Blum, Claro, and Horstmann (2016) make two statements about the balls-and-bins model of Armenter and Koren (2014). First, that using firm-level shipment data changes some of our results. Second, that the balls-and-bins model is not an appropriate statistical method. We respond to the first statement and argue that the second statement is unfounded and unrelated to the first. Indeed, the work of Blum, Claro, and Horstmann (2016) is a perfect example of how to use balls-and-bins in a rich dataset to spot interesting data patterns.

Many of the facts about the extensive margin of trade—which firms export, and how many products are sent to how many destinations— are consistent with a surprisingly large class of trade models because of the sparse nature of trade data. We propose a statistical model to account for sparsity, formalizing the assignment of trade shipments to country, product, and firm categories as balls falling into bins. The balls-and-bins model quantitatively reproduces the pattern of zero product- and firm-level trade flows across export destinations, and the frequency of multiproduct, multidestination exporters. In contrast, balls-and-bins overpredicts the fraction of exporting firms.