Working paper #51
Yüklə 326,23 Kb. Pdf görüntüsü
|
17
Model 1 in Table 4 indicates that a one dollar increase in foundation revenue and non- foundation revenue lead to $.376 and $.283 increase in FB, respectively. The influence of foundation revenue on fund balances intensifies, after controlling for the enrollment and resource allocation variables, as shown in Model 5.
As expected, district enrollment growth also consistently exerts a significant positive impact on district fund balances. Gaining one student increases fund balances by $.03 per total district pupil. This implies that for growing districts the marginal revenue gain associated with an additional student exceeds the marginal cost. Conversely, for declining enrollment districts, revenues decline faster than costs, causing district fund balances to decline. The results for the school choice variables are interesting. In all specifications, as expected, a net inflow of inter-district student transfers significantly increases district fund balances, and correspondingly net out-flows of students lower fund balances. By comparison, the results for districts’ loss of students to charter schools are more tentative. While significant in the parsimonious models (Models 3 and 4 in Table 4), %Charter is only significant at the 90% level in the full model, after the resource allocation variables are introduced (Model 5). Below we test for the possibility that very high and sustained levels of charter penetration have differential impacts on district finances that are not captured in our linear specification of %Charter. It is noteworthy that the point estimate on Enrollment falls by 44 percent when the charter school and inter-district choice variables are added in Model 3. This implies that 44 percent of the enrollment effect is explained by charters and inter-district choice. It also implies that since Enrollment is included with the school choice variables, the point estimates on
18
their effects pass through enrollment. Student characteristics also influence district fund balances. As expected, the percent of a district’s students who are eligible for special education service is strongly related to district fund balances. A one percentage increase in a district’s students receiving special education services in Model 5 generates roughly a $43 decline in district fund balance per student. As noted above (footnote 2), under Michigan’s special education funding arrangements, special education students typically represent a financial burden to local districts, and the more costly a student’s disability, the larger the burden. 13
Meanwhile, increases in the percent of students who are poor (%FRL) do not have a similar negative impact on district fund balances. The percent of students who are eligible for free and reduced priced lunch is positively related to district fund balances, although this relationship is only marginally significant. 14 This implies that for Michigan districts as a whole, the additional revenue they receive to serve low-income students roughly corresponds to their additional spending for these students, producing no significant residual impact on fund balances. In Model 5, three of the five district resource allocation variables are significant (P-T ratio, Teacher_Salary, and %Admin) each with the expected signs. An increase in a district’s pupil-teacher ratio by one student increases its fund balance by $80 per pupil. Meanwhile, a
13 The portion of special education costs that are borne by local districts and charter schools varies across intermediate school districts (mostly county-level agencies) in Michigan, because ISDs differ in the extent to which they provide supplemental special education funding to the local units. 14 We explored the possibility that the impact of student poverty on fund balances is nonlinear, including specifications testing for the possibility of differential budget impacts in districts with very high thresholds of student poverty (e.g., %FRL = 70%, 80% or 90%). In none of these specifications was the measure of student poverty statistically significant.
19
$1000 increase in average teacher salaries decreases fund balances by $29 per pupil. A one percentage point increase in the administrative spending (as a share of total general fund spending) lowers district fund balances by $112 per pupil. Two of the district resource allocation variables, %Health and %Purchases, are not significantly related to district fund balances. These results are noteworthy in view of the fact that within the Michigan policy context both are associated with steps (cutting expenditures on employee health benefits and contracting service provision to private vendors, respectively) that have been widely advocated for local districts in fiscal distress (e.g.,Washburne & Jahr, 2007; LaFaive, 2007 ). Whatever the merits of these strategies as emergency measures for district in fiscal distress, our results fail to indicate that they are systematically related to districts’ fiscal health as measured by their fund balances. 15
Americans are more likely to be in financial distress, as indicated by Figure 3, our results in Table 4 suggest a more nuanced relationship. Districts with high concentrations of African American students are much more likely to be subject to intense charter school penetration, to lose students to inter-district choice, and to have higher concentrations of students with disabilities. 16 Once these and other factors are controlled, however, districts’ level of financial distress is not significantly related to their racial composition as indicated by the %Black variable in Model 6 of Table 4. Moreover the results for other variables are highly stable with the inclusion of %Black in the model.
15
models. MPSERS is a state defined-benefit pension system. The system’s eligibility and benefit provisions are established at the state level, and it is funded by local districts through a uniform statewide contribution rate set annually by the state and assessed against districts’ payroll spending. MPSERS contributions were not significant in any of our models and its inclusion did not change the results for any other variables in the model.
16
and %SpecEd (0.222). The correlations over the entire 1995-2012 period were: %Charter (0.601), %Net_IDC (- .052), and %SpecEd (0.193). %Black is measured by head count data. Yüklə 326,23 Kb. Dostları ilə paylaş:
|