What Really Grows Exports?

Andy Bernard reframes the conversation on how to increase trade from developing nations.

In 2012, the World Bank released a document called the “Trade Competitiveness Diagnostic Toolkit,” which was written to help developing nations increase their exports and thereby grow their economies. The link between exports and economic growth is well established. Studies have shown, for example, that exporting firms are on average more productive, more capital intensive, and pay better wages than non-exporters. As the document notes, “the richer countries become, the more they tend to trade; and countries that are most open to trade grow richer more quickly.” It’s no wonder, then, that advising countries about export-boosting policies falls squarely within the World Bank’s mission to end extreme poverty by 2030.

For years, the World Bank and other similar organizations have looked to academic research to understand the microeconomic factors that increase exports. The stories told in that research suggest that new exporters tend to start small and, if they survive their first year of exporting, grow extremely rapidly. Those conclusions have in turn generated many papers that attempt to explain the start-small-grow-fast phenomenon, attributing it to the benefits of learning, experimentation, and understanding the reliability of partners.

But Andrew Bernard, the Jack Byrne Professor of International Economics at Tuck, has seen a problem with this narrative: it’s based on export data that has been aggregated and then annualized, rendering it easily expressible but also misleading. Bernard and co-authors Renzo Massari, Jose-Daniel Reyes, and Daria Taglioni lay out their corrective analysis in “Exporter Dynamics, Firm Size and Growth, and Partial Year Effects,” a new working paper for the National Bureau of Economic Research.

The problem they describe is surprisingly simple. Say two firms enter the export market in 2012, one in January and the other in November. Assume their monthly performance is identical. The annualized data will show 12 months of sales for the firm that entered in January, and just two months of sales for the firm that started in November. Therefore, the January firm will appear six times bigger for 2012. If the November firm lasts through the end of 2013, the annualized data will show a huge increase in its sales from 2012 (where there were only two months of transactions) to 2013 (where there were 12 months of transactions).

Both conclusions run counter to the reality of exporting. “The idea that exporters start especially small is not consistent with an existing body of work that says there are big costs to enter the market,” Bernard says. “And the rapid growth doesn’t align with the fact that if you start at the correct (larger) level, you will grow at a normal rate.”

Bernard and his co-authors show the perils of annualized data by adjusting for partial-year effects in Peruvian customs data from 1993-2009. They chose Peru because it is a relatively big, middle-income nation that has emphasized exports as a source of economic growth. In all, the data covers 2,352 firms who exported for at least four consecutive years.

The authors first report on export levels of new firms. The annualized data shows that 75 percent of new entrants are smaller than their within-sample average. When adjusted for partial year effects, however, the new exporters look much like other firms: just 46 percent are smaller than average. “Adjusting for the month of entry and allowing first-year exports to represent 12 months for each firm raises the size of entrants substantially,” they write.

The partial-year effect on firm growth rates is even larger. The calendar-year data paints quite an optimistic picture for firms who survive their first year of exporting: 146 percent growth from Year 1 to Year 2. For the next two years, growth evens out at about 30 percent per year. But adjusting for the starting month of exporting has profound repercussions. Firms no longer skyrocket in their first year, but grow at a reasonable pace of 25 percent, which then becomes 20 and 23 percent in years two and three respectively. “That’s a phenomenal difference,” Bernard says. “It’s almost unbelievably big.”

How has this discrepancy managed to evade academics? In part, because researchers often received data from governments in annualized form, and they didn’t think to question it. But it’s also due to human nature. “We are heavily conditioned to think in calendar years,” Bernard says. “We want to know who sold the most in 2012, not from July 2011 to June 2012. We just don’t respond naturally to that.”

For organizations like the World Bank, the implications of the partial-year effect are wide-ranging. New exporters are probably more important for economic growth than was previously thought. That means policy experts will need to reassess the determinants of exporter success, shifting the focus away from learning and experimentation to reducing the barriers that prevent small and medium firms from entering export markets—things like expensive export licenses and onerous incorporation rules.

“I argued for a long time that if you wanted to grow exports in the first two years, you needed to focus on the big firms,” Bernard says. “That’s still true, but this research suggests it’s a little less true. And that’s good news because it means the dynamism in the market is robust.”