Commentary: Evaluating OCIO performance

When funds search for an outsourced chief investment officer, they inevitably consider past performance. Unfortunately, there are no widely accepted standards for measuring OCIO performance, and many common approaches can lead funds to draw the wrong conclusions.

For example, they may rank a truly skilled OCIO lower than others that simply experienced better luck over a (short) time horizon; or they may penalize an OCIO for performance over which the OCIO, in fact, had no control. If funds consider past OCIO performance, it is critical that they minimize the risk of drawing wrong conclusions. Here's how.

Typically, funds delegate the discretion to hire and fire underlying managers to the OCIO, but retain authority over their policy portfolios. As a result, funds — not the OCIO — are accountable for policy portfolio outcomes (i.e., benchmark returns), leaving the OCIO accountable only for alpha.

While the passive, or beta, return component is larger and riskier than alpha, OCIOs are not accountable for it because their asset mix responsibility is limited to providing prudent, non-discretionary advice. OCIOs aren't off the hook for asset allocation or risk management, of course. Instead, evaluations in these important areas should be based on the quality of OCIOs' advice, rather than the outcomes over which they have no discretion.

Performance analysis should start by assessing an OCIO's skill before fees are considered. Any fees included in past performance should be backed out, and replaced by the fees that will apply to accounts in the future.

Here's why. First, fees change over time, so those that apply in the future may differ from past fees. Second, fees are negotiable, while investment skill is not. There's no point considering fees until there's evidence that an OCIO has any skill worth paying for.

An OCIO's ability to add value from active management is constrained by the asset classes the fund permits and the ranges within which the portfolio must be rebalanced. To make performance comparable across OCIOs, then, we need to remove any effects that made it harder — or easier — for some OCIOs to outperform benchmarks.

For example, it is harder to outperform the policy portfolio when:

• Fewer asset classes are permitted.
• Rebalancing ranges are tight.
• Policy weights are high in markets that are more efficient.
• Opportunities to add value are limited.

Inflation-linked bonds might be good for funds with liabilities that rise with inflation, and are therefore included in the policy portfolio, but they offer fewer opportunities to outperform the benchmark (mainly duration) because the benchmark has few bonds and they are of similar quality (i.e., U.S. Treasuries).

Alternatively, OCIOs add value more easily when one or more of these strategies are permitted: tactical asset allocation; active currency hedging; certain hedge fund strategies; or private equity. In these four strategies,there's typically an inverse relationship between the capital used to implement the strategy (low) and tracking error taken (high). This low/high relationship between capital/risk creates an advantage because tracking error is allocated more efficiently in this case, all else equal.

To remove these effects, performance should first be analyzed at the asset class level and only later at the total fund level.

Also, greater care is needed in private markets, where valuations (and therefore returns) are less reliable than public ones, benchmark quality is low and because it takes years to deploy private capital (i.e., "vintage year" plays a big role).

It takes time to gain confidence that "seemingly good" returns were truly due to skill, rather than luck, and it involves considering the risks taken as well. An OCIO that takes less tracking error but adds the same value as another OCIO has generated better performance on a risk-adjusted basis (other things equal). Measuring the information ratio (alpha divided by tracking error) allows us to group OCIOs into four performance categories, based on their skill/luck profile:

1. Blessed OCIOs (skilled and lucky).
2. Doomed (unskilled and unlucky).
3. Forlorn (skilled, but unlucky).
4. Insufferable (unskilled, but lucky). (Terms were developed by Richard Grinold and Ronald Kahn in their book, "Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk.)

If median managers outperform half the universe, by definition, we should expect them to add zero alpha before fees (normal distribution) in a given year. Their expected information ration is zero before fees. Top quartile managers, however, are expected to add 67 basis points of value for every 100 basis points of tracking error taken (information ratio of 0.67).

While performance could be aggregated by client type (e.g., endowments only), funds should resist doing so because it provides less evidence of skill (or lack thereof) due to a smaller sample size. Asset class strategies pursued by an OCIO are typically similar across different client types, so there's little benefit from a focused analysis on any single investor type.

The final step is to apply appropriate weights to OCIOs' asset class performance, along with the fees that would apply (including underlying managers' fees).

Our OCIO search clients find "historical replays" useful, where adjusted performance figures are aggregated using the fund's policy weights to determine how performance would look had each of the OCIOs been engaged by the fund. This historical replay is easy to understand and more useful than the (unadjusted) performance against a 60/40 mix, which includes far too much noise.

Valter Viola is managing director and principal at Cortex Applied Research in Toronto. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines but is not a product of P&I's editorial team.