This is a guest post from my colleagues Brandon Bartels and Jake Haselswerdt, which is substantially more interesting than the title that I gave the post:
In response to Suzanne Mettler’s post on Monday, commenter Josh asked for clarification regarding the importance of policy delivery mechanism (e.g., direct cash payment vs. tax break) to citizen understanding and support of government programs. As Mettler noted in her response, the focus of her and Guardino’s experiment is not the effect of the delivery mechanism but the ways in which providing additional information can impact citizens’ evaluations of indirect or “submerged” programs.In a new paper (available upon request) inspired in part by Mettler’s work, we tackle the effect of delivery mechanism directly with a survey experiment of our own. Our findings confirm that the way a policy is delivered to beneficiaries can have a profound impact on public support for that policy, and that this effect is conditioned by ideology.
We presented survey respondents with a description of a federal housing program, after which they were asked to rate their approval of the program on a seven-point scale. About half of respondents received a description of the real-life Home Mortgage Interest Deduction:
“We’re going to ask you your opinion on a government program intended to help Americans afford to own homes. Under this program, individuals who take out a mortgage to buy a home are eligible to deduct the monthly mortgage interest from their taxable income, thereby reducing their tax burden. The total savings for individuals under this program are estimated to be $94 billion for fiscal year 2011.”
The other half of the respondents were shown a description that differed in two respects: first, the words “eligible to deduct the monthly mortgage interest from their taxable income, thereby reducing their tax burden” were replaced with “eligible for a grant from the federal government to help them afford the monthly payment;” second, the words “The total savings for individuals under this program…” are replaced with “The total government expenditures to individuals under this program…” We believe these contrasts in language were reasonable given the way these types of programs are often framed by elites.
The effect of this manipulation of delivery mechanism is displayed in this bar graph, which displays the percentage of respondents in each treatment group who expressed at least some approval of the program. The effect is considerable, as support drops by about 24% when the program is described as a grant.
The effect of this manipulation was especially pronounced for conservatives. Conservatives appear to be just as willing as liberals to support a government program, provided that it is delivered through the tax code, but less willing to support this program when described as a “grant.”
Is the extra support we observe for the tax expenditure due to a lack of understanding about how such programs actually work? Here our findings are less clear. We included a second experimental factor that varied the amount of information provided, though we focus on program cost rather than distributive impact (Mettler and Guardino’s focus). About half of the respondents in both experimental conditions described above were shown an extra sentence at the end of the program description: “It is estimated that this program will add around $390 billion to the national debt over the next four years.” We expected that this information would come as a bigger revelation to respondents in the tax expenditure group due to the opaque nature of tax expenditures as a policy tool, and that this would reduce the advantage of the tax expenditure. There was little evidence of this when we examined all respondents, however; the debt information reduced support across the board, but no more so for the tax expenditure than the grant, as shown in this graph.
Examining conservative respondents specifically, we do find some evidence that providing debt information modestly tempers the strong positive effect of the tax expenditure condition. In substantive terms, this suggests that while conservatives are much more likely to support a tax expenditure than a comparable direct spending program, this difference is due in part to a lack of understanding about the fiscal impact of tax expenditures.
There is clearly more at work here than conservatives’ failure to account for costs, however. The positive reaction to a tax break as compared to a functionally equivalent grant is present even for liberals, and remains strong for conservatives even in the presence of extra information about program cost. It is quite possible that what we are observing here is a general American antipathy toward “government,” or at least government that looks like government. Alternatively, perhaps it’s about the beneficiaries; Americans may look more generously upon “taxpayers” than they do upon other potential recipients of policy benefits.