Articles - Active Management and Survivorship Bias


There is a very simple and powerful argument in favour of passive management. This argument holds more broadly across asset classes, but is simplest to consider in equities, so I’ll focus there. Essentially, the combined portfolio of all active investors must aggregate back up to the market portfolio. Therefore, the average active manager should simply return passive performance, less fees and transaction costs.

 By Alex White, Head of ALM Research, Redington
 
 In one sense this is true. At any point in time, the aggregate portfolio will be the market portfolio. But this misses a key detail. Not every active investor is an active manager. Investors with structural home biases, for example- and retail investors holding personal portfolios- are, in this sense, active investors. Unless there is truly no information to be had, it is hard to see that these investors are competing on a level playing field against a well-equipped and resourced team of professionals.

 So what does the evidence say? There is empirical evidence that active management can lead to positive returns. Looking at the 20 year period from 1997 to 2017, the Morningstar database suggests 31bps of net alpha from 1.65% tracking error; the eVestment database for institutional equities implies 1.16% of net alpha (with 0.5% fees assumed), for 1.42% active risk (sources AQR, Morningstar, eVestment).

 That these values are so different is something of a flag. At least part of the explanation is likely to come from survivorship bias.

 Essentially, managers will grow more, attract more investors, and publish their results more freely if they’ve had strong performance. Those with weaker performance are more likely to fail, and thus disappear from the records. In particular, the eVestment database is self-published, so is likely to suffer from more survivorship bias.

 So we should haircut the evidence, but it’s not obvious by how much. However, we can test this, at least in theory. Assuming a logistic distribution (essentially a normal distribution but with fatter tails, and simpler computationally) for alpha, we can simulate 10,000 funds, with volatilities in line with the history, expected returns of zero, and prescribe rules for when funds fail. Essentially, we ascribe an x% chance (at the end of each period) of excluding any fund that has a period of n years with negative total active performance.

 Clearly, this won’t solve the active-passive debate. But it does show a few things. Firstly, that survivorship bias can be significant; but also, for it to explain a 0.2 Sharpe entirely, we would have to see a fairly punchy half to two-thirds of the original funds drop out of the reckoning. It seems unlikely that survivorship bias could explain all the alpha.

 

 

Back to Index


Similar News to this Story

Stars of the Future Voting Is Now Open
Stars of the Future highlights the rising talent in the actuarial industry. We know how hard you all work to qualify as actuaries and this is the pe
Delivering Future Readiness for Insurance Carriers
Ever the double-edged sword, advances in technology have arguably driven the greatest change to the industry that insurance has ever witnessed. As tec
LGPS Legislative Logjam
The queue is building. Apologies for the alliteration in the title, but a recent client discussion got me thinking about the huge volume of outstandin

Site Search

Exact   Any  

Latest Actuarial Jobs

Actuarial Login

Email
Password
 Jobseeker    Client
Reminder Logon

APA Sponsors

Actuarial Jobs & News Feeds

Jobs RSS News RSS

WikiActuary

Be the first to contribute to our definitive actuarial reference forum. Built by actuaries for actuaries.