ABSTRACT
This paper analyzes how changes in school expenditures affect dropout rates based on data from 466 school districts in New York during the 2003/04 to the 2007/08 school years. Past traditional regression approaches show mixed results in part because school expenditures are likely endogenous, so that one cannot disentangle cause and effect. The regression discontinuity design used in this study isolates exogenous variation in school expenditures per pupil by comparing school districts where budget referenda passed and failed by narrow margins. The results indicate that increases in school expenditures reduce New York State dropout rates.
Acknowledgements
The authors thank Carmen Carrion-Flores, Ronald Ehrenberg, Alfonso Flores-Lagunes, David Slichter, David Slusky, two anonymous referees, as well as Colin Green, the editor of this journal, for valuable comments and suggestions.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 It is worth mentioning that Hanushek does not argue that increasing school expenditures makes no difference. He claims additional school spending does not make an apparent difference on average, and expenditure increases are likely to raise school/student performance only if they are related to performance incentives for schools and students (Hanushek Citation1989; Levin Citation1997).
2 More detailed information for the New York State school budget process is available in ‘New York State School Aid Budget Process’ at The New York State School Finance Reform Archive (http://finance.tc-library.org/Content.asp?uid=8356), and the following explanation for the budget process is largely based on this web-based article. Also see Ehrenberg et al. (Citation2004), Feld and Grossman (Citation1984), and Romer, Rosenthal, and Munley (Citation1992), each of which describe the New York State school budget referenda voting process.
4 The New York State School Finance Reform Archive provides a comprehensive description which is available at http://finance.tc-library.org/.
5 Chart A2.1, Education at a Glance 2014 OECD Indicators (www.oecd.org/edu/Education-at-a-Glance-2014.pdf).
6 NCLB requires states, school districts, and schools to report performance measures for their overall student population. The reported performance measures include standardized test scores, dropout rates and graduation rates.
7 Another alternative method for calculating dropout rates is using enrollment counts. But in this case there are numerous inconsistencies because student mobility cannot be fully controlled.
8 The results for these robustness checks are presented in Appendix A (online supplemental file).
9 According to the AIC, the most suitable model is the quadratic one. In the 12th grade case in , for instance, the AIC value for the cubic model is 9394.37, for the quadratic model it is 9393.25, for the linear model with the interaction it is 9403.00, and for the simple linear model it is 9355.15. The cases of other grades show very similar patterns. Although the cubic model has the smallest AIC value, the difference between it and the quadratic model is quite small.
10 State law allows school districts to raise the upper age limit from 16 to 17 for the students in their district. The New York City, Buffalo, and Brockport school districts raised their compulsory age to 17.
11 State Education Department, Fiscal Analysis and Research Unit. Available at www.oms.nysed.gov.