Policy Research Working Paper 11025 Mapping Returns of Private Equity Investments in Emerging Markets Florian Mölders Edgar Salgado International Finance Corporation January 2025 Policy Research Working Paper 11025 Abstract This paper fills a gap in research on private equity invest- yielding extremely high returns alongside others gener- ments in emerging markets and developing economies. It ating little or no return, highlighting the importance of provides descriptive evidence and examines the distribution diversification. The analysis reveals that firm-specific fac- of returns across sectors such as finance, technology, and tors account for the largest share of return variability, with resource-intensive industries like mining, where significant country and sector factors playing a smaller role. Gross variation exists. Using data from the International Finance domestic product growth and real exchange depreciation Corporation, the analysis finds that return distributions are significantly related to returns, with median elasticities exhibit “fat tails”, with a notable presence of investments of 0.35 and −0.67, respectively. This paper is a product of the International Finance Corporation. It is part of a larger effort by the World Bank Group 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 authors may be contacted at fmoelders@ifc.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 Mapping Returns of Private Equity Investments in Emerging Markets* ¨ Florian Molders † Edgar Salgado‡ JEL Classification: F21, G24, G32, O16 Keywords: equity markets, private equity, private sector development, venture capital * We thank Kostas Kollias, Susan Lund, Paolo Mauro, Cesaire Meh, Tristan Reed, and the IFC Economics BBL Series participants for comments and suggestions. The authors would like to thank Soumya Agrawal for her invaluable assistance in ensuring data accuracy. The views expressed in this paper are those of the authors and do not necessarily represent those of the World Bank Group. All results have been reviewed to ensure that no confidential information is disclosed. † International Finance Corporation, email: fmoelders@ifc.org, corresponding author. ‡ International Finance Corporation, email: esalgadochavez@ifc.org 1 Introduction Emerging markets and developing economies (EMDEs) increasingly attract interest from private equity investors seeking high returns. However, data on returns to these investments remain un- derexplored, especially sector-specific performance, owing to data limitations. Using data from the International Finance Corporation (IFC) over the past 60 years, our paper is the first to present sectoral equity returns across EMDEs, including finance, technology, and resource-intensive in- dustries such as mining. Although previous research on EMDEs has predominantly focused on overall private equity performance, this study explores sectoral returns and examines key macroe- conomic drivers, such as economic growth and the real exchange rate. Since Kaplan and Schoar (2005), the private equity returns literature has commonly used the S&P 500 index as a benchmark against public markets, using metrics like the Public Market Equiv- alent (PME), for evaluating private equity performance. Building on Cole et al. (2024), this paper expands on that framework by exploring how private equity returns in EMDEs compare to pub- lic market indices. In this study, we apply the PME to compare performance against the MSCI Emerging Markets Index, which permits a comparison of returns on private equities versus pub- licly listed companies in the same country.1 Our exploration of returns across different sectors finds that some industries, such as tech- nology and finance, exhibited particularly strong performance during the sample period. These results add depth to our understanding of sectoral dynamics within EMDEs by providing a more granular view of private equity returns in these markets. Enhancing the transparency of private equity data in EMDEs also offers investors and policy makers a clearer understanding of equity returns across different sectors within these economies. We also examine the relative importance of firm-specific, sector, and country factors. Firm- specific factors account for the largest share of variation of returns, far surpassing the influences of country and sectoral aspects. Indeed, the role of idiosyncratic firm factors is greater than in studies for publicly listed equities. Using data from 12 European countries, Heston and Rouwen- horst (1994) find that country and industry characteristics explain 24% and 5%, respectively, of the variance of returns.2 Norges Bank (2019) built on the return decomposition framework by Heston and Rouwenhorst (1994) using an international dataset of firm-level public equity returns from developed and emerging markets, finding that country and industry factors explain about 30% of return variation, with firm-specific dynamics dominating. Using an ANOVA decomposition, our analysis extends this methodology to private equity in EMDEs. Our results reveal that sectoral effects account for 4.3% and country factors for 6.5%. The importance of firm-specific factors is consistent with studies that have examined the un- derlying determinants of profitability and performance. For example, Esho and Verhoef (2021) 1 Indeed,recent literature has emphasized the need for benchmarks with a closer risk profile to those in emerging markets (Jeffers et al., 2024). 2 This observation aligns with the findings of the volatility decomposition of Campbell et al. (2001). The authors highlight that between 1962 and 1997, the proportion of firm-specific volatility in total common stock return volatility rose from 65% to 76%, while the contributions of market volatility and industry volatility declined from 20% to 14% and from 15% to 10%, respectively. 2 and Rumelt (1991) highlight the dominance of firm-specific determinants over industry and coun- try effects in explaining profitability. Rumelt (1991) finds that business-unit effects explain a much larger share of profitability variance compared to industry effects. Hawawini et al. (2003) ex- tend this view by demonstrating that outlier firms—those with uniquely strong or poor perfor- mance—can disproportionately skew overall performance metrics, a factor particularly relevant in emerging market contexts where firm performance may vary dramatically within the same sector. This observation is supported by our data, where sectors such as finance exhibit a high number of ’home run’ investments or those that significantly outperform sectoral averages. Throughout this paper, we define home run investments as those that generate a return of at least 50% annually, on average, over the duration of the investment. Even so, macroeconomic variables play a role. In particular, GDP growth and real exchange rate depreciation significantly influence returns. Our study extends the analysis of Cole et al. (2024) and highlights that the variance in returns is notably influenced by real exchange rate (RER) movements, which significantly impact dollar-denominated returns, consistent with earlier work by Solnik (1974) and Phylaktis and Ravazzolo (2005) who demonstrate that exchange rate volatil- ity plays a significant role in dollar-denominated returns. This is especially pertinent in EMDEs, where currency stability varies considerably. Thus, although firm-specific factors are dominant, sectoral and country/macroeconomic factors remain relevant. 2 Data This paper analyzes the cash flows associated with every equity investment made by the Interna- tional Finance Corporation (IFC), a member of the World Bank Group. The IFC aims to improve social and environmental outcomes that are aligned with the United Nations’ Sustainable De- velopment Goals. Historically, before these goals were formally established, the IFC focused on providing capital to markets where it was scarce. In 1961, the institution’s operations changed when a charter amendment allowed it to hold equity. This triggered a sharp rise in equity invest- ment to approximately 50 percent of total investment by 1963-64 (Kapur et al., 1997). Since then, the equity share has fluctuated between 15 and 35 percent. Our primary dataset comprises the complete series of cash flows associated with all 2,727 equity investments made by the IFC, from the inception of the first equity investment in 1961 to June 30, 2023 (end of fiscal year 2023).3 Negative cash flows refer to contributions that increase the value of the investment in an equity position. Conversely, distributions are cash flows returned to the investor (IFC) through dividend payments or the sale of a company’s shares. For investments still in the portfolio, our analysis treats the net asset value as of June 30, 2023, as though it was a positive distribution (the residual). We measure returns using the annualized cash multiple, which considers the number of years an investment has been held in the portfolio. Additionally, we employ the PME as this metric accounts for the absolute level of returns and their correlation with a global risk factor, consistent with the principles of the capital asset pricing model. 3 We exclude any loans converted into equity from the analysis. 3 The dataset includes detailed information on the exact month and value of each cash flow in U.S. dollars and the most recent mark-to-market valuation of investments that remain in the portfolio. This information allows for precise tracking of the performance of each investment over time. Additionally, each investment is assigned a ”vintage year”, defined as the year when the first cash flow was made to the company or fund, providing a clear timeline for the start of each investment. This enables robust investment performance analysis across different periods. Our dataset encompasses investments in funds and individual companies. We employ IFC’s internal sector classification, resulting in 22 sectors.4 The “Finance and Insurance” sector has concentrated close to 30 percent of the overall IFC portfolio and has the largest share in the historical IFC portfolio. Next come “Collective Invest- ment Vehicles” or Funds with 27 percent, followed by “Oil, Gas & Mining” with 7 percent (Figure 1).5 Collective investment vehicles are pooled investments that gather capital from multiple in- vestors to invest in a diversified portfolio of assets. These vehicles enable IFC and other investors to access a range of investments, including venture capital and growth equity funds. By pooling capital, funds offer investors the benefits of diversification, professional management, and shared risk. Our data, however, do not include information about the specific sub-sectors that funds invest in. The prominence of Finance and Insurance is recent, even though IFC’s portfolio has been heav- ily concentrated in 5 sectors from the start. The composition of this concentration, however, has changed over time. From 1970 to the early 1990s, IFC invested heavily in equity in Manufacturing. From the second half of the 1990s, the Finance sector and Utilities rose to the top of IFC’s equity portfolio. 3 Measuring equity returns This section presents performance metrics for investment sectors in EMDEs viewed through the lens of a well-diversified international investor. We utilize metrics that focus on public market comparisons and take into account annualized returns. Public Market Equivalent (PME): The PME is a financial metric used to evaluate the perfor- mance of private equity investments by comparing them to the returns of public market indices. 4 Thesesectors are the following: Accommodation & Tourism Services, Agriculture and Forestry, Chem- icals, Collective Investment Vehicles, Construction and Real Estate, Education Services, Electric Power, Finance & Insurance, Food & Beverages, Health Care, Industrial & Consumer Products, Information, Non- metallic Mineral Product Manufacturing, Oil, Gas and Mining, Plastics & Rubber, Primary Metals, Pro- fessional, Scientific and Technical Services, Pulp & Paper, Textiles, Apparel & Leather, Transportation and Warehousing, Utilities, Wholesale and Retail Trade. 5 As a result of its focus on accelerating an equitable and just energy transition, the World Bank Group has not financed any new coal-fired power project since 2010, and no oil and gas projects since 2019 that involve exploration, drilling, and extraction (upstream). The World Bank has not financed any oil pipelines since 2014. 4 Figure 1: Share of IFC Equity Investment of Top 5 Sectors (1961-2023) Note: The share represents the proportion of total IFC investments accounted for by each sector. It shows how much each sector contributed relative to the entire IFC investment portfolio over the entire period. Nominal contributions have been deflated to constant USD (2010) to compute the shares. In doing so, it answers the question: ”How would the returns from private equity investments6 compare if the same amount of money were invested in the public markets over the same period?” The PME allows investors to evaluate whether higher returns justify the risks and illiquidity as- sociated with private equity compared to the more liquid and often less risky public markets. The idea of PME involves replicating the cash flows of a private equity investment with an equivalent investment in a public market index. This means tracking the timing and amount of cash flows (investments and returns) and applying them to the public market index, that is, investing in the public market index whenever a cash contribution to the private equity fund is made and selling (or withdrawing) from the index whenever the private equity fund distributes cash back. The PME is computed as: dist(t) ∑t 1+ R ( t ) PME = cont(t) (1) ∑ t 1+ R ( t ) R(t) is the realized total return of the market index (for example, MSCI EM or S&P 500) from the year of the first cash flow (t = 0) to the time of the distribution or contribution (t). If the ratio exceeds one, the investor prefers the private equity investment to the public index. 6 This includes investments through funds, such as growth equity, infrastructure, or venture capital funds. 5 When using the PME, our study is focused on the MSCI EM Index7 as a benchmark because it more accurately reflects the countries where IFC invests and provides a more relevant point of comparison. When available, country-specific MSCI indices are used to estimate PME returns. For investments in emerging markets without a country-specific MSCI index, the MSCI EM index is used. Total Value to Paid-In Capital (TVPI): The TVPI is a performance metric used in private eq- uity to assess the value created by an investment relative to the capital invested. It measures the total value generated by a private equity investment to the capital that has been invested (paid- in capital). It helps investors understand the overall return on their investment, combining both realized returns (cash distributions) and unrealized returns (the current value of remaining invest- ments). TVPI reflects the total return on investment. It considers both the interim cash flows (dis- tributions) and the remaining value of the asset (the residual), providing a holistic view of the investment’s performance. The TVPI is computed as: Cumulative Distributions + Residual Value TVPI = (2) Cumulative Contributions Anything above 1 means an investment grew in nominal value. Anything below 1 means the investment shrank in nominal value. In addition to the TVPI, we calculate an annualized version of this metric: 1 Cumulative Distributions + Residual Value years TVPI A = −1 (3) Cumulative Contributions For ease of interpretation, we subtract 1 so that annualized returns above (below) zero indicate growth (or loss) in nominal value. “Years” denotes the number of years that any equity investment has been in IFC’s portfolio. Annualizing the TVPI allows us to discount investments by their respective duration, changing the distribution of this series compared to the standard TVPI. The frequency chart of equity returns in EMDEs (Figure 2), as measured by the PME over the (country-specific) total return MSCI EM index, exhibits a highly skewed distribution, which indicates significant deviations from a normal distribution. The mean return of 1.16 suggests that, on average, equity investments in these markets have yielded positive returns 16% higher than the public benchmark over the entire duration of the investments8 . However, this average is heavily influenced by extremely high values, as evidenced by the substantial skewness of 46.8. The returns to private equity are highly concentrated at their tails, indicating many investments with low returns and some “home runs” (positive outliers that will be further discussed in the subsequent section). The high level of skewness is also found in the context of a developed market in Korteweg (2023), especially in the context of returns to VC funds. 7 Source: https://www.msci.com/documents/10199/c0db0a48-01f2-4ba9-ad01-226fd5678111. Figure B.1 in the appendix plots the time series of the Index. 8 Using the country-specific PME price (instead of total return) index yields a mean return of 1.27. 6 The median return of 0.9, considerably lower than the mean, reflects a more typical return experience for most investors, suggesting that the distribution is highly asymmetric. This indicates that while a few extraordinarily high returns skew the average upwards, the median highlights the prevalence of more modest outcomes. This is also driven by a high number of very low returns, such as 278 investments with a PME of 0.05 or below, which includes investments that have been written off. The skewness confirms the presence of a long right tail in the distribution, driven by occasional extreme positive returns that are uncommon across the sample. This, however, underscores the substantial upside potential of EMDE equities and highlights the rarity of such outsized gains. Figure 2: Frequency plot of IFC equity returns as measured using the PME vs MSCIEM, 1990-2023 Note: Investments with a PME equal to or greater than 5 are given a PME of 5. Investments with vintage years from earlier decades tend to exhibit higher annualized returns, as measured using the annualized TVPI (Figure 3). The historical median annualized return is 2.2%, while the median for the recent period has dropped to 0.3%. Investment-weighted mean historical returns stand at 6%.9 The investment-weighted mean return has also declined in recent 9 Annualized 1 TVPI is computed by taking the investment-weighted TVPI ratio to the power of years , where “years” is the investment-weighted average number of years the typical IFC investment lasts. Using the internal rate of return (IRR) as an alternative return metric, IFC’s portfolio generated 13.2% annually between 1961 and 2023. In comparison, the S&P500 and MSCI EM indices generated a CAGR of 10.4% and 9.5%, respectively, between 1961 and 2023 (S&P500) and 1988-2023 (MSCI EM). 7 years (2010-2023) to 2.6%. Despite this downward trend, the portfolio still produces occasional high-return investments, or ”home runs” (investments for which we calculate an annualized re- turn equal to or larger than 50%). Notably, 2.4% of all investments made from 1961 to 2023 were home runs, with 2.1% from 2010 to 2023 achieving similar standout performance.10 Figure 3: Frequency plot of IFC equity returns as measured using the annualized TVPI, 1961-2023 vs 2010-23 Note: Investments with an annualized TVPI equal to or greater than 100% are given a return of 100%. IFC’s equity investments are structured through direct investments in individual companies or commitments to private equity funds. Investments in private equity funds generally provide diversification across a portfolio of companies, which tends to smooth out volatility and reduce the likelihood of extreme value returns. By pooling capital across multiple investments, funds can mitigate the impact of poor performance from individual companies, making fund investments less susceptible to large fluctuations in value. This lower volatility offers more stable and less extreme returns over time. In contrast, direct investments in companies carry higher levels of idiosyncratic risk, although these are diversified in the overall portfolio. This heightened risk increases the likelihood of sub- stantial home runs—extraordinary returns from companies that experience significant growth or market success— or the possibility of major losses when individual companies underperform or fail. ”Home run” scenarios are often driven by the success of firms operating in high-growth 10 Aswith Figure 2, we observe a high number of investments that have resulted in losses of 100% (67 for the period between 1961-2023). 8 industries or emerging technologies with higher potential for exceptional returns. However, high- growth opportunities also come with inherent uncertainty, making the return profile of direct investments more uneven. Between 1961 and 2023, the mean and median returns for direct investments and funds dis- played little variation over time. However, the distribution of returns differed markedly: funds exhibited a narrower range, characterized by a lower incidence of extreme values at both ends of the spectrum (Figure 4), compared with direct investments. Figure 4: Frequency plot of IFC equity returns, 1961-2023, direct vs indirect (Funds) in- vestments Note: Investments with an annualized TVPI equal to or greater than 100% are given a return value of 100%. Between 2010 and 2023, funds have shown stable performance, compared with previous decades, whereas direct investments have underperformed (Figure 5). Within funds, the median returns from 2010-2023 are statistically unchanged (1%) from 1961-2023. However, for direct investments, median returns in recent years have dropped significantly compared to historical levels, declining from 3.7% in 1961-2023 to 0% in 2010-23.11 The overall recent underperformance of IFC’s equity portfolio is primarily attributed to the weaker returns in direct investments. 11 Using the IRR as an alternative metric, IFC’s direct investment portfolio achieved a return of 13.5% between 1961 and 2023, whereas its fund investment portfolio delivered a return of 8% over the same period. 9 Figure 5: Frequency plot of IFC equity returns, direct vs indirect (Funds) investments, 1961-2023 vs 2010-23 Note: Investments with an annualized TVPI equal to or greater than 100% are given a return value of 100%. The period from the end of the 1990s to the early 2000s was the most profitable period for IFC. Investments made since the 2010s have been less profitable. Figure 6 further elaborates on this by showing the 10-year returns of the portfolio initiated in each respective vintage year (noting that the median holding period of IFC equity investments is around 9 years), with the x-axis representing the starting year of an investment. The performance metrics reflect the cumulative effect of investments that have been fully realized and those that are still active within the 10-year window. The metrics provide insight into the ”mature” phase of private equity investments since most investments in the portfolio would have had enough time to develop and produce returns. 10 Figure 6: PME (RHS) & annualized TVPI (LHS) of investments held up to 10 years Note: The blue horizontal line references the PME=1, with values below the line indicating underperformance versus the public index. The red line similarly depicts the profitability threshold for the annualized TVPI. Years on the horizontal axis refer to the investment’s vintage year. Performance measures are calculated based on the vintage year of the invest- ments. For instance, the last year with a value (2013) represents the return measure(s) for all investments initiated in 2013, reflecting a 10-year investment period. 4 Sectoral equity returns in EMDEs The sectoral performance (measured by annualized TVPI in percent) displays a high level of het- erogeneity, as reflected by the sectors’ medians and 25th and 75th percentiles (Figure 7). The highest median returns throughout the investment history are found in the Electric Power and Plastics & Rubber sectors. However, both sectors have average returns below their respective me- dians, indicating that the distribution of returns is skewed to the left. Most of these returns are higher than the average. Still, some relatively low outliers or very poor-performing investments drag the average down.12 A more detailed depiction of industry differences in IFC’s equity returns is presented in Table 1. Unlike Figure 7, where metrics are calculated at the individual investment level, the table’s figures are computed at the sector-portfolio level from 1961 to 2023. This means that the TVPI’s contributions and distributions are summed across all investments within each respective sector. The annualized TVPI is calculated based on the average duration of each sector’s investments. The table also highlights the total number of home runs (an annualized TVPI of 50 percent or 12 See Figure B.2 in the appendix for sectoral return distributions. 11 greater) and the total number of equity investments in each sector over the entire sample period. Finance & Insurance, one of the largest sectors in overall contributions by IFC, has the most home runs, skewing the distribution of its returns to the right. Returns to investments in funds are more evenly distributed than in most other sectors, but even here, home runs play a significant role in boosting average performance. Figure 7: Median and quartiles of returns by sector, 1961-2023 Note: Figure depicts median, 25th, and 75th percentile of annualized returns, by sector, according to IFC sector classifica- tion. Information and Oil, Gas & Mining are among the sectors with the highest annualized returns, with 21 percent and 13 percent, respectively. These sectors are also more dependent on home runs for their returns. While the Wholesale and Retail Trade sector is the second most profitable, it does not have as many home runs as other sectors, and its overall contribution to IFC’s equity portfolio is relatively small.13 In contrast, Funds and Finance & Insurance exhibit many home runs, largely due to their significant weight in the overall portfolio. 13 Itis important to note the impact of disruptive technology on equity returns. The performance in the Wholesale and Retail sector demonstrates this trend. A significant portion of the capital invested in this sector has been directed toward venture capital, resulting in higher returns than traditional equity investments in that sector. 12 Table 1: Returns, duration, home runs, investments by sector, 1961-2023 Table includes two metrics to assess investment performance: the Total Value to Paid-In (TVPI) ratio and its annualized counterpart (TVPI A ). TVPI measures the overall return on investment, calculated as the sum of cumulative distributions and the residual value (the current value of the remaining investment) divided by the cumulative contributions (the total capital invested). A TVPI greater than 1 indicates that the investment has grown in nominal value, while a value less than 1 implies a nominal loss. To provide a time-adjusted perspective, the annualized TVPI (TVPI A ) adjusts the TVPI to reflect an average annualized growth rate 1 over the life of the investment. It is computed by taking the TVPI ratio to the power of years , where ”years” is the duration of the investment in years, and then subtracting 1. This metric enables comparisons across in- vestments of different durations by standardizing returns on an annual basis. Sector names are according to IFC sector classification. Home runs are investments with an annualized TVPI greater than or equal to 50%. Sector TVPI (annualized) TVPI Average years Home-runs Projects Construction and Real Estate -3.13% 0.81 6.69 0 29 Professional, Scientific and Technical Services -0.69% 0.96 6.64 0 65 Textiles, Apparel & Leather -0.33% 0.96 12.52 0 71 Accommodation & Tourism Services -0.06% 0.99 11.10 0 63 Education Services 0.39% 1.02 6.00 2 30 Pulp & Paper 0.58% 1.07 11.36 1 43 Primary Metals 2.74% 1.36 11.32 1 39 Transportation and Warehousing 2.95% 1.26 8.04 1 83 Collective Investment Vehicles 3.51% 1.34 8.50 8 694 Electric Power 5.05% 1.53 8.61 1 89 Chemicals 5.12% 1.62 9.70 3 107 Utilities 5.54% 1.48 7.28 0 31 Health Care 5.84% 1.51 7.25 0 60 Nonmetallic Mineral Product Manufacturing 5.86% 1.86 10.90 2 89 Plastics & Rubber 6.97% 1.90 9.52 0 17 Industrial & Consumer Products 7.20% 1.77 8.26 1 111 Finance & Insurance 7.20% 1.85 8.87 15 654 Agriculture and Forestry 7.75% 1.78 7.75 5 84 Food & Beverages 8.60% 1.97 8.22 2 102 Oil, Gas and Mining 13.39% 2.80 8.18 6 162 Wholesale and Retail Trade 16.56% 2.11 4.89 1 32 Information 21.29% 3.12 5.90 13 72 The results suggest that private equity returns are superior in sectors characterized by high growth potential or resource availability, partially confirming the findings by Bai et al. (2012).14 Sectors like mining and information technology benefit from structural growth drivers in EMDEs, such as urbanization, digitalization, and resource exploitation. These factors create opportunities for private equity investors to generate outsized returns by capitalizing on inefficiencies or unmet 14 Theauthors use data on 1,537 individual firms from selected emerging markets, consisting of monthly adjusted prices of ordinary shares for the period from August 1984 to July 2004, which includes the 1997 Asian financial crisis. They find superior returns in listed companies in sectors such as Technology and Financials. 13 demand in these markets. On the other hand, sectors that underperform, such as professional services and accommodation, may reflect more mature industries with higher levels of competi- tion and less differentiated investment opportunities. In these sectors, public market investments might offer more stability and comparable or better returns, reducing the relative advantage of private equity. 5 Econometric analysis What factors underlie the variation in investment returns? This analysis involves ANOVA (vari- ance decomposition) of the annualized TVPI, the probability of achieving home runs, and the annualized TVPI calculated without including home runs or extreme losses15 (Figure 8). The re- sults are qualitatively similar and suggest that observable investment-specific characteristics like investment size and holding period (duration) explain the largest share of the explained returns variation (between 22% and 29%).16 Unobservable investment-specific characteristics explain be- tween 54% and 60% of the returns’ variance. Country characteristics explain investment returns to a lesser degree (between 6.1% and 7.5%), while industry-specific factors reflect an even lower ex- planatory power of return variance. This is consistent with earlier results by Heston and Rouwen- horst (1995), who found the absolute value of country effects was twice as large as the value of the industry effects in their analysis of stock returns.17 According to our findings, firm-specific factors matter more for private equity investments than those in listed companies.18 Listed stocks are less likely to experience home runs or bankrupt- cies. In the context of IFC returns, firm-specific characteristics account for about 83% of the vari- ance. This aligns with findings from Heston and Rouwenhorst (1995), who attribute 71% of the variance in returns across 12 European countries to firm-specific factors, and Bekaert et al. (2009), who highlight the influence of industry factors during sector-specific volatility but emphasize the continued importance of country and firm-specific factors in driving international stock returns. Notably, a substantial portion of equity return variation remains unexplained, underscoring the critical role of firm-specific (or investment-specific) factors. Moving to a more general analysis of the determinants of returns, including macroeconomic variables, we employ quantile regressions. This method is particularly suited to our analysis due 15 I.e., excluding annualized TVPI smaller than -50% or greater than 50%. 16 For ANOVA purposes, size at inception and duration are split into 20 equally distributed groups. Their interaction is then included in the ANOVA estimation. 17 The authors note that the variance of the country effect is larger than that of the industry; the average correlation between equities within a country is higher than among firms within the same industry. 18 Note that our analysis assumes static country and sector factors. However, some sectors may have been very profitable during one period but were outdated and less profitable in another. Cat˜ ao and Tim- mermann (2010) leverage a multi-factor model that incorporates static and time-varying components to capture how the importance of country-specific and industry-specific factors evolves over time. Using rolling regressions, the authors estimate how the influence of country and industry factors fluctuates, iden- tifying periods when one factor dominates. Their findings suggest that the common country factor is the most persistent compared to industry factors, even though the latter has increased since the late 1990s. 14 Figure 8: ANOVA Decomposition for Measures of Returns (1988-2023) Note: Regional investments are excluded, resulting in a lower number of investments compared to the overall total of 2,727. Column (1) is based on the total sample of investments; column (2) estimates the probability of achieving a home run investment; column (3) includes only investments that are neither home runs nor the lowest-performing. Sector names follow the IFC sector classification. Home runs are defined as investments with an annualized TVPI of 50% or higher. The ANOVA decomposition for home runs provides a linear approximation to evaluate the relative contributions of factors such as country, sector, year, size, and duration to investment outcomes. Due to the large number of categories within these factors, logistic or probit models could not be reliably estimated. to the presence of extreme values and skewed distributions in investment-level returns, which could otherwise distort results obtained from mean-based methods like ordinary least squares (OLS) regression. Unlike OLS, which estimates the mean relationship between independent and dependent variables, quantile regression models relationships at different points of the conditional distribution, enabling us to capture heterogeneous effects across the return spectrum. This is especially valuable given that returns often exhibit heavy tails, where extreme values may affect traditional regression results. Consistent with the methodology outlined in Cole et al. (2024), this approach allows for a robust examination of median and tail behavior. We also include alternative estimations in the appendix (Table B1), which yield results similar to those obtained through quantile regression. We exploit differences in investment timing to estimate the regression between investment returns and macroeconomic variables: yi,t,t−k = α + β 1 GDP growthc,t,t−k + β 2 Ec,t,t−k + β 3 In f lationc,t,t−k + ε i,t,t−k (4) 15 Where yi,t,t−k are the annualized returns of investment i over years t and t-k.19 Returns are measured as annualized TVPI or PME-MSCI-adjusted returns. The analysis period is from 1990 to 2023 to match the MSCI series. Returns are regressed on real GDP growth, nominal exchange depreciation (E), and inflation in country c where the investment occurred over period k. These variables are expressed as changes over the investment cycle using the natural log of the ratio between end-of-investment and start-of-investment values.20 We next focus on the relationship between equity returns and currency depreciation, a key factor in driving USD returns for international investors. An alternative estimation considers Real Exchange Rate (RER) annual depreciation instead of nominal exchange rate depreciation and inflation. Whereas nominal exchange rate and inflation capture direct and immediate changes in currency value and price levels, they do not account for the relative inflation rates between the investee country and the US (considering that IFC is a USD investor). RER, on the other hand, also considers changes in relative price levels. yi,t,t−k = α + γ1 GDP growthc,t,t−k + γ2 RERc,t,t−k + ε i,t,t−k (5) Panel a of Table 2 shows the correlations between equity returns measured as annual TVPI and macroeconomic variables.21 Column (1) indicates that the semi-elasticity of median returns to GDP growth, estimated using quantile regression at the median, is 0.47. This implies that a 1 percent increase in GDP over the time of an investment is associated with an increase in median returns of 0.47 percentage points. Columns (2) and (4) indicate that nominal exchange rate depre- ciation is negatively associated with returns, while inflation is positively correlated. Specifically, for every 10% nominal exchange rate depreciation, median returns decrease by 7.05 percentage points, whereas a 10% increase in inflation is linked to a 6.76 percentage point increase in median returns. As nominal exchange rate depreciation and high inflation often occur together (and eq- uities represent a claim on real assets), their significance with opposite signs is not unexpected. Column (3) shows that real exchange rate depreciation is also negatively associated with returns, with a 10% RER depreciation resulting in a 6.71 percentage point decrease in median returns. 19 The median investment duration is around 9 years. 20 Lack of country-level data for some periods excluded 142 investments, mostly in Argentina, the Russian ´ Federation, the Republica Bolivariana de Venezuela, and Zimbabwe. 21 An alternative estimation that excludes home runs yields qualitatively similar results, not shown here but available upon request. 16 Table 2: Correlates of Equity Returns (1990-2023) Panels (a) and (b) present results from quantile regressions. Regressions in columns (1)–(5) are estimated at the median (50th percentile) of the return distribution, while columns (6)–(9) extend the analysis to other quantiles based on the specification used in column (5). Panel a uses annualized TVPI as a dependent vari- able, while panel b uses returns adjusted by PME-MSCI. All macroeconomic variables are annual changes over the duration of the investment cycle. Size is the initial contribution, and duration is the investment holding period measured in years. Bootstrapped errors in parenthesis. * 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) (6) (7) (8) (9) Quantile Regression 0.5 0.5 0.5 0.5 0.5 0.1 0.25 0.75 0.9 Panel a. Dependent Variable: Annual TVPI GDP growth 0.465*** 0.259** 0.349*** 0.194 0.284 0.465*** 0.521 (0.108) (0.123) (0.113) (0.611) (0.320) (0.151) (0.374) E -0.713*** -0.705*** (0.062) (0.066) Inflation 0.685*** 0.676*** (0.064) (0.067) RER -0.671*** -0.678*** -0.242 -1.092*** -0.880*** -0.923*** (0.061) (0.063) (0.341) (0.179) (0.084) (0.209) Ln(Size) 0.003* 0.007*** 0.006*** 0.006*** 0.005*** 0.026*** 0.007 0.005** -0.012** (0.002) (0.002) (0.002) (0.002) (0.002) (0.010) (0.005) (0.002) (0.006) Ln(Duration) 0.014*** 0.017*** 0.003 0.020*** 0.008* 0.300*** 0.085*** -0.028*** -0.123*** (0.005) (0.005) (0.005) (0.005) (0.005) (0.027) (0.014) (0.007) (0.016) R 2 0.005 0.019 0.018 0.020 0.020 0.117 0.036 0.033 0.093 Observations 1668 1668 1668 1668 1668 1668 1668 1668 1668 Panel b. Dependent Variable: PME-MSCI GDP growth 0.562 0.321 0.524 0.228 1.362* 1.871 3.667 (0.729) (0.752) (0.737) (0.359) (0.781) (1.589) (4.055) E -0.222 -0.254 (0.384) (0.404) Inflation -0.025 0.048 (0.395) (0.408) RER -0.097 -0.167 0.043 -0.720* -0.079 1.196 (0.392) (0.411) (0.200) (0.436) (0.887) (2.263) Ln(Size) 0.029** 0.031*** 0.031*** 0.029** 0.029** 0.008 0.026** -0.018 -0.150** (0.012) (0.012) (0.011) (0.012) (0.012) (0.006) (0.012) (0.025) (0.064) Ln(Duration) -0.013 -0.025 -0.019 -0.013 -0.012 0.025 -0.053 0.263*** 0.654*** (0.032) (0.031) (0.030) (0.032) (0.032) (0.016) (0.034) (0.069) (0.177) R2 0.003 0.004 0.003 0.004 0.003 0.003 0.005 0.014 0.050 Observations 1668 1668 1668 1668 1668 1668 1668 1668 1668 Columns (5) - (9) show that real exchange rate depreciation is negatively associated with equity 17 returns throughout all quantiles. The size of an investment at inception has a positive, albeit small, elasticity on annual TVPI, except in the regression for the 90th percentile (column 9), where returns at the top end of the distribution are negatively associated with size at inception. Duration is also positively associated with returns, except for the top returns, where this association becomes negative. The highest returns are linked to smaller, short-lived investments. Results from Panel B, where PME-MSCI is used as the measure of equity returns, reveal almost no statistically significant association between returns adjusted by PME-MSCI and the macroeco- nomic variables. This lack of significance suggests that the macroeconomic factors influencing annual returns do not affect IFC investments differently than their market equivalents. In other words, while macroeconomic variables such as GDP growth, exchange rate changes, or infla- tion may drive fluctuations in absolute returns, IFC’s relative performance—measured against the PME-MSCI benchmark—remains unaffected by these factors. This finding indicates that IFC investments behave in line with the broader market in terms of their sensitivity to macroeconomic conditions. IFC’s ability to mitigate risks or leverage opportunities presented by macroeconomic factors appears comparable to what is observed in the market. The coefficient on duration in column 9 indicates that IFC returns with longer duration will outperform the MSCI market equiv- alent. Table B1 in the appendix presents three alternative estimations: OLS, RREG, and winsorized. OLS is the standard OLS estimation, RREG is the robust regression that screens for outliers and then performs Huber iterations followed by biweight iterations, as suggested by Li (2006). Win- sorized estimation involves capping values of TVPI and PME-MSCI at the 10th and 90th per- centiles before estimating OLS. Instead of removing outliers altogether, winsorization replaces values above the upper percentile threshold with the value at that threshold and values below the lower percentile threshold with the value at that threshold. Results are qualitatively similar to those reported in table 2. Table 3 explores whether there are differences in these associations by type of investment vehi- cle. The positive association with GDP growth starts from the median return of direct investments. The coefficients for the interaction with funds are insignificant. Conversely, the negative associa- tion with RER depreciation occurs mainly in direct investments. Table B2 in the appendix suggests that macroeconomic variables have no heterogeneous association by investment vehicle with re- turns adjusted by PME-MSCI. Only duration and size at inception are positively and negatively associated with PME-adjusted returns in the 90th percentile. 18 Table 3: Correlates of Annual Equity Returns by Investment Vehicle (1990-2023) Dependent variable is annualized TVPI. All macroeconomic variables are annual changes over the duration of the investment cycle. Size is the initial contribution, and duration is the investment holding period mea- sured in years. The base category for the interaction is direct investment. The ”Funds” dummy indicates the investment was in a fund (“collective vehicle” in IFC nomenclature). Bootstrapped errors in parenthesis. * 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) Quantile Regression 0.1 0.25 0.5 0.75 0.9 GDP growth 0.479 0.328 0.434*** 0.352* 0.674* (0.696) (0.372) (0.131) (0.195) (0.403) RER -0.568 -1.035*** -0.808*** -0.973*** -0.893*** (0.385) (0.206) (0.073) (0.108) (0.223) Ln(Size) 0.051*** 0.019*** 0.005** 0.001 -0.020*** (0.012) (0.006) (0.002) (0.003) (0.007) Ln(Duration) 0.304*** 0.133*** 0.009 -0.036*** -0.143*** (0.031) (0.017) (0.006) (0.009) (0.018) Funds × GDP growth 0.257 -0.125 -0.333 0.092 -0.210 (1.480) (0.791) (0.279) (0.415) (0.857) Funds × RER -0.127 0.681 0.662*** 0.630*** 0.495 (0.861) (0.460) (0.162) (0.241) (0.498) Funds × Ln(Size) -0.041* -0.013 0.002 0.006 0.012 (0.025) (0.013) (0.005) (0.007) (0.014) Funds × Ln(Duration) -0.263*** -0.117*** -0.002 0.027 0.088** (0.060) (0.032) (0.011) (0.017) (0.035) R2 0.171 0.061 0.023 0.038 0.109 Observations 1668 1668 1668 1668 1668 Table 4 investigates whether these associations between macroeconomic variables and returns are heterogeneous by sector. IFC sectors are grouped into three main categories. FIG: collective investment vehicles and finance and insurance. MAS: Manufacturing, Agribusiness and Services. INFRA: electric power; information; oil, gas and mining; primary metals; transportation and ware- housing; and utilities. FIG is the base category in the results presented in table 4. For this sector, annual returns are positively associated with GDP growth, especially in the higher percentiles: results are statistically significant for the 75th and 90th percentiles of annualized returns. RER de- preciation is negatively associated with annual returns in the FIG sector. For MAS, GDP growth is not associated with higher annualized returns except for returns in the 90th percentile, where this association is negative. RER, on the other hand, is not associated with returns for sectors grouped under MAS. Even though returns in the INFRA sector are not statistically associated with GDP growth, RER depreciation is associated with lower returns in the INFRA sector. This is likely due to the local currency generation in the associated sub-sectors. Any real depreciation of the local currency then translates into lower USD returns. Table B3 in the appendix suggests that 19 macroeconomic variables have no heterogeneous association by sector with returns adjusted by PME-MSCI. Table 4: Correlates of Annual Equity Returns by IFC Sector (1990-2023) Dependent variable is annualized TVPI. All macroeconomic variables are annual changes over the duration of the investment cycle. Size is the initial contribution, and duration is the investment holding period measured in years. The base category for the interaction is the FIG sector: collective investment vehicles (Funds) and finance and insurance. MAS: Manufacturing, Agribusiness and Services. INFRA: electric power; information; oil, gas and mining; primary metals; transportation and warehousing; and utilities. Bootstrapped errors in parenthesis. * 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) Quantile Regression 0.1 0.25 0.5 0.75 0.9 GDP growth 0.544 0.514 0.311 0.646*** 1.098* (0.764) (0.433) (0.192) (0.197) (0.577) RER -1.155*** -0.594** -0.512*** -0.908*** -0.811*** (0.413) (0.234) (0.104) (0.106) (0.312) Ln(Size) 0.011 0.007 0.007** 0.007** -0.004 (0.012) (0.007) (0.003) (0.003) (0.009) Ln(Duration) 0.112*** 0.031* 0.009 -0.027*** -0.097*** (0.032) (0.018) (0.008) (0.008) (0.025) MAS × GDP growth 0.754 0.419 0.025 -0.477 -1.546* (1.179) (0.668) (0.297) (0.304) (0.890) MAS × RER 1.272* 0.422 -0.224 0.454** 0.547 (0.720) (0.408) (0.181) (0.185) (0.543) MAS × Ln(Size) 0.085*** 0.034*** -0.000 -0.005 -0.023 (0.021) (0.012) (0.005) (0.005) (0.016) MAS × Ln(Duration) 0.273*** 0.155*** -0.008 0.002 -0.044 (0.055) (0.031) (0.014) (0.014) (0.041) INFRA × GDP growth -0.544 -0.286 -0.641 -0.663 -1.271 (1.728) (0.979) (0.435) (0.445) (1.304) INFRA × RER 1.621** -0.449 -1.283*** -0.662*** -0.980 (0.796) (0.451) (0.200) (0.205) (0.601) INFRA × Ln(Size) 0.040* 0.033** 0.003 -0.017*** -0.061*** (0.023) (0.013) (0.006) (0.006) (0.018) INFRA × Ln(Duration) 0.149** 0.193*** 0.025 -0.032* -0.078 (0.065) (0.037) (0.016) (0.017) (0.049) R2 0.194 0.082 0.025 0.045 0.118 Observations 1668 1668 1668 1668 1668 20 6 Conclusion The scarcity of empirical evidence on sectoral private equity returns in EMDEs has left critical questions unanswered for international investors seeking to diversify their portfolios and assess risks effectively. This study comprehensively examines sectoral private equity returns in EMDEs using an extensive dataset from the International Finance Corporation. By decomposing returns into firm, sectoral, and macroeconomic components, we highlight the dominance of firm-specific factors in explaining return variability. This finding aligns with existing literature on idiosyncratic risk and underscores the critical role of firm-level dynamics in investment performance. Our analysis also reveals the heterogeneity in returns across sectors, with high-growth industries such as technology and finance demonstrating superior returns and higher incidences of ”home run” investments. Conversely, more mature sectors showed lower, albeit more stable, performance. Macroeconomic factors, particularly real exchange rate movements, emerge as critical determi- nants of returns. Real exchange rate depreciation negatively impacts dollar-denominated returns, emphasizing the need for robust risk management strategies in currency-sensitive investments. Furthermore, the role of GDP growth in boosting returns varies by sector and investment type. Our analysis has several limitations. The data do not account for certain variables, such as an investee’s reliance on exports or the currency in which it generates revenue, which will likely influence the sensitivity of equity returns to currency depreciations. These are important con- siderations for future research, as understanding these factors could provide insights into the relationship between currency movements and returns. Future research on private equity returns could also benefit from deeply exploring how real and financial components influence industry cycles, particularly in competitive sectors. Hoberg and Phillips (2010) highlight that firms in competitive growth industries often face negative cash flows and abnormal stock returns after periods of high industry financing and overvaluation. This indicates that private equity investors may encounter greater downside risk in these envi- ronments, underscoring the importance of understanding how macroeconomic conditions, such as financing booms, interact with firm valuations and long-term returns. Additionally, Zhang (2005) points out that growth-oriented firms, especially in their early stages, can deliver high re- turns during prosperous times but may disproportionately suffer during downturns. Therefore, future research should examine how private equity investments perform across various business cycles. Similarly, private equity returns may vary across sectors based on their stage of penetra- tion in the economy. Sectors in early growth stages may offer higher returns due to greater growth potential and associated risks, while more established industries yield lower returns. 21 References B AI , Y., C. J. G REEN , AND L. L EGER (2012): “Industry and country factors in emerging market returns: Did the Asian crisis make a difference?” Emerging Markets Review, 13, 559–580. B EKAERT, G., R. J. H ODRICK , AND X. Z HANG (2009): “International stock return comovements,” The Journal of Finance, 64, 2591–2626. C AMPBELL , J. Y., M. L ETTAU , B. G. M ALKIEL , AND Y. X U (2001): “Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk,” The Journal of Finance, 56, 1–43. ˜ , L. AND A. T IMMERMANN (2010): “Volatility Regimes and Global Equity Returns,” in C AT AO Volatility and Time Series Econometrics: Essays in Honor of Robert Engle, Oxford University Press. ¨ C OLE , S., M. M ELECKY, F. M OLDERS , AND T. R EED (2024): “Long-run Returns to Private Equity in Emerging Markets,” Mimeo, World Bank Group and NBER. E SHO , E. AND G. V ERHOEF (2021): “Variance decomposition of firm performance: past, present and future,” Management Research Review, 44, 867–888. H AWAWINI , G., V. S UBRAMANIAN , AND P. V ERDIN (2003): “Is performance driven by industry-or firm-specific factors? A new look at the evidence,” Strategic Management Journal, 24, 1–16. H ESTON , S. L. AND K. G. R OUWENHORST (1994): “Does industrial structure explain the benefits of international diversification?” Journal of Financial Economics, 36, 3–27. ——— (1995): “Industry and country effects in international stock returns,” Journal of Portfolio Management, 21, 53–53. H OBERG , G. AND G. P HILLIPS (2010): “Product market synergies and competition in mergers and acquisitions: A text-based analysis,” The Review of Financial Studies, 23, 3773–3811. J EFFERS , J., T. LYU , AND K. P OSENAU (2024): “The risk and return of impact investing funds,” Journal of Financial Economics, 161, 103928. K APLAN , S. N. AND A. S CHOAR (2005): “Private equity performance: Returns, persistence, and capital flows,” The Journal of Finance, 60, 1791–1823. K APUR , D., J. P. L EWIS , AND R. W EBB (1997): The World Bank: Its First Half Century, Brookings Institution Press. K ORTEWEG , A. (2023): “Risk and return in private equity,” in Handbook of the economics of corporate finance, Elsevier, vol. 1, 343–418. L I , G. (2006): Robust Regression, John Wiley & Sons, Ltd, chap. 8, 281–343. N ORGES B ANK (2019): “Country and Industry Effects in Global Equity Returns,” Working Paper 01/2019, Norges Bank. 22 P HYLAKTIS , K. AND F. R AVAZZOLO (2005): “Stock prices and exchange rate dynamics,” Journal of International Money and Finance, 24, 1031–1053. R UMELT, R. P. (1991): “How much does industry matter?” Strategic Management Journal, 12, 167– 185. S OLNIK , B. H. (1974): “An equilibrium model of the international capital market,” Journal of Eco- nomic Theory, 8, 500–524. Z HANG , L. (2005): “The value premium,” The Journal of Finance, 60, 67–103. 23 A Appendix B Tables Table B1: Correlates of Equity Returns (1990-2023): Robustness Dependent variable is either the annualized TVPI or the equity return adjusted by PME-MSCI. All macroe- conomic variables are annual changes over the duration of the investment cycle. Size is the initial contribu- tion, and duration is the investment holding period measured in years. OLS denotes ordinary least squares, RREG is an alternative to OLS designed to reduce the influence of outliers by iteratively reweighting obser- vations for a more robust fit. Winsorized involves capping values at the 10th and 90th percentiles, reducing the impact of outliers on the regression estimates.* 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) (6) TVPI PME OLS RREG Winsorised OLS RREG Winsorized GDP growth 0.330 0.398*** 0.267** 1.708* 0.982 1.269* (0.229) (0.108) (0.122) (0.930) (0.635) (0.654) RER -0.789*** -0.607*** -0.427*** 0.268 -0.117 0.145 (0.128) (0.060) (0.068) (0.519) (0.354) (0.365) Ln(Size) 0.005 0.004** 0.003* -0.019 0.034*** 0.006 (0.004) (0.002) (0.002) (0.015) (0.010) (0.010) Ln(Duration) 0.027*** 0.001 0.017*** 0.189*** 0.023 0.100*** (0.010) (0.005) (0.005) (0.041) (0.028) (0.029) R 2 0.033 0.077 0.037 0.014 0.009 0.008 Observations 1668 1668 1668 1668 1668 1668 24 Table B2: Correlates of PME Returns by Investment Vehicle (1990-2023) Dependent variable is the equity return adjusted by PME-MSCI. All macroeconomic variables are annual changes over the duration of the investment cycle. Size is the initial contribution, and duration is the in- vestment holding period measured in years. The base category for the interaction is direct investment. The funds dummy indicates the investment was in a fund (“collective vehicle” in IFC nomenclature). Boot- strapped errors in parenthesis. * 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) Quantile Regression 0.1 0.25 0.5 0.75 0.9 GDP growth 0.054 0.560 1.359 2.115 3.938 (0.386) (0.776) (0.927) (1.740) (4.286) RER -0.068 -0.686 -0.351 0.150 1.775 (0.214) (0.429) (0.513) (0.962) (2.370) Ln(Size) 0.011* 0.066*** 0.050*** -0.034 -0.231*** (0.006) (0.013) (0.016) (0.029) (0.072) Ln(Duration) 0.013 -0.003 0.016 0.291*** 0.751*** (0.017) (0.035) (0.042) (0.078) (0.193) Funds × GDP growth 0.742 -0.346 -1.382 -1.031 3.211 (0.821) (1.650) (1.970) (3.697) (9.110) Funds × RER -0.263 0.890 0.462 -0.204 -1.711 (0.478) (0.959) (1.146) (2.150) (5.299) Funds × Ln(Size) 0.027* -0.036 -0.031 0.005 0.078 (0.014) (0.028) (0.033) (0.062) (0.152) Funds × Ln(Duration) -0.157*** -0.111* -0.026 -0.109 -0.286 (0.033) (0.067) (0.080) (0.149) (0.368) R 2 0.032 0.030 0.006 0.019 0.077 Observations 1668 1668 1668 1668 1668 25 Table B3: Correlates of PME Returns by IFC Sector (1990-2023) Dependent variable is the equity return adjusted by PME-MSCI. All macroeconomic variables are annual changes over the duration of the investment cycle. Size is the initial contribution, and duration is the investment holding period measured in years. The base category for the interaction is the FIG sector: collective investment vehicles and finance and insurance. MAS: Manufacturing, Agribusiness and Services. INFRA: electric power; information; oil, gas, and mining; primary metals; transportation and warehousing; and utilities. Bootstrapped errors in parenthesis. * 10%, ** 5% and *** 1%. (1) (2) (3) (4) (5) Quantile Regression 0.1 0.25 0.5 0.75 0.9 GDP growth 0.846 0.929 0.419 1.847 3.117 (0.573) (0.951) (1.146) (2.406) (5.057) RER -0.721** -0.020 -0.085 -0.009 -0.034 (0.310) (0.514) (0.619) (1.300) (2.733) Ln(Size) 0.025*** 0.044*** 0.025 0.008 -0.027 (0.009) (0.015) (0.018) (0.037) (0.078) Ln(Duration) -0.029 -0.070* 0.009 0.276*** 0.559*** (0.024) (0.040) (0.049) (0.102) (0.215) MAS × GDP growth -0.784 -0.112 0.867 0.415 -2.780 (0.884) (1.467) (1.768) (3.712) (7.801) MAS × RER 0.718 0.494 0.661 1.569 4.775 (0.540) (0.895) (1.079) (2.266) (4.762) MAS × Ln(Size) -0.015 0.017 0.054* -0.049 -0.129 (0.016) (0.026) (0.031) (0.066) (0.139) MAS × Ln(Duration) 0.044 0.105 -0.008 -0.151 -0.190 (0.041) (0.068) (0.082) (0.173) (0.363) INFRA × GDP growth -1.080 -0.753 3.683 1.253 -2.430 (1.296) (2.150) (2.591) (5.441) (11.435) INFRA × RER 0.725 -0.302 -1.793 -2.406 -6.142 (0.597) (0.991) (1.194) (2.507) (5.268) INFRA × Ln(Size) -0.017 -0.005 0.021 -0.162** -0.481*** (0.017) (0.029) (0.035) (0.073) (0.154) INFRA × Ln(Duration) 0.029 0.070 0.047 0.132 -0.094 (0.049) (0.081) (0.098) (0.206) (0.432) R 2 0.035 0.043 0.012 0.024 0.089 Observations 1668 1668 1668 1668 1668 26 Figures Figure B.1: MSCI Emerging Markets Index, 1988-2023 27 Figure B.2: Return distributions (annualized TVPI), by sector, investments originated in 1990 or after Note: IFC sector classification. The red vertical line represents the mean, and the black line represents the median. All distributions are winsorized at + 100%. 28