Does accurate financial forecasting improve student achievement?

Death, taxes, and the Browns missing the playoffs are just about the only predictable things on this earth. But far greater uncertainty exists in other aspects of life, including matters of school finance. A new paper by Stéphane Lavertu and Travis St. Clair examines the accuracy of Ohio school districts’ revenue predictions. They also study whether districts’ forecasting errors—a surplus or shortfall relative to predicted amounts—affect student achievement. The analysis relies on state-required financial forecasts produced by Ohio’s 611 districts from 2007–08 through 2014–15.

In terms of forecasting accuracy, Lavertu and St. Clair find that districts underestimated following-year revenues by an average of 2.7 percent during the period of study—what they call a “conservative bias” in districts’ predictions. But as the authors note, it’s not just conservative budgeting that explains the underestimates: Districts also have an incentive to “deflate” revenue predictions in efforts to rally voters to approve local taxes (local news stories illustrate how they can use projected deficits in levy campaigns). But the average masks variation, with a significant number of cases in which districts overestimated revenues, leading to unexpected shortfalls.

Using a battery of analyses, the researchers examine the relationship between districts’ forecasting errors and their value-added scores, a measure of academic performance based on student growth. They conclude that inaccuracies—whether over or under the actual revenues received—have relatively modest effects on value-added results. Specifically, they find that a 1 percentage point increase in forecasting errors is associated with a loss equivalent to one to two days of student learning. But when focusing on districts that overestimated revenues—those that reported rosy predictions but later faced shortfalls—the analysts find somewhat larger negative effects, equivalent to losses of about four to eight days of learning per percentage point increase in errors. Since they run statistical models that control for per-pupil spending, the authors conclude that “the impact of forecast error is unlikely to be attributable to any link between achievement growth and district spending levels.” Finally, based on supplemental analyses, the study offers evidence that teacher attrition and a focus on passing new taxes to fill unexpected budget holes might explain these findings.

Accuracy in revenue projections is surely to be encouraged, and both local and state leaders can take initiatives to this end. On a local level, school leaders can conduct detailed analyses of housing and demographic patterns to better predict local revenues and expenditures. State legislators can do their part as well by creating straightforward, predictable funding formulas that determine districts’ state allocations. A couple steps, described in more detail here, that Ohio can take are: (1) funding districts based on prior-year enrollments, instead of “real-time” headcounts that cause fluctuations in state aid; and (2) phasing out funding caps and guarantees, which are policies that are sharply debated every two years and yield an overly complicated and less predictable formula.

Just like the weather, no organization can perfectly predict future revenues. But helping schools to better plan for the future, so that they don’t end up in financial straits, is a task well worth undertaking.

SOURCE: Stéphane Lavertu and Travis St. Clair, “Beyond spending levels: Revenue uncertainty and the performance of local governments,” Journal of Urban Economics (2018); an open-access version is available here.

Aaron Churchill
Aaron Churchill is the Ohio Research Director of the Thomas B. Fordham Institute.