Standard Measure Of Barriers To Trade Are Seriously Flawed

World Bank and IMF analysts try to measure the height of a country''s trade barriers in order to determine how far they must be reduced before fresh loans can be agreed. But according to Professor James Anderson, writing in the latest issue of the Economic Journal, the standard measures used are highly unsound and quite inappropriate for assessing a country''s worthiness for assistance. In their place, he proposes a new Trade Restrictiveness Index, which can be used both for World Bank loan conditionality evaluation, and to examine the link between openness and economic growth.

How high are trade barriers?, Anderson asks. This seemingly simple question is hard to answer when there are many barriers: some sort of average of the individual tariff rates is needed, but which one? The answer matters for important questions such as the link between policy openness and economic growth. In the belief that there is such a link, the ''Washington consensus'' forces client countries to liberalize trade in return for World Bank/IMF assistance. Continued loans are made contingent on a measure of reduction in the height of trade barriers.

For all these purposes, lacking a theoretically based measure, analysts have employed a variety of indices such as trade weighted average tariffs, arithmetic average tariffs and even dispersion measures of tariffs.

Anderson''s study deploys a theoretically sound measure of the height of trade barriers, the Trade Restrictiveness Index (TRI). The theoretical idea is to find the uniform tariff which lowers welfare by just as much as the initial tariff structure – the ''uniform tariff welfare equivalent''. In other words, take away all tariffs and replace them with a uniform tariff rate on all previously tariffed goods such that welfare remains constant. The index is formed with ''welfare weights'' rather than trade weights or arithmetic weights. Theory shows that this index can differ substantially from the usual measures.

The results of the study demonstrate that benchmark measures of trade restrictiveness using the Trade Restrictiveness Index stand in sharp contrast to standard measures. For a 27 country sample of late 1980s trade barriers, trade weighted average tariffs underestimate restrictiveness measured by the ''uniform tariff welfare equivalent'' by an average of 50%. For a 7 case sample of year on year changes in trade policy, the TRI and changes in average tariffs are uncorrelated.

In more inflammatory words, a World Bank analyst using average tariffs to evaluate, say, Indonesia''s performance pursuant to its next loan might just as well flip a coin. These conclusions appear to be robust with respect to missing data problems and to imprecision in the elasticities of substitution between traded goods and domestic goods. The results argue strongly for replacing atheoretic measures of trade barriers with the TRI in World
Bank loan conditionality and in examining links between openness and growth. The study also demonstrates that the TRI is operational with data that are usually available, and with computational methods requiring only a decent modern computer. The way forward is clear: develop detailed data based on trade distortions; process them with the methods of this study; and use the TRI routinely in World Bank loan conditionality evaluation. With enough work of this kind, analysts can more appropriately examine the link between openness and economic growth.

''Trade Restrictiveness Benchmarks'' by James E. Anderson is published in the July 1998 issue of the Economic Journal. Anderson is Professor of Economics at Boston College.