Forbearance (Regarding Performance) is in the Eye of the Beholder

A new article in the Financial Analyst Journal, “Forbearance in Institutional Investment Management: Evidence from Survey Data,” examines the practices of asset owners when it comes to firing investment managers.  Amit Goyal, Ramon Tol, and Sunil Wahal surveyed more than two hundred institutional investors to come up with their observations.

According to the authors, previous academic work has shown that hiring decisions are primarily a function of performance, personal connections, and the recommendations of consultants, but “we know precious little about the firing process, in part because archival data on terminations are sparse”:

Institutions are disinclined to report terminations lest their shifts affect the transition from legacy to target portfolio or their tactical changes are second-guessed ex post, or because termination may reflect poorly on their initial selection decisions.  Asset managers are similarly disincentivized to report terminations for reputational concerns and for fear of triggering further client withdrawals.

Forbearance

The topic, as reflected in the article’s title, is “forbearance,” so let’s start with that.  The primary definition of the word, as found in the online Merriam-Webster dictionary, is “a refraining from the enforcement of something (such as a debt, right, or obligation) that is due.”  Alternate meanings offered involve “patience” and “leniency.”

Therefore, some sort of standard to be lived up to is implied.  When is something “due,” beyond which point the terms “patience” and “leniency” can be applied?

In this instance, the question involves the investment performance of asset managers who have been hired.  The article introduces some helpful information but gives readers an unfortunate impression.

Expectations

The authors introduce an “obvious question:  How should an institution think about the ‘optimal’ tolerance to underperformance?”  The period of time it takes for statistical significance varies depending on assumptions, but it is well beyond the time horizons applied by the industry, and so such “significance” is almost always ignored by practitioners.  (One calculation cited in the article derived the need for at least sixteen years to determine significance; other assumptions can result in periods much longer than that.)

The question is key, since there are “costs and benefits to asset manager turnover,” with the costs of transitions being “ultimately borne by the claim holders” on the portfolio, while poor performance over time has a negative effect as well.  Overall, the body of evidence shows that frequent changes in managers impede long-term performance, since allocators tend to hire strong performers and fire weak ones not long before their recent fortunes reverse.

From the horse’s mouth

One concern in basing the research on survey data:  “Hearing directly from the horse’s mouth could be problematic because respondents might not do what they say.”  That is an important caveat (which applies generally to surveys), since it is not uncommon for allocators (and consultants) to say a firing is about something other than performance even when the performance is the precipitating factor, since doing so makes them look less like performance chasers.

There is likely less of that tendency in this survey environment than in a situation where the selection process of an allocator or consultant is being evaluated directly.  In any case, the bottom line from the survey is that “underperformance is by far the dominant reason” given for termination.  (The eight options for that question can be found in the survey form.)

The wrong frame

The main problem with the article is evident here:

Respondents report surprisingly long holding periods for their active investment managers.  In public equity and fixed income, over two-thirds of respondents report holding periods of longer than five years.  In contrast, for hedge funds, only 42% of respondents report holding periods of longer than five years.  Regardless of this variation across asset classes, holding periods of this length run counter to the pervasive notion that institutional investors are impatient.

They do?

Another section:

In public equity, almost two-thirds of institutions are willing to tolerate underperformance for three years or longer.  In fixed-income and hedge funds, there is slightly less tolerance, 56% and 50%, respectively.  Notwithstanding these differences across asset classes, this tolerance is strikingly large and goes against the grain of the common narrative that institutions are “trigger-happy.”

And, the article’s conclusion says that “holding periods for the average asset manager are quite long, frequently longer than five years,” and “institutions are surprisingly tolerant of underperformance.”

It is fine that the goal of the research was to describe current practice rather than assert an optimal time horizon after which the concept of forbearance is applied.  But the counterproductive three-to-five year industry time frame which drives the tone of the article is not the standard by which impatience or trigger-happiness or tolerance should be judged.

Those unfortunate characterizations stand in contrast to good advice from the authors about the importance of a “more refined Bayesian approach” that updates “prior beliefs of underperformance based on the arrival of new information about the asset manager,” well beyond the performance numbers coming in.  They favor long-lived, continuous, and multi-faceted assessments — and even mention the importance of tracking fired managers, a praiseworthy practice that a small minority of investors implement.

It is too bad that those good suggestions are buried underneath an unfortunate message — that investors’ time horizons are found to be somewhat longer than a destructive industry “standard.”

Begin with beliefs

As mentioned by the authors, we need more exploration of how and when asset owners terminate managers.  But the most important thing for those doing due diligence and making allocation decisions is to understand their own beliefs about what they should be doing and why.

Start here:

How do you think about signal and noise in investment performance and what kind of significance should you ascribe to performance numbers in your assessment of managers?

What are your expectations about the normal cycles of performance for a given type of manager, including the length of periods of underperformance and their magnitude?

What is your baseline time horizon over which you intend to judge an investment manager?

Without examining these questions — and agreeing on standards that make sense — you are prone to react to interim events in ways that fit the pattern of the industry instead of crafting one of your own that leads to better outcomes.

What is a good baseline horizon?  Start with a decade, even though that also falls short of any statistical proof regarding performance.  That doesn’t mean that you have to be locked in for that time — Bayesian updating based on qualitative factors should play a role in whether you ultimately change your mind — but if you have that as your foundation, both your selection and termination decisions will improve.

These concepts are also addressed in the Advanced Due Diligence and Manager Selection course, other training options, and the consulting work of The Investment Ecosystem.