The U.S. International Trade Commission (ITC) is required by the Bipartisan Congressional Trade Priorities and Accountability Act of 2015 to prepare estimates of the economic effects of trade agreements. In specific:
“Not later than 105 calendar days after the President enters into a trade agreement under section 103(B), the Commission shall submit to the President and Congress a report assessing the likely impact of the agreement on the United States economy as a whole and on specific industry sectors, including the impact the agreement will have on the gross domestic product, exports and imports, aggregate employment and employment opportunities, the production, employment, and competitive position of industries likely to be significantly affected by the agreement, and the interests of United States consumers.”
This statutory language guided the ITC’s analysis of the twelve-nation Trans-Pacific Partnership (TPP). The ITC study was released on May 18, 2016.
It had been several years since the United States concluded a free trade agreement. The previous one with South Korea (Korea-U.S. Free Trade Agreement, or KORUS) dates from 2007. I served as chairman of the ITC at the time and am quite familiar with the KORUS study. The econometric modeling used a “comparative static” analysis. A comparative static approach can be likened to taking two snapshots of the economy. The first photo was of the known baseline economy as it existed in 2007. The second photo also used the 2007 baseline, but this time it was “shocked” by incorporating all provisions of KORUS as if they had been fully implemented. This allowed a conceptually sound – albeit counterfactual – assessment of the likely economic effects of KORUS by analyzing how those reforms would have influenced the 2007 economy. (These issues are explained in this Free Trade Bulletin.) Static modeling has been used in all the ITC’s analyses of trade agreements prior to TPP.
One of the great strengths of the comparative static approach is that it makes no attempt to project the economy into the future. There is no need to speculate on whether a recession will curb trade flows, or whether technological change will make some industries obsolete while spurring new ones into existence. Precisely predicting the future requires a degree of clairvoyance not possessed by economists or anyone else. A comparative static analysis deals with that reality by instead looking backward. It imposes new policy reforms on an old – but well-known – economy. And it allows economists to avoid trying to make forward-looking projections of economic activity that inevitably turn out not to be correct.
However, comparative static modeling is not the only tool in the econometrician’s toolbox. For its analysis of the economic effects of TPP, the ITC has chosen to use a dynamic computable general equilibrium (CGE) model. The Global Trade Analysis Project (GTAP) model “is an appropriate tool for analyzing the effects of trade agreements because it consists of a database with international trade flows and other macroeconomic information, social accounting matrixes that show how different segments of the economy are interlinked, and national income accounts data.” Using a dynamic version of the GTAP model has allowed the ITC to estimate changes in various economic measures (real GDP, employment, exports, imports, etc.) up to 30 years in the future. Most of the analysis focuses on the 15-year period beginning in 2017 and ending in 2032.
In order to evaluate how TPP might influence the economy in the future, it first was necessary to create a baseline projection of what the economy would be like in the years ahead without TPP. The ITC has done this by incorporating projections made by the International Monetary Fund (IMF) and the Organisation for Economic Co-operation and Development (OECD) regarding growth rates in many countries for labor, population, and GDP. Once the 30-year baseline was established, the model was shocked by adding TPP’s annual policy changes. (TPP gradually phases in many reductions in trade restrictions year by year.) The economic effects of TPP then were measured as differences between the original baseline and the baseline following the shocks from TPP. The dynamic GTAP model provides a mathematically sound means to estimate future economic variances caused by policy changes.
The real questions regarding forward-looking estimates have to do with the baseline itself. The IMF and OECD are quite capable when it comes to analyzing historic trends. Generally it’s not unreasonable to project well-established trends a short distance into the future. If global GDP has grown at an average rate of 3 percent over the past ten years, for instance, it may be quite sensible to estimate that growth in the coming year also will be around 3 percent. The problems come as we look further into the future. Life’s inherent uncertainties make it relatively likely that a future projection of U.S. exports or imports of cheese, for instance, will turn out not to be precisely accurate. How much confidence should we have in projections five years into the future? Fifteen years? Thirty?
Making estimates that turn out to be different than actual future outcomes is not a problem to economists and statisticians schooled in economic modeling. They understand well that the ITC’s estimates were made using the best available information and up-to-date econometric techniques, and that the real world economy simply diverged from what had been projected in the baseline. Unfortunately, not everyone interested in trade policy has such depth of knowledge and understanding. My concern is not with the integrity of the modeling, but rather the challenges that trade supporters may face in defending the results of the analysis against criticism.
Even with comparative static modeling, opponents of expanded international trade have been inclined to misinterpret the analysis. There are claims, for example, that the ITC did a poor job with its KORUS study because the U.S. trade deficit with South Korea has gone up since the agreement went into effect. The KORUS study didn’t say anything about what might happen to the trade deficit in the future. However, it did indicate a likely decrease in the deficit in the hypothetical situation in which all provisions of KORUS were somehow implemented during the static 2007 baseline period.
Now that the ITC’s TPP study has used a dynamic CGE approach that actually does make estimates about future trade flows, critics of trade agreements no doubt will be happy to point out how the ITC “got things all wrong.” (In fact, the ITC’s estimates are seldom likely to be “right.”) Trade skeptics are unlikely to bother explaining that the real source of the estimated “errors” is that the underlying economy evolved differently than the IMF and OECD had projected. Most anti-trade NGOs have little interest in raising the quality of the trade policy debate. Rather, they may be inclined to argue that all economic analysis showing positive effects for the United States from trade agreements is suspect and can’t be trusted.
Supporters of trade liberalization will do their best to counter such misinformation by explaining the details of dynamic CGE modeling. But the criticism of the ITC’s estimates will take only a few words; setting the record straight will require several sentences or paragraphs. Protectionist rhetoric may prove to have a greater influence on public opinion than do the substantive explanations.
It will be interesting to see whether analyzing trade agreements via dynamic CGE modeling leads to a more informed public discussion than has been the case for the comparative static technique. With a comparative static approach, the ITC was never wrong, but often misunderstood. With dynamic modeling, the ITC will almost never be right, while still being misunderstood.
Daniel R. Pearson is a senior fellow in the Cato Institute’s Herbert A. Stiefel Center for Trade Policy Studies, and is a former chairman of the U.S. International Trade Commission.