1 Do Households Value Competition Among Public Schools? A Natural Experiment H. Spencer Banzhaf1 Georgia State University Garima Bhalla2 July 2007 1 Corresponding Author. Associate Professor. Department of Economics, PO Box 3992, Atlanta, GA 30302. 404-651-6981. [email protected]. http://aysps.gsu.edu/people/BanzhafS.html. 2 Former Masters student, Public Policy Institute, Georgetown University.2Do Households Value Competition Among Public Schools? A Natural Experiment H. Spencer Banzhaf Georgia State University Garima Bhalla July 2007 Abstract: Perennial discontent with public education is a constant source of proposed reforms. One such reform is school choice. While experience with outright voucher programs is very limited, the Tiebout mechanism suggests that school districts have always operated within a choice context. If this Tiebout competition works, municipalities with more districts should have more effective schools. This paper tests whether households expect these effects and whether they value them. It does so by studying the effect on real estate prices of a natural experiment provided by the surprise breakup of the Los Angeles Unified School District into 11 minidistricts in April 2000. We find consistent evidence that households valued this decentralization, with a 1-4 percentage point increase in housing values in the LAUSD area over pre-existing trends, compared to control districts. The result is robust to a variety of specifications and appears to be fairly homogenous within the LAUSD area. JEL Codes: I2, R21.3Do Households Value Competition Among Public Schools? A Natural Experiment3 H. Spencer Banzhaf Georgia State University Garima Bhalla May 2007 Perennial discontent with the quality of public education is a constant source of new educational reforms. One old idea, only recently being experimented with, is school choice in which households receive free public schooling or vouchers for private schools.4 Theoretically, by increasing competition, this expansion of choices might force public schools to improve, while also allowing parents to take better advantage of private offerings. While such experiments are very limited, the Tiebout mechanism, through which households “vote with their feet” by moving to communities offering their preferred bundle of public goods and taxes, suggests that public school districts have always operated within a kind of choice context. In this case, the choice (and competition) is among public school districts (Borland and Howsen 1992, Hoxby 2000, Hanushek and Rivkin 2003). If this Tiebout competition works, municipalities with more districts, ceteris paribus, should have more effective schools. Hoxby (2000) has noted that school quality and the number/size of districts may be endogenous. However, using an instrumental variables regression, she finds that greater Tiebout school choice does raise achievement, while lowering expenditures. Moreover, she finds that parents appear to acknowledge this improvement, as evidenced by lower enrollment in private schools in municipalities with more districts. However, the results of this research, and in particular the IV strategy she employs, remain controversial (Rothstein 2006a, Hoxby 2005). This paper employs an alternative strategy to evaluate parental valuation of greater 3 We thank Rajashri Chakrabarti, Jason Fletcher, Jesse Rothstein, David Sjoquist, Todd Sorensen, and participants of a session at the 2006 Southern Economic Association meetings for valuable comments. 4 Friedman (1962) is the classic reference. For overviews on voucher programs and school choice more generally, see Hoxby (2003a), Ladd (2002), and Neal (2002). See Cullen et al. (2005), Geller et al. (2006), Hoxby (2003b), Peterson et al. (2003), Rouse (1998), and Sander (1999) for recent empirical evaluations of specific cases.4school choice: the natural experiment presented when, in April 2000, the Los Angeles Unified School District (LAUSD), the nation’s second largest, split into 11 subdistricts. Admittedly, this breakup was only partial, insofar as official authority remained vested in a single school board. Nevertheless, it devolved substantial power to a more local level, with the creation of 11 new superintendents where there had been one and the transfer of some 500 staff from the central office. Thus, the breakup qualitatively increased the range of school choices available to households. We test for parental valuation of this breakup by studying the response of real estate markets to the announcement of the LAUSD breakup. In particular, we use a difference-in-difference hedonic methodology comparing changes in real estate prices before and after the announcement within the LAUSD area and in other LA-area districts just outside it. As an alternative to a discrete announcement effect, we also consider an index of newspaper articles on the breakup as a proxy for expectations. LAUSD real estate prices would be expected to respond positively to news of a breakup if (i) households value high quality schools and if (ii) they perceive the breakup as a move that would enhance school performance. Previous research has shown that households do value high-quality public schools, as revealed by higher real estate prices in school districts with better test scores (e.g. Black 1999). School quality is a function of many variables, including the school’s peer group as well the effectiveness of the teachers and curriculum. Rothstein (2006b) has recently cast doubt on parents’ ability to judge effective schools and/or their valuation of this aspect of quality. However, Figlio and Lucas (2004), using a methodology similar to ours, find that real estate prices respond to clear informational shocks about effectiveness, such as the school report card issued by the State of Florida. Our work speaks to a combination of the two issues. We find that households appear to have valued the increase choice provided by the breakup. Against a background of 9% housing price appreciation in the LA area during 1999-2001, LAUSD prices increased by 1 to 4 percentage points, relative to nearby control districts, during a sixth-month window between the initial proposal of the breakup in October 1999 and its formal adoption in April 2000. This result is robust to a variety of specifications, from simple parametric models using the announcement dates