Our estimates of the impact of policy liberalization, derived from tariffs plus NTBs (i.e., Scenario
2 plus Scenario 5), exceed the remaining amount of US trade growth after independently accounting
for GDP growth and exchange rate changes. However, if we focus solely on Scenario 2—i.e., the impact
of just tariff liberalization since the Tokyo Round—we see that this dimension of policy liberalization
explains roughly 70 percent, or $175 billion, of the $250 billion in two-way trade growth not explained
by GDP growth or exchange rate changes. This works out to roughly 11 percent of total US two-way
trade growth. Changes in transportation costs explain a small portion of US trade growth over the period.
Since the 17 partners accounted for roughly 85 percent of US trade in 2004, we can extrapolate
from the 17 partner results to the whole world. These results suggest that tariff liberalization since
the Tokyo Round has boosted US two-way trade by roughly $200 billion per annum. The decline in
transportation costs adds another $30 billion. Our estimates of NTB liberalization suggest a further
$250 billion impact on two-way trade—quite a large figure. Either this NTB estimate has to be sharply
discounted or we need to reduce the assigned income elasticities of merchandise trade (exports and
imports) with respect to GDP. Since income elasticities have a far stronger econometric basis than NTB
estimates, we are inclined to discount the large NTB figure.
Benefits of Trade Expansion
Our estimates of the US trade expansion induced by policy liberalization can be converted into income
effects. Bradford, Grieco, and Hufbauer (2006) have investigated the benefit for US economic welfare of
US trade expansion since the 1950s. The authors draw on methods and key results from several studies to
produce a range of estimates. We follow one of the methods set out in Bradford, Grieco, and Hufbauer
(2006) to make our estimate of the income effects of trade growth induced by policy liberalization.
In an effort to understand the effect of various policies and characteristics on per capita income
growth, an OECD (2003) study found that a 10 percent rise in a developed country’s long-term trade
exposure leads to a 2 percent increase in the level of annual per capita income (measured by GDP per
capita). A standard measure of trade exposure is exports plus imports divided by GDP.11
Using the OECD (2003) coefficient of 0.2 (2 percent divided by 10 percent), we can estimate the
per capita income effect under each of the six scenarios. To do so we must first scale up the export and
import effects displayed in the last row of table 6, because these estimates cover only about 85 percent of
US trade. We then calculate the actual US merchandise trade exposure in 2004 (20.1 percent) and the
hypothetical merchandise trade exposures if we took away the trade growth suggested by each of the six
11. OECD (2003) uses a slightly different measure, exports divided by GDP plus imports divided by GDP minus exports
plus imports. Bradford, Grieco, and Hufbauer (2006) use the sum of exports plus imports divided by GDP, and we do the
same here.