New Ecological Paradigm meets behavioral economics : on the relationship between environmental values and economic preferences

Many studies examine the effect of environmental values on environmental behavior. In such empirical analyses it is typically at least implicitly assumed that environmental values are independent of economic preferences from behavioral economics like risk and time preferences, trust, or reciprocity, which play an important role for the explanation of indi-vidual behavior. This paper tests whether environmental values are related to economic preferences and examines possible consequences when independence is assumed. The data for this test stem from a large-scale computer-based survey among more than 3700 Ger-man citizens. Our indicators for environmental values are based on the New Ecological Paradigm (NEP), which is a standard instrument in the social and behavioral sciences and increasingly common in the economic literature. The econometric analysis with General-ized Poisson regression models reveals strong correlations between the NEP scales and economic preferences, which are based on established experimental measures: While so-cial preferences (measured in an incentivized dictator game) and positive reciprocity are significantly positively correlated, trust and (less robust) negative reciprocity are signifi-cantly negatively correlated with environmental values, respectively. Only risk and time preferences (also measured in an incentivized experiment) are not robustly significantly correlated with the NEP scales. These estimation results strongly recommend the addition-al inclusion of economic preferences and especially of social preferences, trust, and posi-tive and negative reciprocity in econometric analyses that use a NEP scale as explanatory factor for individual behavior since their non-consideration can lead to strong distortions due to omitted variable biases. This conclusion is illustrated in an empirical example that reveals biased estimation results for the effect of a NEP scale on donation activities if not all relevant economic preferences are included as control variables.