Essays on the role of income in trade

Kwan Yong Lee, Purdue University

Abstract

Much of standard trade theory focuses on supply side explanations for trade and uses homothetic demand structures to rule out income effects. In my dissertation, I examine the role of non-homothetic preferences in driving changes in traded good expenditures. I use a linear expenditure system derived from Stone-Geary preferences in which a subsistence good is subject to a minimum consumption requirement (MCR). The MCR leads to two important properties: expenditure shares for traded goods are rising with income, while income elasticities for traded goods are declining in incomes. I examine the two properties by constructing a synthetic panel of household expenditures using data from the Consumer Expenditure Survey (CEX), and estimating income elasticities of demand that vary across product x income level x time. I find evidence supporting the properties, in line with the theory. I also find that the income elasticities vary across traded goods. By combining data on the distribution of income shocks with the estimates of income-specific expenditure shares and income elasticities, I construct expected changes in expenditures specific to product x income level x time period. I find that income changes alone explain about half of the change in traded goods expenditures. Using the two properties and the expected changes in expenditures, I examine a novel channel driving volatility in traded good expenditures and a sharp decline in US imports during the Great Trade Collapse (GTC) of 2008-2009. Combined with a shock to incomes concentrated in the left tail of the income distribution, I find large movements in expenditures in traded goods. Aggregating the expected changes in expenditure over income bins, I construct a measure of expected expenditure change, at the national level, arising only from income shocks. Using this expected expenditure change as an instrumental variable for the actual expenditure change, I find evidence that income shocks throughout the GTC period generate tremendous cross-product variation in expenditures and imports. Lastly, I compare alternative non-homothetic demand systems widely used in the demand literature, but not in the trade literature. These systems include the linear expenditure system, NBR system, almost ideal demand system, quadratic expenditure system, and quadratic almost ideal demand system. I compare these systems based on two criteria: goodness-of-model-fit and the effect of shocks to the distribution of income on aggregate expenditure. Specifically, using the CEX and US Consumer Price Index data, I examine whether additional components such as curvature in the income, product prices, and an interaction between prices and the income improve goodness-of-model-fit. In addition, I test whether or not including product prices affects income elasticities. I find that goodness-of-model fit is not sensitive to the choice of demand systems. Also, I find that product prices barely affect estimates of income elasticities. Finally, I find that the distribution of income shock affects aggregate expenditure on traded goods in all demand systems except the linear expenditure system.

Degree

Ph.D.

Advisors

Hummels, Purdue University.

Subject Area

Economic theory

COinS