Import demand in heterogeneous panel data with cross-sectional dependence

We investigate the long-run income and price elasticity of import demand functions with a heterogeneous unbalanced panel of 34 countries over the period 1985:q1-2018:q3. To estimate world elasticities the model is tested with the activity variables derived from the theoretical and empirical literature: GDP, GDP minus exports, Private Demand, Aggregate Domestic Demand, National Cash Flow, and Import intensity-Adjusted Demand (IAD). First, we evaluate time series properties using second generation panel unit root and cointegration tests. Second, we rely on the dynamic common correlated effects mean groups (CCEMG) estimator to deal with cross-sectional dependence (CSD). We find that the IAD, whose world elasticity is close to one, is the best performing specification. Our results confirm that the most appropriate activity variable to assess import demand should encompass intermediate goods as suggested by the recent literature on global supply chains. Moreover, we partially solve the puzzle of the recent trade slowdown since, taking stock of the role of intermediates, the time needed to resort to the long run equilibrium in the aftermath of a global turmoil is greater than that predicted by previous studies.