Search

We present a new type of field experiment to investigate ethnic prejudice in the workplace. Our design allows us to study how potential discriminators respond to changes in the cost of discrimination. We find that ethnic discrimination is common but highly responsive to the “price of prejudice,” i.e., to the opportunity cost of choosing a less productive worker on ethnic grounds. Discriminators are on average willing to forego 8 percent of their earnings to avoid a coworker of the other ethnic type. The evidence suggests that animus rather than statistical discrimination explains observed behavior. (JEL C93, J15, J24, J31, J71)

Although many countries are aggressively implementing the One Laptop per Child (OLPC) program, there is a lack of empirical evidence on its effects. This paper presents the impact of the first large-scale randomized evaluation of the OLPC program, using data collected after 15 months of implementation in 319 primary schools in rural Peru. The results indicate that the program increased the ratio of computers per student from 0. 12 to 1. 18 in treatment schools. This expansion in access translated into substantial increases in use both at school and at home. No evidence is found of effects on enrollment and test scores in Math and Language. Some positive effects are found, however, in general cognitive skills as measured by Raven’s Progressive Matrices, a verbal fluency test and a Coding test.

A housing lottery in an Indian city provided winning slum dwellers the opportunity to move into improved housing on the city’s periphery. Fourteen years later, relative to lottery losers, winners report improved housing farther from the city center, but no change in family income or human capital. Winners also report increased isolation from family and caste networks and lower access to informal insurance. We observe significant program exit: 34% of winners never moved into the subsidized housing and 32% eventually exited. Our results point to the importance of considering social networks when designing housing programs for the poor.Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at www.nber.org.

Research on behavioral economics has established the importance of factors such as reference dependent preferences, hyperbolic discounting, and the value placed on non-financial rewards. To date, these insights have had little impact on the way the educational system operates. Through a series of field experiments involving thousands of primary and secondary school students, we demonstrate the power of behavioral economics to influence educational performance. Several insights emerge. First, we find substantial incentive effects from both financial and non-financial incentives on test scores. Second, we find that non-financial incentives are considerably more cost-effective than financial incentives for younger students, but were less effective with older students. Third, and perhaps most importantly, consistent with hyperbolic discounting, all motivating power of the incentives vanishes when rewards are handed out with a delay. Since the rewards to educational investment virtually always come with a delay, our results suggest that the current set of incentives may lead to underinvestment. Fourth, in stark contrast to previous laboratory experiments, we do not see an increased response of effort when rewards are framed as losses. Our findings imply that in the absence of immediate incentives, many students put forth low effort on standardized tests, which may create biases in measures of student ability, teacher value added, school quality, and achievement gaps.

Market outcomes depend on the quality of information available to its participants. We measure the effect of information disclosure on market outcomes using a large-scale field experiment that randomly discloses quality information in wholesale automobile auctions. We argue that buyers in this market are horizontally differentiated across cars that are vertically ranked by quality. This implies that information disclosure helps match heterogeneous buyers to cars of varying quality, causing both good and bad news to increase competition and revenues. The data confirm these hypotheses. These findings have implications for the design of other markets, including e-commerce, procurement auctions, and labor markets. (JEL C93, D44, D82, L15)

We use an RCT to analyze the impacts of microcredit. The study population consists of loan applicants who were marginally rejected by an MFI in Bosnia. A random subset of these were offered a loan. We provide evidence of higher self-employment, increases in inventory, a reduction in the incidence of wage work and an increase in the labor supply of 16–19-year-olds in the household's business. We also present some evidence of increases in profits and a reduction in consumption and savings. There is no evidence that the program increased overall household income. (JEL C93, G21, I38, J23, L25, P34, P36)

This paper examines whether individual and team incentives can improve student learning and performance using a theoretical analysis and a randomized eld experiment. If individual incentives are provided to students based on a specied target or performance standard for all, on the one hand it may not work for the low-ability students if they perceive the target to be too high, while on the other hand the high-ability students will win the prizes without extra eort if the target is too low. Only students with an ability level close to the target will attempt to win by working harder. If the students are grouped in teams and evaluated based on the team average, incentives can have an eect for all students across the spectrum of ability, because the average team abilities are pushed toward the target and there is also an incentive to help each other within teams. A team incentive augmented with a tournament component may be even more desirable because it eliminates the target, inducing all teams to work harder, because they do not know what the quality and eort of their competitors will be. However, team incentives may be prone to free riding and coordination problems. This paper presents the results of a randomized eld experiment for a representative sample of 100 Beninese secondary schools to compare these three methods of incentivizing students.

This paper estimates how experimentally-manipulated experiences with a novel financial product, rainfall index insurance, affect subsequent insurance demand. Using a seven-year panel, we develop three main findings. First, recent experience matters for demand, consistent with overinference from small samples. Second, spillovers also matter, in the sense that the recent payout experience of village co-residents affects insurance demand about as much as one's own recent payout experience. Third, the spillover effect decays as time passes while the effect of one's own experience does not. We discuss implications of this analysis for commercial sustainability of this complicated but promising risk management technology.

Using data from a field experiment in Kenya, we document that providing individuals with simple informal savings technologies can substantially increase investment in preventative health and reduce vulnerability to health shocks. Simply providing a safe place to keep money was sufficient to increase health savings, through a mental accounting effect. Adding an earmarking feature was only helpful when funds were put towards emergencies; earmarking for preventative health reduced savings on average, because the liquidity cost of tying up money was too great. Providing social pressure and credit through a ROSCA-based savings scheme had very large effects.

We outline a strategy for distinguishing rank-dependent probability weighting from systematic risk misperceptions in field data. Our strategy relies on singling out a field environment with two key properties: (i) the objects of choice are money lotteries with more than two outcomes; and (ii) the ranking of outcomes differs across lotteries. We first present an abstract model of risky choice that elucidates the identification problem and our strategy. The model has numerous applications, including insurance choices and gambling. We then consider the application of insurance deductible choices and illustrate our strategy using simulated data.

We propose a new classification of experiments that captures the extent to which the experimental design and analysis are linked to economic theory. We then use this system to classify all published field experiments in the five top economics journals from 1975 to 2010. We find that the vast majority of field experiments (68%) are Descriptive studies that lack any explicit model; 18% are Single Model studies that test a single model-based hypothesis; 6% are Competing Models studies that test competing model-based hypotheses; and 8% are Parameter Estimation studies that estimate structural parameters in a completely specified model. Using the same system to classify laboratory experiments published over the same period, we find that economic theory has played a more central role in the laboratory than in the field. Finally, we discuss in detail three sets of field experiments, on gift exchange, on charitable giving, and on negative income tax, that illustrate both the benefits and the potential costs of a tighter link between experimental design and theoretical underpinnings.

We study the impact of status and social recognition on worker performance in a field experiment. In collaboration with an international non-governmental organization we hired students to work on a database project. Students in the award treatment were offered a congratulatory card from the organization honoring the best performance. The award was purely symbolic in order to ensure that any behavioral effect is driven by non-material benefits. Our results show that students in the award treatment outperform students in the control treatment by about 12 percent on average. Our results provide strong evidence for the motivating power of status and social recognition in labor relations with major implications for theory and applications.

During the night of April 14, 1912, the RMS Titanic collided with an iceberg on her maiden voyage. Two hours and 40 minutes later she sank, resulting in the loss of 1,501 lives—more than two-thirds of her 2,207 passengers and crew. This remains one of the deadliest peacetime maritime disasters in history and by far the most famous. For social scientists, evidence about how people behaved as the Titanic sunk offers a quasi-natural field experiment to explore behavior under extreme conditions of life and death. A common assumption is that in such situations, self-interested reactions will predominate and social cohesion is expected to disappear. However, empirical evidence on the extent to which people in the throes of a disaster react with self-regarding or with other-regarding behavior is scanty. The sinking of the Titanic posed a life-or-death situation for its passengers. The Titanic carried only 20 lifeboats, which could accommodate about half the people aboard, and deck officers exacerbated the shortage by launching lifeboats that were partially empty. Failure to secure a seat in a lifeboat virtually guaranteed death. We have collected individual-level data on the passengers and crew on the Titanic, which allow us to analyze some specific questions: Did physical strength (being male and in prime age) or social status (being a first- or second-class passenger) raise the survival chance? Was it favorable for survival to travel alone or in company? Does one's role or function (being a crew member or a passenger) affect the probability of survival? Do social norms, such as “Women and children first!” have any effect? Does nationality affect the chance of survival? We also explore whether the time from impact to sinking might matter by comparing the sinking of the Titanic over nearly three hours to the sinking of the Lusitania in 1915, which took only 18 minutes from when the torpedo hit the ship.

We explore the use of social comparison theory as a natural mechanism to increase contributions to an online movie recommendation community by investigating the effects of social information on user behavior in an online field experiment. We find that, after receiving behavioral information about the median user's total number of movie ratings, users below the median demonstrate a 530% increase in the number of monthly movie ratings, while those above the median decrease their monthly ratings by 62%. Movements from both ends converge towards the median, indicating conformity towards a newly-established social norm in a community where such a norm had been absent. Furthermore, the social information has a more dramatic effect on those below the median, suggesting an interaction between conformity and competitive preferences. When given outcome information about the average user's net benefit score from the system, consistent with social preference theory, users with net benefit scores above average contribute 94% of the new updates in the database. In both treatments, we find a highly significant Red Queen Effect.

We conduct a field experiment in which highly-ranked chess players play the centipede game in a natural setting. This game represents one of the main paradoxes of backward induction. In the experiment two players alternately are faced with the decision of either taking an exponentially growing pile of money and ending the game, or letting the other player make the decision. The player who decides to stop the game takes the larger portion of the pile, and the other player gets the remaining amount. All standard equilibrium concepts dictate that the player who decides first must stop the game immediately. There is vast experimental evidence, however, that this rarely occurs. Contrary to this evidence our results show that 69% of chess players stop the game immediately. When we restrict attention to chess Grandmasters this percentage escalates to 100%. We also conduct standard laboratory experiments where college students and chess players play ten repetitions of the game. We find that chess players playing versus other chess players rapidly converge to the equilibrium outcome, whereas students playing versus other students systematically depart from it. However, when students play against chess players the occurrence of the backward induction outcome increases tenfold.

We run a large field experiment with an online company specializing in selling used automobiles via ascending auctions. We manipulate experimentally the “price grid,” or the possible amounts that bidders can bid above the current standing price. Using two diverse auction sites, one in New York and one in Texas, we find that buyer and seller behavior differs strikingly across the two sites. Specifically, in Texas we find peculiar patterns of bidding among a small but prominent group of buyers suggesting that they are “cyber-shills” working on behalf of sellers. These patterns do not appear in the New York auctions. (JEL C93, D12, D44, L62, L81)

Micro-entrepreneurs often lack the financial literacy required to make important financial decisions. We conducted a randomized evaluation with a bank in the Dominican Republic to compare the impact of two distinct programs: standard accounting training versus a simplified, rule-of-thumb training that taught basic financial heuristics. The rule-of-thumb training significantly improved firms' financial practices, objective reporting quality, and revenues. For micro-entrepreneurs with lower skills or poor initial financial practices, the impact of the rule-of-thumb training was significantly larger than that of the standard accounting training, suggesting that simplifying training programs might improve their effectiveness for less sophisticated individuals., International Finance Corporation, Ideas42

We study the design of randomized controlled experiments in environments where outcomes are significantly affected by unobserved effort decisions taken by the subjects (agents). While standard randomized controlled trials (RCTs) are internally consistent, the unobservability of effort provision compromises external validity. We approach trial design as a principal-agent problem and show that natural extensions of RCTs--which we call selective trials--can help improve the external validity of experiments. In particular, selective trials can disentangle the effects of treatment, effort, and the interaction of treatment and effort. Moreover, they can help experimenters identify when measured treatment effects are affected by erroneous beliefs and inappropriate effort provision.

Game theory is usually difficult to test precisely in the field because predictions typically depend sensitively on features that are not controlled or observed. We conduct one such test using field data from the Swedish lowest unique positive integer (LUPI) game. In the LUPI game, players pick positive integers and whoever chose the lowest unique number wins a fixed prize. Theoretical equilibrium predictions are derived assuming Poisson- distributed uncertainty about the number of players, and tested using both field and laboratory data. The field and lab data show similar patterns. Despite various deviations from equilibrium, there is a surprising degree of convergence toward equilibrium. Some of the deviations from equilibrium can be rationalized by a cognitive hierarchy model.