Key Points:
- Accurate performance readings of public pension funds and other institutional investors are crucial.
- Two types of benchmarks are commonly used: passively investable benchmark (PB) and custom benchmark (CB).
- Many institutional investors rely solely on CBs, which can present a skewed and inaccurate reading of performance.
- Investors can infer PBs through statistical analysis of the portfolio’s rates of return.
- Public funds typically underperform PBs due to investment expenses and conflicts of interest in benchmarking and evaluating performance.
Accurate performance readings of public pension funds, endowments, and other institutional investors are critical to their trustees and stakeholders. Fund performance is usually evaluated by comparing the portfolio rate of return to that of an index-like benchmark. The following analysis reviews the benchmarking practices of US public pension funds and finds them wanting: In effect, these funds have unleashed their hounds on slow rabbits.
Benchmark Types
Institutional investors rely on two kinds of benchmarks when they measure the performance of the total portfolio:
- A passively investable benchmark (PB) typically comprises several broad market-cap-weighted indexes. These might include the Russell 3000 stocks, ACWI ex-US stocks, and Bloomberg Barclays Aggregate bonds. These indexes don’t tend to overlap and pretty much cover the waterfront. Sometimes the PB is described as a policy portfolio.
A PB expresses the investor’s risk tolerance and concept of diversified investing. It might also reflect a home-country bias or currency preference. Through the PB, the investor is saying, “If I had no information about mispricing of markets or assets, this is the portfolio I would be most comfortable with.”
As the name implies, the benchmark is investable and passively so: It is feasible rather than hypothetical. It provides a baseline to determine whether portfolio management adds value in excess of purely passive implementation. Finance scholars and serious practitioner researchers invariably use PBs to evaluate investment performance. Indeed, the PB is the essential benchmark for performance evaluation.
- A custom, or strategic/composite, benchmark — I’ll go with “custom” — is, in principle, derivative of the PB. The custom benchmark (CB) generally consists of additional asset class components that describe how the portfolio manager intends to depart from the PB at the asset class level to achieve a strategically superior, better-performing portfolio.
In addition to stock and bond allocations, the CB may include weights for private equity, hedge funds, real estate, commodities, and other alternative assets. Sometimes the traditional and alternative components have multiple subcomponents, which can make the CB complex, sometimes opaque, and sometimes difficult to replicate.
The CB can help measure the investment strategy’s effectiveness at the asset class level. If, over time, the CB generates greater returns than the PB, it indicates the strategic allocation was better than the passive baseline. And if the portfolio’s actual return is greater than that of the CB, it indicates that implementation decisions also had a positive effect.
Using the two benchmarks in this way helps to differentiate between strategy and implementation in performance attribution. In a perfect world, this is how the two benchmark types would be determined and applied. Unfortunately, things rarely work like this in the real world.
In practice, the PB — the essential benchmark — has gone by the wayside. Among most institutional funds, the CB has become the sole benchmark in use — or at least the sole visible benchmark in public performance reporting. As a result, insight into the merit of strategic decision-making versus the policy baseline is lost.
As we shall see, exclusive use of the CB has another, even more perverse effect: It tends to present a rosy, rather than accurate, reading of performance.
Hugging the Portfolio
Institutional portfolios often exhibit close year-to-year tracking with their CB. This results in part from how CBs are revised over time. Sometimes revisions are motivated by a change in asset allocation, which may warrant adjusting the benchmark. Often, though, the revisions are more a matter of periodically tweaking the benchmark to more closely match the execution of the investment program.
No doubt the benchmarkers see such tweaking as a way of legitimizing the benchmark so that it better aligns with the actual market, asset class, and factor exposures of the fund. It accomplishes that, to be sure. But it also reduces the value of the benchmark as a performance gauge, because the more a benchmark is tailored to fit the process being measured, the less information it can provide. At some point, it ceases to be a measuring stick altogether and becomes a mere shadow.
We talk about “hugging the benchmark” in portfolio management. Here we have another twist on that theme: forcing the benchmark to hug the portfolio.
Inferring PBs
We stated that PBs are rarely reported. We can, however, infer them through a statistical analysis of the portfolio’s rates of returns. We do this by regressing portfolio returns on those of multiple independent variables, such as the three stock and bond indexes mentioned earlier. This process provides the appropriate weights, or allocations, for the individual broad market indexes to infer the best passively investable benchmark (I-PB). We can use these I-PBs and the reported CBs to give a fuller, more accurate picture of total portfolio performance.
The multiple regression benchmarking technique, originated by William Sharpe, is a powerful means of estimating I-PBs.
CalPERS: A Case Study
CalPERS is fairly typical in its approach to performance reporting: It uses a CB and tweaks it with some regularity. So in addition to being large and prominent, CalPERS serves as a good representative for the sector as a whole. Thus what follows is not intended to single CalPERS out or present it in an unfavorable light, but rather to demonstrate how public funds present their investment results.
The table below compares CalPERS’s total fund rate of return with that of its CB and an I-PB of the type described above. The I-PB comprises 79% US and non-US stocks and 21% US investment-grade bonds.
CalPERS Benchmarking and Performance: An Analysis
| Fiscal Year Ending | CalPERS Total Fund | Custom Benchmark | Difference | Inferred Passive Benchmark | Difference |
| 2011 | 21.7% | 21.8% | -0.1% | 23.6% | -1.9% |
| 2012 | 0.1% | 0.7% | -0.6% | 2.2% | -2.1% |
| 2013 | 13.2% | 11.9% | 1.3% | 13.8% | -0.6% |
| 2014 | 18.4% | 18.0% | 0.4% | 18.6% | -0.2% |
| 2015 | 2.4% | 2.5% | -0.1% | 3.8% | -1.4% |
| 2016 | 0.6% | 1.0% | -0.4% | 1.4% | -0.8% |
| 2017 | 11.2% | 11.3% | -0.1% | 13.3% | -2.1% |
| 2018 | 8.6% | 8.6% | 0.0% | 9.2% | -0.6% |
| 2019 | 6.7% | 7.1% | -0.4% | 7.5% | -0.8% |
| 2020 | 4.7% | 4.3% | 0.4% | 5.5% | -0.8% |
| 10 Years | |||||
| Annualized Return | 8.54% | 8.51% | 0.03% | 9.68% | -1.14% |
| Annualized SD/TE | 7.4% | 7.1% | 0.5% | 7.3% | 0.7% |
| R2 with Total Fund | .995 | .991 |
CalPERS’s portfolio return tracks that of the CB to an extraordinary degree. The 10-year annualized returns differ by all of 3 basis points (bps), 8.54% versus 8.51%. Year to year, the two-return series move in virtual lockstep, as demonstrated by the measures of statistical fit — an R2 of 99.5% and tracking error of just 0.5% — and even by a simple visual inspection of the annual return differences. For example, excluding 2012 and 2013, the annual return deviations from the CB are no greater than 0.4%. This is a skintight fit.
The table also shows CalPERS I-PB’s return series. This, too, has a close statistical fit with CalPERS’s returns in terms of the R2 and tracking error, though not as snug a fit as with the CB. Moreover, there is an important difference in the level of returns. Whereas CalPERS’s 10-year annualized return is virtually identical to that of its CB, it underperforms the I-PB by 114 bps a year. And it does so with remarkable consistency: in each of the 10 years.
The return shortfall is statistically significant, with a t-stat of -2.9. And it is of huge economic significance: A 114 bps shortfall on a $440 billion portfolio is about $5 billion per year, a sum that would fund a lot of pensions.
It’s Not Just CalPERS
To reiterate, CalPERS is not an outlier or an exception. Its approach and results are representative of what my reviews of public fund performance have found. For example, I compared the same three returns series for each of the 10 largest US public pension funds. The results are presented in the following table.
Benchmark Return Comparisons: Simple Averages, 10 Years to 30 June 2018
| 10-year Simple Average Return | Row 1 Minus Row 2 | Row 2 Minus Row 3 | |
| 1. Average Reported (10 Funds) | 6.56% | ||
| 2. Custom Benchmark Average | 6.58% | -0.02% | |
| 3. Investable Benchmark Average | 8.11% | -1.53% |
The simple average CB return essentially matches the simple average return earned by the funds. It differs by a mere 2 bps. Both of those series, however, lag the I-PB average by approximately 1.5% per year. At the individual-fund level — particulars not reported here — none of the CBs had a return greater than that of the corresponding I-PB. In other words, the benchmarking shortfall is both large and pervasive.
What we observe with CalPERS is not an isolated problem but a chronic one: CB returns tend to lag I-PBs by a wide margin. The funds are chasing slow rabbits.
What’s Happening Here?
Finance scholars have a dictum that, to the extent markets are reasonably efficient, diversified portfolios can be expected to underperform properly constructed (passive) benchmarks by the approximate margin of cost. I estimate the annual cost of investing public funds at 1.1% of asset value. We can reasonably conclude that investment expenses account for much of public funds’ performance shortfall relative to their I-PBs.
CBs underperform I-PBs because they are devised and revised to be in sync with portfolio structure. Thus, they are essentially shadows, or echoes, of the portfolios themselves. As a result, public funds appear to deliver rather neutral performance relative to their benchmarks. But the overall effect of using CBs as benchmarks is to mask underperformance in excess of 100 bps per year.
Ultimately, trustees are responsible for how funds are managed and for their reporting. In practice, though, staff and consultants conduct the performance reporting, including devising and revising benchmarks. There is a conflict here. These are the parties that formulate strategy, conduct the investment program, and select the investment managers. They are benchmarking and evaluating their own work.
To make matters worse, some public funds pay staff bonuses based on performance relative to the CB. Fund trustees should direct their staff and consultant to incorporate a PB in all performance reporting.
The management of public pension funds is bedeviled by agency problems. Here is an opportunity to ameliorate an important one of them. Public pension funds need to find faster rabbits to chase.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©Getty Images / Dgwildlife
Professional Learning for CFA Institute Members
CFA Institute members are empowered to self-determine and self-report professional learning (PL) credits earned, including content on Enterprising Investor. Members can record credits easily using their online PL tracker.
