For more than a decade, an MIT-trained economist has meticulously recorded the results of nearly 75,000 deliveries of bagels and donuts to corporate offices. These bagels and donuts are left in workplaces in the morning, along with a list of prices and a lockbox. Office workers purchase the products on the honor system. Later that day, the uneaten goods and the lock box are collected and the relevant data for each office are recorded in a spreadsheet. These data provide a unique opportunity for understanding the extent to which individuals behave honestly in this setting and the factors that influence the level of honesty. An analysis of the discrepancy between the prices of the goods consumed and the actual payments deposited in the lock box yields a number of insights. First, the base payment rate is quite high. On average, payments represent almost 90 percent of the posted price. In a model of pure self-interest, one would expect very low payment rates, because there would seem to be many opportunities for individuals to take the products with almost no chance of their nonpayment being detected. Moderating the tendency to take the products without paying is the fact that low payment rates will increase the probability the delivery company stops serving the company. Paying for one’s bagels and donuts, viewed in this light, is a public good for other office workers. Thus, one might expect a strong negative relationship between office size and payment rates, which is not present in the data. Rather, the data appear more consistent with a model in which there are internal, nonpecuniary costs associated with stealing the bagels and donuts, i.e., people overwhelmingly pay for the products because it would be wrong not to do so. The observed payment rates are systematically related to a variety of observable factors. For instance, payment rates fall in response to increases in the posted price. This is consistent with the increase in financial costs pushing some marginal consumers to be willing to sustain the moral cost associated with not stealing the goods or paying less than full price. Payment rates are higher when many of the bagels and donuts go uneaten, suggesting that the marginal consumer pays a lower share of the posted price than the inframarginal consumer. Purchasers of donuts appear to be more likely to pay less than the full price. There is a sharp and persistent increase in the payment rate following the September 11 terrorist attacks.

School choice has become an increasingly prominent strategy for enhancing academic achievement. To evaluate the impact on participants, we exploit randomized lotteries that determine high school admission in the Chicago Public Schools. Compared to those students who lose lotteries, students who win attend high schools that are better in a number of dimensions, including peer achievement and attainment levels. Nonetheless, we find little evidence that winning a lottery provides any systematic benefit across a wide variety of traditional academic measures. Lottery winners do, however, experience improvements on a subset of nontraditional outcome measures, such as self-reported disciplinary incidents and arrest rates.