Association for Public Policy Analysis & Management

In this paper I estimate the effect of a change in tax base on local governments’ choice of tax instrument. Municipalities in Ohio may choose to impose an income tax, a property tax, or a mix of the two. Interestingly, the income tax generally applies both to income earned within the jurisdiction, and to income earned by residents of the jurisdiction. In other words, the income tax is both source-based and residence based. A city’s choice between income and property tax could be motivated by several considerations. In a city where a larger proportion of the population is unemployed or not in the labor force, residents may prefer an income tax (whose incidence falls on their higher income neighbors or commuters) to the property tax, and the municipality’s choice of tax instrument will likely reflect that preference. A city that experiences a large inflow of commuters may wish to “export” its tax burden to commuters; however, the tax competition literature suggests that such tax exporting will be limited by the fear of driving economic activity out of the district. In this paper, I seek to determine which of these factors prove most important to city administrators and voters.

 I consider a model in which a municipality’s current tax policy depends upon inflow of commuters into the municipality, the number of residents in the municipality who are employed, and the share of the residential population employed, in previous periods. I instrument for changes in these explanatory variables using changes in employment shares by industry at the macroeconomic level. By interacting these shares with base-year employment by industry in the municipality, I create a set of “Bartik-type” instruments for the potentially endogenous commuting and employment variables.

I then estimate the model in first differences, regressing two-year lagged differences in the explanatory variables upon four dependent variables: current property or income tax rate, and current share of tax revenue from the property or income tax.  Preliminary results from a 2002-2012 panel of Ohio cities show that a 10% increase in residents’ employment causes a 0.37 standard deviation (1.7 mill) increase in property tax rate, while a one standard deviation (0.066) increase in share employed decreases the income tax rate by 0.10 standard deviations (0.06 percentage points) and increases the property tax rate by 0.04 standard deviations (0.18 mills). Given these effects, it is unsurprising that a one percentage point increase in share of the population that is employed increases share of revenue from the property tax by about 0.88 percentage points. Commuting inflow, on the other hand, has no statistically significant effect on choice of tax instrument, perhaps suggesting that tax competition offsets the tax exporting motivation for imposing an income tax that falls upon commuters.