Abstract:
A large number of recent papers employ value-added trade data alongside traditional gross measures of trade to estimate the impact of various trade costs on bilateral trade. Value-added gravity equations are typically justified by referencing the theoretical and empirical merits of traditional gravity equations for gross trade. Contradicting this notion, we use theory and simulations to show that value-added gravity equations are misspecified when the gross trade gravity equation is correct. Consequently, estimates from value-added gravity equations are difficult to interpret and prone to omitted variables bias.