Spending in Lean Times: School-Level Budget Allocations During the Great Recession in Texas

ABSTRACT

The intention of this paper is to add to existing knowledge of how building-level spending is prioritized toward horizontal and vertical equity during severe economic downturns. Using a sample of all public schools in Texas during the Great Recession, we examine how schools undergoing the greatest spending reductions reallocated their spending on academic programs. Results demonstrate that schools undergoing financial shocks respond mainly by reapportioning regular, accelerated, and special education spending, rather than simply enacting across-the-board cuts. High-poverty, low-performing, and urban schools tended to prioritize reallocations toward targeted group support, while lower poverty, higher performing, suburban schools tended to prioritize reallocations toward regular education support. Furthermore, results of fixed and random effect regression models suggest that while spending allocations are in part determined by district-level characteristics, reactionary changes to spending are more explained by school or leadership characteristics. These results support the notion that site-level budgeting is an important factor in ensuring that spending is calibrated to current student needs when undergoing periods of financial uncertainty.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ To maintain scope, we focus on standard definitions of horizontal and vertical equity in school finance, but recognize that there are several definitions, as well as other conceptions of equity in school finance (Allbright et al., 2019).

² This notion is echoed in the school leadership literature as turbulence theory, can be seen as a situation that disrupts normal operations (Shapiro & Gross, 2013). When turbulence is severe or extreme, it can require that school leaders make decisions based on authenticity and care rather than overriding theoretical or historical frames (Beabout, 2012; Gross & Shapiro, 2004; Shapiro & Gross, 2013).

³ Given research on differences in both accounting and costs structures, charter schools were omitted from the analysis to keep results generalizable to public schools (Knight & Toenjes, 2020). We also censored the first two years for schools that had opened during the sampling period, as well as the last year for schools that closed or were consolidated, as these changes may have been due to other factors.

⁴ $\mathrm{\Delta }\gamma =1-\left(\frac{\gamma }{{\gamma }_{t-1}}\right)$, where $\gamma$ is the current year's spending allocation and ${\gamma }_{t-1}$ is the prior year's allocation. The same approach is used for calculating total budget reduction as a proportion (${\beta }_1)$.

⁵ Student achievement is set as the percentage of students in a school passing the Texas Assessment of Knowledge Skills (TAKS) exam, which is then standardized by year. Accountability ratings are from the Texas Academic Excellence Indicator System (AEIS). While related to student achievement, the AEIS is a rating based on a combination of scores including growth for the lowest performing sub-group, graduation rates, completion rates, and academic progress. We consider them substantively different metrics that did not contribute to issues of multicollinearity, with a VIF < 3.

⁶ Although both fixed and random effects models help with interpretation, results from Hausman tests demonstrated that fixed effect models were preferred.

⁷ For example, Houston ISD covers 280 schools, while Valentine ISD covers one school.

⁸ The Texas Education Agency recommends that budgetary decisions are developed with site level input including a budgetary planning committee including parents, community members, and others (Texas Education Agency, 2010). However, the extent to which each level influences the actual budget likely varies widely between schools and districts as the budgeting process in Texas is both highly related to leadership styles and organizational history (Peternick & Sherman, 1998).

⁹ While smaller districts, who frequently only have a few schools, often all experienced similar expenditure changes, larger districts demonstrated considerable variance by school, with very few districts all experiencing the same magnitude of reduction across schools. Only 141 of 1,016 districts were fully shocked or non-shocked in 2010. For example, in 2010, 17 schools in Austin ISD were ”shocked” with an average reduction in per-pupil spending of $1,309, while 72 others saw increases in their total per-pupil expenditures between $159 and $1,709.

¹⁰ Indeed, those schools that were categorized as never experiencing a shock had roughly 60 more students on average than those that were categorized as experiencing a shock in one or more years. While we cannot directly observe or control for these issues given that the data does not contain expenditure objects, robustness checks comparing the results of models restricted to the smallest and largest schools yielded generally analogous results in terms of coefficient direction and magnitude of variance explained, although overall model fit was generally stronger for larger schools. Similar to the results presented in Table 4, models restricted to small and large schools (based on enrollment quintiles) for accelerated spending level with district fixed effects accounted for the largest proportion of variance for both small (r² 0.37) and large (r² 0.41) schools, and those with principal fixed effects accounted for the largest proportion of variance (small r² 0.13, large r² 0.26) for accelerated change compared to other fixed effect models. These results suggest that while unobserved factors, such as staffing changes, may have more bearing in smaller schools, our substantive results hold across school size. Full results of school size models available upon request.

¹¹ Given the associative nature of these tests, terms such as ”explain” and ”association” should be taken in a statistical sense of accounting for variance, rather than in the causal sense of ”as a result” or ”because of.”

¹² Total reallocation represents the absolute percentage-point change in allocation across all categories. For example, if regular education spending increased from 62% of expenditures to 67% (a five point increase) and ESL spending went from 8% to 6% (a two point decrease), expenditures would have been reallocated by seven percentage points from the prior year.

¹³ Full results, with standard errors and per-pupil spending changes available in Appendix A.

¹⁴ The Texas Education Agency provides a host of compensatory education programs (Texas Education Agency, 2020). Discussions with several current and former principals in Texas highlighted that extended day, week, or year programs were most likely to be used in high-need schools trying to provide support to at-risk, low income, or other vulnerable student populations, as a means to provide nutrition and before/after school supervision and support. These funds may also be used to cover the compensatory interventions that are not often funded, such as the requirement that students not passing the STAAR reading or math benchmark tests in grades 5 and 8 receive accelerated instruction.

Additional information

Notes on contributors

Andrew Pendola

Andrew Pendola, Ph.D., is an assistant professor of Educational Leadership at Auburn University. His work centers around educator labor markets, specifically leadership retention, turnover, and policy.