Many U.S. school infrastructure projects are financed with municipal bonds, which must be authorized in local referenda. I construct a new bond referenda dataset using online newspaper archives, and use it to estimate how school infrastructure spending impacts local economies by comparing areas where bond authorizations were (narrowly) approved with areas where they were defeated.
I find that bond approval has large, positive effects on local income and employment. The long-run income multiplier is above 2, and an additional job is created per $40,000 of school spending. In addition, school construction leads to higher net migration, and to persistent increases in population, house prices, residential construction, and mortgage refinancing. Overall my results are consistent with infrastructure spending acting both as a temporary shock to labour demand and as a persistent shock to local amenities, and suggest an important complementarity between fiscal multipliers and the marginal utility of government spending.
This award was used to collect the data on vote shares in each referendum. The tasks that required funding where
- Writing a computer script to search and download relevant articles
- Writing a computer script to parse the article text for voting shares
- Running a data validation project on Amazon Mechanical Turk
Tasks (1) and (2) were completed. Task (3) was not completed due to technical difficulties using Amazon Mechanical Turk, and the lack of reliability of the results obtained using those services. As a result, the award’s funding was not fully used.
This award resulted in a working paper titled The Effects of Public Infrastructure Spending: Evidence from U.S. Bond Referenda.
The next steps will be to finalize the working paper for publication. The dataset could also be used in future research projects looking at the effect of school construction on educational achievement and inequality.
