The Influence of Fiscal Information on Preferences for City Services WILLIAM SIMONSEN* University of Oregon MARK D. ROBBINS University of Georgia
This article looks at the effects of fiscal information on citizen preferences for city services. We ask the question: Does support for taxes for government services change if citizens are presented with the tax amount? We explore this question using surveys of registered voters in Eugene, Oregon. Understanding how fiscal information sways support for taxes for services could influence how much and what type of information is provided to citizens. That is, less principled individuals could “rig the game” by withholding or providing fiscal information in order affect citizen responses. First, we review the literature about citizen attitudes toward taxes and services. Next, we discuss our methodology followed by our research findings. Finally, we discuss the implications of our findings.
THE “FREE LUNCH,” FISCAL INFORMATION, AND SUPPORT FOR TAXES FOR SERVICES Typically, when citizens are asked whether they prefer continuing or enhancing services, lowering taxes, or increasing efficiency they usually choose lower taxes and increased efficiency. Several notions of efficiency may be represented by these responses. In one case, efficiencies are possible where service levels are increased and revenues remain constant. In the other scenario, service levels are maintained while revenues are reduced. Commonly it is a third and more extreme interpretation of efficiency with which govern*Direct all correspondence to: William Simonsen, 1209 University of Oregon, Eugene, OR 97403-1209, Telephone: (541) 346-3859. The Social Science Journal, Volume 37, Number 2, pages 195–214. Copyright © 2000 by Elsevier Science Inc. All rights of reproduction in any form reserved. ISSN: 0362-3319.
196
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
ments are faced. This is the belief that services may be increased or maintained while taxes are substantially decreased. This represents a belief that taxpayers can find and enjoy the elusive ‘free lunch.’ Citizens are asking for simultaneously lower taxes and increased efficiencies, suggesting that many citizens think government is inefficient and want their tax dollars used more wisely. The ‘free lunch’ notion that citizens prefer both increased services and decreased taxes was explored by Susan Welch (1985). She investigated this ‘more for less paradox’ in her research on public attitudes toward taxes. Welch concluded that those truly wanting more for less were a minority. Her work suggests that such a paradox exists, but makes up a small relative portion of the attitude toward taxes and spending. This work also suggests that citizens respond with more sophistication when presented with the trade-offs for taxes and services. Wilson (1983) looked at the more for less paradox by composing a measure of sincerity. Sincere spending preferences are defined by this author as a willingness to increase city taxes, which corresponds with their preferred budget increases. Wilson surveyed residents in Tempe, Arizona and Eugene, Oregon in order to determine whether preferences for spending occur with corresponding willingness to pay. He concludes that citizens are willing to pay more in order to receive more government services. Winter and Mouritzen (1997) examined the “free lunch” concept in Denmark. They surveyed 3000 randomly selected adults in a Danish municipality with about 175,000 people. Winter and Mouritzen found that citizens demand both more spending, item by item, and also a smaller tax. The notion of the “free lunch” is directly related to the influence of fiscal information on preferences for government services in that citizen support may be influenced by an understanding of the price they pay for those services. Arrington and Jordan (1982) used data from a survey of Mecklenburg County, North Carolina registered voters to determine the influence of fiscal information on citizen’s willingness to pay for the costs of government services. The results, which do not control for demographic variables or service use, reveal that “for virtually every [government service] activity the support was less when respondents were asked whether they would pay the costs directly.” (p. 169). Citizen judgments are complex and are based on a multiple set of influences.1 A key lesson from this literature is that greater efficiencies and/or lower taxes, while maintaining existing services, is the preferred option when it is offered.
BUDGET CONSTRAINTS Many surveys that ask citizens about local government services do not contain a budget constraint or, at least, information on the cost of the service. This leads to an overstatement of preference since this information is not presented and the resulting opinion is volatile once other conditions are imposed (Clark, 1974). The imposition of a budget constraint can be helpful in two ways. Constrained respondents must trade-off support or allocation increases against corollary decreases. This ideally checks the overstatement, particularly if the other items are something about which respondents care. Budget constraints also provide the opportunity to present respondents with a ‘reality-based’ experience, one that reflects some of the essential dynamics of the actual policy choice facing decision makers.
The Influence of Fiscal Information
197
In this way the choices of respondents to a survey on budget allocation may be presented in much the same fashion—and with the same trade-offs—as those confronting the decision makers themselves. Budget pies, for instance, present one method of imposing a constraint, by requiring that respondents visually indicate how they would allocate funds for various services with totals equaling 100% (a circle is presented to facilitate this choice). Based on the concept of budget pies as presented by Clark (1974) authors McIver and Ostrom (1976) explore the application of the process to a field experience based on police spending. Respondents divided a circle into pieces representing their preferred allocations for aspects of this service. To correct for overstatement (respondents may rank things highly because they like a service or because they are dissatisfied and want more emphasis in that area) control variables based on negative and positive experiences with the police were introduced into the model. Once differentiated, some positive relationships were observed between the budget allocations of those served and local police expenditures.
METHODOLOGY In 1991 and 1992 the city of Eugene, Oregon completed a citizen participation process to ascertain public sentiment for a variety of spending and budget balancing alternatives. An $8 million (approximately 10%) budget shortfall was projected, requiring either service reductions, revenue enhancements, or both. Revenues had been growing slower than expenditures for several years due, in part, to property tax growth (the major revenue source), lagging behind expenditure demand. Property tax levies could increase six percent annually without voter approval (a simple majority). In addition, in November 1990, Oregon voters approved Measure 5, a constitutional amendment further limiting property taxes.2 Measure 5 limited annual property taxes levied for all local governments other than schools to 1% of real market value (schools have a separate limit). Measure 5 does not void the earlier six percent limit; both were in effect after its passage. It is in this context— expenditures growing faster than revenues and a new property tax limit—that the city of Eugene decided to undertake an ambitious financial planning exercise that included substantial citizen involvement. This process was called Eugene Decisions and employed an innovative combination of public forums, budget balancing exercises, and representative surveys. Eugene Decisions sought to bring citizens together to render public judgment about alternative strategies for addressing the city’s budget deficit. By presenting citizens with information about the actual amounts of fees, taxes, and the budget shortfall, participants were presented with a picture of the true dilemma facing policy makers. There were three random sample surveys used to provide the City Council with a realistic picture of how citizens viewed city services and revenues. One survey was a budget template that led respondents through a realistic budget balancing exercise where they were provided an opportunity to balance the budget according to their own priorities, but respondents were constrained by real choices.3 Two other surveys were used to gauge citizen support for services and taxes. The Blue survey (named after the color of the paper on which it was printed) asked respondents to express their support for taxes for a variety
198
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
of government services, while it also presented the amount of the taxes paid by an average household. The third survey (called the Ivory survey) was similar to the Blue survey except it did not present the tax amounts. These surveys provided the city of Eugene with a wealth of information to help solve their budget problems. Although originally designed for this purpose, the Ivory and Blue surveys also provide the opportunity to test whether fiscal information, in the form of household tax amounts, influences support for taxes for services. For the purpose of this analysis, the two surveys (Ivory and Blue) were combined together and analyzed collectively. Both samples were drawn through the same process (a random sample of registered voters) at the same time. Both questionnaires asked questions about the same set of services, and asked respondents whether they strongly support, somewhat support, somewhat oppose, or strongly oppose the use of taxes for the services (see appendix A).4 The survey response rates for these mail surveys were extraordinarily high: 83% for Ivory and 76% for Blue.5 The influence of information on respondent’s preferences for services and taxes is examined through analyses of the combined Ivory and Blue surveys using logistic regression. Logistic regression uses independent variables to predict the probability that a choice will be made. In this case the choices predicted are the selection of a given level of support for taxes to pay for an indicated service. Logistic regression allows us to see the differences in responses between those questions with and without fiscal information, controlling for other important factors. In other words, it allows us to answer the question: “What effect does fiscal information have on the likelihood of support for taxes to pay for services, holding all else equal?” The two surveys that we analyzed were part of an effort by the city of Eugene to help them get citizen input about how to solve a structural budget problem. As such, the surveys represent a real life effort by the city of Eugene to solve their budget problem. The surveys, because of their structure, provide an interesting opportunity to see if providing information on the cost of providing public services influences public support. These surveys were designed to help solve a real-time fiscal problem, not as an experiment. As such, the questions are not exactly symmetrical, although they are extremely close. The main difference is that in the Ivory survey (the one without information) the questions are neutral whereas for the Blue Survey (with information) the questions are framed to maintain current levels (avoid reduction). We feel that implicit in the neutral question is the current level of service. Also, Kahneman and Tversky (1982) demonstrate that the value function for losses is steeper than for gains. In other words, if a question is presented as a loss avoidance it may be preferred to the same question presented as a gain. The effect of this for our work is that if a bias did exist, it would be toward a higher level of preference for the Blue survey that is framed as loss avoidance, and would make it more difficult to find significance.
Fee Services and Tax Services Questions were asked about 23 different services. They have been separated for analysis here into two distinct groups: those services for which fees are in place or could conceivably be charged (11) and those for which they could not be as easily imagined (12). Throughout this discussion these will be referred to as fee services (Library,
The Influence of Fiscal Information
199
Bookmobile, Swimming Pools, Athletics Programs, Community Centers, Outdoor Recreation, Hult (performing arts) Center, Specialized Recreation, Building Permits, Senior Citizen Center, Park Concerts) and tax services (Police, Fire, Special Events [e.g., street festivals], Sister Cities, Metropolitan Partnership, Downtown Development, Affordable Housing, Land Use, Neighborhood Groups, Human Services, Public Buildings [repair and maintenance]). These could also be conceptualized as individually consumed versus collectively consumed services. As a practical matter, the tax services are the ones for which the city of Eugene was not considering the possibility of a user fee. We understand that the distinction between collective (tax) and individual (fee) services is open to interpretation. See Appendix B for a full description of these services. The logistic regression model predicts support for taxes for the various services as a function of the characteristics of the survey respondents, their use of services, and whether or not they received information about the specific tax amounts. Tax Support ⫽ f (Fiscal Information, Respondent Characteristics, Service Use) Logistic regressions are estimated for each of the 11 fee and 12 tax services described earlier. Three different dependent variables are used: (1) whether the respondent strongly supports taxes for the service, (2) either somewhat or strongly supports taxes for the service, and (3) whether the respondent strongly opposes taxes for the service. The logistic regressions include the following independent control variables: 䡠 Age 䡠 Gender 䡠 Home ownership 䡠 Low Income 䡠 High Income 䡠 Household Size 䡠 Higher Education 䡠 Less Education 䡠 Often Use 䡠 Never Use 䡠 Use Often 䡠 Use Never
Expressed as natural logarithms 1 if female, 0 if male 1 if renter, 0 if home owner 1 if household income under $20,000, 0 if not 1 if income over $50,000, 0 if not Number of people in the household 1 if respondent had a bachelor degree or higher, 0 if not 1 if respondent has high school diploma or less education 0 if not 1 if respondent (or household member) often uses the service, 0 if not (fee services only) 1 if respondent (or household member) never use the service, 0 if not (fee services only) Number of fee services (0–11) which respondents report using often (tax services only) Number of fee services (0–11) which respondents report using never (tax services only)
Four of the variables listed refer to service use. For fee services, the Eugene Decisions surveys asked respondents to indicate their frequency of use. In the case of the tax services, the use variables are a count of the number of services that respondents report using often or never.
200
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Research Questions and Hypotheses We seek to answer two research questions: 1. 2.
What are the effects of fiscal information on citizen preferences for taxes to support city services? Does fiscal information influence different groups of people differently? For example, how do women with fiscal information differ from women without this information?
Based on the literature, we hypothesize that presenting fiscal information will reduce support for taxes for (or increase opposition to) government services. Therefore: H0 3 respondents with fiscal information have the same level of support for taxes for (or opposition to) public services as do respondents without fiscal information HA 3 respondents with fiscal information have a lower level of support for taxes for (or higher level of opposition to) public services compared to respondents without fiscal information Are one or two groups of respondents with strong antitax or antifee sentiments driving this effect, or are all groups equally deterred in the face of actual taxes? In addition to the general effects, we also test the differential effects of information among groups of individuals.6 If the influence of information is ubiquitous, we would expect it to show up across demographic groups, therefore: H0 3 Across demographic groups, respondents with fiscal information have the same level of support for taxes for (or opposition to) public services as do respondents without fiscal information HA 3 Across demographic groups, respondents with fiscal information have a lower level of support for taxes for (or higher level of opposition to) public services compared to respondents without fiscal information
FINDINGS—INFORMATION EFFECTS When the tax amount is known respondents are significantly less likely to support and more likely to oppose taxes for these services. The general effects of information are presented based on simple descriptive statistics. The percentages listed in Table 1 show the base rates for support for taxes for both surveys (Ivory survey does not present specific tax information, the Blue survey does provide tax information). The percentage of respondents indicating that they strongly support taxes for these services decreases for the Blue compared to the Ivory survey for 19 of the 23 services. The percentage of respondents indicating that they strongly oppose taxes for services increases for the Blue survey for all 23 of the services presented! The relative amounts of support for taxes varies for each of these services, but the pattern remains clear: whatever the initial level of support, it drops in the face of fiscal information. These differences are visible even in the absence of controls for other factors. The effect of information is measured by the logistic regressions. How much does fiscal information matter to respondent preferences? The short answer is a lot. For the vast majority of services, presenting the tax amount results in reduced support for taxes to
The Influence of Fiscal Information
201
Table 1. Percent Supporting or Opposng Services* Strongly Oppose Somewhat Oppose Somewhat Support Strongly Support Blue Ivory Survey Survey
Blue Survey
Ivory Survey
Blue Survey
Ivory Survey
Blue Ivory Survey Survey
Tax Services Police Fire Special Events Sister Cities Business Assistance Metro partnership Downtown Development Affordable Housing Land Use Planning Neighborhood Grps Human Services Public Bldg.
13.9 8.3 41.9 45.3 32.1 29.9 47.0 14.3 53.7 27.0 17.5 36.9
0.6 0.3 21.4 33.8 13.4 15.0 30.0 7.6 12.0 15.8 4.3 3.7
9.6 10.0 20.1 18.1 19.7 18.8 27.0 13.0 23.5 23.0 16.2 23.2
2.4 0.3 23.3 28.0 22.0 17.6 24.3 13.0 25.9 26.0 10.5 7.8
27.2 27.2 21.1 20.1 31.4 28.5 17.0 24.9 11.4 29.4 30.1 24.8
14.8 8.0 37.1 31.2 42.5 35.1 28.1 33.5 45.3 42.1 37.2 43.0
49.0 54.2 13.4 12.4 14.0 20.1 7.0 47.2 7.7 17.9 35.1 11.7
82.2 91.4 18.2 7.1 22.0 32.3 17.7 45.9 16.8 16.1 48.0 45.5
Fee Services Library Bookmobile Swimming Pools Adult Athletics Community Centers Outdoor Athletics Hult Center Specialized Rec. Bldg. permits Senior Centers Park Concerts
17.6 20.3 18.5 34.4 24.0 36.6 37.2 21.8 36.6 14.3 34.7
5.1 10.4 2.7 20.7 8.5 26.6 20.5 4.9 12.0 4.8 25.7
12.5 10.0 20.5 23.8 25.3 29.7 25.3 12.8 30.0 15.9 16.3
10.6 23.6 12.9 30.7 13.1 24.9 23.2 9.5 25.9 8.1 26.3
25.3 27.1 33.2 24.8 31.8 20.3 24.6 28.4 20.7 34.9 22.4
33.8 43.3 44.4 34.7 42.4 31.6 32.1 44.0 45.3 41.4 32.6
43.6 40.2 25.7 14.6 17.8 10.7 12.6 33.6 9.7 33.2 25.5
50.5 22.7 39.9 13.9 36.0 15.8 24.2 41.6 16.8 45.6 15.4
* Blue survey contains fiscal information; Ivory survey has no fiscal information.
provide the services. Table 2 shows that knowing the amount of the tax makes it significantly more likely that respondents will strongly oppose taxes for 11 of 12 tax services (all at least at the 0.01 level of significance)! Fiscal information also makes it significantly less likely that respondents will strongly support taxes for eight of twelve tax services. Somewhat or strong support for taxes decreases for eleven of the twelve services. The sole service that does not show this effect, Sister Cities, bears a tax amount of only $1. Among the fee services the information effect is equally stunning. Strong opposition significantly increases for every service when respondents know the tax amount. Strong support for taxes wanes once the tax amount is known for 7 of 11 services. Therefore, based on the aggregate results of the logistic regressions, for the vast majority of services we reject our null hypothesis and conclude that fiscal information lowers support for taxes for services (or increases opposition). Table 3 shows the change in the predicted probability of support for (or opposition to) taxes for services when fiscal information is provided. In many cases the changes are huge. For example, provision of fiscal information is predicted to increase opposition to
202
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Table 2. Information Effects Significant Coefficients Strong Support B Tax Service Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Planning Neighborhood Grps Human Services Public Bldg. Fee Services Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Athletics Hult Center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
Sig.
⫺1.6322*** ⫺2.1942*** 0.9033*** ⫺0.4465** ⫺0.499** ⫺0.9730*** ⫺0.7491*** ⫺0.5513*** ⫺1.7938***
Somewhat or Strong Support B
Sig.
⫺2.5236*** ⫺4.2004*** ⫺0.8213*** ⫺0.7088*** ⫺0.6605*** ⫺0.9309*** ⫺0.5182** ⫺1.9736*** ⫺0.5031*** ⫺1.2711*** ⫺2.5504*** ⫺0.9646***
1.0528*** ⫺0.5615*** 0.1217*** ⫺0.8866*** ⫺0.4562* ⫺0.7742*** -0.5130** ⫺0.4386** 1.0396***
⫺1.3081*** ⫺0.3900*** ⫺1.2829*** ⫺0.7682*** ⫺0.8539*** ⫺1.4230*** ⫺1.3599*** ⫺1.3131***
Strong opposition B
Sig.
3.9679*** 1.160*** 0.5719*** 1.2544*** 0.8876*** 0.7246*** 0.8654*** 2.5480*** 0.8665*** 1.6367*** 2.8588*** 1.5015*** 0.9105*** 2.6227*** 0.7386*** 1.2480*** 0.5150*** 0.9701*** 1.1960*** 1.6918*** 1.5766*** 0.3978*
*** Significance at the 0.01 level; ** significance at the 0.05 level; * significance at the 0.10 level. B is the logistic regression coefficient.
spending on land use planning by 32% (based on a starting probability of opposition of 50%). We can see clearly that once the amounts of taxes are known, citizens are less supportive. This finding is consistent with our common understandings of human behavior as well as the empirical research which has preceded this work. We all recognize the person who supports a variety of causes and efforts without hesitation, until it becomes time to do the work or to pay the bill. It seems to make sense that people might support generally that which they would not like to pay for specifically.
Influence of Fiscal Information on Sociodemographic Groups We again use logistic regression to assess the influence of information on different groups of individuals. In this iteration, respondents are compared among their own group. Women with fiscal information, for instance, are compared to women without information and those over 65 with fiscal information are compared to those over 65 without it, and so forth. Tables 4 and 5 show that all groups are affected, but some slightly more than
The Influence of Fiscal Information
203
Table 3. Information Effects Somewhat or Strong Support
Strong Support B
Sig.
Change From .5
B
Sig.
Change From .5
Strong Opposition B
Sig.
Change From .5
Tax Services Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Planning Neghborhood Grps Human Services Public Bldg.
⫺1.6322*** ⫺2.1942*** ⫺0.1265 0.9033*** ⫺0.4465** ⫺0.4991** ⫺0.9730*** 0.0891 ⫺0.7491*** 0.1645 ⫺0.5513*** ⫺1.7938***
⫺0.3365 ⫺0.3997 ⫺0.0316 0.2116 ⫺0.1098 ⫺0.1222 ⫺0.2257 0.0223 ⫺0.1790 0.0410 ⫺0.1344 ⫺0.3574
⫺2.5236*** ⫺4.2004*** ⫺0.8213*** ⫺0.0944 ⫺0.7088*** ⫺0.6605*** ⫺0.9309*** ⫺0.5182** ⫺1.9736*** ⫺0.5031*** ⫺1.2711*** ⫺2.5504***
⫺0.4258 ⫺0.4852 ⫺0.1945 ⫺0.0236 ⫺0.1701 ⫺0.1594 ⫺0.2173 ⫺0.1267 ⫺0.3780 ⫺0.1232 -0.2809 ⫺0.4276
3.9679*** 10.0333 1.1601*** 0.5719*** 1.2544*** 0.8876*** 0.7246*** 0.8654*** 2.5480*** 0.8665*** 1.6367 2.8588***
0.4814 0.5000 0.2614 0.1392 0.2781 0.2084 0.1736 0.2038 0.4274 0.2040 0.3371 0.4458
Fee Services Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Athletics Hault Center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
⫺0.1331 1.0528*** ⫺0.5615*** 0.1217*** ⫺0.8866*** ⫺0.4562 ⫺0.7742*** ⫺0.2957 ⫺0.5130** ⫺0.4386*** 1.0396***
⫺0.0332 0.2413 ⫺0.1368 0.0304 ⫺0.2082 ⫺0.1121 -0.1844 ⫺0.0734 ⫺0.1255 ⫺0.1079 0.2388
⫺0.9646*** 0.1197 ⫺1.3081*** ⫺0.3900** ⫺1.2829*** ⫺0.7682*** ⫺0.8539*** ⫺1.4230*** ⫺1.3599*** ⫺1.3131*** 0.2538
⫺0.2240 0.0299 ⫺0.2872 ⫺0.0963 ⫺0.2829 ⫺0.1831 ⫺0.2014 ⫺0.3058 ⫺0.2957 ⫺0.2880 0.0631
1.5015*** 0.9105*** 2.6227*** 0.7386*** 1.2480*** 0.5150*** 0.9701*** 1.1960*** 1.6918*** 1.5766*** 0.3978
0.3178 0.2131 0.4323 0.1767 0.2770 0.1260 0.2251 0.3721 0.3445 0.3287 0.0982
*** Significance at the 0.01 level; ** Significance at the 0.05 level; * significance at the 0.10 level. B is the logistic regression coefficient. Change from .5 denotes the predicted change form .5 due to the respondent having fiscal information.
others. Regardless of group, knowing the tax amount generally decreases the support for taxes for (increases opposition to) services. Therefore, for the vast majority of services, we reject our null hypothesis and conclude that fiscal information generally lowers support for taxes for services (or increases opposition to services) regardless of the demographic group. These results speak to the ubiquity and power of tax information’s influence on citizens. Those who know the tax amounts are significantly more likely to oppose the service than other members of their sociodemographic cohort, regardless of whether that group is women, men, seniors, homeowners, those from low or middle income households, or college graduates. However, some groups are more consistently affected than others. The elderly, women, and men seem particularly affected by fiscal information. Fiscal information has somewhat less of an affect on those who have a high school degree or less (fee services), renters, and high income respondents. The power behind this finding is simple: we are all the same, even with our differences, when it comes to how the amount of taxes affects our decisions.7
204
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Table 4. Information Effects Within Sociodemographic Groups Tax Services Logistic Regression Significant Coefficients HS Diploma with Information SERVICE Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Neighborhood Grps Human Services Public Buildings
College or More with Information
Strong Strong and Strong Strong Strong and Strong Support Somewhat Support Opposition Support Somewhat Support Opposition ⫺*** ⫺***
⫺*
⫺*** ⫺**
⫺*** ⫺** ⫺**
⫺*** ⫺*** ⫹**
⫺***
⫹***
⫺** ⫺* ⫺*** ⫺* ⫺** ⫺***
⫹*** ⫹* ⫹*** ⫹*** ⫹*** ⫹***
⫺* ⫺* ⫺*** ⫺** ⫺** ⫺***
65 or Over with Information SERVICE Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Neighborhood Grps Human Services Public Buildings
Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Neighborhood Grps Human Services Public buildings
⫺** ⫺** ⫺*** ⫺** ⫺*** ⫺*** ⫺*** ⫺***
⫹*** ⫹*** ⫹*** ⫹*** ⫹** ⫹* ⫹*** ⫹*** ⫹*** ⫹***
Men with Information
Strong Strong and Strong Strong Strong and Strong Support Somewhat Support Oppposition Support Somewhat Support Opposition ⫺*** ⫺***
⫺** ⫺** ⫺**
⫺* ⫺***
⫺*** ⫺*** ⫺** ⫺** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺** ⫺*** ⫺***
⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹** ⫹*** ⫹*** ⫹*** ⫹***
High Income with Information SERVICE
⫺*** ⫺*** ⫺***
Strong Strong and Strong Support Somewhat Support Opposition ⫺*** ⫺***
⫺** ⫺**
⫺* ⫺** ⫺***
⫺***
⫺** ⫺***
⫹** ⫹*
⫹* ⫹*** ⫹** ⫹*** ⫹**
⫺*** ⫺***
⫺***
⫹***
⫺**
⫹*** ⫹*** ⫹*** ⫹*** ⫹* ⫹* ⫹*** ⫹*** ⫹*** ⫹***
⫹** ⫺* ⫺**
⫺*** ⫺*** ⫺**
⫺* ⫺***
⫺*** ⫺** ⫺*** ⫺***
The Influence of Fiscal Information
205
Table 4. Continued Homeowners with Information SERVICE Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Neighborhood Grps Human Services Public Buildings
Renters with Information
Strong Strong and Strong Strong Strong and Strong Support Somewhat Support Opposition Support Somewhat Support Opposition ⫹*** ⫹***
⫺*** ⫺** ⫺*** ⫺*** ⫺*** ⫺***
⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺** ⫺*** ⫺*** ⫺*** ⫺***
⫺*** ⫺*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹** ⫹*** ⫹*** ⫹*** ⫹***
Police Fire Special Events Sister Cities Business Assistance Metro Partnership Downtown Dev. Affordable Housing Land Use Neighborhood Grps Human Services Public Buildings
⫹**
⫺***
⫹***
⫺*** ⫺***
⫹* ⫹***
⫺***
⫺***
Women with Information SERVICE
⫺***
Low Income with Information
Strong Strong and Strong Strong Strong and Strong Support Somewhat Support Oppposition Support Somewhat Support Opposition ⫺*** ⫺*** ⫺** ⫺* ⫺* ⫺** ⫺*** ⫺** ⫺***
⫺*** ⫺*** ⫺***
⫺*** ⫺***
⫺*** ⫺***
⫹***
⫹*** ⫹*** ⫹***
⫺** ⫺** ⫺** ⫺*** ⫺*** ⫺*** ⫺*** ⫺***
⫹*** ⫹*** ⫹*** ⫹** ⫹*** ⫹** ⫹*** ⫹***
⫺*
⫺***
⫹** ⫹*** ⫹*
⫺***
⫹***
⫺***
⫹***
⫹** ⫺***
*** Significance at the 0.01 level, ** significance at the 0.05 level, * significance at the 0.10 level (⫹) ⫽ positive coefficient, (⫺) ⫽ negative coefficient.
Does Fiscal Information Matter in the Ordering of Preferences? We find that providing fiscal information lowers support and increases opposition to taxes for city services. But does the presence of fiscal information change the ordering of preferences? That is, we now know that fiscal information lowers the level of support for taxes, but do the services that enjoy the greatest (or least) amount of support remain the same? Table 6 presents the rank ordering of the services from highest to lowest. Generally speaking, the popular services remain popular in the presence of fiscal information. The same is true of the least popular services. Table 7 shows the correlation coefficients for the percentage strongly supporting, somewhat or strongly supporting, or strongly opposing taxes for the services. The significant and high correlations indicate that although fiscal information typically reduces levels of support for taxes, the ordering of the preferences are not significantly changed.
206
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Table 5. Information Effects Within Sociodemographic Groups Fee Services Logistic Regression Significant Coefficients HS Diploma with Information SERVICE Library Bookmobile Swimming pools Adult Athletics Community Ctrs. Outdoor Prog. Hult Center Specialized Rec. Bldg. Permits Sinor Centers Park Concerts
Strong Support
Strong and Somewhat Support
Strong Oppose ⫹*
College or More with Information Strong Support ⫹*** ⫺*** ⫹*** ⫹*** ⫹*** ⫺*
Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Prog. Hult Center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
Strong Support ⫺**
Strong and Somewhat Support ⫺*
⫺***
⫺***
⫺***
⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺**
⫺** ⫺***
Strong Opppose
⫹*** ⫹** ⫹*** ⫹* ⫹*** ⫹*** ⫹*** ⫹**
Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Prog. Hult Center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
Strong Support
⫹* ⫺* ⫺** ⫺** ⫺*** ⫺*
⫹**
Strong Support
Strong and Somewhat Support
Strong Oppose
⫺**
⫹**
⫺* ⫺**
⫹* ⫹*** ⫹*
⫹*
Strong Oppose ⫹*** ⫹** ⫹*** ⫹*** ⫹* ⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹***
Men with Information
⫹***
High income with Information SERVICE
⫺*** ⫺* ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺***
⫹***
65 or Over with Information SERVICE
Strong and Somewhat Support ⫺***
Strong and Somewhat Support
Strong Oppose
⫺***
⫹*** ⫹** ⫹*** ⫹*** ⫹*** ⫹** ⫹*** ⫹*** ⫹*** ⫹***
⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫹**
The Influence of Fiscal Information
207
Table 5. Continued Homeowners with Information SERVICE Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Prog. Hult Center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
Strong Support ⫹*** ⫺***
Strong and Somewhat Support
Strong Oppose
⫺***
⫹*** ⫹*** ⫹*** ⫹*** ⫹*** ⫹** ⫹*** ⫹*** ⫹*** ⫹*** ⫹**
⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺*** ⫺***
⫺*** ⫺*** ⫺** ⫺** ⫹***
Renters with Information Strong Support
Library Bookmobile Swimming Pools Adult Athletics Community Ctrs. Outdoor Prog. Hult center Specialized Rec. Bldg. Permits Senior Centers Park Concerts
Strong Support ⫹*** ⫺**
⫺** ⫹** ⫺** ⫺*
⫺*** ⫺*
⫺** ⫹***
Strong Oppose
⫺**
⫹*** ⫹*** ⫹***
⫺*** ⫺* ⫺*** ⫺*** ⫺*** ⫺***
⫺*** ⫺* ⫺** ⫺***
⫹***
⫹***
⫹**
Low income with Information
Strong and Somewhat Support
⫺***
Strong Oppose
⫹**
Women with Information SERVICE
Strong and Somewhat Support
⫹*** ⫹*** ⫹*** ⫹*** ⫹***
Strong Support
Strong and Somewhat Support
Strong Oppose
⫺**
⫹** ⫹* ⫹* ⫹*
⫺** ⫺*
⫺*
⫹***
⫺*** ⫺* ⫺*** ⫺*** ⫺***
⫹*** ⫹*** ⫹***
⫹*
*** Significance at the 0.01 level, ** significance at the 0.05 level, * significance at the 0.10 level (⫹) ⫽ positive coefficient, (⫺) ⫽ negative coefficient.
DISCUSSION Fiscal information is ubiquitous in its effect on the tax and service preferences of citizens. The implications are that the very presence of the amount of a tax will, all else equal, decrease the likely support and increase the likely opposition for taxes for a service. The effect stands for citizens generally and also when comparing among groups. Fiscal conservatives who wish to limit government spending would be well served by publicizing the true costs of government services. On the other hand, those wishing to increase spending would do better with their program if the citizenry remained in financial ignorance. What does this tell us about what to expect from citizens in the polling booth or responding to opinion surveys? The decision about whether to include fiscal information is crucial to the level of support expressed for it. We could reasonably expect more no votes, particularly from those on the margin. These findings also suggest that an information campaign geared to influence voters’ opinions about services would not need to be
208
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Table 6. Service Rankings Strong Support Tax Services Fire Police Human Services Affordable Housing Public Buildings Metro Partnership Business Assistance Special Events Downtown Development Land Use Neighborhood Groups Sister Cities Fee Services Library Senior Centers Specialized Recreation Swimming Pools Community Centers Hult (Performing Arts) Center Bookmobile Building Permits Outdoor Program Park Concerts Athletic Programs Tax Services Fire Police Public Buildings Human Services Affordable Housing Metro Partnership Business Assistance Land Use Neighborhood Groups Special Events Downtown Development Sister Cities Fee Services Senior Centers Specialized Recreation Library* Swimming Pools* Community Centers Bookmobile Building Permits Hult (Performing Arts) Center Athletic Programs Park Concerts Outdoor Program
Ivory Survey
Blue Survey
1 2 3 4 5 6 7 8 9 10 11 12
1 2 4 3 10 5 7 8 12 11 6 9
1 2 3 4 5 6 7 8 9 10 11
1 4 3 5 7 9 2 11 10 6 8
1 2 3 4 5 6 7 8 9 10 11 12
1 2 8 4 3 5 7 12 6 9 11 10
1 2 3 3 5 6 7 8 9 10 11
2 4 1 5 6 3 11 9 8 7 10
The Influence of Fiscal Information
209
Table 6. Continued Tax services Fire Police Public Buildings Human Services Affordable Housing Land Use Business Assistance Metro Partnership Neighborhood Special Events Downtown Development Sister Cities Fee Services Swimming pools Senior Centers Specialized recreation Library Community Centers Bookmobile Building Permits† Hult (Performing Arts) Center Athletic Programs Park Concerts Outdoor Program
Ivory Survey
Blue Survey
1 2 3 4 5 6 7 8 9 10 11 12
1 2 8 4 3 12 7 6 5 9 11 10
1 2 3 4 5 6 7 8 9 10 11
3 1 5 2 6 4 9 11 7 8 9
Note: 1 Lowest percent, 12 (or 11 for fee services) highest percent. Ivory survey: contains no fiscal information. Blue survey: contains fiscal information. * Library and swimming pools both had 84.3% somewhat or strong support for the Ivory survey. † Building permits and Outdoor Program both had 36.6% strong oppposition for the Blue survey.
sophisticatedly targeted in order to be effective in eroding support for taxes for services In a small number of instances, the coefficients for both strong support and strong opposition are positive for the same service. An example of this is the Sister Cities program where knowledge about the $1 tax actually leads to an increased likelihood of
Table 7. Correlations Ivory and Blue Tax Services (N ⫽ 12) Strong Support Strong or Somewhat Support Strong Opposition Fee Services (N ⫽ 11) Strong Support Strong or Somewhat Support Strong Opposition
Correlations 0.88*** 0.78*** 0.72*** 0.66** 0.77*** 0.87***
** One-tailed significance at the 0.05 level. *** One-tailed significance at the 0.01 level. Correlations are based on the percent strongly supporting, somewhat or strongly suppporting, or strongly opposing each service.
210
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
both strong support and strong opposition to taxes. One explanation may lie in the distinction between two possible reactions from citizens faced with the amount of a tax. One sentiment is absolute tax aversion. The other resembles “sticker shock.” Those who are absolute in their aversion to taxes are affected minimally, if at all by the amount of such a tax because for them any tax is unacceptable. Knowing that some specific amount of tax will be charged, regardless of its size, makes these individuals more likely to oppose a service. At the same time, other citizens seem to have a threshold for tax support. For these individuals, the amount of a tax is assessed relative to some standard of reasonableness. If the tax is within some common limits of their spending patterns or expectations for a service they support it. If they are startled by how large a tax is, they don’t support it, just as potential car buyers may shake their heads and walk out of the showroom after seeing the price of a new vehicle. Where this gets interesting is when both groups are responding differently to the same service. Faced with a $1 tax to continue the Sister Cities program some citizens responded with aversion, despite the minimal size of the tax. At the same time, some portion of the citizenry were more likely to strongly support this tax only after knowing how much (or little) it cost. Perhaps for these people $1 was not a shock, but a reasonable amount to pay for something. This effect surfaces for three services in this research. An increased likelihood of support as well as opposition is seen for one of the tax services (Sister Cities with a $1 tax) and two of the fee services (Bookmobile with a $1 tax, Park Concerts with a $2 tax). Further explanation for the fee services may lie in the fact that respondents receiving information were presented elsewhere with the amount of fees necessary to eliminate the tax ($100 for Bookmobile and $4 for Park Concerts). To some the tax may have seemed a far more acceptable amount of money than the fee necessary to supplant it, particularly in the case of the Bookmobile. Our findings also suggests some support for the notion of the “free lunch.” Citizens may or may not always want ‘more for less’, but our research indicates they surely seem to prefer more when it is offered for free. We are reminded of the pizza commercials where focus groups of chimpanzees are asked whether they prefer one pizza for $9.95 or two pizzas for $9.95. We as citizens are no different than these chimps, we prefer more of a service as opposed to less of it. Our findings also suggest that citizen surveys are often methodologically flawed. The methodological flaw is the absence of a budget constraint—the link between service level and how much it costs. Or, stated another way, the absence of a budget constraint means that citizen surveys often systematically overstate how much respondents support a service. Our findings are consistent with the earlier work concerning budget constraints (Clark, 1974). The lesson for survey researchers is that the surveys that systematically overstate support for services because of the lack of fiscal information do not provide a realistic view of support for government services. Therefore, researchers should consider including some type of budget constraint if they wish to avoid overstatement of support. However, it may be that the ordering, or ranking, of services is not greatly affected by such overstatement.
The Influence of Fiscal Information
211
APPENDIX A QUESTION WORDING The following is an example of a question about fire from the Ivory survey: How much do you support or oppose using taxes to pay for. . . Fire and Emergency Medical Services. This includes fire protection, emergency medical, ambulance, and other rescue operations. Strongly Support [ ]
Somewhat Support [ ]
Somewhat Oppose [ ]
Strongly Oppose [ ]
No Opinion [ ]
The Blue survey asked similar questions but provided substantially more information on the fiscal impact on households, as follows:8 How much do you support paying the yearly tax to offer the current service level for. . . Fire and Emergency Medical Services include fire protection, emergency medical, ambulance, and other rescue operations. To avoid reducing these services, how much do you support an additional tax of $10 paid by every household in the city (including yours)? Strongly Support [ ]
Somewhat Support [ ]
Somewhat Oppose [ ]
Strongly Oppose [ ]
No Opinion [ ]
APPENDIX B TAX AND FEE SERVICES Tax Services 1. 2. 3. 4. 5. 6. 7. 8. 9.
Police Services. Police respond to police calls, enforce laws, and investigate criminal cases. Fire and Emergency Medical Services. Fire protection, emergency medical, ambulance, and other rescue operations. Special Events. Promotes events for the community. Sister Cities. Provides cultural exchange for Eugene’s sister cities in Korea, Nepal, Japan, and Russia. Business Assistance Team. Provides assistance to new and existing small businesses. Metro Partnership Contract. Works to attract new business to the Eugene area. Downtown Development. Promotes private development of downtown Eugene. Affordable Housing and Neighborhood Development. Develops new low cost and low rent housing and provides emergency shelter for homeless families. Land Use Planning Services. Manages the city’s land use rules including the historic preservation program.
212
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
10.
Neighborhood Groups. Provides opportunities for citizens to participate in decision making with the city. 11. Human Services Contracts. Assists homeless and low income people with food, shelter, medical attention, and safety. 12. Public Buildings and Facilities. Renovation and repair of city owned buildings and equipment.
Fee Services 1. Public Library. Public library services for the city. 2. Library Bookmobile. The bookmobile provides library services to those who have difficulty getting to the library on their own. 3. Swimming Pools. Swimming pools provide recreational opportunities, water safety instruction, and fitness activities. 4. Adult Athletics. Adult recreational programs such as softball and soccer. 5. Community Centers. The city’s four community centers provide year-round recreational classes (such as aerobic dancing) and meeting places for a wide range of community groups. 6. Outdoor Programs. These programs provide instruction in various outdoor activities such s rock climbing and skiing. 7. Hult Center. The Hult Center consists of two performing arts theaters that provide a place for local, national, and international artists to perform. 8. Specialized Recreation. This program provides recreational opportunities for people with disabilities. 9. Building Permits. This includes checking plans and issuing permits for residential construction. 10. Senior Centers. Senior Centers provide a range of recreational and social activities for city residents over 55 years old. 11. Concerts in the Park. Throughout the summer the city provides concerts at various parks.
NOTES 1.
2. 3.
4. 5.
For additional information on the of the determinants of citizen attitudes on taxes and spending see Citrin (1979); Courant et al. (1980); Ladd and Wilson (1982); Wilson (1983); Beck and Dye (1982); Beck et al. (1987); and Beck et al. (1990). Measure 5 has recently been superseded (1997) by Measure 50 that further limits property taxes. The results of the budget balancing exercise are not presented in this manuscript. For further information about the results of that survey see William Simonsen, “Citizen Preferences and Budget Policy,” paper presented at the Shaping the Urban Future: International Perspectives and Exchanges Conference jointly organized by the Urban Affairs Association, School for Advanced Urban Studies, University of Bristol and the Department of City and Regional Planning, University of Wales at Cardiff, July, 1994. Care was taken to appropriately match question and response categories. This resulted in 333 completed survey instruments for the Ivory survey and 304 completed instruments for the Blue survey.
The Influence of Fiscal Information
213
6.
The influence of information is tested using a dummy variable (1 if the respondent had fiscal information, 0 if not). The effects of information on various groups is tested by a series of dummy variables. Each set of analysis required the creation of a different set of dummy variables based on the questions being asked. When comparing information effects between groups, for instance, each group has at least two variables associated with it. In the case of seniors with information, variables were created for seniors with information while seniors without information was the comparison case. Logistic regressions were run, controlling for all other factors, for each level of support for each service for this group and the entire process was repeated for each of the subsequent groups (e.g., women, men, high income, low income, etc.). 7. But what about the differential effects of information between groups? Are there different patterns of influence when comparing, for instance, men with information versus women with information or seniors with information versus non-seniors with information? If the mere presence of information is powerful enough to dissuade those within most demographic groups, we might expect to see only a small effect when comparing different groups (such as men and women, those 65 and over and those younger than 65, etc.). This is largely true (results not shown). The only group that is consistently more likely to strongly oppose taxes for these services (and less likely to strongly support them) across both the fee and tax services, is the senior set. Those 65 and over with information are significantly more likely than those under 65 with the same information to strongly oppose taxes for five of the tax supported and eight of the fee services. They are also significantly more influenced in their decisions to strongly, or strongly or somewhat support services for eight of the tax services and nine of the fee services. Interestingly, one of the nine is Park Concerts, for which seniors with information are more likely to support it than nonseniors. (We can only surmise, but suppose that this particular effect may vary from place to place with the type of music being played!). For fee services, men with information are consistently different from women with information in this analysis. For those with fiscal information, men are less likely than women to support (strongly or somewhat strongly) taxes for Bookmobile, Outdoor Recreation, Hult (performing arts) Center, Specialized Recreation and Park Concerts. Men are significantly more likely than women to strongly oppose these programs and services: Bookmobile, Pools, Community Centers, Outdoor Programs, Specialized Recreation and Park Concerts. This data suggests that men (with or without information) are distinctly less supportive and more opposed to services. An interesting observation about gender differences is that in no instances are there services that men are more likely to support than women. 8. The service description is not included for fee services (described later) in the Blue survey since the descriptions are presented earlier in the survey.
REFERENCES Arrington, T. S. and D. D. Jordan. (1982). Willingness to Pay Per Capita Costs As A Measure of Support for Urban Services. Public Administration Review, 42: 168 –171. Clark, T. N. (1974). Can You Cut A Budget Pie? Policy and Politics, 3: 3–31. Kahneman, D. and A. Tversky. (1982). The Psychology of Preferences. Scientific American, 246: 160 –175. McIver, J. P. and E. Ostrom. (1976). Using Budget Pies to Reveal Preferences: Validation of a Survey Instrument. In Clark, T. N. (Ed.), Citizen Preferences and Urban Public Policy. Beverly Hills, CA: Sage.
214
THE SOCIAL SCIENCE JOURNAL Vol. 37/No. 2/2000
Welch, S. (1985). The ‘More For Less’ Paradox: Public Attitudes on Taxing And Spending. Public Opinion Quarterly, 49: 310 –316. Wilson, L. A. (1983). Preference Revelation and Public Policy: Making Sense of Citizen Survey Data. Public Administration Review, 43: 335–343. Winter, S. and P. E. Mouritzen. (1997). Why People Want Something For Nothing: The Role Of Asymmetrical Illusions, unpublished manuscript.