Essays on Asset Pricing and International Finance

Abstract

In this dissertation, I empirically analyze how asset prices, especially exchange rates and stock prices, are determined. In chapter 1, I develop and estimate an empirical model of exchange rates that can explain four puzzles about the relationship between exchange rates and interest rates. The delayed reaction model of Bacchetta and Van Wincoop (2021) or the latent financial shock models of Itskhoki and Mukhin (2021) and Valchev (2020) can explain various puzzles in international finance, including the Fama puzzle, the delayed overshooting puzzle, the predictability reversal puzzle, and the Engel puzzle. In this paper, we develop an empirical model of exchange rate dynamics that encompasses these models as well as the UIP model and the risk premium model. Our empirical results based on the G6 portfolio and the individual currencies suggest that most of the persistence in the expected exchange rate change stems from the delayed reaction feature. We also show that the delayed reaction and the missing premium, only when combined together, can empirically explain the above-mentioned puzzles. In particular, the missing premium plays a critical role in our explanation of the Engle puzzle. These results are robust regardless of whether we employ the interest rate differential data [1980:M1-2007:M12] or the year-on-year inflation differential data [2000:M1-2019:M12] as our explanatory variable. Chapter 2 extends Bansal and Yaron’s (2004) long-run risks model by allowing for regime switching in the steady-state volatilities of the shocks. In addition to the persistence of the long-run component, the persistence of both inter-regime and intra-regime volatilities play important roles in our model. Additional extensions include allowing for i) serial correlation in the idiosyncratic component of the dividend growth process and ii) non-zero correlations between shocks to the common long-run and the idiosyncratic components. We first provide empirical relevances of our extensions and provide a solution to the model. We then calibrate our model based on the choice of the preference parameter values in the literature and our estimates of the model parameters. The resulting equity premium ranges between 2.36% and 5.7%, suggesting that, in order to explain the historical equity premium of 6%, a considerably higher risk aversion parameter is necessary than is assumed by Bansal and Yaron (2004). Chapter 3 investigates the time variation of the relative importance of the news on future cash flows and news on future returns in explaining the stock return variance, by combining the VAR-based stock return decomposition in Campbell (1991) and Campbell and Ammer (1993) and the time-varying parameter VAR with stochastic volatilities (TVPVAR-SV) in Primiceri (2005). This new approach provides two empirical findings. First, the stock market crashes in recessions because of high discount rates news, and recovery of the stock market in booms is associated with high cash flow news. Second, there is a reversal in the relative importance of CFs and DRs in explaining the stock return variance in the 1990s. Then we propose two explanations for this reversal. First, the importance of the cash flow news in stock return variance may have increased since the 1990s because of the recent developments in information technology. Second, a decline in macroeconomic volatility since the 1990s can cause less volatile DRs, if investors make inferences on the macroeconomic condition and the corresponding expected return by extracting signals from observable variables.