基于广义有序模型的“Easterlin悖论”验证

“Easterlin Paradox” Verification Based on Generalized Ordered Model

对于幸福感这种有序因变量的计量回归基本上采用Ordered Probit(Logit)等传统有序模型,国内研究尚未出现讨论并使用广义有序模型的论文。与传统有序模型相比,广义有序模型考虑了解释变量对于被解释变量的影响随着潜在变量阈值的不同而不同,放松了传统有序模型须满足平行线假设条件,广义有序模型具有更大的适用性。文章在传统有序模型的基础上引入广义有序模型,并以农民工幸福感为例,讨论广义有序模型的应用及其与传统有序模型估计结果的联系与区别。实证研究发现:对于幸福感较低的农民工人群,增加收入能够提升农民工幸福感,当收入达到一定程度时,幸福感开始降低,即收入与幸福感之间存在倒U型特征;但对于幸福感较高的农民工群体,增加收入后,农民工幸福感反而降低,存在Easterlin悖论。传统有序模型不加区分不同幸福感的农民工群体笼统地指出Easterlin悖论的存在,显然具有局限性。

Traditional ordered models such as Ordered Probit or Ordered Logit are basically used to meter ordered dependent variable like happiness, but there are no domestic research papers that discuss and use generalized order model. Compared with traditional ordered model, generalized ordered model considers that the influence of explanatory variable on explained variable varies with the threshold of potential variable, breaking through the condition that the traditional ordered model must satisfy the assumption of parallel lines, thus having greater applicability. The author bases his paper on the traditional order model to introduce the generalized order model, and then takes the happiness of migrant workers as an example to discuss the application of the generalized order model, as well as the relation and difference between the estimation results of the generalized order model and the traditional order model. Empirical research finds that for migrant workers with low happiness, increasing income can improve their happiness;when income reaches a certain level, happiness starts to decrease, that is, there is an inverted u-shape between income and happiness;however, for the migrant workers with high happiness, their happiness decreases after increasing their income, which presents Easterlin paradox. Obviously the traditional order model has limitations when pointing out the existence of Easterlin paradox in general without distinguishing the migrant workers with different sense of happiness.