Originally published January 17, 2018
In this, our private equity benchmarking deep dive, we have taken a methodical approach to better understanding how investors make use of benchmarks for their private equity program(s). First, we surveyed institutional investors to learn how they viewed their private equity returns in context. Next, we explored differences between selected, major benchmarks and implications for how investors make use of them. We further investigated how index composition and leverage considerations affect the suitability of an index’s use as a benchmark. Finally, we outline a series of principles to hold when selecting an appropriate private equity benchmark.
Comparing returns in private equity is not as straightforward as it might first seem – IRRs and multiples are a good measure but not perfect.
To see how institutional investors view their returns in context, we surveyed over 100 public pension plans to understand their choice of public targets (Figure 1).
The Russell 3000 (market-weighted US large cap) is by far the most popular, with only 3% using the Russell 2000 (US small cap). This was a somewhat surprising finding as we feel a small cap index might be more representative of the underlying holdings of any given PE program (even with a healthy allocation to mega-buyout); we’ll be exploring this more below.
Institutions are about evenly split when it comes to benchmark adjustments (adding bps to base returns) – those pegging their returns against the S&P 500 clearly expect their managers to outperform by a decent margin (at minimum).
We also wanted to assess the use of blended benchmarks (separate weightings of say, an equity index and a debt index). Given portfolio diversity (and leverage risk) within any alternatives program, we had anticipated greater usage. That less than 1/5 of institutions rely on a blended benchmark may be the result of the analytical complexities of assembling a blended benchmark using only Excel.
We’ve spent a lot of time thinking about public market equivalent (PME) returns, and Chronograph’s platform offers investors a number of ways to measure their performance, including extremely flexible and painless benchmark blending.
We wanted to explore some of the differences between selected benchmarks and how investors make use of them for portfolio construction.
Drawing on our prior survey of over 100 US public pension plans, approximately 20% use a pure private equity benchmark (Cambridge and State Street All-PE were the most popular). As you can see, the annualized 10 year return average for the private set is materially higher – almost twice that of the public indices’ average.
While private markets have consistently outperformed public markets over the past two decades – and should, given the illiquidity and leverage risk of the asset class – we wanted to drill down further on this large difference.
We deconstructed underlying sector exposures, revealing some notable differences (Figure 2); namely significantly greater exposure to the higher-beta technology and highly-cyclical materials sectors. Given post-financial crisis bull markets, it’s not necessarily surprising that PE indices have outperformed consistently, with some additional outperformance resulting from this underlying compositional difference.
Because the PE benchmarks are constructed using a private dataset, we weren’t able to perform a perfect look-through decomposition of the indices, but the aggregate figures line up nicely with Cambridge’s own deconstruction as well as Preqin total deployment data.
We exercise caution with any of the ‘public’ private equity datasets such as Preqin, Cambridge Associates, and others for several reasons: (1) inherent selection biases (in the case of Preqin, overweight mega- and large-buyout given their reliance on large public pension disclosures), (2) mixing of gross/net performance, and (3) the pitfalls of voluntary manager self-reporting. Having said that, these sources are currently the best available and we give them their due respect (Chronograph does not provide a benchmark product to avoid conflicts of interest).
Finally, we wanted to understand if benchmark preference had any measurable impact on asset allocation – are investor portfolio construction decisions influenced by choice of benchmark? This is a reasonable hypothesis if an investor’s PE portfolio consistently and significantly outperforms their chosen benchmark. At least for the selected data set, the answer seems to be a fairly resounding no. The correlation between selected benchmark performance and PE asset allocation is essentially zero, as Figure 2’s Allocation to Private Equity chart shows.
To review the facts: About a fifth of pension plans use a PE index to benchmark returns; these benchmarks have outperformed public indices consistently, but at least part of this difference can be attributed to compositional differences. It does not appear that benchmark selection has a measurable impact on PE allocations.
We wanted to be more prescriptive in assessing choice of benchmarks.
As a brief refresher, approximately 80% of the 100+ public pension plans we surveyed use a public equity index to benchmark their private equity returns. Of these, the Russell 3000 (market-weighted US large cap) was by far the most popular with 25% of the surveyed institutions using it as their primary index (S&P is used by 17%, followed by MSCI World at 14%).
What stood out most from the data is how overweight this selection is toward large / mega-cap indices. Only 3% of public pension plans make use of the Russell 2000 (US small cap) as their private equity benchmark.
We feel a small-cap index is actually a far better comparable for any given private equity program.
The first chart below shows the distribution of enterprise values in the total PE buyout universe, as compared with the major indices listed above.
Though not a perfect comp, the overall distribution of enterprise values in the Russell 2000 is much closer to that of the total PE buyout universe, particularly in the ‘upper middle market’ of companies with TEVs of $1b – $4b. Conversely, 85% of companies in Russell 3000 index have a TEV > $10B – not even in the same state, much less zipcode.
Given the higher relative risk of smaller cap companies, we feel that ‘size matters’ in this exercise. To underscore this point we quantified a ‘size premium’ using the Fama-French three factor model, which you can see in Figure 3.’s Average Return by Market Cap Decile chart.
The Fama-French framework utilizes three factors to explain diversified portfolio returns, one of which is a size factor (decile data available here). In short, they demonstrate that smaller companies yield greater returns over time, in part due to increased risk.
By extension, Figure 3’s Relative Return Overstatement chart shows the degree to which public indices under- / overstate relative private equity returns as a direct result of this size premium.
There are some important qualifications for this analysis, namely: we only accounted for size utilizing the Fama-French model, ignoring the beta and value considerations. Additionally, we used a data set going back to 1927, whereas size premiums have been compressing since 1980 leading to some additional overstatement in the contemporary context.
Even with these qualifications, one can get a directional sense of what we had hypothesized: a small cap index, in this case the Russell 2000, provides a truer comparable for assessing relative performance of a typical private equity portfolio.
In conclusion, size is not the only premium associated with private equity – there are liquidity, leverage, and other risks that must be accounted for – but when measuring PE portfolio performance, an index which implicitly incorporates size as a consideration is likely a better proxy.
We wanted to test the application of leverage to a public equity index to consider whether this might provide a truer proxy for comparing private equity returns.
To recap, over 80% of public pension plans use a public index to benchmark private equity returns; there are important differences in the underlying sector compositions of indices; and underlying market capitalization is an even more-important consideration when selecting a benchmark. For these reasons and more, we view the Russell 2000 as the best public equity index for relative comparison of private equity returns across strategies.
What is missing from our analysis so far is consideration of leverage – certainly a key ingredient for many private equity strategies as demonstrated in Figure 4’s Avg. Capitalization Ratio chart.
Per 2016 Pitchbook leverage measures, the average private equity deal (excluding real estate) features approximately 50% debt-to-total capitalization, versus a 26% simple average for public companies. PE debt usage has increased to over 56% of capitalization as of Q3 2017 (up from 50% at the end of 2016).
Though one might expect higher risk to result from higher leverage used in PE investments, Figure 4’s Risk vs. Return chart demonstrates the opposite. In fact, illiquid investments have realized a relatively low risk (<10% standard deviation) and high return profile (>9% for all strategies aside from ex-US buyout) when compared against major public indices.
Further, there is not a significant difference in returns volatility between levered and unlevered illiquid strategies (e.g. US buyout vs. US venture capital).
This trend applies not only to the current expansion of the past ~10 years: the same holds true when taking a 25-year view.
Evaluating the drivers of lower volatility within private equity portfolios is beyond the scope of this post, but we offer two high-level explanations: quarterly Net Asset Value ‘smoothing’ and positive selection bias.
On the first point, the lagging nature of private equity valuation marks means that managers don’t have to react to every ‘flash-crash’ or short-term change in investor sentiment. The truth reveals itself over time in underlying operating performance.
On asset selection: buyout managers (and their lenders) explicitly select lower risk assets with greater cashflow potential in recognition of the need for debt paydown and covenant buffers. In aggregate, this has the implicit effect of introducing selection bias (in this case, a good thing for limited partners).
Again, these may be simplistic explanations, and are certainly beyond the scope of the current investigation. Regardless, low volatility for PE strategies becomes even more important when evaluating the use of a levered index.
On a 10 year basis, leverage-adjusting the Russell 2000 with PE debt levels ultimately yields the same return profile as the ‘Average PE’ grouping – at least in the current bull market. However, the effect works both ways: the leverage-adjusted index underperformed significantly in market down drafts (see: ‘hedge’ funds).
This increased volatility in the leverage-adjusted index resulted in a lower ex-post Sharpe ratio (returns / standard deviation) and drawdowns than the standalone Russell 2000 – a poor proposition for most portfolio managers.
As the ‘return metrics’ table more-fully demonstrates, a levered public equity index – in this case the Russell 2000 – ultimately proves to be a poor proxy for benchmarking relative private equity returns despite comparable headline performance.
We have taken a methodical approach to better understanding how investors make use of benchmarks for their private equity program(s).
First, we assessed choice and use of benchmark across 100 US public pension plans to understand current practices. We subsequently deconstructed the most widely-used indices to examine the impact of various compositional factors, namely, underlying constituent size, leverage, and sector mix.
In Figure 5, we have synthesized our conclusions as a scored matrix to encourage further discussion among allocators.
Fortunately we are not operating in a vacuum in this exercise: the Chartered Alternative Investment Analyst Association (CAIA) has published some fundamental research on what constitutes an ideal benchmark.
In short, CAIA identifies three desirable properties: transparency, investability, and ‘appropriateness’ (which we have deemed ‘comparability’ in our summary Benchmark Score below).
Per the CAIA, the ‘transparency’ consideration concerns components, prices, and methodology. Does the benchmark provider disclose the underlying constituents? Are constituent prices verifiable and unambiguous? Is the methodology used for constructing the benchmark transparent and well understood?
Applying these considerations to the most commonly used public and private benchmarks, the results are relatively straightforward: public indices are, in essence, fully transparent (though we slightly ding the BAML Global High Yield index given some ambiguity about constituent names). Conversely, the private equity indices are almost entirely opaque, not only in their constituents but also in construction methodologies (though we give some credit to Cambridge for providing high-level commentary on methodology).
For our own analysis, we do not give a lot of weight to CAIA’s second principle: investability. As we described in our second series post, choice of benchmark does not have an empirical impact on asset allocation, nor is there strong evidence allocators use benchmarks as a ‘substitution’ for portfolio allocations.
The third CAIA principle (‘appropriateness,’ which we call ‘comparability’) is the most critical in our view. Transparency is only useful to the extent it provides further insight on the appropriateness of an index as a given comparison. While CAIA’s definition speaks more to investment style and coverage, we define a set of subcriteria that our own research has demonstrated to have more meaningful implications, once again: size, leverage, and sector mix. Across these subcriteria, relative strengths and weaknesses become more apparent.
Ultimately, we view the Russell 2000 as an overall good proxy, though utilization as a private equity benchmark remains low (only ~3%). Where it falls short however is in its accounting of leverage. Its constituents are less levered, and simply applying additional leverage does not result in a better benchmark.
Taking all of this into account, we arrive at our conclusion: in short, a weighted blend that reflects the underlying allocations of a given PE program is the best benchmark for comparing relative performance of private equity. At a high-level, a blend of the Russell 2000 and the BAML High Yield index would provide a more-useful analysis given that any balanced allocation to alternatives is likely to be diversified across equity and credit strategies. Further, a blended approach more-closely captures the capital mix and geographic dynamics of the underlying companies (we re-emphasize small market cap indices for like-to-like comparability).
The natural question becomes, why isn’t this approach more widely used? (<20% of institutions utilize a blended benchmark per our original survey.) We would argue this has to do with the manual complexities of regularly assembling and updating a blended benchmark using only traditional tools like Excel.
At Chronograph, we’ve solved this problem by making it seamless to construct blended benchmarks in real time, and with little effort. Users can add an unlimited number of indices and weightings to construct an optimized program benchmark, and compare performance across their total portfolio or any dynamic subgrouping.
With more new capital entering private markets than at any prior time, it is critical that investors understand the relative risks and opportunities of private capital investing. Seamless benchmark construction is one way in which the Chronograph platform helps institutional investors discover deeper portfolio insights. To learn more, visit www.chronograph.pe or reach out to [email protected]
Get updates in your inbox
Learn how Chronograph can streamline your private capital investment monitoring and diligence