Combining the economic, site-specific, and common-to-world components we obtain:
where GDPjkt is measured by real GDP/km , KL^ is measured by the capital to labor ratio, Ikt is one period lagged three year moving average of GDP per capita, Rijk is a dummy indicating site ijk is in a rural location, Bijk is a dummy indicating site ijk is in a suburban location, Mjkt is average temperature in city j at time t, Mjkt is the variation in precipitation in city j at time t, Okt is measured by the ratio of exports and imports to GDP, Okt RKLkt and On (RKLkt) are interactions of openness with country k’s relative capital to labor ratio and its square, and Okt RIkt and Ot (RIkt) are interactions of openness with country k’s income per capita and its square. Link In addition to these determinants we include a dummy for communist countries in all of our estimations.
Our empirical strategy has four steps. We first estimate (3.2) excluding the terms involving openness to determine whether our simple model specification capturing scale, composition and technique effects is useful in explaining pollution concentration levels around the world. We then take a second step by adding several measures of “openness” to our basic model and noting the consequences. Our purpose here is to investigate whether a simple and definitive relationship exists between openness to international markets and pollution concentrations (after controlling for differences across countries in scale, factor endowments, etc.)
In our third step, we include our openness interactions to allow trade’s effect to differ across countries. Our theory would suggest that conditioning the impact of further openness on country characteristics is the key to determining how trade affects the pollution intensity of national output. In our fourth and final step we combine our scale, technique and trade intensity elasticities to provide a preliminary assessment of how trade affects SO2 concentrations.