Exploring School Board Budget Cut Mentions with Minute Metrix

Budget Cuts Analysis Overview

A topic that’s been particularly salient for schools this past year is budget cuts. With the ongoing public school enrollment declines and the expiration of federal pandemic relief aid, districts nationwide have been grappling with the reality of holes in their budgets.

Minute Metrix enabled me to mine school board documents for insights on important questions related to budget cuts. In particular, I am interested in the district characteristics associated with budget cut discussions and the importance of enrollment declines in driving these discussions. While I can’t answer these questions definitively, I found some interesting patterns that shed light on the dynamics of budget cuts this past school year.

I took a highly simplistic approach to generating the data for my analysis. I conducted a semantic search on Minute Metrix with “budget cuts” as the topic. I had the platform return a dataset with every district that had discussed budget cuts in the 2024 calendar year and the number of times the topic was discussed. This approach does not do justice to the much more powerful capabilities of the platform, but it was still quite effective for my purposes.

To then get the full sample of school districts on the platform, I generated another dataset but with “budget” as the topic. As expected, this yielded a much larger set of school districts. I merged the “budget cuts” dataset onto this larger set of districts to construct my final dataset that identifies the districts that did and did not discuss budget cuts.

I must caveat the analyses below by noting that this is not a random representative sample of districts. The results I present can therefore not be generalized beyond this particular sample of districts. Still, the patterns I uncover shed light on the time-sensitive topic of budget cuts. Administrative data on district finances typically lags multiple years behind the present, so the ability to generate large-scale insights on district finances in real-time would otherwise not be possible.

Characteristics of Districts That Discussed Budget Cuts

To start, I created a table of average characteristics of districts that did and did not discuss budget cuts.

Districts that discussed budget cuts have higher average enrollment, but more notably, they have also seen larger average enrollment declines. Between the 2022-23 and 2023-24 school years, districts that discussed budget cuts saw a 1.1% average enrollment decline, more than double that of districts with no budget cut discussions. The declines have also been larger in these districts since the first pandemic year of 2019-20 (with enrollment measured in the fall of that year).

Districts that discussed budget cuts also tend to operate more schools, and they have a lower average enrollment per-school despite the higher overall enrollment.

Additionally, nearly one-third of districts that discussed budget cuts are urban, compared to just 14% of districts that did not discuss budget cuts.

The average demographic composition of the two sets of districts are relatively similar, with the districts that discussed cuts enrolling a slightly smaller share of White and Black students and slightly larger share of Latino and Asian students.

Budget Cut Mention Regression Models

I took the district characteristics analysis a step further by running regression models. These models take into account not just average characteristics, but also the amount of variation in these characteristics among districts in each group. They indicate whether the average differences are noteworthy relative to the overall variation.

The models also take into account the correlations among the district characteristics, telling us the importance of each individual measure above and beyond the others. For instance, urban districts may have seen larger enrollment declines, and they may also be more likely to discuss budget cuts for reasons unrelated to declining enrollment. The regression allows us to see whether enrollment declines are related to budget cut discussions regardless of whether a district is urban or not.

One more technical point is that school districts are nested within states. I am interested in the school district characteristics that are related to the likelihood of budget cut discussions, but there may also be state-level factors, like state budgets, that explain some of the budget cut discussion frequency. I therefore use a state “fixed effects” regression model, which removes the between-state variation by only comparing districts within the same state.¹

The regression model is set up as follows: the outcome, or “Y” variable, is an indicator that equals 1 if a district discussed budget cuts in the prior year and 0 otherwise. The predictors, or “X” variables, are the variables in the descriptive table above.² This is known as a “linear probability model” because the result attached to each predictor variable indicates the change in probability of discussing budget cuts associated with a one-unit increase in the variable.³

I also ran a second model version where the outcome is each district’s total budget cut discussion count. For simplicity, I ran this model in the same fixed effects linear regression framework.⁴

The table below displays the results of these regression models. The first number in each cell is the coefficient, and the number in parentheses is the standard error, a measure of variance. The larger a coefficient is relative to its standard error, the more certainty we have that the coefficient is different from zero and therefore truly related to discussions of budget cuts. Generally, if a coefficient is close to double the size of its standard error, it is considered “statistically significantly” different from zero, and we can be relatively confident that there is a true relationship there. The “+” symbol next to a coefficient indicates a statistically significant result.⁵

Before interpreting the results, it’s important to remember that these are correlations. We would need a more rigorous research design to say whether any of these factors actually cause budget cut discussions.

The most interesting set of results is in the “Enrollment & Operations” section. For the model where the outcome is whether a district had any mention of budget cuts, the strongest predictors are districts’ total number of schools and the number of students per school. A one percent increase in a district’s number of schools is associated with a 3.6% increase in the probability of mentioning budget cuts, and an increase of 100 students per school is associated with a 0.4% reduction in the probability of mentioning budget cuts. Enrollment change does not quite reach conventional levels of statistical significance, though it comes close, with a one percent decline in enrollment associated with a 0.6% increase in the likelihood of mentioning budget cuts. In the model where the outcome is total budget cut mentions, the enrollment change variable becomes the stronger predictor.

These results seem to suggest that enrollment declines are related to budget cut mentions, and that having under-enrolled schools may be a particularly important dynamic in the context of declining enrollment.

For locales, urban districts are more likely to discuss budget cuts than suburban districts, though the variance is also relatively large, so the results do not quite reach statistical significance. Rural districts, though, had statistically significantly fewer total mentions of budget cuts than suburban districts.

Finally, the demographic variables were generally not strong predictors of budget cut mentions, with some notable exceptions. Districts with larger shares of Asian students relative to White students had fewer total mentions of budget cuts. Also, districts with larger shares of Latino students relative to White students were less likely to discuss budget cuts, though these do not quite reach statistical significance.

Takeaways

These results highlight interesting patterns relevant to current trends affecting district budgets. The enrollment-related results underscore the compounding budgetary challenges of declining enrollment while operating a large number of under-enrolled schools. The results do not indicate disparities by race or poverty in budget cut discussions, which may provide some solace given the many other educational inequities that fall along these lines.

It is important to remember that these results are far from definitive. Rather, they are intended to showcase the possibilities that Minute Metrix provides to conduct rapid, real-time analyses that can help understand important contemporary education issues.

This analysis is just the tip of the iceberg. I am excited to continue exploring the platform and gaining familiarity with the far more powerful capabilities that the platform offers. For example, as a next step, I want to search for both budget cuts and enrollment declines within the same document excerpts. This would provide much stronger evidence as to the degree to which enrollment declines are driving budget cuts. I could similarly conduct a search for both budget cuts and federal pandemic relief aid in the same excerpt to gauge the relative contribution of these two factors to budget cuts. There is a vast ocean of potential insights in these school board documents waiting to be explored.