We examine the relationships among water pollution, income, and political institutions using country-level global water quality data over the period 1980 to 2012. In order to address concerns about the highly nonlinear relationship between pollution and income, the endogeneity of income, and the discrete nature of political variables, we use a nonparametric instrumental variable approach that allows for the inclusion of continuous and discrete variables to identify these relationships. Results indicate an inverted-U shaped relationship between pollution and income consistent with an environmental Kuznets curve for one pollutant (lead), a cubic shape for three pollutants (nickel, mercury, and fecal coliform), and more highly nonlinear relationships for many of the other pollutants. For several stock pollutants (nickel, mercury, and arsenic), we find that pollution levels may continue to increase at higher levels of income, suggesting that stock pollutants may continue to accumulate in productive high-income countries even when marginal emissions have been reduced. We also find suggestive evidence that levels of pollutants resulting from industrial activity (e.g., chemical oxygen demand) may increase with income while those that are more driven by residential activity and population levels (e.g., fecal coliform) do not. By estimating a nonparametric relationship between pollution and political institutions and by accounting for the categorical nature of the political variables, we are able to detect a nonlinear relationship between pollution and political institutions as well, which for some pollutants is an inverted-U shaped curve.

我们使用1980年至2012年期间的国家/地区级全球水质数据，研究了水污染，收入与政治机构之间的关系。为了解决有关污染与收入，收入的内生性和收入之间的高度非线性关系的担忧。政治变量的离散性质，我们使用非参数工具变量方法，该方法允许包含连续变量和离散变量以识别这些关系。结果表明，污染与收入之间呈倒U型关系，与一种污染物（铅）的环境库兹涅茨曲线，三种污染物（镍，汞和粪便大肠菌群）的立方关系一致，并且对于许多污染物，其非线性关系更为密切其他污染物。对于几种污染物（镍，汞和砷），我们发现，随着收入水平的提高，污染水平可能会继续增加，这表明即使减少了边际排放量，生产性高收入国家中的污染物仍可能继续积累。我们还发现了有力的证据，表明工业活动（例如化学需氧量）导致的污染物水平可能随收入增加而增加，而居民活动和人口水平（例如粪大肠菌群）驱动的污染物水平却没有增加。通过估计污染与政治制度之间的非参数关系，并考虑到政治变量的分类性质，我们还能够检测出污染与政治制度之间的非线性关系，对于某些污染物，这是一条倒U型曲线。