Looking for the premium

Academics and practitioners seem to be disagreeing about the level of outperformance of the private equity asset class. Whereas the former are often arguing that there is no premium once the fees to the manager have been taken into account, the latter are strong defenders of a high premium over the public market. What is even more worrying is that both parties seem to be able to make their point with convincing statistics and beautiful charts, planting doubt in the minds of many investors.

In this article we try to solve this conundrum by analysing the risk premium associated with private equity in different contexts. We first start with an investigation at single fund level, which shows that the “average” fund actually does not beat the public market, making the point of many academics. Then we analyse what happens at market level, i.e. when all cash flows from the funds considered previously are pooled together. This second analysis speaks in favour of a premium associated with the private asset class, as supported by practitioners. Finally, given that investors are neither investing in one single fund, nor in the entire market, we repeat our analysis with “diversified” portfolios. These diversified portfolios are meant to represent investment programs somewhere between those two extremes, to investigate cases which are relevant to investors.

Our findings are the following. Though pooled results clearly indicate the existence of a premium for both buyout and venture capital funds, there is a fundamental difference between the two strategies. The outperformance of the buyout portfolios stems from the positive skewness of distribution of returns for this strategy, and hence may be achieved by diversification. For the venture capital funds the outperformance of the pooled sets is actually not due to a diversification effect, but to the fact that historically large funds have clearly outperformed small funds. This size bias does not necessarily result from the ability of managers of large funds to produce higher returns, but may be from the timing at which the large funds have been raised, as it was more favourable compared to venture capital.

Table 1: Result of the public market equivalent analysis performed at individual fund level for the three different data sets. As the median private equity fund IRR is lower than the median PME+, one deduces that the “average” private equity fund does not out-perform the public benchmark. As only about 40%-45% of the funds do actually beat the benchmark, the median premium is also negative in all cases.

Sample Vintages # Funds Benchmark Median IRR Median PME+ Median Premium % Out – performing
BO Global 1978-2005 616 S&P500 TR 7.2% 9.3% -175 bps 45%
BO Mature 1980-2000 512 S&P500 TR 7.8% 8.0% -106 bps 47%
BO Liquidated 1980-1995 237 S&P500 TR 12.3% 14.8% -277 bps 41%
VC Global 1969-2005 1150 NASDAQ 4.4% 10.6% 433 bps 39%
VC Mature 1980-2000 1023 NASDAQ 5.0% 10.7% -375 bps 40%
VC Liquidated 1980-1995 634 NASDAQ 8.7% 12.9% 460 bps 37%

Table 2: Result of the public market equivalent analysis performed at pooled level with the three different data sets. The difference between the pooled private equity fund IRR and the pooled PME+ is a measure of the premium achieved by investing in the private asset class at market level. Depending on the sample and on the strategy, the premium ranges from 200bps to 400bps.

Sample Vintages # Funds Benchmark Pooled IRR Pooled PME+ Pooled Premium
BO Global 1978-2005 616 S&P500 TR 12.3% 8.8% 350 bps
BO Mature 1980-2000 512 S&P500 TR 11.9% 8.8% 310 bps
BO Liquidated 1980-1995 237 S&P500 TR 17.2% 14.8% 240 bps
VC Global 1969-2005 1150 NASDAQ 16.0% 12.0% 400 bps
VC Mature 1980-2000 1023 NASDAQ 15.7% 12.1% 360 bps
VC Liquidated 1980-1995 634 NASDAQ 16.4% 14.7% 170 bps

BUYOUT FUNDS
In order to find out if a size bias is at the origin of the premium measured on buyout funds, we split the fund sample in four categories depending on the fund size: $0 to $250 million, $250 to $500 million, $500 to $1000 million and larger than $1000 million. Then we pool all cash flows by category and perform the PME+ analysis at pooled level for all three Global, Mature and Liquidated samples. The results are summarized in Figure 3, which represents the premium as a function of the size category. In each case the premium turns out to be positive. It is also worthwhile noting that the premium associated with the smallest funds, namely $0 to $250 million is about twice the size of the premium of the three other categories. This clearly shows that a size bias is not of the origin of the outperformance of the pooled cash flows at market level.

The origin of the outperformance of the smaller funds is to be found in the timing over which these funds have been raised and the growing fund size over time. A large percentage of these funds have been raised in the 1980s, years where the buyout industry delivered its highest returns. Over time general partners have been able to raise larger funds, which have been performing quite well, but with a lot more dispersion and not always as successful as over the previous decade. This explains why the premium is larger for smaller funds on Figure 3, which does not reflect this timing element.

Having verified that the premium is not due to a size bias, we still need to check whether it can be explained by diversification. For that purpose, we set a Monte-Carlo simulation where a given number of funds are selected at random within the sample. We repeat the simulation 10,000 times and look at the median of the distribution of premiums as a function of the number of funds in the portfolio. For a single fund portfolio, the result of the simulation is the same as when looking at the statistics for single funds. For very large portfolios, the result will be different than the pooled results at market level, because independently of its size each fund gets the same weighting within our portfolio. By increasing the number of funds in the portfolio, we are then in a position to measure the influence of diversification on the premium achieved over public.

Figure 4 represents the premium as a function of the number of funds in the portfolio for all three samples. The result for one fund portfolio matches the result obtained at single fund level and indicates a negative premium. Increasing the number of funds clearly reduces the underperformance of the portfolio. The break-even point seems to happen around portfol ios comprising six to seven funds. Portfolios with ten funds and above show a positive premium. This simulation clearly demonstrates the key role that diversification is playing when investing in buyout funds: diversification reduces idiosyncratic risk and enhances returns due to the positive skewness of the distribution of returns.

This result is highly relevant to investors, as it highlights that the origin of the premium for buyout funds is the strong positive skewness of returns, which is directly exploitable by diversifying investments.

VENTURE CAPITAL FUNDS
We repeat the analysis for venture capital funds, so as to understand whether the mechanism at the origin of the premium is the same as in the case of the buyout funds. We split the universe of venture funds in to four categories driven by the size of the fund. As venture funds tend to be generally smaller than buyout funds, we use slightly different categories: $0 to $50 million, $50 to $100 million, $100 to $250 million and larger than $250 million.

Figure 5 represents the premium as a function of the size category for venture capital funds. The result is quite different from what was obtained for buyout funds, with this time a negative premium for the smallest funds and a large positive premium for the larger funds. This shows that large funds did actually outperform the benchmark whereas small funds did not. It is then obvious that, when pooling all categories together, the overall premium at market level is going to be dominated by the contribution from the larger funds and will turn out to be positive. This clearly indicated that size bias is at least one of the effects at the origin of the outperformance of the pooled data.

To investigate the effect of diversification on the premium, we run our Monte-Carlo simulation with the venture capital fund data. Figure 6 stands for the size of the median premium as a function of the number of funds in an investor portfolio with equal allocation. Surprisingly, results are significantly different from the results obtained for buyout funds. The premium for the one fund portfolio is again the same as the one in the case of the single fund investment by definition, and hence is also negative. Increasing the number of funds in the portfolio definitely improves the overall performance, but fails to turn the premium into positive territory, except for the VC global sample, where a portfolio with more than 20 funds starts showing some positive premium over public market.

The interpretation of the simulation is less straightforward than in the previous case. As for buyout funds, we know that the distribution of returns is positively skewed and hence that diversification will improve the odds of achieving better returns. What is different with venture capital is that although the improvement effect is there, it is not large enough to turn the premium positive.

“The difference in multiple becomes huge: 16.4x multiple with private equity as opposed to 9.6x multiple when investing in the public index.”

Having shown that the outperformance at market level is mainly due to size bias, it might look like the immediate consequence for investors is to prefer large venture capital funds to small funds, as they lead to better returns. Our analysis however does not take into account the time dimension of the investments and the growing average fund size. Figure 7 represents the number of funds in each size category as a function of the vintage starting from 1978 to 2005. What the figure shows is that a majority of the small funds between $0 and $50 million have been raised during the 1980s, whereas the majority of the funds above $100 million have been raised during the 1990s. This means that the size of the fund is probably only an indirect cause of the outperformance of the large funds. Put in different terms, it might imply that managers of large funds have partly beaten the market because of the timing at which the funds have been invested, and not necessarily because of the skill of those managers.

CONCLUSION
Private equity is a long term asset class, and the disadvantages of holding an illiquid position are actually compensated with a premium over public market. We have shown that the average premium that an investor can expect with a diversified private equity portfolio has been historically of the order of 300 bps.

There is a fundamental difference between buyout funds and venture capital funds. Whereas diversification is critical to improve returns and beat the market in the case of buyout funds, it is only one of the elements which help achieving better returns in the venture capital case, where a strong historical size bias has been observed.

The impact of a 300 bps premium is far from negligible. Assuming an average 12 percent return for the public market – the average S&P 500 total return over the last 20 years – and compounding interest over the same 20 years timeframe the difference in multiple becomes huge: 16.4x multiple with private equity as opposed to 9.6x multiple when investing in the public index.

This simple computation highlights the benefits of investing in the asset class for investors who can afford to hold positions over a very long horizon. Furthermore, these results have been obtained under the assumption of random fund selection, whereas experience shows that thorough due diligence broadly increases the chances to invest with best performers and hence to achieve a stronger premium.

Christophe Rouvinez is a managing director of Capital Dynamics, one of the world's largest private equity asset managers, with an exclusive focus on private equity. Capital Dynamics has offices in Zug (Switzerland), Birmingham (UK), London, New York, San Francisco and Hong Kong.