I just returned from the AFA 2016 Annual meeting in San Francisco. I found two papers to be significant in regards to return predictability:
Dash for Cash: Month-end Liquidity Needs and the Predictability of Stock Returns, by Kalle Rinne, Matti Suominen, Lauri Vaittinen
Abstract. This paper uncovers strong return reversals in the US value weighted stock market index around the last monthly settlement day, T-3, which guarantees liquidity for month-end cash distributions. Similar reversals in market returns around T-3 are documented internationally. The return reversals are stronger in countries where the mutual fund ownership is large, and in the US they have become stronger over time as the mutual fund ownership of stocks has increased. Using data that contains all trades of a subset of institutional investors, we show direct evidence that institutional trading contributes to the market reversals. Finally, we find that in the cross-section of stocks, return reversals around the turn of the month are stronger for stocks more commonly held by mutual funds and for liquid stocks. These market reversals help explain the previously documented abnormally high market returns around the turn of the month. Key words: asset pricing, limits of arbitrage, mutual funds, short-term reversals, turn-of-the-month effect JEL classification: G10, G12, G13
The authors make a strong case for a causal relationship of pension funds needing to sell by t-4 and the pensioners having extra money to buy after the first day of the month. We have been using this indicator since 2006 and will continue to use it. What was most interesting is that the magnitude of the effect can be predicted by the TED spread and the return of SPX.
Good and Bad Variance Premia and Expected Returns, by Mete Kilic, Ivan Shaliastovich
Abstract. We measure “good” and “bad” variance premia that capture risk compensations for the realized variation in positive and negative market returns, respectively. The two variance premium components jointly predict excess returns over the next 1 and 2 years with statistically significant negative (positive) coefficients on the good (bad) component. The R2 s reach about 10% for aggregate equity and portfolio returns and about 20% for corporate bond returns. We show that an asset pricing model that features distinct time variation in positive and negative shocks to fundamentals can explain the good and bad variance premium evidence in the data.
We have been using the Variance Risk Premium in our models for some time. All one needs is the VIX and a real time feed that allows you to calculate the realized variance from five minute returns. The problem that I have with good and bad variance is the overhead in obtaining the individual option prices. We estimate this will be about $50,000/year and we are studying to see if there is enough added information to justify the expense.
Although there were fewer papers this year involving aggregate return predictability, all sessions that I attended on the subject were positive. I find it ironic that as I was observing 80 academics arguing over the nuances to predict the equity risk premium using options, just 30 miles away 80 employees of Wealthfront were trying to figure out how to preserve and promote buy and hold.