This post was updated at 5:43 pm to reflect additional analysis.

Today, the Washington Post fact checker, Glenn Kessler, claimed that Public Citizen’s analysis of the Korean Free Trade Agreement (KORUS) is based on flawed economics and faulty math. Kessler accepts the White House claim that the employment effect of the KORUS should be based only “on a gain in merchandise exports,” and then claims that “the most appropriate way to look at export flows would be on an annual basis, which shows a net gain of about $2.3 billion. That’s theoretically a gain of 15,000—a far cry from the loss of 85,000 [jobs],” as estimated by Public Citizen. By ignoring imports, Kessler completely ignores one of the most important factors in the effects of trade on employment.

Imports reduce the demand for domestic goods and services. This is a fundamental assumption in introductory (and applied) macroeconomics. By ignoring it, Kessler denies his readers critical information needed to evaluate Public Citizen’s claim.

The Fact Checker approach (and the White House’s KORUS trade and job estimate) is a form of bookkeeping which counts only the credits and ignores the debits. It would earn a failing grade in any basic accounting class. Kessler spends a lot of time talking about things that make job and export calculations difficult (overall economic health, and the state of the business cycle), and yet he glosses over the impact of imports. It’s really important to calculate the jobs impact of both exports and imports, and it’s easy to do.

We begin by calculating the change in trade since the agreement took effect.  The KORUS agreement took effect in 2012, so most analysts, including the U.S. Trade Representative, use 2011 as the base year before KORUS.  Apparently, Mr. Kessler used 2012 as a base year, but the agreement had already been in effect for more than 9 months in 2011 (Because total U.S. exports to Korea were higher in 2011 than in 2012, using 2011 as a base yields a smaller increase in exports, and a smaller number of jobs supported by exports than claimed by Kessler.

Public Citizen’s analysis implicitly assumes that the same (export) jobs multiplier is applied to imports and exports. If we divide the increase in general imports between 2011 and 2014 ($12.9 billion) and the increase in total exports ($1.1 billion, not $2.3 billion as report by Kessler) by the jobs multiplier ($150,000 per job), we get 86,000 fewer jobs because of increased imports and 7,000 more jobs (not 15,000, as estimated by Kessler) from increased exports. You get exactly the same result (a net loss of 79,000 jobs) if you divide the change in the trade deficit ($11.8 billion) by the jobs multiplier. This just reflects the mathematical law of transitivity. Public Citizen makes a few other minor assumptions that give them a slightly bigger net job loss (85,000), but these estimates just show how unimportant these other assumptions are to the big picture, which is that growing trade deficits with Korea have eliminated more than 75,000 U.S. jobs in the first three years alone (this is exactly the number I came up with in my most recent blog on KORUS trade and jobs last month).

The deeper point is that the import jobs are usually characterized as being lower-skilled and lower paid, so they could have a higher multiplier than that for exports—since each job pays less, more jobs would be lost per dollar of imports. In that regard, Public Citizen’s jobs calculation is conservative—the use of separate import jobs multipliers would likely inflate net job losses.

Lastly, Kessler also argues that the export multiplier should not be used to evaluate the employment impacts of imports because some of those goods are not produced in the United States. However, 97.1 percent of U.S. goods imports from Korea are manufactured products—goods which we either do or could produce but for the offshoring of manufacturing.

For more than twenty years, both Democratic and Republican administrations have claimed that Free Trade agreements like KORUS and the North American Free Trade Agreement will lead to growing U.S. exports and will stimulate creation of goods jobs in the United States. If that were true, we would have experienced shrinking trade deficits and net job creation, but that has not been the case with most, if not all, of these agreements. Proponents of trade agreements have asserted that these deals will lead to trade-related job creation. In order to test that assertion fairly, it is essential to evaluate these agreements based on the employment effects of changes in both imports and exports.