Schools and Stimulus

Abstract

This article analyzes the impact of the education funding component of the American Recovery and Reinvestment Act of 2009 (Recovery Act) on public school districts. We use cross-sectional differences in district-level Recovery Act funding to investigate the program's impact on staffing, expenditures, and debt accumulation. To achieve identification, we use exogenous variation across districts in the allocations of Recovery Act funds for students with special needs. We estimate that $1 million in grants to a district had the following average effects: Expenditures increased by $570,000, employment changed little to none, and debt increased by $370,000. Moreover, 70 percent of the increase in expenditures was in the form of capital outlays. Next, we build a dynamic, decision-theoretic model of a school district's budgeting problem, which we calibrate to district-level expenditures and staffing data. The model can qualitatively match the employment and capital expenditure responses from our regressions. We also use the model to conduct policy experiments.

Introduction

The American Recovery and Reinvestment Act of 2009 (the Recovery Act) was signed into law with a primary goal of creating and saving millions of jobs during and following the 2007-09 recession. A large share of the appropriations from the act consisted of grants. Public school districts constituted one of the largest groups of these recipients, receiving $64.7 billion in Department of Education Recovery Act funds.

The act's education component has been touted as one of the success stories by the law's supporters. Shortly after its passage, Vice President Joe Biden stated that funds from the act would "help to keep outstanding teachers in America's schools." According to the Executive Office of the President of the United States (2009, p. 4), "the rapid distribution of SFSF [State Fiscal Stabilization Funds] funding helped fill the gaps and avert layoffs of essential personnel in school districts and universities across the nation." The act's official website, Recovery.gov, used surveys of recipient organizations to track the number of jobs payrolled by the act's funds. The Council of Economic Advisers (various quarterly reports) used the job-count data from these surveys as evidence of the act's success. According to these reports, Department of Education Recovery Act dollars alone directly created and saved over 750,000 jobs during the first two school years following the act's passage. For context, Figure 1 plots the demeaned growth rates of education employment and noneducation, that is, total net of education employment between 1991 and 2016. Noneducation employment is the much more cyclical of the two.

Figure 1: Education and Noneducation Employment Growth in the United States, 1991-2016

NOTE: Shaded bars indicate recessions as determined by the National Bureau of Economic Research. Growth rates are demeaned.

Bureau of Labor Statistics.

This article analyzes the act's impact on schools, using cross-sectional differences in district-level Recovery Act grants and expenditures, staffing, and debt accumulation. We compare the behavior of districts receiving relatively little grant money with that of districts receiving generous grants. From this comparison, we infer what districts would have done without the grants.

To address the potential endogeneity of spending, we employ two instruments. Our first instrument is the ratio of the number of special-needs students to the total number of students in each district. Our second instrument is the Recovery Act dollars received by a district through the act's Special Education Fund (SEF). The SEF was one category of the Recovery Act education component and constituted one-fifth of the education grants. Its allocation across districts was determined primarily by the requirement that districts finance their special-needs programs. Although each instrument is highly correlated with overall Recovery Act education spending, each is plausibly uncorrelated with the short-run business cycle and tax revenue situations faced by school districts.

We have four main findings. First, the grants had either no or only a small impact on education jobs. Each $1 million of aid to a district resulted in roughly an average of 1.5 additional jobs within that district. The point estimate implies that, in the first two school years following its passage, the act increased education employment by 95,000 persons nationwide. Moreover, this estimate is not statistically different from zero.

We find no evidence that the grants increased the number of classroom teachers. Intuitively, district administrators may have shown a strong preference for maintaining teacher-to-student (teacher-student) ratios and, to a lesser extent, staff-to-student (staff-student) ratios. As such, school officials may have found margins other than firing or hiring with which to cover shortfalls or spend surpluses.

Second, each $1 million of grants to a district increased its expenditures by $570,000. Because districts already had substantial funds from local and state sources, the additional Recovery Act funds were effectively fungible. Thus, upon receipt of Recovery Act funds, state and local funding sources may have reduced their own contributions to district funding, thereby offsetting the act's grants.

Third, districts that received grants tended to accumulate more debt. Roughly 70 percent of the spending increases were capital expenditures, that is, spending on construction, land, or existing structures (CLS) and equipment. Why might districts have used these funds for capital improvement? Since this aid was temporary, school districts may have smoothed the benefits of the aid over time by making long-lived physical investments.

Fourth, we build and calibrate a model of dynamic decision-making by a forward-looking school district. We show that the small employment effect and relatively large investment effect fall out of a fully specified and realistic dynamic programming problem.

We also use our theoretical model as a laboratory to understand the effects of different types of policy. Our main finding is that forcing school districts to use all the stimulus money on labor has no additional effect on the employment outcome. School districts that are forced to use stimulus money only on employment reduce their labor spending from state and local funding sources and substitute this shortfall with stimulus money, leaving the net employment outcome unchanged. We show that an alternative policy requiring districts to spend most of their revenue (from both stimulus and state and local sources) has a more significant effect on employment.

With respect to existing work, there is little economic research on the act's education component. Two exceptions are Dinerstein et al. (2013), who study the impact of the act on universities, and Chakrabart and Setren (2011), who examine the impact of the recession and the early part of the Recovery Act on school districts in the state of New York. More generally, other studies using microeconomic evidence that study the overall impact of the Recovery Act have focused mainly on economy-wide labor market outcomes. These include Chodorow-Reich et al. (2012); Conley and Dupor (2013); Dupor and McCrory (2018); Feyrer and Sacerdote (2012); and Wilson (2012).

Another line of research studies how federal grants to schools influence school spending. Gordon (2004) studies the impact of additional federal grants to school districts serving economically disadvantaged children through the No Child Left Behind Act of 2001. She finds that, although the additional federal grants initially caused a dollar-for-dollar increase in school spending, over time school districts offset those increases with reductions in their own contributions to education funding.

Lundqvist, Dahlberg, and Mork (2014) study the impact of intergovernmental grants to local governments in Sweden and find that the grants do not stimulate local public employment. Evans and Owens (2007) study the extent to which federal grants to fund new police hires increased the size of local police forces versus simply supplanting local funding. They found that for every four officers payrolled by a grant, in an accounting sense, a police force actually increased by only a little over two officers.