By far the best overview of the happiness literature from an economics and policy perspective is Frey & Stutzer's Happiness and Economics. Frey, a first class constitutional theorist, explores the big impact policy can have on happiness, but warns against construing the happiness function as an approximation of that unicorn of social science, the social welfare function, and then trying to maximize it. F&S point out that the goal of maximizing social welfare has traditionally faced three problems: (1) empirical emptiness; (2) aggregation (Arrow social choice stuff); (3) incentive compatibility (Buchanan public choice stuff). The happiness data perhaps solves (1), but it does not solve (2) or (3). Folks like Layard are particularly inept with respect to (3). Here's what F&S say:
Missing incentives. Deriving optimal policies by maximizing a social welfare function only makes sense if the government has an incentive to apply the optimal policies in reality. This is only the case if a “benevolent dictator” government is assumed. (Brennan and Buchanan 1985). From introspection as well as from empirical analysis in political economy (see, e.g., the collection of papers on political business cycles in Frey 1997), we know that governments are not benevolent and do not follow the wishes of the population, even in well-functioning democracies, not to mention authoritarian and dictatorial governments. Hence to maximize social welfare corresponds to a “technocratic-elitist” procedure, neglecting the crucial incentive aspect.
This criticism applies particularly when one tries to derive optimal policies by maximizing happiness.
This point cannot possibly be emphasized enough. Even if your theory of value, or your philosophical standard for policy evaluation, is “correct” in some metaphysical sense, this gets us almost nowhere. Why? First, people, especially agents of the state, must understand and broadly agree that it is correct. This is exceedingly unlikely. Second, people, especially agents of the state, must be motivated to reliably act in accordance with its prescriptions. But if understanding and agreement is unlikely, then motivation based on them is unlikely. It may be conceivable to structure incentives so that it is as if agents of the state were motivated by a commitment to a single normative standard, but it is usually unrealistic.
Here is F&S's approach:
There is a solution on hand that overcomes the problems posed by the impossibility theorem and by the government's missing incentives. Constitutional political economy (e.g., Buchanan 1991, Frey 1983, Mueller 1996, Cooter 2000) redirects attention to the level of social consensus where, behind the veil of ignorance, the basic rules governing a society–the fundamental institutions–are chosen or emerge. At the same time, the approach shifts from a (vain) effort to directly determine social outcomes to shaping the politico-economic process by setting the institutions.
. . . The fundamental institutions shape the incentives of policymakers. Once these basic institutions are in place and the incentives are set, little can be done to influence the current politico-economic process. Economic policy therefore must help to establish those fundamental institutions, which lead to the best possible fulfilment of individual preferences. Research in positive constitutional economics helps to identify which institutions serve this goal, and whether they do in fact systematically affect happiness.
F&S's research shows that procedural aspects of government, such as the directness of democratic participation and the degree of decentralized federalism, are themselves determinants of happiness. The Swiss are among the happiest people in the world not only because they are fabulously wealthy, but because the canton system allows for direct democratic rule of over local jurisdictions, with little interference from the central state.