Working Papers

This study provides the first comprehensive analysis of the distributional effects of Head Start, using the first national randomized experiment of the Head Start program (the Head Start Impact Study). We examine program effects on cognitive and non-cognitive outcomes and explore the heterogeneous effects of the program through 1st grade by estimating quantile treatment effects under endogeneity (IV-QTE) as well as various types of subgroup mean treatment effects and two-stage least squares treatment effects. We find that (the experimentally manipulated) Head Start attendance leads to large and statistically significant gains in cognitive achievement during the pre-school period and that the gains are largest at the bottom of the distribution. Once the children enter elementary school, the cognitive gains fade out for the full population, but importantly, cognitive gains persist through 1st grade for some Spanish speakers. These results provide strong evidence in favor of a compensatory model of the educational process. Additionally, our findings of large effects at the bottom are consistent with an interpretation that the relatively large gains in the well-studied Perry Preschool Program are in part due to the low baseline skills in the Perry study population. We find no evidence that the counterfactual care setting plays a large role in explaining the differences between the HSIS and Perry findings.

Does the environment have a causal effect on economic development? Using meteorological data, we reconstruct every country's exposure to the universe of tropical cyclones during 1950-2008. We exploit random within-country year-to-year variation in cyclone strikes to identify the causal effect of environmental disasters on long-run growth. We compare each country's growth rate to itself in the years immediately before and after exposure, accounting for the distribution of cyclones in preceding years. The data reject hypotheses that disasters stimulate growth or that short-run losses disappear following migrations or transfers of wealth. Instead, we find robust evidence that national incomes decline, relative to their pre-disaster trend, and do not recover within twenty years. Both rich and poor countries exhibit this response, with losses magnified in countries with less historical cyclone experience. Income losses arise from a small but persistent suppression of annual growth rates spread across the fifteen years following disaster, generating large and significant cumulative effects: a 90th percentile event reduces per capita incomes by 7.4% two decades later, effectively undoing 3.7 years of average development. The gradual nature of these losses render them inconspicuous to a casual observer, however simulations indicate that they have dramatic influence over the long-run development of countries that are endowed with regular or continuous exposure to disaster. Linking these results to projections of future cyclone activity, we estimate that under conservative discounting assumptions the present discounted cost of "business as usual" climate change is roughly $9.7 trillion larger than previously thought.

Copyright theorists often ask how incentives can be designed to create better books, movies, and art. But this is not the whole story. As the Roman satirist Martial pointed out two thousand years ago, markets routinely ignore good and even excellent works. The insight reminds us that incentives to find content are just as necessary as incentives to make it. Recent social science research explains why markets fail and how timely interventions can save deserving titles from oblivion. This article reviews society’s long struggle to fix the vagaries of search since the invention of literature. We build on this history to suggest policies for the emerging world of online media.

Homeric literature was produced and disseminated through direct interactions between audiences and authors. Though appealing in many ways, the process was agonizingly slow. By the 1st Century AD commercial publishers had moved to the modern model of charging readers above-cost prices to pay for search and marketing. Crucially, the new model was only sustainable so long as firms could suppress copying. We argue that Roman and early modern publishers developed remarkably successful self-help strategies to do this. However, their methods did little to suppress copying after the first edition. This seemingly modest defect made publishers profoundly risk averse. Ancient best-seller lists were invariably dominated by authors who had been dead for centuries.

Publishers’ self-help systems collapsed under a wave of piracy in the mid-17th Century. This led to the first modern copyright statutes. Crucially, the new laws extended protection beyond the first edition. This encouraged modern business models in which publishers gamble on a dozen titles for each that succeeds. The ensuing proliferation of titles helped fuel the Enlightenment. It also promoted a rich new ecosystem of search institutions including libraries, newspaper critics, and editors.

The Digital Age has changed everything. As copyright fades, the old institutions for finding titles are drying up. We explore several possible responses. First, society can shore up current publishing models by expanding copyright and technical protections. We argue that these methods cannot save book search but might be adequate for music and movies. Second, search engines could pay for editors. We argue that an on-line Digital Bookstore can suppress copyists long enough to fund reasonable search efforts. Finally, society can return to the Homeric pattern of harvesting advice directly from audiences. We explore various commercial and open source institutions for organizing the work.

For the past twenty years, large corporations have routinely developed and enforced industry-wide standards to address problems that are only distantly related to earning a profit. This includes writing detailed private regulations for environmental protection, national security, working conditions, and other topics formerly reserved to governments. At the same time, the US Supreme Court has said that the Sherman Act forbids any “extra-governmental agency” that “provides extra-judicial tribunals for the determination and punishment of violations.” This seems to ban enforceable rules. Despite this, many US policymakers continue to argue that private standards are efficient and desirable. Many corporations are sympathetic but fear legal liability and are reluctant to participate unless and until the law is clarified.

This article asks how existing law can be reformed to arrive at principled rules for deciding when private standards violate the Sherman Act. We begin with an historical account of recent private initiatives to regulate food processing, fisheries, forestry, and coffee production. We argue that these private rules are often just as effective – and burdensome – as government regulation. We then generalize from this evidence to explain when and how large corporations are able to impose their preferences through industry-wide standards. We also describe the politics that determines how large corporations use their power. We argue that the need to earn positive profit and defend market share frequently encourages – and sometimes forces – large companies to choose standards that please consumers. In these cases, consumers act as a shadow electorate that constrains private power in much the same way that real voters constrain elected officials. Finally, our examples show that big corporations often decide to share power with smaller rivals, suppliers, NGOs, and other stakeholders. We argue that these delegations are genuine and make private standards more accountable.

The article concludes by asking how current law can be reformed. We argue that the Sherman Act serves two goals. The first is economic efficiency. We argue that private standards advance this goal by addressing problems (“externalities”) that lack well-defined market prices. We argue that private bodies should be allowed to address such problems in the first instance knowing that government may later step in to change or supplement policy. The second goal is to protect democracy from private power. We argue that this danger is minimal so long as (a) market structure encourages corporations to make choices that please consumers and other shadow electorates, (b) the standard setting body represents a wide range of affected stakeholders, or (c) industry selects the prevailing standard from multiple competing proposals. Significantly, all of these tests can be determined from objective evidence without obscure metaphysical inquiries into when private power becomes “illegitimate” or “poses a threat” to democratic politics.

Do public service reforms improve citizen services? Over the last two decades both public-private partnerships and information and communication technologies have been promoted as tools for reforming service delivery in developing countries. However, observational studies of policies intended to promote these reform models are hindered by selection bias. Experimental evaluations, on the  other hand, can be limited in their potential for generalization to broader populations. In this study, I adopt a combined experimental and observational approach to evaluate the independent effects of privatization and computerization in an initiative to improve citizen services in the south Indian state of Karnataka. Through the use of a citizen survey and field experiment, I show that privatization of service delivery, combined with computerization, has a larger positive effect than computerization alone on a number of service quality measures, including the demand for, and size of, bribes from citizens. While private, computerized centers do not improve all facets of service delivery and, interestingly, do not engender higher levels of satisfaction from citizens, their effect on corruption in the service delivery process is substantial.

Constituency service is an important element of Indian legislator activity. Early interest in the importance of the personal vote in India paid particular attention to the relevance of Indian electoral institutions for promoting the supply of constituency service to Indian citizens. Yet analysts have paid little attention to the potential effects of other institutional characteristics, such as the major decentralization reforms of the 1990s, on the nature of politician-citizen interactions. In this paper I use survey evidence from citizens in the south Indian state of Karnataka to show that, despite nearly two decades of formal political and fiscal decentralization in the state, in the majority of cases citizens continue to rely on the assistance of state-level politicians to navigate the state bureaucracy rather than their local counterparts. In addition I find that party affiliation, rather than demographic characteristics such as gender or caste, plays a predominant role in shaping both whom citizens have asked for help in the past and who they expect they would ask for help in the future. These findings contrast with the literature on decentralization that emphasizes the importance of decentralization for increasing the representation of minority groups and highlights the important role of party politics in linking constituents to their representatives and the resources controlled by those representatives.