Note: The elasticities shown in the above diagrams correspond to the random-effects regression presented in table 3, evaluated at corresponding sample means. Countries with less than five observations in the data set were excluded from the above diagrams.
Next consider the trade intensity elasticity. The trade intensity elasticity measures the predicted change in concentrations for a 1% change in the ratio of exports plus imports to GDP. This measure indicates that a 1% change in the share of trade in GDP should reduce concentrations by .53% in the random effects model and .86% in the fixed effects model. These seem rather large in isolation, but the estimates from table 3 also indicate that technological progress in abatement technology or changing knowledge and attitudes toward pollution appear to be driving concentrations down by 3-4% per year.
Note our trade intensity estimate (evaluated at the mean of our sample) is negative and significantly different from zero in both formulations. This is a somewhat surprising result because it indicates that trade has an overall negative composition effect rather than a close to zero effect we may have expected. Proposition 1 indicates that the sign of the trade intensity elasticity should reflect a country’s comparative advantage in clean versus dirty goods. Therefore it is not plausible that all countries in the world have a negative composition effect. Although we have only a sample of countries it seems reasonable to expect both positive and negative elasticities.
As a check on our theory we calculated each country’s trade elasticity. We find that the country specific elasticity estimates are both positive and negative. About 1/3 of the countries have trade elasticities indistinguishable from zero. We find some positive elasticity estimates, but the majority of elasticities in our sample are negative and statistically different from zero. These findings are roughly consistent with our theory, because our theory only predicts that there should be a distribution of these elasticities around zero. payday loans with no bank account