Tuesday, June 22, 2010

The Kramer effect

I taught a class yesterday and, as I typically do, when I got to the topic of peer groups I mentioned one of its disadvantages: questionable makeup of the universe. That is, does the manager belong here?

And, as I typically do, I cited an episode of Seinfeld to make my point: the episode where Kramer is telling Jerry and Elaine about his Karate class, and how he's the #1 student. What he fails to explain is that the other students are all small children.

This raises the question of the appropriateness of the peer group. Perhaps management assignment to peer groups isn't always so obvious when in error, but there is often a concern as to the appropriateness of placement. Thus, one of the many shortcomings of peer group benchmarks.


  1. Stephen Campisi, Intuitive Performance SolutionsJune 22, 2010 at 10:01 PM

    Like it or not, peer groups are a fact of life in the competitive landscape for active fund managers. Clients want to know how a manager stacks up against other managers in the same area of the market. Are there problems with the dispersion of strategies represented by a single peer group? Of course there are, and it's hard to see how much of this is caused by style inconsistency, style drift, abuse of discretion, or simply incorrect manager classification.

    You can ask: "Who cares?" After all, it's really just the human emotions of envy and regret that motivate these peer comparisons of oneself against others. Clients should ask themselves: "Am I investing with the goal of beating the other guy?" Or am I investing to achieve my personal financial goals? Well, don't hold your breath waiting for an answer to those questions, because clients are not likely to own up to having such pedestrian emotions, much less letting these emotions drive the important decision of selecting and retaining fund managers. Rather, they'll say things like: "Benchmarks are abstract. They have no fees or transaction costs. They face no practical problems on the availability of securities. But a peer group is a real-life comparison."

    Personally, I get a laugh when fund managers justify underperforming a benchmark by noting that their peers also underperformed. As if that made the relative losses any easier to bear! That's the real life example of Kramer in the dojo, only this time Kramer is getting his butt kicked by a nine-year-old.

    Truth be told, the performance measurement and analysis processes are bedeviled by human emotions, herd behavior, behavioral biases and everything else that tends to skew the investment process away from rationality and common sense.

    Peer groups? They're likely to be with us for a long time. The same is true for benchmarks. Want a better yardstick for investment success? Try using the client's financial goals. Don't be surprised if you don't start looking at individual fund managers in the context of the client's total portfolio, rather than in isolation - or that you start looking at the ability to meet the client's actual goals, such as funding liabilities, upcoming expenses and the like - and (perish the thought) that you start looking at money and not just returns, or (most shocking) that you start using money weighted returns. The upside is that you'll start producing analysis that actually means something to your client, and which gets used in the investment decision making process. That would be a welcome change.

  2. "I get a laugh when fund managers justify underperforming a benchmark by noting that their peers also underperformed. As if that made the relative losses any easier to bear!"

    It can raise valid questions about the appropriateness of that benchmark to that peer group, though.

  3. Steve, thanks for your insightful comments.

    It's interesting, I think, that peer groups fail six of the seven criteria offered by Richards & Tierney:

    Unambiguous (no)
    Investable (no)
    Measurable (yes)
    Appropriate (no)
    Reflective of current investment opinions (no)
    Specified in advance (no)
    Owned (no).

    But in spite of this, peer groups remain popular.

  4. Thanks for your note. Yes, this does raise further questions as to the appropriateness of peer groups. As the Myners Review (available on the Internet) pointed out, "peer group comparison is unacceptable." But, it remains!

  5. I actually think peer groups are fine, if they are properly specified and populated.

    My point was that if the market index that is being used to benchmark everyone in a peer group consistently mismatches the average performance of that peer group, then it may be that the peers aren't investing in what the index measures. In which case, you probably need a better benchmark.

    In the absence of a good benchmark then peer groups CAN provide at least some measure of opportunity cost, if not an objective estimate of investment management skill or investor experience.

    I think you are too hard on peer groups.

    Ambiguity is unavoidable as you need to capture a range of peers, not just a specific one. The question is whether the degree of ambiguity is acceptable or not. For example, just having a peer group called equity funds isnt specific enough. A peer group that is Peruvian Leveraged Large Cap Biotech Specialist funds is much more specific and far less ambiguous, and much more useful (if you care about the PLLCBS sector).

    Peer groups are investable. Now, you cant actually go out and buy a single security that gets you peer group exposure. But this was true before index funds as well. Just as you could have theoretically bought the market back then, you can theoretically buy a portfolio that matches the asset weighted portfolio of the peer group. Manager alpha is easily purchased by buying the same stocks as the manager. It is not easy to purchase, but it is doable.

    Appropriateness is in the eye of the beholder, and has more to do with application than any feature of peer groups per se.

    Peer groups are reflective of current investment opinions, as they are composed of the manifestations of current investment opinions. Not sure this is necessarily a good thing, though.

    Peer groups can be specific in advance, and ownership is just to avoid arbitrary, non-natural changes if I recall correctly, so it comes down to peer group specification and construction again.

  6. Kimble, some very good points. Thanks!

  7. Stephen Campisi, Intuitive Performance SolutionsJune 25, 2010 at 6:39 PM

    With all this talk about peer groups (especially the comments that relegate them to the trash heap) it's a good idea to consider what peer groups actually represent. Peer groups are really the results of managers who claim to invest according to a certain mandate... or who are assigned to that mandate, either by qualitative or quantitative means. Presumably, the mandate itself is clearly identified and there is agreement on the definition of the mandate. Unfortunately, none of these assumptions and statements are absolutely true. So why would we expect the peer group to be consistent and cohesive if we lack clear definitions on the groupings (i.e. the benchmarks) themselves?

    To prove that we lack even consistent definitions of the groupings, just go out and ask someone to define what a "core bond index" allocation should include. Or what an "emerging markets equity benchmark" looks like. Or what "foreign developed equity" should include by way of countries and securities. You won't get a consistent answer. You can ask people what certain benchmarks look like, but who declared the benchmark providers to be the arbiters of what truly represents true market exposure in each sector? The various providers use their own arbitrary methodologies and they get different results.

    So we must conclude that benchmarks are not markets; they are simply our best efforts to represent these markets. As proof that there is no agreement on how to define markets, just go ask a few portfolio managers to set the capitalization limits for U.S. small cap, mid cap and large cap. You won't even get them to agree on whether mid cap is truly a sector (some believe that it's really "large + mid" and small, while others think that mid cap is a sector that BOTH large cap and small cap managers use.) So where does small cap begin and end? $1 billion? $2 billion? $3 billion? All of these answers have some degree of support among equity managers. Same is true for large cap. Is the S&P 500 really large cap? When the index was developed back in the 1950s it was meant to be a proxy for the entire U.S. equity market, not just the largest companies. That's why a close look at the S&P 500 will show that it is about 85% large cap with the rest split between both mid cap and small cap stocks.

    So, is there any wonder that there is a lot of dispersion in the styles represented by a single peer group? This is perhaps not so much a criticism of peer groups as a device, since there is a breadth of opinion on what constitutes each market sector benchmark. As a result, some of the performance difference within a peer group is really structural in nature, showing the value of holding certain market exposures (both strategically and tactically) as opposed to others. The remainder is a "selection" effect.

    With this in mind, perhaps we might all treat peer groups with a little more respect. After all, are we really going to say that ALL clients who request information on peer group rankings and performance are COMPLETELY wrong in their requests? Or might it be that they understand the relevance of each manager's performance relative to the consensus on what constitutes a market segment?

    Finally, we might do well to refresh our perspective by reading Bill Sharpe's classic paper called "The Arithmetic of Active Management" where he uses common-sense (rather than Nobel-winning mathematics) to show that in aggregate the performance of active managers must equal the performance of the market, minus their active fees. So, if the managers ARE the market, then the criticism of peer groups must be "allocated" in part to a criticism of the benchmarks we use to proxy "the market."

    Perhaps we need a little less faith in the benchmarks...

  8. Some excellent points...sorry for my delay in posting this (it got lost). Being aware of all of these potential differences which can in of themselves contribute to differences is important but unfortunately not visible. So I guess "caution" or "buyer beware" might be appropriate statements to accompany the results.


Note: Only a member of this blog may post a comment.