This document presents a model for predicting (1) school district spending per pupil, and (2) the effect of alternative forms of school aid. Constrained maximization equations take account of (1) real expenditure per pupil; (2) real school taxes, income taxes, and nonschool property taxes per household; and (3) homeowner's proportion of property tax. The model shows that matching grants stimulate more local expenditure than the customary lump sums. Regardless, the impact on spending per pupil is proportional to the number of pupils per household--not to aid per pupil. To predict results of matching grants, States can judge by the response of spending per pupil to changes in costs over time. An incomplete version of the model accounted for some 80 percent of the variance in school expenditure among States during 1954-1956, and, unlike earlier models, gave results consistent from year to year. (Author)