14
is a vector of two variables reflecting the share of a district’s resident students who participate in
each of Michigan’s two choice policies . It includes %Charter--the percent of students residing
in district i who attend charter schools--and %Net_IDC, the net gain or loss of students to inter-
district school choice as a percentage of students residing in district i.
11
We expect increases in
charter penetration to decrease district fund balances, and a net inflow of out-of-district students
to increase fund balances. (Conversely, net outflows of students through inter-district choice are
expected, like the loss of students to charter schools, to decrease fund balances.)
StudentCharacteristics
it
captures the net influence of a vector of student attributes that
are associated with added categorical funding, as well as additional costs of service provision.
Within the model specified by Equation 1, these variables can be interpreted as the extent to
which the added funding covers the service provision costs for high-cost students.
StudentCharacteristics
it
includes the percent of students who receive special education
services, %SpecEd, and the percent of students eligible for free and reduced-priced
lunch, %FRL.
12
For both variables, a negative sign on their estimated coefficient implies that the
additional revenue associated with these special-need students is less than the additional
expenditures for services they receive, whereas a positive implies additional revenues exceed
additional expenditures.
ResourceAllocation
it
is a vector of five variables reflecting key local resource allocation
decisions by district i in year t. Unlike the other variables in Equation 1, which are entirely or
largely outside local districts’ control, the resource allocation variables fall squarely within the
domain of local district decision makers. We expect increases in average teacher salary,
11
Specifically,
%Net_IDC = (incoming out-of-district students minus resident students enrolling in other districts)
as a percentage of students residing in district i.
12
Special education enrollment is measured by individual education plans (IEPs).
15
Teacher_Salary, to have a negative impact on fund balances. Increases in class size represent an
important option to reduce spending. We consequently expect increases in a district’s average
pupil-teacher ratio, P-T ratio, to have a positive impact on fund balances. High levels of
administrative spending are often taken as indicators of inefficiency in school operations,
although empirical support for this view is limited (Brewer, 1996). We test the hypothesis that
administrative “bloat” is associated with declines in district fund balances with %Admin, the
percent of spending devoted to administration.
The rising cost of employee health insurance has squeezed school district budgets across
the nation in recent years and prompted a variety of local strategies to curtail these costs through
changes in coverage and payments. We include the percent of spending devoted to employee
health insurance, %Health, with the expectation that it is negatively related to district fund
balances. Finally, private contracting of services, especially support services such as
transportation, custodial, food service, has been advocated as a promising strategy for local
districts to lower costs and diminish fiscal stress (LaFaive, 2007). Private contracting by
Michigan school districts has increased over the past decade (Holman & Fryzelka, 2014).
We
test this hypothesis with %Purchases, the share of spending devoted to purchased services, with
the expectation that it is positively related to fund balances.
θi is the time-constant district fixed effect, which picks up all unobserved characteristics
of a school district that are stable over time, including historical differences in local cost of living
or other geographical or structural influences on district revenues or spending that influence fund
balances. The idiosyncratic error that changes across time for each district is u
it
. All Equation 1
variables which are expressed in monetary units ( FB, Foundation, Nonfoundation, and
16
Teacher_Salary) are converted to real 2012 dollars using the U.S. Department of Commerce’s
GDP price deflator for state and local government purchases.
Our panel dataset covers the years 1995 to 2012, with the exception of the two school
choice variables (%Charter and %Net_IDC) which are only available for the years 2001 to 2012.
We estimate our models that do not include the choice variables over the entire 1995-2012
period, and those that include the choice variables over 2001 to 2012. We estimate Equation 1
for all Michigan school districts with enrollment of at least 100 students.
The data were assembled from two main sources: the Michigan Department of Education
(MDE) and the State of Michigan’s Center for Educational Performance and Information (CEPI).
MDE was the source for data on district foundations, fund balances (Bulletin 1011), and district
enrollment, average salaries and pupil-teacher ratios (Bulletin 1014). All other district financial
data came from CEPI’s Financial Information Database. CEPI was also source for the school
choice and student demographic variables.
RESULTS
Table 4 presents the estimates of the district fund balance models. Models 1-5 in Table 4
introduce the variables specified in Equation 1 sequentially. Model 5 is the full model.
In all specifications, both revenue variables have highly significant positive impacts on
district fund balances. As districts receive more revenue, their fund balances increase.
Moreover, the estimated coefficients indicate that an additional dollar of discretionary foundation
revenue is equivalent to roughly two dollars of non-foundation revenue. This is intuitively
plausible since non-foundation revenue is primarily categorical funding which comes with
corresponding expenditure requirements.
[Table 4 about here]
Dostları ilə paylaş: |