Author
The benefits to consumers from freer trade—and the welfare loss from higher trade barriers—depend crucially on trade elasticity: For consumers, to what extent are domestic and imported goods substitutable for one another?
If this trade elasticity is high, consumers will turn more readily to domestic goods if imports become less available or affordable because of an increase in trade costs (such as tariffs). A lower trade elasticity, on the other hand, means consumers place a higher relative value on those imported goods or varieties. Trade rises or falls relatively less in response to changes in trade costs, and trade barriers impose more pain.
The intrinsic connection with consumer welfare makes trade elasticity an essential parameter to pin down to understand the costs and benefits of trade policies. Minneapolis Fed Monetary Advisor Michael Waugh and Ina Simonovska of UC Davis demonstrate that different canonical models of trade—when informed with the same, real-world data on trade flows and product prices—imply a wide range of trade elasticities (Staff Report 674, “Trade Models, Trade Elasticities, and the Gains from Trade”). Specifically, the economists find that models with richer microfoundations—that is, more components capturing economic complexities—imply welfare gains from trade that are approximately 50 percent larger than models that lack these features.
Simonovska and Waugh assess three prominent “fixed variety” models of trade: The classic Armington model (as formalized by Anderson and van Wincoop in 2003) assumes perfect competition, differentiated products, and “iceberg” trade costs that decrease with factors like proximity and contiguous borders. A model from Eaton and Kortum (2002) additionally allows for changes in traded goods at the extensive margin (that is, new goods are traded as trade costs come down). A subsequent model from Bernard et al. (2003) layers on an environment of imperfect competition, with more efficient firms able to charge a higher markup over their marginal cost.
They also apply their exercise to two models of a different class. “Endogenous variety” models by Paul Krugman (1980) and Marc Melitz (2003) allow different goods to arise within the model, in a context of monopolistic competition. The richer Melitz model, however, also features firms that differ in productivity, that form and dissolve, and that enter and prosper in export markets depending on their productivity.
To measure trade costs, Simonovska and Waugh focus on cross-country differences in price for identical manufactured goods. While not a perfect proxy for trade costs, these price gaps, they argue, convey valuable information about trade barriers and correlate with other variables that explain bilateral flows of trade. Larger price gaps for identical products would tend to indicate higher trade frictions; at the other extreme, with no frictions at all, real prices would converge in both places.
In the theoretical portion of the paper, the economists demonstrate that the distribution of such price gaps in models with richer microfoundations—given their additional margins of adjustment—should be categorically smaller than in simpler models. Price gaps, however, are an empirical fact, verifiable in historical trade data. If all models must generate the same price gaps and aggregate trade flows, amid the same trade barriers, richer models would necessarily imply a lower trade elasticity (and thus a higher intrinsic value of trade by consumers).
To estimate the trade elasticity parameters implied by each model, the economists require each model to conform to actual trade flow, price, tariff, and geographic data from the 30 largest countries (in terms of goods and services consumption). They focus on 2004, 2011, and 2017, and utilize up to 70 categories of identical goods from World Bank pricing data. Specifically, they ask each set of trade models to reckon with the maximum price gap (adjusted for each country’s price level) observed between each pair of countries in each period, building on a method they established in an earlier paper.
Physical trade frictions, such as distance, are known, as are policy frictions, such as tariffs. This leaves trade elasticity as the parameter that is allowed to vary as needed such that each model will generate the same aggregate trade outcomes. The elasticity estimates are notably consistent across the three sample years (see figure).
Among the three fixed-variety models, they estimate a trade elasticity for 2017 of 3.37 in the richest model (Bernard et al.); this is approximately 25 percent less than the middle model (Eaton and Kortum), and about 50 percent less than the simplest models (Krugman/Armington). The inverse correspondence between elasticity and welfare suggests that the welfare effects of trade are roughly 50 percent stronger in the model with the richest microfoundations. The difference is similarly large between the richer and simpler endogenous-variety models.
The economists perform an overidentified estimation to confirm the better fit of the richer models to the actual trade and price-gap data. They also compare their approach to a method developed by Caliendo and Parro (2014) that uses cross-sectional tariff data to estimate trade elasticities. They find the trade elasticity estimates from that method vary widely over time, which they attribute to the noisiness of tariff data. By contrast, the estimates generated by Simonovska and Waugh using their technique are largely unchanged across the three periods they study.
Read the Minneapolis Fed staff report: “Trade Models, Trade Elasticities, and the Gains from Trade”
Jeff Horwich is the senior economics writer for the Minneapolis Fed. He has been an economic journalist with public radio, commissioned examiner for the Consumer Financial Protection Bureau, and director of policy and communications for the Minneapolis Public Housing Authority. He received his master’s degree in applied economics from the University of Minnesota.
