Using rich data linking federal cases from arrest through to sentencing, we find that initial case and defendant characteristics, including arrest offense and criminal history, can explain most, but not all, of the large raw racial disparity in federal sentences. Across the sentence distribution, blacks receive sentences that are almost 10% longer than comparable whites arrested for the same crimes. Most of this disparity can be explained by prosecutors’ initial charging decisions, particularly the filing of charges carrying mandatory minimum sentences. Ceteris paribus, the odds of black arrestees facing such a charge are 1.75 times higher than those of white arrestees.

[go to paper][Data Appendix]
[Originally circulated as:   "Racial Disparity in Federal Criminal Charging and Its Sentencing Consequences"]

This Article presents new empirical evidence concerning the effects of United States v. Booker, which loosened the formerly mandatory U.S. Sentencing Guidelines, on racial disparities in federal criminal cases. Two serious limitations pervade existing empirical literature on sentencing disparities. First, studies focus on sentencing in isolation, controlling for the “presumptive sentence” or similar measures that themselves result from discretionary charging, plea-bargaining, and fact-finding processes. Any disparities in these earlier processes are excluded from the resulting sentence-disparity estimates. Our research has shown that this exclusion matters: pre-sentencing decision-making can have substantial sentence-disparity consequences. Second, existing studies have used loose causal inference methods that fail to disentangle the effects of sentencing-law changes, such as Booker, from surrounding events and trends.

In contrast, we use a dataset that traces cases from arrest to sentencing, allowing us to assess Booker’s effects on disparities in charging, plea-bargaining, and fact-finding as well as sentencing. We disentangle background trends by using a rigorous regression discontinuity-style design. Contrary to other studies (and in particular, the dramatic recent claims of the U.S. Sentencing Commission), we find no evidence that racial disparity has increased since Booker, much less because of Booker. Unexplained racial disparity remains persistent, but does not appear to have increased following the expansion of judicial discretion.

[go to paper]

In this Essay, Professors Starr and Rehavi respond to the U.S. Sentencing Commission's empirical staff's criticisms of their recent article, which found, contrary to the Commission's prior work, no evidence that racial disparity in sentences increased in response to United States v. Booker.  As Starr and Rehavi suggest, their differences with the Commission perhaps relate to differing objectives.  The Commission staff's reply expresses a lack of interest in identifying Booker's casual effect; in contrast, that is Starr and Rehavi's central objective.  In addition, Starr and Rehavi's approach also accounts for disparities arising throughout the post-arrest justice process, extending beyond the Commission's narrower focus on disparities in adherence to the Sentencing Guidelines.  Beyond these core disagreements, Starr and Rehavi point to several ways in which the reply's other criticisms inaccurately describe their claims, their methods, and the scope of their study's sample.

[go to paper]

Political ties and the need to cultivate support for nominations to higher office create a conflict of interest for U.S. attorneys and the prosecutors they supervise in political corruption cases. How severe is this problem? Contrary to previous research, prosecutors do not appear to bring weaker cases against opposition party defendants before elections; we find no measurable difference in conviction rates and some evidence that co-partisans received less favorable treatment in plea bargains and sentencing until recently.  However, we observe partisan differences in the timing of public corruption case filings that we attribute to the career incentives facing prosecutors. Relative to the president’s co-partisans, opposition defendants are more likely to be charged immediately before an election than afterward. We find a corresponding difference in case duration, suggesting prosecutors move more quickly to file cases against opposition partisans. These timing differences are associated with greater promotion rates to appointed political office.

[go to paper] [data appendix]

This paper provides new evidence on physician induced demand (PID). Using rich microdata on childbirth, we compare the treatment of physicians when they are patients with that of comparable non-physicians. We also exploit a unique institutional feature of California to determine how inducement varies with the obstetrician's financial incentive. Consistent with PID, physicians are almost 10 percent less likely to receive a C-section, with only a quarter of this effect attributable to differential sorting of patients to hospitals or obstetricians. Financial incentives have a large eect on C section probabilities for non-physicians, but physician-patients are relatively unaected. Physicians also have better health outcomes, suggesting overuse of C-sections adversely impacts patient health.

[go to paper] [Data Appendix] [Supplementary Tables]
[NBER working Paper Version]

The Trade Adjustment Assistance Reform Act of 2002 profoundly changed the nature and scope of U.S. policy for addressing dislocations of workers caused by trade. Although the program covers only a small (and peculiarly selected) fraction of the unemployed, these changes affect the incentives faced by many workers and may have far-reaching consequences in health and labor markets.

[go to paper]

More details to come.

Of necessity, many tax-price elasticities are calculated from tax return data. Survey data on families' charitable giving and reporting on their tax returns indicate that reporting is a significant part of tax-price elasticity estimates. Roughly a quarter of the tax-price elasticity estimated from tax returns represents changes in reporting as opposed to changes in actual donations. These results are consistent with a model in which individuals face unit costs of reporting in addition to any fixed cost of itemization.
[go to paper]

More details to come.

Recent papers have documented strong party effects consistent with policy divergence at the national level in the United States, but have generally failed to find such effects locally. These differences have generally been attributed to Tiebout sorting and government competition at the local level.  Using data on both roll call voting scores and fiscal policy outcomes at the state legislative level, I find large party divergence in roll call voting (nearly identical to that documented in Congress) and no measurable effects on fiscal policy outcomes (the outcome used in studies of local government party divergence).   Divergence in roll call voting does not produce divergence in policy outcomes.  These patterns hold even when roll call votes are restricted to those likely to have affected the fiscal policy of interest suggesting it is not simply a product of the different nature of the policy spaces at the local and national levels.  They are consistent with bargaining models in which the content of bills (and associated logrolling) are negotiated to secure marginal votes.