Association for Public Policy Analysis & Management

While the fields of ecology, epidemiology, and criminal justice have a robust history of utilizing spatial autocorrelation to expound their understanding of population fluctuation, disease outbreak, and criminal activity, the field of tax policy has been slow to adopt this technique. Currently, little is known about the spatial concentration of federal tax expenditure activity and its spread spatiotemporally. In this study I explore the spatial dimension of participation into one specific tax expenditure, the Saver’s Credit. This is a nonrefundable credit introduced in 2002 to encourage low- to middle-income households to contribute to retirement savings plans. Previous research into this credit’s effect have reported a lackluster response. However, they were all nationally focused over short time horizons and completely ignored the spatial aspect of credit uptake and participation. For this research, I employ the universe of tax filers from 2002 through 2013 to isolate spatially autocorrelated clusters across zip codes and document the change in concentrations over time. I use four varying definitions of population, each being progressively more targeted at those ever more likely to be induced to participate in the credit. I estimate the existence and magnitude of spatial autocorrelation both globally and locally across the U.S. and over time for each of these four populations. Moreover, I develop hot spot analysis maps to identify the exact geographic locations exhibiting particularly high or low activity. I find that for all population groups Saver’s Credit participation is not evenly distributed with quite significant pocketing of high and low activity in the first year with most low-activity clusters focused around urban areas. As time passes, I find that these pockets of activity dissipate with larger declines in areas of low activity than in areas of high. This research marks one of the first studies into the spatial autocorrelation of tax expenditures and offers significant insight into the flow of tax expenditure participation over time.