Does simultaneity exaggerate empirical tests of the trade-growth relationship?

Simultaneity between the volume of trade and national output may have biased past empirical tests of the export-growth relationship. To judge the extent of such bias, this paper compares results of single- and simultaneous- equation regression models of trade and growth that closely resemble the specifications of previous studies. In order to enhance the analysis, a number of other alleged shortcomings of previous statistical studies of trade and growth are addressed: omitted variable bias and non-stationarity are dealt with by using modern time-series regression procedures, measurement error is reduced by using more accurate measures of capital and labour available for six Latin American countries over the period 1960-1990, and both imports and exports are used to proxy trade. The results shown that simultaneity bias causes single equation results to understate, not over state, the trade-growth relationship.