Knowledge that Transforms

We conduct tests for reference dependent loss aversion using slot machine gamblers' decisions on when to quit playing for a visit to a casino. Evidence for a lagged status-quo reference point is found in the aggregate, while endogenously determined reference points are found when conditioning on betting intensity choices. Significant deviations from the distributions implied by random quitting support the loss aversion and diminishing sensitivity hypotheses.

We document the correlation between the workplace diversity in bureaucratic organizations and public service delivery. We do so in the context of Nigeria, where ethnicity is a salient form of self-identity. We thus expand the empirical management literature highlighting beneficial effects of workplace diversity, that has focused on private sector firms operating in high-income settings. Our analysis combines two data sources: (i) a survey to over 4,000 bureaucrats eliciting their ethnic identities; (ii) independent engineering assessments of completion rates for 4,700 public sector projects. The ethnic diversity of bureaucracies matters positively: a one standard deviation increase in the ethnic diversity of bureaucrats corresponds to 9 percent higher completion rates. In line with the management literature from private sector firms in high-income countries, this evidence highlights a potentially positive side of ethnic diversity in public sector organizations, in the context of Sub-Saharan Africa.

In the repeated prisoner's dilemma, predictions are notoriously difficult. Recently, however, Blonski, Ockenfels, and Spagnolo (2011)—henceforth, BOS—showed that experimental subjects predictably cooperate when the discount factor exceeds a particular threshold. I analyze individual strategies in four recent experiments to examine whether strategies are predictable, too. Behavior is well summarized by “Semi-Grim” strategies: cooperate after mutual cooperation, defect after mutual defection, randomize otherwise. This holds both in aggregate and individually, and it explains the BOS-threshold: Semi-Grim equilibria appear as the discount factor crosses this threshold, and then, subjects start cooperating in round 1 and switch to Semi-Grim in continuation play. (JEL C72, C73, C92, D12)

Energy-efficient technologies offer considerable promise for reducing the financial costs and environmental damages associated with energy use, but these technologies appear not to be adopted to the degree that appears justified, even on a purely private basis. We present two complementary frameworks for understanding this so-called “energy paradox” or “energy efficiency gap.” First, we build upon previous literature by dividing potential explanations for the energy efficiency gap into three categories: market failures, behavioral anomalies, and model and measurement errors. Second, we examine the elements of cost-minimizing energy efficiency decisions, the typical benchmark used in assessing the gap's magnitude.

Apparently mistaken decisions are ubiquitous. To what extent does this reflect irrationality, as opposed to a rational trade-off between the costs of information acquisition and the expected benefits of learning? We develop a revealed preference test that characterizes all patterns of choice “mistakes” consistent with a general model of optimal costly information acquisition and identify the extent to which information costs can be recovered from choice data. (JEL D11, D81, D83)

We study whether people became less likely to switch Medicare prescription drug plans (PDPs) due to more options and more time in Part D. Panel data for a random 20 percent sample of enrollees from 2006--2010 show that 50 percent were not in their original PDPs by 2010. Individuals switched PDPs in response to higher costs of their status quo plans, saving them money. Contrary to choice overload, larger choice sets increased switching unless the additional plans were relatively expensive. Neither switching overall nor responsiveness to costs declined over time, and above-minimum spending in 2010 remained below the 2006 and 2007 levels.

The US Social Security Disability Insurance (SSDI) program is designed to provide income support to workers who become unable to work because of a severe, long-lasting disability. In this study, we use administrative data to estimate the effect of labor market conditions, as measured by the unemployment rate, on the number of SSDI applications, the number and composition of initial allowances and denials, and the timing of applications relative to disability onset. We analyze the period of the Great Recession, and compare this period with business cycle effects over the past two decades, from 1992 through 2012.

A commonly held view is that laboratory experiments provide researchers with more “control” than natural field experiments. This paper explores how natural field experiments can provide researchers with more control than laboratory experiments. While laboratory experiments provide researchers with a high degree of control in the environment which participants agree to be experimental subjects, when participants systematically opt out of laboratory experiments, the researcher's ability to manipulate certain variables is limited. In contrast, natural field experiments bypass the participation decision altogether due to their covertness, and they allow for a potentially more diverse participant pool within the market of interest.

We examine how participation in social safety net programs differs by income-to-poverty levels, and how that relationship changed after the Great Recession. We define income-to-poverty based on the average of 2 years of merged CPS data, and investigate program participation among households with income less than 300 percent of poverty. We find changes in both the level and distribution of safety-net program participation during the Great Recession, with SNAP expanding most at the bottom, the EITC expanding most in the middle, and UI expanding most at the top of the income ranges that we investigate; TANF did not expand.

Researchers often report estimates and standard errors for the object of interest (such as a treatment effect) based on a single specification of a statistical model. We propose a systematic approach to assessing sensitivity to specification. We construct estimates of the object of interest for each of a large set of models. Our proposed robustness measure is the standard deviation of the point estimates over the set of models. Each member of the set is generated by splitting the sample into two subsamples based on covariate values, constructing separate parameter estimates for each subsample, and then combining the results.