From table 1 it also appears that our strategy for identifying the separate, but related, impacts of changes in scale and technique is successful. Recall that since scale is measured in the intensive form GDP/km there is within-country heterogeneity in the scale variable for most countries. If, as we assume, pollution policy is determined by average income per capita in a country, then variation in the scale variable across cities within the same country can be used to separate the influence of scale from that of technique. Therefore the recognition that scale should be measured in intensive form together with a theoretical restriction linking policy to national income allows us to disentangle these two effects in our data.
In addition to these observations table 1 also reports that it may be important to distinguish between communist and non-communist countries. This would appear to support our concerns to distinguish carefully across countries according to the type of political system. Link If we investigate further and interact the communist dummy with our income terms reflecting the technique effect we find that pollution concentrations in communist countries are much less responsive to increases in real income.
This result is consistent with our theory as it implies that eT is must smaller for communist countries. In the fixed effects case, the elasticity of concentrations to an increase in per capita income in communist countries has a point estimate of 0.594 but the 95% confidence interval includes zero and is given by (-0.139,1.326). And hence we cannot reject the hypothesis of no technique effect in communist countries! In the random effects case, the point estimate is -0.587 with a 95% confidence interval of (-1.062,-0.111). We have excluded the communist-income interaction terms from table 1 to avoid clutter, but include them in all subsequent regressions.
It also appears that weather has a significant affect on concentrations. We find an increase in average temperature reduces concentrations as we may expect, and an increase in the concentration of yearly precipitation raises concentrations. Finally, the estimates indicate that locations in less dense areas, either suburban or rural locations, experience less pollution than locations at city center (our excluded category).