(1995), "Which Types of School Spending are Most Effective? New Evidence on the School Quality Debate", Discussion Paper 95-03, Department of Economics, UCSD.

The paper examines students' achievement in standardized math and science tests using the Longitudinal Study of American Youth. Most standard measures of school inputs, such as class size and teacher experience, bear no relation to student achievement. However, computer resources specifically devoted to the given subject are strongly related to student performance. Second, the teacher's field of study in some cases significantly affects outcomes. The hypothesis that public school expenditures are efficiently allocated across types of inputs is rejected. A cost-benefit analysis reveals that too much is being spent on reducing class size relative to spending on classroom computers.