Katrina's Children: Evidence on the Structure of Peer Effects From Hurricane Evacuees
September 09, 2009
National Bureau of Economic Research (Working Paper Series)
Scott Imberman, Adriana Kugler, and Bruce Sacerdote
This complex but intriguing paper asks to what extent did the arrival of Hurricane Katrina (and Hurricane Rita) student evacuees adversely affect the academic performance and behavior of “native” students in the schools into which they transferred? Analysts examined 2003-2007 student-level data from the Houston Independent School District and districts not affected by the storm in the state of Louisiana (they excluded six districts, four affected by Katrina and two by Rita, as well as schools outside those districts with more than 70 percent evacuees, since they were likely in or near an affected area), which collectively took in roughly 200,000 students after the August and September 2005 storms. They found that, on average, an inverse relationship between the number of evacuees and Houston elementary math test scores and Louisiana state secondary reading test scores. In other words, more evacuees meant a bigger drop in scores in those subjects. They also found strong evidence in Louisiana’s evacuee-receiving districts that the arrival of low-performing evacuees hurt native children’s scores in all quartiles; those most hurt by the presence of low performing evacuees were high-performing students, while low-performing natives were least hurt by the arrival of low-performing peers. The influx of evacuees also increased absenteeism and disciplinary problems of the native students--which analysts term the “bad apple” effect. After conducting a number of mini experiments to test the robustness of their findings, they largely ruled out various alternative explanations for the drop in scores, such as a reduction in resources. And since the vast majority of Katrina evacuees were unable to select their schools (some were shipped out on buses without knowledge of their destination), the analysts believe this “natural experiment” helps to lessen the selection bias that often occurs in studies of peer effects. The study brims with other interesting data relevant to peer impact on student performance, discipline, and attendance. You can find it here (for a small fee).