Working Paper: NBER ID: w25519
Authors: John B. Donaldson; Rajnish Mehra
Abstract: We evaluate the properties of mean reversion and mean aversion in asset prices and returns as commonly characterized in the finance literature. The study is undertaken within a class of well-known dynamic stochastic general equilibrium models and shows that the mean reversion/aversion distinction is largely artificial. We then propose an alternative measure, the ‘Average Crossing Time’ that both unifies these concepts and provides an alternative characterization. Ceteris paribus, mean reverting processes have a relatively shorter average crossing time as compared to mean averting processes.
Keywords: mean reversion; mean aversion; average crossing time; asset pricing; dynamic stochastic general equilibrium
JEL Codes: C13; C53; E3; E44; G1
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| mean reverting processes (C22) | shorter ACT (Y60) |
| mean averting processes (D81) | longer ACT (C41) |
| ACT (Y20) | ranking strength of mean reversion or aversion (D81) |
| negative autocorrelation (C29) | mean reversion (C22) |
| positive autocorrelation (C22) | mean aversion (D11) |