Working Paper: NBER ID: w31204
Authors: David Hirshleifer; Dat Y. Mai; Kuntara Pukthuanthong
Abstract: Using a semi-supervised topic model on 7,000,000 New York Times articles spanning 160 years, we test whether topics of media discourse predict future stock market excess returns to test rational and behavioral hypotheses about market valuation of disaster risk. Media discourse data address the challenge of sample size even when disasters are rare. Our methodology avoids look-ahead bias and addresses semantic shifts. Our discourse topics positively predict market excess returns, with War having an out-of-sample R² of 1.35%. We call this effect the war return premium. The war return premium has increased in more recent time periods.
Keywords: war; disaster risk; asset pricing; media discourse; stock market returns; bond market returns
JEL Codes: G00; G01; G02; G1; G10; G11; G13; G4; G41
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| war discourse (H56) | stock market returns (G17) |
| war discourse (H56) | mid to long-term high-yield corporate bonds returns (G12) |
| war discourse (H56) | short-term government bonds returns (E43) |
| war discourse (H56) | investment-grade corporate bonds returns (G12) |
| war discourse (H56) | market behavior (D40) |