[Excerpts taken from the article “The Sad Truth about Happiness Scales” by Timothy Bond and Kevin Lang, forthcoming in the Journal of Political Economy]

“A large literature has attempted to establish the determinants of happiness using ordered response data from questions such as ‘Taking all things together, how would you say things are these days—would you say that you are very happy, pretty happy, or not too happy?'”

“We…reach the striking conclusion that the results from the literature are essentially uninformative about how various factors affect average happiness.”

“The basic argument is as follows. There are a large (possibly infinite) number of states of happiness that are strictly ranked. In order to calculate a group’s “mean” happiness, these states must be cardinalized, but there are an infinite number of arbitrary cardinalizations, each producing a different set of means. The ranking of the means remains the same for all cardinalizations only if the distribution of happiness states for one group first-order stochastically dominates that for the other.”

“…we do not observe the actual distribution of states. We instead observe their distribution in a small number of discrete categories…Without additional assumptions we cannot rank the average happiness of two groups if each has responses in the highest and lowest category. Using observed covariates to achieve full nonparametric identification of the latent happiness distributions would require making assumptions that happiness researchers generally claim to reject.”

“We are therefore forced to follow the standard approach and assume the latent distributions are from a common unbounded location-scale family (e.g., an ordered probit). If we do, it is (almost) impossible to get stochastic dominance, and the conclusion is therefore not robust to simple monotonic transformations of the scale.”

“…we outlined the conditions under which the rank order of happiness for groups can be identified using categorical data on subjective well-being. We now put these into practice for nine key results from the happiness literature: the Easterlin (1973, 1974) paradox for the United States, whether happiness is U-shaped in age, the optimal policy trade-off between inflation and unemployment, rankings of countries by happiness, whether the Moving to Opportunity program increased happiness, whether marriage increases happiness, whether children decrease happiness, the relative decline of female happiness in the United States, and whether disabilities decrease happiness.”

“Table 1 summarizes the results.”

“None of these results are identified nonparametrically. Moreover, in the eight cases for which we can test for equality of variances under a parametric normal assumption, we reject equality. Thus we never have rank-order identification and can always reverse the standard conclusion by instead assuming a left-skewed or rightskewed lognormal.”

“…if researchers wanted to draw any conclusions from these data…they would have to argue that it is appropriate to inform policy based on one arbitrary cardinalization of happiness but not on another…”

“Researchers who wish to continue to interpret such questions more broadly need to…justify their particular cardinalization or parametric assumption…At a bare minimum, we would require a functional form assumption that survived the joint test of the parametric functional form and common reporting function across groups. Certainly calls to replace GDP with measures of national happiness are premature.”

To read the full article, click here. (NOTE: Article is behind a paywall.)