Comparison of Welfare Gains in the Armington, Krugman and Melitz Models : Insights Based on a Structural Gravity Approach

How large are the estimated welfare gains from changes in trade costs in a Melitz based trade model compared with models based on Armington and Krugman? We examine model features and scenarios that are important for policy that are unexamined in the literature. Starting with the stylized model of Akolakis et al. (2012), we produce fourteen new results by progressively introducing model structures and scenarios that are relevant to international trade policy analysis. We show that there are policies or model features that result in global welfare gains in the monopolistic competition models that are several multiples of the Armington model results and one relevant policy that leads to opposite signs; on the other hand, we identify other extensions of model features that have very small impacts. In multi-sector, comparative static models, we are the first to compare the welfare impacts of market structure, while holding the trade responses constant based on structural gravity estimates. In response to global changes in trade costs, in all models except a stylized one-sector model, the global welfare changes are largest in absolute value in Melitz and smallest in Armington; and the Krugman model captures between 75 and 95 percent of the additional gains (losses) above the Armington model that are estimated by the Melitz model. We construct parameters that facilitate intuitive explanations of the general pattern of results and for reversed welfare rankings that occur for individual regions