Asymmetries in trade flows and international specialization: Evidence from the Asian Input-Output tables

Laura Puzzello, Purdue University

Abstract

Chapter 1 of this thesis revisits Reimer (2006), and Trefler and Zhu (2005, 2006) (RTZ) tests of the Vanek proposition in the presence of international differences in production techniques and global production sharing. In this framework, knowing the bilateral details of each country's input-output structure is key to the correct calculation of the factor content of trade. Because input-output tables typically lack this detail, RTZ impute the relevant input-output coefficients using a method that implicitly assumes that international flows of goods respond to trade determinants independently of their end-use ( Symmetry Hypothesis). This chapter uses survey-based input-output coefficients from the Asian Input- Output tables (AIO) that do provide bilateral detail. Exploiting methodological differences in the compilation of the AIO tables and the data underlying RTZ studies, the symmetry hypothesis is empirically tested and found to fail. This failure causes input-output data imputed following RTZ methodology to overstate the gross quantity of both domestic and foreign factors' services embodied in a country's trade. However, both biases are systematic and tend to cancel each other out resulting in only a small positive bias on net flows of factors and in the performance of the Vanek proposition. Chapter 2 examines the determinants of input trade using data from the Asian IO tables. We employ an extension of a widely used model of intermediate input trade in which inputs and final goods are considered symmetric up to differences in expenditure shares. This provides a null hypothesis that inputs and final goods are determined by the same factors. Our estimates provide the following insights. One, the extent of industrial absorption relative to final consumption as measured by input-output tables does help explain the input share of trade. Two, contrary to the maintained assumption of symmetry from our null hypothesis, this is not the only determinant of input trade. Input trade is more likely to be characterized by zeros, less sensitive to factor endowment differences than final goods trade, and more sensitive to trade costs as measured by home bias, contiguity and common language. However, the role of home bias and contiguity is eliminated by the year 2000, consistent with the popular view of the internationalization of input trade. Chapter 3 is in response to Adda and Cornaglia (AC, 2006) article which analyzes the effect of cigarettes taxes on smokers' behaviors. Their work is widely cited because their results suggest taxes on cigarettes may be harmful to smokers' health. Taxes induce smokers to reduce the number of cigarettes smoked but make them smoke each cigarette more intensively (e.g. smokers decrease time between puffs, smoke each cigarette up to the filter or take off the filter altogether). In this chapter, we consider the interpretation and robustness of the empirical results in AC. After replicating AC's original results, we repeat the analysis using a larger sample now available from the NHANES III dataset that covers the same years as the original AC data. For both the original AC sample and the new expanded sample, we fail to find any strong evidence of compensatory behavior from the estimated cigarette and cotinine elasticities. As a robustness check, we also estimate models using cigarette-price data rather than cigarette-tax data. We find that all of the elasticity estimates become statistically insignificant in the price specifications. Failing to find strong support for AC's claim of compensatory behavior in the overall sample, we proceed to estimate cigarette and cotinine elasticities for different subsamples of the data. We do find little systematic evidence of compensatory behavior among the subsamples.

Degree

Ph.D.

Advisors

Hummels, Purdue University.

Subject Area

Economics

COinS