Those EPI job loss estimates

Both Clinton and Obama have based dramatic claims about trade-related job loss on Economic Policy Institute (EPI) analysis (Where are these numbers coming from?).

The EPI uses Bureau of Labor Statistics input-output tables to calculate the number of jobs required to produce a million dollars of output for each of about 200 industries.  They then calculate the value of net imports in each of those industries and multiply that by the number of jobs per million dollars in output for that industry.  This, they say, provides an estimate of the number of jobs gained or lost by trade policy in each industry.  Summing across the industries gives them an aggregate figure for "jobs that would otherwise have been created," but that "were lost" because the trade deficit with a country widened since some policy action (NAFTA, trade with China, China's accession to the WTO) was taken.

In the paper quoted by Clinton (Costly Trade With China) an EPI analyst estimates that trade with China has cost 2.2 million U.S. jobs since 1997.  Clinton quoted an EPI figure of 1.8 million jobs since China joined the WTO in 2001.  Obama drew on EPI's NAFTA stats (Revisiting NAFTA) and asserted that NAFTA has cost a million U.S. jobs.

For the whole economy, however, the proportion of the population employed has actually grown as the trade deficit widened. 

In the figure below, the ratio of net exports to GDP is on the vertical axis, and the ratio of civilian employment to the civilian population is on the horizontal (since I've used net exports rather than net imports, a trade surplus is associated with positive ratios and a trade deficit with negative ratios):

 

The proportion of the civilian population in the labor force increased as the the ratio of net exports to GDP moved from positive to negative (as the trade deficit emerged and grew).  I don't think this increase was trade related, but it certainly isn't what I'd expect on the basis of the EPI's assumption that deficits cause job losses.

Philip Levy of the American Enterprise Institute (AEI) recently pointed to some problems with the EPI approach (Doing a Job on NAFTA).  First, as suggested by the figure, the number of jobs in the U.S. is not really determined by our trade laws:

The number of jobs in an economy is set by the size of the work force, the health of the labor markets, and macroeconomic fluctuations. Trade can certainly create new jobs with export opportunities or cheaper inputs. It can also destroy jobs when firms succumb to import competition. Lots of job creation and destruction occurs every year in the U.S. economy. In an average year, 17 million jobs are created and 15 million are destroyed, with a net job creation of 2 million. When net job creation matches growth in the labor force, the unemployment rate stays constant.

So why couldn’t trade push the unemployment rate up or down? Among other reasons, the Federal Reserve is watching and would offset any trade-driven economic swings. In the decade before 1994, unemployment averaged 6.6 percent in the United States. In the decade after, it averaged 5.1 percent, which is near the level at which central bankers begin to worry about inflation.

What of Obama’s claim that NAFTA cost the United States 1 million jobs? Imagine this were right. Then, without NAFTA we would have had 1 million more jobs. In the year 2000, this would have made the unemployment rate just under 3.3 percent, rather than the 4 percent we actually enjoyed. But Federal Reserve governors would have been in a panic long before we got down to that level and would have raised interest rates to slow the economy. They would have known they had gone far enough when unemployment increased to a level they were comfortable with – the same as with NAFTA.

Second, the procedure ignores all the factors that have contributed to our trade deficits, and assumes that they are due to the relevant trade agreements:

This is, in fact, the basis for the largest estimates of NAFTA-related job losses. The U.S. trade deficit ballooned after NAFTA. That change occurred broadly, not just with NAFTA countries, and it was accompanied by a drop in the unemployment rate and significant job creation. But if one ignores these broader lessons, ignores the pre-existing free trade agreement with Canada, and ignores everything else that happened, one can come up with the number of 1 million jobs lost....

NAFTA did some wonderful things in bolstering Mexico politically and economically. Most responsible estimates show that it had a small but positive effect on the U.S. economy. The minimal size of the impact was to be expected, since Mexico was economically small relative to the U.S. economy and the United States had very low tariffs even before NAFTA.

But what about the factory workers in Ohio? Are they just imagining those lost jobs? Of course not. Manufacturing employment in the United States did hit a peak and then begin a steady decline. The problem is that the peak was in 1979, 15 years before NAFTA came into force. The long-term decline of American manufacturing jobs has much more to do with technological change than with trade. We’re producing more stuff with fewer workers.

The EPI is concerned about a number of other issues - transitional job losses associated with changing trade patterns, median income levels, and income distribution.  These are important issues.  But their job loss calculations are very misleading.

h/t Jonathan Dingel at Trade Diversion: NAFTA Numbers.

Comments

[rickc]

I beg to differ. Check out my post on associatedcontent to see how job growth since 2000 compares to previous decades:

[rickc]

Sorry here is the link:

http://www.associatedcontent.com/article/1300166/modern_society_threatens_the_american.html?cat=9