The Impact of Exporting: Evidence from a Randomized Experiment

The evidence clearly suggests that exporting enterprises are more productive than enterprises that only sell domestically. The traditional explanation is that since export market participation is costly, the best enterprises self-select into export markets. Many policymakers believe, however, that export-oriented growth raises productivity through a variety of channels, such as learning from buyers, the need to exert greater effort in competitive foreign markets and the ability to exploit economies of scale. Understanding whether only "good" firms export or whether exporting helps firms become "good" is difficult, because firms endogenously choose to export.

By using a randomization methodology, these researchers hope to disentangle the causal impacts of exporting. The experimental design and the small-scale enterprise context in Egypt allows them to answer three questions. First, by focusing on the carpet industry, where productivity can be accurately measured, they can assess the causal impact of exposure to foreign markets on productivity by comparing outcomes between the treatment and control groups. The data also allows them to understand the mechanisms through which productivity changes (product mix, product quality, input choices, labour composition, investment, etc.). Second, by using subgroup analysis, they can analyze the key firm-specific factors that are conducive for export success by exploring how treatment outcomes co-vary with skills, education and risk aversion. Third, the team can explore the impact of export participation on welfare measures such as income, investments in child education and income/consumption volatility.

Given the difficulties in generating sustainable export orders ourselves, the researchers chose Egypt as their project site since they found an NGO whose mission is to link small-scale enterprises in developing economies to Western markets. The first step was to carry out a baseline survey of 305 carpet producers and select a random sub-set of treatment and control firms from this sample. The NGO and a local carpet distributer have since then been marketing handmade carpets to retailers and importers in the United States and Europe: only the treatment group firms are shown a product sample and provided with the opportunity to fill these orders. Analysis of the impact of this intervention is ongoing. 


David Atkin

Massachusetts Institute of Technology (MIT)

Amit Khandelwal

Columbia University

Adam Osman

University of Illinois at Urbana Champaign