Our methodology relies on using subjectively reported happiness measures in empirical work. But how much can such measures be trusted? Economists worry about the validity of such questions and to some extent the scientific evidence supports these worries (Bertrand and Mullainathan 2000). A large array of evidence has shown that subjective survey questions are prone to significant reporting error. For example, studies have found that the placement of wellbeing questions affects how they are answered. If they are preceded by a question, for example, that asks about dating behavior, people are more likely to report unhappiness. Beyond order choice, instantaneous mood at the time of survey is also found to have a large effect on how people answer such questions. Schwarz and Strack provide a nice survey of these effects.
Yet such results only tell us that there is measurement error in these questions. There is also measurement error in the numerous other variables that economists study. What is more relevant for our purposes is that the evidence is clear that these questions also contain significant true signal about well-being. Evidence of this kind comes in several varieties but they all follow a similar methodology: find a more objective measure of well-being and see how well this measure correlates with the self-report. And strong positive correlations have been found for a large set of such variables.
For example, outsider’s assessments of a person’s happiness or independent counts of smiles correlate positively with self-reported happiness. Moving to much more physiological measures, everything from heart rate, blood pressure, skin resistance measures of responses to stress, to even level of activity in the left versus right prefrontal lobe all are found to correlate with subjective reports of well being (Kahneman 1999; Gardner and Oswald, 2001). These studies all suggest that despite the measurement error inherent in this attitudinal question, it nevertheless correlates effectively with well being. so
Moreover, the small happiness literature in economics also has uncovered interesting patterns further bolstering the idea that these variables in fact measure well-being. In crosssections, happiness generally rises with factors that economists would associate with improved well being, such as higher incomes. The income effect appears to be causal, as it is present for lottery winners and those receiving inheritances (Gardner and Oswald, 2001). Self-reported well-being is also lower for the unemployed, and for those who are divorced (Blanchflower and Oswald, 2000); interestingly, however, the reduction in happiness due to unemployment is mitigated when there is a larger “reference group” of unemployed (Clark, 2000).
Despite the increased use of this measure, there has been no attempt to date of which we are aware that uses these subjective well-being measures to attempt to either distinguish models of behavior, or to draw welfare conclusions about particular tax or spending interventions. As a whole, therefore, the available evidence suggests that while subjective well-being measures do contain noise, they also contain significant signal and are a fruitful area for empirical exploration.