Identifying welfare effects from subjective questions

We argue that the welfare inferences drawn from answers to subjective-quali tative survey questions are clouded by concerns over the structure of measu rement errors and how latent psychological factors influence observed respo ndent characteristics. We propose a panel data model that allows more robus t tests and we estimate the model on a high-quality survey for Russia. We f ind significant income effects on an individual's subjective economic welfa re. Demographic effects are weak at given income per capita. Ill and becomi ng unemployed lower welfare at given current income, although the unemploym ent effect is not robust, and returning to work does not restore welfare wi thout an income gain.