Determinants of local school spending: Some consistent estimates

ConclusionThe traditional median voter model predicts that median income and taxprice are the primary determinants of local school spending levels. It further implies that lump-sum grants wills be utilized like private income, and thus income and grant coefficients should be the same. Our comprehensive search technique shows that, excluding equalized assessed value, these predictions hold up quite well over 2048 specifications. Income and taxprice elasticities are reasonable and tightly distributed, with income and blockgrant effects of the same general size. As note above, any hint of flypaper effect is completely dependent on the inclusion of equalized assessed value in the regression equation. Since the state of New York has annual district level tax levy elections, it would be hard to rationalize bureaucratic power or agenda control as capable of producing flypaper-type effects, so this absence is not surprising. Future research should (1) Investigate the specification sensitivity of flypaper effects with other datasets and wealth variables; and (2) Examine election and tax rules for the possibility of budget-maximizing actions by school officials. Finally, the consistently positive effects of private schooling, % nonwhite and % in poverty, suggest that future work should attempt to model non-linear income effects.