Attention that has been directed toward the effect of trade on foreigner's
holdings of domestic currency comes exclusively from time-series evidence.
Here we extend a model of currency substitution that incorporates a trade m
otive for foreigners to hold domestic currency. It uses time-series and cro
ss-sectional information for a panel of 17 industrialized countries testing
two-way fixed-effects models against pooled and random-effects alternative
s. The cross-sectional information is significant, revealing that pooling o
f data could result in misleading inferences, since country-specific effect
s, regional group effects, and distance are all important determinants of d
omestic currency holding by foreigners. (JEL F41, F36, F14).