This dissertation presents two lines of research, each aimed at developing and assessing psychologically-motivated economics research in the realm of public policy. In the first chapter I present a theory of tax sheltering activities motivated by prospect theory (Kahneman and Tversky, 1979), where a loss-averse citizen frames a refund as a gain and a tax payment as a loss. A unique implication of this theory is a discrete drop in the marginal benefit of tax sheltering once crossing the threshold into the gain domain. This drives excess tax sheltering among individuals owing money on tax day, and an excess mass of individuals to shelter precisely to the gain/loss threshold. I investigate these implications in 19791990 IRS panel of individual returns and find strong support for loss aversion. A mixture-modeling approach is developed to estimate model parameters and conduct policy simulations. Estimates suggest that psychologically-motivated framing effects can have substantial impact on tax revenue. I discuss the implications of these results for the detection and deterrence of tax evasion, the implementation of tax-incentivized public programs, and forecasting behavioral response to tax policy changes. The second and third chapters assess current uses of happiness or subjective well-being (SWB) data in economic settings. Economists and policy makers often estimate the tradeoffs individuals accept and forecast the choices they will make. An increasingly-used approach to this exercise uses survey responses to SWB questions as a direct measure of economists' notion of utility. The research presented here directly assesses these practices across a variety of settings. Chapter 2 reports the results of three surveys eliciting choice and SWB over alternatives in a battery of hypothetical scenarios. Chapter 3 reports the results of a field study of medical residency choice, allowing the side-by-side comparison of choice-based and SWB-based tradeoff estimates. Across these studies, we find that while choice and SWB rankings are often reasonably well aligned, systematic differences exist, and are particularly problematic for inference on marginal rates of substitution. We discuss the implications of our results for the use of SWB measures in economic applications and the comparative performance of different SWB-based approaches.