School quality and Massachusetts enrollment shifts in the context of tax limitations

New England Economic Review, July-August, 1998 by Katharine L. Bradbury, Karl E. Case, Christopher J. Mayer

These measures of initial cuts are unrelated to net enrollment changes a decade later (column 2). But Proposition 2 1/2's limits on revenue growth affected increasing numbers of communities toward the end of the 1980s, and a variable measuring the stringency of the growth limit does help to explain enrollment changes from 1990 to 1995.(16) Specifically, communities at their levy limits in 1989 experienced smaller enrollment increases between 1990 and 1995, 3 to 4 percentage points less, than did communities less constrained by Proposition 2 1/2.(17) Whether by cutting the quality of schools in ways not captured by the test score measure, reducing the quality of nonschool public services, or inhibiting community flexibility more generally, Proposition 2 1/2 made constrained communities relatively less attractive to families with children, both in the early 1980s and the early 1990s.(18)

Land use data for Massachusetts communities are available only for 1984. A 1984 measure of open and residential land relative to the 1980 housing stock ("developable" land) was associated with net enrollment growth in both periods.(19) Several explanations are possible for the finding of a larger estimated effect of developable land in the 1980s. First, space to accommodate added families was probably more at issue in the early 1980s than in the early 1990s. Two pieces of evidence suggest that the number of households was expanding more in the earlier period despite the fact that enrollments were shrinking: (1) The state experienced net in-migration between 1980 and 1985 while population moved out, on net, in the early 1990s. (2) based on permits, more housing was constructed in Massachusetts in the 1980-85 period than in the 1990-95 period. A second explanation has to do with measurement: The measure of developable land may relate more strongly to growth in the 1980s because the year land use was tallied, 1984, obviously precedes the 1990-95 period by six-plus years. To the degree that the mid-decade 1980s housing boom elicited residential development on open land, this 1984 measure may not reliably represent differences among communities in the availability of land as the 1990-95 period opened.(20)

Other Local Amenities

Because of peer effects in local schools as well as more general neighborhood effects, families with children might be attracted to communities whose residents have higher incomes or are more highly educated. Location is another key local characteristic; proximity to jobs, retail services, and other economic activity would be viewed as a plus by most households, although some household types may be willing to pay more for increased access than others.

The regressions reported in columns 4 and 5 of Table 9 add measures of these other local amenities to the analysis. In both periods, communities with higher-income residents at the beginning of the period realized greater net enrollment gains. By contrast, the percentage of residents with a college or higher education is unrelated to enrollment changes.(21) An indicator of geographic location is also included in the equations. Coefficient estimates on a dummy variable for the Boston metropolitan area imply that communities near Boston experienced larger net enrollment declines in the 1980s and smaller increases in the 1990s than communities outside the Boston area.(22) The effect of location was somewhat stronger in the 1990s than in the 1980s.