APPENDIX A
DATA SOURCES AND METHODS
This appendix outlines our data sources and methods. The creators and maintainers of the GTAP database
work tirelessly to make their data as current as possible. However, our analysis requires both “present”
and “past” data to examine the impact of policy reforms (or transportation costs) over the last 25 years.
We summarize below the methods we used for creating our “past” and, where necessary, “present”
estimates.
Country Coverage
The CGE analysis we have used (like the simple partial equilibrium analysis) considers only 18 countries:
Australia, Brazil, Canada, China, the European Union, Hong Kong, India, Indonesia, Japan, Korea,
Malaysia, Mexico, the Philippines, Singapore, Taiwan, Thailand, the United States, and Venezuela.
Table A.1 shows US exports and imports with these 17 partners in 1980 and in 2006. Of total US two-
way trade, trade with these 17 partners made up 73 percent in 1980 and 84 percent in 2006. The list of
countries was chosen based primarily on their importance in US trade but also reflects data availability.
Sector Coverage
The GTAP database contains data for 57 merchandise and service sectors. Our analysis covers only the 42
merchandise sectors. Table A.2 shows US exports and imports in 2006 by GTAP merchandise sector.
Applied Rates
Most favored nation (MFN) applied tariff rates are the rates that any WTO member (and at times non-
WTO members) apply to all nonpreferential trade partners—namely, trade with countries that are not
linked by a free trade agreement or customs union. To calculate “present” MFN applied rates, we take
the simple average of MFN applied rates for each of the 18 countries for three years of available data,
selected from the period 2002 to 2005, with a preference for the most recent data. For “past” tariff rates
we consider three years of available data, selected from the period 1988 to 1993, with a preference for the
oldest data.1,2 The reason for using three-year averages is that applied rates bounce around for aggregate
categories even when there is no policy change, simply because the importance of trade in individual tariff
lines changes from year to year. When necessary, we include ad valorem equivalents of specific tariff rates.
1. The phrase “three years of available data” effectively means if a country has only two years of data in the listed time
frame, then only data from those two years is considered. If data from all the years in the time frame are available, the
most recent years are preferred to calculate present tariff rates, and oldest years are preferred to calculate past tariff rates.
2. In the case of Mexico, we felt the increase in MFN applied tariffs after the Peso Crisis of 1995 inaccurately skewed the
comparison of past to present tariffs. To compensate, we use Mexican 1998 MFN applied rates as the past rates and 2006
MFN applied rates as the present rates.
39