Enhancing private equity returns via diversification

Everybody knows that diversification decreases risk. But what if diversification could also enhance returns? For private equity investors, this question may not just be very relevant but actually have a positive answer. Christophe Rouvinez and Thomas Kubr suggest that diversification of private equity investments can indeed be a great way to achieve the returns the asset class promises.

Investments in private equity have never been for the faint hearted. Although large gains can be achieved, losses are common and severe. In this article the return characteristics of investments into single private equity funds are assessed and the effect of diversification on these returns is shown. The significance of these findings is considerable:

  • • Investments in single funds are highly risky. The probability of achieving outsized returns is surprisingly low and the most likely outcome is less than twice your money back.
  • • Diversification leads to higher median returns and significantly increases the most likely investment returns.
  • • Diversification substantially reduces the probability of low performance (bottom quartile or worse) while only slightly reducing top quartile returns.
  • • Sufficient diversification is achieved with 20 – 30 funds.
  • Individual private equity investments are risky
    There are many ways to measure private equity fund performance. The most common are internal rates of return (IRR) and return multiples. The following analysis focuses on the return multiple, defined as the sum of distributions and net asset value relative to the total capital contribution to a fund. The analyses and results can also be generated using IRR, but this is at the expense of transparency due to the more intensive computations involving individual cash flows. The analysis is based on return multiples to keep the message simple and the computations easily verifiable. Furthermore, the return multiple of a portfolio of fund investments reduces to the weighted average of the underlying return multiples, which simplifies the computations at the portfolio level.

    Managers' skills seem to be responsible for a significant component of the return

    The reference universe considered for this illustration consists of the 118 US venture capital and buyout partnerships in the Venture Economics database* that were formed from 1990 to 1992 (vintages 1990, 1991 and 1992). The corresponding return multiples as per September 30, 2002 are used as indicators of longterm performance. The 10- to 12- year period has been chosen to be representative of the long-term nature of the asset class. A set of three consecutive vintages has been selected to provide a significant statistical sample. It also serves as a reasonable proxy for a fund of funds.

    Figure 1 represents the histogram of multiples for this universe. It is striking that 18 out of 118 partnerships show negative performance (in other words, have a multiple smaller than 1.0x) even over such a long horizon, confirming the high-risk nature of individual investments in the asset class.

    Large gains from good funds more than compensate for the loss incurred with bad funds

    At the other extreme, six funds are showing multiples higher than 5.0x (last histogram bin), indicating a very large, fat tail on the positive side. The vast majority of the funds are reporting multiples between 1.0x and 3.0x. The average equals 2.28x and reflects the performance of the overall private equity market over the selected period.

    The large dispersion of performance is symptomatic of alternative asset classes, where managers' skills seem to be responsible for a significant component of the return. Such dispersion is also indicative of a high level of idiosyncratic or non-systematic risk, meaning that a large component of the risk can be diversified away by investing with several managers.

    Table 1: Average and standard deviation of return multiples for diversified investments in 1, 3, 10 and 30 funds.

    1 Fund 3 Funds 10 Funds 30 Fund;
    Average multiple 2.28x 2.28x 2.28x 2.28x
    Std Deviation 1.82x 1.05x 0.58x 0.34x
    Source: Venture Economics, Capital Dynamics

    Table 2: Multiples for diversified investments and single funds*

    Vintages 84/85 85/86 86/87 87/88 88/89 89/90 90/91 91/92
    Top Quartile Diversified Single fund 2.17x 2.38x 2.30x 2.40x 2.14x 2.38x 2.09x 2.40x 2.09x 2.39x 2.24x 2.60x 2.39x 2.75x 2.61x 2.53x
    Median Diversified Single fund 1.90x 1.55x 2.02x 1.80x 1.94x 1.68x 1.91x 1.69x 1.92x 1.69x 2.02x 1.68x 2.16x 1.86x 2.28x 1.91x
    Bottom Quartile Diversified Single fund 1.72x 1.18x 1.84x 1.29x 1.76x 1.16x 1.74x 1.12x 1.75x 1.15x 1.83x 1.09x 1.95x 1.11x 2.03x 1.41x
    * Diversified means 20 funds selected at random from two consecutive vintage years between 1984 and 1992.
    Source: Venture Economics, Capital Dynamics

    Funds of funds require proper benchmarking
    As a further direct consequence of the quartile analysis presented in Figure 2, fund of funds should not be benchmarked against the quartiles of individual funds. Comparing the quartiles for a single fund investment with an investment in a portfolio of 30 funds on Figure 2, one immediately notices that bottom quartile fund of funds perform better than median single funds. Thus, comparison with the quartiles of individual funds only makes sense to demonstrate the superiority of a diversified strategy in reducing downside risk, but delivers very little information on the performance of the fund of funds manager.

    The diversification mechanism reduces both the positive and negative tails

    Proper fund of funds benchmarking requires comparison with fund of funds performance. Where no fund of funds data are available, one could rely on synthetic fund of funds returns generated through Monte Carlo simulation, where the underlying funds are drawn at random while respecting the composition of the original fund. As an example, Table 2 provides a comparison between the quartiles of single fund investments and diversified investments over two consecutive vintages. This procedure is non-trivial but provides information about the ability of the fund of funds manager to assemble good funds together.

    The mean, the median… and the mode
    A third statistical measure often used as an attempt to characterise a probability distribution with a single number is the mode. The mode is defined as the most probable event. It is especially relevant for this analysis, as it represents the most likely performance with a single trial. Here the mode is approximated by the value corresponding to the maximum on the histograms on Figure 3. Whereas the histogram for a single fund peaks around 1.85x with a probability of 8 per cent, the histogram corresponding to 30 funds has a maximum around 2.15x with a probability of 12 per cent, indicating a greater chance to achieve better performance with a single fund of funds investment. As Figure 3 illustrates, the diversification mechanism reduces both the positive and negative tails, and also shifts the bulk of the distribution to the right for the benefit of the investor.

    Conclusions
    The high-risk nature of private equity fund investing actually lies in a large non-systematic risk component that can be reduced when running a diversified programme or when investing in fund of funds. Statistical measures clearly demonstrate the superiority of a diversified private equity fund investment strategy. Beyond the mathematics, private equity naïve diversification works because the high gains associated with a few funds more than compensate for the reduced gains and losses of the others. By diversifying investments in multiple funds, one can simply reduce nonsystematic risk and simultaneously increase the chance to capture some of the high-performers.

    Christophe Rouvinez and Thomas Kubr are partners in Capital Dynamics, an asset management firm specialized in private equity with offices in Zug, Switzerland and New York.