Seminar

Pareto Distributions in International Trade: Hard to Identify, Easy to Estimate

Marnix Amand (HEC Lausanne)

May 17, 2016, 17:00–18:30

Room MS 001

Macroeconomics Seminar

Abstract

We show that many forms of heterogeneity and uncertainty that are common in the heterogeneous firm international trade literature drive a wedge between observed exports in the data and firm productivity. This implies that testing for an exact Pareto distribution for firm productivity is impossible, barring unreasonable assumptions. Furthermore, the presence of this wedge, which is possibly correlated with the productivity distribution, means that econometric methods that rely on exports to directly estimate the productivity distribution are misspecified. To overcome this, we provide a formal result with a straightforward economic interpretation that allows to estimate the tail of a power law irrespective of the presence of such a (correlated) wedge. Lastly, we show that this wedge can distort the left end of the observed exports, often making it \look" log-normal, even if productivity is exactly Pareto distributed. Hence our interpretation of recent empirical work that rejects a Pareto distribution in exports data is that these results do not necessarily reject the assumption of a Pareto distribution for productivity.