Robert McMillan, University of Toronto

Abstract: This paper develops a framework that allows us, for the first time, to compare the cost effectiveness of incentive-based education reforms with other widely-discussed alternative policies. Central to the cost-effectiveness calculation, we propose an empirical strategy to separate the overall impact of teachers on student achievement into an incentive-varying component – labeled teacher effort – and incentive-invariant teacher ability while imposing minimal assumptions. The strategy draws on exogenous variation in the incentive strength of a well-known federal accountability scheme and rich administrative data covering all public school students in North Carolina over time. Our results from identifying contemporaneous teacher effort and ability separately indicate that a one standard deviation increase in ability is equivalent to 18 percent of a standard deviation increase in student test scores, versus a 5 percent increase in scores for a one standard deviation increase in effort. Further, the overall effect of teachers (measured by value-added) is increasing in accountability incentive strength. We then implement a structural estimation procedure to uncover the persistence of ability and effort, showing that effort affects future scores positively, though by less than ability. Combining these estimates with a model of school decision-making allows us to place incentive-based reforms on a common footing with a range of alternative education policies. For illustration, we show that incentive-based reforms can be more cost-effective than reforms that target teacher ability as a means of improving average student performance.

Friday, Dec. 1, 2017
3:30pm – 5:00pm
Fronczak 444