Essays on International Finance Puzzles

This dissertation studies two puzzles in international finance which have become more relevant since the outbreak of the financial crisis in 2007. The first puzzle is the U.S. valuation effect (VE) channel. I show that a positive valuation effect – when changes in the value of external assets are higher than changes in the value of external liabilities – mitigated the impact of the persistent current account deficit on the NIIP until 2007. Since 2008, however, valuation losses coming from changes in asset prices and exchange rates have contributed to the deterioration of the NIIP despite the current account improvement. The reason for this, as I demonstrate, is that the U.S. NIIP tends to decrease if the return on its external assets is less than the return on its external liabilities. To explain these empirical findings, I build a two-country, two-asset model of portfolio balancing where the interaction between investors\u27 risk-return preferences, optimal demands for assets, and home bias can account for the post-2007 drop in the U.S. return differential. The return differential is explained by an increase in foreign asset risk, a decrease in domestic asset risk, an increase in the risk aversion for both domestic and foreign investors, or any combination of these factors. The second puzzle is the failure of the uncovered interest rate parity (UIP) condition for major currencies. After running the Fama regression over low and high volatility periods, contrary to the common view, I demonstrate that an arbitrageur can make profits by borrowing in high-interest-rate currencies and lending in low-interest-rate currencies during high-volatility times. The opposite is true. Moreover, my results from the time-varying conditional correlation analysis suggest that the correlation between the expected depreciation and interest rate differential is volatility-regime dependent, and even positive from 2008-2017. This finding contradicts that the UIP puzzle comes from the negative unconditional correlation between these two variables. Lastly, I document deviations from the CIP in the post-crisis period which show that there are arbitrage opportunities at very short horizons. An important implication is that forward premiums and interest rate differentials cannot be used interchangeably in the Fama regression