Association for Public Policy Analysis & Management

While capital spending is one of the largest expenditure categories in schools, we know little about how improvements to capital assets affect schools and students. The best evidence to date on the effects of capital spending focuses on very large-scale projects (particularly, bond-funded, multimillion-dollar projects that face referenda votes). However, research on smaller projects, which are more common and have more limited consequences for future tax burden or borrowing capacity, has been lacking. Presenting challenges to researchers, funding for these projects are usually endogenous because they are initiated under the discretion of principals, school boards, or other local officials and depend upon schools’ needs and financial resources. Researchers have struggled to find ways to estimate causal effects. In this paper, we evaluate the impacts of typically-sized capital projects on schools and students, taking advantage of participatory budgeting (PB) votes to overcome potential endogeneity.

We exploit the results of PB votes in New York City (NYC). Our approach is similar to previous research on large-scale bonded capital spending (Cellini, Ferreiera, Rothstein 2010; Martorell, Stange, McFarlin Jr. 2016). These studies use a regression discontinuity framework, comparing outcomes for districts in which school capital bond referenda passed or failed by narrow margins. Crucially, close referenda create arguably quasi-random treatment and control groups. PB offers the same research benefits as bond referenda because winners and losers of marginal votes are plausibly exogenously assigned. However, unlike bond referenda for huge, multimillion-dollar projects, PB involves smaller projects, approximately $300,000 on average, related to, for example, technology, libraries, bathrooms, air conditioning, electric and water systems, security, etc.

PB is a process that some NYC council members use to allocate a portion of their capital discretionary funds. Within this process, council districts’ residents propose ideas for improvements to schools, parks, streets, public housing, libraries, and other community spaces, and then vote on those projects through a balloting process (residents can vote for no more than five projects). Projects receiving the most votes earn funding until PB moneys run out. Projects with vote totals near the point in which moneys run out provide the plausibly exogenous variation we exploit in this paper.

We match three years of PB vote data (2015-2017) to data on other discretionary capital spending and school expenditures. We then use a regression-discontinuity framework, exploiting the PB vote as an instrument for capital funding. We estimate the extent to which capital spending affects the allocations of expenditures across other functions of schools. We will also examine if capital spending has any effect on student performance. We then will explore whether government officials, including city council members or the NYC Department of Education, find alternative means to fund losing PB projects, softening the PB effect. Finally, we will characterize PB projects by types of spending to explore potential consequences for children. For example, do investments in laptops improve student academic performance to a greater degree than gymnasiums? Taken together, this paper will provide novel evidence on the consequences of capital spending for school budgets and students.