The implications of alternative import assumptions on multipliers: Should we be concerned?

The basic tool for studying the effects of a policy or a structural change in multi-sectorial general equilibrium models is the multiplier matrix. No matter how we dress it up, the results always depend on the multiplier matrix. Therefore, the robustness of this matrix is of paramount importance when we want to report sensible estimates of changes. In an input-output (I-O) model with no external sector—a closed economy assumption—the multiplier matrix is unique and there are no interpretive doubts about the information it provides. But imports exist and are always present in real-world data. And this is where the difficulty arises since the introduction of imports in a model requires establishing assumptions regarding their incorporation with the domestic side of the economy. As a result, the domestic and total multiplier matrices will no longer be unique, and each type of imports' incorporation will give rise to a different estimate of the multiplier matrix. We study in this note three different incorporation of imports that cover the vast majority of cases: non-competitive and competitive imports, and imports that follow a hybrid Armington-Leontief assumption. For each of them we calculate the multiplier effects to find out whether the assumptions about imports matter or not. This is empirically illustrated using OECD I-O data for the U.S.