Three essays on international economics

Abstract

This dissertation comprises of three essays in international macroeconomics. The first essay investigates the competition between two city states, both of which will stand in place of countries in the global scheme. Under the framework of the three-stages-game, we assume that there are two cities competing for dominance over two sectors: the manufacturing sector and the financial sector. In addition, the government of each city state can build infrastructure to increase the competitiveness of the financial and distributive firms of its city. Under this framework, we are able to show that the amount of resources, the start-up costs of providing services, and the relative effectiveness of their infrastructures determine the optimal amounts of infrastructures the cities decide to build, and thus also decide the equilibrium outcome of this game. In my second essay, we examine the relationship between income distribution and import patterns. The Linder hypothesis states that countries with similar economic characteristics should trade more often. However, although the total volumes of trade between these countries are similar, the traded goods may be different. This paper investigates the trading patterns of countries with similar characteristics. Specifically, we analyze the relationship between the import patterns and income distributions of importers. We develop an import similarity index to portray the composition of imports and utilize the idea of a "market overlap," a theoretical concept proposed by Bohman and Nilsson (2007), to represent the similarity of income distributions across different importing countries. We provide empirical evidence to support the notion that countries with similar income distributions display similar import patterns. We also separate countries by income level and find that income distribution exerts a positive impact on the similarity of import patterns for all but low income countries. Finally, we incorporate the characteristics of goods into our analysis and show that the positive relationship between income distributions and import patterns holds for differentiated and reference-priced goods, but not for homogeneous goods. In my final essay, we look into another aspect of international literature: the exchange rate. In the literature, we find that vector autoregressive (VAR) models and impulse response analyses are common tools to study the relationship between monetary policy and exchange rate movements. Therefore, it is important to investigate the accuracy of the VAR model. In the first part of this essay, we assume that the true, underlying, data-generating process is hump-shaped, which is the shape of the impulse response of exchange rate to a monetary policy shock. We show that results estimated from any VAR models applying AIC as their lags selection are biased. We also introduce two possible solutions to remedy this bias: the use of more lags in the VAR models or the use of the proposed loss functions estimations. These results suggest we should be cautious when interpreting empirical evidences on international literature. In the second part of the same essay, we investigate another issue that is closely related to the exchange rate and the VAR model. Under the estimation of the VAR model, the researcher implicitly assumes that the objective loss function is quadratic. However, it is a well accepted fact that monetary authority adjusts the interest rate according to policy. One of the objectives of the monetary authority is to influence the exchange rate in their favor. They estimate the size of the loss caused by deviations from the current exchange rate to the rate they desire, and then they adjust the amount of money in the international market. We propose an asymmetric loss function that monetary authorities may use to estimate the impulse response of the exchange rate to a contractionary monetary policy shock. We then compare these estimated impulse response functions to those estimated by the VAR. We find that while both of these estimated impulse response functions share the same sign, the magnitude and the duration of the shock are quite different. These results suggest that the VAR model may not be appropriate in estimating the exchange rate movement.