Monday, February 16, 2009

Why Can't We Properly Compare Pension Funds?

I recently had a discussion with a senior risk manager about how most journalists like to compare pension funds without understanding what they are comparing.

I asked him if he'd be so kind as to write a piece on this subject. Below are his thoughts on comparing risk-adjusted returns across pension funds:

The large majority of pension funds build their investment portfolio based on their “Policy Portfolio” (asset mix) which sets the target allocations and the benchmarks for each asset class (and is usually reset at longer intervals of time - every 3 years or more).

While benchmarks can be pretty much anything (example : rate of inflation + a spread, T-bills + 5% or the S&P500), in order to better understand the behavior of the asset classes (expected return and risk) it is preferred that benchmarks or, at least, the proxies for each asset class, be “investible benchmarks” – financial variables (like an index) on which financial instruments exist and can be traded. This way, the first and easiest way to invest would be to build a simple portfolio comprised of financial instruments (example: ETFs, futures, Total Return Swaps) whose underlying are the benchmarks of each asset class.

Given that much debate has been going on the choice of benchmarks, it is worth clarifying a couple of facts. The buzzword “benchmark” is used in many instances but may differ in meaning. Among all uses, the main ones are in “performance measurement” and in “portfolio construction”

For the purpose of measuring (and evaluating) the performance of the investment managers, pension funds set various benchmarks, not all of them “investable”. The choice of these benchmarks is entirely a management decision and can vary largely from fund to fund. For example, the performance benchmark for Real Estate can be a fixed rate of return, an inflation index plus a spread, an industry index or a public market sector index.

For the purpose of “portfolio construction” – and the Policy Portfolio is the main “portfolio construction” exercise – the pension funds and the investment consultants use “benchmarks” as proxies for the behavior (expected return and risk) of each asset class. In many instances, this “behavior” or characteristics of the Policy Portfolio are called “passive return” and “passive risk” (obviously derived from the idea that one could simply buy instruments mirroring each asset class – that why the “benchmarks” had to be investible! – and stay “passive” until the next reset of the Policy Portfolio).

Given the distinction between the two uses of “benchmarks”, it is obvious why a fixed rate of return used as “benchmark” in performance measurement cannot be used also in “portfolio construction”: the fixed rate of return has no volatility (and, therefore, risk) and is not an expected rate of return but rather a threshold to be achieved by investment managers.

Therefore, a Policy Portfolio’s expected return and risk can be easily computed and used for comparison among pension funds once the same “benchmarks” are used as proxies for each asset class.

Various criticisms could be brought to such a simple approach. One would be that no two investment portfolios are alike, mainly in the private markets. The simple answer is that asset class is the same for all pension funds and that it is only each fund’s investments that vary within each asset class.

I’ll go back to Real Estate for an example. Two pension funds may advocate that their real estate portfolios are totally different and that no unique Real Estate proxy can be used for both of them.

We shouldn’t forget, however, that at the onset of their investments, when both pension funds had no invested assets, each made an allocation to Real Estate as a “generic” asset class. As each pension fund built their portfolio of Real Estate assets based on each one’s expectations and strategies, the two Real Estate portfolios started to differ, mirroring each fund’s investment philosophy. In the end, however, Real Estate is Real Estate for all funds; what differs is each one’s approach to it. That “personalized approach” is not a Policy Portfolio decision but rather an “active management” decision made by the investment managers after the Policy Portfolio was built.

Therefore, to properly model and compare the Real Estate allocations and their impact on the overall Policy Portfolio, one should use the same “investable” proxy across all pension funds, The same goes for all asset classes.

On top of the Policy Portfolio, the fund’s management and Board agree on some guidelines for the “active management” – the investments that deviate from the “investable benchmark or proxy” set in the Policy Portfolio.

The sum of all these “active management” decisions result in “active return” and “active risk” which serve to assess the investment skills of the “active” investment managers hired to bring additional return on top of the one expected from the Policy Portfolio.

Finally, the “active return” and “active risk” add up (under some mathematical hypothesis, mainly on the risk side) to the “passive return” and “passive risk” to make up the fund’s “total return” and “total risk”. In most pension funds, the “passive risk” represent the huge majority of the “total risk” of the fund and is responsible for the bulk of the fund’s return.

Luckily, there are not so many asset classes (each requiring a proxy), but rather a handful (or two), so that one can easily model and estimate the bulk of a fund’s return and risk – the “passive return” and “passive risk”.

In conclusion, the investment process of a pension fund usually has two steps: the construction of the Policy Portfolio and of the “active portfolio”. In order to assess the efficiency of each of these two steps, one has to evaluate each step’s return and risk and then calculate a simple Information Ratio or Sharpe ratio.

For rapid comparison of the “passive return” and “passive risk”across various pension funds an analyst has to choose and use the same simple and “investable” proxies for each asset class for all funds. The “active return” is derived straight forward from the fund’s published “total return” and the analysts personal estimate (according to its choice of asset proxies) of the “passive return”. The computation of the “active risk” is much more complex. A comparison of “active risks” – and implicitly of “active Information Ratio” - across pension funds would be possible only if there were a uniform method of calculating it. However, its benefits would be worthwhile the effort as it would pinpoint the “active managers’” investment skills (or lack of).

Such a industry-standard method – although not perfect and subject, as always, to criticisms – was created, approved and is globally used by banking institutions and their regulators.

Does the pension world require one?

If you ask me, the pension world first requires a lot more transparency. In order to properly compute risk-adjusted returns, we need to know the benchmarks that govern each and every investment activity and then we need to evaluate whether these benchmarks accurately reflect the risks and beta of the underlying investment.

It amazes me how we trust "pension experts" with billions of dollars and we do not scrutinize the benchmarks they use to justify the risks they take and their compensation!

Sure, they will tell you all sorts of reassuring things about how they "invest for the long-run" and how they "added value" in this or that asset class. But did they really add value of just beat some bogus benchmark? In most cases, it's the latter, not the former.

When I see a pension fund manager killing his or her benchmark by double digit returns - typically in alternative investments - then I know they are pulling the wool over their stakeholders' eyes. Worse still, they are pulling the wool over the eyes of their board of directors which is entrusted to protect the best interests of stakeholders.

In my mind, it all amounts to sophisticated pension fraud - an institutional scam that has been going on for years so pension fund managers can reap huge bonuses by beating bogus benchmarks.

Enough is enough already. If we are going to demand transparency and accountability from private banks, we should demand even more from our public pension funds.

Finally, if senior pension fund managers want to get compensated for performing, then the onus is on them to prove to stakeholders that the benchmarks they are using to evaluate all internal and external investment activities accurately reflect the risks and beta of those investments.

No comments:

Post a Comment