Second, we then need to condition this impact on country characteristics. To condition the impact of openness on country characteristics we interact the trade intensity measure with our model’s predicted determinants of comparative advantage. Within our framework the most important country characteristics determining trade patterns are a country’s capital to labor ratio and its income per capita. Moreover, because comparative advantage is a relative concept, we express our measures of country characteristics relative to their corresponding world averages. This procedure allows us to condition the predicted environmental impact of further openness on our theoretical determinants of comparative advantage.
Finally, we need to somehow account for the two possible motives for trade when we introduce our interaction terms with country characteristics. In general the trade-off between the factor endowment basis for trade and the pollution haven motive is exceedingly complex and not amenable to simple formulation. Rather than imposing specific functional forms that arise from some tractable special cases of our model, we instead rely on the results presented in Proposition 2 and 3.
Because our theory does not tell us at what point further increases in the capital to labor ratio raise pollution (via the composition effect) or when increases in per capita income finally lower pollution (via the composition effect), we adopt a flexible approach to estimation. We interact a quadratic in a country’s (relative) characteristics with its trade intensity ratio.