schedules require extensive cleaning because the tariffs are not uniformly coded. The United States, for
example, lists tariff codes predominantly at the HS 8-digit level; however, in some instances an 8-digit
code contains several underlying tariff lines denoted by letters, usually with different tariff rates. We
assign the same 8-digit code to all of these underlying codes. Once we have the bound tariff schedules
appropriately cleaned at the tariff line level (usually at the HS 8-digit level but occasionally at the 10- or
12-digit level) we collapse the tariffs into the HS 6-digit level and take the simple average of all tariff lines
under each code.
Countries also have specific bound tariffs. Rather than go through the laborious exercise of
calculating our own ad valorem equivalents of bound specific rates, we splice into the HS 6-digit bound
tariff schedule the ad valorem equivalent of MFN applied tariffs from a corresponding time period (circa
1990 for pre-Uruguay Round and circa 2003 for post-Uruguay Round) whenever there was a specific
bound tariff at the tariff line level below any HS 6-digit code. The ad valorem equivalents of the MFN
applied rates are derived in the same manner as the MFN applied rates discussed above and displayed in
tables 3 through 20.
After splicing in the ad valorem equivalents of specific applied rates we use a concordance between
HS 1988/1992 and GTAP (HS 1996 and GTAP for China and Taiwan) to collapse the tariff schedules
into the 42 GTAP merchandise sectors, using the simple average of HS 6-digit codes.
In some instances the applied rates displayed in tables A.3 through A.20 actually exceed the bound
rates for corresponding countries, sectors, and periods (circa 2004 for Uruguay Round and circa 1990 for
Tokyo Round) displayed in tables A.21 through A.38. This is a coding and averaging problem rather than
malfeasance by WTO members.6 To control for this type of misreporting, we substitute the applied rate
for the bound rate whenever the applied rate is higher. There is another similar problem: Occasionally,
the post-Uruguay Round bound rates are higher than pre-Uruguay Round bound rates. This is matter
of coding revisions and the tariffication process under the Uruguay Round. To control for this type of
misreporting, we replace the pre-Uruguay bound rate with the post-Uruguay bound rate whenever the
post-Uruguay bound rate is higher.
Preferential Tariffs
The method used for calculating preferential tariff rates parallels the applied tariff rate method. Using
the WITS software and UNCTAD TRAINS database we query effectively applied tariffs, which gives
preferential tariffs when they are in effect and MFN applied tariffs otherwise. Again due to coding
updates and tariffication under the Uruguay Round, there are some instances where even the preferential
6. For China and Taiwan, applied rates may be higher than bound rates because some bound rates were implemented after
our sample period for applied rates.
42