The Economist’s Buttonwood column this week highlights an interesting paper by Shahin Shojai of consultancy Capco, George Feiger of Contago Capital and Rajesh Kumar of the Institute of Management Technology in Dubai. The paper essentially lambastes active management by pointing out the lack of standardization in performance measurement. As the trio laments,
We find that very few, if any, portfolio managers look for the efficiency frontier in their asset allocation processes, mainly because it is almost impossible to locate in reality, and base their decisions on a combination of gut feelings and analyst recommendations. We also find that the performance evaluation methodologies used are simply unable to provide investors with the necessary tools to compare portfolio managers’ performances in any meaningful way.
Instead of using countless performance metrics like the Sharpe Ratio, Information Ratio, Treynor Ratio etc. they suggest a brutally simple, yet curiously elegant solution: simply compare the manager’s performance to what they would have achieved had they not made any trades all over the past year (or for that matter, over the past 2, 3 or more years)…
Using the concept of inertia, an asset manager‘s end of period performance is compared to the performance of their portfolio assuming their initial portfolio had been held, without transactions, during this period. We believe that this will provide clients with a more reliable performance comparison tool and might prevent unnecessary trading of portfolios.
Although the Economist says the measure was “first suggested by Paul Myners, a minister in Britain’s last Labour government”, the paper itself simply says that the measure was merely “suggested to one of the authors during a meeting with Lord Myners a number of years back, when he had finished compiling his report on the U.K. asset management industry [Myners (2001)].” A quick Google search reveals the concept was apparently used by NatWest as early as 1997 – in fairness, perhaps at the behest of Myners himself who was with NatWest at the time.
The concept seems to have generated the usual cacophony of accusations against asset managers (and hedge fund managers, even though the article was essentially about private equity and mutual funds). This kind of research is bound to bring out the hedge fund bashers (e.g. “Let ‘em all fry in boiling oil till they clean their act!!”). A couple of the comments, however, hinted at the curious implications of such a measure. We thought we’d explore some of of these here.
The first issue is the rather arbitrary choice of the benchmark as the portfolio holdings on January 1st of a given year. Calendar years are the root cause of much of consternation regarding hedge fund fees since they can pervert the manager’s decision making depending on the calendar date (see previous post on the propensity of hedge fund managers to dial down risk after June, for example).
But let’s say that the “inert” benchmark is just any arbitrarily chosen date. What if the portfolio was a total dog on that date? Could the manager game this metric by window dressing on Day One (or on every successive January 1st). Except, unlike regular window dressing, this display would be the ugliest, meanest thing you ever saw. By setting their own benchmark, managers are in total conflict of interest. Ergo, the inertia benchmark” only really seems to work as long as the manager is oblivious to it (or at least that they are adequately incentivized to ignore it). Back in 1997, The Independent wrote that:
…outperforming the inertia index does not in itself guarantee that a trust will be among the best performing in its sector when you measure absolute returns. The reason for this is that no amount of skill can make up for being in the wrong markets in the first place…
But let’s assume the manager actually puts their best foot forward on Day One and creates the best portfolio they can. Let’s say it’s full of Apple Inc., for example. Then let’s say Steve Jobs needs a liver transplant at some point after Day One and the manager dials back on their Apple exposure as a result. Then let’s say the manager locks in the portfolio and sits on their couch the rest of the year.
Now, let’s assume Jobs goes to, say, Tennessee, to pick up a new liver. He then stages a triumphant return and Apple continues to take over the world. The fund manager gets up from the couch to check emails and finds a barrage of email from clients demanding he step down because he underperformed his “inertia benchmark.”
To be sure, the manager seems to have made a bad call. But no manager is perfect and who could have known what the future held for AAPL? In fact, the best managers are right only a small majority of the time (indeed, many are right less than half the time, but make up for it with hefty winners and modest losers). So the manager can’t really be criticized too heavily for under performing an arbitrarily-selected previous “state” of their own portfolio.
Let’s now say the manager got up from the couch part way through the year and made a few trades. The result is a new portfolio state. They can now either flop back on the couch until Christmas or could keep trading and create a new portfolio state. Essentially, there is a binary decision to be made: lock-in the portfolio (make it “inert”) until the end of the year or remain active. This decision point could come once, twice, three times, or perhaps many times each day throughout the year.
Like TV’s “Lost”, there could be multiple parallel universes, each one sprouting from a discrete decision to render the portfolio inert in each portfolio state. The problem for the manager is that any one of those portfolios stands a chance of blowing away the managers actual annual return if kept inert until year end. In this way, a demanding investor might come to expect a kind of lookback call option with a floating strike on the fund, where the strike price would be the return of the best performing discrete portfolio state during the year.
While this sounds much more complex than the “inertia benchmark” mentioned in the articles above (which is simply cast on January 1st), the inertia benchmark might just as well have represented the n-th portfolio state in a long string of January 1st portfolio states before it.
The bottom line is that a single inert portfolio benchmark stands a good chance of being either a star or a dog, and that the more inert portfolios benchmarks (/previous states) you have, the more likely its is that one of them will make all future end-states look terrible. Besides, shouldn’t the manager get some credit for picking a winning Day One portfolio?
Even if you did adopt a simple inertia benchmark as described above, you’d have to reset the clock for every new investor subscription, lest those new investors use a previous portfolio state from before they invested in the fund. (This would be akin to the problem hedge fund administrators face when new investors enter a fund that has a fund-wide high water mark or hurdle).
In any case, our concerns about the inertia portfolio shouldn’t be taken as an endorsement of portfolio churning, however. As the paper points out:
The fact that managers know that they can only benefit from trading excessively means that they might end up making trades that they might otherwise have not had they also been forced to pay part of that fee themselves.
And judging from the following chart buried deep in the Roger Ibbotson paper we mentioned yesterday, more than a few churn-happy portfolio managers could use a little more down time on the couch…