Antidumping policy moves incrementally in the right direction only on rare occasions. In that regard, last week was nothing short of historic. In addition to the U.S. International Trade Commission deviating from its conventional script and revoking 15 longstanding antidumping measures on key steel products (described here), the Office of the U.S. Trade Representative announced to Congress the administration’s decision to implement a critical change to the Commerce Department’s antidumping calculation methodology, which, if implemented in good faith, will likely reduce the incidence and disruptive impact of antidumping measures henceforth.


In response to a series of rulings from the dispute settlement body of the World Trade Organization, which found a U.S. methodological practice known as “zeroing” to violate Article 2.4.2 of the WTO’s Antidumping Agreement, Commerce decided (albeit, grudgingly) to change it’s policy. I have described zeroing and its impact in a few previous papers and in this blog post, but here’s a brief summary.


In a typical antidumping investigation, the sales and cost data of each foreign company under investigation are subject to a series of calculations before the bottom line “dumping margin” is produced. Usually, the Commerce Department calculates average net prices for each product (i.e., widget model 1, widget model 2, etc.) sold in the U.S. and home markets. The average U.S. and average home market prices of widget model 1 are compared, the average prices of widget model 2 are compared, and so on. In some cases there may be few comparisons, and in others there may be hundred or even thousands of comparisons. Some of those comparisons may generate positive dumping margins (when average home market price exceeds average U.S. price) and some may generate negative dumping margins (when U.S. price is higher).


Commerce then calculates from all of these model-specific comparisons an overall weighted-average dumping margin. But before calculating the overall average, Commerce tinkers with the mathematics by zeroing. Zeroing refers to the practice of assigning a value of zero to all of the comparisons that generate a negative dumping margin. Only after zeroing does Commerce calculate the average dumping margin. So, in other words, zeroing precludes the negatively dumped sales from having the proper impact on the “average” dumping margin. Thus, if 99 of 100 comparisons generate large negative dumping margins and 1 of 100 produces a positive dumping margin, zeroing ensures that the average dumping margin calculated is positive. Pretty fair, huh?


In research that Brink Lindsey and I conducted a few years ago, we found that zeroing is highly distorting. In a sample of 18 actual antidumping determinations, we found that calculated dumping margins would have been on average 86% lower had zeroing not been employed. Five of those 18 cases would have resulted in the cases being dropped, and antidumping measures never having been imposed. So the change in policy is laudable and potentially very significant.


I say “potentially” because zeroing reform remains incomplete. The policy change announced last week pertains to zeroing in what are called average-to-average comparisons. In some cases, the Commerce Department compares average prices to transaction-specific prices and in others it compares transaction-specific to transaction-specific prices. It is possible that Commerce will use these methodologies more frequently now and continue to zero (at least until zeroing under these comparison methodologies is found in violation of our WTO commitments as well).


And there is one other possible obstacle on the road to implementing this change: Congress. Although zeroing is not mandated by law, the practice has been in use for a very long time. Cases have been heard in the Court of International Trade and the Court of Appeals for the Federal Circuit concerning the question of whether zeroing is even permitted under the statute. Both courts have ruled that zeroing is a permissible interpretation of the statue, which has been taken by some in Congress to mean, wrongly, that zeroing is a requirement of the statute.


Congress, which is bipartisan in its broad support of a strong (i.e., menacing and unfair) antidumping law, may seek a fight with the administration over the propriety of changing the zeroing practice without input from the legislative branch. But, by and large, last week’s zeroing announcement was another rare victory for antidumping reform.