Policy Research Working Paper 10276 What Makes an Investment Promotion Agency Effective? Findings from a Structural Gravity Model Victor Steenbergen Finance, Competitiveness and Innovation Global Practice January 2023 Policy Research Working Paper 10276 Abstract Although many countries have established investment pro- to define two groups: high-performing investment pro- motion agencies over the past two decades, there is little motion agencies (those with positive, significant effects in evidence on what characteristics make them effective in attracting foreign direct investment) and other investment attracting foreign direct investment into their home coun- promotion agencies (those with insignificant or negative try. To provide new insight into this question, this paper significant effects). Using t-tests, the study considers which brings together sectoral foreign direct investment data with investment promotion agency characteristics significantly survey data on investment promotion agency characteristics. differ between the two groups. The findings suggest that Using a structural gravity model framework, it explores the effective investment promotion agencies are more likely to effect of investment promotion agencies’ sectoral targeting be private or semi-private agencies. Their mandate tends on inward foreign direct investment stocks over 2013 to to be focused narrowly on foreign investment and exclude 2018, across a sample of 36 middle- and high-income coun- responsibilities for domestic investment promotion. Such tries. The study finds that investment promotion agency investment promotion agencies are more likely to have a sectoral targeting provides a significant positive effect on board of directors, and their staff tends to be better com- the sector’s foreign direct investment stock in that coun- pensated. Finally, high-performing investment promotion try. Yet, a gravity model with country-interaction effects agencies tend to provide more investor services, partly by suggests that not all countries are equally effective at pro- engaging smart, sectoral analytics and adopting systems for moting investment. The results from the model are used identifying investor complaints or disputes. This paper is a product of the Finance, Competitiveness and Innovation Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The author may be contacted at vsteenbergen@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team What Makes an Investment Promotion Agency Effective? Findings from a Structural Gravity Model Victor Steenbergen 1 Keywords: foreign direct investment, investment promotion, gravity model JEL-Codes: F23, F13, F14 1 This note is an output of the Investment Climate Unit of the World Bank. It also features as a background paper for the project “Impact Assessment for Invest in Pomerania, Poland (2011-2020)”. Special thanks to Maximilian Eltgen for his support, and for Armando Heilbron, Jose P. Vasquez and Devaki Ghose for their review and thoughtful comments and suggestions. Asya Akhlaque, practice manager for the Investment Climate Unit, provided guidance and supervision throughout. 1. Introduction Investment promotion agencies (IPAs) stand out as one of the most favored policies by countries to attract foreign direct investment (FDI). Between 2002 and 2018, the number of national and subnational IPAs registered in the World Association of Investment Promotion Agencies (WAIPA) grew from 112 to 170 (Heilbron and Kronfol, 2020). These IPAs also receive considerable public funding. A survey of IPAs in Latin America and the OECD found the median IPA in these countries manages a budget of US$7 million, has 100 employees, and performs roughly 40 different promotional activities. Resources available to IPAs in certain countries can exceed US$200 million and 1,000 employees (Volpe Martincus and Sztajerowska, 2019). Despite the proliferation of IPAs in the last two decades, their impact on stimulating economic development has varied considerably across countries. On average, IPA activity is often associated with greater FDI inflows in targeted sectors (see for example Charlton and Davis, 2007; Harding and Javorcik, 2011). High-profile cases have further illustrated the influence that investment promotion can have in attracting FDI and shaping economic development, as shown in the cases of Malaysia, Costa Rica and Morocco (Freund and Moran, 2017). Yet, in many other developing countries, IPAs have failed to replicate these examples and have made little impact in attracting FDI (Heilbron and Kronfol, 2020). To better understand what drives these differences in IPA performance, this paper brings together new data on sectoral FDI and IPA characteristics. It uses this to construct a structural gravity model framework with country-interaction effects that evaluates the effectiveness of each country IPA’s sectoral targeting on their FDI stock in that sector. Results from this model are then used to define two groups: high- performing IPAs (those with positive, significant effects in attracting FDI) and other IPAs (those with insignificant or negative, significant effects). It then considers which IPA characteristics significantly differ between the two groups, and in turn, reflects on what type of institutional setting, strategic focus, organizational framework and service provision is most conducive for IPAs to attract FDI. Outline Section 2 of this report reviews the literature. It first considers the broader literature on the effect of IPAs in attracting foreign investment. It then also considers the potential that gravity models hold for the analysis of FDI dynamics. To identify the characteristics that may shape the functioning of IPAs we build on the qualitative literature reviewed by Heilbron and Kronfol (2020) and consider four broad pillars: institutional arrangement; strategic alignment and focus; organizational framework and resourcing; and investor service delivery. Section 3 provides a summary of the data used in this analysis. This describes the process for harmonizing different types of sectoral FDI, and merging this to the WB-WAIPA survey on IPA characteristics. Section 4 summarizes the empirical strategy, which includes the structural gravity model of FDI and the gravity model with country-interaction effects. Section 5 provides the results, and Section 6 concludes. Added value of this paper This paper provides a modest contribution to two distinct bodies of literature. The first relates to economic literature analyzing the impact of IPA to attract FDI. These studies commonly rely on an IPA’s specific sectoral targeting strategy combined with sectoral FDI data. Yet, because sectoral FDI is often not available for most countries, researchers either restrict their analysis to FDI inflows from the United States 2 (e.g. Harding and Javorcik 2011), or rely on newspaper-reported greenfield FDI announcements (e.g. Crescenzi, Di Cataldo and Giua 2021). In both cases, this only accounts for a subset of FDI and can deviate significantly from actual FDI dynamics. In addition, their models commonly do not account for the relative effects that IPA targeting may have on relative competitiveness vis-à-vis alternative host countries. Such omissions can bring about endogeneity concerns and potentially over-estimate the effect of investment promotion on FDI inflows. This challenge was first raised by Anderson and van Wincoop (2003), showing the importance of ‘multilateral resistance’ general equilibrium effects in correctly estimating the response of trade flows to trade costs. In contrast, this paper makes use of a wide set of FDI source countries and makes use of actual FDI stock data (rather than newspaper reported FDI announcements). It also explicitly considers how an IPA’s targeting of a priority sector in one country may affect the relative effectiveness of IPAs in other countries targeting the same sector, by including sectoral IPA targeting within a structural gravity model framework. The paper also contributes to the more qualitative literature around identifying high-performing IPAs and considering which characteristics contribute to their performance. Other such studies have historically relied on IPAs’ self-reported statistics (e.g. Lim, 2018) which can be biased, or have defined IPA performance based on a set of “good practices” exhibited by their favored IPAs, which then runs the risk of circular reasoning (e.g. Whyte, Ortega and Griffin, 2011). Other, more empirical estimates tended to correlate FDI inflows with a narrow indicator of IPAs (e.g. Harding and Javorcik, 2012 show a relationship between an IPA’s information handling and aggregate FDI inflows). Yet, it can be difficult and misleading to reduce the effectiveness of an IPA to a single outcome indicator. This paper avoids some of these limitations by taking a data-driven approach to identifying ‘effective’ IPAs. Namely, it uses a model to identify those countries which see a statistically significant relationship between their IPA’s sectoral targeting and increased FDI stock in that sector. For this, the paper uses an empirical specification with country-pair, host-sector, source-, host-, and sector-year fixed effects. It then compares these to other IPAs across a highly detailed set of characteristics using the 2019 WB-WAIPA survey, to assess the potential dynamics that may shape the effectiveness of IPAs in attracting FDI. While this paper provides a wide range of controls to address endogeneity concerns, the non-randomness of priority sectors may still happen for reasons not accounted for by the comprehensive set of fixed effects. An additional limitation comes from the analysis of characteristics of effective IPAs. Using t-tests to distinguish between high-performing and other IPAs does not allow us to make causal inference about IPA characteristics to drive FDI dynamics. As such, these results should be considered as indicative only. 2. Literature Review 2.1 The effect of IPAs on attracting foreign investment The rationale for investment promotion agencies stems from the existence of information asymmetries and transaction costs in capital markets (Williamson, 1985; Wells and Wint, 2000, Loewendahl, 2001). International investors, who intend to invest in a foreign market, often lack specific information on the potential host country’s available business partners, government regulation, and wider investment climate (OECD, 2015). IPAs are expected to influence the investment decisions of multinational corporations (MNCs) by solving these information/perception gaps about the host economy (OECD, 2018). 3 IPAs may further help to attract foreign investment by reducing the transaction costs of foreign companies seeking to relocate. Through proactive investment facilitation, their activities may help to cut operational and search costs of new investors (for example by accelerating their business licensing or by identifying viable domestic suppliers to collaborate with). Broader advocacy within the government may further help to remove regulatory obstacles or encourage new policy to improve the host country’s investment climate. For example, the IPA agencies may lobby to reform visa regulations or advocate the introduction of specific training programs to match a priority sector’s need for specialist skills (Wells and Wint, 2000). The economic literature has mixed results on the effectiveness of IPAs to attract foreign investment. This literature tends to adopt one of three approaches to assess investment promotion: a binary variable indicating the existence of an IPA in the host or in the home country, the presence of an IPA interacted with a certain organizational characteristic (such as their budget), or by using an IPA’s specific sectoral targeting strategy (Volpe Martincus et al. 2020; Crescenzi, Di Cataldo and Giua, 2021). Head et al. (1999) exemplify the first approach, using the existence of an overseas office in Japan between 1980 and 1992 to explore the effectiveness of US states’ investment promotion in attracting Japanese FDI. The authors found no evidence of any effect of IPAs, and suggested that Japanese investors may have been already well informed about the states, thus limiting the IPAs’ impact through information provision. In contrast, Bobonis and Shatz (2007) analyzed eight different countries and found that state offices did influence their FDI into US states between 1976 and 1996. Similarly, Anderson and Sutherland (2015) find that the presence of Canadian provincial-level IPAs located in China increases the likelihood of Chinese firms locating in that Canadian province. Hayakawa et al. (2014) focused on countries establishing IPAs in Japan and the Republic of Korea and find that the presence of an IPA does have a positive effect on FDI inflows, but only for “politically risky” countries. The second approach considers the existence of certain organizational characteristics to assess the impact of IPAs. Using this approach, Morisset and Andrews-Johnson (2004) find a positive association between the budget and staff size of IPAs and the proportion of foreign investments towards the country where IPAs are based. Lim (2008) considers the age and the number of staff (domestically and overseas) of investment agencies and finds that additional staff are positively correlated with foreign investment attracted. Harding and Javorcik (2012) find that IPAs that handle investors’ inquiries in a more professional manner and have higher-quality websites that attract larger volumes of FDI. The third approach makes use of sector targeting to identify the impact of investment promotion of FDI inflows (see Charlton and Davis, 2007; Harding and Javorcik, 2011). Such studies tend to be more empirically robust by adopting a difference-in-difference strategy that exploits a change in an IPA’s sectoral targeting and tends to control for a range of confounding factors at the source-country, host- country or year or sector level. Such analysis finds that FDI inflows from sectors prioritized by the IPA grow more strongly than in other sectors. For example, Charlton and Davis (2007) adopt propensity score matching in combination with difference-in-differences techniques and find evidence of a positive effect of IPAs on the volume of investments towards specific industries targeted by IPAs in OECD countries. Similarly, Harding and Javorcik (2011) use sectoral FDI from the US over the period 1990-2004 and find that countries targeting these sectors received 155 percent more FDI after being targeted. However, such findings were found to only materialize in developing economies, thereby indicating that IPAs are most effective in countries where red tape and bureaucratic barriers to investors are prevalent. 4 Similar findings come from Crescenzi, Di Cataldo and Giua (2021), who leverage a survey on national and subnational IPAs in Europe to evaluate the impact of IPA sectoral targeting on announcements of greenfield FDI projects, using both a difference-in-difference and synthetic control methods approach. They find that IPAs help to attract FDI even in advanced economies. They further find that sub-national IPAs are more effective than national IPAs and speculate that this is because their proximity to investors helps them to better address their investment concerns. Finally, an innovative new study by Volpe Martincus et al. (2020) expansd the focus from sectoral to firm-level targeting. The authors use firm-level data on both location decision and IPA assistance status to assess the effectiveness of IPA facilitation on attracting foreign firms in Costa Rica and Uruguay over the period 2000-2016. They find evidence that investment promotion strongly increases the probability that an MNC locates in the host country. FDI Gravity Models A separate literature relevant for the analysis of IPA effectiveness relates to gravity models to study FDI flows. These models originate in international trade, and use the metaphor of Newton’s law of universal gravitation to predict that trade flows between countries is a function of two main variables: their size (more economic activity predicts more trade) and their distance (greater physical/cultural distance and trade frictions predict less trade). Gravity models build on solid theoretical foundations that make them particularly appropriate for counterfactual analysis, such as quantifying the effects of trade policy (Yotov et al. 2016). Their predictive power is another reason why these models have grown in popularity. Empirical gravity equations of trade flows consistently deliver a remarkable fit of between 60 and 90 percent with aggregate data as well as with sectoral data for both goods and services (Yotov et al. 2016). New trade gravity models also hold additional insights for FDI analysis. In their seminal paper, Anderson and van Wincoop (2003) demonstrate that traditional gravity equations can overstate the effect of trade frictions on trade flows because they ignore the fact that countries operate in a multilateral world. That is – a change in the trade cost (say, a tariff) between countries A and B also affects the trade flows of countries A and B with all their other trading partners (because it may create or divert trade flows). New or ‘structural’ gravity models distinguish themselves by directly accounting for this multilateral resistance (MR) effect, and are therefore considered more robust and accurate in evaluating policy reforms. The issue of multilateral resistance is also clearly present within foreign investment dynamics. For example, investment climate reforms or IPA promotion efforts in host country A would affect its relative competitiveness compared to other host countries and can shift FDI across such countries. As such, new gravity models also hold theoretical and empirical benefits to study FDI dynamics. The economic literature has recently started to apply gravity models to FDI. In a literature review of FDI determinants, Blonigen (2005) noted that “As with trade flows, a gravity specification actually fits cross- country data on FDI reasonably well”. Kox and Rojas-Romagosa (2020) note that “given the success of the gravity trade model, one wonders why it has not yet become part of the standard empirical toolkit for the analysis of international patterns in bilateral FDI”. Part of the challenge here lies with the absence of tractable models, but a more important constraint has been the availability of bilateral direct investment data. Fortunately, more countries are now reporting detailed bilateral FDI statistics,2 which is leading to the increased popularity of studying FDI in a gravity context. A recent example of this is Kox and Rojas- Romagosa (2020) which uses a structural gravity model of FDI to estimate the potential impact of preferential trade agreements (PTAs) on the bilateral FDI stocks and flows between the countries signing. 2 Examples include the US Bureau of Economic Analysis, the OECD’s BMD4 database and EUROSTAT’s BPM6 data sets. 5 Another example comes from Echandi, Maliszewska and Steenbergen (2022) which consider the effect of the African Continental Free Trade Agreement on FDI stocks. To our knowledge, no analysis has been done using gravity models to explore the effect of IPAs on FDI inflows. 2.2. Characteristics of effective IPAs In addition to evaluating if IPAs’ sectoral targeting can influence FDI inflows, this study is also interested in identifying the type of characteristics that may explain the relative effectiveness of IPAs. There is significant evidence that IPAs differ in their overall abilities, and that these distinctions may help explain why only a select group of developing countries have been successful in attracting FDI to transform their economy (Whyte, Ortega and Griffin, 2011; Harding and Javorcik, 2012; Freund and Moran, 2017). Yet, there is little economic literature on characteristics of effective IPAs. Where analysis does focus on this, it tends to be restricted to a few very high-level determinants such as staff size (Morisset and Andrews- Johnson, 2004) or responsiveness to investors’ inquiries (Harding and Javorcik, 2012). Instead, to identify the characteristics that may shape the functioning of IPAs we rely on the more qualitative literature reviewed by Heilbron and Kronfol (2020). They propose a general framework that can be expanded into four broad pillars (Table 1): (i). Institutional arrangement focuses on the broad place that an investment promotion agency has within the government, and the overall legal, financial, and managerial pressures that shape the incentive structure of the agency and its interaction with the private sector. Key dimensions assessed under this pillar are the type of an agency (private versus public agency), the existence and structure of a board of directors, as well as subnational-national dynamics. (ii). Strategic alignment and focus is concerned with the coherence of an agency’s mandate and activities with the broader development agenda of a country or region, as well as the level of focus that it displays in its approach. Key dimensions assessed under this pillar are the number and nature of functions included in an IPA’s mandate, the degree to which an IPA engages in sectoral targeting, how it selects priority sectors, and the share of its resources dedicated to different types of firms. (iii). Organizational framework and resourcing more directly zooms in on the specific structure of an agency, how this aligns with its strategic priorities, and whether an agency is sufficiently funded and has qualified staff at its disposal to be able to reach its targets. Key dimensions assessed under this pillar are the professional experience of staff and staff compensation. (iv). Investor service delivery focuses on the different services an IPA performs, whether these are in line with the agency’s strategic priorities, and whether an agency has adopted adequate tools, systems, and M&E to effectively deliver these services. The most important services assessed are marketing services, information services, assistance services, and advocacy services. Key dimensions assess are different ways of how investment promotion plans for priority sectors are implemented, the existence of tools and systems to facilitate operations, and whether an IPA has an electronic database with updated contact information. Table 1: Characteristics of effective IPAs Pillar Best Practice for IPAs Institutional Arrangement • High degree of autonomy, and close connection to the private sector. 6 • Strong partnership within the country (alignment with other government agencies, including sub-national IPAs). Strategic Alignment and • Focus on promoting a specific sectors or business activities. Focus • Mandate is focused narrowly on attracting FDI (rather than on domestic investment or administrative functions) Organizational Framework • Organizational structure aligns with strategic priorities and sufficiently funded. and Resourcing • IPA staff should have relevant private-sector experience with adequate pay. Investor Service Delivery • High-quality service delivery focused on sectoral strategic priorities. • Adopt adequate tools, systems and M&E to best address investor priorities and strengthen service delivery. Source: Author’s adaptation from Heilbron and Kronfol, 2020 The next few sections will go through each pillar and describe the existing literature evaluating the impact of specific characteristics on an IPA’s ability to attract FDI. Institutional Arrangement IPAs with a higher degree of autonomy and a stronger connection to the private sector tend to be more effective3 (ECORYS 2013; Lim 2018; Loewendahl 2001; Nelson 2009; UNCTAD 1997; Wells and Wint 2000). To help realize this focus, Heilbron and Kronfol (2020) argue that more effective IPAs tend to operate as a (semi)-private organization (rather than a public body or ministerial department). They may also report to a separate board of directors, which can safeguard autonomy while ensuring accountability to investment targets. Such IPA boards are further recommended to include the participation of private sector representative, to help better understand investor concerns and deliver relevant services to them (ECORYS 2013; Miškinis and Byrka 2014). Jointly, these characteristics allow IPAs to receive consistent support even during periods of political transition, attain better understanding of investor needs, and work more effectively alongside private sector actors (Bauerle Danzman and Gertz, 2020). Effective IPAs also tend to develop strong partnerships within the wider government to address market failures and promote investment. Issues affecting the investment climate cut across many different ministries and agencies. IPAs pushing for policy reforms thus often benefit from strengthening such collaborations (Qiang, Liu and Steenbergen, 2021). They also benefit from support from the topmost levels of government, sometimes linked to a high institutional status, hierarchy, or attachment to upper ministry levels (Lim 2018; Morisset and Andrews-Johnson 2004; Volpe Martincus and Sztajerowska 2019). Finally, countries with multiple IPAs (for example, at national and sub-national level) also ensure to complement each other, and avoid a “race to the bottom” where multiple IPAs are competing for the same investors on the basis of incentives or concessions. Complementary mandates and protocols of engagement help to realize this (Heilbron and Kronfol, 2020; Fernandez et al. 2021; Phillips et al. 2021). Strategic Alignment and Focus IPAs are more likely to succeed when they focus strategically on promoting specific sectors or business activities, and restrict their mandate narrowly to attracting foreign investment. Evidence for the 3 This finding is in line with the broader literature on industrial policy, which calls on agencies to receive some degree of autonomy to be shielded from both short-sighted political interference as well as unscrupulous business interests. At the same time, agencies need close, ongoing collaborations with the private sector to address the specific market failures holding back performance, and so should be “embedded within a network of linkages with private groups” (Evans, 1995). While this “embedded autonomy” is difficult to realize, Dani Rodrik (2004) notes that “getting balance right is so important that it overshadows all other elements of policy design”. 7 importance of IPA sectoral targeting is extensive (see section 2.1) and is often considered to help the IPA by making it easier to communicate priorities, target investors and advocate necessary policy reforms. Heilbron and Kronfol (2020) further argue that the mandate of effective IPAs is limited mostly to foreign investment promotion. Yet, many countries expand the role of IPAs to focus on other tasks including supporting domestic direct investment (DDI), negotiate investment agreements, issue licenses, promote exports, negotiate public concessions, and administer public- private partnerships (PPPs). This can lead to conflicts of interest when the promoter is under the same roof as the regulator, or incentives approver or when resources favor domestic investment to the detriment of FDI promotion (Heilbron and Whyte 2019). It may also dilute a focus on FDI promotion. As such, Heilbron and Kronfol (2020) find a strong negative association between the number of IPA mandates and FDI inflows in developing countries. Organizational Framework and Resourcing A third component of effective IPAs proposed by Heilbron and Kronfol (2020) relates to an organizational structure that aligns with its vision and strategic priorities. This includes ensuring that the organization has the appropriate guidelines, protocols and key performance indicators (KPIs) that correspond to its strategy. In addition, such IPAs are ensured to receive sufficient and sustained financial resources over the medium-term, to ensure a stable and continued ability to purpose investment promotion (Morisset and Andrews; Johnson 2004; Volpe Martincus and Sztajerowska 2019). A greater overseas presence of IPAs through dedicated offices may also help their functioning (see 2.1; Anderson and Sutherland 2015). Yet the most important resource for IPAs tends to be its staff – and so effective IPAs recruit workers with the appropriate experience and pay. According to Ortega and Griffin (2009) IPAs need both management and staff members who have appropriate private sector experience, service skills, and deep business knowledge (including understanding of investor needs, motivations, challenges, and concerns but also sector terminology and trends). Nelson (2009) further calls for staff with international exposure, that would have the appropriate language and cultural skills to foster translational learning. They further find that traditional civil service recruitment and pay policies hamper IPAs potential to recruit such qualified, specialized staff. More effective IPAs therefore have the operational independence and financial ability to recruit such staff and provide them with adequate compensation (often in line with the private sector). Investor Service Delivery A fourth area of IPA effectiveness relates to their ability to provide high-quality services to increase investor satisfaction, notably in areas considered most relevant to meet sectoral strategic priorities. As an example, Harding and Javorcik (2012) find that better information services (such as website information and inquiry handling) are positively correlated with FDI inflows. According to the World Bank’s 2017 GIC Survey, IPA services were often most appreciated by investors during the establishment, retention, and expansion stages of investment. This included both hands-on assistance with issues during registration and business establishment, but also advocacy to improve the business environment (World Bank 2018). IPAs with quality services also tend to adopt a series of tools, systems and monitoring and evaluation (M&E) framework that help identify investor priorities and strengthen their service delivery. Examples of this include development of standard operating procedures, a customer relationship management (CRM) system or adopting a system for gathering investor complaints or disputes. Using digitalization and emerging technologies was considered an important attribute by IPAs to help reach target investors more efficiently (DCI 2017; WAIPA 2019). The United Kingdom reported several improvements in the functioning of the different IPAs in its union because of a new M&E framework (DIT 2018, 2019). 8 3. Data To better understand what characteristics help explain the effectiveness of IPAs, this paper brings together new data sources on sectoral FDI and IPA characteristics. This section describes these various data sets as well as the study sample used. 3.1 Data on sectoral FDI We create a new sectoral FDI database by combining data on FDI positions from EUROSTAT’s BPM6 database with the US Bureau of Economic Analysis (BEA). EUROSTAT’s database reports information on outward- and inward FDI from 39 OECD countries for the years 2013 to 2018. However, the amount of information reported differs significantly across countries due to different country-level reporting standards. As a result, some countries only report aggregate FDI stock (e.g. Switzerland) or report on only a small number of partner countries or sectors (e.g. Japan and the United States). We therefore only include data from 29 reporting countries. To create the widest available data on a country’s FDI, we use their information on outward FDI positions (as our interest is in both OECD and non-OECD countries). The sectoral data is available for a range of NACE4 sectors at the 1-digit to 3-digit-level (country depending). This data is then combined with the sectoral FDI database from the US BEA, which provides information on US FDI positions across the world since the 1980s. However, this relies on a distinct set of sectors that do not align directly to any NACE code. We therefore create our own concordance table to harmonize this with the EUROSTAT data. We also only keep the years 2013-2018 from this database, for which US BEA has information on over 200 countries. All FDI values are then transformed into constant 2018 USD terms. The summary statistics on FDI reporting countries is given in table 2, showing each country’s count, number of unique sectors and their unique country-pairs (i.e. the number of countries for which they report their sectoral FDI position). Table 2: FDI reporting countries– breakdown by country Data Source Country ISO Country Name Total Count Unique Sectors Unique Country-Pair Count EUROSTAT BEL Belgium 12,181 79 36 BGR Bulgaria 10,930 79 34 CYP Cyprus 10,207 77 26 CZE Czech Republic 11,149 75 32 DEU Germany 8,293 79 35 DNK Denmark 12,186 78 35 ESP Spain 1,313 48 31 EST Estonia 11,246 79 35 FIN Finland 9,674 77 35 FRA France 3,525 66 34 GBR United Kingdom 6,269 77 35 GRC Greece 10,950 79 34 HRV Croatia 11,024 73 26 HUN Hungary 8,598 74 30 IRL Ireland 6,994 79 31 ITA Italy 7,615 75 35 LTU Lithuania 9,859 79 26 LVA Latvia 9,779 82 23 MLT Malta 9,572 79 32 4 NACE is the industry standard classification system used in the European Union, and comes from the French term "nomenclature statistique des activités économiques dans la Communauté européenne." 9 NLD Netherlands 13,114 76 36 NOR Norway 11,349 79 36 POL Poland 10,385 74 35 PRT Portugal 7,756 77 35 ROU Romania 9,373 79 23 SRB Serbia 864 19 24 SVK Slovak Republic 8,972 79 26 SVN Slovenia 11,910 76 30 SWE Sweden 7,459 79 36 XKX Kosovo 359 38 26 US BEA USA United States of America 2,970 15 33 Total 255,875 945 3.2 Data on IPA Characteristics (WB-WAIPA) To study the specific characteristics of IPAs, we rely on a new survey conducted in 2019 by the World Bank and the World Association of Investment Promotion Agencies (WAIPA) that included 91 IPAs – 87 national and 4 sub-national IPAs. This information is then combined with the new FDI database to explore the effect of IPA sectoral targeting on inward FDI. This leads to a study sample of 36 national IPAs from middle- and high-income countries around the world (Table 3). Of these, 26 IPAs have sectoral targeting (group 1, the core sample), while 10 do not target any priority sectors (group 2, used for control only). We align the two data sets by adapting the potential priority sectors of the WB-WAIPA survey to 28 NACE sectors, corresponding to either NACE1 (e.g. for agriculture, mining and services) or NACE2 level (for manufacturing sectors). Table 3: Study sample from WB-WAIPA Survey Study sample Excluded from the study Group 1: Sectoral FDI data Group 2: Sectoral FDI data Group 3: No sectoral FDI data available available, with sectoral targeting available, no sectoral targeting (CORE SAMPLE) (CONTROL ONLY) Austria Canada Albania Haiti Brazil Finland Andorra Iraq Bulgaria France Angola Jamaica Chile Germany Antigua and Barbuda Jordan Cyprus Iceland Azerbaijan Kazakhstan Czech Republic Nigeria Bahrain Kuwait Denmark Portugal Bangladesh Lebanon Egypt, Arab Rep. Russian Federation Belarus Lesotho Bosnia and Madagascar Estonia United States Herzegovina Greece Australia* Botswana Mali Hong Kong SAR, China Cambodia Mauritius Hungary Cameroon Mongolia India China Montenegro Ireland Comoros Mozambique Italy Congo, Rep. Namibia Poland Cook Islands Nicaragua Romania Costa Rica North Macedonia Korea, Rep. Curaçao Pakistan Djibouti West Bank and Slovak Republic Gaza Slovenia Dubai- UAE Rwanda South Africa El Salvador Scotland, UK Spain Eswatini Serbia 10 Switzerland Ethiopia Sri Lanka Türkiye French Polynesia St. Kitts and Nevis United Kingdom Gabon Uganda Uruguay Ghana Vanuatu Guatemala Vietnam Guinea Note: * = No time-varying patterns in sector targeting, and so cannot be included in treatment group. Table 4 then provides a breakdown of the FDI statistics for this country-sample, showing the total count per country and their number of included sectors. It also shows the unique country pairs (that is – the number of countries reporting their FDI position in the host country) which range from a low of 16 (for Uruguay) to a high of 29 countries (for Austria, the Russian Federation, and Switzerland). To consider the share of FDI captured by this new sectoral FDI database, we also compare it to the aggregate FDI stock from the countries included in the World Bank’s harmonized bilateral FDI database.5 This finds that the FDI-reporting countries jointly make up almost two-thirds of all global FDI stock. For the study sample this share is even higher, covering over 70 percent of all countries’ FDI stock (see Annex tables A1 and A2). The three countries with the highest share of FDI covered by our sectoral FDI database are Italy, the Arab Republic of Egypt and Finland (with 96, 95 and 95 percent of FDI covered by the sample, respectively) while the lowest countries are Hong Kong SAR, China, India and Uruguay (at 13, 44 and 50 percent, respectively). In sum, for most target countries the sectoral FDI database provides a reasonable approximation for their aggregate FDI stock. Yet, for selected countries the interpretation is more limited, and focused on only the effectiveness of IPAs at attracting FDI from OECD countries. Table 4: Study sample – breakdown by country Country ISO Country Name Total Count Sectoral Count Unique Country-Pair Count AUS Australia 1,938 28 26 AUT Austria 3,377 28 29 BGR Bulgaria 3,086 28 28 BRA Brazil 2,876 28 25 CAN Canada 3,052 28 26 CHE Switzerland 3,342 28 29 CHL Chile 1,557 28 23 CYP Cyprus 3,204 28 27 CZE Czech Republic 3,096 27 28 DEU Germany 3,102 28 28 DNK Denmark 3,236 28 27 EGY Egypt, Arab Rep. 1,555 26 23 ESP Spain 3,142 28 27 EST Estonia 2,694 28 22 FIN Finland 2,877 28 25 FRA France 3,254 28 28 GBR United Kingdom 3,217 28 28 GRC Greece 2,772 28 26 HKG Hong Kong SAR, China 2,610 28 23 HUN Hungary 3,025 28 27 IND India 2,993 28 25 IRL Ireland 3,086 28 27 ISL Iceland 1,223 28 18 5This harmonizes bilateral FDI from the IMF CDIS database, the OECD BMP4 database and China’s Annual Yearbook. Because aggregate and sectoral FDI positions are not directly comparable, we estimate each country’s share of aggregate FDI stock tha t arise from the countries that report on sectoral FDI (i.e. those in table 2). 11 ITA Italy 3,315 28 28 KOR Korea, the Republic of 1,177 28 20 NGA Nigeria 1,119 28 18 POL Poland 3,228 28 28 PRT Portugal 2,941 28 25 ROU Romania 3,149 28 28 RUS Russian Federation 3,164 28 29 SVK Slovak Republic 3,218 28 27 SVN Slovenia 2,962 28 25 TUR Türkiye 1,721 27 28 URY Uruguay 959 28 16 USA United States of America 3,161 28 28 ZAF South Africa 1,400 28 22 Total 95,828 917 4. Empirical Strategy Following the economic literature on evaluating IPA effectiveness (Harding and Javorcik, 2011; Crescenzi, Di Cataldo and Giua, 2021), our empirical strategy makes use of a staggered difference-in-difference model that exploits information on IPAs’ targeting strategies by sector. As such, the unit of observation is the country-sector-year level. We estimate whether sectors which were classified as an IPA “priority” experienced an increase in their FDI stock compared to their time in which they were not a priority sector, and relative to non-priority sectors. Such an occurrence would provide evidence of the effectiveness of IPAs by demonstrating their ability to attract the type of investment they seek. The model commonly used for this is: = 1 + + + + (1) Where is the outcome variable measuring the total inward FDI (stock or flow) in country c, in sector s, in year t. is a dummy variable that refers to the targeting of sector s in year t by an IPA. 1 is the coefficient of interest, capturing the average effect of the IPA strategy on FDI. and are region-time and sector-time fixed effects to deal with unobserved country- and sector-specific annual shocks (e.g. macro-economic shocks and sector-specific shocks). is a country-sector fixed effect to deal with pre-existing structural conditions. are idiosyncratic error terms. There are two main endogeneity concerns that arise from specification (1). The first lies with the non- randomness of the priority sectors chosen by IPAs for FDI attraction. The priority sectors may differ in their ex-ante competitiveness compared to non-priority sectors, or may experience other interventions (non-IPA related) that are driving the result. We address this issue through the staggered difference-in- difference model exploiting the timing of the sector-targeting, and through the wide set of fixed effects to minimize confounding factors. A second, less explored concern, is that IPAs explicitly use targeting to improve their relative competitiveness vis-à-vis alternative host countries. There is only a finite amount of FDI in the world, and so an IPA’s targeting of a priority sector in one country may affect the relative effectiveness of IPAs in other countries targeting the same sector. This is a common concern in the trade literature, known as “multilateral resistance” (see section 2.1). Anderson and van Wincoop (2003) showed the importance of multilateral resistance general equilibrium effects in estimating the response of trade flows to trade costs. 12 To develop the structural gravity model of FDI, we follow the approach suggested by Kox and Rojas- Romagosa (2020). First, even though trade gravity models usually depend on flows, we use the stock of (sectoral) bilateral FDI as our dependent variable, as it is much less volatile than FDI flows and tends to be non-negative (which would be problematic for some of the estimation methods). We are careful to distinguish between real zero FDI stocks and missing (or non-reported, suppressed) data. Second, we employ a Pseudo-Poisson maximum likelihood (PPML) estimator, because it effectively deals with zero bilateral FDI stocks and accounts for the presence of heteroskedasticity in FDI data (Santos Silva and Tenreyro, 2006). Third, alongside our standard set of gravity variables we also follow the best practice of using country-pair fixed effects to account for any unobservable time-invariant FDI barriers, which has been proven to be a better measure of the bilateral costs (Agnosteva, Anderson, and Yotov, 2014; Egger and Nigai, 2015). Finally, we use source-year and host-year fixed effects to properly account for multilateral resistance terms in panel data gravity estimations (Olivero & Yotov, 2012). The country of origin generates the outward FDI stock and the destination country the inward FDI stock. The origin-time and destination-time fixed effects also absorb the country size variables from the structural gravity model, in addition to all other observable and unobservable time-varying country-specific characteristics, including different national policies, institutions and exchange rates (Yotov et al. 2016). Based on this, our main specification is: = 1 + + + + + + (2) This specification is similar to (1), but is now set up in a bilateral framework where is the inward FDI stock from source country i to the host country j, for sector s in period t. 1 is still the coefficient of interest, capturing the average effect of the IPA strategy of the host country on sectoral FDI stocks. captures the country-pair fixed effects, while , , and reflect the source-year, host-year, sector-year and host-sector fixed effects, respectively. are the idiosyncratic error terms. Standard errors are clustered at the country-pair-sector level and are heteroscedasticity-robust, correcting for possible bias induced by estimation of log-linearized models (Santos Silva and Tenreyro, 2006). Characteristics of effective IPAs In the second part of the analysis, we also consider the relative effectiveness of different IPAs. This is done by incorporating an interaction effect with the IPA strategy in our main specification: = 1 + 2 ∗ + + + + + + (3) Here, the main coefficients of interest are 1 (the average effect of IPA targeting) and 2 (the country- specific effect of IPA targeting). From this, we define three different groups: ‘high-performing IPAs’ are those in which country-interaction effects show a positive, significant effects of IPA targeting on FDI stocks. ‘Other IPAs’ are those in which country-interaction effects either show an insignificant or negative, significant effect of IPA targeting on FDI stocks. Finally, a select group of countries have no sectoral targeting or no variation in the timing of the sector-strategy, thus precluding any country-level analysis. Next, we consider how high-performing IPAs differ from other IPAs across a range of characteristics. This is done by comparing the mean values of both groups, making use of t-test to consider any statistically significant effect. These variables are grouped according to the four pillars identified by Heilbron and Kronfol (2020): institutional arrangement, strategic alignment and focus, organizational framework and resourcing and investor service delivery. 13 Remaining issues A remaining challenge for the analysis may pertain to reported FDI data. A first concern here may be that this does not capture FDI from all countries, but only OECD countries. However, due to the prominence of OECD-countries in outward FDI, we find that in most cases this data set captures a large share of a country’s FDI stock (see data section for details). In cases where the share of FDI from OECD countries is lower (e.g. Hong Kong SAR, China, or India), this would not introduce a bias (as the estimate would still be consistent), but changes the interpretation of the estimate to reflect an IPA’s ability to attract FDI from OECD countries. As such, this may potentially affect the external validity of the estimate, if it is expected that different strategies are more effective for non-OECD countries (but we are not aware of any such theories). An additional limitation comes from the analysis of characteristics of effective IPAs. Using t-tests to distinguish between high-performing and other IPAs does not allow us to make causal inference about IPA characteristics to drive FDI dynamics. Instead, these results should be considered indicative only. This is partly because there may be other factors (e.g. strong political leadership) which may drive both IPA characteristics and FDI performance. These IPAs may then just be part of a government’s wider strategy of FDI sectoral targeting. It also does not allow us to gauge exactly how important specific characteristics are. That said, we do think that if there is a strong, consistent pattern emerging across all countries that see a strong FDI effect in sectors targeted by the IPA that it is not unreasonable to assume that at least part of this contribute this to the underlying IPA characteristics. In addition, to the extent that IPAs can align with the wider government’s sectoral growth strategy (or push sector-specific reforms through internal advocacy), this itself reflects an “effective” IPA. 5. Results 5.1 Sectoral Gravity Model The results for the sectoral gravity model are presented in Table 5. The first equation uses traditional gravity model variables from the CEPII database (Head and Mayer, 2014) for the OLS and PPML model, respectively, and combines this with simple year fixed effects. Both models operate in line with expectations, as GDP source country and GDP host country are significant and positively correlated to FDI stock, while source-host distance is negatively correlated. Countries that are more culturally proximate (e.g. through a common language, colonizer or religion) also appear to be more likely to invest in one another in the OLS model, though the common religion variable is the only cultural approximation variable that is significant in the PPML model. In both cases, IPA sector targeting is highly significant and positive. In the second set of regressions (equation 2 and 4), the traditional gravity variables are replaced with country-pair fixed effects, and time-varying dynamics are captured through source-year and host-year fixed effects. In both cases, the IPA sector targeting coefficient remains significant and positive, though reduces in size (thus capturing part of the unobservable country-pair FDI dynamics through fixed effects). The third set of regressions further include sector-year and host-sector fixed effects. Again, the IPA sector targeting coefficient remains significant and positive, but further reduces in size (suggesting that IPAs are likely targeting those sectors which see larger increases in their FDI stock across the board). Our most conservative, and preferred, estimate is from the PPML model with the widest set of controls (equation 6). This suggests that an IPA’s targeting of a specific sector is associated with around 10 percent increase in the value of their country’s FDI stock in that sector over the timeframe of the analysis. 14 Table 5: Sectoral gravity model OLS PPML (1) (2) (3) (4) (5) (6) FDI Stock FDI Stock FDI Stock FDI Stock FDI Stock FDI Stock VARIABLES (Logged) (Logged) (Logged) (Logged) (Logged) (Logged) IPA Sector Targeting 0.192*** 0.164*** 0.114*** 0.198*** 0.133*** 0.0987*** (0.0558) (0.0278) (0.0290) (0.0387) (0.0208) (0.0201) GDP Source Country (Logged) 0.798*** 0.582*** (0.0292) (0.0171) GDP Host Country (Logged) 0.390*** 0.301*** (0.0348) (0.0216) Source-Host Distance in Km (Logged) -0.371*** -0.292*** (0.0566) (0.0396) Common Language 0.686*** -0.0913 (0.204) (0.0872) Common Colonizer 1.306*** 0.371 (0.305) (0.505) Common Religion 0.549*** 0.571*** (0.165) (0.104) Regional Trade Agreement 0.0216 0.0540 (0.126) (0.0964) Year F.E. YES NO NO YES NO NO Country-Pair F.E. NO YES YES NO YES YES Source-Year F.E. NO YES YES NO YES YES Host-Year F.E. NO YES YES NO YES YES Sector-Year F.E. NO NO YES NO NO YES Host-Sector F.E. NO NO YES NO NO YES Constant -19.45*** 1.080*** 1.143*** -15.72*** 0.960*** 1.150*** (1.060) (0.0354) (0.0369) (0.742) (0.0267) (0.0258) Observations 85,718 92,830 92,830 85,718 85,982 85,982 R-squared 0.367 0.620 0.696 0.399 0.481 0.569 Note: Author’s calculations. Robust standard errors in parentheses, clustered at the country-pair-sector level. FDI stock is defined at bilateral, sectoral level in USD Million (Logged) in constant 2018 terms. OLS = Ordinary Least Squares. PPML = Poisson pseudo-maximum likelihood estimation. FDI = Foreign Direct Investment. IPA = Investment Promotion Agency. KM = Kilometers. GDP = Gross Domestic Product. F.E = Fixed Effects. *** p<0.01, ** p<0.05, * p<0.1 To further gauge the relative effectiveness of IPAs, we use the model from equation 6, and include an additional country-interaction effect with the IPA sectoral targeting coefficient. To maintain the anonymity of respondents in the WB-WAIPA survey, we will not make public the names of the specific countries, but instead focus on aggregate results only. Their results are summarized in Table 6 and presented in full detail in Table 7. These findings suggest that there is considerable heterogeneity between IPAs in their ability to affect FDI stocks. Overall, 10 countries have positive and statistically significant results (which we will classify as “high performing IPAs” in the next section). An additional 13 IPAs have statistically insignificant results, while 3 IPAs have negative and statistically significant results (which we will jointly classify as “other IPAs” in the next section). Finally, 10 IPAs fell out of the country -interaction equation as they either did not have sectoral targeting or exhibited no variation in the timing of their 15 sector targeting strategy (and thus could not be evaluated within a difference-in-differences empirical strategy). In such cases, we were unable to gauge the relative effectiveness of these IPAs and as such they were excluded from the IPA characteristics analysis. Table 6: Results from sectoral gravity model with country interaction effects Country interaction results Count Defining IPA type Positive and statistically significant results 10 High performing IPAs Statistically insignificant results 13 Other IPAs Negative and statistically significant results 3 Other IPAs No country results (no sector targeting or time variation) 10 Not included in analysis of IPA characteristics Total 36 Note: Author’s calculations summarizing the results from Table 7. Significance is defined at the 10 percent level or higher. Table 7: Sectoral gravity model with country interaction effects (1) VARIABLES FDI Stock (Logged) IPA Sector Targeting -0.000775 (0.0681) IPA Sector Targeting * Country 1 0.852*** -0.198 IPA Sector Targeting * Country 2 0.274*** -0.0913 IPA Sector Targeting * Country 3 0.262** -0.129 IPA Sector Targeting * Country 4 0.249* -0.149 IPA Sector Targeting * Country 5 0.230** -0.0905 IPA Sector Targeting * Country 6 0.207* -0.118 IPA Sector Targeting * Country 7 0.199* -0.111 IPA Sector Targeting * Country 8 0.192** -0.0888 IPA Sector Targeting * Country 9 0.190* -0.113 IPA Sector Targeting * Country 10 0.133* -0.0796 IPA Sector Targeting * Country 11 0.195 -0.22 IPA Sector Targeting * Country 12 0.134 -0.105 IPA Sector Targeting * Country 13 0.125 -0.08 IPA Sector Targeting * Country 14 0.121 -0.1 IPA Sector Targeting * Country 15 0.0386 -0.116 IPA Sector Targeting * Country 16 0.0389 -0.109 IPA Sector Targeting * Country 17 0.0309 -0.0913 IPA Sector Targeting * Country 18 0.108 -0.0907 IPA Sector Targeting * Country 19 0.00666 -0.0868 IPA Sector Targeting * Country 20 0.00257 -0.112 IPA Sector Targeting * Country 21 -0.0137 -0.134 IPA Sector Targeting * Country 22 -0.0668 16 -0.105 IPA Sector Targeting * Country 23 -0.0955 -0.255 IPA Sector Targeting * Country 24 -0.281** -0.136 IPA Sector Targeting * Country 25 -0.376* -0.224 IPA Sector Targeting * Country 26 -0.515*** -0.198 Country-Pair F.E. YES Source-Year F.E. YES Host-Year F.E. YES Sector-Year F.E. YES Host-Sector F.E. YES Constant 1.247*** (0.00544) Observations 86,148 Note: Robust standard errors in parentheses, clustered at the country-pair-sector level. FDI stock is defined at the bilateral, sectoral level in USD Million (Logged) in constant 2018 terms. Green-highlighted fields indicate countries with positive, statistically significant IPA effects at the 10 percent level or higher. PPML = Poisson pseudo-maximum likelihood estimation. IPA = Investment Promotion Agency. F.E = Fixed Effects. *** p<0.01, ** p<0.05, * p<0.1 5.2 Characteristics of effective IPAs General Characteristics and Institutional Arrangement A comparison of the two groups finds that high-performing IPAs are found across different regions and income levels. As shown in Table 8, IPAs with sectoral targeting strategies are found to be significant predicters of increasing FDI stock across Europe, Asia and the Americas, and their prevalence is not unlike those of the group of other IPAs. 6 Similarly, we see that across the two groups there is a broad division of IPAs across income levels, suggesting that effective IPAs are not only present in high-income countries but can also be developed in developing countries. Yet, when considering their institutional arrangement, the type of agency and their board of directors are both significant predicters to whether an IPA is in the high-performing group. High-performing IPAs are much more likely to be part of a private or semi-private agency, while other IPAs are more likely to be part of a government agency (a significant difference at the 5 percent level). Another striking finding is that 80 percent of IPAs in the effective group have a board of directors, while only around 40 percent of other IPAs have such a board (significant at the 10 percent level). For those IPAs that do have a board, high-performing IPAs also have a considerably higher average share of private sector representatives (58 percent, versus 38 percent in other IPAs). However, this result is not statistically significant due to the relatively small number of IPAs in both groups. Both of these findings are aligned to the predictions by Heilbron and Kronfol (2020) around the importance of (semi)-autonomous IPAs with close connection to the private sector. For the sub-national dynamics, half of both groups have national IPAs with regional offices, while around a third have separate sub-national IPAs operating independently. While this leaves only a very small group to study, it is striking that all high-performing IPAs that also have a sub-national IPA in their country report to having a close, systematic working relationship with them that includes regular joint activities and information sharing. In contrast, other IPAs were split in their responses and may have had only regular but ad hoc, or occasional contact with sub- national IPAs. These results thus seem to imply that close coordination is important for effective IPAs, but due to the sample size of IPAs with sub-national, no statistically significant findings are present. Strategic Alignment and Focus Table 9 considers the strategic alignment of IPAs in both groups. Interestingly, this finds that there is no significant difference in the number of sectors targeted between high-performing IPAs and other IPAs. Yet, this first group may be slightly more likely to dedicate at least half of their resources to priority sectors (90 percent) versus other IPAs (70 percent). There also seems to be no significant difference in how both groups select their priority sectors, as 6One exception is that no African IPAs are in our group of effective IPAs. However, this may be partly due to the small number of such countries in our sectoral FDI database. 17 both are roughly equal to pick these based on a combination of their national development strategy/policy and based on their own research (including elements such as their FDI trends and export potential). One notable difference does appear to exist in the strategic alignment of these two groups of IPAs, related to their share of resources dedicated to targeting specific types of firms. While both groups allocate considerable funds targeting large foreign firms (upwards of 40 percent), effective IPAs spend an even larger share on attracting small and medium-sized foreign firms (almost 50 percent of all their resources) and allocate almost nothing on domestic investors. In contrast, other IPAs only dedicate around 20 percent of their resources to small and medium-sized foreign firms and dedicate another 20 percent of their resources to stimulating investment from domestic firms. Table 10 confirms that there is one significant difference in the mandate of the two groups of IPAs – the focus on promotion of domestic investment. Only 10 percent of all high-performing IPAs report a responsibility in promoting domestic investment, versus almost two-thirds of all IPAs. They may also be less likely to take on administrative functions such as incentives administration or issuing of licenses and permits, though this is not statistically significant. In general, this does suggest that a narrow focus on foreign investment is important for effective IPAs. Table 8: General Characteristics and Institutional Arrangement High-performing Other Differen Variable IPAs IPAs ce Region Europe 0.500 0.688 -0.188 [0.167] [0.120] Asia 0.300 0.125 0.175 [0.153] [0.085] Americas 0.200 0.063 0.138 [0.133] [0.063] Africa 0.000 0.125 -0.125 [0.000] [0.085] Income Level High income 0.700 0.688 0.013 [0.153] [0.120] Upper middle income 0.200 0.250 -0.050 [0.133] [0.112] Lower middle income 0.100 0.063 0.037 [0.100] [0.063] Type of Agency Semi-Autonomous Public Agency 0.500 0.500 0.000 [0.167] [0.129] Private or Semi-Private Agency 0.300 0.000 0.300** [0.153] [0.000] Government Agency 0.200 0.500 -0.300 [0.133] [0.129] Board of directors Does the IPA have a board of directors 0.800 0.438 0.362* [0.133] [0.128] Share of board represented by public sector 0.411 0.613 -0.203 [0.112] [0.155] Share of board represented by private sector 0.580 0.375 0.205 [0.109] [0.153] Share of board represented by others (e.g., academia, civil society) 0.009 0.011 -0.002 [0.009] [0.011] Sub-national dynamics IPA has regional offices within the country? 0.500 0.500 0.000 [0.167] [0.129] Are any sub-national IPAs operating in the country? 0.300 0.313 -0.013 [0.153] [0.120] 18 Close, systematic working relationships, with regular joint activities and information sharing 1.000 0.400 0.600 [0.000] [0.245] Regular but ad hoc with sub-national IPA 0.000 0.200 -0.200 [0.000] [0.200] Occasional contact with sub-national IPA 0.000 0.400 -0.400 [0.000] [0.245] Note: Author’s calculations. The value displayed for t-tests are the differences in the means across the groups. on the sectoral gravity model with country interaction effects (see section 4.1). IPA = Investment Promotion Agency. Green-highlighted fields indicate statistically significant differences between high-performing IPAs and other IPAs. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 19 Table 9: Strategic Alignment Variable High-performing IPAs Other IPAs Difference Sectoral targeting Number of sectors targeted 11.300 11.438 -0.137 [1.620] [1.231] Dedicate 50% or more of IPA’s resources to priority sectors only 0.900 0.688 0.212 [0.100] [0.120] How are the priority sectors selected Taken from national development strategy or policy document 0.700 0.875 -0.175 [0.153] [0.085] Selected based on research (e.g., FDI trends, export potential) 0.700 0.688 0.013 [0.153] [0.120] Selected by IPA management in consultation with stakeholders 0.500 0.375 0.125 [0.167] [0.125] Selected by the office to which the IPA reports 0.100 0.250 -0.150 [0.100] [0.112] Selected by IPA management alone 0.100 0.125 -0.025 [0.100] [0.085] Share of resources dedicated to type of firms Large foreign firms 42.222 44.333 -2.111 [6.186] [4.574] Small and medium-sized foreign firms 46.667 21.000 25.667*** [6.455] [3.525] Joint ventures between foreign and domestic firms 5.000 7.333 -2.333 [2.357] [1.944] Large domestic firms 2.222 9.400 -7.178** [1.211] [1.983] Small and medium-sized domestic firms 1.111 10.933 -9.822** [0.735] [2.906] Mega deals 2.778 7.000 -4.222 [1.470] [2.000] Note: Author’s calculations. The value displayed for t-tests are the differences in the means across the groups. High-performing IPAs are identified based on the sectoral gravity model with country interaction effects (see section 4.1). IPA = Investment Promotion Agency. FDI = Foreign Direct Investment. Green-highlighted fields indicate statistically significant differences between high-performing IPAs and other IPAs. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Table 10: Strategic Focus – Functions included in the IPA mandate Variable High-performing IPAs Other IPAs Difference Functions included in the IPA’s mandate Foreign Investment promotion 1.000 1.000 N/A [0.000] [0.000] Matchmaking services (foreign investors to local suppliers) 0.700 0.625 0.075 [0.153] [0.125] Innovation promotion 0.700 0.375 0.325 [0.153] [0.125] Policy advocacy / advocate for investment climate reforms 0.600 0.438 0.163 [0.163] [0.128] Export promotion 0.400 0.500 -0.100 [0.163] [0.129] Development of local suppliers 0.400 0.188 0.212 [0.163] [0.101] One-stop shop 0.400 0.500 -0.100 [0.163] [0.129] Outward investment support 0.300 0.313 -0.013 [0.153] [0.120] Screening/approval of investment projects 0.300 0.313 -0.013 [0.153] [0.120] Small and medium-sized enterprise development 0.200 0.313 -0.113 20 [0.133] [0.120] Administration of special economic zones or industrial parks 0.200 0.313 -0.113 [0.133] [0.120] Administration of incentives 0.200 0.438 -0.237 [0.133] [0.128] Promotion of domestic investment 0.100 0.625 -0.525*** [0.100] [0.125] Negotiation of international investment agreements 0.100 0.125 -0.025 [0.100] [0.085] Issue of other licenses or permits 0.000 0.188 -0.188 [0.000] [0.101] Note: The value displayed for t-tests are the differences in the means across the groups. High-performing IPAs are identified based on the sectoral gravity model with country interaction effects (see section 4.1). IPA = Investment Promotion Agency. FDI = Foreign Direct Investment. Green-highlighted fields indicate statistically significant differences between high-performing IPAs and other IPAs. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Organizational framework and resourcing For the organizational framework and resourcing of IPAs, Table 11 shows that high-performing IPAs differ significantly to other IPAs in both the type of staff members employed and their compensation. Around 80 percent of this first group’s professional staff tends to have private sector experience, versus 50 percent in the other group (a significant difference at the 1 percent level). This mainly comes at the expense of public sector experience, which is the case for about one-third of the high-performing staff versus nearly two-thirds of staff for other IPAs (again, a significant difference at the 1 percent level). Both groups also tend to employ staff that is proficient in a foreign language, and the high-performing group may also be slightly more likely to employ sectoral experts, though no statistical difference was found between these groups. Yet, a drastic difference did emergence on staff compensation. 60 percent of high- performing IPAs provided pay that is comparable with the private sector, compared to less than 20 percent for other IPAs (a significant difference at the 5 percent level). In contrast, most other IPAs paid their staff at a rate that was above public but below private sector level. A second organizational characteristic of high-performing IPAs seems to lie with their permanent overseas representation (Table 11). This group is more likely to have overseas representation through their own offices abroad (which was the case for 60 percent of the first group, but less than 20 percent for the second group). Jointly, these findings indicate that a well-resourced IPA matters for effectiveness – most notably in the areas of staffing and overseas representation. Table 11: Organizational framework and resourcing Variable High-performing IPAs Other IPAs Difference Professional experience Share of professional staff with private sector experience 0.787 0.493 0.293*** [0.054] [0.067] Share of professional staff with public sector experience 0.327 0.610 -0.283** [0.109] [0.078] Share of professional staff with specific background in priority sectors 0.676 0.511 0.166 [0.093] [0.080] Share of professional staff proficient in a foreign language 0.804 0.737 0.067 [0.111] [0.065] Staff compensation Comparable with the private sector 0.600 0.188 0.412** [0.163] [0.101] At par with the public sector 0.300 0.375 -0.075 [0.153] [0.125] Above public but below private sector 0.100 0.438 -0.338* 21 [0.100] [0.128] Permanent overseas representation No permanent overseas representation 0.300 0.500 -0.200 [0.153] [0.129] Permanent overseas representation via IPA staff at national embassy 0.100 0.313 -0.212 [0.100] [0.120] Permanent overseas representation via IPA’s own offices abroad 0.600 0.188 0.412** [0.163] [0.101] Note: The value displayed for t-tests are the differences in the means across the groups. High-performing IPAs are identified based on the sectoral gravity model with country interaction effects (see section 4.1). IPA = Investment Promotion Agency. Green- highlighted fields indicate statistically significant differences between high-performing IPAs and other IPAs. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Investor Service Delivery Finally, it appears that high-performing IPAs also differ in the way that they provide their services, partly by engaging in more smart, sectoral analytics (Table 12). While Heilbron and Kronfol (2020) noted the importance of quality service delivery, it is not possible to gauge the quality of such services from an IPA- reported service. Instead, we therefore focus on the way in which IPAs implement their sectoral promotion strategy – an important indicator of quality, as shown in the wider literature (see section 2.1) and in our results (see 5.1). This finds that while high-performing IPAs and other IPAs have a lot in common in how they implement their sectoral promotion plans. Yet, high-performing IPAs are much more likely to engage in sectoral analytics – either by preparing sector profiles that detail their country’s relative advance (100 percent of high-performing IPAs vs. less than 70 percent for other IPAs) or by purchasing sectoral intelligence/research reports (60 percent vs. 25 percent). In both cases, this difference is significant at the 10 percent level. Another striking feature is that high-performing IPAs are more likely to engage in each one of the 11 listed options7 than other IPAs are, thus indicating the breadth of sectoral promotion activities undertaken by such high-performing IPAs. While tools and systems to facilitate IPA operations are commonplace in both groups – high-performing IPAs are more likely to adopt systems for identifying investor complaints or disputes (Table 12) . In both groups, all respondents noted having a customer relationship management software, and were similar in adopting standard operating procedures, and having an electronic database with updated contact information for investors. Yet the high-performing group was considerably more likely in adopting a system for gathering investor complaints or disputes (80 percent vs. 44 percent, significant at the 10 percent level). The use of this tool may therefore also be important helping IPAs to identify investor priorities and strengthen their service delivery. Table 12: Investor Services Variable High-performing IPAs Other IPAs Difference Implementation of investment promotion plans for priority sectors Preparation of sector profiles, detailing country’s relative advantages 1.000 0.688 0.313* [0.000] [0.120] Participation in sector events/conferences 0.900 0.625 0.275 [0.100] [0.125] Participation in sector trade shows 0.800 0.563 0.237 [0.133] [0.128] Organization of events, conferences, and trade shows 0.800 0.500 0.300 7 These include development of sector profiles, sector events, trade shows, sector-specialized staff, investor and communications campaigns, relationship-building, sectoral website sections, purchase of sector intelligence, and investor databases. 22 [0.133] [0.129] Sector-specialized and dedicated staff 0.700 0.625 0.075 [0.153] [0.125] Investor-targeting campaigns for priority sectors 0.700 0.625 0.075 [0.153] [0.125] Communications and public relations campaigns 0.700 0.500 0.200 [0.153] [0.129] Relationship-building with existing investor communities 0.700 0.438 0.263 [0.153] [0.128] Website section for each priority sector 0.600 0.563 0.037 [0.163] [0.128] Purchase of sector intelligence/research reports 0.600 0.250 0.350* [0.163] [0.112] Purchase of investor databases for the sector 0.300 0.250 0.050 [0.153] [0.112] Tools and systems to facilitate operations IPA has customer relationship management (CRM) software 1.000 1.000 N/A [0.000] [0.000] IPA’s CRM software is ‘mostly’ or ‘fully’ used by its staff 0.900 0.688 0.212 [0.100] [0.120] IPA has a system for gathering investor complaints or disputes 0.800 0.438 0.362* [0.133] [0.128] IPA has standard operating procedures for investor queries, aftercare. 0.700 0.688 0.013 [0.153] [0.120] IPA has shared information system cataloging general and sectoral data and information for use by investors 0.600 0.688 -0.087 [0.163] [0.120] IPA has electronic database with updated contact information Database containing current foreign investors 0.900 0.938 -0.037 [0.100] [0.063] Database containing potential foreign investors 0.800 0.688 0.113 [0.133] [0.120] Database containing domestic supplier firms 0.400 0.500 -0.100 [0.163] [0.129] Database containing available local joint venture partners 0.200 0.188 0.013 [0.133] [0.101] Note: The value displayed for t-tests are the differences in the means across the groups. High-performing IPAs are identified based on the sectoral gravity model with country interaction effects (see section 4.1). IPA = Investment Promotion Agency. CRM = Customer relationship management. Green-highlighted fields indicate statistically significant differences between high- performing IPAs and other IPAs. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 6. Conclusion This paper brings together new data and analysis on sectoral FDI and IPA characteristics to analyze what characteristics make IPAs more or less effective in attracting foreign direct investment (FDI) into their home country. It does so by first creating a new data set on bilateral, sectoral FDI positions, using data from the US Bureau of Economic Analysis (BEA) and EUROSTAT’s BPM6 database for the years 2013 to 2018. This is combined with a recent WB-WAIPA survey to explore the effect of IPA sectoral targeting on inward FDI stocks for a sample of 36 middle- and high-income countries around the world. Using a structural gravity model framework, the study finds that IPA sectoral targeting generally provides a significant positive effect on the sector’s FDI stock in that country. The most conversative, and preferred, estimate is from a PPML model that includes country-pair fixed effects, source-, host-, and sector-year fixed effects, and host-sector fixed effects. This suggests that an IPA’s targeting of a specific sector is 23 associated with an increase of around 10 percent in the value of their country’s FDI stock in that sector over the timeframe of the analysis. To gauge the relative effectiveness of IPAs, a country-interaction effect with the IPA sectoral targeting coefficient was incorporated into the same PPML model These findings suggest that there is considerable heterogeneity between IPAs in their ability to affect FDI stocks. Overall, 10 countries have positive and statistically significant results (classified as “high performing” IPAs). An additional 13 IPAs have statistically insignificant results, while 3 IPAs have negative and statistically significant results (which are jointly classified as “other” IPAs). Finally, 10 IPAs fell out of the country-interaction equation as they either did not have sectoral targeting or exhibited no variation in the timing of their sector targeting strategy. The final section compared the 10 “high performing” IPAs to the 16 “other” IPAs across a range of characteristics from the WB-WAIPA survey, and used t-tests to consider which IPA characteristics significantly differ between the two groups. This suggests that effective IPAs are more likely to be private or semi-private agencies. Their mandate also tends to be focused narrowly on foreign investment and exclude responsibilities for domestic investment promotion. Such IPAs are also more likely to have a board of directors, while their staff tends to be better compensated. Finally, high-performing IPAs are also likely to engage in more investor services, partly by engaging smart, sectoral analytics and by adopting systems for identifying investor complaints or disputes. This paper provides a modest contribution to two distinct bodies of literature. The first relates to economic literature analyzing the impact of IPA to attract FDI. Here, the added value is the paper’s use of a wide set of FDI source countries and actual FDI stock data (rather than FDI announcements). It also explicitly considers how an IPA’s targeting of a priority sector in one country may affect the relative effectiveness of IPAs in other countries targeting the same sector, by including sectoral IPA targeting within a structural gravity model framework. The paper also contributes to the literature around characteristics of effective IPAs. This paper avoids some of these limitations of the literature by taking a data-driven approach to identifying ‘high-performing’ IPAs. In addition, it compares ‘high-performing’ IPAs to other IPAs across a highly detailed set of characteristics, so as to provide a detailed assessment of potential dynamics that may shape the effectiveness of IPAs in attracting FDI. New research should focus on replicating the findings with a larger database of sectoral FDI. This would allow the analysis to study more IPAs. Having a broader timeframe would also be important (both on FDI stock, and on FDI sectoral targeting), as many of these dynamics are slow moving. Such analysis will likely be possible in the near-future as major FDI source countries and regions are increasingly publishing sectoral FDI statistics. 24 References Agnosteva, D., Anderson, J. E., & Yotov, Y. V. (2014). Intra-national trade costs: Measures and aggregation (NBER Working Paper No. 19872). National Bureau of Economic Research. https://www.nber.org/paper s/w14056 Alfaro, L., & Charlton, A. (2009). Intra-industry foreign direct investment. American Economic Review, 99(5), 2096–2119. Anderson, J. E., & van Wincoop, E. (2003). Gravity with gravitas: A solution to the border puzzle. American Economic Review, 93(1), 170–192 Anderson, John, and Dylan Sutherland. 2015. “Developed Economy Investment Promotion Agencies and Emerging Market Foreign Direct Investment: The Case of Chinese FDI in Canada.” Journal of World Business 50 (4): 815–25. Bauerle Danzman, Sarah, and Geoffrey Gertz. 2020. “Facilitating Sustainable Investment: The Role and Limits of Investment Promotion Agencies.” In World Trade Forum 2018: Trade, Development, and Sustainability. Cambridge: Cambridge University Press. Blonigen, B. (2005). A review of the empirical literature on FDI determinants. Atlantic Economic Journal, 33(4), 383–403. Bobonis, Gustavo and Shatz, Howard, (2007), Agglomeration, Adjustment, and State Policies in the Location of Foreign Direct Investment in the United States, The Review of Economics and Statistics, 89, issue 1, p. 30-43 Charlton, Andrew, and Nicholas Davis. 2007. “Does Investment Promotion Work?” The B.E. Journal of Economic Analysis & Policy 7 (1): 1–21. Crescenzi, Riccardo & Di Cataldo, Marco & Giua, Mara. (2021). FDI inflows in Europe: Does investment promotion work?. Journal of International Economics. 132. 103497. 10.1016/j.jinteco.2021.103497. DCI (Development Counsellors International). 2017. “A View from Corporate America: Winning Strategies in Economic Development Marketing.” Survey report, DCI, New York. DIT (United Kingdom, Department for International Trade). 2018. “Estimating the Economic Impact of FDI to Support the Department for International Trade’s Promotion Strategy: Analytical Report.” DIT, London. DIT. 2019. “Department for International Trade Inward Investment Results 2018–19.” Statistical report, DIT, London. https://www .gov.uk/government/collections/dit-inward -investment-results. Echandi, Roberto, Maliszewska, Maryla, and Steenbergen, Victor (2022) Making the most of the African Continental Free Trade Area: Leveraging trade and foreign direct investment to boost growth and poverty reduction. World Bank: Washington DC ECORYS. 2013. “Exchange of Good Practice in Foreign Direct Investment Promotion: A Study Carried Out under the Framework Contract ENTR/2009/033.” Final report, 2nd rev., ECORYS, Rotterdam, the Netherlands. Egger, P., & Nigai, S. (2015). Structural gravity with dummies only: Constrained ANOVA-type estimation of gravity models. Journal of International Economics, 97(1), 86–99. Evans, Peter, Embedded Autonomy: States and Industrial Transformation, Princeton, NJ, Princeton University Press, 1995. Fernández, María Jesús; Blanco, Adrián; and Aranda-Larrey, Yago. 2021. National-Subnational Coordination for Investment Attraction: The Case of Spain. EFI Note-Finance. Washington, DC: World Bank Group. 25 Freund, Caroline, and Theodore H. Moran. 2017. “Multinational Investors as Export Superstars: How Emerging-Market Governments Can Reshape Comparative Advantage.” Working Pape r 17-1, Peterson Institute for International Economics, Washington, DC. Harding, Torfinn, and Beata S. Javorcik. 2011. “Roll Out the Red Carpet and They Will Come: Investment Promotion and FDI Inflows.” The Economic Journal 121 (557): 1445–76. Harding, Torfinn, and Beata S. Javorcik. 2012. “Investment Promotion and FDI Inflows: Quality Matters.” CESifo Economic Studies 59 (2): 337–59. Hayakawa, K., Lee, H. and Park, D., 2014. Are investment promotion agencies effective in promoting outward foreign direct investment? The cases of Japan and Korea. Asian Economic Journal, 28, 2. Head, K., & Mayer, T. (2014). Gravity equations: Workhorse, toolkit, and cookbook. In G. Gopinath, E. Helpman, & K. Rogoff (Eds.), Handbook of international economics (vol., 4, pp. 131 –195). Amsterdam, The Netherlands: Elsevier Head, K., Ries, J., Swenson, D., 1999. Attracting foreign manufacturing: investment promotion and agglomeration. Reg. Sci. Urban Econ. 29, 197–218 Heilbron, Armando and Hania Kronfol. “Increasing the Development I mpact of Investment Promotion Agencies.” (2020). Global Investment Competitiveness Report 2019/2020. Doi:10.1596/978-1-4648-1536-2. Heilbron, Armando, and Robert Whyte. 2019. “Institutions for Investment: Establishing a High -Performing Institutional Framework for Foreign Direct Investment (FDI),” Investment Climate In Focus Note, World Bank, Washington, DC. Kox, HLM, Rojas-Romagosa, H. How trade and investment agreements affect bilateral foreign direct investment: Results from a structural gravity model. World Econ. 2020; 43: 3203– 3242. https://doi.org/10.1111/twec.13002 Lim, Sung-Hoon. 2008. How investment promotion affects attracting foreign direct investment: Analytical argument and empirical analyses. International Business Review 17, 39-53. Lim, Sung-Hoon. 2018. “Determinants of the Performance of Investment Promotion Agencies: Evidence form a Mix of Emerging Economies.” Emerging Markets Finance and Trade 54 (8): 1907–23. Doi: 10.1080 /1540496X.2017.1334144. Loewendahl, Henry. 2001. “A Framework for FDI Promotion” Transnational Corporations 10 (1): 1–42 Miškinis, Algirdas, and Mariya Byrka. 2014. “The Role of Investment Promotion Agencies in Attracting Foreign Direct Investment.” Ekonomika 4 (93): 41–57. Morisset, Jacques, and Kelly Andrews-Johnson. 2004. “The Effectiveness of Promotion Agencies at Attracting Foreign Direct Investment.” Occasional Paper No. 16, Foreign Investment Advisory Service, World Bank Group, Washington, DC. Nelson, Roy C. 2009. Harnessing Globalization: The Promotion of Nontraditional Foreign Direct Investment in Latin America. University Park, PA: The Pennsylvania State University Press. OECD (Organisation for Economic Co-operation and Development). 2015. Strengthening Chile’s Investment Promotion Strategy, http://www.oecd.org/investment/investment-policy/chile-investment-promotion- strategy.htm. OECD (Organisation for Economic Co-operation and Development). 2018. “Mapping of Investment Promotion Agencies in OECD Countries.” Inventory and survey report, OECD, Paris. 26 Olivero, M., & Yotov, Y. V. (2012). Dynamic gravity: Endogenous country size and asset accumulation. Canadian Journal of Economics, 45(1), 64–92. Ortega, Celia, and Carlos Griffin. 2009. “Investment Promotion Essentials: What Sets the World’s Best Investment Facilitators Apart from the Rest?” Investment Climate In Practice Note No. 6, World Bank, Washington, DC. Phillips, Joe; Heilbron, Armando; and Kher, Priyanka. 2021. Lessons in Investment Promotion: The Case of Invest India. EFI Note. Washington, D.C.: World Bank Group. Qiang, Christine Zhenwei; Liu, Yan; Steenbergen, Victor. 2021. An Investment Perspective on Global Value Chains. Washington, DC: World Bank. © World Bank. https://openknowledge.worldbank.org/handle/10986/35526 License: CC BY 3.0 IGO. Rodrik, Dani, (2004), Industrial Policy for the Twenty-First Century, No 4767, CEPR Discussion Papers, C.E.P.R. Discussion Papers, https://EconPapers.repec.org/RePEc:cpr:ceprdp:4767. Santos Silva, J., & Tenreyro, S. (2006). The log of gravity. Review of Economics and Statistics, 88(4), 641 –658. UNCTAD (United Nations Conference on Trade and Development). 1997. “Survey of Best Practices in Investment Promotion.” Survey report, United Nations, New York and Geneva. Volpe Martincus, C., Carballo, J., Marra de Artiñano, I. and Blyde, J.S. (2020) How Effective is Investment Promotion? Firm-Level Evidence. IADB DISCUSSION PAPER N o IDB -DP -007 41 Volpe Martincus, Christian, and Sztajerowska, Monika. 2019. “How to Solve the Investment Promotion Puzzle: A Mapping of Investment Promotion Agencies in Latin American and the Caribbean and OECD Countries.” Monograph, Organisation for Economic Co-operation and Development and Inter-American Development Bank, Washington, DC. http://dx.doi.org/10.18235 /0001767. WAIPA (World Association of Investment Promotion Agencies). 2019. “Overview of Investment Promotion: Report of the Findings from the WAIPA Annual Survey of 2018.” WAIPA, Geneva. Wells, Louis T. Jr., and Alvin G. Wint. 2000. “Marketing a Country: Promotion as a Tool for Attracting Foreign Investment.” Revised ed. Occasional Paper No. 13, Foreign Investment Advisory Service, World Bank Group, Washington, DC. Whyte, Robert, Celia Ortega, and Carlos Griffin. 2011. “Investment Regulation and Promotion: Can They Coexist in One Body?” In Practice Note No. 16, World Bank, Washington, DC. Williamson, O.E., 1985. The Economic Institutions of Capitalism. The Free Press, New York. World Bank. 2018. Global Investment Competitiveness Report 2017/2018: Foreign Investor Perspectives and Policy Implications. Washington, DC: World Bank. Doi:10.1596/978-1-4648-1175-3. Yotov, Y. V., Piermartini, R., Monteiro, J.-A., & Larch, M. (2016). An advanced guide to trade policy analysis: The structural gravity model. Geneva, Switzerland: World Trade Organization 27 Annex Table A1. Share of global FDI stock in sectoral FDI database, average 2013-2018 Included in Sectoral Year FDI Database 2013 2014 2015 2016 2017 2018 Average Yes 67.7 65.7 65.3 64.7 63.5 62.1 64.7 No 32.3 34.3 34.7 35.3 36.5 37.9 35.3 Total 100 100 100 100 100 100 100 Source: Author’s calculations using WB Harmonized Bilateral FDI Database Table A2. Share of sample countries’ FDI stock in sectoral FDI database, average 2013-2018 Share of FDI Stock from countries Country ISO Country Name Covered in sectoral FDI Database Not covered in sectoral FDI Database AUS Australia 62.7 37.3 AUT Austria 81.1 18.9 BGR Bulgaria 67.2 32.8 BRA Brazil 80.7 19.3 CAN Canada 83.1 16.9 CHE Switzerland 91.9 8.1 CHL Chile 58.2 41.8 CYP Cyprus 59.8 40.2 CZE Czech Republic 83.0 17.0 DEU Germany 88.0 12.0 DNK Denmark 96.2 3.8 EGY Egypt, Arab Rep. 95.1 4.9 ESP Spain 89.2 10.8 EST Estonia 87.3 12.7 FIN Finland 94.7 5.3 FRA France 92.5 7.5 GBR United Kingdom 74.8 25.2 GRC Greece 89.6 10.4 HKG Hong Kong SAR, China 12.5 87.5 HUN Hungary 69.8 30.2 IND India 44.0 56.0 IRL Ireland 88.9 11.1 ISL Iceland 92.9 7.1 ITA Italy 95.8 4.2 KOR Korea, Rep. 54.4 45.6 NGA Nigeria 58.4 41.6 POL Poland 94.3 5.7 PRT Portugal 90.3 9.7 ROU Romania 81.2 18.8 RUS Russian Federation 75.9 24.1 SVK Slovak Republic 79.8 20.2 SVN Slovenia 68.3 31.7 TUR Türkiye 71.4 28.6 URY Uruguay 49.9 50.1 USA United States 67.9 32.1 ZAF South Africa 77.1 22.9 Total 70.3 29.7 Source: Author’s calculations using World Bank harmonized bilateral FDI database 28