NESTA is the UK's foremost independent expert on how innovation can solve some of the country's major economic and social challenges. Its work is enabled by an endowment, funded by the National Lottery, and it operates at no cost to the government or taxpayer.

NESTA is a world leader in its field and carries out its work through a blend of experimental programmes, analytical research and investment in early-stage companies. www.nesta.org.uk

Executive summary

The importance of a vibrant venture capital industry in supporting growth is widely recognised, and consequently governments across the world have sought to promote the industry. But the development of the VC industry in the UK (and in many other countries) has been hampered by the low returns it delivers to its investors. Understanding how the UK venture capital market compares with other ones, particularly the US market, is the first step towards improving the performance of the UK VC industry.

This report sheds further light on the magnitude of the performance gap between US and UK venture capital funds, its evolution over time, and what the likely drivers of the performance differences are. It uses a novel database that combines data on VC fund performance and their investments in the US and UK for 791 funds raised between 1990-2005. Therefore, it not only reports differences in aggregate performance across countries, but in addition it compares like-for-like funds, with the same focus, vintage year and experience, but located on opposite shores of the Atlantic.

The key contribution of this study to the debate on venture capital fund performance in the UK is thus twofold: first, it aims to identify key factors that are associated with VC fund performance, and second, to examine if any of these factors may explain the performance difference between UK and US-based funds.

The report thus provides useful insights to investors determining their asset class allocation and selecting which VC funds to invest in, VC fund managers choosing their investment strategy and fund structure and, finally, policymakers aiming to support the development of the VC industry.

Key findings

1. The returns performance of UK and US VC funds in recent years has been very similar. UK funds have historically underperformed US funds, but this gap has significantly narrowed. The gap in fund returns (net IRR) between the average US and UK fund has fallen from over 20 percentage points before the dotcom bubble (funds raised in 1990-1997) to one percentage point afterwards (funds raised in 1998-2005). However, this convergence has been driven by declining returns in the US after the burst of the dotcom bubble, rather than by increasing returns in the UK. Average returns for funds raised after the bubble in both the UK and the US have been relatively poor, but VC performance is likely to move upwards as VC funds start to cash out their investments in social media companies (particularly in the US).

2. The wider environment in which UK funds and the companies they finance operate was a major contributor to the historical gap in VC returns. While there are some large differences in the observable characteristics of VC funds between both countries, they cannot account for the historical returns gap.

3. Average returns obscure the large variability in returns within countries. The dispersion in returns across funds was highest during the pre-bubble years, and has fallen significantly since then. But in both periods the gap in returns between good and bad performing funds within a country was much larger than the gap in the average returns across countries. Thirteen per cent of UK funds established since 1990, would have got into the top quartile of US funds by returns (this is 22 per cent for funds established in the post bubble period), while 45 per cent of UK funds outperformed the median US fund. Selecting the right fund manager is thus more important than choosing a particular country.

4. The strongest quantifiable predictors of VC returns performance are (a) whether the fund managers' prior funds had outperformed the market benchmark; (b) whether the fund invests in early rounds; (c) whether the fund managers have relatively more prior experience; and (d) whether the fund is optimally sized (neither too big nor too small). Moreover, historical performance has been higher for funds located in one of the four largest investor hubs (Silicon Valley, New York, Massachusetts and London) and for investments in information and communication technology.

5. UK government-backed funds have historically underperformed their private counterparts, but the gap between public and private returns has narrowed in recent periods. This suggests that in later years governments have become savvier when designing new VC schemes.

Part I: Introduction

Venture capital has been the driving force behind some of the most vibrant sectors of the US economy.¹ It is an important source of funding, expertise and networks for innovative companies (see Section 1.1). Because of this, a well-developed venture capital industry is crucial for entrepreneurs, financiers and policymakers alike. But the development of the venture capital industry around the world has been slow and uneven.

The first venture capital (VC) fund was created in the US back in 1946, but the growth of the US VC industry only accelerated in the late 1970s. Across the Atlantic, the UK venture capital market only began to take off in the 1980s, though 3i was also founded in 1946. And it was not until the late 1990s that venture capital started to take hold in the rest of Europe. Today the US continues to be home to the largest venture capital industry in the world, investing $20 billion in 2010 (Figure 1). In contrast, VC investments made by European-based funds totalled only $5 billion in 2010, a quarter of the US level. Within Europe, the UK has one of the most active VC market, with UK funds investing $1 billion in 2010. Still, this only represents 0.05 per cent of UK GDP, which compares unfavourably to 0.14 per cent in the US.²

The development of the VC industry over time has not been smooth either. Venture capital activity peaked in 2000, fuelled by an internet bubble that was followed by millions of dollars lost, hundreds of bust companies and a massive fall in VC activity. As the industry began to pick up pace again the financial crisis struck to reverse these gains, however Figure 1 reveals that in 2010 VC investments were growing again.

While venture capital continues to spread globally, expanding in China and other emerging markets in recent years, questions are being asked about the industry's ability to deliver acceptable returns in more mature markets. Average VC returns over the last decade have been low, not having recovered from the dotcom crash (Figure 2). Some observers argue this is a permanent shift towards lower returns, raising doubts about the sustainability of the VC model. However, VC returns have not been much different relative to the overall stock market since 2002, so recent returns may reflect the natural evolution of a competitive market.³ As VC funds start to cash out their investments in social networks, returns are likely to move upwards once again.

Whether European VC funds will benefit from this recovery is a different question. Trade associations and VC professionals have long asserted that realised returns of venture investments in Europe have historically been lower than in the US, and pointed to this underperformance as the main obstacle to the development of a strong VC industry in the UK and continental Europe.⁴ The few empirical studies that provide a comparison between US and Europe (and the even fewer that consider US and UK differences) appear to support these claims: in all such studies the UK and the European venture capital industry more generally are perceived to be the poor cousins of the US industry, consistently delivering lower returns to their investors.⁵

However, most of these studies only compare aggregate performance across countries, without considering how the characteristics of each individual fund and its investments can help explain the performance gap across countries.

This report aims to fill this gap, focusing on the performance difference between UK and US funds and its evolution over time. Specifically, we compare the returns in the US and the UK of like-for-like funds, with the same focus, vintage year and experience, among other factors, and explore whether some of these factors may explain the UK's underperformance. To do so, we build a new database that matches fund level performance records with individual investments made by these funds, covering 791 funds raised in the UK and the US between 1990-2005 (See Box 2 for a description of the database and the sources used to build it).

The second aim of this report is to identify which factors are associated with better VC fund performance in the UK. Most of the literature that has explored the drivers of VC returns has only used data for US funds. We examine whether the same factors explain performance on both shores of the Atlantic, and how the impact of these factors has evolved over time.

Finally, the success of venture capital and the potential benefits that it can bring into an economy have not been left unnoticed by policymakers, and thus several governments around the world have attempted to stimulate VC activity. Therefore we also consider the relative performance of publicly backed venture capital funds in an attempt to assess the role of government in this area.

1. See Box 1.

1.1 The contribution of venture capital to innovation and economic growth

The provision of suitable finance to high-growth companies requires the presence of effective financial systems that support innovation and growth. Bank finance is by far the main source of finance for most businesses in developed economies. The majority of firms accessing finance rely on small-scale debt finance: credit cards, overdrafts, and commercial loans. Only a small proportion of UK businesses seeking external finance receives venture capital (3 per cent). A natural question is why venture capital is important if only such a small proportion of firms use it. Empirical evidence on the significance of venture capital for innovation and growth is available at the following three levels.

Venture capital and firms

Venture capitalists provide several critical services in addition to providing finance, such as helping the company to raise more finance, reviewing and helping to formulate business strategy, filling in the management team and introducing them to potential customers and suppliers.⁶ Venture capitalists carefully screen firms and structure contracts to strengthen incentives and monitor firms, promote their professionalisation and induce them to behave more aggressively. Gompers and Lerner (2001) write that venture capital helps entrepreneurial firms to invest more than they would otherwise, to grow more quickly, and sustain performance in the long term – even after going public.⁷

Chemmanur and Loutskina (2005) found that venture capitalists attract a greater number and higher quality of market participants such as underwriters, institutional investors, and analysts to an IPO, thus obtaining a higher valuation for the IPOs of firms backed by them. They also found some evidence that suggests that venture capitalists are able to either select better quality firms to back (screening), or help create such higher quality firms by adding value to them (monitoring) in the pre-IPO stage.⁸

Knowledge regarding the target company's industry allows the venture capitalist to oversee investments more efficiently and more effectively, in part because industry experience enhances the venture capitalist's ability to recognise signs of trouble at an early stage.⁹ As a consequence, venture capital has become an important source of funding for a significant number of innovative companies, not only in the United States, but also increasingly around the world.¹⁰

Venture capital and innovation

The role of venture capital in economic development is increasingly recognised as central to the development of an entrepreneurial economy.¹¹ There is an emerging consensus that venture capital is a key component of the innovation process¹² and that it plays a crucial role in 'new economy innovation systems' formed by highly dynamic sets of interrelationships between VCs, market conditions and new firm incubators.¹³

On the basis of his study of the Internet industry in the US, Zook (2005) concludes that: "financial institutions of innovation are probably more important for economic growth in this knowledge economy than the location of research universities”.¹⁴ Kenney and Florida (1988) argue that VCs play a key role within regional innovation networks and act as catalysts or 'technological gatekeepers' who facilitate and direct innovations within localised clusters¹⁵ making it the "most important institution supporting technology entrepreneurship".¹⁶

A variety of studies suggest that venture-backed firms are responsible for a disproportionate number of patents and new technologies,¹⁷ and they bring more radical innovations to market faster than lower-growth businesses that rely on other types of finance.¹⁸ In fact, venture capital has played a unique role in the information and commercialisation of entire new industries.¹⁹ In the case of biotechnology for example, "it is safe to say that without venture capital and regional agglomeration, the industry would not exist in the form that it does today".²⁰

Venture capital and employment growth

The evidence from the US, where venture activity has a longer pedigree, is compelling. A study undertaken by Puri and Zarutskie (2008) found that despite the extremely small proportion of VC-backed firms (0.1 per cent of all new firms created between 1981 and 2001) they accounted for nearly 10 per cent of employment in the US in the late 1990s and early 2000s. Chemmanur, Krishnan and Nandy (2009) also show that VC-backed firms have higher levels of productivity compared to non VC-backed firms. Hence, VC is widely perceived to be an engine of growth for the economy. The question that arises is why other sources of business finance cannot borrow or duplicate such features and processes of the venture capital industry. Hall and Lerner (2010) summarise these reasons as follows:

“Economists have suggested several explanations for the apparent superiority of venture funds in this regard. First, because regulations limit banks' ability to hold shares, at least in the US, they cannot freely use equity. Second, banks may not have the necessary skills to evaluate projects with few collateralizable assets and significant uncertainty. Finally, venture funds' high-powered compensation schemes give venture capitalists incentives to monitor firms closely. Banks sponsoring venture funds without high-powered incentives have found it difficult to retain personnel”.²¹

A natural question is whether the sample of funds with performance data is representative of the wider population of funds, given that we only have performance data for 791 funds. We follow a number of steps to mitigate these concerns. First, we compare our performance metrics to the returns provided by Thomson One, the most comprehensive source of aggregate returns for the VC industry (even if also incomplete; for a critique see Maats, et al. (2011)³⁵). As seen in Figure A2 in the appendix, the IRRs in the UK and the US follow similar patterns over time regardless of the source used. Second, we compare the characteristics of funds for which we have performance data to the funds for which this is not available. We identify several differences between both samples, reported in Table A3 in the appendix. For instance, funds reporting performance data tend to be larger, from more established GPs, from earlier vintage years and invest larger amounts in more companies. However, these differences appear both in the US and the UK samples, so they are unlikely to bias the estimates for the returns gap. Moreover, the econometric analysis controls for these factors as well. There is only one difference worth discussing in more detail. As seen in Table A3, it appears that our performance sample oversamples funds with a higher share of IPO exits in the UK while it undersamples them in the US, which suggests that we may be underestimating the magnitude of the returns gap. The share of IPO exits is an imperfect proxy of performance, but we try different methods to exploit this information to correct our estimates (i.e., weighting and imputing) and find similar patterns under these corrections.

Part 2: The venture capital industry in Europe and the US

The US VC industry continues to be more developed than the European VC industry, even though both venture capital industries follow similar cycles. In this section we ask whether differences go beyond the aggregate size of the market, examining fund characteristics in the UK, the US and continental Europe. We focus the discussion on a few selected factors, but include more detailed information on several other characteristics in the appendix.³⁶

2.1 There are some large differences in fund characteristics across countries, and they are robust over time

US funds are larger than European funds, and the gap has not narrowed much over time. Funds raised between 2006 and 2009, the most recent year for which we have data, are larger in real terms than funds raised during 1990-2005, the main study period. But convergence between countries in fund size has been limited. Average fund size in the US was twice as much as in continental Europe in 1990-2005, and continues to be 90 per cent higher in the most recent period.

Their larger size allows US funds to make twice as many investments as European funds, and invest larger amounts to each portfolio company on average, particularly relative to continental Europe. US funds invest in a larger number of companies, but also undertake a larger number of investments in each of the companies that they fund. Over 1990-2005, US funds invested in follow-up rounds in roughly 40 per cent of their portfolio companies, which compares to only about 20 per cent in the UK and continental Europe. There are also differences in their syndication practices. US funds tend to invest with a larger number of co-investors than UK and continental European funds (Table 1).

2.2 US funds are more specialised, concentrating investment in fewer sectors

Figure 3 shows the proportion of capital invested by all funds in our sample across nine broad industries.³⁷³⁸ The left panel presents the allocation of capital for funds established in the 1990-2005 period and the right panel focuses on the most recent funds, established in 2006-2009. In the earlier period the industry defined as internet/computers received the highest share of investment in all three examined geographical regions. Biotech/healthcare and communications/electronics were the next most preferred industries. The remaining industries received considerably smaller proportions of investments. If we look at funds raised in the most recent period (2006-2009), there has been a shift of investment away from internet/computers, even if this industry continues to top the ranking.

US funds had, and continue to have, a much larger focus on internet/computers and communications/electronics when compared with European funds, which instead invest relatively larger proportions in other sectors such as business/industrial and consumer. European funds also invest more on energy, a sector that has tripled its share of investment in recent years.

There are also differences in the degree of sectorial specialisation across countries. Using a measure of the concentration of investment

in an industry or a group of industries,³⁹ we find that US funds are more specialised than both UK and continental European funds. Specialisation may be a more difficult strategy in Europe given the higher fragmentation of the European VC market, although in the most recent years (2006-2009) we observe a trend towards higher specialisation in all three markets (Table 1).

2.3 European funds are more internationalised than US funds

Most US funds have traditionally only invested locally, with less than a third of US funds raised between 1990 and 2005 having invested in one or more companies outside the US (Figure 4a).⁴⁰ In contrast, the majority of European funds have invested outside of their home market.⁴¹ This reflects the smaller size of home markets across Europe and, encouragingly, the ability of European funds to overcome, at least partially, this constraint and invest across borders.

Figure 4b reports the location of investments broken up by the fund's nationality. Less than 20 per cent of US funds raised between 1990 and 2005 made an investment in a European company, while about 40 per cent of European funds (both in the UK and the continent) have invested in the US.

The strong geographical preference of US funds for investing in the US becomes even clearer when we consider the amounts invested in each market (Figure 4c).⁴² Ninety-one per cent of all US VC capital raised between 1990 and 2005 stayed in the US, while less than 5 per cent went to Europe. In contrast, only 47 per cent of all capital invested by the UK funds went to UK companies, while as much as 20 per cent went to the US and another 20 per cent was invested in continental Europe. Finally, continental European funds invested a large proportion of their funds in the US (17 per cent), but only 5 per cent in the UK.

The situation has changed somewhat in recent times. A higher proportion of European funds raised in 2006-2009 have chosen to invest locally while US-based funds are becoming more global. As a result, the proportion of European VC capital being invested in the US has halved, falling to 10 per cent, and a slightly larger share of US VC capital is coming to Europe.

Overall, this analysis suggests that Europe did not offer an attractive proposition to US VC funds. Europe has a less developed VC market than the US, so attracting US funds (their

Footnotes

Overview of Venture Capital Investment Geography

money but also, crucially, their expertise) ought to benefit European economies. Instead, the opposite is happening. A much larger share of European VC funds invest in the US than the other way around. While Europe is likely to benefit from its funds investing in the US (for the returns it provides, the network it builds and the experience it generates), the small flow in the opposite direction is a cause for concern.

The analysis also shows that the UK has become an attractive destination to domicile venture capital funds, building on its strong financial industry. This has made it easier for London to become a global hub for venture capital funds.

geographical areas, Silicon Valley, New York, Massachusetts and London, hosted 40 per cent of all VC funds launched in the US and Europe between 1990 and 2005 (Figure 5). This is discussed further in Section 4.4.

2.4 The global venture capital industry is concentrated in very few hubs

Information about potential investment opportunities generally circulates within geographic and industry spaces. ⁴³ This makes it advantageous for fund managers to locate near each other and close to agglomerations of high potential entrepreneurs. As a result, four

Part 3: The performance of venture capital funds

Observers have argued that the low returns delivered by European VC funds to their investors are a major factor hampering the development of the VC industry in the UK and continental Europe. This section aims to understand the magnitude of this performance gap, its evolution over time, and what potential factors may explain it.

3.1 UK funds have historically underperformed US funds, but the gap has narrowed

We consider two different metrics of VC fund performance, fund multiples and the internal rate of return. The fund multiple corresponds to the amount that investors receive from the fund relative to the amount that they had originally contributed.⁴⁴ However, multiples do not take into account the time value of money: a fund which pays back investors after 12 years has the same multiple as one which does so after two years, even though investors would much prefer the latter. Therefore we also use the internal rate of return (IRR), which takes account of this and is thus our preferred metric. ⁴⁵ Performance data is only available for a subset of funds, so we restrict the analysis in this section to the UK and the US, since the data on these regions is the most complete.⁴⁶

Figure 6 reports the average performance of funds raised in the UK and the US between 1990 and 2005, broken in two periods. Whether we look at IRRs or multiples a clear UK-US gap arises for funds raised before the dotcom bubble. The average IRR for funds with vintage year 1990-1997 is 33 per cent in the US and 13 per cent in the UK. That is, returns for pre-bubble funds in the UK were 20 percentage points lower than in the US. Multiples display the same patterns. US VC funds raised in 1990-1997 returned to their investors 181 per cent more than their original investment on average, compared to UK funds which only managed to return an additional 66 per cent to their investors (2.81 vs. 1.66).

But this gap in returns has narrowed significantly since the bubble burst. Funds with vintage year 1998-2005 in the US reported IRRs of -0.21 per cent on average, compared to -1.21 per cent in the UK, a difference of 1 percentage point. Multiples have also converged (1.05 vs. 1).

The volatility of fund returns becomes even clearer when looking at Figure 7, which illustrates the performance of VC funds by vintage year and location. Funds raised in the years before the dotcom crash performed well, while returns for funds that followed have been close to zero. This dynamic is quite typical of the venture capital industry, which is very sensitive to business cycles and public market dynamics, and tends to display a large cyclicality in returns. ⁴⁷

VC returns have followed a similar pattern in the UK and the US over the sample period, although there are two significant but related differences. First, returns in the UK did not grow as fast as they did in the US in the run-up to the dotcom bubble, and as a result the UK was proportionally less affected by the dotcom crash that followed. Second, the magnitude of the UK-US returns gap has changed over time, and the gap appears to have almost closed for funds raised after 1997. However, the convergence of returns has not been driven by UK funds becoming better, but by the worsening performance of US funds.

3.2 The historical UK-US returns gap cannot be explained by observable fund characteristics

As seen in Section 2, there are some important differences in the characteristics of venture capital funds in the US and the UK. We next ask whether these differences can help explain the UK-US performance gap. In other words, we examine whether the UK-US gap in average returns is explained by (a) differences in the characteristics of the VC industry between countries (e.g., the US having larger funds) or (b) differences in the returns that funds with exactly the same characteristics achieve in each country (e.g., large UK funds doing worse than large US funds).

We use regression analysis to examine how the UK-US returns gap changes when we control for factors related to the characteristics of the fund, its managers and its investments, and test how statistically significant any remaining differences are. We split the sample into four periods (1990-93, 1994-97, 1998-2001, and 2002-2005) to examine how the returns gap has evolved over time. This also has the advantage of smoothing some of the year-on-year fluctuations in returns typical of the venture capital industry. Table A5 in the appendix presents the results of this analysis, which is summarised in Figure 8.⁴⁸

Each bar represents the magnitude of the UK-US returns gap in each period when a set of controls has been included in the regression. As in Figure 7, we also observe a narrowing of the gap between UK and US funds returns over time. While before the dotcom bubble, the UK-US gap in annual returns averaged between 15 and 20 percentage points, this gap fell to negligible levels after the dotcom crash, and it is no longer statistically significant. Figure 8 also provides evidence on what factors may explain both the existence of the UK-US gap in returns and its convergence in most recent periods. We discuss them one-by-one next.

The first bar in Figure 8 reports the gap in performance for UK and US funds with the same vintage year.⁴⁹ It shows that UK funds underperformed US funds, even after controlling for the year in which the fund was established. Therefore the time of fundraising does not explain the historical gap in returns. The second bar controls in addition for the size of the fund and its investment stage focus, but the magnitude of the gap remains basically unchanged.⁵⁰

We next consider whether these performance differences are driven by the industries in which funds from the two countries chose to invest.⁵¹ We find that two funds with exactly the same industry exposure would still have performed differently across both shores of the Atlantic. Therefore, the UK VC funds historical underperformance is not due to poor industry selection, but to their choice of companies within industries or the underperformance of some industries in the UK.

We continue this exercise including a set of extended controls associated with the fund manager itself, such as its prior experience relative to the other managers in the market, and its strategic choices, such as the number of companies they invest in, the average round in which they first invest in a company, the amount they commit, the number of co-investors in the round (i.e. the extent of syndication), and an index that measures how industrially specialised the fund is.⁵² In a nutshell, the inclusion of these controls does not narrow the gap either.⁵³ Overall, the gap in the earlier periods remains when like-for-like funds are compared.

Finally, we ask whether the historical returns gap is the result of a lack of exits in the UK, or instead less profitable exits. As shown in the last bar in Figure 8, the gap narrows when we control for the shares of IPO exits and M&A exits,⁵⁴ but it still continues to be large and significant.⁵⁵ Thus, both the smaller number of exits and the lower returns from these exits explain the historical low returns generated by UK VC funds.

3.3 The convergence in returns is not the result of changes in the characteristics of UK funds

Figure 8 also provides additional evidence on what factors explain the convergence of returns in the UK and US or, more precisely, which don't. The collapse in the returns gap over time is of the same order of magnitude regardless of whether controls are included or not. Therefore, this suggests that convergence in returns has not been driven by changes in the observable characteristics of VC funds in the UK and the US, such as size, experience or investment strategy. Instead, it is most likely the result of changes in the wider economic environment (or alternatively in some unmeasured VC funds characteristics).

The question that remains is whether this convergence is permanent or will prove to be just transitory. One hypothesis states that the UK VC industry has matured, becoming more connected with the US and hence more exposed to the US market functionalities. Some evidence appears to support this hypothesis, since younger European and US VC firms are less different from each other than their older counterparts, in terms of fund characteristics, investments strategies and contract and monitoring behaviour. ⁵⁶,⁵⁷

In contrast, the alternative hypothesis suggests that performance differences are just not evident yet; but that they will be once the likes of Facebook and Twitter go public (after taking account of the impact of the recent exits of LinkedIn and Skype). The public markets for entrepreneurial companies have been difficult in recent years, so many of the most promising venture-backed firms in the US have chosen to remain privately held. Accounting conventions in the industry often lead to conservative valuations of private firms. Therefore, their eventual IPO may trigger large gains in the US funds which hold these investments, improving as a result the aggregate performance of US funds in most recent periods.

While only time will tell which of the two hypotheses dominates, examining what explains the historical gap may shed some light on what is likely to happen in the future. We turn to this next.

3.4 The wider UK environment was a major contributor to the historical gap in VC returns

The large difference in the returns enjoyed by investors in the US and the UK in the decade prior to the dotcom crash cannot be explained by a large set of controls capturing several characteristics of the funds, their managers and their strategies. In other words, a large UK-US returns gap remains in that period, even if we compare like-for-like funds. Therefore, the UK historical underperformance can only be the result of some unmeasured fund characteristics or the environment in which funds operated (for example, the number of opportunities available and the barriers to their development, the ambition and ability of entrepreneurs, the background of investors, other cultural issues, etc.).

Examining the performance of funds when they invest in other countries can shed some light on whether performance differences are due to different fund manager characteristics or to the investment opportunities that they face.

We first examine how investing more capital into one country as opposed to another is associated with fund performance. We find evidence that the more capital US-based funds committed to UK-based companies as opposed to the US-based companies, the worse the IRR and multiples they achieved.⁵⁸ However, we only find this relationship for funds raised before the dotcom bubble, and not afterwards. In most recent years US funds investing a larger amount of capital in European markets have instead displayed higher returns.⁵⁹

A caveat to this analysis is that we cannot directly attribute portions of the fund performance to investments in any particular country. In other words, we do not know whether the negative performance of a US fund investing in the UK was due to its investments in the UK or those made anywhere else. To overcome this problem we analyse the success that funds that invest in all three markets had at bringing companies to an IPO exit in each region (UK, US and continental Europe). IPOs are by far the most important (and profitable) means for venture capitalists to exit an investment, and have been used as performance measurements in various studies.⁶⁰ The share of portfolio companies that exited through an IPO thus allows us to measure a given fund's performance in one region and compare it with its performance in other regions.

Figure 9 illustrates that both US and UK funds enjoyed the lowest share of IPO exits when investing in UK companies, while they achieved the highest share of IPOs in their US investments. The share of IPOs for UK portfolio companies were also somewhat lower than for continental Europe, despite having a less developed IPO market than the UK.⁶¹ Similarly, UK funds performed well when investing in the US, while US funds' performance worsened when investing in Europe.⁶²

This suggests that the wider UK environment was a major contributor to the UK's low historical returns. It may also help to explain two facts documented in Section 2, namely the limited appetite of US funds to invest in Europe, and the large share of European VC capital invested in the US. And, finally, it raises the difficult yet important question of what are the factors that influence the environment in which young innovative companies operate.

3.5 Differences in returns within countries are larger than between countries

Average returns do not capture the large variability in returns within countries. Figure 10 illustrates the spread of performance records in the UK and the US over time.⁶³ The gap in returns between good and bad performing funds within a country is much larger than the gap in the average returns across countries. This suggests that investors can find good performing funds everywhere, so that selecting the right fund manager may be more important than choosing a particular country.

Looking at the distribution of returns within countries also allows us to examine the historical UK-US returns gap from a different angle. Was the UK's underperformance driven by the extremely good performance of a small number of US-based funds, which experienced exceptionally high returns during the dotcom era? Or was it instead the majority of US funds that performed better than their UK counterparts, and not just a small proportion of them?

The median performance of UK funds was 0.8 per cent while for US funds it was 2 per cent (in others words, 50 per cent of UK funds had returns above 0.8 percent). Thirteen per cent of UK funds would have got into the top quartile in the US, 24 per cent of UK funds outperformed the average US fund and 45 per cent outperformed the median US fund.

Figure 10(b) reveals that the dispersion of returns has been volatile over time. The range of returns for US funds raised between 1994 and 1997 was large, with the vast majority of such funds performing well. More particularly, 75 per cent of all US funds received positive returns, and more than a quarter reached returns above 55 per cent. In contrast the range of returns for UK funds was mostly concentrated below 10 per cent. The dispersion of returns for funds established between 1998 and 2005 has been much lower, concentrated around zero on both sides of the Atlantic. Few funds appear to have done very well or very badly. The fall in the dispersion of returns over time is also clear if we examine the evolution of the interquartile range. The difference in net IRR between top and bottom quartile funds in the US was 40 percentage points pre-bubble and only 14 after. For the UK it was 22 and 13 percentage points respectively.

These comparisons do not take in account the different characteristics of UK and US funds. Since we are interested in estimating the magnitude of the gap at different percentiles of the returns distribution (instead of just the average) we use a quantile regression to compare the returns of like-for-like funds in the UK and the US. We find that in the years prior to the dotcom boom, the very best US funds indeed outperformed their UK equivalents by a whopping 89 percentage points. ⁶⁴ But the returns for the median US fund were also 13 percentage points higher than in the UK (adding additional controls leads to the similar conclusions). Therefore, while the outstanding returns of top US funds account for some of the performance gap in early periods, they do not on their own fully explain the UK's underperformance in that period.

Part 4: Identifying the best-performing funds

The last section shows that, while there are large differences in the average VC returns between the UK and the US, investors can find good performing funds in both countries. The question is how to identify them. Thus in this section we go beyond examining cross-country differences in returns and aim to explore which factors can help predict VC returns once cross-country differences have been accounted for. We consider several potential drivers of returns and use a regression model to examine how their role varies across countries and over time.

Specifically, we investigate the importance of the following variables: the size of the fund; the experience of the fund manager; the fund's industry specialisation; its propensity to syndicate; its past performance; its location in an investor hub; some other characteristics of its investment strategy; and finally, the fund manager's structure (i.e., its number of partners and offices).

We estimate the effect that each of these variables has on the two measures of fund performance (net IRR and multiples), controlling in all regressions for the nationality of the fund manager, the funds' vintage year and its industry and stage focus. We identify a few of these variables which are associated with better performance, whether considered separately or simultaneously with other factors: size, experience, past performance, hubs and earlier rounds. However, we find differences on their impact over time, with some of them having become weaker predictors in recent years.⁶⁵ We next discuss the main findings in detail, and present the full results of the regression analysis in the appendix.⁶⁶

4.1 Small funds underperform medium-sized funds, but larger is not always better

The average IRR for the bottom quartile of funds by size (those with assets under management below $84 million) are about 7 percentage points lower than those achieved by larger funds.⁶⁷ Economies of scale in VC investing are, however, limited. Above this threshold, increasing the size of the fund is not associated with better performance, and in some circumstances it can even be counterproductive.

A larger scale allows funds to cast a wider net, follow-up their most successful investments and spread the costs more widely. But the size of the fund is positively related to the experience of the fund manager itself. Successful fund managers are able to raise larger funds, so it could be experience and not size that is driving returns. In order to test this, we control for the experience of the fund manager and we find that medium-sized funds ($84-$365 million)⁶⁸ continue to have a superior performance in comparison to small funds. However, we also find that larger funds (those above $365 million, the top quartile) no longer achieve better returns than small funds.⁶⁹ Beyond a certain size diseconomies of scale arise, damaging performance. Therefore, while experienced fund managers may be able to raise very large funds, the analysis suggests that their fund returns would have been higher on average if they had kept their funds smaller. Or, in other words, it is medium-sized funds that perform better.⁷⁰

4.2 More experienced fund managers achieved higher returns

There are several reasons why more experienced venture capitalists could be able to deliver better returns. First, experience in screening potential deals helps fund managers select better companies to invest. Second, venture capitalists with more experience may be able to add more value to the firms they fund, for instance due to improved monitoring skills or by helping new ventures set business strategy and attract critical resources.⁷¹,⁷² Finally, more experienced fund managers also have stronger and wider networks. These can be a source of comparative advantage when sourcing deals and looking for co-investors, but also when working with clients and suppliers or when engaging with potential acquirers or underwriters at the time to exit an investment. In fact, a large proportion of the benefits accrued from greater experience can be explained by the better networks more experience fund managers have access to.⁷³

We examine the importance of experience and find that more experienced fund managers deliver higher returns to their investors.⁷⁴,⁷⁵ We measure the experience of a fund manager by the number of prior investments it has made relative to the average for all the other funds with the same vintage year.⁷⁶ However, using other measures of experience, such as the number of previous funds raised by that fund manager (sequence number) or the number of years it has been investing also leads to the same conclusion.⁷⁷ Similarly, experience predicts returns regardless of the controls that we include in the regression.⁷⁸

We also test whether experience is equally important for both US and UK funds, and find this to be case.⁷⁹ However, when we break down the effect of experience by period,⁸⁰ we find that the benefits of experience appear to have faded over time, perhaps reflecting the fact that information on potential deals is becoming easier to access. Alternatively, this could potentially reflect the greater importance of experience in booming periods when opportunities abound, relative to more normal times when the differences between the best and the worst performing funds is much smaller.

4.3 Past performance predicts future performance

We capture the performance of the preceding fund raised by each fund manager and find strong evidence of persistence in returns over time.⁸¹ A fund raised by a fund manager whose previous fund performed well is more likely to exhibit superior performance too.⁸²,⁸³ This sets venture capital apart from some other asset classes, such as mutual funds, where often the returns are more the result of luck than skill, and so display very limited persistence.⁸⁴

When comparing the persistence of returns in the UK and US, we find a significant relationship between past and present returns in the US, but not for the UK. This however could be driven by the relatively small number of UK fund managers for which we have performance data for multiple funds. Finally, we observe a varying degree of persistence across different periods. It was highest during the earliest period (1990-93) and lowest for funds raised during the dotcom bubble (1998-2001), although it increased again in the latest period.⁸⁵

4.4 Funds in investor hubs had better returns

Fund managers located in one of the four largest investor hubs (Silicon Valley, New York, Massachusetts and London) achieved on average 7.4 percentage points higher IRRs than those based elsewhere.⁸⁶ This may be explained by the fact that these regions concentrate a large share of the world's venture capital activity, so funds can benefit from agglomeration economies, both directly (since proximity to portfolio companies and other investors helps minimising information asymmetries) and indirectly (since portfolio companies can have access to better networks and a more developed supporting infrastructure among others).⁸⁷

However, performance varies across hubs and over time. Figure 11 displays the average IRR for funds based in each region relative to the average IRR achieved by US funds located outside the major hubs mentioned.⁸⁸ Funds in Silicon Valley or Massachusetts have significantly outperformed all other US funds. Instead, London's performance has lagged behind that of all US hubs, on par with US funds outside hubs, but still better than for UK funds outside London.

The superior performance of funds based in hubs has faded over time though. It was mostly driven by funds raised in 1994-1997, which were able to benefit from the dotcom bubble, but it has almost disappeared since then. The falling trend is common for all hubs except New York, which on the contrary has seen improving relative performance for funds based there over the last two decades. In 1990-93 New York underperformed the rest of the US while in 2002-05 it outperformed the rest of the US, including all the other hubs. In contrast, in the most recent period the advantage from being located in Silicon Valley, Massachusetts and London is no longer significant.

4.5 Investing in earlier rounds leads to better performance

Fund managers face several strategic choices that may impact fund performance, such as the number of companies to fund, the round at which to invest, the amount of initial funding to commit or whether to syndicate with other venture capitalists. We find that only one of these choices is significantly correlated with performance: funds investing in earlier rounds, admittedly a more risky strategy, performed better than those investing in later rounds.⁸⁹,⁹⁰ This is true both in the UK and the US. This relationship is also quite robust over time (albeit weaker in the earliest period). On average, making the first investments one round earlier is associated with 2.4 percentage points higher returns.⁹¹

Which industries to target is an important investment strategy decision too. Funds can follow a narrow strategy, concentrating on a few sectors, or a wide strategy, spreading their investments across many sectors. We test the impact of specialisation with an index that measures how concentrated the funds' portfolio is on a few sectors,⁹² but we do not find a robust relationship between specialisation and fund performance.⁹³ This is not to say that all sectors perform equally. There are large differences in the historical returns generated by different sectors. Specifically, funds with a larger share invested in internet and computers, communications and electronics, and biotechnology and healthcare delivered the highest returns over the study period.⁹⁴ But the extent to which concentration of the portfolio matters seems ambiguous.

4.6 A larger number of partners was associated with higher returns

We also investigate the relationship between performance and the number of partners and offices that the fund has. Venture capital organisations have on average five general partners and two offices, but there is a wide variation among them.⁹⁵ Most venture capital general partners have a decade or more of experience.⁹⁶ However, any given partner can only monitor a certain number of deals and effectively serve on only a limited number of boards, so a larger number of partners may be an advantage.

We find that funds with more partners outperform, even after controlling for fund size and experience among others.⁹⁷ However, the positive effect associated with a higher number of partners has been eroded over time. A larger number of offices are also correlated with better performance, but very weakly and only in the earliest period.⁹⁸ Altogether, it appears that having more resources, whether partners or offices, may help the fund manager take advantage of more opportunities during boom periods (as prior to the dotcom bust), but it does not make a difference in non-boom times.

4.7 Much of the variability in returns is not explained by these factors

A note of caution to conclude this section, the model that we have estimated is successful at identifying several factors that help to explain performance. But differences across funds regarding these factors can only account for about 30-40 per cent of the variation in returns.⁹⁹ In other words, much of the variability of returns continues to be unexplained. This is most likely accounted for by a combination of unmeasured (or unmeasurable) factors and serendipity. After all, investing in new ventures is, and will continue to be, an uncertain business.

Part 5: Public interventions to support VC funds

Governments around the world have taken a strong interest in facilitating access to finance for innovative high-growth companies, and the UK has been no exception. Several schemes in support of the venture capital industry have been set up in response to the belief that there are significant funding gaps not being addressed by the market, particularly for small high-technology start-ups or in particular regions.¹⁰⁰ The attempts to stimulate the supply of new sources of finance have followed different approaches, from ensuring that each region has access to regional-based VC funds to trying to demonstrate that investors in early-stage funds can make robust returns, thereby promoting the private sector venture capital industry.¹⁰¹ Many past UK interventions have fallen foul of a few common problems: trying to achieve too many goals; being sub-scale; limiting the pool of potential investments; and having unrealistic time horizons.¹⁰² However, recent government schemes (i.e. ECFs) have avoided many of these pitfalls and are aiming to make use of the best of private sector experience and establish a credible policy in the area.

Despite their growing importance in the market,¹⁰³ the evidence on the performance of UK government VC schemes is limited. A recent evaluation by the National Audit Office (NAO)¹⁰⁴ concluded that the financial performance of public funds had been very poor (although not untypical when compared with private VC returns over the same period), and pointed to poor design and geographical restrictions on the funds' investment activity as factors to address. But while the financial performance of the UK public funds has not been good, their impact on supply of finance may have been significant, since many recipients of public funding claimed that in the absence of the schemes they would have not undertaken their investments.¹⁰⁵ International studies also provide a mixed picture. For instance, a recent report by the World Economic Forum found that companies with moderate government VC support outperform (in terms of value and patent creation) both companies with only private venture capital support and those with extensive public support. This effect was however stronger for programs associated with national governments (and international organisations) rather than sub-national entities (e.g. state and provincial governments).¹⁰⁶

5.1 Publicly backed funds have delivered lower returns than private funds, but the gap has narrowed

In this section, we assess the financial performance of publicly backed venture capital funds.¹⁰⁷ Public schemes are generally set up to support investment in the least profitable segments of the market, where private VC investors will not go. In addition, they are often motivated by spillover effects and other non-commercially driven outputs. Therefore, the financial returns generated by publicly supported funds are a useful but incomplete metric to assess the success of these interventions. Also, when evaluating their returns, they need to be benchmarked with private funds operating in the same space, since otherwise there is the risk of underestimating their performance.

We analyse performance records of 93 (50 UK and 43 US) publicly backed funds¹⁰⁸ raised between 1990 and 2005, which have made over 3,100 investments. Our coverage of public funds is limited, but we do have access to very detailed information about their investment activities, and thus we can compare their performance with that of solely private funds with the same characteristics. This is particularly useful because the average public and private fund are different in several dimensions. As reported in Table A4 in the appendix, public funds tend to be smaller, undertake fewer investments in fewer companies, and invest smaller amounts. They also have fewer exits and achieve markedly lower returns, about 10 percentage points lower IRR than the average private fund in the UK and 6 percentage points lower in the US.¹⁰⁹

We next use the same regression approach as in previous sections to examine what factors may account for this public-private gap, as well as to estimate the real magnitude of the gap when like-for-like funds with the same characteristics are compared. Table A11 reports the full results of this analysis, summarised in Figure 12.

In line with previous studies we find that publicly backed funds historically underperformed private funds, particularly in the UK.¹¹⁰ However, the public-private returns gap shrinks when we control for vintage year,¹¹¹ which indicates that many public funds were established in years preceding poor market performance. Still, large differences in returns remain for some periods even after we control for vintage year.

Publicly backed funds in the pre-bottom bubble period heavily underperformed private funds, with average returns for public funds being up to 20 percentage points lower in the UK. The results suggest that in later years governments have become savvier when designing new schemes.¹¹² Based on the regression analysis, there may still be some margin for improvement as in the latest period there continues to be a gap in the returns between public and private funds, even if it is small and only statistically significant for the US.

5.2 The public-private gap in returns is not fully explained by observed fund characteristics

Observers have often pointed to the small size of many public funds as a culprit for their underperformance. But it is not just fund size that explains the difference in returns. In the earliest period the public-private gap decreases slightly after we control for size, but it continues to be large and significant.¹¹³ Controlling for stage, another factor commonly cited given publicly backed funds' usual focus on early-stage, does not alter the results either. Similarly, not much changes when we add industry controls.¹¹⁴ It is not the industries that public funds choose to target, but their selection of companies within industries (and what they do with them) that leads to underperformance. Adding additional controls, such as experience of the fund manager, industry specialisation and other measures of investment strategy reduces the historical public-private gap for the US, but not for the UK.¹¹⁵

Many publicly backed funds had a regional focus that may have prevented them from locating in hubs and taking advantage of the benefits associated with them, but controlling for investor hubs does not eliminate the gap. Controlling for the proportion of companies that a given fund successfully exits also does not explain the public underperformance.

If the historical public-private returns gap is not accounted for by these factors, what does explain it? One hypothesis is that the restrictions that publicly backed schemes impose on fund managers may hinder their performance. An alternative hypothesis is that governments may fund less capable fund managers, either because they are unable to distinguish good from bad fund managers or, alternatively, because only managers that cannot raise funding in the market seek support from government schemes.¹¹⁶ Whatever the answer, the narrowing of the public-private performance gap in most recent years suggests that this may have become less of an issue.

A question that remains is what has been the role of public funds in explaining the historical UK-US returns gap. While the US has more government schemes to encourage VC than many realise, the UK has been particularly active in using public funding to support the VC industry. The design of the schemes has also been different. We find that the underperformance of publicly backed government funds is one of the factors that contribute to the aggregate UK-US returns gap.¹¹⁷ However, the US continues to experience higher historical returns than the UK when we exclude public funds from the sample.

Footnotes

Part 6: Policy implications

This report examines the performance differences between UK and US venture capital funds. UK funds have historically underperformed US funds but this gap has narrowed to one percentage point (net IRR) for funds raised in the post bubble period. The historical performance gap is robust to different measures of venture capital activity, fund characteristics or investments strategies, suggesting that the UK environment was a major contributor to this gap.

The evidence shows the progress that the UK has made over the last two decades, but it also highlights the important challenges that lie ahead. Whether the UK venture capital industry will be able to match (or, why not, surpass) the performance of the US venture capital industry in the next decade will depend on the decisions taken by investors, fund managers and policymakers among others. The findings in this report can help inform investors' and fund managers' decisions, but they have particularly important implications for UK policymakers:

• Remember venture capital activity does not exist in a vacuum. Venture capitalists are tremendously dependent on their partners. Without entrepreneurs who conceptualise visionary businesses, experienced lawyers able to negotiate agreements, skilled marketing gurus and engineers, and customers who are willing to take a chance on a young firm, success is unlikely. The evidence shows that the environment that UK start-ups faced was a major contributor to the UK historical gap in VC returns. While there are encouraging signs that the UK is becoming a better place for entrepreneurial ventures, this cannot be taken for granted. Efforts to improve the conditions faced by those young innovative companies that could become the giants of tomorrow should be stepped up.
• Resist the temptation to overengineer public support schemes. In many instances, government requirements that limit the flexibility of entrepreneurs and venture investors have been detrimental. It is tempting to add restrictions on several dimensions: for instance, the locations in which the firms can operate, the type of securities venture investors can use, and the evolution of the firms (e.g. restrictions on acquisitions or secondary sales of stock). These types of restrictions have been present in a number of past UK public programs, and probably can help explain their large historical underperformance. More recent programs appear to have mostly overcome the temptation to micromanage the entrepreneurial and venture capital process, but pressures to do so in the future should be resisted.
• Avoid initiatives that are too small. Policymakers must be sure that their venture initiatives have critical mass. Too small a program will do little to improve the environment for pioneering entrepreneurs and venture funds. It is also likely to lead to poor financial returns (i.e. small funds tend to underperform), which may create a backlash that impedes future efforts. Too often, UK efforts have led to the creation of undercapitalised funds which do not have the staying power to back their companies through development.

Over the coming months NESTA will continue its work to develop further evidence on what measures policy makers can take to improve the environment for UK high-growth companies.

Appendix 1: Data sources and definitions

Table A1: Data sources and definitions

Fund characteristics: Preqin provides information on fund characteristics such as fund size, type, location and focus. In order to provide accurate information on fund characteristics (including management characteristics) we captured individual investment deals made by each fund. We used three different databases to do so. Thomson One Reuters was our primary source of individual investment deals data. Thomson One captures investments deals for over 14,000 US, UK or European private equity funds. Investments deals for the funds we have performance records for were captured using the Thomson One database. This was supplemented with data acquired from VentureSource Dow Jones (when funds were not included in Thomson One database). These two databases (fund returns and fund individual investment outcomes) were merged and any discrepancies between the two were reconciled. These related to a minority of cases where the sources reported a different vintage year, size or stage focus for a fund. To find the true values we consulted other sources or contacted the fund managers themselves.

Fund type: Our study sample contains funds that are classified by Preqin as seed-stage, early-stage, early-stage: seed, early-stage: start-up, venture (general), expansion, development, later-stage, balanced and mezzanine.[^122] We aggregate these into five new categories: i) 'early-stage' which includes all seed-stage, early-stage, early-stage: seed and early-stage: start-up funds; ii) 'generalist' which includes all venture (general); iii) 'expansion' which includes all expansion, development and later-stage funds; iv) 'balanced' funds; and v) 'mezzanine' funds. Again with this variable, some discrepancies appeared between what our two main sources were reporting. We again endeavoured to attain more information on what the correct stage focus for each fund is by checking other external sources such as industry experts, as well as contacting the funds themselves through phone calls or emails. Using the combination of the information received from these sources and our knowledge of what investments the fund made, we are confident we have a variable that accurately reflects the investment focus of the funds in the sample.

Fund size and vintage year: Two main sources were used to locate data on the size of (amount of capital raised by) the VC funds. These sources were Thomson Reuters and Preqin. Discrepancies appeared between the sizes reported by the two sources so the following actions were taken to reconcile the figures.

1. If the difference between the fund sizes was less than 5 per cent of the larger estimate, using the home currency the amount was reported in from Preqin, we use the value in US dollars from Thomson.
2. If the difference between the fund sizes was greater than 5 per cent using the same criteria, we search for a third source, e.g. VentureSource, fund website, to confirm the value given by one of the main sources. The criteria for confirming a fund size using a third source is, again, a difference of less than 5 per cent of the larger estimate in the reported currency.
3. On the rare occasion where two or more sources confirm a size not within 5 per cent of the size reported in Thomson or Preqin, this confirmed size is used.
4. Where the size in the reported currency being used comes from a source other than Thomson, the Thomson exchange rate corresponding to that fund is used to convert the size into US dollars. This is only necessary when the reported currency isn't already US dollars and is due to the investment sizes corresponding to these funds coming from Thomson and using the same rates. If the Thomson exchange rate for the fund is unavailable, the Preqin rate is used.
5. In instances where the fund size is reported by only one of Preqin and Thomson, this size is used.
6. All amounts are converted to 2010 US dollars (and so they are adjusted for inflation).
7. Funds with assets under management less than $1 million have been excluded from the analysis (due to inconsistencies in the data).

Similarly discrepancies were found relating to the vintage years reported by the two main sources. These were reconciled as follows:

1. If a difference of only one year was found between the sources, the year from Thomson is used. If a difference of more than one year is found, a third source is found to confirm either of the vintage years reported by Thomson or Preqin.

Partners and office data: We use information from three directories: Pratt's guide to Private Equity and Venture Capital sources, Galante's Venture Capital & Private Equity Directory and information from the Magdalen Centre at the Oxford Science Park, and from the European Venture Capital Association, in order to collect data on GP partners and offices. Our intention was to analyse the effect of partners/offices for the period in which the fund was active. The difficulty this presented was obtaining historic data on the number of partners and offices the GP had on the fund's vintage year (since accessing not current data was enough).

Performance data: The two most commonly used performance metrics are fund multiple, which measures the total distribution relative to the total takedown, and the internal rate of return (IRR), which takes into account the time value of money (McKenzie and Janeway 2008,[^123] for a critique see Phalippou 2009[^124]). The most commonly used source of VC fund performance data in the literature is Reuters Thomson One. The principal shortcomings of these two databases have been outlined by Ljungqvist and Richardson (2003):[^125] (i) the performance data is largely provided by VC firms on a voluntary basis and thus potentially subject to selection biases; and, (ii) the data is based on unrealised as well as realised investments, which introduces noise and potentially biases due to subjective accounting treatment.

The multiple reports the return received by the fund's LPs compared to the amount of capital they committed to the fund. A weakness of this measure is that it does not take account of the time value of money, i.e. it does not account for the length of time it took for the fund to deliver the return to the LPs. Net IRR overcomes this by using discounted cash flow data to take account of both the return the LP receives and how long it took to be realised therefore making it a more accurate measure. Net IRRs are reported net of fees and carried interest and include returns from dividends.

The analysis of performance records represented an empirical challenge. Several funds report either Net IRR or Multiples (but not both) and in order to conduct a meaningful analysis the missing value (Net IRR or Multiple) was imputed using a regression analysis with the reported value (Net IRR or Multiple) and controls for industry, country and vintage year. The resulting measures of returns are winsorized at the 1 per cent level to reduce the impact of outliers on the estimation.

The primary source of VC fund performance data was the Private Equity Intelligence database (Preqin). Preqin contains performance data for over 4,000 private equity funds based in the US, UK or continental Europe, from which over 1,000 are operating in the VC market. Their performance records are attained either directly from the LP and/or GP of the fund or calculated internally by Preqin based upon cash flows and valuations, provided for individual partnerships. We supplemented this with performance records from other commercial databases such as LP Source Dow Jones, EurekaHedge, Pitchbooks and desk research.

LP Source is a newly established (by Dow Jones) research tool that provides data on active fund managers, their investments strategies and preferred industries. Although the dataset is mainly focused on the US market, a tailored dataset with performance data for UK funds was made available to the researchers. EurekaHedge provides information on the global alternative fund industry and contains over 20,000 funds across all assets class. Its European Private Equity Database provided us with a number of VC fund performance records that were not available in the other databases used.

Performance records for quoted funds (i.e. VCTs) were acquired by the Tax Shelter report published by the Allenbridge Group plc, an independent UK tax efficient investment consultancy.[^126] Venture Capital Trusts (VCTs) are companies listed on the London Stock Exchange, and are similar to investment trusts. They are run by fund managers who are usually members of larger investment groups. They invest in a range of small higher-risk companies whose shares and securities are not listed on a recognised stock exchange, by investing through VCTs. However, companies whose shares etc. are dealt in solely on the Alternative Investment Market (AIM) of the London Stock Exchange or on two of the Plus Markets are regarded as unquoted companies.[^127]

Appendix 2: Tables

Table A2: Summary statistics - All funds (UK, US and continental Europe)

Note: The last six columns report whether there are significant statistical differences in the means across countries (positive or negative). +,++,+++ (-,--,---) indicate significance at the 10 per cent, 5 per cent and 1 per cent level respectively.

Table A3: Summary statistics - Performance funds 1990-2005 (UK and US)

Note: The last three columns report whether there are significant statistical differences in the means across countries and samples (positive or negative). +,++,+++ (-,--,---) indicate significance at the 10 per cent, 5 per cent and 1 per cent level respectively. Perf and Non-Perf refer to the sample of firms for which performance data is available and unavailable respectively. A dot indicates missing data in the non-performance sample.

Table A4: Summary statistics - Public vs. Private funds (Performance funds 1990-2005 in the UK and US)

Note: The last four columns report whether there are significant statistical differences in the means between public and private funds in the UK and the US (positive or negative). +,++,+++ (-,--,---) indicate significance at the 10 per cent, 5 per cent and 1 per cent level respectively.

Table A5: UK-US performance gap over time (with and without controls)

Note: Columns 1 and 2 look at how controlling for fund vintage year affects the historic underperformance of UK funds. From column 3 onwards we analyse how the performance changed over time, which allows us to observe the convergence in performance. We then add additional controls to examine how they affect our convergence findings. Size variables refer to quartiles of the size distribution with the bottom quartile omitted as a reference quartile. The regression models in this table have been estimated with ordinary least squares (OLS). The table reports standard errors in parentheses clustered at the fund manager level. *, indicate statistical significance at the 10 per cent, 5 per cent and 1 per cent level respectively. The right panel reproduces the same regressions but considering multiples instead of net IRR as dependent variable.

Table A6: Investments location and fund performance

Note: Columns 1 and 2 examine how investing in different regions as opposed to the US affected the performance of both US and UK funds. Columns 3 and 4 look at how this evolved over time. Size variables refer to quartiles of the size distribution with the bottom quartile omitted as a reference quartile. The regression models in this table have been estimated with OLS. The table reports standard errors in parentheses clustered at the fund manager level. , *, *** indicate statistical significance at the 10 per cent, 5 per cent and 1 per cent level respectively. The right panel reproduces the same regressions but considering multiples instead of net IRR as dependent variable.

Table A7: Distribution of returns (Quantile regression)