Moochers and Makers in the Voting Booth Who Benefits from federal spending and hoW

the 2012 election campaign popularized the notion that people who benefit from federal spending vote for democrats, while people who pay the preponderance of taxes vote republican. a survey conducted during the election included questions to test this hypothesis and to assess the accuracy of voters’ perceptions of federal spending. Voters’ perceptions of their benefit from federal spending are determined by family income, age, employment status, and number of children, as well as by party identification and race. Voters aged 65 and older who believe they are net beneficiaries of federal spending are more likely to be democrats and vote for Barack obama than seniors who believe they are net contributors to the federal government. however, the 77.5 percent of voters under age 65 who believe they are net beneficiaries of federal spending are as likely to vote for romney as for obama and as likely to be republicans as democrats. Voters who live in states that receive more in federal funds than they pay in federal taxes are less likely to vote for obama or to be democrats. for most of the electorate, dependence on federal spending is unrelated to vote choice. harold lasswell (1936) defined politics as “who gets what, when, and how.” to voters, it is the federal budget—the combination of taxes paid and benefits received—that defines who gets what. the conflict over who gets what in the federal budget captured attention in the 2012 election campaign after the release of a videotape in which republican candidate mitt romney told a group of supporters, “there are 47 percent who are with [president obama], dean lacy is a professor of government at dartmouth college, hanover, nh, Usa. the editors and three anonymous reviewers provided valuable comments on earlier versions of this paper. Brice acree, harry enten, James Khun, david rufful, and Jessica taub provided research assistance. the dartmouth program in politics & law funded dartmouth college’s participation in the cooperative congressional election survey. *address correspondence to dean lacy, hB 6108, hanover, nh 03755, Usa; e-mail: [email protected]. Public Opinion Quarterly, Vol. 78, special issue, 2014, pp. 255–275 doi:10.1093/poq/nfu016 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from who are dependent upon government, who believe that they are victims, who believe that government has a responsibility to care for them, who believe that they are entitled to health care, to food, to housing, to you name it.” this quote haunted romney during the campaign, yet he reaffirmed it after the election when he told a group of donors that obama won because he “focused on giving targeted groups a big gift...that add up to trillions of dollars” (reston 2012). romney’s comment about the “47 percent” is derived from the percentage of americans who did not pay income taxes in 2011 (tax policy center), but it reflects a broader perception that the republican party represents people who pay more in taxes than they receive in federal benefits—the makers—while the democratic party represents people who receive more in government benefits than they pay in taxes—the moochers (e.g., eberstadt 2012; sykes 2012). campaign rhetoric aside, perceptions of the federal budget constitute an important test case for economic theories of voting. theoretically, voters who benefit from federal spending should vote for candidates who pledge to protect or expand federal programs, while voters who lose money to the federal government should prefer candidates who promise to cut taxes and reduce spending. the public’s perceptions of who gets what from federal spending is understudied, as is the effect of such perceptions on vote choice. a 2012 election survey included a question asking people whether their family is a net beneficiary of or net contributor to the federal budget. this article assesses the content of those perceptions—they do seem to be partly accurate—and then uses them to explain vote choice in the 2012 presidential election. surprisingly, there is little correlation between voters’ perceptions of whether they benefit from federal spending and their vote. Both moochers and makers are about equally likely to vote democratic or republican, which contradicts popular accounts and campaign rhetoric about the types of voters who support each of the parties. section one briefly reviews the literature on federal spending and elections. section two describes the 2012 election survey instruments and assesses their validity. section three presents results from a model of vote choice and party identification that includes public perceptions of benefit from federal spending. section four discusses implications for party competition, budget politics, and theories of voter behavior. What We know about Federal spending and Voter choice aggregate-level evidence suggests that federal spending affects elections. Kriner and reeves (2012) examine county-level data from 1980 through 2008 and find that increases in federal grant spending are associated with increases in the incumbent presidential party’s share of the vote. they conclude that Lacy 256 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from “presidents are rewarded at the ballot box for federal spending” (Kriner and reeves 2012, 363). a few studies examine the relationship between presidential elections and a state’s net benefit from federal spending, taking into account both taxes paid and spending received. lacy (2002) shows that states that received more in federal spending than their residents paid in taxes were more likely to vote for republican george W. Bush than for democrat al gore in 2000 (see also pink [2004]). states that increased their ratio of federal spending to tax dollars during the clinton administration were even more likely to vote for Bush. francia and levine (2006) show that states that increased their vote share for Bush from 2000 to 2004 saw increased federal spending per tax dollar over the same period. lacy (2013) shows that a state’s ratio of federal spending to taxes has an increasingly positive correlation with republican vote margin since 1996. in contrast to the findings of Kriner and reeves (2012), only republican presidential candidates, not incumbents generally, benefit from increases in federal spending (lacy 2013). Kriner and reeves (2012), francia and levine (2006), and lacy (2002, 2013) examine aggregate data, not the individual-level relationship between benefit from federal spending and vote choice. several studies show that individuals may not be aware of the benefits they receive in the federal budget. howard (1999) argues that the tax code creates a hidden welfare state through tax breaks and loopholes such as the home mortgage interest deduction, college loan interest deductions, and employer pensions. mettler (2011, 38) finds in survey data that about 50 to 60 percent of americans who benefit from tax loopholes report that they “have not used a government social program.” similarly, about one-quarter of people who receive food stamps or subsidized housing and over 40 percent of people who receive medicare or social security claim that they “have not used a government social program.” many americans appear unaware that they benefit from federal spending (appelbaum and gebeloff 2012). these findings raise at least two questions that need answers. first, are people’s perceptions of their net benefit from federal spending accurate or, at the minimum, do their perceptions coincide with the variables that determine who gets what from the federal government? We do not know whether the people who believe they benefit from federal spending are the ones who do. theoretically, people’s net benefit from federal spending should matter to them, not whether they benefit from specific social programs. presumably most people who receive social security, medicare, or government salaries do not pay more in taxes than they receive in benefits, but some might. people who benefit from tax loopholes may still pay a lot more in federal taxes than they receive in federal dollars. it is the balance of taxes paid and spending received that should in theory create political divisions among voters. second, what is the impact of perceptions of federal spending on the vote? mitt romney’s comment that the 47 percent of americans who are dependent on the government would vote for obama is consistent with a conventional Moochers and Makers in the Voting Booth 257 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from perception that people who benefit from federal spending vote democratic. this is not only mere campaign rhetoric. it is consistent with self-interest if beneficiaries of federal spending vote for the party that will protect federal coffers while net contributors to those same coffers vote for the party that promises to cut taxes and spending. no academic study examines whether people who benefit from federal spending are more likely to vote for one political party over the other. Measuring Perceptions of Benefit from Federal spending data for this study come from the 2012 cooperative congressional election survey (hereafter cces). the cces is an opt-in survey conducted over the internet by yougov/polimetrix.1 the computer format of the cces survey allows respondents to complete surveys at their leisure and often results in more reliable and valid responses than telephone interviews (chang and Krosnick 2009). the study solicited 77,357 email addresses, of which 29,182 yielded a completed interview and 3,267 a partial interview. the sample is weighted to match the 2008 american community survey conducted by the U.s. census Bureau. the questions used in this study were part of a two-wave panel given to a randomly selected 1,000 respondents in the cces sample. Wave 1 was fielded between october 1 and november 3, 2012, N = 1,000. all respondents to wave 1 were invited to participate in wave 2. Wave 2 was conducted between november 7, 2012, and January 6, 2013, with 96 percent of interviews completed before december 1, 2012, N = 822. the survey methodology, questions, and descriptive statistics are in the appendix. to measure public perceptions of benefit from federal spending, wave 1 included the following question: “think about all the taxes you and your family pay to the U.s. federal government, such as income taxes and social security taxes. also think about how much money you and your family get from the federal government in social security, medicare or medicaid, salary or wages if you are employed by the federal government or military, farm subsidies, education loans, veterans’ benefits, or anything else. Would you say that you and your family get more money from the federal government than you pay in taxes, do you pay more in taxes than you get back, or do you pay about the same amount in taxes as you get back from the government?” ( ) pay more in taxes than we get back from the federal government 1. the cces is a nonprobability sample. therefore, no accepted aapor response rate can be calculated and reported for the survey. see ansolabehere and schaffner (2012) and http://projects. iq.harvard.edu/cces for details on the cces survey. Lacy 258 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from ( ) get back from the federal government more than we pay in taxes ( ) pay about the same amount in taxes as we get back from the government such a question could present a family’s benefit from federal spending in retrospective, prospective, or current terms. a retrospective approach would ask, “have you received more in benefits than you have paid in taxes over the course of your lifetime?” a prospective question might ask, “does the benefit you expect to receive over the course of your lifetime exceed what you expect to pay in taxes over your lifetime?” the cces survey question taps current benefit: “at this point in time, are you paying more in taxes or receiving more in benefits?” the current perspective on the question is simpler and more straightforward for respondents to answer, requires them to make fewer assumptions about the past and present, and is also more consistent with common conceptions of “moochers.” given a prospective version of the question, different respondents could answer differently not due to their beliefs about their budget status, but due to their expectations about the future, such as the continued viability of social security. in the retrospective question, the past is past and may no longer be relevant to current fiscal preferences. of the 986 respondents who answered the question in the october preelection survey, 56.5 percent of respondents believe they pay more in taxes than their family receives in spending, making them net contributors to federal coffers. another 24.8 percent believe they pay about the same in taxes as they receive in spending. only 18.7 percent believe they are net beneficiaries of federal spending since they receive more from the federal government than they pay in taxes.2 the fact that less than 20 percent of americans believe their families receive more in federal spending than they pay in taxes says much about the federal budget. republican presidential candidate mitt romney’s remark that 47 percent of americans are dependent on the government is based on the percentage of americans who do not pay income tax. But most of those americans do pay social security, medicare, and excise taxes. even if 47 percent of the public does not pay income taxes, a much smaller share perceive of themselves as net beneficiaries of federal spending. it is difficult to know if these perceptions of taxes and spending are accurate. families may underestimate their benefit from federal spending, overestimate their taxes paid, or both. We do not know how much each individual really receives in federal spending compared to their taxes paid. it is especially difficult to calculate an individual’s indirect benefits from federal spending, such as law enforcement, national defense, environmental protection, and subsidies for roads, education, agriculture, and housing, 2. these percentages are based on the weighted yougov/polimetrix sample. descriptive statistics for the variables in the analysis are included in the appendix. Moochers and Makers in the Voting Booth 259 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from one way to assess the accuracy of these perceptions is to test whether they are explained in a multivariate model by variables that generally determine a person’s eligibility for many federal programs. the dependent variable in this ordered logit model has three values: a respondent believes her family (1) pays more in taxes than it receives from the government; (2) pays about the same in taxes as it receives in spending; or (3) receives more from the federal government than it pays in taxes. the explanatory variables are socioeconomic factors that determine eligibility for some of the larger federal spending programs, plus political variables that are known to filter perceptions. family income should be negatively related to net benefit from federal spending. higher-income families generally pay a higher income tax rate, though a lower percentage of income in payroll taxes, and they are less likely to be eligible for means-tested social programs. having a full-time job should also be negatively related to net benefit. conversely, people who are retired, unemployed, or working part time should be more likely to benefit from federal spending. age should have a curvilinear relationship with benefit from federal spending, first declining in its effect on net benefit, then increasing as a person hits retirement age and older due to the availability of retirement benefits such as social security. the number of children in a family under age 18 should be positively related to net benefit from federal spending due to the income tax credit per child, tax credits for child care, the children’s health insurance program, food stamps, head start, college loans, and other programs. people who served or have immediate family members who served in the military are likely to benefit from federal spending. a three-category dummy variable separates respondents into those who served in the military, those who have family members who served, and a baseline of respondents who have not served and have no family members who served. education level is also in the model, but its expected effect on federal benefit is less clear. political and demographic variables may also predict a person’s perception of benefit from federal spending. race will predict perceived benefit if white respondents are less likely than others to believe they benefit from federal spending. the media and politicians frequently portray welfare recipients as minorities, and whites tend to associate welfare programs with african americans (gilens 1999). a voter’s party identification may also explain their perceived benefit from federal spending. one reason for this link may be self-interest. democrats may be more likely to know about or to seek federal benefits, while republicans may be more likely to forego federal benefits. partisans may also rationalize whether they benefit from federal programs. democrats may be more likely than republicans to claim a personal benefit from federal spending, in support of the party’s reputation for protecting federal programs. republicans may be more likely to deny that they benefit from federal spending when they really do. alternatively, when partisans mimic the rhetoric of their party’s leaders, Lacy 260 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from republicans may claim that they are net contributors to the government while democrats may express they are net beneficiaries. an additional variable may explain people’s perceptions of their benefit from federal spending: their home state’s actual benefit from federal spending. information about total federal taxes paid from each state in each year is available from the internal revenue service.3 total federal spending in each state is reported by the U.s. census Bureau in the consolidated federal funds report (2011). a state’s benefit from federal spending is measured as total spending in a state divided by total taxes paid from the state, ranging from a low of $0.53 for delaware to a high of $3.67 for new mexico. the median states, tennessee and north carolina, receive $1.55 and $1.58 in federal spending for every dollar paid in taxes.4 table 1 reports the results of the ordered logit model.5 people’s perceptions of their family benefit from federal spending vary systematically with the sociodemographic and political variables that should predict them. people who are older, working full-time, or earning higher incomes believe they pay more in taxes than they receive in federal spending, since those coefficients are all negative and statistically significant. the square of age has a positive coefficient, indicating a nonlinear effect where older americans begin to receive more in federal spending than they pay in taxes.6 the number of children under 18 in a family is positively related to perceived benefit from federal spending. military service—by either the respondent or a family member—is unrelated to perceived benefit from federal spending. gender and marital status have no effect on perceptions of federal spending benefit. a state’s federal spending ratio is unrelated to a voter’s perceived family benefit from federal spending. When estimating that benefit, people do not account for the indirect effect of federal spending near their homes: military bases, national parks, federal roads and public works projects, grants, and more. state spending ratio aside, the “objective” predictors of personal benefit from federal spending strongly suggest that people’s perceptions are partly accurate or at least responsive to the things that determine “who gets what.” 3. total taxes paid are reported separately from total tax refunds to a state in a year. We calculate total taxes as taxes paid minus refunds received. 4. congress ended reporting of federal spending in the consolidated federal funds report after 2010. therefore, the 2010 taxing and spending data are used to calculate a state’s net take from the federal government. federal spending ratios per state change very little from year to year, although over decades they change somewhat (lacy 2013). 5. models are estimated on five multiply imputed data sets because of missing observations on family income. estimates are averages of the estimates from the five data sets, with standard errors adjusted (King et al. 2001). 6. estimating the model with age in years replaced by a dummy variable for people over the age of 65 shows a large positive and statistically significant effect for seniors. Moochers and Makers in the Voting Booth 261 by gest on O cber 0, 2014 http://poqrdjournals.org/ D ow nladed from