Finally we may ask what country characteristics are tied with a trade elasticity that is negative or one that is positive. If the compositional effects of trade were primarily driven by the simple pollution haven hypothesis we would expect a strong negative correlation between relative income and the magnitude of the trade elasticity. In fact as shown in Figure 1, there is no such relationship between the size of a country’s trade elasticity and its relative income level review.
Similarly, if the compositional effects of trade were primarily driven by the simplest factor endowment hypothesis we would expect a strongly positive relationship between relative capital abundance and the sign of a country’s trade elasticity. In fact as shown in Figure 2, there is little apparent relationship between the strength of a country’s trade elasticity and its relative capital abundance.
The explanation for these finding is simple: low-income countries typically have both low income per capita and low capital to labor ratios. The pollution haven hypothesis suggests that a low-income economy should be made dirtier by trade, but if pollution intensive industries are also capital intensive then whatever benefits accrue from lax pollution regulation could be largely undone by the relatively higher price of capital in this capital scarce country. As a result, further openness to trade will have a very small effect on the pollution intensity of output for low-income countries.
Similarly, high-income countries have both high income and high capital to labor ratios. The former argues in favor of trade lowering the pollution intensity of output, while the latter argues in favor of trade raising it. It is not that the (ceterus paribus) pollution haven hypothesis is wrong, or that the (ceterus paribus) factor endowment driven basis for trade is absent. Rather it is that given the relationship between income per capita and capital to labor ratios (summarized for example by the one-sector neoclassical growth model) these two partial theories work against each other. Consequently, the potentially very large composition effects predicted by either theory turn out to be relatively small in practice.