manufacturing (Figure 8B). Transport, administrative services, and ICT report export-to-sales ratios of about 10 percent. Traditionally non-tradable activities, such as financial services, construction, and real estate, have the lowest export-to-sales ratios. Because service exports in accommodation and food services are not captured by the tax records, they are understated. Figure 8. Exports by Sector a) Percent of total exports Manufacturing Wholesale and retail trade Information Mining and Transport 45.1 24.1 and communication quarrying and 7.4 Storage 4.7 Utilities 6.5 Construction & Real estate 2.7 Administrative and support act. b) Share of exports in total sales by sector Mining and quarrying 36 Utilities 27 Manufacturing 22 Transport and storage 13 Information and communication 11 Administrative and support act. 11 Average 6 Professional, scientific and technical 5 Wholesale and retail trade 2 Construction and Real Estate 2 Financial services 2 Accommodation and food service 1 Source: KTR and KPS data, World Bank staff calculations. Note: Exporters/importers are those companies that report €10,000 or more in exports/imports in the year of reference. “Importer and exporter” reflects the company reports both imports and exports within the same period. Only exporters (importers) are firms that export (import) but do not import (export). The share of exports in total sales corresponds to total exports divided by total sales. Figures report the average values for 2012–18. Activities: B-N (excluding K64.2) according to NACE Rev. 2. See Annex 1 for more detail on sector classification. Only 6 percent of Kosovan firms have foreign shareholders (Figure 9A) and the share of firms with at least 10 percent of foreign shareholding is far lower than in the other Western Balkans countries. Foreigners hold more than 10 percent of the assets in less than 6 percent of firms. About 10 percent of shareholders in construction and transport firms are foreign, as are 8 percent in utilities and 5 percent in manufacturing (Figure 9B). According to the Enterprise Survey data, only 1.3 percent of firms have at least 10 percent foreign ownership, compared to 11.8 percent of firms elsewhere in the region.14 14 World Bank Enterprise Survey (2019). The figure is the average for Albania, Bosnia and Herzegovina, Montenegro, North Macedonia, and Serbia. 18 Firm Characteristics and Dynamics Figure 9. Firms with Foreign Shareholders, percent of firms, 2018 a) Firms with foreign shareholding b) By sector Mining and quarrying With no foreign shareholders 93.8 Manufacturing 5.29 Utilities 8.27 With foreign shareholders 6.2 Construction 10.17 Wholesale and retail trade 6.78 With foreign shareholders Transport and storage 9.60 5.7 (>10% assets) 3.86 Accommodation and food services With foreign shareholders Information and communication 4.61 (>50% assets) 0.8 Real Estate 2 technical act. 2.36 Professional, scientific and 2 act. Administrative and support 2.17 Source: KPS data, World Bank staff calculations. Notes: Fraction of firms that report having foreign shareholders (> 0%). Figures are weighted by firm weights. Only 3.4 percent of firms invest in R&D, far less than in the EU and peer economies; investment in training and management is also low. Table 1 shows that only 3.4 percent of firms report investing in R&D against an EU average of 14 percent and, typical of comparators, Croatia with 7.4 percent (World Bank 2018). The proportion is especially low among micro (3.7 percent), and small (12 percent), and medium enterprises (12.6 percent). Across all firm size classes, investment in R&D is also lower than in the EU, Western Balkans comparators, and aspirational peers (World Bank 2018). Only about 6 percent of enterprises report investing in employee training and 8 percent in management training; the percentage increases, however, with firm size. Table 1.  Selected Characteristics of Firms in Kosovo, 2017, Percent Class Category Invest in Receive Have Tertiary Have Unpaid Share of plants Share of firms R&D Government Educated Employees that provide that Train Subsidies Workers Employee Managers and Training Supervisors Size Micro (0-9) 3.7 1.6 31.6 14.7 5.4 7.9 Small (10-49) 12 2.7 26.5 2.7 15.6 13.1 Medium (50-249) 12.6 4.0 23.6 1.7 29.9 23.7 Large (250+) 18.2 14.0 36.5 0.2 27.6 27.2 Shareholding Nonforeign 4.3 1.8 31.0 14.1 6.1 8.6 shareholding Foreign 3.8 0.9 33.8 8.8 9.3 6.1 shareholding Average All firms 3.4 1.7 31.2 13.8 6.3 8.4 Source: KPS data, World Bank staff calculations. Note: figures were calculated using firm sample weights. On average micro have 0–9 employees, small firms 10–49 employees, medium firms 50–249 employees and large firms 250 or more employees. Only about 33 percent of workers have tertiary education and about 14 percent are unpaid—a proxy for labor informality. Large firms received a larger proportion of government subsidies than smaller firms. There are no significant differences between firms by size in workers’ skills, although the share of tertiary workers is higher in larger firms (36.5 percent). The share of unpaid workers reaches 15 percent among micro firms, although it is only 1.7 percent for medium and 1.7 percent for large ones. Although only 1.7 percent of all firms reported receiving subsidies, among these more beneficiaries were large firms. Firm Characteristics and Dynamics 19 Firms with foreign shareholders tend to have more skilled workers and fewer unpaid workers but the share of companies investing in R&D is similar to domestic-owned companies—close to 4 percent (Table 1). About 9 percent of companies with foreign shareholders invest in employee training compared to just 6 percent in solely domestic firms. In contrast, investment in management and supervision is higher in enterprises with a foreign shareholder. Those companies also have a slightly larger proportion of skilled workers (34 vs. 31 percent of domestic-owned firms), and far fewer unpaid workers (8.8 vs. 14 percent). In only 9 percent of the firms is the decisionmaker female15 (Figure 10A). Sectors with the highest proportion of female decisionmakers (Figure 10B) are administrative services 15 percent; professional services 10 percent; and W&R 10 percent. ICT, accommodation and food services. and administrative and support services employ the largest shares of female workers, about 30 percent each, followed by W&R 15 percent and professional services 10 percent. Mining, electricity and gas, and construction have the fewest female workers (Figure 10C). Figure 10. Gender of Firm Decisionmakers by Sector, Percent a) Gender of firm decisionmaker Percent of total firms 91% men women 9% b) Firms with female decisionmaker by sector c) Female employees in 2017 Administrative and support act. 14.7% Information and communication 33.1% Professional, scientific and technical 10.4% Accommodation and food services 31.8% Wholesale and retail trade 10.3% Administrative and support services 30.3% Real estate activities 27.1% Real Estate 9.3% Wholesale and retail trade 26.2% Manufacturing 8.7% Professional, scientific and technical activities 23.7% Accommodation and food service 7.7% Manufacturing 18.7% Information and communication 6.9% Transportation and storage 17.9% Transport and storage 5.6% Water supply; sewerage, waste 10.9% mgmt. and remediation activities Construction 3.8% Construction 6.9% Electricity, gas, steam and Utilities 1.3% air-conditioning supply 6.7% Mining and quarrying 4.0% Mining and quarrying 0% Source: KPS data, World Bank staff calculations. Note: Figures are weighted using firm sample weights. See Annex 1 for more detail on sector classification. Female-run companies have a higher proportion of unpaid workers and received proportionately less in government subsidies. According to the KPS data, firms with female decision-making have more skilled workers (43 vs. 30 percent in male-run enterprises). In female-run firms about 19 percent of workers are unpaid vs. 13 percent in male-run companies, but the latter receive 1.8 percent of government subsidies and the former 1.4 percent. Between 2010 and 2018 the proportion of firms in digital-intensive sectors consistently averaged 8.5 percent of the firm population despite a surge in 2017 (Figure 11).16 15 The decision maker is defined in response to the following question in the KPS: “Who makes the main decisions in the company?” Multiple responses were allowed, including founder, family member of the founder, manager, members of executive board, and gender. 16 In 2013 and 2017, changes in the number of firms in professional services, information and communication, and administrative services activities accounted for more than 90 percent of the variation in the share of digital- intensive enterprises. 20 Firm Characteristics and Dynamics Figure 11. Digital Firms, 2010–10, Percent of Total Firms 10 9.70 9.41 9.07 8.60 8.30 7.89 8 7.61 7.15 7.26 6 4 2 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: KAS data and Calvino et al. 2018; World Bank staff calculations. Notes: Digital-intensive sectors are those with a high level of digitalization. Kosovo’s economy has relatively high and persistent levels of concentration. For the economy as a whole and proxied by the normalized Herfindahl Index (HHI) of sales, since 2013 concentration has been declining only marginally (Figure 12A). By sector, economic concentration is highest in electricity and gas (close to 1), followed by information and communication, and mining (each 0.53) and financial services (0.43), and is lowest in accommodation and food services, construction, and W&R (Figure 12B). Figure 12. Concentration in the Kosovar Economy a) Herfindahl Index, 2010–18 b) Herfindahl Index by sector, 2018 0.35 Electricity, gas, and related act. 0.92 Information and communication 0.44 0.30 0.38 0.280 Mining and quarrying 0.268 0.274 0.262 0.263 0.251 0.253 0.245 Financial services 0.27 0.25 0.246 Real estate activities 0.25 0.20 Administrative and support services 0.24 Water supply; sewerage, waste 0.15 manag. and remediation activities 0.15 Manufacturing 0.13 Wholesale and retail trade 0.11 0.10 Professional, scientific and 0.09 technical activities Transportation and storage 0.05 0.05 Construction 0.04 0.00 Accommodation and food services 0.01 2010 2011 2012 2013 2014 2015 2016 2017 2018 0 0.20 0.40 0.60 0.80 1.00 Source: KTR; World Bank staff calculations. Note: The normalized HHIS is 1 minus the squared sum of market shares of firm sales at four-digit level NACE Rev. 2. Shares are weighted by total sector sales. The index ranges from 0 to 1, where 1 is the highest level of economic concentration. See Annex 1 for more detail on sector classification. Firm Characteristics and Dynamics 21 1.2. Firm Dynamics Firm dynamics have a crucial bearing on productivity growth. According to economic theory and the available empirical evidence (Maloney and Cusolito, 2017), entry of new more productive firms, exit of companies with lower productivity or the increase in the average productivity of incumbent firms are key drivers for productivity growth in any economy. Thus, this section delves into the recent dynamics of firms in Kosovo with the aim of setting the background for the analysis of firm productivity presented below in Section 1.3. For 2012–18 Kosovo’s net rate of firm creation was close to zero because firm entry and exit rates have been similar. Although the firm entry rates are also similar to some of the most dynamic aspirational economies (e.g. Lithuania and Latvia), it has much higher firm exit rates than the EU average and tracks some of the most developed aspirational peers, resulting in an average net annual firm turnover rate of just 0.3 percent (Figure 13A). During this period, only five sectors posted positive net firm turnover rates: manufacturing 3.12 percent, mining 3.10 percent, utilities 3 percent, accommodation and food services 2.5 percent, and construction 2.3 percent (Figure 13B). As expected, entry and exit rates are much higher for micro and small firms than for larger companies (Figure 13C). Figure 13. Net Firm Turnover, Entry. and Exit Rates, Percent of Firms, Average 2012–18 a) Entry, Exit, and Net Firm Turnover in Kosovo and Aspirational Peers 25 Entry rate % of firms 21.2 Exit rate 20 Net turnover rate 17.3 15 15.0 13.5 13.2 12.8 10.6 10.0 11.1 10 8.3 9.2 8.0 8.9 8.8 8.8 8.3 5 3.9 4.4 2.8 2.8 0.3 1.2 0.0 0.5 0 Lithuania Latvia Kosovo Slovakia Estonia EU (2020) Croatia Czechia b) Net Firm Turnover Rate by Sector Manufacturing 3.1 Mining and quarrying 3.1 Utilities 3.0 Accommodation and food services 2.5 Construction 2.3 Average 0.3 -0.3 Real estate activities -0.7 Information and communication -0.7 Wholesale and retail trade -1.4 Transportation and storage -1.8 Administrative and support services -1.9 Professional, scientific and technical activities -3.0 Financial services -4 -3 -2 -1 0% 1 2 3 4 c) Entry and Exit Rates by Firm Size, Percent of Total Active Firms 16 Entry Exit 14 14 13 12 10 8 6 5 4 4 2 2 2 2 2 0% 0-9 10-49 50-249 250+ Source: KTR and Eurostat data; World Bank staff calculations. Note: Entry and exit rates are total active units that enter and exit the market divided by total active enterprises. A unit enters the market in t if it is active in t and it was not active in t-1 and t-2; and exits the market i if it is active in t and it is not active in t+1 and t+2. EU entry and exit rates are based on the same definitions and criteria. Activities: B-N (excluding K 64.2) of NACE Rev. 2. See Annex 1 for more detail on sector classification. Micro are firms with an average of 0–9 employees, small 10–49 employees, medium 50–249 employees, and large 250 and more. 22 Firm Characteristics and Dynamics Kosovo’s new business density is slightly above what its income per capita would predict but lower than in some comparators and aspirational peers. Figure 14 plots new business density, a proxy for firm entry used because international data are available and comparable, against income per inhabitant adjusted for purchasing power parity (PPP). It shows that the number of new firms per thousand people is far higher than Kosovo’s GDP per capita would predict. Nevertheless, some Western Balkans and aspirational peers have higher new business density indicators. Unfortunately, there are no internationally comparable data on firm exit rates. Figure 14. New Business Density and GDP per Capita in Constant US$ PPP, 2018 New registrations per , people aged - Kosovo Comparators Aspirational peers Upper-middle income countries High income countries R² = . . . . . . . . . GDP per capita (PPP, in logs) Source: Entrepreneurship Survey and database, http://www.doingbusiness.org/data/exploretopics/entrepreneurship Administrative barriers to exit are high. Figure 15A shows the difference between the number of firms that terminate their legal registration and the number that stop reporting sales, the latter being the standard economic definition of firm exit.17 Between 2010 and 2018, 3,400 companies “exited” annually on average, whereas fewer than 200 firms formally deregistered. The discrepancy between economic entry and administrative registration rates is far less pronounced, suggesting fewer major regulatory obstacles to creating new companies (Figure 14B). Figure 15. Economic and Administrative Firm Entry and Exit, Number of Firms, 2012–18 a) Entry b) Exit Economic Economic Administrative Administrative 6,000 5,426 6,000 5,000 4,410 5,000 4,320 4,009 3,982 3,907 3,739 3,601 3,449 4,000 4,000 3,422 3,419 3,368 3,285 3,238 3,056 2,971 2,913 2,809 2,784 2,774 2,508 3,000 3,000 2,000 2,000 1,000 1,000 388 323 200 192 191 158 141 0 0 2012 2013 2014 2015 2016 2017 2018 2010 2011 2012 2013 2014 2015 2016 Source: KTR; World Bank staff calculations. Note: Activities: B-N (excluding K 64.2) of NACE Rev. 2. Administrative defines entry and exit using the firm’s registration year in the Tax Agency registry; Economic defines entry and exit according to firm sales. See Annex 1.A for more detail. 17 See Annex 1 for further technical details on the economic definition of firm entry and exit rates. Firm Characteristics and Dynamics 23 Entrants are much larger than firms that exit but smaller than incumbent firms. New firms are only about one-third of the size of incumbent companies and about twice as large as firms exiting. The relatively large size of firms at entry compared to companies already in the market (Figure 16) could be related to high barriers to entering the market, perhaps connected to informal competition (see World Bank 2018). Figure 16. Size of Firms Entering a Market Relative to the Average Firm Already Active, 2012–18, Percent Entry 39 32 Exit 21 22 Incumbent 121 129 Source: KTR; World Bank staff calculations. Notes: The average number of employees and turnover (in real 2018 euros) are calculated as the total number of employees and turnover of each category divided by the number of firms. Activities: B-N (excluding K64.2) of NACE Rev. 2. See Annex 1 A for definitions of entry, exit, and incumbent. The change in entry rates was driven by the increasing participation of sectors with higher rates of firm creation; the change in exit rates was driven by a rise in the average exit rate for all sectors. More than 90 percent of the change in the entry rate between 2012 and 2018 was explained by an increasing share of sectors with higher firm entry rates (“between-sectors”). In contrast, most of the change in the aggregate exit rate resulted from a generalized increase in exit rates across all sectors (“within-sectors”). Moreover, the change of about 1 percent in the entry rate was far lower than the 0.72 percent change in the exit rate (Figure 17). Figure 17. Drivers of Entry and Exit Rates, Percentage Points 0 Percentage point 1 Within-sectors Entry rate change Between-sectors (2012-2018) Cross-change 0.09 0.64 +0.72 Change Exit rate change (2010-2016) -0.04 1.01 +0.97 Source: KTR data; World Bank staff calculations. Notes: The figure depicts changes in entry rates due to variation “within-sectors”, changes in the participation of industries with different level of dynamism (“between-sectors”), and the covariance between changes in sector shares and level of firm dynamism (“cross- change”). Activities: B-N (excluding K64.2) of NACE Rev. 2. 24 Firm Characteristics and Dynamics Firm size at entry is lower than would be predicted by Kosovo’s income per capita and in aspirational peers but similar to Western Balkans comparators. New firms experience most of their employment growth in the first year, with growth extremely low thereafter. Firms enter with an average of 3.3 employees, jump to 4.6 workers in their first year, then add workers very slowly (Figure 18A). Moreover, entrants have a smaller median size than would be predicted by income per capita and the aspirational economies, although similar to most Western Balkans comparators (Figure 18B). Figure 18. Firm Size at Entry a) Median number of employees after entry by firm age* Employees 7 6 5.8 6.0 5.5 5.0 5 4.6 4 3.3 3 2 1 0 0 1 2 3 4 5 Age b) Median number of employees at entry and GDP per capita in US$ PPP. 2018 Median size (employees) Kosovo Comparators Aspirational peers Rest of the countries R² = . ALB KOS MKD SRB MNE BIH GDP per capita (PPP constant %, log) Source: KTR and Enterprise Surveys data; World Bank staff calculations. Notes: Average entry size and average size of active firms are the total number of employed persons divided by the number of total active unit of each category. Activities: B-N (excluding K64.2) of NACE Rev. 2. Over a five-year period, 33 percent of firms are born as and remain micro enterprises, 44.5 percent of the micro enterprises continue to be micro, and other firm classes similarly lack dynamism. Focusing on the cohorts of firms in 2012 and 2013, Table 2 shows that about 58 percent of small, 64 percent of medium, and 74.4 percent of large enterprises stay in the same class over a five-year period. Only about 2 percent of micro enterprises became small firms, 5.5 percent of small become medium, and 4.6 percent of medium firms become large. Firm Characteristics and Dynamics 25 Table 2.  Five-Year Transition Matrix by Firm Class, 2012 and 2013 Firm Cohorts, Average Size class five years after the year of reference Micro Small Medium Large Economic Exit Size class at reference year (0-9) (10-49) (50-249) (250+) Births 33.7 2.6 0.4 0.1 63.2 Micro (0-9) 44.5 2.3 0.0 0.0 53.2 Small (10-49) 17.9 57.7 5.5 0.0 18.9 Medium (50-249) 5.0 12.9 63.6 4.6 13.9 Large (250+) 3.5 0.0 8.1 74.4 14.0 Source: KTR data; World Bank staff calculations. Notes: This table shows the share of firms by size class in the year of reference and five years later; we use data for the 2012 and 2013 firm cohorts to identify entrants and incumbents and track changes in the number of employees over a five-year period. To control for potential idiosyncratic shocks in a particular year, calculations are made by taking the average between the two firm cohorts. Year of reference refers to the first year the firm enter the panel, either 2012 or 2013. Economic exit are classified as those units that report positive turnover in year of reference but do not report turnover in the next two years. Size class is defined according to the number of employees of the firm in the year of reference, as noted in parentheses. Firm mortality is higher for smaller companies and a notable proportion of small and medium enterprises become smaller over time. In line with the international evidence (Hsieh and Klenow 2014; Cusolito and Maloney 2019) closure rates are highest among new firms (63 percent) and micro enterprises (53 percent) and lowest among medium and large firms (about 14 percent). Moreover, about 18 percent of small companies become micro-sized and about 13 percent of medium become small (Table 2). Micro firms have the lowest employment growth rates; small, and especially medium, firms show higher and more stable growth. In line with the international evidence (Hsieh and Klenow 2014), younger firms have higher employment growth than older firms, especially in their first two years after birth (Figure 19A). Over time micro firms consistently have lower employment growth than small and medium enterprises, and their growth rates may turn negative or close to zero in 10 years (Figure 19b). Figure 19. Firm age and Employment Growth b) Firm Age and Employment Growth by Number of Employees at Birth a) Firm Age and Employment Growth (Percent) (Size Class) 10% 35% Micro (0-9) Employment growth (%) Employment growth (%) Small (10-49) 30% Medium (50-249) 25% 5% 20% 15% 10% 0% 5% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Firm age Firm age Source: KTR data; World Bank staff calculations. Notes: Annual growth rate is calculated as the number of employed persons in the period of reference relative to the number employed in the year before. To increase sample size and analyze the relationship between firm age and employment growth more robustly, we calculate age as the difference of a given year relative to the year the firm was registered. Figure 19B is based on data only available for five years. Both figures include the entire universe of firms (e.g. incumbents, entrants and exiters). 26 Firm Characteristics and Dynamics Conditional on surviving five years after their creation, companies born micro (0–9 employees) and small (10–49) grow faster and more sustainably than those born medium (50–249), which suggests that there are barriers to growth for larger firms. Figure 20 shows that micro enterprises are able to increase the number of employees 2.5 times over a five-year period, but small companies only 2 times. Medium companies increase employment by 1.7 times up to the third year but then become practically stagnate. Although this declining rate of growth relative to firm size has been observed in many countries (Syverson 2011; Cusolito and Maloney 2019), in Kosovo it suggests the presence of restrictions on firm growth beyond a certain threshold of workers. Figure 20. Employees by Firm Age and Size over Five Years, Average 2012–18, Index, Entry Year = 1 2.5 Micro (0-9) 2.0 Small (10-49) Medium (50-249) 1.5 1.0 0 1 2 3 4 5 Source: KTR data; World Bank staff calculations. Notes: The figure depicts the fraction of persons employed in active units relative to entry, by size class at birth solely including firms that survived after their first year. Since there are many entry cohorts, we average employment figures across cohorts. Figures in parenthesis in the x axis labels refer to number of employees. Only about 74 percent of firms survive the year after they enter the market, 10 percentage points less than such firms in aspirational peers and the EU. Considering all firms irrespective whether they reported or not sales in previous years, this difference is also fairly constant over the next four years of firm operations (Figure 21A). Conditional on surviving in the previous year, firm survival in the first year after entry, the “valley of death”, is also markedly lower than in the EU and in the aspirational peers, although it gradually converges in the subsequent years (Figure 21.b). Figure 21. Firm Survival in Kosovo, Average 2012–18, Percent a) Firms that survive by age after entry Percent of firms b) Firms that survive for another year (Conditional on surviving) 100 Kosovo EU (2020) 84.0 Czechia Croatia 89.4 84.0 89.0 Slovenia 86.0 70.6 83.6 60.5 73.8 54.7 85.0 84.5 83.6 48.5 83.7 61.8 51.6 73.8 43.9 37.1 0 1 2 3 4 5 Age 0 1 2 3 4 5 Age Firm Characteristics and Dynamics 27 Figure 21. Firm Survival in Kosovo, Average 2012–18, Percent c) Survival rates by firm size (number of employees) 100.0 95.0 89.6 86.1 83.3 95.5 81.9 87.5 80.9 73.2 72.2 61.1 63.2 58.0 Micro (0-9) 50.9 Small (10-49) 43.2 36.5 Medium (50-249) 30.8 0 1 2 3 4 5 6 Source: KTR and OECD Business Dynamic Indicators data; World Bank staff calculations. Notes: the figure depicts the percent of active units that survive at different ages after entry. Since there are many entry cohorts, we average the cohort figures. For example, the survival rate at year 1 is the number of units in the reference period (t) newly born in t-1 having survived to t divided by the number of enterprise births in t-1. For selected EU countries, we use 2010-2017 averages. Activities: B-N (excluding K64. 2) of NACE Rev. 2. While Figure 20 a reports the annual survival rate for all firms, irrespective whether they report sales in the previous years, Figure 20 b calculates the rate of survival per year conditional on the firm reporting sales in the panel in the previous year Only about 70 percent of micro firms survive the first year, and in subsequent years their survival rates are much lower than those of small and medium enterprises. In the first year, 95.5 percent of small and 100 percent of medium enterprises also survive. In years two to five, the survival differential between micro and medium companies is on average 30 percentage points (pp) and between micro and large firms 40 pp (Figure 20C). According to the administrative data employed in this note, between 2010 and 2018 Kosovo’s economy created about 80,000 new jobs. Between 2011 and 2014 jobs created averaged just under 5,000, and from 2015 up to 2018, jobs created annually averaged 10,000 (Figure 22). Employment creation did accelerate starting in 2015 and by 2018 had reached 180,600.18 Figure 22. Thousands of Employees in Active Firms, 2010–18 Thousands of YoY change employed persons YoY change 200 40,000 181 169 Total employed 160 145 155 32,000 133 persons 125 130 120 118 24,000 100 18,278 80 13,764 16,000 11,801 9,988 11,827 40 6,436 5,461 8,000 3,058 0 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Notes: Employed persons in active firms (based on sales). Activities B-N (excluding K64.2) of NACE Rev. 2. Source: World Bank’s own calculations based on KTR. 18 The administrative data broadly coincides with KTR information; according to KSA national accounts data, gross employment, measured by number of employees, went up by about 50 percent between 2009 and 2019 and 20 percent between 2015 and 2019 (https://askdata.rks-gov.net/PXWeb/pxweb/en/ askdata/). 28 Firm Characteristics and Dynamics Job creation was partly related to workers being registered by firms that had not reported employees before. The proportion of active firms reporting employees went from 63 percent in 2014 to 71 percent in 2018 (Figure 23), and new registrations explained about 22 percent of net job creation. Similarly, Table A1.1 in Annex 1 shows that the average number of employees per firm practically doubled (from 3 to 6) between 2010 and 2018 but the median number basically did not vary. Finally, Table A2.4 in Annex 2 also indicates that firms that started reporting employees experienced faster employment growth in the years thereafter. Figure 23. Firms Reporting Employees, Percent of Total Firms, 2010–2018 Employers Non-employers 49 50 54 60 63 71 69 69 71 51 50 46 40 37 29 31 31 29 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: KTR; World Bank staff calculations. Notes: Activities B-N (excluding K64.2) of NACE Rev. 2. Younger firms with higher productivity and access to credit had higher net job creation rates. Firms with higher revenues per capita tend to experience higher net job growth rates (Table A2.4. The magnitude of the coefficient is even larger when controlling for industry (NACE 4 digits), geographic (municipality), and time (year) fixed effects. The negative sign of the age coefficient suggests that younger firms experience higher net job creation. Access to credit is also positively correlated with more rapid employment growth. By sector, employment growth was mainly driven by W&R and other services and to a lesser extent by manufacturing and construction. Other services and W&R accounted for about 25 pp out of nearly 35 percent of total job growth in both 2010–14 and 2015–18. Manufacturing and construction explained between 4.3–4.9 and 2.8–4.8 percent of total job growth, depending on the period; mining and utilities (electricity, gas, water and sewerage services) accounted for only a minor share (Figure 24). Firm Characteristics and Dynamics 29 Figure 24. Contribution to Employment Growth by Sector, 2010–14 and 2015–18, Percent 2010- 2014 1.35 10.2 13.9 4.3 2.8 33.2 Other services Wholesale and retail Manufacturing Construction and real estate Mining Utilities 2015- 2018 -0.22 13.1 12.3 4.9 4.8 35.6 Source: KTR; World Bank staff calculations. Note: Activities B-N (excluding K64.2) of NACE Rev. 2. ICT and professional services are the most dynamic job creators. The data on job creation in the five years after firm entry reveal that the most dynamic activities are information and communication, which averages 6.7 employees, and professional services, 4. In industries with large numbers of workers and contribution to total job creation, such as W&R, construction, and manufacturing, new firms managed to create only half as many jobs as the more dynamic industries (Figure 25). Figure 25. Employment Growth, Firms by Sector Surviving 5 Years, Average Number of Employees, 2017–18 Manufacturing Gap between entry size 2.8 and size at age 5 Construction 2.4 Wholesale and retail trade 2.4 Transportation and storage 3.2 Accommodation and food services 1.4 Information and communication 6.7 Professional, scientific and technical activities 4.0 Employees 0 2 4 6 8 10 12 14 Source: KTR; World Bank staff calculations. Notes: Difference between t the number of employees and the number of employees five year later conditional on the firm “surviving” or reporting sales in the previous year. We report activities in which more than 10 companies enter the market. See Annex 1 for more detailed sector classification. From 2015 on, annual net job creation doubled, driven by rising job creation and falling job destruction, almost all by incumbent firms. As a result, the economy has been creating an average of about 10,000 net new jobs annually, about twice the net jobs created annually between 2011 and 2014 (Figure 26A). Rising net job creation among incumbent firms, and to a lesser extent, entry of new firms, along with relatively stable exit rates, largely explain the rise in aggregate net job creation (Figure 26B). 30 Firm Characteristics and Dynamics Figure 26. Net Job Creation, 2011–18 a) Net change in number of jobs Employed persons 35,000 32,803 31,393 Job Creation 30,000 28,964 27,725 Net Job Change 25,949 25,000 22,606 23,772 Job Destruction 20,679 20,000 19,566 15,000 18,311 17,621 17,737 16,170 15,200 14,525 14,148 10,000 5,000 0 2011 2012 2013 2014 2015 2016 2017 2018 b) Incumbents entering and exiting 30,000 Incumbent (creation) 20,000 Entry Net by incumbents 8,639 8,691 9,745 8,066 Exit 10,000 6,705 Incumbent (destruction) 3,090 2,563 18 0 -10,000 -20,000 2011 2012 2013 2014 2015 2016 2017 2018 Source: KTR data; World Bank staff calculations. Note: Net job change equals gross job creation minus gross job destruction. Activities: B-N (excluding K64.2) of NACE Rev. 2. See Annex 1.A for detailed definitions of incumbent, entry, and exit. Young SMEs are driving job creation. Although they contribute just 25 percent of total employment, young SMEs accounted for nearly 40 percent of the jobs created annually between 2016 and 2018. As firms mature, they lose some dynamism. Even among small and medium companies, mature and old firms explain 50 percent of total employment but only one-third of gross job creation. Furthermore, large, mature firms contribute less than proportionately to gross job creation (Figure 27). Firm Characteristics and Dynamics 31 Figure 27. Job Creation, Destruction, and Employment by Firm Age and Size, Average 2016–18 Contribution to job creation 52 51 Contribution to job destruction Contribution to employment 41 36 30 24 22 21 16 2 2 3 Young (1-5) Mature and Old (>5) Young (1-5) Mature and Old (>5) SMEs (1-249) Large (250+) Source: KTR; World Bank staff calculations. Notes: Using the same criteria as Criscuolo, Gal, and Menon 2014 we exclude entrants (0 years). Young are companies 0–5 years of age; mature and old firms have lasted more than five years. Activities: B-N (excluding K64.2) of NACE Rev. 2. Micro firms average 0–9 employees, small 10–49, medium 50–249, and large 250 and more. In 2010–18 young firms and startups, including entrants, accounted for about 45 percent of total employment. On average, startups accounted for about 33 percent of total employment, young firms for about 16 percent, and mature firms for the rest. However, there is considerable heterogeneity by sector. Setting aside entrants, the proportion of jobs in younger firms is largest in accommodation and food services, 74 percent, and professional and administrative services, about 60 percent each. In contrast, about 95 percent of workers in financial services and water supply and 90 percent in mining are employed by mature firms (Figure 28). Figure 28. Employment by Firm Age, Percent of Total, Average 2016–18 Accommodation and food services Entrants (0 years) Start-ups (1-2) Administrative and support services Young (3-5) Construction Mature and old (6+) Electricity, gas and related act. Financial and insurance services Information and Communication Manufacturing Mining Professional, sci. and tech. act. Real Estate Transportation and Storage Water supply, sewerage and waste mgmt. Wholesale and Retail trade 0% 20% 40% 60% 80% 100% Source: KTR; World Bank staff calculations. Notes: Entrant firms have less than a year in the market, startups up to 2 years, young up to 5 years, and maturing more than 5 years. Activities: B-N (excluding K64.2) of NACE Rev. 2. Calculations for this figure do not match exactly the estimates in Figure 26 because they incorporate entrants and startups and do not consider the contribution of age by firm size class. See Annex 1 for more detailed sector classification. 32 Firm Characteristics and Dynamics Firm Characteristics and Dynamics 33 02 Firm Productivity and Access to Credit This section first presents the main results of our estimates of TFPR and LP for 2013–17.19 It then examines the drivers of TFPR growth and compares TPFR in domestic firms, firms with foreign shareholders, and exporters. It also assesses productivity differences by the gender of the firm decisionmaker. Finally, it delves into the links between access to credit and productivity and the main barriers to productivity growth. 19 We exclude 2018 from the productivity analysis because reporting of key variables, such as investment and purchases of inputs is significantly lower in that year than in previous years. 1.3. TFPR and Labor Productivity Between 2013 and 2017, TFPR grew by 0.5 percent annually, but LP (value added per worker) fell by –0.25 percent. Although TFPR growth in Kosovo was negative in 2013 (–0.5 percent), between 2014 and 2017 it grew at an annual average of about 0.7 percent. While LP growth, proxied by value added per worker, was negative in 2013–15, it was marginally positive in 2016 at 0.17 percent and 2017 at 0.26 percent. During this period proxies for LP, sales per worker, and average wages per worker, exhibited similar dynamics (Figure 29A). This timid recovery in productivity growth rates coincided with the acceleration in net job creation (see Figure 21). Between 2012 and 2017, TFPR grew by 2.4 percent, while LP measured as sales fell by 1.1 percent, as value added per worker by 1.2 percent, and as sales per worker by 0.5 percent (Figure 29B). Figure 29. TFPR, Sales, and Value Added per Worker, 2013–17 a) Annual Percent Change TFPR 0.7% 0.8% Value added per worker 0.8% 0.6% Sales per worker 0.0% 0.2% 0.3% 0.1% 0.1% -0.3% 0.5% -0.2% -0.5% -0.9% -1.0% 2013 2014 2015 2016 2017 b) Index 2013=100 103 TFPR Value added per worker 102 Sales per worker Wage per worker 101 100 99 98 97 2012 2013 2014 2015 2016 2017 Source: KTR; World Bank staff calculations. Notes: TFPR and value added per worker are adjusted for firm entry and exit (Melitz and Polanec, 2015). TFPR is estimated following Levinsohn and Petrin (2003). Figures are weighted by firm employment in the entire sample dataset (see Box 1 for details). We include the total business economy, repair of computers, personal and household goods, except financial and insurance activities. 36 Firm Productivity and Access to Credit Labor productivity in Kosovo is only one-third that of the EU; and it is much lower than in aspirational countries. Kosovo’s LP (sales per worker) of 33.5 percent of the EU average compares badly to Lithuania’s 53 percent and the even higher LPs of Slovakia, Czechia, Estonia, and Slovenia (Figure 30). Figure 30. Labor Productivity Relative to the EU, 2018, Percent 89.3 84.9 81.8 73.4 53.2 31.8 Kosovo Lithuania Slovakia Czechia Estonia Slovenia Source: KTR; World Bank staff calculations. Notes: Labor productivity is calculated as the ratio of total turnover (sales) in euros to the total number of employees. We include the total business economy, repair of computers, and personal and household goods, except financial and insurance activities. The figure depicts LP in each country relative to the EU average LP as percent. Between 2012 and 2017 use of capital and intermediate inputs by Kosovar firms declined. Figure 31 suggests that during the period TFPR growth was partly driven by that decline, although firm output decreased to a much lesser extent; firms were able to produce the same output with less inputs of production. In parallel, the use of labor fell less than proportionately to the use of capital, suggesting that firms also became more labor-intensive. Figure 31. Labor, Capital, and Intermediate Inputs, 2012–17, Index 100=2012 100 Labor Sales 80 Intermediate consumption Capital stock 60 2012 2013 2014 2015 2016 2017 Source: KTR; World Bank staff calculations. Notes: Labor is calculated as the number of employees times the average wage (wage bill) in order to control for possible heterogeneities in worker labor productivity. The figure only comprises the sample of firms with data available to estimate TFPR. Activities: B-N (excluding K64.2) of NACE Rev. 2. Firm Productivity and Access to Credit 37 Manufacturing is the activity with the highest TFPR—about 26 percent higher than the average sector, closely followed by retail (24.7 percent) and wholesale (23.2 percent). Construction and other services are the industries with the lowest TFPR, about 30 percent lower than the average sector (Figure 32A). Looking at LP, Figure 32B shows wholesale trade has the highest value added per worker, follow by construction and manufacturing. Figure 32. TFPR and value added per worker by Sector and Average a) TFPR by Sector and Average in logs TFPR (log) Average 4.0 3.5 3.0 2.5 2.40 2.34 2.37 2.0 1.5 1.31 1.28 1.0 0.5 0.0 Manufacturing Construction Other services Wholesale trade Retail trade b) Value Added per Worker by Sector and Average in Euros Value added Average per worker 16,000 14,421 14,000 12,854 12,000 11,663 10,903 10,412 10,000 8,000 6,000 4,000 2,000 0 Manufacturing Construction Other services Wholesale trade Retail trade Source: KTR; World Bank staff calculations. Notes: Labor productivity is calculated as total value added (total sales minus input purchases) divided by the number of employees. Like the TFPR calculations, LP estimates were adjusted by for firm entry and exit (Melitz and Polanec 2015). Sector averages were calculated using firm employment weights (see Box 1 for details). Activities: B-N (excluding K64.2) of NACE Rev. 2. TFPR growth has been driven by market reallocation toward more productive firms. Box 2 explains how the sources of TFPR growth are decomposed in this note following Melitz and Polanec (2015). Figure 33 shows that in the 2012–17 period about 80 percent of average TFPR annual growth rate explained by the increasing productivity of incumbent firms (covariance), reflecting a reallocation of labor and other inputs to the most productive companies (Foster, Haltiwanger, and Krizan 2001). 38 Firm Productivity and Access to Credit Figure 33. Decomposition of TFPR Growth, 2014–17, Percent Change -0.4% 0.6% 1.5% 1.8% 0.1% Entry Within Covariance Aggregate Exit productivity growth Source: KTR; World Bank staff calculations. Notes: TFPR was calculated following Levinsohn and Petrin (2006), adjusted for firm entry and exit (Melitz and Polanec 2015). Covariance refers to the change in the share and average productivity of incumbent firms; within denotes the change in average firm productivity; and entry and exit refers to the contribution to aggregate TFPR of firms exiting and entering the market. Activities: B-N (excluding K64.2) of NACE Rev. 2. The contribution of firm productivity, typically related to the adoption of technology, innovation, and improved management techniques, was much lower in Kosovo than in other developing countries; firm dynamism had only a marginal role. “Within” firm productivity growth only accounted for about a third of TFPR growth, far less than in other developing countries for which there is comparable data.20 Underscoring once again the lack of dynamism of Kosovo’s economy, exit of less productive firms had only a relatively minor, though positive, role, and entry of new firms contributed negatively to aggregate TFPR growth (Figure 32). Box 2. Decomposing the Sources of TFPR Growth21 In this note, TFPR is decomposed into different components following Melitz and Polanec (2015) to gauge how much surviving, entering, and exiting firms contribute to aggregate productivity changes. The contribution of incumbent firms is decomposed further into the reallocation of resources across firms by the between component (measured by the covariance term); and increases in productivity within firms. The between component is positive when resources are allocated from less productive to more productive firms and the within component is positive when existing firms become more productive due to technology adoption, innovation, and better managerial skills. Among barriers to reallocation of resources might be trade barriers, excessive regulations, and overbearing state-owned enterprises. Kosovo’s economy is producing low-quality entrants: new firms are smaller, pay lower wages, and are far less productive than the median company. Figure 34 illustrates22 that entrants’ sales are only 33 percent of a median company’s sales. Similarly, employment of a Kosovar entrant is only about 40 percent of that of a median company’s. Moreover, entrants pay 30 percent less in wages, sales per worker is only about 20 percent, and TFPR is 11 percent lower when compared to the median firm. 20 According to Iacovone et al. (2018), the “within” component accounted for more than 40 percent of the variation in TFPR growth in their sample of Latin American economies. Similarly, Cusolito and Maloney (2018) found that the “within” component accounts for more than 33 percent of TPFR variation in their sample of developing countries. 21 Based on Cusolito and Maloney (2018) 22 It refers to the median LP and TFPR of incumbents, entrants, and exiting firms. Firm Productivity and Access to Credit 39 Figure 34. Characteristics of Entrants Relative to the Median Firm. Percent Employment 38.8 Turnover 32.3 Wages 71.1 Sales per worker 79.3 TFP 88.7 Source: KTR; World Bank staff calculations. Notes: TFPR was adjusted for firm entry and exit (Melitz and Polanec 2015). TFPR is estimated following Levinsohn and Petrin (2003). Figures are weighted by employment in the entire sample dataset (see Box 1 for details). Activities: B-N (excluding K64.2) of NACE Rev. 2. (See Annex 1A for more detail of the definitions of incumbent, entry and exit. There are large and persistent differences in TFPR and LP by firm and sector. For instance, median TFPR in the 90th percentile is 1.6 times higher than in the 10th percentile. As for LP, the differences in productivity are even starker: about 90 times between leading (90th percentile) and laggard firms (10th percentile). By sector, dispersion in TFPR is largest in wholesale trade (1.9 times) and manufacturing (1.8), but less marked in construction (1.3) and services (1.34). Similar results are obtained for LP but with larger differences. By firm size, micro companies (0-9 employees) exhibit the largest distance between leading and laggard firms in LP (about 98.7 times), although the difference is smaller in TFPR. By contrast, productivity dispersion is largest in medium (50-249 employees) and large (250+) companies with respect to median TFPR than with respect to median LP (Table 3). Table 3.  Ratio of Productivity in “Laggard” (90th percentile) to “Frontier” (10th percentile) Firms by Sector and Size, Average 2012-18 Dimension Category TFPR Sales per worker Sector Other services 1.445 52.5 Manufacturing 1.797 52.1 Retail trade 1.646 65.9 Wholesale trade 1.878 126.7 Construction 1.328 149.7 Size class Micro (0-9) 1.456 98.7 Small (10-49) 1.540 27.4 Medium and large (50+) 1.879 23.1 Sector average 1.595 89.370 Source: KTR; World Bank staff calculations. Notes: TFPR was calculated following Levisohn and Petrin (2006), adjusted for firm entry and exit (Melitz and Polanec 2015). LP is also adjusted by firm entry and exit. TFPR and LP dispersion is calculated as the ratio between mean TFPR/LP in the 90th percentile relative to mean TFPR/LP in the 90th percentile at NACE 2-digit. 40 Firm Productivity and Access to Credit Micro firms are much less productive than small, medium, and large firms, especially in wholesale and manufacturing. Median TFPR in small firms is more than double than median TFPR in micro firms and is 1.3 times larger in large companies. These differences are even more pronounced in wholesale, in which median TFPR in medium and large firms is 1.6 times higher than in micro firms, and in manufacturing, where median TFPR in medium and large firms is 1.5 times higher than micro firms; the productivity gap between micro and other firm classes is smaller in construction (1.03) and other services (1.07) (Table 4). Table 4.  TFPR in Micro-Enterprises Relative to Other Firm Classes Sector Size class Ratio Manufacturing Small 1.136 Medium and Large 1.467 Construction Small 0.999 Medium and Large 1.038 Services Small 0.976 Medium and Large 1.070 Wholesale Small 1.147 Medium and Large 1.591 Retail Small 1.132 Medium and Large 1.407 Average Small 1.078 Medium and Large 1.315 Source: KTR; World Bank staff calculations. Notes: TFPR was calculated following Levisohn and Petrin (2006), adjusted for firm entry and exit (Melitz and Polanec 2015). LP iis also adjusted by firm entry and exit. Dispersion is the average logged difference between TFPR in micro firms and the other firm classes. The TFPR ratio equals elnT(FPRm) -ln(TFPRj) where ln(TFPRm) is the logged median TFPR in m micro-firms and ln(TFPTj) is the logged median in j, the other firm classes. Higher TFPR is correlated with higher employment, revenues per worker, and to a lesser extent, value added per worker. The simple correlations plotted in Figure 35 indicate that more productive firms have far more workers and sales per worker, and to a lesser extent, more value- added per worker. Table A2.5 in Annex 2 shows that these correlations are robust to controlling for firm size and capital stock as well as sector, location (municipality), and year fixed effects. Figure 35. TFPR, Employment, Revenue, and Value Added per Worker, Logarithms, 2012–18 TFPR and Employment (log) TFPR and Revenue per worker (log) TFPR and VA per worker (log) TFPR (log) relative to average 1.5 1.5 1.5 1 1 1 .5 .5 .5 0 0 0 -.5 -.5 -.5 -4 -2 0 2 4 6 -5 -2.5 0 2.5 5 -3 -2 -1 0 1 2 3 Employment (log) relative to average Source: KTR; World Bank staff calculations. Notes: TFPR, employment, revenue, and value added per worker (in logs) relative to industry-by-year average. Activities: B-N (excluding K64.2) of NACE Rev. 2. Firm Productivity and Access to Credit 41 Wholesale was the activity that had the highest TFPR growth rate in 2013–17. Between 2013 and 2017, TFPR in wholesale grew by 6.55 percent, followed by retail (5 percent) and construction (3.26 percent). In contrast, TFPR in manufacturing fell by –2.2 percent, but growth was also low in other services (0.6 percent). During that period, wholesale also had the highest annual average TFPR growth, 1.3 percent, almost triple the economy’s average of 0.5 percent, followed by construction (1 percent). Manufacturing had negative average annual TPFR growth of –0.5 percent), and for other services at 0.7 percent it was just above the average (Figure 36A). Figure 36. Total Factor Productivity Revenue (TFPR) by Sector, 2013–17 a) Year-on-year percent change 4.0% Wholesale Construction Retail 3.0% Manufacturing Other services 2.0% 1.0% 0.0% -1.0% -2.0% -3.0% 2013 2014 2015 2016 2017 b) Contribution to TFPR Growth by sector, percent change, 2013–17 1.0% 0.03 0.07 0.82 0.74 0.37 0.75 0.03 0.24 0.10 0.13 0.56 0.06 0.08 0.5% 0.26 0.36 0.26 0.66 0.11 0.07 0.19 0.20 0.27 0.13 0.11 0.0% 0.02 -0.53 -0.24 -0.36 Manufacturing Construction -0.50 Other services -0.5% Wholesale -0.23 Retail Aggregate productivity growth -1.0% 2013 2014 2015 2016 2017 Source: KTR; World Bank staff calculations. Notes: TFPR is calculated following Levisohn and Petrin (2006), adjusted for firm entry and exit (Melitz and Polanec 2015). See Box 1 and Annex 1 for details. Activities: B-N (excluding K64.2) of NACE Rev. 2 Wholesale and retail were the main engines of aggregate TFPR growth between 2013–17 (Figure 36B). Wholesale accounted for about 50 percent and retail for 40 percent of yearly aggregate productivity growth. Construction explained another 17 percent and other services 5 percent, but manufacturing dragged down aggregate TFPR performance with negative growth in 2013 and 2015 and only marginal positive contributions in the other years. 42 Firm Productivity and Access to Credit Firms engaged in international trade are larger, more productive, more capital-intensive, and pay higher wages than domestically oriented companies. Firms that export and import have on average more employees and sales, higher value-added, and higher capital stock and wages per worker (Table A2.6 in Annex 2). Similar results, though with less statistically significant coefficients, are obtained for firms engaged exclusively in either exporting or importing. These results are robust to controlling for time, municipality, and industry fixed effects. Firms in which more than 50 percent of shareholders are foreign are also larger, more productive, more capital-intensive, and pay higher wages (Table A2.8). In contrast, firms with lower foreign-shareholding are statistically indistinguishable from domestic-owned companies except for higher wages and capital per employee. It appears that in Kosovo managerial control by foreign investors rather than just foreign shareholders might be correlated with higher productivity. Results are robust even when time, municipality fixed effects and firm size are controlled for. Firms with female decisionmakers, especially in manufacturing and W&R, tend to be smaller, less productive, and more labor-intensive (Table A2.8). Female-run companies have 20 percent fewer employees than male-run firms and 57 percent lower total sales. These differences are much larger for manufacturing and W&R enterprises, though not statistically significant for other sectors. Sales per worker are on average 10 percent lower in female-run firms in general, and about 30 percent lower in female-run manufacturing companies. Moreover, female-run companies average 30 percent less capital per worker, with larger differences in W&R, manufacturing, and professional services. 1.4. Access to Credit and Productivity Limited access to finance can affect firm productivity in several ways. It can prevent hiring workers, innovation, and investment. On the other hand, it can induce exit of less productive firms, which may benefit aggregate productivity growth. Credit constraints can discourage entry of potentially more productive new firms, which may also induce a fall in total productivity growth. Finally, restricted access to credit may have reallocation effects, by forcing less productive firms to shrink and by restricting expansion of more productive companies (Linarello et al. 2019) This section analyzes the links between access to credit and TFPR based on detailed micro data from the KCR, combined with firm-level information from the KTR (see Box 1 and Annex 1 for details). It first examines the recent evolution and sector distribution of credit to the private sector. It then explores two dimensions of the credit-productivity nexus: (1) what determines access to credit; and (2) the links between access to formal finance and firm TFPR growth. Between 2010 and 2018, domestic credit to the private sector in Kosovo increased by 24 percent, from 35.4 to 44.1 percent of GDP. Despite this positive performance, other comparator countries, such as the Slovak Republic, Estonia, and Croatia, registered much higher and faster growth in their credit-to-GDP ratios (Figure 37). Figure 37. Domestic Credit to the Private Sector, Percent of GDP % GDP Kosovo 95 Serbia Estonia Croatia 85 Slovak Republic Montenegro 75 Latvia 65 55 45 35 25 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Source: WDI 2021. Firm Productivity and Access to Credit 43 Less than a third of firms have access to financing in Kosovo through formal channels. The activities with the highest proportion of firms with that access are mining 47 percent, and construction 42 percent; the industries with the least access to credit are accommodation and food services and professional services, all about 20 percent (Table 5). Table 5.  Credit in Kosovo, Summary Statistics, 2018 Sector Firms Accessing Loans per Firm Total Loans Share of Total Nominal Interest Credit (%) (¤‘000) (¤ million) Loans (%) Rate (pp) Mining and quarrying 47.7 163.2 10.8 0.9 5.0 Manufacturing 32.5 139.9 198.7 18.5 4.9 Electricity, gas, steam and air-conditioning supply 28.4 449.6 5.0 0.4 2.7 Water supply; sewerage, waste mgmt. and 38.3 59.9 2.9 0.3 5.1 remediation act. Construction 42.3 195.6 171.0 15.0 3.2 Wholesale and retail trade 25.8 144.4 575.1 52.2 4.3 Transportation and storage 34.1 61.4 24.1 2.1 6.9 Accommodation and food services 20.1 46.9 31.2 2.9 6.1 Information and communication 25.2 154.5 25.2 2.4 5.9 Financial services 34.3 126.1 5.1 0.4 3.0 Real estate activities 14.5 319.7 3.3 0.3 3.5 Professional, scientific and technical activities 21.2 61.2 17.2 1.5 5.4 Administrative and support services 33.6 189.2 31.2 3.1 3.6 All firms 27.7 135.6 1,100.9 100.0 4.3 Source: KCR data; World Bank staff calculations. Note: Firms not reporting location (municipality) were excluded from the sample. More than 50 percent of banking credit is directed to W&R, followed by manufacturing (19 percent) and construction (17 percent). No other activity absorbs more than 3 percent of total financing. Accommodation, food services, and transportation have the lowest share of credit (Table 5). Credit to the private sector is therefore allocated to the economic activities with the largest shares in total employment (Figure 38). Qualitatively similar results are obtained when plotting the sectoral distribution of credit against the share of each activity in total sales or value added.23 Figure 38. Credit Disbursed and Employment by Sector, 2010–18, Percent of Total 60 Wholesale & retail trade 50 40 Mining and Transportation & Construction 30 quarrying storage 20 Manufacturing 10 Real estate Administrative and support services Information & communication 0 Accommodation & food services Water supply; sewerage, Financial services waste mgmt. and Electricity, gas, steam and rel. act. remediation act. Professional, sci. & tech. act. -10 -5 0 5 10 15 20 25 30 35 Source: KCR data; World Bank staff calculations. 23 Results not presented here in lieu of space but are available upon request to the authors. 44 Firm Productivity and Access to Credit By type of financial instrument, loans account for more than half of credit to the private sector. Overdrafts account for another 20 percent, followed by guarantees with 17 percent. Other forms of financing account for the remaining 11 percent, led by other credit assets at about 5 percent and letters of credit at about 3 percent. (Figure 39). The relative share of guarantees in total credit to the private sector has been relatively constant since 2015, averaging 15 percent. Figure 39. Credit to the Private Sector by Type, Average 2010–18, Euros and Percent of Total Loan Overdraft Other 52.8 22.3 credit assets 4.3 Guarantees 16.5 Source: KCR data; World Bank staff calculations. Note: Firms not reporting municipality were excluded from the sample. In Kosovo, 70 percent of credit to the private sector consists of short-term loans: 44 percent have a term of up to one year, 13 percent 1–2 years, and 12.7 percent 2–3 years (Figure 40). Clearly, long-term finance, which is typically linked to productivity-enhancing investments, is scarce in Kosovo. Figure 40. Loans to the Private Sector by Term (Years), Average 2015–18, Percent of Total Up to 1 year inclusive 5 years or more 2-3 years 4-5 years 43.9 13.4 12.7 10.4 1-2 years 13.0 3-4 years 6.7 Source: KCR data; World Bank staff calculations. Note: Firms not reporting municipality were excluded from the sample. Micro enterprises, firms oriented to the domestic market, and startups are particularly affected by the lack of access to credit. According to the latest BEEPS, 74 percent of SMEs in Kosovo report having needed a loan but being either rejected or discouraged from applying— the largest share among Western Balkans countries.24 Only 24 percent of micro firms have access to credit (Table 6). Moreover, access to financing is 7 pp lower for small enterprises (63 percent) than for medium and large firms (about 70 percent). 24 See https://www.beeps-ebrd.com/indicators/ Firm Productivity and Access to Credit 45 Table 6.  Access to Credit by Firm Class, Average, 2015–18 Characteristic  Type Percent of Firms Size class Micro (0–9) 24.3 Small (10–49) 63.4 Medium (50–249) 72.8 Large (250 and more) 71.2 Age class Start-ups (0–2 years) 21.5 Young (3–5 years; excludes Start-ups) 28.2 Mature (6+ years) 31.2 Foreign trade Non-exporter 26.6 Exporter 57.8 Non-importer 20.4 Importer 53.9 Foreign shareholding No foreign shareholders 45.2 0.1–24.9% of total assets 42.0 25–49.9% of total assets 36.9 50–100 of total assets 54.2 Gender of decision-maker Male-run companies 45.6 Female-run companies 38.1 Sources: Data from Kosovo Tax and Credit Registries and KPS; World Bank staff calculations Of the factors that affect creditworthiness, startups have the lowest proportion of firms with access to credit at about 22 percent, followed closely by firms not engaged in foreign trade: only 20.4 percent of non-exporters and 26.6 percent of non-importers were able to secure financing, compared to 57.8 percent of exporters and 53.9 percent of importers. Less than 40 percent of female-run companies are acceptable to the financial system compared to more than 45 percent of male-fun firms. While firms with some foreign-shareholding seem to have similar or even less access to credit than firms with only domestic shareholders, companies in which foreigners exert management control with more than 50 percent of the shares have about 10 pp greater access to finance, 54 percent, compared to 45 percent for firms that have only domestic shareholders. High performance and larger firms belonging to more concentrated sectors and located in non- majority municipalities are more likely to get access to credit.25 The determinants of access to credit are spelled out econometrically in Table A2.10: Note that the probability of obtaining credit is higher for firms with higher TFPR growth. As expected, the correlation between productivity growth and access to credit operates with some lag; financial institutions typically grant financing based on observed firm performance in the past, not on firm productivity growth at the time of granting the credit. The probability of getting access to finance is higher the more concentrated the sector the firm is in as proxied by the normalized HHI. It may be that market power drives access to credit more than firm efficiency. The results also suggest that larger firms also have a 1 percent higher probability of gaining access to finance. Enterprises in non-majority municipalities are less likely to obtain credit. Gender of the firm decisionmaker does not have a statistically significant correlation with firm access to finance. And while firm age is statistically correlated with access to finance, the magnitude of the coefficient is not economically significant. Access to credit is poorly correlated with higher firm productivity, which suggests that in the local financial system, capital is misallocated. Figure 41 examines the links between credit allocation and firm productivity in Kosovo by plotting the results of a simple regression analysis of access to credit with respect to TFPR, controlling for firm size, capital stock, sector, municipality, sector concentration (proxied by the HHI), and exporter status. The findings suggest that access to finance is not well correlated statistically with firm productivity. 25 Non-majority municipalities are municipalities with more than 40 percent of Serbian population. See Box 1 for further details. 46 Firm Productivity and Access to Credit Figure 41. Productivity and Access to Credit 60% Credit access (%) 50% 40% 30% 20% 1 TFPR (log) 1.5 2 2.5 3 Source: KCR and KTR; World Bank staff calculations Notes: The figures plot access to credit, a dichotomic variable equal to one if the firm was granted any financing and zero otherwise, against TFPR (in logs) controlling by firm size, capital stock, sector, municipality, and exporter status. 1.5. Barriers to Productivity Firms reported informality and unreliable electricity supply as the main barriers to productivity growth (Figure 42 A). One-third of Kosovar firms cited informality as the main barrier, a considerably higher proportion than in Albania and Croatia. Moreover, 64 percent of the enterprises in Kosovo cited as a barrier having to compete with an informal firm, the second largest proportion among Western Balkans countries after North Macedonia. About 25 percent of companies also cited unreliable electricity supply as a major impediment; that was a problem for only 8 percent of firms in Albania.26 While tax rates and administration in Kosovo are perceived as less stringent for investment than in comparator countries, more firms in Kosovo were concerned about political instability and corruption. Figure 42. Barriers to Investment and Exports. Percent of firms a) Barriers to investment b) Obstacles to exporting to the EU Informal competitors Lack in international 59 Electricity market linkages Tax rates Lack of skilled Political instability 38 workforce Corruption Tax administration Negative Kosovo 31 brand reputation Transport Access to finance Lack of internal 27 Customs & trade… manufacturing capacity Poorly educated workers Licensing and permits Low product quality 17 Labor regulations Crime, theft and disorder Kosovo Excessive technical 12 Access to land Albania requirements on products Croatia Courts Other 8 0% 10% 20% 30% 40% 50% Source: Enterprise Survey (2019). Source: Kosovo’s American Chamber of Commerce, “Free Movement of Goods and Conformity Assessment” 2018. 26 See https://www.beeps-ebrd.com/indicators/. Firm Productivity and Access to Credit 47 Exporters to the EU market cited lack of international links, shortages of skilled workers, and a negative brand reputation as the main barriers to growth. About 60 percent of respondent firms identified the lack of international market links as a major obstacle to exporting to the EU, and about 38 percent reported the lack of skilled workers as a major impediment. For instance, the IT sector does not have enough well-trained programmers, or even graduates with basic IT skills, which prevents companies from pursuing larger outsourcing contracts (World Bank 2016). Moreover, about 33 percent of the companies singled out “Negative Kosovo brand reputation” as a formidable barrier to trade with the EU. Lack of internal manufacturing capacity and low product quality were also named as major barriers (Figure 41B). Firms reported not only shortages of skilled workers but also a poor match between training systems and employer needs. According to STEP 2019, 75 percent of firms that had attempted to fill a high-skill position and 60 percent firms that tried to fill a medium to lower skill position encountered problems finding workers with the appropriate skill or experience. Furthermore, nearly 60 percent of employers think that general education does not give students practical experience, and almost 50 percent think it does not build up-to-date knowledge or soft skills like discipline, timeliness, and people skills. Although employers are more satisfied with the TVET outcomes, 35–42 percent of employers think that TVET systems do not provide the skills the market needs. Infrastructure gaps, too, limit firm productivity growth. Kosovo has invested heavily in new physical capital. Most main roads and local roads have been upgraded, but many national and regional roads did not receive much attention. The railway infrastructure is outdated, but only the north-south section of it merits investment because that connects with European and Asian markets. The mining sector requires, among other improvements, an upgrade of the railway system (World Bank 2016). Improving firm productivity will depend on more private investment in R&D, physical capital, and management training. Firms with higher R&D- and capital stock-to-sales ratios are more productive in term of sales and value added per worker (Table 7). Investment in training, especially management techniques, is also correlated with more productive firms. By contrast, firms subsidized by the government are less productive, as are firms with a higher proportion of unpaid workers in total workers, though the latter relationship is less statistically robust. Table 7.  Potential Correlates with Productivity in Kosovo Investment in R&D Capital Stock Unpaid Workers Government Investment in (Percent of Sales) (Percent of Sales) (Percent of Total Subsidies Training Workers) Sales per worker 0.618*** 0.218*** -0.031 -0.978*** 0.293** (0.147) (0.025) (0.121) (0.233) (0.121) Value added per worker 0.488*** 0.175*** -0.338* -1.256*** 0.248 (0.182) (0.040) (0.185) (0.342) (0.160) Source: KPS data; World Bank staff calculations. Notes: This table shows the results of a cross-sectional regression of sales and value added per worker on each of the covariates, controlling for firm size, age, gender of the decisionmaker, worker skills, firm capital stock, exporter status, municipality, and sector. Government subsidies is a dummy variable equal to one if the firm receives any government subsidy and zero otherwise. Investment in training is a dummy variable equal to one in the firm reports investing in training and zero otherwise. The Impact of COVID-19 on Firms This section assesses the economic effects of the COVID-19 shock on Kosovo’s firms based on the recent World Bank Business Pulse Survey (BPS) and updated administrative data. The BPS covered more than 100,000 businesses in 49 countries; it was conducted in Kosovo on a sample of about 1,200 in June and July 2020 to understand the economic impact of the crisis in April 2020 and firm expectations for the next six months. As in Western Balkans peers, in Kosovo sales dropped in domestic firms affected by lockdown and MSMEs. Sales eroded by 55 percent in April 2020 relative to April 2019, a slightly larger decline than in neighbors like Moldova (54.7 percent) and Albania (46.7 percent). For the whole year, based on tax records, revenues dropped by 9.5 percent. As expected, accommodation and food services, transportation and storage, real estate, and arts and entertainment suffered the most, and by firm size, smaller companies had the worst experience based on BPS. As for jobs, firms largely responded by granting leave and reducing hours and wages, but the BPS found that MSMEs experienced higher net job losses: About 40 percent of firms granted leave, 13 percent reduced hours, and 9 percent reduced wages. Only 6 percent fired workers; 2 48 Firm Productivity and Access to Credit percent hired. Firms in sectors that were shut down experienced net job losses similar to firms in sectors not shut down and used other means of adjustment, such as reducing hours and wages less intensively; new job losses were higher in medium and small firms. Consistent with the BPS findings, tax records also show that formal employment weathered the impact of the downturn. It appears that wage-subsidy and other policy measures not only helped preserve formal jobs but also encouraged formalization because employment increased, based on administrative data. However, unemployment jumped, probably from losses of informal jobs: registering as unemployed was a prerequisite for benefiting from government social protection measures. Firms globally integrated through FDI and exports dealt with the shock better than domestic- oriented firms. Average drops in sales in April 2020 were 43 percent for companies with foreign shareholders, 50.9 percent for exporters, and 55.2 percent in domestic-oriented firms. The crisis demonstrated the importance of global integration if Kosovo firms are to become more resilient. Digital adoption was weaker in Kosovo than in comparator countries, with responses highly heterogenous by firm size and manager gender. Firm responses focused on increasing use of platforms (25 percent), and to a lesser extent innovating in products and services (12 percent), although these reactions were weaker than in Moldova and Albania (Figure 32). Large firms invested the most in digital solutions and innovation and used digital platforms most intensively. As did female-run companies. Figure 43. Digitalization and Innovation in Response to the COVID-19 Outbreak Compared, Percent of Firms 73 Kosovo Moldova Albania 29 25 19 20 12 19 Increased use of Invested in Invested in digital platforms digital solutions digital solutions Source: BPS data; World Bank staff calculations. Note: *Percent of firms that report affirmatively in each category. **Figures may not add up to 100 percent because type of question varies. More than half of firms in Kosovo are already or are expecting to be in arrears, with SMEs most financially stressed, slightly more than in Moldova (53 percent) and much more than in Albania (33 percent). Arrears are likely to affect medium-size companies the most. Primary production, services and construction are the sectors in most financial distress. More firms in Kosovo reported benefiting from policy measures than in comparator countries; wage subsidies were the most common measure at end July 2020. At first complex procedures hindered access to benefits for MSMEs. The number of firms in Kosovo that reported benefiting from policy measures (65 percent) was much higher than in comparator countries (e.g., 2 percent in Moldova, and 39 percent in Albania). More than 95 percent of beneficiary firms reported having received wage subsidies. A smaller proportion of larger firms received wage subsidies than other firm classes, but large firms benefited more from eased financial conditions (e.g., deferral of payments due on loans, rent, mortgages, and utilities). Going forward increasing awareness and clarifying eligibility requirements are likely to increase coverage of future public assistance, especially for MSMEs. Small firms reported complex procedures as the main barrier to benefiting from policy measures. Firms in shutdown sectors and companies in majority municipalities received more public assistance. Of firms in shutdown sectors, 70 percent received public assistance, compared to 63 percent in non-shutdown sectors. While 65 percent of firms located in majority-municipalities benefited from policy measures, compared to only 50 percent of those in non-majority regions. Firm Productivity and Access to Credit 49 Conclusions and Tentative Policy Implications Kosovo’s economy is dominated by low-productivity micro and small enterprises with low survival rates. A striking feature is the lack of dynamism in creating new firms: for the last decade the net rate of firm creation has been close to zero and the country has far fewer firms per million inhabitants (firm density) than the EU and aspirational peer averages as well as what might be expected given Kosovo’s income per capita. Kosovar firms are smaller than EU and aspirational peers, especially medium and large enterprises. Moreover, Kosovo is also not well integrated into global markets through FDI and exports. Firms also seem to find it difficult to expand beyond small and medium size, and survival rates are far lower than in the EU and aspirational economies. Kosovo has a markedly lower proportion of exporters and firms with foreign shareholders than in the EU, aspirational peers, and even other Western Balkans economies. The importance of FDI and international trade for fostering economic growth and employment creation is clear from the fact that importers and foreign-controlled enterprises on average have more formal employees, are more productive and capital intensive, and pay higher wages than companies oriented to the domestic market. Furthermore, in Kosovo productivity trails that of aspirational peers and the EU. For instance, labor productivity is only one-third of the EU average. In the decade up to the COVID-19 pandemic, average annual growth rates of labor productivity were negative, and annual growth of TFPR was relatively low. Kosovo’s substandard productivity is explained by lower firm productivity and the small and largely negative contribution of firm dynamics. Confirming the economy’s lack of dynamism, exit of less productive firms contributed only marginally to aggregate TFPR growth, and entry of new firms had a negative contribution. Finally, the large dispersion in productivity between “laggard” and “frontier” firms suggests that there are distortions in product and input markets, especially in services. Despite its dreary productivity performance, between 2015 and 2018 Kosovo’s economy speeded up the creation of jobs, driven by higher job registrations among young SMEs. Higher firm productivity is associated with higher job creation. In 2015–18, compared to 2011–14 annual net job creation practically doubled, driven by rising net job creation by incumbent firms, especially young SMEs. Formal employment growth was also driven by higher worker registration rates: firms reporting having employees went from about 50 percent in 2010 to about 70 percent in 2018. Highlighting the centrality of productivity for fostering formal employment growth, higher firm net job creation was associated with the most productive firms, exporters, and younger and larger firms with access to credit. Despite its crucial role in fostering firm growth and productivity and job creation, access to formal finance is limited to only a third of the companies and mostly to short-term loans. Micro and small enterprises, startups, domestic-market-oriented firms, companies in less concentrated economic sectors, and enterprises in non-majority municipalities are hindered by the lack of formal finance. About 70 percent of loans have a term of no more than three years. To foster firm productivity, Kosovo needs a multidimensional policy strategy. Based on the main findings of this and other background notes prepared for the Kosovo CEM, Table 8 organizes this note’s main policy recommendations along the three main sources of firm TFP growth: (1) growth of the firm (“within”); (2) market reallocation (“between” firm productivity); and (3) dynamics related to firm entry and exit. 52 Conclusions and Tentative Policy Implications Table 8.  Main Drivers of Lower TFPR Growth and Recommended Policy Changes Not Enough “Within” Firm Limited Market Reallocation Lackluster Firm Dynamics TFPR Sources Productivity (“Between” Firms) (Entry and Exit) •• Low private R&D and capital •• Deficient railway and road •• High barriers to entry in investment infrastructure some services sectors (e.g., •• Inadequate firm management •• Informality telecommunications and practices and room for •• Cumbersome and costly electricity) improvement in quality procedures for MSMEs to access •• Cumbersome asset liquidation infrastructure and standards the electricity grid procedures and inadequate •• Weak numeracy and literacy skills •• Unequal access to financial insolvency laws Drivers of worker’s not matched with incentives and public •• Lack of access to formal young employer demands, especially in procurement for publicly owned firms and SMEs access to formal TVET (Technical and Vocational enterprises (POEs) finance Educational Training) •• Lack of commercial courts and •• Poor integration with GVCS via poorly trained legal staff exports and FDI •• Inefficient, costly, and opaque •• Limited access to finance for inspection procedures young firms •• Devise a strategy to attract •• Prevent allocation of •• Facilitate business registration by: FDI, strengthen the investment discriminatory financial (a) introducing the silent consent agency, and create an investor advantages to POEs. principle; and (b) developing an grievance mechanism •• Ensure equal treatment of effective e-signature system to •• Foster reforms in TVET and private firms and POEs in public allow full online registration. general education (e.g., improve procurement processes. •• Streamline asset liquidation teacher wages and training, and •• Enhance the quality and control procedures. education infrastructure, and of the regulatory process (e.g., •• Foster the development of match curricula with firm needs) draft a clear regulation on the commercial courts and provide •• Strengthen the National interaction between regulators appropriate training to legal staff) Innovation System (NIS) and and interest groups; and •• Expand the role of microfinance Proposed Policy provide management training for strengthen the Regulatory Impact and foster digital-enabled Changes MSMEs. Assessment process). business •• Expand the role of microfinance •• Streamline procedures and •• Promote digitalization and (e.g., payment services, reduce fees for MSMEs to access technology adoption, especially remittances, access to equity the electricity grid, draft and in MSMEs and female-run capital and e-services) and use implement a new and greener companies of the Kosovo Credit Guarantee energy strategy •• Incentivize piloting programs Fund. to improve diffusion of good •• Foster the development of management practices among Venture Capital Funds targeted MSMEs. to startups and dynamic young firms. Sources: Cusolito and Maloney (2019) and other background notes prepared for the forthcoming Kosovo CEM. See the other background notes for more detailed actions plans. Table 8 summarizes some the drivers behind the three sources of firm TFPR growth identified in this note and the other Kosovo CEM background papers. It also outlines some policy actions to address the main obstacles to firm productivity in Kosovo. (The background papers provide more detailed actions plans.) Conclusions and Tentative Policy Implications 53 This multidimensional pro-productivity strategy should have three main pillars: • First, the low productivity of Kosovar firms relative to the EU, aspirational economies, and Western Balkans peers suggest the urgent need to (1) improve firm capabilities related to human capital skills (e.g., matching TVET curricula to firm demands, strengthen the general education system to better build basic numeracy and literacy skills), and emphasize management and organizational practices (e.g., provide training and technical assistance to MSMEs and strengthen the quality of infrastructure); (2) encourage adoption of technology and innovation and investment in R&D (e.g., by strengthening the National Innovation System [NIS]27); (3) facilitate microenterprise access to finance (e.g., among other measures, expand the role of microfinance institutions and SME guarantee funds); and (4) adopt strategies to attract FDI and promote exports (e.g., build up the investment agency ) The country also needs to foster digitally enabled businesses: only 9 percent of firms engage in digital-intensive activities, and the digital response of Kosovar firms, especially MSMEs, to the COVID-19 shock trailed other Western Balkans countries. • Second, the declining contribution to TPFR growth of reallocation of labor to the most productive firms reinforces the need to remove barriers that prevent movement of capital, labor, and other inputs and the barriers in product markets related, for instance, to informality (e.g., the urgency of streamlining registration, inspection, and tax procedures); competition policy (allowing private firms easier access to financial incentives and public procurement and improving the quality and control of regulation) unreliable and costly energy supply (e.g., streamlining access for MSMEs to the Kosovo Energy Distribution and Supply Company electricity grid); and inadequate transport infrastructure (e.g., promote investment in some national and regional roads and upgrade the North-South railway corridor). Kosovo also needs to improve the gender agenda: only 9 percent of firms are female run and firms with a female decisionmaker tend to be smaller and less productive, especially in manufacturing and wholesale and retail. • Third, the negligible contribution of new firms to TFPR growth illustrates how important it is to foster entry of highly productive and fast-growing enterprises, just as the negative contribution of firm exits to productivity growth could be indicating the presence of high barriers to exit. The barriers already identified (see Table 8) to entry of more productive enterprises, especially access to formal finance, and the barriers to exit of nonviable firms, such as simplified and accelerated liquidation procedures, clearly need to be dismantled. More targeted policies, improved communication, and streamlined application procedures are necessary to support MSMEs dealing with the economic effects of the COVID-19 economic crisis. Recognizing the limited fiscal resources, policies should be targeted especially providing support to displaced workers and minimizing layoffs by firms in financial stress and confronted by heightened demand or supply shocks. Targeted policies, such as the Kosovo Credit Guarantee Fund, can be influential in addressing MSME liquidity constraints. Improved communication of public support measures and simplified application procedures to enhance uptake and reduce the regulatory administrative burden (especially inspections) not only to encourage local companies but also to attract FDI are key. 27 See Cirera and Maloney 2019. The NIS incorporates local and international R&D suppliers (universities and research institutions, among others), government agencies related to R&D and innovation, and private firms investing in R&D and adopting product, process, and organization innovations. 54 Conclusions and Tentative Policy Implications References American Chamber of Commerce in Kosovo. 2018. Free Movement of Goods and Conformity Assessment. Pristina. Bloom, Nicholas, Aprajit Mahajan, David McKenzie, and John Roberts. 2010. “Why Do Firms in Developing Countries Have Low Productivity?” American Economic Review 100 (2): 619–23. Brown, J. David, Gustavo A. Crespi, Leonardo Iacovone, and Luca Marcolin. 2018. “Decomposing Firm-level Productivity Growth and Assessing its Determinants: Evidence from the Americas.” Journal of Technology Transfer 43(6): 1571–1606 Calvino, F., et al. (2018), “A taxonomy of digital intensive sectors”, OECD Science, Technology and Industry Working Papers, No. 2018/14, OECD Publishing, Paris, https://doi.org/10.1787/f404736a-en. Cirera, Xavier, and William F. Maloney. 2017. “The Innovation Paradox: Developing-Country Capabilities and the Unrealized Promise of Technological Catch-Up.” Overview booklet. World Bank, Washington, DC. Cojocaru, A. 2016 Jobs Diagnostic: Kosovo. World Bank. Washington DC. Criscuolo, C., P. Gal and C. Menon (2014), “The Dynamics of Employment Growth: New Evidence from 18 Countries”, OECD Science, Technology and Industry Policy Papers, No. 14, OECD Publishing, Paris, https://doi.org/10.1787/5jz417hj6hg6-en Cusolito, A. P., and W. F. Maloney. 2018. Productivity Revisited. Shifting Paradigms in Analysis and Policy. Washington, D.C.: World Bank Group. Foster, Lucia, John Haltiwanger, and C. J. Krizan. 2001. “Aggregate Productivity Growth: Lessons from Microeconomic Evidence.” In New Developments in Productivity Analysis, edited by. Charles R. Hulten, Edwin R. Dean, and Michael J. Harper, 303–63. Chicago and London: University of Chicago Press. Haltiwanger, John. 2007. “Measuring and Analysing Aggregate Fluctuations: The Importance of Building from Microeconomic Evidence.” Federal Reserve Bank of St. Louis Review 79 (3): 55–78. Hsieh, Chang-Tai, and Peter J. Klenow. 20214. “The Life Cycle of Plants in India and Mexico.” The Quarterly Journal of Economics 129 (3): 1035–84, https://doi.org/10.1093/qje/qju014 Levinsohn, James, and Amil Petrin. 2003. “Estimating Production Functions Using Inputs to Control for Unobservables.” Review of Economic Studies 70 (2): 317–41. https://EconPapers.repec.org/RePEc:oup:restud:v:70:y:2003:i:2:p:317-341 Linarello, Andrea. Andrea Petrella, and Enrico Sette. 2019. “Allocative Efficiency and Finance. Bank of Italy Occasional Paper No. 487. https://ssrn.com/abstract=3432990 or http://dx.doi.org/10.2139/ssrn.3432990 Melitz, M. J., and S. Polanec. 2015. “Dynamic Olley-Pakes Productivity Decomposition with Entry and Exit.” The Rand Journal of Economics 46(2): 362–75. OECD. 2009. Measuring Capital, OECD Manual 2009: 2nd ed. Paris: OECD Publishing. https://doi.org/10.1787/9789264068476-en Syverson, Chad. 2011. “What Determines Productivity?” Journal of Economic Literature 49:2: 326–65. World Bank. 2018. Country Economic Memorandum: Serbia’s New Growth Agenda. World Bank, Washington, DC. ———. Forthcoming. “Chapter 4: Human Capital Through Education System Improvement.” Kosovo: Country Economic Memorandum. World Bank, Washington, DC 55 Annex 1. Technical Appendix A. Firm Dynamics Data sources: Tax Registry from Kosovo Statistics Authority and Kosovo Productivity Survey Period covered: 2010-2018 Unit-level: Establishment Methodological definitions and benchmarking: Definitions follow the Eurostat Manual on Business Demography Statistics (OECD 2007), which ensures international comparability. We also draw on Eurostat and the OECD, which provide comprehensive and relevant standardized information on business and employment dynamics: Business Demography Statistics and, the Annual Enterprise Statistics (Eurostat), and the Structural and Demographic Business Statistics (SDBS, OECD). Size class: Micro (0–9 employees), small (10–49), medium (50–249), and large (250 and more) Age class: Entrants (0 years), start-ups (1–2 years), young (1–5 years), and mature and old (6+) Sector: B-N (excluding K64.2; Activities of Holding Companies in Financial and Insurance Activities) according to NACE Rev. 2m as followed in Eurostat and OECD manuals unless otherwise stated. Employed persons: Eurostat defines an employed person as a person aged 15 and over who performs part-time or full-time work for pay, profit, or family gain. Active unit: Reported positive turnover in the reference year. Entry: A unit enters the market in t if it is active in the period of reference but was not active in t-1 and t-2. Exit: A unit exits the market in t if it is active in t and is not active in t+1 and t+2. Incumbent: Unit is active in t if it reports activity in t-1, t and t+1. Table A1.1. Methodological Definitions of Firm Status in Market Status in t t-2 t-1 t t+1 t+2 Entry No No Yes Activation No Yes Exit Yes No No Inactivation Yes No Incumbent Yes Yes Yes Entry rate in t: Sum of all entries in t divided by total active units in the period of reference. Exit rate in t: Sum of all exits in t divided by total active units in the period of reference. Turnover rate in t: Sum of entry and exit rates in t. Incumbent share in t: Difference between gross job creation in t and gross job destruction in t. Survival unit in t+s: Share of units that entered the market in t and survives s periods ahead, percent. Survival rate in t+s: Share of total units that entered the market in t that survived in t+s, percent. Entry size in t: Number of employed persons of entrant units. Post-entry growth in t+s: Total employed persons in an active unit in t+s relative to total employed persons of entrants in t that survived to t+s. Age: Each firm is given an entry cohort year if the year of entry is known. Age is then calculated as the difference between the year of reference and the entry year. For firms in the database before 2012, their entry year unknown, that were active throughout the periods of analysis, we assigned a ‘symbolic’ age in 2017 and 2018 in order to carry out the age analysis. Sector and main sector: Unit activities are classified according to NACE Rev. 2 at the four-digit level. Because some companies started operations in one sector but then switched to another activity, we generated the ‘main sector’ variable, which reports the sector in which the firm spends most of its life cycle. Gross job creation in t: Sum of all positive unit-level variations of employed persons in active units in t relative to t-1. 58 Technical Appendix Gross job destruction in t: Sum of all negative unit-level variations of employed persons in active units in t relative to t-1. Net job change in t: The difference between gross job creation and gross job destruction in t. Total employed persons (total employment): Sum of employed persons of active units. Gross job creation rate in t: Gross job creation in t divided by total employed persons. Gross job destruction rate in t: Gross job destruction in t divided by total employed persons, B. TFPR Estimation Following Levinsohn and Petrin (2003), we estimate TFPR using semi-parametric methods. We use this approach as we do not have access to data on firm initial capital stock and only a small number of firms in the KAS dataset report investment. Hence, we use data on costs of production (intermediate consumption) reported in the personal income tax and corporate income tax forms in the KAS dataset. To estimate TFPR, we take the firm’s total wage bill as the free-standing variable and our estimate of the capital stock as the state variable. Production costs enter into the production function as a proxy for capital stock. Because data on worker skills and number of hours worked are unavailable, we use the total wage bill to control for heterogeneity in labor productivity and labor intensity (hours worked per year). The total wage bill is the result of multiplying workers’ hourly wages by the number of hours worked. If production factors are paid according to their marginal productivity, the total wage bill should account for labor productivity differences across firms (Cusolito and Maloney 2018). To estimate the capital stock, we follow OECD methodology (2009) by using data on investment in durable goods from the KAS dataset, combined with self-reported information on firm assets provided by the KPS. We compute firm i average investment over time , and estimate initial capital stock by dividing the average investment by the depreciation rate . Hence, the initial capital stock of firm i is expressed as follows: . As does OECD (2009), we assume that the depreciation rate is constant over time and across industries. Given the relatively small size of our sample, we estimate TFPR for five highly aggregated sectors: manufacturing (firms classified as letter C in NACE Rev. 2); construction (F and L); other services (H transport and storage, I: accommodation and food services, J: information and communications, M: professional, scientific, and technical services, and N: administrative and support service activities); wholesale trade (division 46: wholesale trade, except of motor vehicles and motorcycles, and class 4531: wholesale trade of motor vehicle parts and accessories); and finally retail trade (division 47: retail trade, except of motor vehicles and motorcycles, plus activity classes 4511: sale of cars and light motor vehicles, 4519: sale of other motor vehicles, 4520: maintenance and repair of motor vehicles, 4531: retail trade of motor vehicle parts and accessories, and 4540: sale, maintenance and repair of motorcycles and related parts and accessories. Technical Appendix 59 Annex 2. Additional Tables and Figures Table A2.1. Summary Statistics, Kosovo Statistics Authority Firms Year Total (‘000) Micro (%) Exporters (%) Importers (%) 2010 27.918 95.268 - - 2011 30.299 94.789 1.8 18.9 2012 30.866 94.450 1.8 20.5 2013 28.123 93.372 2.3 23.8 2014 28.556 93.059 2.4 24.1 2015 29.041 92.542 2.9 26.3 2016 28.709 91.821 3.4 17.3 2017 30.25 91.537 3.5 17.3 2018 29.441 90.282 4 18.3 Average 29.245 93.011 2.7 20.7 2010-2018 Firm Turnover Wage per worker Sales per worker Employed persons (‘000 €) (‘000 €) (‘000 €) Year Total (‘000) Mean Median Total Mean Median Mean Median Mean Median 2010 100.035 3.583 0 5064265.99 181.398 9.059 4.585 3.521 47.613 16.193 2011 118.313 3.905 1 7967761.229 262.971 10.719 4.763 3.731 61.871 33.341 2012 124.749 4.042 1 8256137.147 267.483 9.818 4.908 3.804 62.089 35.278 2013 130.21 4.630 1 8372110.563 297.696 12.101 4.859 3.809 60.643 35.498 2014 133.268 4.667 1 8028598.981 281.153 12.758 4.738 3.663 58.236 33.589 2015 145.069 4.995 1 8496580.504 292.572 13.642 4.616 3.608 57.606 30.902 2016 155.057 5.401 1 9001941.122 313.558 15.318 4.709 3.757 57.073 31.038 2017 168.821 5.581 1 10042692.799 331.99 16.358 4.854 3.949 57.521 29.667 2018 180.648 6.136 2 10564413.334 358.833 16.989 5.015 3.977 57.526 28.123 Average 139.574 4.771 1 8421611.297 287.517 12.974 4.783 3.758 57.798 30.403 2010-2018 Source: KTR, World Bank staff calculations. Notes: Active units are those with positive turnover in the relevant period. Activities included: B-N (excluding K64.2) according to NACE Rev. 2. Wages and sales per worker are weighted by employment. Monetary values in 2018 constant prices, using Kosovo’s CPI deflator. 62 Additional Tables and Figures Table A2.2. Summary Statistics. Kosovo Productivity Survey. 2017 Firms Management Workers Sector Surveyed (units) Share (%) Size (workers) Women (%) W/ tertiary education (%) Mining and quarrying 25 0.452 9.45 0 18.310 Manufacturing 682 13.960 6.397 8.7 25.114 Electricity, gas, steam and air-conditioning supply 12 0.181 46.077 0 35.733 Water supply, sewerage, waste mgmt. and 33 0.395 30.001 1.9 27.747 remediation act. Construction 337 8.139 8.095 3.8 28.253 Wholesale and retail trade 1217 50.220 4.248 10.3 31.766 Transportation and storage 154 3.996 7.873 5.6 30.418 Accommodation and food services 366 11.046 6.306 7.7 39.498 Information and communication 149 2.777 11.762 6.9 61.368 Real estate activities 21 0.330 4.198 9.3 36.755 Professional, scientific and technical act. 205 6.183 3.36 10.4 71.657 Administrative and support services 113 2.319 7.094 14.7 42.716 All firms 3314 100 5.654 9 32.925 Foreign Foreign trade Sales, value added and assets shareholding Sector Unpaid Exporter Importer Share Sales per VA per worker Average assets (%) (%) (%) (%) worker (‘000 €) stock (‘000 €) (‘000 €) Mining and quarrying 0.714 0.6 28.3 0 14.961 4.987 2317.909 Manufacturing 9.231 4.4 36.6 5.29 27.415 21.027 493.623 Electricity, gas, steam 0 1.8 59.2 18.14 31.886 26.562 3555.544 and air-conditioning supply Water supply, 1.813 2.4 47.8 3.74 12.973 5.388 2942.46 sewerage, waste mgmt. and remediation act. Construction 3.262 1.4 31.4 10.17 26.576 19.364 307.793 Wholesale and retail 8.650 1 36.4 6.78 25.294 17.705 195.244 trade Transportation and 5.314 2.4 31.3 9.6 15.281 9.639 155.401 storage Accommodation and 5.917 0 25.9 3.86 8.378 5.692 88.864 food services Information and 1.787 2.7 44.7 4.61 28.261 17.696 915.221 communication Real estate activities 6.977 7.3 11.3 0 28.304 17.164 1244.783 Professional, scientific 8.504 0 29.6 2.36 18.308 14.589 61.055 and technical act. Administrative and 3.605 0 34.4 2.17 23.84 19.573 353.758 support services All firms 6.67 1.4 34.4 6.15 22.974 16.526 277.497 Source: KPS, World Bank staff calculations. Notes: Surveyed units are the effective number of firms that responded the survey. The remaining variables are weighted using sample weights. Additional Tables and Figures 63 Table A2.3. Summary Statistics: TFPR Sample, Average 2012–18 Sector Manufacturing Construction Wholesale and retail Services All firms trade Firm sin sample 708 622 1,719 676 3,725 Employment 6.6 7.8 4.9 5.3 5.5 Value added 109.2 138.2 90.2 81.5 95.3 Turnover 391.8 526.3 655.3 307.2 513.2 Materials 282.6 388.1 565.1 225.7 417.9 Wage bill 26.1 31.7 20.8 22.5 23.0 Capital stock 715.6 931.4 577.9 358.2 576.3 Investment 84.1 111.4 70.2 42.6 69.4 TFP (log) 2.4 1.3 1.7 1.1 1.6 Value added per 16.2 19.4 17.8 16.9 17.5 worker Turnover per worker 59.3 83.8 128.4 81.9 99.9 Source: KTR, World Bank staff calculations. Notes: TFPR is weighted by employment while Employment, VA, Turnover, Materials, Wage bill, Capital stock and Investment are weighted using firm weights. 64 Additional Tables and Figures Table A2.4. Net Job Creation and Labor Productivity Dependent Variable: Net Job OLS FE FE Creation (1) (2) (3) Log Turnover 0.399*** 0.684*** 0.684*** (0.066) (0.124) (0.124) Log Capital stock -0.138** -3.127 -3.127 (0.059) (2.516) (2.516) Firm age -0.151*** -0.181** -0.181** (0.018) (0.075) (0.075) Firm starts to report employees 2.518*** 2.791*** 2.791*** (0.194) (0.265) (0.265) Exporter status -0.451 -1.078 -1.078 (0.566) (0.896) (0.896) Access to credit 0.158 0.039 0.039 (0.219) (0.238) (0.238) Municipality majority -0.877 -0.086 -0.086 (1.178) (6.410) (6.410) HH Index (four-digit) -1.596 76.464*** 76.464*** (8.660) (22.549) (22.549) Small (10-49) 1.009*** 1.796*** 1.796*** (0.320) (0.367) (0.367) Medium (50-249) 7.657*** 13.644*** 13.644*** (1.317) (2.256) (2.256) Large (250+) 44.678*** 66.635*** 66.635*** (12.921) (20.641) (20.641) Constant -5.549 30.027 30.027 (9.312) (31.415) (31.415) Industry effects (four-digit) Yes Yes Yes Geographic effects (municipality) Yes Yes Yes Year effects Yes Yes Yes Firm fixed effects No Yes Yes R-squared 0.04 0.02 0.02 Number of firms 14,534 14,534 14,534 Observations 66,291 66,291 66,291 Source: KTR, World Bank staff calculations. Notes: the table presents the results of regressing firm net job creation (log difference between number of employees in t0 and t1) on a set on covariates, controlling by industry (4 digits), geographic (municipality) and time (year) fixed effects and using the entire dataset of firms provided by KRT. Entrants and exiters are excluded from the panel. Micro (0-9 employees) firms are the omitted variable. Columns (2) and (3) estimate the model using a fixed effects estimator. Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. Additional Tables and Figures 65 Table A2.5. TFPR and Employment, Revenue, and Value Added per Worker (in Logs Relative to Industry-year Average) Dependent variable: TFPR (log) OLS OLS OLS (1) (2) (3) Log Employment 0.064***     (0.002)     Log Turnover per worker   0.096***     (0.003)   Log Value added per worker     0.064***     (0.003) Constant 2.223*** 1.383*** 1.779*** (0.017) (0.033) (0.030) Industry effects (macrosector) Yes Yes Yes Year effects Yes Yes Yes Geographic effects (municipality) Yes Yes Yes R-squared 0.92 0.93 0.91 Observations 6,774 6,774 6,774 Source: KTR, World Bank staff calculations. Notes: TFPR, employment, revenue and value added per worker (in logs). Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. Table A2.6. Exporting and Importing Premiums, Kosovo   Log Employment Log Turnover per Log Turnover Log Capital per Log Wage per   worker worker worker (1) (2) (3) (4) (5) Exporter only 0.115*** 0.558*** 0.838*** -0.085*** 0.043** (0.024) (0.037) (0.040) (0.024) (0.021) Importer only  0.196*** 0.423*** 0.764*** -0.180*** 0.110*** (0.005) (0.009) (0.011) (0.007) (0.004) Importer and exporter 0.353*** 0.621*** 1.155*** -0.320*** 0.141*** (0.014) (0.021) (0.023) (0.015) (0.009) Constant 0.649*** 9.588*** 9.638*** 11.288*** 7.342*** (0.125) (0.554) (0.473) (0.158) (0.207) Industry effects (four-digit) Yes Yes Yes Yes Yes Geographic effects (municipality) Yes Yes Yes Yes Yes Year effects Yes Yes Yes Yes Yes Firm fixed effects Yes Yes Yes Yes Yes R-squared 0.08 0.05 0.08 0.13 0.04 Observations 161,949 161,949 262,231 72,569 160,353 Source: KTR, World Bank staff calculations. Notes: Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. In each column, we regress the dependent variable on foreign trade dummies, which relate to the type of link that the firm has with foreign markets. “Exporter only” equals 1 if the firm exports but does not import; “Importer only” follows the same criteria for importing firms. Finally, the “importer and exporter” dummy values 1 if the firm both imports and exports in the year of reference. Each variable is constructed based on the VAT forms. Exporting/importing firms are those firms that report a positive export/import value in the year of reference while importers. Furthermore, we control for industry effects (four-digit level), geographic effects (municipality), year and firm effects. 66 Additional Tables and Figures Table A2.7. Foreign-shareholding Premiums in Kosovo Dependent variables Per worker (col) / Employment Turnover Turnover Value added Capital Wages Independent variables (row) (1) (2) (3) (4) (5) (6) Foreign shareholding -0.007 -0.265** -0.207*** -0.103 0.720*** 0.007 (% assets: 0.1%-24.9%) (0.080) (0.116) (0.063) (0.096) (0.070) (0.028) Foreign shareholding -0.128** 0.144 -0.099 -0.293*** 0.814*** 0.082*** (% of assets: 25%- 49.9%) (0.060) (0.111) (0.073) (0.108) (0.075) (0.030) Foreign shareholding 0.978*** 1.625*** 0.537*** 0.151** 0.860*** 0.425*** (% of assets: 50%-100%) (0.077) (0.134) (0.076) (0.074) (0.152) (0.041) Constant 4.501*** 14.670*** 10.242*** 9.261*** 9.383*** 8.336*** (0.109) (0.172) (0.106) (0.158) (0.164) (0.058) Industry effects (two- Yes Yes Yes Yes Yes Yes digit) Geographic effects Yes Yes Yes Yes Yes Yes (municipality) Year effects Yes Yes Yes Yes Yes Yes R-squared 0.20 0.20 0.27 0.17 0.25 0.23 Observations 17,285 20,381 17,285 9,976 16,644 17,227 Source: KTR, World Bank staff calculations. Notes: Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. In each column, we regress the detailed dependent variable on a set of dummies that breaks down the share of assets owned by foreign shareholders. Table A2.8. Gender Premiums in Kosovo Per worker Employment Turnover Turnover Value added Capital Wages Sector (1) (2) (3) (4) (5) (6) All sectors -0.205*** -0.053 -0.053 -0.176** -0.076 -0.017 Manufacturing -0.643*** -0.081 -0.081 -0.445*** -0.221** 0.003 Wholesale and retail -0.224*** -0.029 -0.029 -0.013 -0.296*** 0.038 trade Professional and -0.122 -0.028 -0.028 0.508*** 0.426** -0.141*** administrative services Other services -0.109 0.005 0.005 0.194 -0.133 0.030 Industry effecsts (two- Yes Yes Yes Yes Yes Yes digit) Geographic effects Yes Yes Yes Yes Yes Yes (municipality) Source: KTR, World Bank staff calculations. Note: Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. In each column, we regress the detailed variable on a female-run company dummy variable, which equals 1 if the establishment is run by a female and 0 otherwise. Each specification is estimated by OLS and includes geographic (at the municipality level) and sector effects (two-digit level). Additional Tables and Figures 67 Table A2.9. Access to Finance: Descriptive Statistics, Average, 2015–18 Borrowers Share of disbursed Share of disbursed Approved amount amount by borrower (%) amount by lender (%) (‘000 €) Total Legal Physic Non- Banks Insurance MFI Mean Median Year residents company 2015 11,137 98.3 1.6 0.0 91 7 1 0 0 2016 10,962 98.4 1.6 0.1 88 10 2 0 0 2017 10,516 98.2 1.7 0.1 90 8 2 0 0 2018 10,569 98.0 2.0 0.0 86 13 1 0 0 Disbursed amount Nominal Duration (months) Type of credit (‘000 €) interest rate (pp) (% of total disbursed amount) Year Mean Median Mean Median Mean Median Loan Overdraft Guarantee Other 2015 0 0 9.7 8.0 42 24 51 26 17 5 2016 0 0 9.1 7.6 45 13 51 24 17 8 2017 0 0 7.5 7.4 43 21 56 20 14 9 2018 0 0 8.1 7.6 43 24 52 20 17 11 Source: KTR, World Bank staff calculations. Note: TFPR is weighted by employment while Employment, VA, Turnover, Materials, Wage bill, Capital stock and Investment are weighted using firm weights. 68 Additional Tables and Figures Table A2.10. Drivers of Access to Credit in Kosovo (1) TFP (log) change 5.591*** (1.235) TFP (log) change (t-1) 0.167*** (0.043) Serbian-majority municipality -0.550*** (0.050) Exporting firm -0.030** (0.013) Age 0.003** (0.001) Small (10-49) 0.174*** (0.010) Medium (50-249) 0.244*** (0.020) Large (250+) 0.326*** (0.043) Market concentration (HH index) 0.257*** (0.065) Constant 0.698*** (0.052) Year effects Yes Industry effects (four-digit) Yes R-squared 0.35 Number of firms 1,513 Observations 3,536 Source: KTR and KCR, World Bank staff calculations. Note: Robust standard errors. (*), (**) and (***) indicates significance level at (10), (5) and (1) per cent level. Dependent variable is a dummy variable that is 1 if the firms was approved a positive loan amount during the period of analysis and 0 otherwise. Additional Tables and Figures 69 WORLD BANK GROUP Raising Firm Productivity Kosovo Country November 2021 Economic Memorandum