A balls-and-bins model of trade: Comment

We show that the Armenter and Koren model's firm-product-country results rely on the assumption that export shipment size is independent of firm size, and this assumption is contradicted by the data. When actual shipment sizes are used in the balls-and-bins model, it cannot reproduce the data on single product/single country exporters. Beyond just showing that the shipment size assumption matters to balls-and-bins outcomes, our results highlight the important fact that shipment size is an economic decision, co-determined with other export choices. For this reason, we argue that a balls-and-bins model cannot be a purely statistical benchmark model.