Making Gravity Great Again

Martin, Will

World Bank, Washington, DC

关键词: GRAVITY MODEL;    EATON-KORTUM MAXIMUM-LIKELIHOOD;    TRADE STATISTICS;    MISSING DATA;   

DOI  :  10.1596/1813-9450-9391
RP-ID  :  WPS9391

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

The gravity model is now widely used forpolicy analysis and hypothesis testing, but differentestimators give sharply different parameter estimates andpopular estimators are likely biased because dependentvariables are limited-dependent, error variances arenonconstant and missing data frequently reported as zeros.Monte Carlo analysis based on real-world parameters foraggregate trade shows that the traditional Ordinary LeastSquares estimator in logarithms is strongly biaseddownwards. The popular Poisson Pseudo Maximum Likelihoodmodel also suffers from downward bias. An Eaton-Kortummaximum-likelihood approach dealing with the identifiedsources of bias provides unbiased parameter estimates.