Measuring Portfolio Diversification is Not Easy
 

In this research brief we explore the pitfalls in measuring portfolio diversification in private capital. To this end we examine two very different methodologies -  temporal diversification and cross-sectional diversification - and discuss the shortcomings of each. In particular, we show two things: first, how temporal diversification makes it hard to identify cross-sectional diversification; second, how econometric challenges make time-series approaches problematic.

Diversification allows investors to reduce their risk without compromising their expected returns by allowing the specific risks of a wide variety of investments to cancel out. In the public markets it is straightforward to show that the risk of a portfolio can be reduced (as a result of diversification) by distributing its capital among different securities. While we don’t believe private capital to be any different in this regard, in this research brief we are primarily interested in quantifying this effect for private capital portfolios.

 

Private Capital Combines Two Different Types Of Diversification – Cross-Sectional And Temporal – Complicating Its Analysis

 
 

Complete the fields below to request a copy of this Research Paper.

Name *
Name
 
james kocis
Commitment Pacing: Targeting a Fixed Valuation
 

If you were to take over the management of a private capital portfolio at a large pension fund and were tasked with maintaining a valuation of about $1B, how would you approach it? At what rate should you commit capital to best achieve this goal? It turns out that this question can be formulated in a number of different ways, each leading to an algorithm that dictates the pace at which commitments should be made:

  • Constant Commitment –Assumes the investment manager would make commitments of a constant size every quarter.

  • Expectation Pacing –Combines the expected valuation curve of funds in such a way as to achieve the target, using an algorithm that tries to make the expected valuation of the portfolio equal to the target valuation.

  • Stochastic Pacing – Takes into account the uncertainty in future valuations, and employs an algorithm that minimizes the expected deviation from the target valuation.

  • Dynamic Pacing – This is a dynamic approach which, when making the current decision, takes into account that all future decisions will be made optimally; it minimizes the expected deviation from the target valuation, further assuming that after the current commitment all future ones are made in the same optimal way.

In this paper we define each of these formulations rigorously, construct optimal algorithms, and quantify their performance.

 

Expected Valuations and Standard Deviations of Buyout Funds as Function of Fund Age

 
 

Complete the fields below to request a copy of this Research Paper.

Name *
Name
 
The Burgiss Team
What Happens To Fund Ranks When the Measure of Performance is Changed?
 

In this research brief, we compared fund ranks produced by these various performance measures:

  • Internal Rate of Return (IRR)

  • Total Value To Paid-In (TVPI)

  • Kaplan Schoar Public Market Equivalent (KS-PME)

  • Direct Alpha (DA)

  • Time-Weighted Rate of Return (TWRR)

We found that most measures tend to produce similar ranks. For example, 80-90% of funds fall in the same quintile, whether ranked by IRR or DA. The least similar are TVPI and  KS-PME, with only 60% agreement on average. The one exception is TWRR, which, as expected, produces significantly dissimilar ranks because it’s not money-weighted.  We also show that most of the information in these four measures – IRR, TVPI, KS-PME and DA –  can be summarized using just two numbers.

Scatter plot of IRR-, and TVPI-based ranks. The 45-degree and best- fit lines are shown in black and blue respectively .

Scatter plot of IRR-, and TVPI-based ranks.
The 45-degree and best- fit lines are shown in black and blue respectively
.

 
 

Complete the fields below to request a copy of this Research Paper.

Name *
Name
 
The Burgiss Team
Single-Period Brinson-Style Performance Attribution for Private Capital
 

Performance attribution is the process of decomposing a portfolio's return into subcomponents that are each the result of the decisions that went into the construction of that portfolio. In the latest paper from Burgiss Applied Research, the team examines Brinson-style attribution, which decomposes a portfolio's active returns into allocation and selection effects, and supplies a methodology to carry out this exercise for private capital.

In addition to describing the attribution methodology, the working paper also provides two detailed case studies that illustrate the results of attribution applied to two concrete portfolios. Several aspects of their performance are explained naturally via this paper’s attribution methodology; without it such an analysis is non- trivial.


image.png

Get a copy of the Executive Summary of this report here.

Complete the fields below to request a copy of this Research Paper.

 
 
Name *
Name
 
The Burgiss Team
What Happens to Cash Flows During a Crisis?
 

In this brief we focus on the effect of two crises — the dot-com crash and the global financial crisis (GFC) — on private capital cash flows, disentangling these effects from those arising from returns.

Cash flows are strongly influenced by market crises. Distributions are both delayed and, of course, reduced. However, even contributions are reduced by about one half; venture capital contributions were reduced by somewhat more than that during the dot-com crisis, but by much less during the global financial crisis, only dropping by about a quarter. How market returns affect the performance of private capital has received a great deal attention, including by us (see O’Shea and Jeet (2017)). However, much less has been written about how crises, namely extreme market returns, affect cash flows (contributions and distributions).

 

Pooled Contribution and Distribution Fractions

Note: The shaded areas represent the dot-com crash and the global financial crisis.

Note: The shaded areas represent the dot-com crash and the global financial crisis.

 
 

Complete the fields below to request a copy of this Research Paper.

Name *
Name
 
james kocis
Budgeting for Capital Calls: A VaR-Inspired Approach
 

This is the second of two papers on modeling cash flows. 

The results in the first paper showed that contributions and uncalled capital, in addition to age, are a useful predictor of future cash flows. Additionally, we demonstrated that the approach outlined by Takahashi and Alexander underperforms our data-driven models. In this second paper, we deepen this examination, focusing our attention on budgeting for future capital calls.

We find that while it is useful to anticipate capital calls from an investor’s portfolio, we find it more practical to estimate a likely upper-bound of those calls. To this end, we introduce a new concept, the maximum probable contributions. This statistic is subject to a user-specified confidence level and serves as this upper bound. We explore in detail a historical methodology for its computation, illustrate typical model predictions, and document its out-of-sample performance by backtesting both funds and portfolios of funds.

Forecast of capital calls in one quarter for a portfolio with commitments to 55 funds (the portfolio includes buyout, venture capital, and real estate funds, as well as primary and secondary Funds of Funds; the commitments range from the very recent to those from ten years in the past.)

Forecast of capital calls in one quarter for a portfolio with commitments to 55 funds (the portfolio includes buyout, venture capital, and real estate funds, as well as primary and secondary Funds of Funds; the commitments range from the very recent to those from ten years in the past.)

Get a copy of the executive summary of this report here.

Complete the fields below to request a copy of this Research Paper.

 
 
Name *
Name
 
james kocis
Modeling Cash Flows for Private Capital Funds
 

This is the first of two papers on this topic.

In this paper we focus on predicting cash flows for private capital funds. We start by discussing the characteristics of cash-flow data that make these predictions challenging, and then examine several models for expected contributions and distributions, and evaluate their performance, both in-sample and out-of-sample.

We find that for contributions, uncalled capital, in addition to age, is a useful predictor. Distributions are more difficult to predict; we find that disentangling the effect of a fund’s performance on distributions produces better forecasts than other simpler approaches. We compare the models explored in this paper with those outlined by Takahashi and Alexander and find that they underperform our models by a wide margin. Finally, we draw some lessons regarding how to model cash flows, and how to measure model performance. We also make some observations regarding the intersection of risk and prediction with regard to cash flows.

Backtesting Comparison of Several Models Against That of Takahashi and Alexander

 
 

Complete the fields below to request a copy of this Research Paper.

Name *
Name
 
The Burgiss Team
Better Than The Buffet Bet?
 

A decade ago Warren Buffett bet that public equities would outperform hedge funds. In this month’s research brief, we revisit his bet, this time pitting private equity against public equity. Over the same period as the original bet, Buffet would most likely have lost had his counterparty chosen buyout funds, venture capital funds, or, perhaps, venture funds of funds; against generalist funds of funds Buffett would have won.

image.png

Warren Buffett, in Berkshire’s 2005 annual report, argued that hedge funds and private equity tend to reduce returns in the aggregate*. He subsequently wagered $500,000 that no investment professional could pick a portfolio of five hedge funds that would beat the S&P 500. Shortly after, Ted Seides took up his challenge, and chose five funds of hedge funds. The bet ran from the beginning of 2008 to the end of 2017**. Ted lost***.

In the brief, we examine what would have happened had a bet been made on the other class of investment Buffett mentioned in 2005, namely private equity.

*Berkshire Hathaway 2005, How to Minimize Investment Returns, pp. 18 – 19.
**Buffet and Protégé 2008.
***Berkshire Hathaway 2016; Seides 2017.

Click here to download your copy.

 
james kocis
Can We Improve on Roll-Forward Valuations?
 

Roll-forward valuation errors are significant. This is not surprising since they ignore important sources of information. Examples of relevant information include the returns of public markets, the returns of private markets (represented by the Burgiss Manager Universe), and the past behavior of funds. In this brief we explore various techniques that improve on roll-forward valuations, and discuss in more detail the best-performing of these. Our recommended technique leverages the Burgiss Manager Universe to reduce errors by a factor of about one half.

 
 

Click here to download your copy.

james kocis
How Accurate Are Roll-Forward Valuations?
 

Roll-forward valuations are commonly used in lieu of missing reported valuations. However, the error in roll-forwards is significant. Quarterly errors in recent years have been about 6% as a fraction of fund valuation. But end-of-year errors are larger: about 5% for equity fund-of-funds, 7% for real estate funds, 9% for buyout, and 10% for venture capital funds.

 
 

Click here to download your copy.

 
james kocis
Endurance of Internal Rate of Return
 

Performance measurement of private capital investments has been an ongoing challenge for asset owners, asset managers, and investment consultants alike.

The fundamental challenge in measuring the performance of private capital investments is that investors do not know precisely at a given point in time 1) how much capital is invested, or, colloquially speaking, is “in the ground”, 2) how long the capital is invested, or 3) what is the rate-of-return on the capital. For these reasons, the performance of private capital investments is not understood in the same way as returns in the public markets, where everything is known precisely at every point in time.

In this paper we investigate a time-duration measure – dubbed endurance of IRR – to accompany IRR so that it can be understood in the same way as returns in the public market. We define endurance of IRR as a length of time for which IRR effectively compounds the total contributions to yield the total distributions.

Endurance of IRR is always a positive number less than or equal to the age of the underlying fund at the time of performance measurement. We believe mentioning IRR along with its endurance can enhance its understanding thus making it a more useful measure of investment performance.

Click here to download your copy.

 
The Burgiss Team
Estimating Public Market Exposure of Private Capital Funds Using Bayesian Inference
 

The market exposure (beta) of private capital to the public market is important for a variety of reasons, including asset allocation and risk estimation. However, estimating beta is challenging. The root cause of this difficulty is the illiquid and private nature of the asset class. This, in turn, leads to data issues, namely limited access to high-quality, granular data. It also leads to statistical problems caused by valuations that are smoothed (leading to returns that are autocorrelated) and valuations that are imprecise (leading to noisy per-fund returns).

This paper uses a high-quality dataset of per-fund cash flows and valuations (the Burgiss Manager Universe) to fit Bayesian hierarchical models to per-fund returns. These provide precise estimates of per-fund betas for buyout and venture capital funds. In addition, given the size of our dataset, the authors are able to estimate how beta changes with the age of a fund, thereby resolving some puzzles regarding previous estimates of beta based on pooled data.

Click here to download your copy.

 
james kocis
Private Capital: Indexes, Investability and Self-financing Strategies
 

Traditional indexes for private capital are constructed by compounding short-term pooled returns (IRRs or TWRRs) giving a time-series of index levels over a long period of time. Creating an index in this way gives rise to a number of natural questions centered on whether such an index is investable, and the related question of whether it accurately tracks wealth outcomes.

This paper starts by examining whether such an index, based on pooled data, can be replicated. Because of the dependence on interim fund valuations it turns out that the index is not replicable, although one can come close. The paper continues to examine investability by investigating the funding requirements of an index. It emerges that traditional indexes tend to not be self-financing, generating large swings in how much capital is needed, or how much capital is returned to the investor. Finally, the paper compares the behavior of an index with that of a “strategy”, namely a simulation that explicitly tracks investments in private capital by accounting for commitments and all cash flows, while keeping undeployed capital in a more liquid form (such as a public equity index). The behavior of such a strategy is illustrated in the diagram at right.

Click here to download your copy.

 
The Burgiss Team
Inside Private Equity
 

Inside Private Equity (Wiley, 2009) was written for institutional investors to help provide a general understanding of how to invest in, monitor, and measure the performance and risk of private equity. Here, you’ll become familiar with everything from traditional industry measurements to a structured approach to portfolio management. 

image.png
 
 

Inside Private Equity was co-authored by Burgiss’ own James M. Kocis, Founder and CEO, and James C. Bachman, COO, as well as Austin M. Long, Head of Alignment Capital, and Craig J. Nickels, Head of US Investments at Abu Dhabi Investment Authority.  

Purchase here.

 
james kocis