The performance of asset managers

I have made only one specific investment recommendation on this weblog so far and it has turned out to be rather poor advice, at least in the short term. I questioned the wisdom of investing in the Carlyle IPO almost two years ago. But Carlyle has comfortably outperformed both the market and its sector since then. Naturally, if you read my disclaimer, you know that I am quite adamant that you should not listen to anything I say about future market movements. But despite Carlyle’s good performance, I remain skeptical about listed asset managers. The fundamental issues that I mentioned in that post remain in place.

But now there seems to be empirical evidence to prove me wrong. Guinness Asset Management, a UK fund manager, have a fund that invests exclusively in asset managers. In their marketing literature, they show that asset managers outperformed the MSCI World index by over ten per cent per annum over the past twenty-four years.

This would be a spectacular result if it were true. But is it? The fund was launched only in 2011, so their results are based on a backtest. The quality of backtests are notoriously tricky to assess so we need to look at the small print. There we find that the investment universe consists of 36 listed asset managers that were in existence at the end of the year 2000, a completely arbitrary date. If we were to move that date just a few months in either direction, we would come up with a somewhat different universe. As an example, UAM, the asset management conglomerate, was purchased in late 2000 by Old Mutual. Its performance in the years before the acquisition was rather poor. Looking at the 36 firms, it seems that there is a further, unstated condition for inclusion: survival until the date when the analysis was performed (sometime in 2010?). They are also equally-weighted.

It is not necessary to be particularly cynical to suspect that the admittedly spectacular track record that Guinness claim may have benefited from a few built-in favorable biases. The equally-weighted portfolio is compared to a market cap weighted benchmark in a period when small caps comfortably beat large caps. Sticking to survivors also helps – if we could be certain that all our equity investments survived, they would be much less risky. There is ex post selection bias as well as survival bias. I am not suggesting that the manager is actively trying to deceive. But they seem to exploit the leeway that regulators give asset managers in presenting simulated performance numbers rather aggressively. Of course, they may be unaware of the biases that they have introduced into their results.

How could we avoid or at least reduce these biases? Ideally, the investment universe would include all asset management firms that were listed at any point within the sample period. That would ensure that disasters like Artio Global, the Julius Baer spinoff, which lost about ninety per cent of its market value between its IPO and the time when it was put out of its misery by Aberdeen, would also have been included in the sample. We would also want to report both value- and equally-weighted numbers to illustrate the impact of any small cap bias and the effect of rebalancing.

Despite the data problems, it is quite likely that asset managers did outperform over the past twenty-four years. It is difficult to estimate the performance impact of the various biases. But if we take the results on the impact of the ex post selection and look back biases in hedge funds as a guide and add the small cap effect, it is likely that rather more than half of the claimed outperformance would disappear. In addition, asset management firms are essentially levered bets on the stock market, so they ought to have outperformed in a period in which equity markets showed above average performance.

To conclude, I remain unconvinced on the merits of investing in listed asset managers. I would need to see more convincing, cleaner data – less polluted by data mining – and better arguments in order to change my mind.