On The Performance Of Bootstrap Confidence Intervals For Multilateral Resistance Terms Obtained From A Structural Gravity Model

Structural gravity models offer a tractable approach to estimate trade policy effects on international trade in a general equilibrium framework.We assess the performance of bootstrap pivotal confidence intervals for the conditional general equilibrium effects of structural gravity initially proposed by Anderson and Yotov(2016).With real data obtained from WTO and CEPII websites and Monte Carlo simulated trade flows that are consistent with the structural gravity model,we illustrate a general equilibrium gravity analysis by defining a counterfactual scenario where a free trade agreement is abolished across all countries using the Poisson Pseudo-Maximum-Likelihood(PPML)estimator.We construct percentage changes in the multilateral resistance(MR)terms of every country in our sample.By construction,the true values of the effects are obtained using the actual trade flows.With the Monte Carlo simulated trade flows,we follow the suggested procedure by Anderson and Yotov(2016)to construct bootstrap pivotal 95%confidence intervals for the MR terms.Their bootstrap algorithm is repeated 1,000 times to assess whether the empirical coverage probability of the said interval is close to its specified coverage probability.Our results reveal a 100%empirical probability coverage of the 95%pivotal confidence intervals for both indexes.