As discussed in section II, the level of urbanization increases the efficiency of
employment matches, decreasing the skills variation of principals. The final urbanization
characteristic, the number of school districts by metropolitan area, is modeled to
determine match efficiency through Tiebout competition. Depending on the number of
alternatives, residential mobility could give rise to inefficiently administered school
districts declining in student population.5
The zi covariates, district student size and percent non-white, are included to
capture possible heterogeneity within school districts. Both variables should be
positively correlated with the dependent variable. The mean skills level by district is also
included as a covariate to account for whether the mean administrator quality the school
district seeks also determines skills heterogeneity.
Metropolitan area density, population size and the proportion over age twenty-
five with at least a master’s are taken from the 1990 US Census. Population density is
calculated as total population in the metropolitan area divided by total land area. The
1997 Census of Governments provides data on the number of school districts by county,
which were aggregated to the metropolitan area. Among the school district
characteristics, the number of students in 1990 (district size) and the percent non-white
are taken from the National Center for Education Statistics, which also produced the
Schools and Staffing Survey (SASS). Mean teaching experience is calculated by school
district only for the principals sampled in this study and comes from the SASS data.
5 Hoxby (2000) finds Tiebout competition induces greater education productivity among school districts.