By David Wang LR_WangDavid
School Quality and Property Values
The American education system spans communities of extremely diverse populations, across many socio-economic and ethnic lines. It is likely that the successes of students in these communities hinge on the combination of the students’ innate intelligence, living environment, and school quality. Standard models suggest and many have provided empirical evidence that housing prices have a correlation to this school quality. However, since these factors could be interrelated, to isolate and identify the actual effects of school quality on property values requires careful consideration of student, house, and neighborhood-specific attributes. It is also important to note how school quality is measured. With the passage of the No Child Left Behind Act of 2001 and the shift away from using per pupil expenditures as a proxy for the quality of education, attention has been drawn to the use of standardized test scores as the marker of education quality.
Recent literature reconfirms the positive correlation of school performance on house prices, using basic hedonic models with added controls for test scores of local schools. However, in older papers, the authors have difficulties controlling for neighborhood characteristics that are correlated with the test scores and house prices. Black (1999) develops a new method for assessing school quality by using attendance district boundaries to account for neighborhood characteristics. This method allows her to compare school to school differences in test scores with house prices. Crone (2006) uses a model on a full unrestricted sample that allows for testing of house price and test score relationships on both a school and district level. In addition, he adapts Black’s boundary model to allow for this district level analysis. In contrast to Black, Crone argues that it is a district-wide educational quality, not individual school quality that affects house prices. Finally, Clapp, Nanda, and Ross (2007) also consider Black’s model, but instead use a time-based fixed effects model over the period from 1994 to 2004 to control for the neighborhood characteristics. Despite using different methods, all three papers agree that a positive correlation exists between school test scores and housing prices.
Black’s (1999) measurement of differences across attendance district boundaries enables the use of fixed effects in her model. This district boundary is the line that separates the respective attendance areas of schools. This line provides a discrete point at which standardized test scores should change. However, the line may run through continuous neighborhoods, allowing Black to compare any sudden jump in test scores with houses that are situated in similar neighborhoods. By using dummy variables to account specifically for the districts, Black avoids the omitted variable biases of property taxes, public goods, and neighborhood characteristics. Using MEAP testing data from Massachusetts elementary schools, Black focuses on the fourth grade level. Under the basic, unrestricted model, she must control separately for house level characteristics, distance from the CBD, in addition to other school quality characteristics, such as per-pupil expenditures. She finds that per-pupil expenditure is positively correlated with house prices while higher pupil/teachers ratio is negatively correlated with house prices. Nevertheless, the crux of the problem involves the unobservable characteristics of a neighborhood. Black examines different subsets of her data, restricting the samples to houses nearer and nearer the boundary and increasing the probability that the houses on opposite sides of the boundary differ in only the elementary school quality. Her study reveals that if neighborhood characteristics are not carefully controlled, the marginal value of school quality as measured by test scores on housing prices will be overestimated. Black concludes that parents will pay higher house prices for better schools, but does not examine whether there exists a district level effect of school quality on prices.
Newer researchers incorporate Black’s boundary model and conclusions as supplements to their models. However, unlike Black, Crone (2006) argues that home buyers actually value local public education at the district level rather than the neighborhood school level. Using fifth and eleventh grade Pennsylvania System of School Assessment (PSSA) data from Montgomery County, Crone makes findings that differ from Black’s. While Black argues that differences on an individual school basis affect home prices, Crone claims otherwise. Crone argues that for fifth grade test scores, differences are only significant on the district-level. In fact, he finds that fifth grade test scores are better predictors of house prices than eleventh grade scores. Perhaps this discrepancy could be attributed to the location of families with young children and the subsequent lack of relocation as the children grow up. Crone’s differing results could also be due to his use of the full sample rather than a boundary restricted sample. Crone’s more comprehensive dataset allows him to make district level regressions, while Black’s dataset is restricted to individual schools.
Crone’s study also provides additional factors that may affect school quality and thus house prices. For example, class size is not significant at the elementary school level, but it makes a significant difference at the high school level. By incorporating this measurement into the main model, Crone reduces the chance of omitted variable bias from Black’s neighborhood fixed effects model. The neighborhood fixed effects do not account for differences in the schools, such as per-pupil expenditure or class size, of which the latter was not included in Black’s model, which only seeks to explain the impact of school quality differences. As an additional test, Crone uses Black’s boundary method to estimate the effect of both school and district test scores on housing prices. He finds that with this smaller sample, there is no significant coefficient on fifth grade scores, further conflicting with the results given by Black. However, on the high school level, the results become more significant with the smaller sample with boundary dummies. This result differs from the result when controlling for detailed characteristics in the model with a full sample. Finally, Crone’s study also finds that per-pupil expenditures do not affect the house prices above their effect on student test scores or achievement. Overall, Crone brings the conclusion that school district quality should be considered over the quality of individual schools when determining the effect on house prices.
Clapp, Nanda, Ross (2007) introduce a twist on the examination of test scores and housing prices by suggesting that the quality of school districts is a function of both test scores and demographic composition. Since people tend to use the most accessible signals to judge school quality, they often rely on the demographics of a school, which are very visible. Thus, Clapp examines the significance of the test score and demographic composition effects on house prices. Like Black and Crone, Clapp also finds a statistically significant, though very small, effect of test scores on house values. Clapp also agrees with Black’s finding that failing to control for unobservable characteristics in the neighborhood leads to overstatement of the test score effect. However, Clapp extends this argument by also including the effects of race percentages on home prices. He finds that an increase in percent African-American and percent Hispanic leads to a decline in property values. Nevertheless, over this time period, people appear to be placing more importance on test scores and less on demographics when evaluating school quality.
Of the three papers, Clapp’s is the only one to use a time-based fixed effects model. While the studies of Black and Crone use averages of a single three-year period and district boundaries as fixed effects, Clapp instead exploits the cross time variation in the 1994-2004 panel data to separate school attributes from neighborhood quality. Clapp also incorporates additional neighborhood fixed effects by comparing sales occurring in different neighborhoods, but the same school district. This combination of time variation based identification strategy and also neighborhood fixed effects should yield more accurate estimates than either strategy alone. However, one downside to Clapp’s method that does not appear with Black’s or Crone’s is the possibility of unobservable changes over the sample’s long time period.
Through these three papers, we see a wide variety of techniques used to analyze public school test scores and house prices, yet arrive at the conclusion that standardized test scores do impact housing prices. In these papers, however, we assume that district boundaries are fixed and that students must attend schools in their attendance zone. With the rise of charter schools, students no longer are limited to the public schools near their homes. As the population of students going to charter schools increases, we may begin to see a declining importance of neighborhoods under the models discussed in the three papers reviewed here. It may be worthwhile to examine the effects of charter schools on home prices, both in the area of the school and of the students.
Black, Sandra E. “Do Better Schools Matter? Parental Valuation of Elementary Education*.” Quarterly Journal of Economics 114.2 (1999): 577-99. Web. 7 Feb. 2013.
Clapp, John M., Anupam Nanda, and Stephen L. Ross. “Which School Attributes Matter? The Influence of School District Performance and Demographic Composition on Property Values.” Journal of Urban Economics 63.2 (2008): 451-66. Web. 7 Feb. 2013.
Crone, Theodore M. “Capitalization of the Quality of Public Schools: What Do Home Buyers Value?” Working Paper Series, Federal Reserve Bank of Philadelphia (2006): n. pag. Statistical Insight [ProQuest]. Web. 7 Feb. 2013.