Agricultural Export Subsidies and Intermediate Goods Trade: Comment
评论Paarlberg关于出口补贴对中间品贸易影响的分析，指出其模型存在笔误和计算繁琐，并提出使用出口供给函数和受限利润函数来更简洁地推导核心结果。
Paarlberg examines the impact of an export subsidy policy when one good is an intermediate input in production of a higher-value product (HVP) which is also exported. His analysis suggests that the sector receiving the subsidy is one that benefits, while an export subsidy on one good can cause the home country's price for the other good to rise or fall, depending on the relative subsidized commodity excess demand and supply elasticities, and the relative composition of the home and foreign sectors. Paarlberg's analysis relies on a two-country, two-good, sector-specific factor partial equilibrium model. In each country, the two goods are produced under competitive conditions using technologies with constant returns to scale. The comparative statics of a home country export policy are determined through holding the industry-specific factor supplies constant and by assuming no policy intervention initially. Furthermore, in order to facilitate interpretation of the comparative static results, Paarlberg assumes that the technologies in the processing sectors of the two countries are identical, i.e., that per unit factor uses of the home and foreign HVP industries are identical for all equal factor prices. Deriving Paarlberg's comparative static results can be difficult, due to typographical errors and/or cumbersome calculations, particularly in the appendix. Perhaps this comment can then be of some assistance in so far as we show that Paarlberg's central results may readily be obtained and interpreted from the definition and the properties of export supply functions (Dixit and Norman, Feenstra). Our analytical framework is based on the use of sectoral restricted profit functions rather than unrestricted cost functions. It is more general than Paarlberg's framework in so far as we do not assume that elasticities of substitution in the HVP industries for the two countries are identical and we do not necessarily assume that returns to scale are constant. We define