Presentation on theme: "Measuring Trade Restrictiveness Stephen Tokarick The views expressed herein are those of the author and should not be attributed to the IMF, its Executive."— Presentation transcript:
Measuring Trade Restrictiveness Stephen Tokarick The views expressed herein are those of the author and should not be attributed to the IMF, its Executive Board, or its management." International Center For Trade and Sustainable Development October 28, 2009
Disclaimer The views expressed in this presentation are the views of the presenter and do not necessarily represent the views of Executive Directors, the Executive Board, or Management.
Introduction Great deal of interest in measuring whether a country has become more open to international trade for at least two reasons: Researchers need a way to measure whether a country has made progress on trade reform. Some IMF programs have included trade reforms, although trade conditionality has diminished in recent years. (Note: IMF does NOT attach conditionality to an index). Trade reforms are occasionally elements of World Bank programs.
Introduction Second reason is that there has been a great deal of interest in whether the question: Is there a relationship between openness to trade and economic growth? So need a measure of openness. Large number of empirical papers that have tried to uncover a relationship between openness and growth. Examples: Sachs and Warner (1995), Dollar and Kraay (2001), Warcziarg and Welch (2003), and DeJong and Ripoll (2006).
Introduction Each of these papers adopts a somewhat different variable to measure openness to trade in some way: (i) average tariff rates; (ii) import duties/Imports among others.
Introduction Do three things in this presentation: 1. Provide an overview of how the IMF has measured trade restrictiveness in the past 2. What I have done in the way of research on alternative measures of restrictiveness. 3. Issues of relevance for the Composite Index of Market Access (CIMA).
How has the IMF Measured Restrictiveness? In the past, some IMF country reports used to include an overall “trade restrictiveness index” until about The overall index was comprised of two parts: (i) a tariff index; and (ii) a non-tariff index. The tariff index was essentially a country’s unweighted average MFN tariff rate. Based on the magnitude of this rate, the country was assigned a tariff rating between 1 and 5:
Tariff Index RatingSimple Average Tariff plus Surcharges 1Between 0 and 10 percent 2Between 10 and 15 percent 3Between 15 and 20 percent 4Between 20 and 25 percent 5Greater than 25 percent
Non-Tariff Barrier Index RatingPrevalence of NTBs 1Less than 1 percent of production or trade are subject to NTBs 2Between 1 and 25 percent of production or trade are subject to NTBs 3More than 25 percent of production or trade is affected.
Overall Index A country’s overall trade restrictiveness index is computed according to the following formula: Z = 1 + (T-1) + 3*(N-1) Where Z is the overall index, T is the tariff index, and N is the NTB index. Note that NTBs are weighted 3 times more heavily than tariffs.
IMF’s Trade Restrictiveness Index Further details can be found in a paper on the IMF’s website: Decision was taken in 2005 to stop reporting the IMF’s TRI in country reports because of weaknesses with the index and information on which it was based, especially with respect to non-tariff barriers.
Research on Some Alternatives There has not been a decision to replace the IMF’s old TRI with a new one. Have undertaken a project that compares the performance of alternative indices of trade restrictiveness to answer some research questions mentioned earlier. There are problems with commonly used measures of restrictiveness.
Introduction There are problems with some of the commonly used indicators of openness: 1. Simple average tariffs: attaches same weight to each tariff rate. Some may be more important than others. 2. Weighted average tariffs: weight tariffs by shares in imports. Problem: if a tariff is very restrictive, then that good would get a low weight when you want it to get a high weight.
Introduction 3. Tariff Revenue/Imports: This is an endogenous variable—other factors besides tariffs affect imports. 4. NTB Coverage Ratios: Main issue: need to know whether an NTB is binding and gives rise to a price gap. Using coverage ratio does not reveal this. A NTB may not be binding!
New Measure Anderson and Neary have introduced a trade restrictiveness index that has a solid basis in economic theory and can be implemented empirically. They define the trade restrictiveness index (TRI) as the uniform scaling factor by which prices must be deflated to leave welfare unchanged. Alternatively, the uniform tariff equivalent (UTE) as the uniform tariff rate that has the same effect on welfare as a country’s differentiated structure.
TRI TRI adopts the notion that trade restrictions, i.e. tariffs, should be aggregated in a way that captures how tariff changes affect welfare. That is, the tariffs weighted according to their marginal contribution to welfare. Note that there are other ways to measure restrictiveness, such as contribution to trade volumes (Mercantilist trade restrictiveness index, the MTRI).
Project Goals What we do in this project: Given the interest in measures of openness, we analyze the performance of various openness indicators over a long time horizon: early 1970s to the Specifically, we construct four (4) indicators of openness: (i) simple average tariff; (ii) import-weighted tariff; (iii) tariff revenue/Imports; and (iv) a partial equilibrium version of the Anderson/Neary UTE.
Project Goals We calculate these indicators for 163 countries (plus EU) from 1973 to 2008 (with missing data). We find: 1. Tariff revenue/Imports is negatively correlated with the AN-UTE for about 39 percent of the countries studied. This is significant because Rodriquez and Rodrik (2000) argue this it the “preferred indicator”:
Project results Our results tend to reject the claim of Rodriquez and Rodrik. 2. There is in fact a high degree of correlation between the UTE and average tariffs and weighted-average tariffs. For about 90% percent of the countries, the correlation between the UTE and the simple average and import- weighted tariff is 0.5 or greater, 3. However, there are cases for which the correlations are negative. UTE is negatively correlated with the simple average tariff for about 10% of countries.
Definition of theTRI
TRI Big thing to take away is that the TRI: Defines “trade restrictiveness” in relation to the impact of a trade barrier on real income. Uses marginal welfare weights to aggregate trade barriers, rather than trade weights (level of imports).
TRI One point emphasized by AN-UTE is that the height of a tariff does NOT necessarily tell us how restrictive the tariff is, in terms of welfare or in terms of import volume. Effects depend on elasticities, as well as magnitudes of distortions. High tariffs may confer no welfare loss if applied to goods that have low, or zero, elasticities. Tariff rates are indeed helpful, but need more information to assess openness or restrictiveness.
Price Quantity P P* P Quantity A A B B GOOD 1 GOOD 2 Q1 Q2 50% tariff 5% tariff
Diagram Initially, tariff on good 1 is 50%, while the tariff on good 2 is zero. Therefore, the simple average tariff is 25%. Eliminate the tariff on good 1 and raise the tariff on good 2 to 5%. Simple average tariff declined to 2.5%--a large reduction. However, in welfare terms, real income went down. As well, imports fell.
Method Use the “Feenstra” Shortcut, which is a partial equlibrium measure:
Implementing TRI To implement the TRI, three things are needed: 1. Shares of imports in GDP (sj) 2. Tariff rates by product (tj) 3. Own-price elasticities of demand for imports.
Data Sources 1. Trade Data: Main source is COMTRADE. Level of disaggregation is an issue that had to be confronted. More below. 2. Import Demand Elasticities: These were taken from a paper by Kee et al. (2008) in REStat. Estimated for about 88 countries and 4,800 products per country.
Data Sources 3. Tariff Data: Since we wanted a long time series, had to assemble database on tariff rates. The level of disaggregation needed would need, in theory, to be consistent with the elasticities, since we need to apply them.
Summary of Results Correlations: Percent of countries that fall into each category -1.0 to to 0 0 to to 1.0 TRI-simple TRI-weighted TRI-R/M Simple-weighted Simple-R/M Weighted-R/M
Summary of Results UTE and simple average tariff are negatively correlated in about 10% of the cases. UTE and Revenue/Imports are negatively correlated in about 38% of the cases. Comment of Rodriquez and Rodrik.
Future Work 1. Add more countries and periods—in progress. 2. Concording data from various sources and nomenclatures has been a huge task and will try to do more. 3. Specific Tariffs: a real challenge since these are used by the United States in agriculture. Typical method is to divide the specific tariff by the import unit value. 4. Non-tariff barriers could be taken into account in the Anderson/Neary formula, but would need tariff equivalents.
Final Thought Having said a lot about indicators, it seems what policymakers are ultimately interested in is the impact of different policy measures. For this, focus on models rather than indicators. There are many: 1. ATPSM: Vanzetti or other PE models. 2. GE models: GTAP 3. Anderson: update of Krueger, Schiff, and Valdes. Compares domestic and world prices from 1955 to the present.
Issues For CIMA 1. What question does the CIMA answer? not the impact of policies for that requires a model. 2. How does it address the issue of a change in the composition of protection? Example of 2 goods with different elasticities.
Issues for CIMA 3. Meaning of “measuring market access” needs to be clarified. Seems to suggest that this means that if tariffs in import markets were reduced, how would exports to that market change. But this requires a model or at least, elasticities.
Issues for CIMA 4. “Any reduction in market access barriers, whether by tariff reductions or by less costly ways of meeting standards, should therefore show up as an increase in the market access index.” As shown in the above example, looking simply at the average tariff would imply that market access has increased. However, TRI (and trade restrictiveness) rose when average tariff declined.
Issues for CIMA 5. Seems to include excise taxes. Agreement on subsidies in WTO allows exporters to receive a rebate on taxes paid on imported intermediate inputs (tariffs and excise taxes). I.e. “duty drawback schemes”