Working Paper: CEPR ID: DP16958
Authors: George Korniotis; Stefanos Delikouras
Abstract: Testing consumption-based asset pricing models is a triple hypothesis, which requires selecting the consumption data and examining the assumptions for investor preferences and consumption dynamics. We formalize the triple-hypothesis problem within a GMM framework that relaxes the CRRA assumption, jointly estimates consumption dynamics with Euler equations, and includes the risk-free rate variance as a target moment. We find that consumption measures proposed as alternatives to the BEA consumption (e.g., garbage) do not address the shortcomings of the canonical CRRA model with i.i.d. consumption growth. Instead, a model with BEA consumption, Epstein-Zin preferences, and non-i.i.d. consumption dynamics performs equally well.
Keywords: cross-section of expected returns; Epstein-Zin preferences; risk-free rate; GMM; consumption growth
JEL Codes: D51; D91; E21; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| choice of consumption measure (D12) | estimated risk aversion coefficients (D11) |
| choice of consumption measure (D12) | cross-sectional fit of asset returns (C58) |
| variance of the risk-free rate (G17) | estimation of risk aversion and EIS parameters (D81) |
| choice of utility function (D11) | empirical fit of asset pricing models (C58) |