Essays in Empirical Asset Pricing and Macro-Finance

This dissertation is composed of three essays in Empirical Asset Pricing and Macro-Finance. In the first essay, titled "Can Time-Varying Risk Premia and Household Heterogeneity Explain Credit Cycles?," I use micro-level data from almost 50 million mortgages to measure the dispersion in the credit quality of borrowers in the housing market. I show that credit dispersion forecasts regional real economic activity and provide empirical evidence that associates the predictive power of dispersion with heterogeneity in the exposure of households' labor income to economy-wide shocks. I explain these observations in a model featuring time-varying risk premia, incomplete markets, and household heterogeneity. Due to risk aversion, the consumption and investment responses of households have a convex association with their labor income exposure to aggregate risks. As a result, dispersion forecasts the aggregate output more strongly in more heterogeneous regions, consistent with the data. The second essay is joint work with Mete Kilic and Sang Byung Seo. We develop a model that generates slowly unfolding disasters not only in the macroeconomy but also in financial markets. In our model, investors cannot exactly distinguish whether the economy is experiencing a mild/temporary downturn or is on the verge of a severe/prolonged disaster. Due to imperfect information, disaster periods are not fully identified by investors ex ante. Bayesian learning induces equity prices to gradually react to persistent consumption declines, which plays a critical role in explaining the VIX, variance risk premium, and put-protected portfolio returns. We show that our model can rationalize the market patterns of recent major crises, such as the dot-com bubble burst, Great Recession, and COVID-19 crisis, through investors’ belief channel. In the last essay, "Is There a Macro-Announcement Premium?," co-authored with Sang Byung Seo, we argue that the average excess return over macro-announcement days substantially exaggerates the true risk premium. The conditional volatility of returns barely drops at macro-announcements. This is at odds with virtually all models that justify high macro-announcement returns through a high announcement premium. We propose an alternative explanation: macro-announcement days are, on average, with good news in existing sample periods. We develop a novel estimation approach, which reveals that high macro-announcement returns are not a manifestation of high conditional equity premiums but positive return innovations that are not averaged out in-sample. We find that macro-announcement days do not seem to operate with a separate mechanism: the patterns of macro-announcement days are well replicated by random samples from non-announcement days. Our analysis suggests that the large average macro-announcement return might not be compensation for perceived uncertainty.