Testimony of Ms. Elliot--free trade, unilateral and economic trade sanctions

Testimony of Kimberly Elliot
Research Fellow
Institute for International Economics

May 14, 1998

MS. ELLIOTT: Thank you very much. I am delighted to be here. What I was going to focus on is the empirical research that the Institute for International Economics has been doing on economic sanctions for the last 15 years. I will sort of just briefly summarize the major points of that research and then try to discuss in somewhat more detail the methodology that we used in a working paper last year analyzing the effects on U.S. exports of sanctions.
The second edition of our research and economic sanctions was released in 1990, and at that time we had 115 cases of economic sanctions going back to World War I. About twothirds of those cases involved the United Sates as a leading sender. We were not looking at only the U.S. in the broad sanction study, but I will focus here mainly on the results for the United States.
Going back to the sample as a whole, of the 115 cases, we concluded that in about a third the sanctions had achieved some degree of foreign policy success. We also found, however, that success rates had declined sharply over time, and here is where I will look especially at U.S. unilateral sanctions since that has been a topic of much of the discussion this morning.
We identified 55 cases of unilateral U.S. economic sanctions between 1945 and 1990. Breaking that period roughly in half, we found that U.S. unilateral sanctions succeeded almost 70 percent of the time from 1945 to 1970. In the 1970s and 1980s, however, that success rate dropped quite sharply to only 13 percent in that period. If you look at the results for all U.S. sanctions, including those in which it was part of a multilateral coalition, there the success rates drops from 53 percent in the earlier period to 21 percent more recently.
For our third edition, which we are currently working on and hope to have available early next year, we have added another 41 cases overall of which 33 involve the United States. The 33 cases, and that is not in this decade that is 1990 to 1996 is an increase from 20 cases, new cases initiated during the 1980s, and as of the end of 1997 by our count the U.S. was involved in more than 30 ongoing unilateral and multilateral sanctions.
Contrary to the conventional wisdom, however, we have not found a large increase in the number of unilateral U.S. sanctions in the 1990s. We count 23 unilateral sanctions in the 1970s, 11 in the 1980s and another 11 in the 1990s. However, these have accumulated over time because it is often difficult to list sanctions so that there were more than a dozen ongoing as of last year.
In addition, as has been noted earlier, there are a number of sanctions at the state and local level which are not included in the numbers that I have just given you. Those are all national level sanctions.
Finally, our preliminary and tentative results for the 1990s suggest that effectiveness has not improved with practice. The success rate in the 1990s looks to be similar to the very low of the 1970s and 1980s. While the benefits of economic sanctions, especially in recent years, especially when they are unilateral by the U.S., appear to be elusive, the costs are more tangible, and that is what we tried to look at in some more recent work.
In this working paper that I mentioned that has also been cited in the earlier panels, we calculated that economic sanctions against 26 target countries in effect in 1995 reduced U.S. exports by between $15 and $19 billion. That would translate into a loss of 200,000 export jobs and about $1 billion in export job premia, given that export jobs by a number of calculations pay more than the average manufacturing job. I would note also that that is an annual cost, and that cost continues each year that those sanctions or similar sanctions remain in place.
Now I want to turn to the methodology that we used to calculate these costs. What we did was to use a gravity model, which is a common analytical tool for analyzing bilateral trade flows. We then applied an even more common statistical technique, an ordinary Lee squares regression analysis, in order to isolate the effects of sanctions on bilateral trade flows while holding other factors constant.
The gravity model, just as the name suggests, derives from Isaac Newton's theory of gravitational force which states that the gravitational force between two celestial bodies is positively related to the product of their masses and inversely related to their distance apart. In its simplest form, the gravity model as applied to trade predicts that the amount of trade between two trading partners will be positively related to the product of their outputs, which is used as a measure of their mass, and negatively related to the distance between them.
The particular form of the model that we used was adopted from Jeffrey Frankel, who has used it extensively to study regional trade. That version of the model predicts that trade will increase as the combined size and per capita incomes increase, decrease as the distance between the two increases, and then increase if the two countries share a common border, a common language or are both members of the same trade block, such as NAFTA or the European Union. Now to that basic model we then added a series of dummy variables to indicate the presence of trade sanctions in the current or previous year. Because trade sanctions do take a wide variety of forms, we differentiated between limited, moderate and extreme or extensive comprehensive sanctions basically.
In addition to those variables, we also tried to get to this issue of lingering effects. On supplier effects we are having a series of logged variables, which indicated the presence of sanctions that had been lifted one to two years previously or three to four years previously. Those periods were chosen because we were dealing with data that was every five years. We had 1985, 1990 and 1995. The data set included 88 countries, which worked out to over 3,5000 country pairs.
I will start first just briefly summarizing the general results, which is based on bilateral trade among all 88 countries, but then note that the numbers that I cited before, $15 to $19 billion, are based on a subsample of the data set which looks just at U.S. exports to the other 88 countries and the sample. In general, the basic gravity model variables, the size, income, distance, adjacency and language, all did have the expected sign and were all highly significant statistically, and also expected, sanctions with broad coverage, our extensive sanctions, had very large effects on bilateral trade flows consistently reducing them on all of the tests by around 90 percent.
There was more variance among the different tests that we tried on the estimated impact of moderate and limited sanctions, but those sort of on average suggested a reduction in bilateral trade of roughly a quarter to a third from limited to more modest sanctions.
Now, the problem in applying the method to U.S. exports alone is that you very significantly reduce the number of observations that are available from more than 3,000 to just 84. As a consequence, the results for the United States are less robust than these other results that I have just noted. Nevertheless, several of the coefficients on the variables are statistically significant and so we do think that the $15 to $19 billion is sort of a reasonable order of magnitude estimate for the impact of sanctions on U.S. exports.
The analysis of the lagged sanctions variables as we did it produced little evidence, but the negative impact of sanctions lingers over time. However, I have to say that there were few observations. In some of the years it was very spotty in terms of observations available to test and also the fact that the data that we used was very aggregated. That may also have affected those results.
What I will discuss very briefly now are the advantages and disadvantages of this particular approach. The major benefit of the gravity model and the OLS technique is that it allows us to analyze the indirect, as well as the direct, effects across a large number of countries and while holding other factors constant.
It is helpful for estimation sort of, as I said, a broad order of magnitude of the impact of sanctions on exports, but there are some disadvantages. There are some weaknesses of the approach as well.
The first is on this issue of unreliable supply or effects. Because this is a top down analysis and using aggregated data, it is probably best for drawing general economy wide conclusions, but it may well underestimate the burden on the sectors hardest hit by economic sanctions.
The variables we devised to look at those longterm and reliable supplier effects were not statistically significant, but you would expect that the effects would be relatively more severe for particular sectors of the economy rather than for exports as a whole. We were not able to test for that, given the aggregate data.
A second limitation is I think it was Frank Kittredge asked me about services. We did not in fact have services during the model, nor did we have investment. The question is whether or not this model, which has been very good at predicting trade flows, would function as well for predicting service. We did not try that, so it is possible it could be applied in that way, but we did not try to do that.
Thank you.

Return to Top