The overall theme of this dissertation is the explanation of puzzles in international finance. Empirically, exchange rates seem to be disconnected to the economic fundamentals, and it is referred to as the exchange rate disconnect puzzle. Another puzzling feature in foreign exchange market is that high interest rate currency tends to appreciate, and it is called as ``uncovered interest rate parity puzzle.'' Chapter 1 and 3 of this dissertation examine the exchange rate disconnect puzzle, while chapter 2 investigates the explanation for the UIP puzzle. The first chapter, ``Imperfect Proxies for Market Expectations and the Exchange Rate Disconnect Puzzle'', develop an econometric framework which can capture the relation between exchange rate and economic variables. Conventional empirical studies assume the linear relation between exchange rate and its determinants implied by the theory. I show that this linear modeling strategy leads to the spurious instance of the exchange rate disconnect puzzle and propose the new model which allows imperfectness of the macro variables as a predictor for market expectation. The proposed model provides empirical evidence that the domestic currency appreciates in response to an unanticipated increase in domestic output growth or inflation. Furthermore, results for out-of-sample predictability tests suggest that the proposed model outperforms the random walk model over various horizons less than two years, for most of the countries under investigation. The second chapter of my dissertation, ``Is It Risk or Expectational Error? Explaining Deviation from Uncovered Interest Parity'' explores the behavior of ex-ante excess return to explain the UIP puzzle. Implementing empirical models of ex-ante excess return has proven to be very difficult and previous attempts have not been successful in explaining what makes ex-ante excess return. In this chapter, I propose the new framework which estimates the ex-ante excess return more efficiently by incorporating information in economic variables. The extracted series show that high inflation or output in the foreign country raises the ex-ante excess return for holding foreign currency, while high inflation or high output in home country lowers it. Moreover, using the survey-based forecast of exchange rate data, I find that ex-ante excess return is strongly connected with the market's systematic forecast error instead of with the implied risk premium. These empirical findings suggest that the market's expectation is not fully rational, and this systematic expectational error results in the UIP puzzle. Lastly, the third chapter, ``Commodity Currency Predictions: the Role of Expectations'', examines the dynamic linkage between commodity prices and exchange rate. Even though exchange rates and commodity prices are highly correlated contemporaneously, commodity prices are not shown to have predictive power for exchange rates. With several time-series techniques and alternative data, such as survey-based forecast of exchange rate and foreign exchange option prices, I show that commodity price is linked to the future exchange rate through the market expectation: markets consider aggregate commodity prices when they form expectations of the exchange rates. These empirical findings suggest that commodity price movements are incorporated into the nominal exchange rate with lasting impact beyond one quarter.