Environmental Engel Curves: A Neural Network Approach

Abstract Environmental Engel curves describe how households' income relates to the pollution associated with services and goods consumed. This paper estimates these neural networks using novel dataset constructed in Levinson O'Brien. We provide further statistical rigor empirical analysis by constructing prediction intervals obtained from network methods such as extra-neural nets MC dropout. The application of techniques for five different pollutants allow us confirm statistically that are upward sloping, have elasticities smaller than one shift down, becoming more concave, over time. Importantly, last year sample, we find an inverted U shape suggests existence a maximum medium-to-high levels household beyond which flattens or decreases top earners.