28
Finally, readers might attribute the null result to selection bias. Perhaps investors
refrained from lending to countries that had little commercial intercourse with the United
Kingdom or the United States, such that the ones who actually attracted capital were
satisfactorily vulnerable to trade sanctions. If this objection is valid, then it might explain why
we found no relationship between trade and debt in the sample of borrowers.
Three pieces of evidence argue against this possibility, however. First, as Tomz (2004)
shows, investment primers of the interwar period almost never mentioned the direction of trade
or the prospect of an embargo as a factor in lending decisions. It therefore seems unlikely that
investors rationed credit on the basis of this variable. Second, the wide range on “trade with
creditors” provides a considerable degree of reassurance. British and American investors lent to
countries that conducted less than one percent of their foreign trade with the motherland and
would therefore have been relatively invulnerable to an embargo. Third, countries that borrowed
from the British actually conducted less trade with the United Kingdom, on average, than
countries that did not attract sterling loans. The behavior of US lenders seemed slightly more
consistent with rationing: American investors did indeed show a slight preference for countries
that traded extensively with the United States. The difference was not statistically significant,
however, either in a t-test of means or in a probit regression that explored whether dependence
on the United States could explain which countries got dollar loans. Thus, investors of the
interwar period did not use dependence on trade as a criterion for allocating credit.
Patterns of Discrimination
For additional evidence I turned to the subset of countries that defaulted. If the threat of
an embargo truly loomed, then defaulters should have pushed the cost of non-compliance onto
Convention