Economists have strong theoretical predictions about how in-kind transfers, such as providing vouchers for food, impact consumption. Despite the prominence of the theory, there is little empirical work on responses to in-kind transfers, and most existing work fails to support the canonical theoretical model. We employ difference-indifference methods to estimate the impact of program introduction on food spending. Consistent with predictions, we find that food stamps reduce out-of-pocket food spending and increase overall food expenditures. We also find that households are inframarginal and respond similarly to one dollar in cash income and one dollar in food stamps.

This paper uses a forecasting model to estimate the need for, supply of, and shortage of doctors, nurses, and midwives in thirty-nine African countries for 2015, the target date of the United Nations Millennium Development Goals. We forecast that thirty-one countries will experience needs-based shortages of doctors, nurses, and midwives, totaling approximately 800,000 health professionals. We estimate the additional annual wage bill required to eliminate the shortage at about $2.6 billion (2007 $US)—more than 2.5 times current wage-bill projections for 2015. We illustrate how changes in workforce mix can reduce this cost, and we discuss policy implications of our results.

Nonrandom assignment of students to teachers can bias value-added estimates of teachers’ causal effects. Rothstein (2008, 2010) shows that typical value-added models indicate large counterfactual effects of fifthgrade teachers on students’ fourth-grade learning, indicating that classroom assignments are far from random.This article quantifies the resulting biases in estimates of fifth-grade teachers’ causal effects from several valueadded models, under varying assumptions about the assignment process. If assignments are assumed to depend only on observables, the most commonly used specifications are subject to important bias, but other feasible specifications are nearly free of bias. I also consider the case in which assignments depend on unobserved variables. I use the across-classroom variance of observables to calibrate several models of the sorting process. Results indicate that even the best feasible value-added models may be substantially biased, with the magnitude of the bias depending on the amount of information available for use in classroom assignments.

We performed a pilot study examining the patterns of recovery from severe mental illness in a model integrated service delivery system using measures from the Milestones of Recovery Scale (MORS), a valid and reliable measure of recovery outcomes which ranges from 1 to 8 (8 levels). For purposes of presentation, we constructed an aggregate MORS (6 levels) where the levels are described as follows: (1) extreme risk; (2) unengaged, poorly selfcoordinating; (3) engaged, poorly self-coordinating; (4) coping and rehabilitating; (5) early recovery, and (6) self reliant. We analyzed MORS data on individuals followed over time from The Village in Long Beach, California (658 observations). Using Markov Chains, we estimated origindestination transition probabilities, simulating recovery outcomes for 100 months. Our models suggest that after 12 months only 8% of ‘‘extreme risk’’ clients remain such. Over 40% have moved to ‘‘engaged, poorly self-coordinating.’’ After 2 years, almost half of the initial ‘‘extreme Risk’’ clients are ‘‘coping/rehabilitating’’, ‘‘early recovery’’ or ‘‘Self reliant.’’ Most gains occur within 2 years.

Data from college admissions tests can provide a valuable measure of student achievement, but the non-representativeness of test-takers is an important concern. We examine selectivity bias in both state-level and school-level SAT and ACT averages. The degree of selectivity may differ importantly across and within schools, and across and within states. To identify within-state selectivity, we use a control function approach that conditions on scores from a representative test. Estimates indicate strong selectivity of test-takers in “ACT states,” where most college-bound students take the ACT, and much less selectivity in SAT states. To identify within- and between-school selectivity, we take advantage of a policy reform in Illinois that made taking the ACT a graduation requirement. Estimates based on this policy change indicate substantial positive selection into test participation both across and within schools. Despite this, school-level averages of observed scores are extremely highly correlated with average latent scores, as across-school variation in sample selectivity is small relative to the underlying signal. As a result, in most contexts the use of observed school mean test scores in place of latent means understates the degree of between-school variation in achievement but is otherwise unlikely to lead to misleading conclusions.

Throughout the world, interest in gauging learning outcomes at all levels of education has grown considerably over the past decade. In higher education, measuring “learning outcomes” is viewed by many stakeholders as a relatively new method to judge the “value added” of colleges and universities. The potential to accurately measure learning gains is also viewed as a diagnostic tool for institutional self-improvement. This essay compares the methodology and potential uses of three tools for measuring learning outcomes: the Collegiate Learning Assessment (CLA), the National Survey of Student Engagement (NSSE), and the University of California’s Undergraduate Experience Survey (UCUES). In addition, we examine UCUES 2008 responses of seniors who entered as freshmen on six of the educational outcomes self-reports: analytical and critical thinking skills, writing skills, reading and comprehension skills, oral presentation skills, quantitative skills, and skills in a particular field of study. This initial analysis shows that campus-wide assessments of learning outcomes are generally not valid indicators of learning outcomes, and that self-reported gains at the level of the major are perhaps the best indicator we have, thus far, for assessing the value-added effects of a student’s academic experience at a major research university. UCUES appears the better approach for assessing and reporting learning outcomes. This is because UCUES offers more extensive academic engagement data as well as a much wider range of demographic and institutional data, and therefore an unprecedented opportunity to advance our understanding of the nature of self-reported learning outcomes in higher education, and the extent to which these reports can contribute as indirect but valid measures of positive educational outcomes. At the same time, the apparent differences in learning outcomes across the undergraduate campuses of the University of California without controls for campus differences in composition illustrates some of the limitations of self-reported data.