Yesterday, we covered the first half of an interesting academic study on hedge fund redemption terms. We learned that the auto-correlation of monthly returns was positively related to the actual redemption terms of a fund. Today, we cover the second half of this paper by MIT’s Amir Khandani and Andrew Lo – in which the duo estimate an actual premium associated with redemption terms.
Rather than comparing actual redemption terms to see if illiquid funds perform better than liquid ones, the duo can now compare autocorrelation to (risk adjusted) returns. In the table from their paper reproduced below, they create 5 separate hypothetical portfolios based on autocorrelation buckets (click to enlarge). Lo and behold (pardon the pun), the funds with the highest autocorrelation of returns generally also had the highest risk-adjusted returns.
Note that while high auto-correlation is generally associated with higher risk-adjusted returns, there are some exceptions. In particular, Khandani and Lo highlight the fact that the Global Macro funds with the highest auto-correlation (lowest liquidity) actually posted below average results. Also, we noted that very few of the relationships were monotonic (with the best performing funds of funds, for example, being the ones with the second lowest autocorrelation).
Okay. So you read yesterday’s post and you get the idea above. But so what, right?
It turns out that by essentially going long the illiquid (better performing) portfolios and short the liquid (worse performing) portfolios, you can squeeze out the following return streams for each strategy:
Again, the weird results of the global macro funds sticks out like a sore thumb. But the rest of these hypothetical portfolios perform positively over the time period examined. Note that the strategies that we know intuitively as being highly illiquid had the highest cumulative returns.
Overall, the CAGR of the average fund (across all samples) was 3.96%. In other words, the risk premium associated with hedge fund illiquidity – at least according to this methodology – is 3.96% per annum.