Analyzing fund structure differences and impact sector allocation in emerging market private equity performance

0 MIF: Master Thesis

Christopher Knoll Thesis Advisor – Dr Jens Martin Master in International Finance Amsterdam Business School, University of Amsterdam September 2017

ANALYZING FUND STRUCTURE DIFFERENCES AND IMPACT SECTOR

ALLOCATION IN EMERGING MARKET PRIVATE EQUITY PERFORMANCE

1

ABSTRACT:

This paper analyses historical returns data on 545 emerging market private equity funds using the Preqin private equity database. It makes use of several regional and structural classifications in order to understand the potential performance drivers in emerging market private equity. Making use of aggregate historical Net-IRR, it is observed find that fund size, holding period, firm specialization, and form of ownership are all individually useful in explaining private equity performance in these markets. When tested in a multiple regression, we conclude, in line with our literature review findings, that in emerging market private equity investments, smaller funds that have management expertise tend to exhibit the strongest link with overall performance. We find that impact sector allocation does not have any significant relationship with overall performance, and that returns to this sector are lower as the definition of ‘impact’ becomes narrower. Data on emerging market private equity performance is still limited, which makes drawing absolutely certain predictive conclusions unrealistic. However, this paper highlights several key trends that do provide clarity into the world of emerging market private equity.

2

1 T

ABLE OF

C

ONTENTS

2 INTRODUCTION: ... 3

3 LITERATURE REVIEW AND HYPOTHESIS: ... 6

3.1 Private Equity Investments in Emerging Markets: ... 6

3.1.1 Emerging Market Private Equity Overview: ... 6

3.1.2 Regional Breakdown of Emerging Market Private Equity Fund Performance: ... 8

3.2 Structural Differences in Emerging Market Private Equity Funds: ... 11

3.3 Impact Investment in Emerging Market Private Equity: ... 13

4 DATA AND DESCRIPTIVE STATISTICS: ... 15

4.1 Preqin Private Equity Database: ... 15

4.1.1 Potential Data Problems ... 17

4.2 Descriptive Statistics: ... 18

4.2.1 Performance Analysis by Vintage: ... 18

4.2.2 Regional and Industry Statistics: ... 19

4.2.3 Structural Differences: ... 20

4.2.4 Impact Sector Allocation: ... 23

5 EMPIRICAL ANALYSIS & METHODOLOGY ... 24

5.1 Analysis Structure and Method: ... 24

5.2 Empirical Analysis and Results: ... 25

5.2.1 Regional and Industry Statistics: ... 26

5.2.2 Structural Analysis:... 27

5.2.3 Impact Sector Allocation: ... 32

5.2.4 Robustness Checks: ... 36

6 CONCLUSIONS: ... 38

7 REFERENCES: ... 41

3

2 INTRODUCTION:

Private equity investment in emerging markets has seen strong growth in both aggregate

performance as well as assets under management in recent years. As of September 2015, private equity assets under management had increased to a record high of $297bn, and in 2014, capital distributions in these markets began to exceed capital calls (Preqin, 2016).

However, the heterogeneity of emerging market countries results in differences in perception and interpretation of regional emerging market private equity performance. In a report conducted by the Boston Consulting Group, it is predicted that growth opportunities in private equity investment in Africa is set to continue its strength in decades to come (Boston Consulting Group, 2016). In contrast to this sentiment, African private equity capital flow has tapered off recently, with African

investments falling from $8.1bn in 2014 to $2.5bn and $3.6bn in 2015 and 2016 respectively (The Economist, 2017).

In Central and Eastern Europe, as well as emerging Asia, a similar trend has can be observed, with bullish sentiment being met with declining deal and fundraising activity (Preqin, 2016). Despite this, emerging market private equity activity continues to increase, with capital flows to emerging

markets turning positive in the second half of 2016 (Johnson, 2017), highlighting the importance of a cross-regional analysis of emerging market private equity performance.

To further our understanding of these performance differences, we analyse literature and data on the performance of private equity funds in four emerging market regions; Africa, Emerging Asia, Emerging Europe and Latin America. Following this, structural differences in these funds, such as fund size, holding period, fund manager expertise, and various shareholding classifications are identified, and we analyse whether these classifications and investment preferences have a statistically significant link with the overall performance in emerging market private equity. In addition to this, a prevalent trend that has developed in recent times has been the uptake of “impact” or “socially responsible” investments, as funds target social improvement as well as financial return.

Impact investment, which is defined by the Global Impact Investment Network (GIIN) as

“investments made into companies, organizations and funds with the intention to generate social and environmental impact alongside financial return” has experienced strong growth worldwide, with deal volume growing at a compound annual growth rate (CAGR) of 18% from 2013-2015 (Mudaliar, Pineiro, & Bass, 2016). This growth has been supported by an increasing number of institutional investors (both public and private) such as BlackRock, Goldman Sachs and Bain Capital

4 joining the likes of FMO and PGGM of the Netherlands in launching impact investment strategies (The Economist, 2017).

The sector has been bolstered by improved regulation and the development of acceptable guidelines and standards (The Economist, 2017). However, there is still some debate surrounding impact investments, particularly around definitions of what exactly constitutes an impact investment, and what levels of return are acceptable given the additional focus of generating positive social and environmental impacts.

In addition, Cambridge Associates and the GIIN have, in recent years, developed the Impact Investing Benchmark, which provides “comprehensive analysis of the financial performance of market rate private equity and venture capital impact investing funds” (Cambridge Associates, 2015). The benchmark includes 61 funds (Q3 2016) with total assets under management of $9.9bn, across vintages from 1998 to 2014.

Developments such as this benchmark, as well as the increased availability of performance data from sources such as Burgis, Venture Economics, Preqin and Cambridge Associates (Harris, Jenkinson, & Kaplan, 2014) have made it possible to make inferences about the risk and return characteristics of impact investments in private equity.

Given the uncertainties surrounding private equity investments, both from an academic and industry perspective, as well as the increased popularity and growth potential in emerging market and, in particular, impact investing in these regions, it will be valuable to investigate not only the overall standing and current performance statistics of the above mentioned markets, it will also be

interesting to investigate whether there are any statistically significant links between some of these factors and overall performance.

Hence this study will identify and analyse potential performance determining characteristics in emerging market private equity. Specifically, the study focuses on:

1. Structural differences in emerging market private equity funds such as the fund’s regional and industry focus, the fund’s size, level of firm specialization, fund ownership and investment holding period.

2. Impact Investing – Analysing the performance of various sectors that can be classified as having a social or environmental impact.

The results of the study will provide important insight into the risk and return characteristics impact investments in emerging market private equity. The analysis will quantify and determine if there are

5 any links between a) the fund structure and overall long-term performance, and b) the performance of funds investing in impact investment sectors.

For investors, historically limited data, and the subsequent lack of clarity about performance serve as a large barrier to entry in emerging market private equity investments, and as such, an analysis of the different performance factors mentioned above should be helpful in providing analytical insight into a generally vague sector.

6

3 LITERATURE

REVIEW

AND

HYPOTHESIS:

3.1 P

E

I

E

M

:

For the purpose of this study, Preqin’s definition of Emerging Markets1 _{is used as this definition is} the same as our definitions in the data analysis portion of this report. We further segment this into four primary regions; Africa, Asia, Emerging Europe, and Latin America.

Klonowski, 2013, highlights the increasing amount of private equity capital that has been dedicated to emerging markets in recent years. Strong economic growth and better business conditions have seen foreign investors allocate more capital to these regions. This mirrors the assets under

management and deal growth which is noted by Preqin in its 2016 Emerging Market Private Equity report. As stated in this report, the attractiveness of emerging market private equity to international investors has increased dramatically with the average foreign interest in emerging market deal closures growing from 23% in 2011 to 67% in 2016 (Preqin, 2016).

Figure 1 above shows the relative proportions of funds with an emerging market focus based on the manager’s location. What is observable from the above graphic is that there is a clear trend that has been developing in the last 5 years as mentioned earlier, in which we are seeing a greater amount of foreign investment into these regions.

1 _{Emerging Markets includes all countries in Africa, Asia (excluding Japan, Hong Kong and Singapore), Central} and Eastern Europe (excluding Isreal) and Latin America.

7 Corruption and weak regulatory enforcement remain deterrents to investment in these markets, however it has been found that private equity firms are able to mitigate the costs involved with these market flaws and use their specialist knowledge to work around the complexities of emerging market investment (Johan & Zhang, 2016). Subsequently, an increasing amount of foreign based private equity investment has been taking place in recent years as these investors foresee improvements in the investment landscape in years to come (Klonowski, 2013). This increase in foreign based investment in emerging market private equity highlights these firms’ focus on gaining a local footprint, “on the ground” knowledge, and access to local networks (Gejadze, Giot, & Schwienbacher, 2017) – which we take a deeper look at in section 3.2 of the literature review.

In terms of overall performance, Figure 2 above shows the aggregate assets under management (AUM) in emerging market private equity. As can be seen, there has been strong growth in AUM, however, a large portion of this AUM number, particularly in recent years, has been left as dry powder, or uninvested capital. Further data from Preqin shows that while this number has increased, so has the amount of capital distributions to limited partners – aggregate capital distributions exceeded the amount of capital called up in 2014 and 2015, and the median Net-IRR by vintage year has been increasing steadily from ~5% in 2006 to just over 10% in 2013 (Preqin, 2016). Given this performance and fundraising background, we will analyse further academic research on regional and structural differences, which in conjunction with our data from Preqin should provide good insights as to the drivers and deterrents of emerging market private equity performance.

8 3.1.2 Regional Breakdown of Emerging Market Private Equity Fund Performance:

Regional studies on have shown that geographical location is an influential determinant of emerging market private equity performance. Research across cases, such as within Latin America (Iturralde, 2014), Asia (Oberli, 2014) and Africa (Boston Consulting Group, 2016 & Mataen, 2014) point at differences in investment cases and overall performance. These studies show that while Asia has dominated fund-raising and investment activity historically, additional growth opportunities do exist in the African and Latin American markets.

Africa:

According to the Boston Consulting Group, as of 2016, there were over 200 private equity funds with an African focus, managing upwards of $30bn. This has been the result of growing amount of capital raised in the region, which has averaged around $2,3bn per annum since 2010 (Preqin, 2016). In recent years however, deal activity has lagged somewhat, with the aggregate capital raised mid-way through 2016 only reaching $0.6bn. The Economist, 2017, confirms this in a report showing that African investments have fallen from $8.1bn in 2014 to $2.5bn and $3.6bn in 2015 and 2016 respectively. This has primarily been due to declining commodity and resources prices, which have resulted in less business through the Africa-Asia channel (Klonowski, 2013). Weakening economic conditions in South Africa, partially attributable to commodity prices driving down business with China (Sammut, 2016) has also contributed to the recent decline.

However, despite the short-term weakness in performance, there are indications that strong growth in African private equity on an aggregate capital raised and performance basis is expected. High (5 - 7%) GDP growth, off the back of countries such as Ethiopia, Ghana, Kenya and Rwanda, combined with improvements in governance, regulation and ease of doing business form the basis of the bullish sentiment around Africa amongst private equity investors (Boston Consulting Group, 2016). These political and structural improvements should lower the overall investment risk (Klonowski, 2013), and given that investment valuations are still relatively low (Mataen, 2014), form further basis for the expectation that Africa should perform better than the other emerging market regions. It is expected that as conditions in the region improve, a larger consumer market will emerge, driving growth in FMCG, financial services and technology related industries (Klonowski, 2013), which will add to the growth already being experienced in industries largely funded by government spending such as infrastructure, natural resources, transport, energy and real estate.

Overall, there is an expectation that attractive investment opportunities in Africa exist outside of the traditional South African and Nigerian markets (Boston Consulting Group, 2016), and given the

9 smaller average fund size in the region, require smaller, specialized funds with a local presence in order to fully capitalize on the expected growth opportunities.

Emerging Asia:

With an annual average capital raised of ~$43bn (Preqin, 2016), Emerging Asia has been the dominant region in emerging market private equity investment in recent years, and this has been driven in particular by growth in China and to a lesser degree, India (Oberli, 2014). Klonowski, 2013, echoes this sentiment, citing large populations, a growing middle class, and a robust entrepreneurial sector as some of the key drivers for the economic growth in the region.

While China and India account for ~70% of private equity investment in the region, countries such as Indonesia, South Korea, Singapore and Malaysia have strong economic growth, robust capital markets and large foreign direct investment inflows. (Klonowski, 2013).

Sector specialization varies across countries; with China’s manufacturing sector, India’s services sector and Thailand’s agricultural industry have been examples of this country specialization. In addition to these sectors, there has been strong growth recorded in the consumer goods and services, technology and financial services sectors.

Overall, the private equity sector in the region remains attractive, although, as Klonowski notes, the private equity sector in the region faces its own unique challenges too. Given the fact that the region has been the most popular emerging markets investment destination, there is intense competition for private equity deals as the market becomes more crowded. Additionally, there exist a number of exit barriers; investments in the region typically involve family held businesses, which has been seen to make successful exits more challenging. The most popular exit route in the region is through IPO, however, these have proven to be difficult, especially for smaller private equity funds (Oberli, 2014). While the region has certainly in the past been the top investment destination for emerging market investors (Oberli, 2014), challenging business and legal environments, combined with tricky

shareholding structures and increasing political and regulatory risks are making the region less attractive for private equity investments in comparison to other regions (Klonowski, 2013).

Emerging Europe:

Similar to emerging Asia, emerging Europe has also seen strong private equity investment activity in recent years, averaging ~20 new funds per year, with an aggregate deal value of ~$3bn (Preqin, 2016). This accounts for ~15% of emerging market fundraising activity annually (Klonowski, 2013).

10 While Russia has seen some of the largest individual transactions (Preqin, 2016), Turkey has been the largest private equity destination, with an aggregate amount of ~$3bn invested in Turkish private equity in the 5-year period from 2008 (Klonowski, 2013). The region has also seen increased private equity penetration in Estonia, Bulgaria and Poland. In addition to Russia’s deal size, it has also seen significant investment returns (~20% p.a. over 10 years (EMPEA, 2013)). However, these returns are overshadowed by poor transparency, high levels of corruption and general difficulties doing business in the region, which provides some clarity as to why other Emerging Europe countries have seen larger deal volume despite Russia’s historical performance.

Poland’s private equity sector has grown the fastest, through institutional infrastructure

improvements, increased opportunities and good exit routes. This trend has been mirrored in Russia, Ukraine and the Czech Republic, and has been driven primarily by the consumer goods, retail, communications and life sciences sectors, with financial services and energy investments also contributing (Klonowski, 2013).

Exits in the region tend to be primarily through trade sale and IPO, which account for 35% and 25% of the region’s exit routes respectively. These investments can be characterised as the most attractive in terms of ease of doing business, availability of investment opportunities and access to capital (Klonowski, 2013). Subsequently, limited partners view the region as the most developed and most investable of the emerging market regions, however, this has resulted in a reduction in risk, and therefore lower investment returns.

Overall, while the region is seen as the most developed, least risky, and most capital abundant of the emerging market regions (Klonowski, 2013). Emerging market private equity fund managers tend to have a higher risk appetite than what the region provides, and are as a result preferring investment in other emerging markets – Indonesia, Vietnam and Nigeria are cited as examples of this trend in Klonowski’s 2013 Emerging Market overview.

Latin America:

The second largest region in terms of deal volume, Latin America has averaged ~32 new funds per year since 2011, raising over $7bn in that time period (Preqin, 2016). Strong economic growth in regions such as Brazil, Chile, Columbia, Mexico and Peru (Iturralde, 2014) and an increased number of local, knowledgeable fund managers has resulted in the region being viewed as an attractive private equity investment destination (Klonowski, 2013).

A particular driver of growth in terms of capital raised has been through institutional investment regulation, where pension funds in countries such as Columbia and Brazil can allocate as much as

11 20% of their portfolios to private equity. Industries such as Infrastructure and renewable energy, transport, healthcare, and financial services have been the primary drivers behind the private equity growth in the region (Iturralde, 2014).

Overall, factors such as access to reliable information, inconsistent exit opportunities, legal and regulatory issues remain a deterrent for investment. This can be seen in the political uncertainty in Brazil, Columbia and Chile, the labour unrest in Peru’s mining sector and Mexico, as well as the socio-economic challenges in Argentina and Venezuela (Iturralde, 2014). This risk, however, is counterbalanced by the heterogeneity of countries within the region, generally positive economic outlook and strong investor interest in the private equity sector.

3.2 S

D

E

M

P

E

F

:

Fund Size:

Fund size is a potential performance determining characteristic in emerging market private equity. Fang’s paper on fuse length in private equity points to the idea that there is a tendency for managers to opt for smaller funds in emerging market private equity. While this tendency exists, previous literature indicates that larger funds tend to outperform (Meerkatt & Liechtenstein, 2010) their smaller emerging market counterparts.

Da Rin’s 2013 paper notes that prior to 2000, there was significant information asymmetry in emerging market private equity. This is confirmed in our regional study, in which local expertise is cited as a key performance determinant in each of the different emerging market regions. Firms that are large enough to be able to afford the costs associated with imperfect information, are able to gain competitive advantages over their smaller counterparts, and tend to outperform as a result (Da Rin & Phalippou, 2016).

Since 2000, the information asymmetry in these markets has subsided significantly, especially given the large inflows of foreign based investment into private equity in these regions over that time. (Da Rin & Phalippou, 2016). This inflow is confirmed by the Preqin data mentioned in section 3.1.1. Figure 1 in which we show the large proportion of emerging market private equity investment that has come from North America and Europe. As a result, the study finds no indication that fund size has an influence on performance (Da Rin & Phalippou, 2016). Lopes’ 2013 paper echoes this, in which it is shown that private equity investment actions are not easily scalable, and these

12 subsequent diseconomies of scale result in there being no link between fund size and fund

performance.

Holding Period:

Fang, 2016, argues that private equity fund lifespans are currently lower than their developed market counterparts. This is based on the premise that in these markets, management have

incentives to game returns through opportunistic investments to boost management compensation. Reducing the holding period, or fuse length, of the investment in the firm reduces these

opportunities and thus tend to perform better. This bias for shorter term funds is mirrored in a paper by Lopes de Silanes, 2013, which shows that shorter length funds – “quick-flips” tend to outperform.

Contrary to this position, Fang also predicts that as the institutional investment environment in emerging markets improves, funds will start targeting longer holding periods. This is seen in a report from Boston Consulting, in which it advocates for longer or more flexible fund holding periods (or even evergreen) funds when investing in Private Equity in Africa (Boston Consulting Group, 2016). This potentially reflects the improving institutional environment in emerging markets, but also highlights the fact that in these markets some funds may reach their performance target prior to the exit time, while some funds may require additional time beyond their planned exit date to realize their value (Boston Consulting Group, 2016).

Fund Specialization and Ownership Structure:

Gejadze, et al, 2016, find that specialized firms raise funds more quickly than generalized firms only if the new fund coincides with the current area of the firm’s expertise. While this study was

performed on a set of US based funds, the conclusion is in line with what is noted in the regional review in that a strong local presence, and regional and sector knowledge is seen as a key performance determining factor.

Finally, ownership structure is also seen as a potential performance determining characteristic. Castellaneta, 2013, finds that ownership, defined as corporate affiliation, influences firm performance. While Boston Consulting’s 2010 report finds that specifically funds with minority shareholding positions outperform, it is seen that this link with performance holds true regardless of the form of ownership (Castellaneta & Gottschalg, 2014).

Overall, we find that factors such as fund size, holding period, firm expertise, and ownership in various forms has historically been seen to influence private equity performance. We make use of

13 our Preqin dataset later in the study to assess the performance of funds classified by the above-mentioned characteristics.

3.3 I

I

E

M

P

E

:

“Impact Investment”, as mentioned, is categorised by the GIIN as “investments made into

companies, organizations and funds with the intention to generate social and environmental impact alongside financial return”. However, there is debate as to what exactly constitutes an impact investment, some industry participants argue that it is strictly when the social impact can be measured while others employ much broader “negative screens” which simply filter out firms such as oil or tobacco companies (The Economist, 2017). Others argue that these negative screens are not impactful enough, and classify them under socially responsible investing.

In order to provide clarity as to what constitutes impact investing, the GIIN identifies three central core characteristics of impact investing (Morata, 2017). These are:

1. Intentionality – The investor intends to have a positive social or environmental impact. 2. Range of Asset Classes – Impact investments can be made in both emerging and developed

markets, across asset classes including fixed income, venture capital and private equity. 3. Impact Measurement – The commitment of the investor to measure and report the social

and environmental performance and progress of underlying investments.

Impact measurement has received some scrutiny from the investment community, which makes it difficult to gauge actual performance of these investments (The Economist, 2017). To this end, the GIIN has developed the IRIS Metrics, which measure the social, environmental and financial performance of impact investments across several defined sectors including but not limited to energy, agriculture and financial services.

These IRIS metrics include sector classifications, which provide impact measurement for the following sectors: Agriculture, Education, Energy, Environment, Financial Services, Health, Housing, Land Conservation and Water2_.

As mentioned earlier, differences in the definition of impact investments are a source of continued critique of the sector, particularly in terms of intentionality in prioritizing non-financial impact or prioritizing financial returns (Hӧchstädter & Scheck, 2014). Further debate exists when it comes to the performance of impact investments, some investors are frequently willing to accept a lower

2 _{IRIS Metrics.}

14 return for the sake of having a positive social or environmental impact, while others still target near-market returns for these investments (Brooks, 2015).

Cambridge Associates and the GIIN, in 2015, created their Impact Investing Benchmark, which describes the historical performance of funds classified as “The focus of this report is PI3 _{funds with a} social impact objective to allow for a clear aggregation of similarly motivated funds”. This is further categorized as funds that seek a risk-adjusted market return – thereby excluding funds where investors concede return to generate social impact. The focus of the funds in the benchmark excludes environmental impact, and focuses solely on social impact, which includes financial inclusion, employment, economic development, sustainable living, agriculture, and education (Cambridge Associates, 2015),

This benchmark shows the performance of impact investments in emerging markets using the strictest definition of impact. This paper makes use of this benchmark, as well as benchmarks created using the GIIN classifications and the Preqin database to shed light on these performance discrepancies depending on the definition of “impact investment”. We expect to see decreasing returns as this definition becomes narrower.

3 _{Private Investment}

15

4 DATA

AND

DESCRIPTIVE

STATISTICS:

4.1 P

P

E

D

:

This thesis studies the performance of private equity funds within the Preqin private equity database through Preqin’s private equity online service. We analyse the effects of the various fund structure and investment strategy characteristics mentioned earlier on private equity returns. The data in our sample for these funds’ performance are specified per vintage year dating back to 1982, with historical annual performance data each year from December 2005 through to December 2016. Preqin’s private equity online service is viewed as an industry leading source of private equity data. The graph below indicates the coverage achieved by the Preqin database by comparing the net asset value of the worldwide private equity industry to the aggregate net asset value of the funds within the Preqin database. As can be seen by the graph, there is a relatively large coverage gap, with Preqin’s coverage in 2015 only accounting for ~53% of total private equity investments. While the gap is sizeable, it is understandable given the widely documented lack of transparency in private equity performance, and given the number of funds in the sample set, we believe the data are sufficient for the purpose of this analysis.

Using the Preqin data, we identify 545 private equity funds which focus on emerging markets and have historical performance data available. Similar to previous Preqin emerging market private equity studies, we have extended the emerging market definition to include all countries in Central and Eastern Europe, Emerging Asia (excluding Japan), Africa and Latin America (Preqin, 2016). Further to geographic classification, we also classify these funds by industry and by various structural

0,00 500,00 1 000,00 1 500,00 2 000,00 2 500,00 3 000,00 3 500,00 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Preqin Data Coverage

Aggregate NAV of Funds Used in PrEQIn (USDbn) Aggregate NAV of Global PE Industry (USDbn)

16 differences. These structural classifications include fund size, investment position, board

representation & shareholding preferences, holding period, and firm expertise. Firms are then classified by whether or not they are involved in “clean-tech” investments. This classification is one prescribed in the Preqin database, and includes industries such as renewable energy, waste

management and infrastructure. A further classification is then made based on a broad definition of impact investing derived from the IRIS definition, which includes firms invested in Agriculture, Education, Energy, Environment, Financial Services, Health, Housing, Land Conservation and Water4_. These classifications form the basis of the analysis, in which for each classification a historical mean return and standard deviation can be obtained. These descriptive statistics will form the basis for both the empirical analysis.

As mentioned earlier, the measurement of private equity performance has been the subject of some debate in recent years. As such we have chosen to evaluate multiple performance sources in this study:

1. Net Internal Rate of Return (Net-IRR)

Net-IRR measures the LP’s annualized internal rate of return, and includes all realized and unrealized returns (Harris, Jenkinson, & Kaplan, 2014).

2. The Investment Multiple (TVPI)

The investment multiple compares all contributions from investors with fund distributions and the value of unrealized investments.

Both Net-IRR and TVPI are both presented net of fees and carried interest, and represent the most widely used performance measures in private equity.

For each classification, a pooled Net IRR and TVPI figure is calculated for each year in the period 2005 to 2016 – the maximum time period for which there is a consistent data history. We use the mean historical return and standard deviations thereof for each classification in our analysis. Finally, as a control, we test the historical public market equivalent (PME) performance, a definition of which is given below, as a control for our impact investment classification.

3. Public Market Equivalent (PME):

PME compares the investment in a private equity fund to an equivalently timed investment in the relevant public market. This study makes use of the Kaplan and Schoar public market equivalent (Kaplan & Schoar, 2005), which returns a market adjusted multiple figure, with 1

4 _{IRIS Metrics.}

17 being the base (exact same performance as a public market equivalent). Deviations above and below 1 can therefore be seen as returns.

Unfortunately, the funds for which we have historical Net-IRR and TVPI data do not overlap with those for which we have historical PME performance data. In this instance, we analyse the PME performance of 21 funds meeting the previously mentioned impact investing criteria, compared to 80 global private equity funds on which we have historical PME data.

Additional impact investing data are obtained from Cambridge Associates, which posts a publicly accessible quarterly benchmark of impact investment returns. We generate similar descriptive statistics from this data, based only on Net-IRR, and serves to further clarify the performance characteristics of impact investment.

This model follows and builds on the private equity performance evaluation presented in Harris et. al, 2014 in which the performance of each vintage year is aggregated and presented for vintages from 1984 through 2008. The data presented in this case are point in time data, presented by vintage. This paper makes use of historical returns data, that are aggregated to include all vintages for which a useful performance history exists. This is similar to the methods used both by Preqin and by Cambridge Associates in presenting their respective private equity benchmarks. This method is described by the Private Equity International as an Opportunity Benchmark, and is used to mirror all available investment opportunities (Private Equity International, 2011).

4.1.1 Potential Data Problems

I. Gaps in Coverage:

As mentioned earlier, the Preqin database’s coverage captures just more than half of the funds within the worldwide private equity industry. This coverage is adequate given the secretive nature of private equity performance. However, once we begin filtering for emerging market countries and making the aforementioned classifications, the number of funds on which there are performance histories available diminishes significantly. That being said, we believe that the database and coverage as it stands is enough to make meaningful inferences about a sector that otherwise remains opaque.

II. No Cash Flow data available:

While the performance measures used in this study (Net-IRR, TVPI and PME) do capture cash flows at the fund level. It would add significant value to the study to analyse fund

18

III. PME Performance and Classification gaps:

There are a number of potential problems with the PME dataset. Given that PME

performance measures are relatively new, there were not enough overlaps between funds with Net-IRR and/or TVPI performance histories for a direct comparison to be drawn. Additionally, classification criteria for the funds with only PME performance data were not available for the majority of the funds, which makes a full cross comparison of returns on a PME vs Net-IRR basis redundant, and beyond the scope of this study.

Overall, this study recognizes the limitations of the data, however, we believe that the data available, particularly the Net-IRR and TVPI performance data, are sufficient for the purpose of this study.

4.2 D

S

:

The initial analysis of this study focuses on the historical performance and standard deviation of returns for each of the categories mentioned earlier. We analyse performance differences across the four major regions in emerging markets, which is followed by a similar analysis across industries. 4.2.1 Performance Analysis by Vintage:

The above graph presents the historical average Net-IRR and Investment Multiples (TVPI) for private equity funds in emerging markets. We refer to this graph in conjunction with full performance histories for these funds in the annexure of this report. What is noticeable in the above chart, and in the data in the annexure, is the large spike in returns in the 2000-2002 vintages, followed by significantly lower returns in the 2003 and 2004 vintages. This highlights the

-5,00 0,00 5,00 10,00 15,00 20,00 25,00 30,00 35,00 0,00 0,50 1,00 1,50 2,00 2,50 3,00

Emerging Market Private Equity Performance by Vintage

19 significance of vintage selection in private equity investment. It also highlights cyclicality in returns, which we attempt to account for in our later analysis by grouping the investments together across vintages so as to isolate the structural performance determining characteristics from cyclical or vintage related returns.

The KS-PME data shown in annexure 5 also highlights a potentially interesting trend which is in line with our research on emerging market private equity noted earlier. This trend can be seen in the graph, in that more recent vintages, 2010 onwards, are starting to see growing returns on both an IRR and TVPI basis, and in more recent years, on a PME basis too. PME returns have increased from 1,04 in 2011, to 1,08 in 2012 and to 1,16 in 2013. Naturally, past performance is never an indication of future performance, and it is therefore necessary to identify and analyse the potential drivers of this performance.

4.2.2 Regional and Industry Statistics:

The above table describes the historical performance of private equity funds in emerging markets from 2006 to 2016, which we have categorised both in terms of region and industry focus of the funds. We calculate the return of these funds based on a pooled net-IRR and average investment multiple (TVPI) basis. The variance in net-IRR over the observation period is used to determine the standard deviation of returns.

Regionally, Africa and Emerging Asia have returned the best performance over the observation period, however, African data only dates to 2008, and as can be seen, the number of funds in Africa are significantly less than the other regions. It is likely that the shorter performance history, and fewer funds make any real inferences from these data misleading. Emerging Asian performance is in line with our previous research, the region accounts for more than half of the funds in our study, and has returned the best performance if we exclude the Africa based funds. Table 1: Regional & Industry Statistics Pooled Net-IRR (%) Investment Multiple (TVPI)

Regional # Funds

Mean Annual Return

Standard

Deviation Min Max Average Min Max

Africa 33 9,28% 4,49% 0,00% 14,98% 1,70 0,73 2,66 Emerging Asia 284 8,72% 4,60% 0,60% 14,42% 1,30 1,02 1,73 Emerging Europe 71 7,53% 3,62% 0,00% 15,68% 1,27 0,35 1,57 Latin America 100 7,04% 4,37% 1,52% 16,16% 1,27 0,93 1,87 Emerging Markets 545 7,10% 5,09% 0,00% 16,16% 1,31 0,93 1,78 Industry # Funds Mean Annual Return Standard

Deviation Min Max Average Min Max Consumer Goods & Services 45 12,25% 5,61% 5,29% 23,74% 1,33 0,81 1,62

Diversified 295 9,28% 2,79% 7,94% 13,57% 1,34 1,08 1,62

ICT 16 11,85% 3,56% 2,60% 16,10% 2,03 1,11 3,34

Mining & Natural Resources 13 -0,23% 6,87% -7,50% 10,86% 1,26 0,62 2,10

Property 72 10,09% 6,84% 4,32% 23,20% 1,25 1,07 1,69

Renewable Energy & Infrastructure 22 7,19% 4,56% -3,10% 12,77% 1,46 0,98 1,98

20 The returns from Latin America and Emerging Europe lag the rest of the regions somewhat, which is likely due to the various structural issues in Latin America described earlier, and the decreasing investor appetite for Emerging European private equity exposure which we also allude to in the literature review.

The standard deviation values for these classifications show the variation in returns for a portfolio of all the investable opportunities in the region, and thus does not reflect the risk inherent with a single private equity investment in the region. Rather, the low standard deviations, combined with the high dispersion between minimum and maximum values shown across all regions potentially indicates the importance of manager skill in that while the “market as a whole” tends to vary relatively little, an individual investment’s returns can fluctuate significantly.

In terms of industry classification, we note that diversified funds make up almost half of the funds within our emerging markets universe. This also explains the low standard deviations somewhat, as it is clear that managers in this region have a preference for geographically and industrially diversified funds. These diversified funds show a much lower dispersion in returns and standard deviation, another indication of the potential diversification benefits to be had. The best performing industries are Technology, Consumer Goods and Services, and ICT investments. This is in line with our research findings, as these industries are all consistently mentioned as the top performing (or near to) funds in each of the different regions. Renewable Energy and Infrastructure funds have underperformed the other industries somewhat; we believe this is a possible indication of what was noted regarding impact investing (which renewable energy and infrastructure forms part of) in that investors offer up some return in exchange for impact. Mining and Natural resources has been the worst performing sector, actually producing a negative Net-IRR over the observation period, and this return would be even lower if the 10% positive Net-IRR recorded in 2016 was excluded from the average. 4.2.3 Structural Differences:

Similar to the geographic and industry analysis, we look at the 15-year historical performance of each category. The table(s) below illustrate the historical performance of emerging market private equity funds based on the categorizations we identify earlier. We initially analyse the performance of these funds based on size, investment holding period and firm expertise.

21

Size:

From our first classification, it is observed that the majority of emerging market private equity funds are either Small or Mid-Size funds, ie less than $500m, and these funds have

outperformed their larger counterparts with almost linear decreasing returns to scale. The larger number of smaller funds, and the decreasing returns as the fund size increases are both in line with previous literature stating a tendency for a greater number of smaller funds in emerging markets (Fang, 2016), and decreasing returns as fund size increases (Lopes de Silanes, Phallipou, & Gottschalg, 2013).

Holding Period:

The holding period research conducted suggested a preference for shorter holding periods in emerging markets, but suggested a lengthening of fuse length as the institutional infrastructure improves in individual markets (Fang, 2016). Similarly, other research indicated while “quick-flips” do tend to perform (Lopes de Silanes, Phallipou, & Gottschalg, 2013), there is also an argument for longer holding periods to be made. The descriptive data show a similar trend, the majority of funds tend to have shorter holding periods and with the shortest holding period outperforming the mid-length funds. In addition, we see a smaller number of longer life funds, but these funds have exhibited the best performance, albeit with significantly higher standard deviation.

Firm Expertise:

The classification of expertise in the dataset is somewhat broad, it is based on Preqin’s “expertise” classification in the dataset, and includes funds with any form of value adding expertise, including (but not limited to) financial, operational, industry and management

expertise. Funds excluded from this classification were funds in the dataset that did not have any one of the numerous characteristics listed in the database. This breadth of definition explains the high number of funds in the ‘expertise’ category. Notwithstanding, the aggregate

Table 2: Structural Classification Statistics (1) Pooled Net-IRR (%) Investment Multiple (TVPI)

Size # Funds

Mean Annual Return

Standard

Deviation Min Max Average Min Max

Small (<$100m) 160 9,70% 5,83% -5,00% 16,00% 1,75 0,90 2,70 Mid-Size ($100m - $500m) 217 6,65% 6,36% -5,77% 13,85% 1,20 0,86 1,63 Large (>500m) 168 7,90% 3,66% 1,10% 14,97% 1,26 1,26 1,26 Holding Period 4-5 Years 69 13,97% 3,87% 5,68% 20,00% 1,49 1,22 1,78 6-7 Years 86 7,77% 3,72% 1,45% 14,20% 1,29 0,83 2,01 8-10 Years 42 18,38% 9,01% 11,48% 36,23% 1,91 1,52 2,53 Firm Expertise Firm Expertise 359 12,70% 2,76% 9,96% 17,48% 1,59 1,22 1,98 Generalists 89 4,98% 3,72% -0,57% 9,78% 1,12 0,87 1,50

22 performance of the funds with a form of expertise significantly outperforms the ‘non-expertise’ funds.

Next, we analyse further structural differences amongst these funds, in this instance we focus on the investment position taken by firms in emerging market private equity. To this end, we have classified firms by their investment position (ie lead investor, co-investing, or a mix of the two), board representation preference, and shareholding (controlling vs minority stake) preference. Table 3, below, highlights these differences:

What can be seen from the first classification (investment position) is that funds which have only been involved as a co-investor have clearly underperformed the rest of the market. Therefore, while the differences between pure lead investor funds and mixed funds are small, it is likely that funds taking a leading or mixed investment position will tend to perform better than their co-investing counterparts.

Similarly, firms requiring a seat on the board have also been seen to have outperformed emerging market private equity investments as a whole. This is in line with the previous literature on the subject, and as such we expect this to be a consistent trend when conducting the empirical analysis.

Interestingly, the opposite is the case when analysing shareholding preference. In this case it is noted that funds with no shareholding preference, or a preference for a minority position, have seen higher than average performance in terms of both Net-IRR and TVPI. This, however has come with above average standard deviation, indicating perhaps a tendency of risk-aversion in funds where the investor has a controlling stake.

Table 3: Structural Classification Statistics (2) Pooled Net-IRR (%) Investment Multiple (TVPI) Investment Position # Funds

Mean Annual Return

Standard

Deviation Min Max Average Min Max

Only Co-Invest 28 9,90% 2,37% 6,24% 13,53% 1,53 1,17 1,90 Mix: Co-Invest/Lead 117 15,44% 3,85% 12,62% 24,77% 1,81 1,51 2,10 Lead 91 13,42% 3,72% 8,43% 19,73% 1,49 1,12 2,50 Board Representation Preferred 98 11,09% 2,89% 7,94% 17,28% 1,37 1,16 1,70 Required 192 14,43% 3,39% 10,93% 22,71% 1,81 1,47 2,58 Shareholding Preference Controlling 151 11,83% 3,15% 8,78% 17,52% 1,79 1,43 2,21 Minority 90 10,88% 1,20% 8,00% 12,18% 1,41 0,97 1,87 No Preference 78 18,82% 4,16% 14,68% 28,11% 1,64 1,41 1,91

23 4.2.4 Impact Sector Allocation:

As the final part of the study, we analyse the performance characteristics of socially beneficial, or impact investments in emerging market private equity. As described earlier, we make three different classifications of these types of investment. The first, a broad “clean-tech” definition, which is provided as a classification in the Preqin dataset. Next, we analyse the performance of what we have called “Socially Beneficial” investments, these are investments that aren’t necessarily related to clean technology, but do have, arguably, a positive impact in society. These are investments that fall into the broader IRIS5 _{measures mentioned earlier. Finally, we} analyse the performance of the Cambridge Associates Impact Investing Benchmark for emerging markets, which we describe in the literature review section.

Table 4 (above) shows the historical performance of these “impact” classifications. “Clean-Tech” funds have clearly outperformed the rest of the funds in this classification, while this is biased upwards as a result of one fund in the classification reporting a Net-IRR of ~105%, however, even if this fund is excluded from the calculation, the outperformance remains distinctive. The performance of the non “Clean-Tech” funds does not differ materially from the performance of the

Impact/Socially beneficial funds, returning similar Net-IRR and TVPI values. Finally, the Cambridge Associates benchmark returned a significantly lower Net-IRR, with a much larger distribution of returns. This is expected given the young age of the benchmark and small, yet growing, number of funds within the benchmark.

5 _{Agriculture, Education, Energy, Environment, Financial Services, Health, Housing, Land Conservation and} Water

Table 4: Impact Classification Statistics Pooled Net-IRR (%) Investment Multiple (TVPI)

# Funds

Mean Annual Return

Standard

Deviation Min Max Average Min Max

Preqin "Clean-Tech" 74 12,19% 13,28% -18,40% 27,94% 1,77 0,59 2,97 Non "Clean-Tech" 374 7,23% 4,24% -3,90% 11,84% 1,32 0,95 1,75

Impact/Socially Beneficial 74 6,96% 7,07% -7,40% 14,80% 1,26 0,92 1,62

24

5 EMPIRICAL

ANALYSIS

&

METHODOLOGY

5.1 A

S

M

:

Given the dataset’s short history and relative minimal granularity in that the returns data are annual returns dating back to 2005 for our country and industry classifications, this study conducts the empirical analysis through use of an ordinary least squares (OLS) regression model for these measures. Our structural and impact classifications are tested over a 15-year period, both using individual linear regressions and a final multivariate linear regression incorporating each factor. For each factor, we can determine the extent of the relationship between that factor and returns by regressing the returns the funds within that category against the returns of the overall emerging market private equity universe. Linear regression attempts to model the relationship between two or more independent variables (X) in explaining the dependent variable (Y) by fitting a linear equation to observed data. The OLS multiple linear regression formula is given as:

𝑦_𝑖 = 𝛽₀+ 𝛽₁𝑋_𝑖,1+ 𝛽₂𝑋_𝑖,2+ ⋯ + 𝛽_𝑛𝑋_𝑖,𝑛+ 𝜀_𝑖 𝑓𝑜𝑟 𝑖 = 1,2, … , 𝑛

Where the 𝛽 coefficients are the estimated parameters of each factor specified in the equation. The regression coefficient is calculated as:

β = ∂E[𝑦𝑖|𝑋𝑖] ∂ 𝑋_𝑖

For each factor, we aggregate the returns of all the funds in our universe that fit that factor, and determine the mean returns over our observation time-period. We then regress the average returns of each factor (the X variables in this case) against the average returns in emerging markets as a whole (the Y variable). From this, we calculate the 𝛽 coefficient, standard deviation, t-statistic and subsequent P-Value for each factor. We first analyse the P-value to determine whether the variance in factor X is statistically significant in explaining some of the variation on overall emerging market private equity returns (Y). Once we have determined its statistical strength, we then test the extent of the relationship by looking at the 𝛽 coefficient itself, a higher coefficient (approaching 1) will imply a strong positive relationship, with the opposite being true when the coefficient drops below zero and approaches -1.

25 We then compare our statistical tests to our descriptive results in order to draw conclusions

regarding the extend and statistical significance of these relationships, as well as their economic (or real-world) implications. This method is repeated for each of the previously mentioned factors.

The conclusions we expect to draw should firstly quantify any statistically significant structural relationships prevalent in emerging market private equity, allowing inferences to be made as to which investment strategy/structures perform the best in this sector. Our “Impact” sector

conclusions are tested by running a regression of a similar portfolio of “Impact/Socially Beneficial” funds against emerging market private equity performance measured on a Kaplan-Schoar Public Market Equivalent basis. This comparison is not expected to be very strong, given the lack of overlap in the data we mention earlier, however, it should add to the robustness of the study. Similarly, we make use of Preqin’s Emerging Market Special Report, 2016, to compare our results to, and this is also intended to add to the robustness checks of the paper.

Finally, we test the overall significance of all of the structural and impact classifications in order to produce a reliable multivariate linear regression – we expect this to identify which investment characteristics best explain (on a statistically significant level) the variation in overall emerging market private equity returns.

This paper recognizes the limitations of OLS regression analysis, in that it is likely that some of the assumptions required to have unbiased estimates in the equation will not be satisfied. As a result, these results, even if they do show a strong, statistically significant relationship, cannot be deemed causal relationships. However, even with these limitations, we believe this analysis provides useful information and justification for a more in-depth analysis should more granular data become available.

5.2 E

A

R

:

Similar to the structure of the literature review and descriptive analysis, we conduct our empirical study in three phases. We begin by analysing the performance differences between regions and industries, followed by a structural and investment position analysis, finishing with an analysis of the various impact classifications we have identified. We report the relevant regression coefficients below, and include the full regression outputs in the annexure to this report.

26 5.2.1 Regional and Industry Statistics:

The first of our regression analyses generates the linear regression coefficient statistics based on the regional Net-IRR classification. The regression equation is as follows:

∆𝐸𝑀_𝑖𝑡 = 𝛼 + 𝛽₁∆𝐴𝑆𝐼𝐴_𝑖𝑡 + 𝛽₂∆𝐸𝑈_𝑖𝑡 + 𝛽₃∆𝐿𝐴𝑇𝐴𝑀_𝑖𝑡 + 𝛽₃∆𝐴𝐹𝑅𝐼𝐶𝐴_𝑖𝑡 + 𝜀_𝑖𝑡

The table shows each region’s correlation with emerging markets, and reports a P-value showing the highest level of confidence at which the null hypothesis of no statistically significant link between variance in the independent variable (region) and the dependant variable (emerging market) can be rejected.

Again, our research into the regional performance of emerging market private equity is confirmed, we find that the regression coefficients for funds in Emerging Asia and Europe are the largest, and are both significant at the 95% confidence level. Similarly, Latin America also exhibits a statistically significant relationship with emerging markets as a whole, albeit a weaker relationship. Africa shows a negative regression coefficient, however given the high P-value, it is not statistically significant at any confidence interval, and so we can draw no inferences from Africa’s regression coefficient. Overall, on a regional basis, the data show a strong link between Emerging Asia and Emerging Europe and overall emerging market private equity performance, and this confirms what our

research noted earlier. Interestingly, there appears to be no relationship between Africa and the rest of the emerging markets, however this is likely due to the significantly smaller dataset, hence the lack of statistical significance.

Table 5: Regional Regression

Statistics Coefficients

Standard

Error t Stat P-value

Lower 95% Upper 95% Intercept 0,0131 0,0043 3,0545 0,0552 -0,0005 0,0267 Africa -0,0045 0,0255 -0,1770 0,8708 -0,0858 0,0768 Emerging Asia 0,4192 0,0564 7,4266 0,0051 0,2396 0,5988 Emerging Europe 0,3330 0,0584 5,6982 0,0107 0,1470 0,5190 Latin America 0,2256 0,0322 7,0089 0,0060 0,1232 0,3280

27 The above table shows the regression results of the performance of emerging market private equity funds, classified by industry. Through this analysis we attempt to determine if there are any

statistically significant links between any sector individually and emerging markets as a whole. To do this, we regress the returns of each industry against the returns of emerging markets as a whole using the following equation:

∆𝐸𝑀_𝑖𝑡 = 𝛼 + 𝛽₁∆𝐶𝐺&𝑆_𝑖𝑡 + 𝛽₂∆𝐷𝐼𝑉𝐸𝑅𝑆𝐼𝐹𝐼𝐸𝐷_𝑖𝑡 + 𝛽₃∆𝐼𝐶𝑇_𝑖𝑡+ 𝛽₁∆𝑀𝐼𝑁𝐼𝑁𝐺_𝑖𝑡 + 𝛽₂∆𝑃𝑅𝑂𝑃𝐸𝑅𝑇𝑌_𝑖𝑡 + 𝛽₃∆𝑅𝐸𝑁𝐸𝑊𝐴𝐵𝐿𝐸_𝑖𝑡+ 𝛽₃∆𝑇𝐸𝐶𝐻_𝑖𝑡 + 𝜀_𝑖𝑡

We observe that the historical performance of each individual industry is not statistically significant in explaining the variation in emerging market private equity returns when looking at the P-values of each regression. The only categorization in this instance that exhibits any relationship with emerging market private equity returns is the diversified funds. This we do not find particularly meaningful, in that we expect the performance of funds investing across industries should have a relationship with the performance of the broad market average. However, given the higher return and lower standard deviation noted in these diversified funds in the descriptive data section, this statistical significance we identify here does solidify the argument for a preference for diversified funds over individual exposure in emerging market private equity.

5.2.2 Structural Analysis:

The first part of our structural analysis focuses on the investment position which funds would take (or prefer to take) in their portfolio companies. We test the significance of the firms’ investment position, board representation and shareholding preferences, in order to determine which, if any, factors are important in explaining emerging market private equity returns. The results of our regression analysis are presented in table 7 and commented on below.

Table 6: Industry Regression

Statistics Coefficients

Standard

Error t Stat P-value

Lower 95%

Upper 95%

Intercept 0,0165 0,0181 0,9088 0,4304 -0,0412 0,0741

Consumer Goods & Services 0,1166 0,0920 1,2675 0,2944 -0,1761 0,4093

Diversified 0,4891 0,1447 3,3797 0,0431 0,0285 0,9497

ICT 0,1022 0,1757 0,5819 0,6015 -0,4569 0,6614

Mining & Natural Resources 0,0222 0,0365 0,6085 0,5858 -0,0940 0,1385

Property 0,1190 0,1066 1,1169 0,3454 -0,2201 0,4582

Renewable Energy & Infrastructure 0,0010 0,1250 0,0083 0,9939 -0,3969 0,3990

28 • Investment Position:

We test the statistical relationship between the type of investment position and returns by using the following equation:

∆𝐸𝑀_𝑖𝑡 = 𝛼 + 𝛽₁∆𝐿𝐸𝐴𝐷_𝑖𝑡 + 𝛽₂∆𝑀𝐼𝑋𝐸𝐷_𝑖𝑡 + 𝛽₃∆𝐶𝑂𝐼𝑁𝑉𝐸𝑆𝑇_𝑖𝑡+ 𝜀_𝑖𝑡

The first observation we make from the above table is that there are statistically significant links between the performance of these funds and the type of investment position taken. In assessing the strength of these relationships, we find that Co-Investing and Mixed strategies explain the greatest amount of variation in emerging market returns. Funds with only Lead-investment positions exhibits the weakest relationship. Combining this finding with the increased amount of investment in emerging markets coming from outside the market itself (Preqin, 2016), we believe that the stronger relationship between co-investing funds and emerging markets is an indication of foreign funds investing with local partners to access local market knowledge.

• Board Representation

We regress the returns of (1) funds requiring board representation and (2) funds preferring board representation against our emerging market universe with the following equation:

∆𝐸𝑀_𝑖𝑡 = 𝛼 + 𝛽₁∆𝑅𝐸𝑄𝑈𝐼𝑅𝐸𝐷_𝑖𝑡 + 𝛽₂∆𝑃𝑅𝐸𝐹𝐸𝑅𝑅𝐸𝐷_𝑖𝑡 + 𝜀_𝑖𝑡

The strongest observed coefficient in this analysis is that related to a requirement for board representation, with a coefficient of 0.67, which is statistically significant given its P-value.

Table 7: Fund Structure Regression

Statistics (1) Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Investment Position Intercept 0,0005 0,0137 0,0355 0,9725 -0,0312 0,0322 Co-Invest 0,3837 0,0817 4,6963 0,0015 0,1953 0,5721 Mixed 0,3461 0,0473 7,3154 0,0001 0,2370 0,4552 Lead/Sole 0,1990 0,0773 2,5744 0,0329 0,0207 0,3772 Board Representation Intercept -0,0005 0,0027 -0,1734 0,8646 -0,0063 0,0054 Required 0,6732 0,0321 20,9638 0,0000 0,6047 0,7416 Preferred 0,3226 0,0316 10,2147 0,0000 0,2553 0,3900 Shareholding Preference Intercept -0,0087 0,0108 -0,7994 0,4396 -0,0323 0,0150 Control 0,5402 0,0908 5,9465 0,0001 0,3423 0,7381 Minority 0,2055 0,1108 1,8545 0,0884 -0,0359 0,4469 No Preference 0,1402 0,0532 2,6328 0,0219 0,0242 0,2562

29 This, again is in line with industry theories, in that having a degree of influence through a seat on the board, should at the very least diminish the fund’s risk, but should also show a stronger link with positive returns.

• Shareholding Preference

Finally, we analyse the link between shareholding preferences and overall emerging market returns by the following regression equation:

𝛼 + 𝛽₁∆𝐶𝑂𝑁𝑇𝑅𝑂𝐿_𝑖𝑡 + 𝛽₂∆𝑀𝐼𝑁𝑂𝑅𝐼𝑇𝑌_𝑖𝑡 + 𝛽₂∆𝑁𝑂 𝑃𝑅𝐸𝐹𝐸𝑅𝐸𝑁𝐶𝐸_𝑖𝑡 + 𝜀_𝑖𝑡

Similarly, when analysing the performance of funds based on their shareholding preference in their portfolio companies, we notice that funds with a preference for control show the strongest link with emerging market performance. Minority shareholding funds show a lesser link with emerging market performance, and is only significant at the 90% confidence interval.

Overall, we believe these regression results present a telling case for fund structure analysis in our dataset. From the descriptive section, we notice stronger investment returns attributed to mixed and co-investing funds, where there is no preference as to the type of shareholding taken, and when there is a requirement for board representation. This we believe is a crucial nuance in emerging market private equity; it is important to have “skin in the game”, however, it is also important for the fund investor to not take away control of the portfolio company from the founder. These results confirm what is noted in our research, in that ownership, regardless of form, does matter (Lopes de Silanes, Phallipou, & Gottschalg, 2013).

Continuing the structural analysis, table 9 (below) shows the regression results for the final three structural factors. We test whether or not a fund’s size, holding period and/or investment manager expertise have an effect on the overall performance of emerging market private equity funds. The regression outputs of each equation are listed below:

30 • Fund Size:

In our structural analysis, we split the funds in the emerging market private equity database into four distinct size categories based on the overall fund value. Funds worth less than $100m are classed as “Small”, those between $100m and $500m are classed as “Mid-Size” and “Large” funds constituting any fund worth more than $500m.

Hence, our regression equation for determining the effects of fund size is: ∆𝐸𝑀𝑖𝑡 = 𝛼 + 𝛽1∆𝑆𝑀𝐴𝐿𝐿𝑖𝑡 + 𝛽2∆𝑀𝐼𝐷𝑖𝑡 + 𝛽3∆𝐿𝐴𝑅𝐺𝐸𝑖𝑡 + 𝜀𝑖𝑡

What we noted in the descriptive analysis section of the paper on fund size was a confirmation of previous academic research in that there are decreasing returns to scale when it comes to fund size. Our regression results show that the smaller funds – both the small and mid-size categories – exhibit the strongest link with the dependant variable, emerging market private equity returns. We note highest regression coefficients with these funds, 0,53 and 0,30

respectively, showing that when we classify emerging market private equity investments by size, the smaller funds tend to have the most significant link with emerging market private equity performance. This conclusion is particularly interesting given the large amount of dry powder in emerging market private equity; our data, and previous literature on the subject show a strong leaning towards smaller funds, therefore we believe that the large amounts of dry capital noted earlier could more efficiently be invested should it be focused on a greater number of smaller funds.

Table 8: Fund Structure Regression

Statistics (2) Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Fund Size

Intercept -0,0095 0,0032 -3,0144 0,0108 -0,0164 -0,0026

Small 0,5279 0,0371 14,2326 0,0000 0,4471 0,6087

Mid-Size 0,2993 0,0347 8,6325 0,0000 0,2238 0,3749

Large 0,2158 0,0333 6,4728 0,0000 0,1432 0,2884

Investment Holding Period

Intercept -0,0277 0,0526 -0,5275 0,6075 -0,1423 0,0868 < 5 Years 0,0745 0,3539 0,2105 0,8368 -0,6966 0,8455 5-6 Years 0,6534 0,3966 1,6473 0,1254 -0,2108 1,5176 >7 Years 0,2529 0,2153 1,1743 0,2631 -0,2163 0,7221 Firm Expertise Intercept 0,0047 0,0016 2,9174 0,0106 0,0013 0,0081 Firm Expertise 0,7167 0,0133 53,9203 0,0000 0,6884 0,7450 Generalists 0,1964 0,0252 7,7952 0,0000 0,1427 0,2501

31 • Investment Holding Period

Based on the Preqin data we use for this analysis, we were able to extract holding period data for 197 of the 545 funds where a specific holding period preference was noted. We analyse the returns of funds with holding periods of less than 5 years, 5-to-7 years, and greater than 7 year periods (classified as short, mid and long respectively) and provide commentary in reference to table 9 below:

Similar to our size regression, our holding period equation is as follows: ∆𝐸𝑀_𝑖𝑡= 𝛼 + 𝛽₁∆𝑆𝐻𝑂𝑅𝑇_𝑖𝑡 + 𝛽₂∆𝑀𝐼𝐷_𝑖𝑡 + 𝛽₃∆𝐿𝑂𝑁𝐺_𝑖𝑡+ 𝜀_𝑖𝑡

Our literature review notes two trends in emerging market private equity investments with regards to investment holding period; there is a tendency to favour shorter holding term periods (Fang, 2016), whilst at the same time, there is growing investor interest in utilizing more flexible, longer holding periods (Boston Consulting Group, 2016). We find no statistically significant link between the 4-5 and 6-7 year holding period classifications and emerging market private equity returns, nor do we find any statistically significant link between the longer holding period funds. While this result is telling, and we expect it to be confirmed in the multiple regression, it must also be noted that for the longer funds the data history is very limited, making drawing any inference from this analysis unreliable.

• Firm Expertise

The final factor in our analysis is on what is classed in the Preqin dataset as funds with some level of expertise, the exact definition of which we cover earlier in this paper. We make use of the following regression equation in our analysis of firm expertise:

∆𝐸𝑀_𝑖𝑡= 𝛼 + 𝛽₁∆𝐸𝑋𝑃𝐸𝑅𝑇_𝑖𝑡 + 𝛽₂∆𝐺𝐸𝑁𝐸𝑅𝐴𝐿𝐼𝑆𝑇𝑆_𝑖𝑡 + 𝜀_𝑖𝑡

Both factors (“expert” firms and “generalist” firms) are statistically significant in explaining the variance in overall emerging market private equity performance, however, the strength of the relationship between “expert” firms and overall emerging market private equity performance (0,72) vastly outweighs the “generalist” category.

Our previous literature notes that specialized firms are more capable and efficient in raising new capital (Gejadze, Giot, & Schwienbacher, 2017), and these results tend to suggest that these funds tend to perform better than their non-specialized counterparts. However, given the breadth in the definition of “expertise” in our database, these results do not indicate if there are