Amid all the talk of U.S. trade recently, The Economist just published a series on the importance of exports. A piece entitled “Export or Die” described how a New York-based architecture firm barely avoided massive layoffs by finding projects in China, Korea, and the Middle East, where demand has not faltered as sharply over the last two years. In other words, service exports prevented unemployment. One wonders: Is this just an anecdote, or is it representative of an important trend?

As it turns out, it is a trend. Service exporting has saved thousands of jobs during this recession.

The metros that have seen the most stability in their service export growth rates are Los Angeles, Baton Rouge, Austin, Provo, Utah, and Madison, Wisc. None of these places lost as much as 10 percentage points in service export growth over the pre- to post-recession period. Those that have seen the largest drop offs in service exporting are Modesto, Chattanooga, Charlotte, McAllen, Tex., Greensboro, and Riverside, Calif. All of these metros saw estimated declines in service export growth rates of 20 percentage points or more--with some going from rapid growth to rapid decline. Detroit and Las Vegas were not far behind at 19 percentage points. One interpretation of this is that service exporting is leading metros out of recession, and pointing to an important and growing source of competitive economic activity for much of the United States.

Stepping back for a moment, it is not hard to believe that service exports are especially competitive in the United States. For the last 60 years, the U.S. economy has been becoming more service oriented, and we have long had a positive trade balance for service exports. Meanwhile, increased urbanization has led to a concentration of services in densely populated and innovation-rich metros. We estimate that 77 percent of U.S. service exports are produced in the 100 largest metro areas (and 8 percent are non-metropolitan), but just 60 percent of goods come from those same 100 metros (with 16 percent being produced outside of metros).

With all that, the most obvious advantage of service exports during this recession is that they have performed better than goods exports. According to the BEA, from the second quarter of 2008 to the fourth quarter of 2009, aggregate U.S. service exports fell by 7 percent, but goods exports fell by 14 percent. Of exportable services, film and television production, education, insurance, telecommunications, and business, professional, and technical services have done the best, even expanding since the second quarter of 2008. By contrast, exports of travel (mostly generated by foreign tourists), industrial process services, and to a lesser extent, financial services have lost value over the same period. Yet, none of these losses have been nearly as severe as the worst performing goods sectors, including wheat, engine parts, corn, cars, semiconductors, and vehicle parts, in order of severity.

And so it is service export growth that has really helped with unemployment. I was able to look at this more systematically by using data we’ve been gathering on exports from metropolitan areas using a variety of sources (Moody’s Economy.com, the U.S. International Trade Commission, and Bureau of Economic Analysis). Controlling for gross metropolitan product (GMP, essentially a regional version of GDP) growth, service export growth has a strong negative effect on unemployment changes (meaning it limits joblessness), as does GMP growth. Surprisingly, however, total export growth and growth in goods have no effect.

But this way of doing the analysis is biased because there are many metropolitan level characteristics that are correlated with service exporting and higher employment rates, such as a better educated workforce and a more innovative set of industries. To cancel out the effects of metropolitan characteristics, I relate the change in unemployment to the difference in the growth rate of service exports (that is I subtract the pre-recession growth rate from the post-recession growth rate—starting in the second quarter of 2008).

The more stability in the growth rate of service exports, the less unemployment increased. The unemployment rate was 1 percentage point higher for every 5 percentage point loss in the service export growth rate pre and post-recession. This is basically the same result as before, but now we can feel confident that unchanging metro characteristics--like the share of college educated adults--are not driving the results. To confirm this, adding controls for college educated adults, the pre-recession service intensity, and the percentage of jobs in services does not explain away the connection (though the education effect is significant and does explain half of the service export effect). Moreover, the effect of changes in GMP growth before and after the recession does not matter, only changes in service export growth.

In so far as this analysis is valid, growth in service exports over the second half of 2009 should predict unemployment decreases over the first half of 2010. By this reckoning, Madison, San Antonio, Columbia, S.C., Des Moines, and Austin should see some of the largest decreases in unemployment by June of 2010, while those metros that have performed the worst on service export growth (such as Las Vegas, Detroit, Sacramento, Honolulu, and Salt Lake City) are likely to see little change to their unemployment rates. It appears that The Economist’s anecdotal evidence is largely correct; exportable services amount to nice work for the metros that can get it.