Working Paper: NBER ID: w24270
Authors: Shmuel M. Bartram; Gregory W. Brown; Ren M. Stulz
Abstract: Since 1965, average idiosyncratic risk (IR) has never been lower than in recent years. In contrast to the high IR in the late 1990s that has drawn considerable attention in the literature, average market-model IR is 44% lower in 2013-2017 than in 1996-2000. Macroeconomic variables help explain why IR is lower, but using only macroeconomic variables leads to large prediction errors compared to using only firm-level variables. As a result of the dramatic change in the number and composition of listed firms since the late 1990s, listed firms are larger and older. Larger and older firms have lower idiosyncratic risk. Models that use firm characteristics to predict firm-level idiosyncratic risk estimated over 1963-2012 can largely or completely explain why IR is low over 2013-2017. The same changes that bring about historically low IR lead to unusually high market-model R-squareds.
Keywords: idiosyncratic risk; firm characteristics; macroeconomic variables
JEL Codes: G10; G11; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Macroeconomic Variables (E19) | Idiosyncratic Risk (D81) |
| Firm Characteristics (L25) | Idiosyncratic Risk (D81) |
| Historical Low Idiosyncratic Risk (N21) | Firm Characteristics (L25) |
| Macroeconomic Variables (E19) | Prediction Errors (C52) |
| Firm Characteristics (L25) | Low Idiosyncratic Risk (G39) |