J. OF PUBLIC BUDGETING, ACCOUNTING & FINANCIAL MANAGEMENT, 14(3), 445-461 FALL 2002 A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION Mark D. Robbins and Bill Simonsen* ABSTRACT. In this article we explore two citizen-based approaches to solving the problem of selecting a desirable level of public goods for a jurisdiction. The first approach seeks to inform decision-makers about citizens’ preferences by observing the choices of citizens faced with the actual budget constraint facing the government and asking them to choose service levels within that constraint. The second approach gauges citizens’ willingness-to-pay for their share of the cost of a desired level of public expenditure. In an effort to foster discussion and research into new modes of citizen participation in resource allocation we pose a model that combines both the constraints of the jurisdiction with the tax share of the respondent into a survey methodology that will reveal the underlying demand for government services in ways that are useful for public managers. INTRODUCTION Typically, public managers or elected officials select the amount of public goods and services to be provided and how much citizens will pay for them. This is because there is no observable demand schedule for many public services. That is, the underlying demand schedules for collectively consumed public services, such as police protection, are masked because they are paid for collectively through taxes. The resulting disconnect between the level of service provided and the amount preferred by citizens is one form of what is sometimes referred to as the ‘public goods ………………….. * Mark D. Robbins, Ph.D., is an Assistant Professor in the Institute of Public Affairs and Department of Political Science, University of Connecticut. His research focuses on municipal finance and citizen preference revelation. Bill Simonsen, Ph.D., is Professor and MPA program director for the Institute of Public Affairs, University of Connecticut. His research focuses on public finance and budgeting, and citizen preference revelation. Copyright 8 2002 by PrAcademics Press A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 446 problem.’ Citizen participation mechanisms help to reveal to officials the underlying preference structure of the citizenry and provide decisionmakers with a picture of the underlying demand for public services. However, the state of the art for involving citizens in these resource allocation decisions is in its early stages of development. There has been only a limited amount of research on citizens’ willingness-to-pay for various services. Further, only a small group of methodologies provides citizens with a realistic approximation of the complex trade-offs facing decision-makers. In this article we discuss the above literature and propose extensions of this research. Specifically, we propose a new survey method that queries respondents within a truly realistic decision environment; one that includes both government trade-offs and individual costs. In the end, it is left to elected and appointed public management officials to select service and revenue levels on behalf of a jurisdiction. The process of selecting these officials (voting) is the most rudimentary citizen participation and preference mechanism. However, the study of voting behavior is limited in its ability to reveal citizen tax and service preferences. As a revelation mechanism it is crude, and rarely directly addresses questions of taxation and service level. Even when making choices with direct tax implications (such as when casting votes in a bond referenda) citizens do not have complete information about the decision environment facing the government (what are the trade offs?), and the personal implications of their choices (how much will this choice cost this particular voter?). In most circumstances, however, voting for representatives may be enough to satisfy citizens that their interests are acted upon. In practice, the public goods problem is only a public management problem when the need for citizen participation in resource allocation is acute (high levels of controversy and large impacts on taxpayers). When the stakes are small, or general agreement seems pervasive, public managers may be comfortable imposing their choices on the municipality without investing in a citizen participation mechanism. However, many of the decisions that governments face, particularly about resource allocation, are complicated, have high long run (and perhaps short run) costs, and generate passionate disagreement. These are the times when knowledge of the ‘true’ preference structure of an informed citizenry seems most valuable. The purpose of this article is to propose a A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 447 methodology that best provides decision-makers with such a ‘true’ picture of citizen preferences and the consequent demand for public services. In the pages that follow, we lay the groundwork for a preferable method of preference revelation by examining the literature regarding citizen preference surveys generally. Next, we describe techniques that present budget decisions in their complexity. We then propose a methodology that improves on the current state of the art in citizen preference revelation. Finally, we discuss the implications of our proposed methodology for budgeting and policy-making. WHAT WE HAVE LEARNED FROM CITIZEN PREFERENCE SURVEYS Citizen involvement in resource allocation decisions can take place in many ways. In most cases, citizen input is not systematically included. These methods include public hearings, citizen committees, and advisory boards.1 Attendance at public hearings is generally low. Further, and more importantly, these methods do not represent what citizens at large would choose—rather they indicate the preferences of the individuals who are most motivated to participate. Citizen preference surveys typically ask respondents how much they support various services or how satisfied they are with them. Miller and Miller (1991) analyzed 261 citizen surveys about service delivery in forty states and found consistently good rankings. Without being faced with the real budget constraints and service needs facing decisionmakers, or the cost of their choices, citizens appear to find acceptable any government services that don't offend them. Little is learned from such efforts. There are a group of methodologies that present citizens with the realistic constraints governments' face when making resource or budget allocations. We describe many of these methods in greater detail in our book Citizen Participation in Resource Allocation, and so we will only discuss a few here (Simonsen & Robbins, 2000).2 In this article we are interested in the citizen involvement techniques most likely to be helpful to governments trying to solve the public goods 448 ROBBINS & SIMONSEN problem in accordance with citizen desires. These techniques must go beyond the 'typical' citizen survey and attempt to provide the most realistic representation of the public will. These more sophisticated techniques fall into two broad categories: those that present the constraints that governments face in their complexity and those that attempt to determine the willingness-to-pay of respondents based on information about the prices they face for government services. We will discuss these two techniques in turn, but pause first for a word about the information rich environments of these methods. The theoretical assumption underlying our approach is that the inclusion of information changes the decision environment for citizen respondents. This assumption has been established empirically in the research described below. Our normative assertion is that in order to obtain meaningful citizen input, citizens must be informed. The assertion that citizens, absent information, are not prepared to represent their own desires is, in our opinion, less arrogant than it may at first sound. Just as the heart patient would not be asked to pick between animal and synthetic mitrol valves without first learning more about the implications, it seems inappropriate to query citizens about taxes and services without facing them with their corresponding consequences. Pig or plastic? Service cuts or tax hikes? To shrug and choose either is to set in motion in a cavalier manner a set of actions that have profound implications. The decision method (frivolous and simplex) is mismatched to the decision impact (profound and complex). The methods described in this article gauge citizen preferences under information conditions of various levels of complexity. We now turn to those methods that take into account these various levels of complexity. Techniques That Reflect the Complexity of Budget Decisions The thread that ties together this group of techniques is their attempt to represent the trade-offs that government officials face when making budget decisions. Pioneering work in this area was completed by Terry Clark using budget pies (Clark, 1974). Budget pies are one way to impose a budget constraint, by requiring that respondents allocate funds for various services with the total equaling 100% (a circle, or pie, is presented to illustrate this choice). The strength of the budget pie approach is that it presents the budget to the respondent in a way that A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 449 represents the actual constraints facing the government. Two potential weaknesses are the resource and financial cost of conducting such processes, and the difficulty in getting members of less educated and lower income groups to participate (McIver & Ostrom, 1976). Citizen juries and panels are also techniques that attempt to present government decisions in their complexity. Users of these techniques recognize that citizens generally do not have the time and energy to understand complex government decisions but that a smaller group of systematically selected people can deliberate about these issues. Since these people are representative of the general population, their judgments reflect what the larger group of citizens would choose if they were also presented with the same information. A noteworthy example of the citizen panel technique was designed by Kathlene and Martin (1991) and took place in Boulder, Colorado. In this case, a panel of representative citizens deliberated transportation policy choices, some of which were very controversial (the process also included interviews and surveys). The panel participants were provided with information about the transportation choices important for making informed decisions. In this way, the panel represented the sentiment of an informed citizenry. The Jefferson Center for New Democratic Processes implemented a notable example of the citizen jury technique on the federal budget deficit (Jefferson Center, 1993). A small (representative sample) group of Americans heard testimony about the budget deficit, including statements by Democrats and Republicans. After hearing testimony, the citizen jury deliberated and came to a judgement. (In this case, they chose to cut more than Republicans would have, and raised taxes higher than the level supported by the Democrats.) Techniques that reflect the real decision environment place respondents in a position to understand the difficult trade-offs that characterize much of the government resource allocation process. They also rely upon sampling techniques to secure the claim of representativeness for the choices made by respondents on behalf of the citizenry. However, they don't focus directly on the respondents' willingness-to-pay for government services. In the next section we 450 ROBBINS & SIMONSEN discuss those methodologies that measure respondent’s willingness-topay. Techniques Assessing Willingness-to-Pay Another approach to gauging preferences is to ask about willingnessto-pay for services. One way to measure citizen willingness-to-pay is through contingent valuation; a complex system designed to reveal each respondent’s personal willingness-to-pay. Respondents are taken through a series of nested decision trees designed to determine the dollar value they attach to certain public assets. At each subsequent decision point the boundaries of respondent’s willingness-to-pay are narrowed until a clear maximum is determined for each benefit. Contingent valuation was developed to establish the benefit levels derived from purely public, and intangible goods not traded on markets, such as the benefits derived from the existence of a wilderness area. This technique is most helpful where benefits are unknown or hard to measure. This is the case for many public goods in addition to natural resources. A simpler approach was adopted by Arrington and Jordan (1982) who estimated citizens’ willingness-to-pay for government services in Mecklenburg County, North Carolina. Respondents were asked whether they were willing to pay certain amounts for services if government did not provide them. They were presented with the per capita cost of 20 services. A control group was given questions that corresponded to these without the fiscal information (Table 1). Their results revealed that “for virtually every [government service] activity the support was less when respondents were asked whether they would pay the costs directly.” (p. 169). The authors theorize that the large drop in support for police, tax office, court system and elections services is likely explained by the absurdity of directly paying for them. The results are consistent for people of varied gender, party affiliation, race and age. The only areas where these differences were not significant were for the lowest cost services, and for fire, which the authors note “still seems like a bargain at $30 per year.” (p. 169). Glaser and Hildreth (1996) relate willingness-to-pay taxes to the use of park and recreation services. Many of their respondents' willingness- A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 451 TABLE 1 Arrington and Jordan Survey If local government did not currently provide for it, would you be willing to pay: 1. $1.60 per year to support an animal shelter? 2. $1.40 per year for parks and recreation? [Experimental Group] Do you believe that it is the responsibility of local government to: 1. Operate an animal shelter? 2. Provide parks, golf courses and recreational facilities? [Control Group] Source: Arrington, T. S., Jordan, D. D. (1982). “Willingness to Pay Per Capita Costs as a Measure of Support for Urban Services.” Public Administration Review, 42 (2), 169. to-pay corresponded to service use. However, they also found many heavy users with low levels of willingness-to-pay as well as the reverse: low service users with higher levels of willingness-to-pay. In our book we report on three surveys of Eugene, Oregon, residents. This effort combined the two above mentioned approaches: budget pies and willingness-to-pay. Eugene was facing a substantial budget shortfall and wanted citizen input into how to achieve budget balance.3 One of the three surveys allowed respondents to build their own balanced budget (called the BOB survey). This survey was similar to the budget pie method. Respondents were presented with the current cost of services. They were then allowed to reduce services or raise revenues in order to erase the city’s $8 million projected budget deficit. The second survey (called “Blue” because the paper it was printed on was blue) asked respondents about their support for various services and provided the approximate household cost. The third survey (named “Ivory”) asked respondents about their support for services without providing a budget 452 ROBBINS & SIMONSEN constraint or cost of services. Response rates for these surveys were quite high; in excess of 50% for the BOB survey and about 80% for the Blue and Ivory surveys. One drawback of the budget pie approach mentioned earlier is achieving a high response rate among groups with low socio-economic status. The BOB survey substantially overcame this problem evidenced by satisfactory and uniform response rates for individuals with household incomes under $20,000. Similar to the result of the citizen jury that deliberated the federal budget deficit, BOB survey respondents generally favored both revenue increases and service reductions. The median respondent favored 38% service cuts and 62% revenue increases to solve the budget problem. Further, respondents most often chose to cut services where user fees were also possible and where the benefits were obscure. Public safety and social programs were the most preferred. We combined the Blue and the Ivory surveys to test the cost effects on willingness-to-pay for services, since only one survey contained cost information. The survey results indicated a remarkable difference in support depending upon whether the service cost was revealed to respondents. In the vast majority of cases, support fell significantly in the face of cost information. These findings mirror Arrington and Jordan's results. So, what have we learned about the use of preference surveys for gauging citizen demand for government services? We know that unless faced with some form of budget realism, either through a budget pie approach or by being given fiscal information directly, citizens are likely to appear globally satisfied with government service provision. Once faced with cost information, service support decreases, so omitting it will bias upward the levels of support. We have also seen that complicated surveys can still yield high response rates. We propose a further variant of the survey of citizen preferences for government revenue and spending. Our approach faces each respondent with his/her own tax price for each of the service levels selected. Next, we review the notion of tax price and then describe this survey design. A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 453 DESIGNING A DYNAMIC METHOD OF PREFERENCE REVELATION Using Information About Tax Prices Citizens have an incentive to understate their demand for services when they believe that such expression will influence how much they pay for them. At the same time, we have learned that citizen support for services is significantly weaker in the face of information about the magnitude of the taxes and fees necessary to support them. We cannot solve directly the problem of understatement, but we can control for it by allowing citizens to identify directly the service mix and level that they desire, given the costs that they individually would face. The primary revenue source for most local government services is the property tax. The property tax faced by any given household can be expressed as their share of the total amount of taxes collected by their jurisdiction. This is referred to in the public finance economics literature as the tax share. The tax share for an individual in a property tax supported system is determined by the assessed value of their property relative to the total assessed value in that jurisdiction. Ti = E* [AVi / 3AV] If E* is the chosen jurisdiction wide expenditure on a particular public good then the tax share for individual i (TSi) is the product of that expenditure level and the ratio of i's property value to the total assessed property value in the jurisdiction. Each individual's cost for an additional dollar of public service consumed by their household (assuming that services are consumed in equal proportion by households) is their tax price for that unit (Stiglitz, 2000, p. 142). For the purposes of survey research, this means that a change in the desired expenditure level E* will affect the budget constraint of each individual differently, based on their tax price. Survey methods that seek to present citizens with the cost implications of their tax and service selections should face them with information that reflects the change in tax price associated with changes in service consumption. However, this 454 ROBBINS & SIMONSEN requires a dynamic approach where the change in tax price due to service changes is calculated and presented to the respondent. A failure to do so forces the assumption that all individuals in the jurisdiction face the same tax price. The weakness of this assumption increases with the heterogeneity of the jurisdiction. Much of the work looking at tax price describes their variation and distribution across communities. Some have used this cross-sectional approach to estimate a demand curve for local public services based on the relationships between price and government expenditures (O'Sullivan, 1996; Inman, 1979). In other words, it assumes that the preferences of the population (or the median voter) are a key determinant of the amount of service provision. Average effects across communities are difficult to gauge since communities have different service responsibilities. There is evidence that people act as if they understand their tax price and the differences in service provision. Specifically, there is evidence that people sort themselves into communities based on their demand for public services (O’Sullivan, 1996). This is particularly true the greater the number of municipalities available from which to choose (Gramlich & Rubinfeld, 1982). A Dynamic Approach Two of the studies we mentioned that attempt to measure willingness-to-pay used different measures of tax price. Arrington and Jordan used per capita costs of the services in their survey. This is a tax price for the average person if all persons shared the tax–which they don’t. The Blue survey asked about services in relation to average household taxes. This is the tax price if you live in the average household–which no one does. This is a better measure because the household figure is more likely to correspond with a respondent’s household tax share. In both cases individual respondents were not faced with their own tax price, but a proxy based on some average for the jurisdiction. Responses, therefore, are different from what they would be if the respondents’ individual tax price were presented. Our interest is in determining citizen willingness-to-pay for public services in a full information environment. In other words, what set of public services would a representative set of citizens be willing to buy? A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 455 The best data for definitively revealing these preferences would be broadly representative, would contain much information about the respondents, would record their service preferences in a full information setting, and would contain a complete set of environmental control variables. The willingness-to-pay methodology and the provision of a budget constraint are related approaches. The budget pie presents the respondent with information that reflects the government’s budget constraint. The willingness-to-pay approach presents the respondent with an approximation of his/her tax price (per capita cost or average household cost, for instance) and support for services. This assumes that the tax price is equivalent for all households, an assumption that is almost always false. (In the case of contingent valuation methodologies, all possible prices are potentially considered regardless of the actual price facing the respondent). Therefore, it would be better to combine these approaches–present the situation given the real trade-offs facing the government, present the respondent with his/her tax price, and find out his/her service preferences given these constraints. Given the recent advances in computer and Internet technology, combining these elements is now possible. FIGURE 2 Surveying Citizen Preferences for Public Services Public Service dynamics Government specific Budget constraint Respondent specific budget constraint based on tax price Computer and Internet survey technology Respondent preferences Willingness to pay and demand for public service 456 ROBBINS & SIMONSEN The government specific and respondent specific budget constraints can be combined using computer technology that provides real time computation of the effects of respondent choices. This leads to a better measure of willingness-to-pay and demand for public services. An ideal survey instrument would retain key features of the ‘Build your Own Budget’ (BOB) survey from Eugene as it would query residents about their preferred service and revenue combinations given a balanced budget requirement and real cost information from the jurisdiction. The interactive computer based citizen budget survey method would query each respondent about his/her desired combination of taxes and services in a graphical environment that continuously presents information about the tax price for that service and the combined service package. We present below a more detailed proposal of how such a methodology should be constructed. This technique is likely to be costly, but in all likelihood not more so than the sophisticated methods we mentioned earlier.4 Proposed Elements of a Dynamic Tax Price Survey Sampling Frame. A random sample of citizens (perhaps voters) is useful for informing the public and elected officials of the desires of the electorate. While a national sample has appeal for making broad conclusions about citizen preferences it is less useful when researching attitudes towards local government. Another reason that a national sample is not optimal is due to the possibility of Tiebout (moving and sorting) behavior. In other words, as mentioned earlier, there is evidence people may select a jurisdiction because of the service level and tax price present in that municipality. In this context, it is not meaningful to assign the same underlying preference structure to the residents of different jurisdictions. Pre-Survey Information Gathering. Some information about the respondents themselves needs to be included in the dataset–even prior to surveying the respondents. Under this approach, information about the property tax burden of each household is collected in advance so that each respondent’s service selections during the survey is accompanied by his/her tax price for the service.5 A complete set of demographic A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 457 information is collected as well, including age, race, income, education, housing tenure, vocation and family constellation. Jurisdiction specific information regarding revenue sources and the costs of services is gathered from the local government. Additional information about the way that public services are constructed and delivered is necessary to determine the marginal effects of the revenue increases or decreases that respondents may select. A unique aspect of this approach is the record of respondents’ choices when faced with the host of local services in their jurisdiction, their true costs, the costs to the respondents, and the administrative and legal constraints facing the jurisdiction. Dynamic Survey Design. One innovation in this survey design is the real-time update of the information provided to the respondent. Respondents completing a survey face a computer screen with windows revealing 1) their total tax bill, 2) their share of the cost for the current service under consideration, and 3) the amount (quality) of the service to be provided. The information in these windows is updated with each choice. Respondents increasing the allocation for police services, for instance, see a corresponding increase in the number of police patrols. Survey Process. Respondents are contacted by phone and asked to complete a budget balancing exercise for their government (likely in exchange for a small fee). They are given a code that they can use to access the survey over the Internet, or are scheduled for an appointment at a centrally located computer lab. Computer assisted telephone interviewing (CATI) is an alternative used to complete the same exercise while the respondent is on the phone; the interviewer verbally provides verbally the information from the dynamically updated windows. Survey Elements. Using this method, respondents are queried in three areas: 1) Service use and experience. These questions focus on the satisfaction with a service and a record of how often in the past year the respondent has used it.6 2) Preferences for service allocation levels: Respondents choose the mix of services that satisfies them, given the impacts each choice 458 ROBBINS & SIMONSEN makes on their overall tax bill. Service reduction and expansion is permitted, as is indefinite iteration through the survey. Each iteration reveals the impacts of both tax price and service level on the quality level (outcomes) of services selected. 3) Respondent demographics: Survey participants are asked to provide information about themselves, their families, and their occupations. Survey Analysis. The surveys could be analyzed descriptively, using frequencies and measures of central tendency and dispersion. However, the core analysis of such data should be completed using a host of multivariate techniques focused on the identification of key determinants of different citizen choices. The core set of regressions would examine allocation levels preferred by citizens given their tax price, and control for their experience with the service, the quality of the service and respondent demographics. Unlike many preference surveys where strength of support is indicated creating an ordered hierarchy of responses for analysis, the data resulting from this survey method would produce completely continuous dependent variables. DISCUSSION Public decision-makers face the problem of selecting a mix of public services and offsetting revenues for their jurisdiction. In circumstances where the need for citizen participation is most acute, methods are needed that provide a reliable and representative sense of what an informed citizenry would choose. Surveys of representative blocks of the public can be designed to produce an accurate view of citizen preferences. Surveys that have attempted to measure willingness-to-pay for government services typically do not present respondents with their actual tax price, nor do the surveys capture the complexity facing government decision-makers. Conversely, surveys designed to present respondents with the government decisions in their complexity typically do not include their tax price. Both approaches are tepid when they try to provide useful guidance for government decision-makers. The benefit of our proposed survey design is that it reveals the demand schedule for public services more accurately than has ever been A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 459 done before, through the use of interactive, real time computer (or Internet–based) technology for survey completion. Our approach faces respondents with their own costs for the services they select and the impacts of their choices on the level of services to be provided. This information is provided immediately in response to the choices made. This allows the respondents to settle into the set of revenues and expenditures that they find satisfactory while at the same time reflecting the constraints facing decision-makers. Consumers of data produced by such a method should recognize that respondents are made 'different' from the general population because they are basing their decisions on their actual tax price. In effect, the respondents’ choices are representative of what an informed set of citizens would choose if they knew the tax price and service implications of their choices. As such, this method is likely to be a poor predictor of voting behavior since most citizens do not posses such detailed information. We should also recognize that there might be a price to be paid for the realism of such an exercise. The more realistic the method, the greater the risk that respondents will view their expressed preferences as decisive. This creates an incentive for the respondent to understate the strength of a respondent’s demand for services in the hope of avoiding the need to pay for it. To the degree that this occurs, willingness-to-pay might be understated in the resulting data. A dynamic survey method of preference revelation satisfies many of the concerns about the quality of citizen input (that other methods cannot) by gathering data from a representative sample of the opinion of truly informed citizens. Ideally, it would be the information that decision-makers use to guide them through the difficult choices when the stakes are high and agreement is low. The method is complex, would be difficult to implement, and would be costly. It would also give decisionmakers something that they have heretofore gone without: a reasonable claim to the informed will of the public about revenue and budget allocations. NOTES 1. Non-representative methods, such as 'come one, come all ' public forums, cut out newspaper surveys, non-representative focus groups, etc. are not designed to provide an accurate representation of the 460 ROBBINS & SIMONSEN public will, and therefore could be misleading to decision-makers. Such efforts, however, may be justified on process grounds to the extent that they help engage the public. 2. Specifically, see Chapter 2 “Contemporary Techniques for Citizen Involvement” of Simonsen and Robbins (2000). 3. The surveys were co-designed by Edward C. Weeks and William Simonsen. 4. Computer assisted telephone surveys could accomplish the same result. In this model the person administering the survey uses a specially written computer program to calculate the changes in tax price resulting from the respondents' choices. The respondent is then told the service and tax impacts of his/her service preferences. 5. Ideally, this would mean knowing the market value and assessed value of each property and the degree to which the property tax was capitalized into the respondent’s housing costs. Information on nonresidential property values would also be needed. The renter’s tax burden could be determined by a preliminary set of regressions of rent upon a set of community characteristics including property tax. 6. We found in our analysis of the Eugene, Oregon, surveys that service use influenced citizen support for services. REFERENCES Arrington, T. S., Jordan, D. D. (1982). “Willingness to Pay Per Capita Costs as a Measure of Support for Urban Services.” Public Administration Review, 42(2) 168-171. Clark, T. N. (1974). "Can You Cut A Budget Pie?" Policy and Politics, 3(2), 3-31. Glaser, M. A., & Hildreth, W. H. (1996). “A Profile of Discontinuity Between Citizen Demand and Willingness To Pay Taxes: Comprehensive Planning For Park and Recreation Investment.” Public Budgeting & Finance, 16(4), 96-124. A DYNAMIC METHOD OF CITIZEN PREFERENCE REVELATION 461 Gramlich, E. M., & Rubinfeld, D. L, (1982). “Microestimates of Spending Demand Functions and Test of the Tiebout and MedianVoter Hypothesis.” Journal of Political Economy, 90(6), 536-560. Inman, R. (1979). “The Fiscal Performance of Local Governments” (pp. 270-321). In P. Mieszkowski, & M. Straszheim (Eds.). Current Issues in Urban Economics. Baltimore, MD: Johns Hopkins University Press. Jefferson Center. (1993). Americas Tough Choices Citizens Jury; Report On Federal Budget Panel With Appendices. Minneapolis, MN: Author. Kathlene, L., & Martin, J. A. (1991). “Enhancing Citizen Participation: Panel Designs, Perspectives, and Policy Formulation.” Journal of Policy Analysis and Management, 10(1), 45-63. McIver, J. P., & Ostrom, E. (1976). “Using Budget Pies to reveal Preferences: Validation of a Survey Instrument” (pp. 87-110). In T. N. Clark (Ed.). Citizen Preferences and Urban Public Policy. Beverly Hills, CA: Sage. Miller, T. I., & Miller, M. A. (1991). "Standards of Excellence." Public Administration Review, 51(6), 503-514. O’Sullivan, A. (1996). Urban Economics. Chicago, IL: Irwin. Simonsen, W., & Robbins, M. D. (2000). Citizen Participation in Resource Allocation. Boulder, CO: Westview Press Stiglitz, J. E. (2000). Economics of the Public Sector (3rd ed.). New York: W.W. Norton and Company.
© Copyright 2024