Public Finance in Democratic Process: Fiscal Institutions and Individual Choice
By James M. Buchanan
- Ch. 1, Introduction
- Ch. 2, Individual Demand for Public Goods
- Ch. 3, Tax Institutions and Individual Fiscal Choice
- Ch. 4, Tax Institutions and Individual Fiscal Choice
- Ch. 5, Existing Institutions and Change
- Ch. 6, Earmarking Versus General-Fund Financing
- Ch. 7, The Bridge Between Tax and Expenditure in the Fiscal Decision Process
- Ch. 8, Fiscal Policy and Fiscal Choice
- Ch. 9, Individual Choice and the Indivisibility of Public Goods
- Ch. 10, The Fiscal Illusion
- Ch. 11, Simple Collective Decision Models
- Ch. 12, From Theory to the Real World
- Ch. 13, Some Preliminary Research Results
- Ch. 14, The Levels of Fiscal Choice
- Ch. 15, Income-Tax Progression
- Ch. 16, Specific Excise Taxation
- Ch. 17, The Institution of Public Debt
- Ch. 18, Fiscal Policy Constitutionally Considered
- Ch. 19, Fiscal Nihilism and Beyond
Some Preliminary Research Results
Little is known about individual actions and attitudes in collective, and specifically fiscal, choice. Scholars have simply not been interested in the behavior of individuals as political decision-makers. Once a democratic model for the political order is accepted, however, the gaps in our knowledge become apparent, and the need for many man-years of research is evident. Once we acknowledge that individuals as voters, or potential voters, in a broadly democratic political order ultimately determine the size of the public economy along with its composition, we are obligated to try to find out as much as we can about their choices.
Very little research has been done; what has been done is widely scattered; much of this remains incomplete, and the questions asked have been the wrong ones. Despite all this it may be helpful to report provisionally on what has been accomplished; this will in itself point up the need for further effort rather than indicate definitive conclusions.
Information and Ignorance: The Market Analogue
Several preliminary steps must be taken before systematic attempts can be made to formulate testable hypotheses. In this respect, the earmarking analysis used as our reference example in the preceding chapter is not characteristic. Elemental gaps in our knowledge must be filled in before anything like a sophisticated set of hypotheses about fiscal behavior can be developed.
The first thing that we need to know is the amount or degree of information possessed by individual citizens when they make actual or potential fiscal choices. How much do individuals know about the fiscal institutions under which they live? Earlier we have utilized individual behavior in ordinary market choice as a sort of benchmark with which comparisons of behavior in nonmarket choice can be made. It will be useful to follow the same procedure with respect to the information content of choice situations. Traditionally, economists have assumed that choosers in the market possess substantially full information about the alternatives that they confront. Only within the last decade has the whole set of problems summarized in the term, “theory of information,” come to be examined thoroughly.
It is widely recognized that even for day-to-day market choices the individual may not be in command of anything approaching complete knowledge about the alternatives that he faces. There are several reasons for this ignorance. First of all, given the fact that securing information is costly, the optimal degree of investment in search may produce results that fall far short of genuine omniscience. Secondly, choices may be such that uncertainty cannot be eliminated even under maximum investment in information gathering. Thirdly, the individual may operate under an illusion that he is more informed than he actually is; he may be ignorant but not aware that he is. It becomes difficult, if not wholly impossible, for the external observer of individual choice behavior to make distinctions between these several situations. In any one of them, the rationally motivated behavior of the individual may produce results that are not desired or intended.
When we look at the individual’s behavior in fiscal-collective choice, all of the problems of information emerge with renewed force. We know that the average person possesses far less information about the costs and benefits of public goods and services than he does about the costs and benefits of private or market goods and services. But just how ignorant is the average voter-taxpayer-beneficiary about the fiscal alternatives that he confronts? How much does he really know about the impact upon him exerted by the various fiscal institutions, existing or potential? How much is he obligated to pay in taxes? What are the costs of the public services that he enjoys? How much value does he place on the benefits from these services? Does he make any effective translation of benefits into tax-costs?
To raise such questions as these suggests the paucity of research that has been aimed at answering them empirically. Consider the most direct question of the group: How aware is the individual of the amount of taxes that he pays? How much do public services cost him, computed in dollars of tax liability? Even this elemental question must be factored down into several subsidiary ones before it can be partially answered. The degree of information possessed by the individual will vary with the tax institutions under which he pays. Earlier chapters have demonstrated that the tax awareness should be greater under direct than under indirect taxes. A logical starting point for research might be, therefore, the individual’s estimation of personal tax liability under the most widely employed and most important direct tax, that on personal incomes.
Estimated Liability Under the Federal Income Tax
Enrick Studies. This was the purpose of the studies carried out by Norbert Enrick of the University of Virginia in 1961, 1962, and 1963. This research involved interviews and questionnaires circulated in the Charlottesville-Waynesboro, Virginia, area in 1961 and 1962, followed up by nationally circulated questionnaires in 1963. In each case, samples were drawn by accepted randomizing procedures.
Enrick asked two simple questions of each person. First, the individual was asked to
guess the total amount of federal income tax that he had paid in the preceding year. Secondly, he was asked to look at his personal records and to find out how much he
actually paid for that same year. Questioning was conducted in the last half of the calendar years so as to examine tax awareness at some time other than the springtime settling period.
The results of Enrick’s study were not surprising. Even under the individual income tax, people are not well informed concerning their own personal liabilities. Only slightly more than one-half of the respondents (55 per cent) were able to estimate their own tax liability within plus or minus limits of 10 per cent. More than one-fourth of the respondents erred in estimating their liability by more than 20 per cent. Over-all, there was some slight tendency for the respondents in Enrick’s sample to underestimate their tax liabilities, although this finding was not a dominant one. The samples were drawn from all income levels, but there was no demonstrable relationship between the percentage error of tax estimation and the level of income of the respondent. This provisional finding was, itself, of some importance since it tends to refute the hypothesis that withholding, as an institution, makes people less conscious of the taxes that they pay. Corroboration of this hypothesis would have required that high-income receivers (who have a smaller share of total tax liability withheld) demonstrate more accurate estimates.
Estimated Liability Under Withholding: The Wagstaff Study. To some extent at least, the personal income tax was converted into an “unconscious” tax for many taxpayers through the inauguration of withholding provisions in 1943. It seems plausible to suggest that this institution, in and of itself, affects the tax awareness of the individual, despite the fact that this suggestion was not corroborated by Enrick’s limited survey.
To ascertain more specifically liability awareness under withholding was the purpose of a second University of Virginia study completed by J. V. Wagstaff in 1963, some months prior to the tax legislation of 1964. Wagstaff surveyed more than a thousand taxpayers whose income was subject to withholding. He asked each of them to estimate both his gross income per pay period, before any deductions, and the amount of tax withheld from this income per pay period. This information on taxpayers’ estimates was then compared directly with payroll and withholding records made available to Wagstaff by employing firms.
As in the comparable Enrick study, Wagstaff found that individuals are not well informed as to their income-tax liabilities. In this case, some 52 per cent of the respondents were able to estimate tax liability within an error range of plus or minus 10 per cent, thus confirming the general validity of Enrick’s similar result. Some 30 per cent erred in their estimates by more than 20 per cent, plus or minus. As with the Enrick study, there was no dominant tendency for either overestimation or underestimation by the group taken as a whole, although here, in contrast with Enrick’s results, there was a slight indication of overestimation. Wagstaff produced more interesting results when he broke his respondents down by income groups. He found that, as a subgroup, respondents with incomes lower than the median for his whole sample tended to overestimate the amount of taxes paid. By contrast, the above-median income subgroup tended to underestimate the amount of tax. Both of these findings were significant statistically.
An additional feature of Wagstaff’s study that is highly interesting concerns the relationship between attitudes toward tax fairness or equity and tax consciousness. As a preliminary question, Wagstaff asked all respondents whether or not they considered the personal income tax to be “fair.” He then isolated those respondents who held the tax “fair” from those who claimed it to be “unfair.” He found that there was a surprisingly accurate estimation of tax liability among individuals in the group who considered the income tax to be “fair.” By contrast, individuals who responded that the tax was “unfair” tended to have a significantly wide margin of error between estimated and actual tax liabilities.
Since the Enrick and Wagstaff studies are not comparable in any direct sense, no definitive conclusions as to the differential influence of the institution of withholding on tax awareness seem warranted. Since the errors in estimation were somewhat greater in Wagstaff’s survey, some corroboration of the “unconscious tax” hypothesis is suggested. But additional comparative data on tax awareness of persons not subjected to withholding is required before any conclusions here can be definitely established. Wagstaff’s data need to be compared with closely comparable data for persons whose income is not subject to withholding; in this way, a reasonably clear testing of the “unconscious tax” hypothesis would be possible.
Schmölders’ Survey. One of the most persistent workers in the whole area of fiscal consciousness for many years has been Professor Günter Schmölders of the University of Cologne, Germany. Schmölders’ primary concern is simply that of learning more about how individual citizens think about the fisc, and he has explicitly called for a new branch of public finance, which he calls “fiscal psychology.”
*56 His research, for the most part, has been directed toward ascertaining taxpayer attitudes. We shall refer to this at several points later in this chapter, but at this point one of Schmölders’ surveys may be mentioned in connection with estimation of liability under personal income taxes.
Schmölders asked his respondents what percentage of their gross income they thought they paid in income tax, and then he compared these results with external estimates for the appropriate percentages actually paid. These external estimates were based on general rates assigned on the grounds of occupational category and income class. As in all of the studies of this type, Schmölders found that taxpayers were not well informed. In addition, his results indicated that individuals, on the average, tended to overestimate their liabilities under the tax.
Awareness of High-Income Taxpayers of Marginal Rates. A recent Michigan study attempted to ascertain whether or not high-income taxpayers were aware of the marginal rates of personal income tax. The major finding was that almost one-third of the respondents (27 to 31 per cent), all with annual incomes in excess of $10,000, were unaware of the marginal rate of tax which they paid.
On the basis of the limited studies that have been completed, the only conclusion that seems possible concerns the limited extent to which individuals are informed, even about their own liabilities under the personal income tax, surely the one tax in the structure upon which we should expect a relatively high degree of accuracy.
Tax Awareness Under Indirect Taxation
We know that individuals are likely to be less informed about the costs that indirect taxes impose on them than they are about the costs of direct taxes. Our knowledge concerning the magnitude of their ignorance and the direction of their errors in estimation is even more limited than that relevant to direct taxation. Schmölders found that, in many cases, taxpayers are not able to distinguish between commodities and services that are subjected to specific excise taxes and those that are not. For the most part, individuals are aware that taxes are important components in the final prices of the standard sumptuary goods, such as liquor or tobacco. However, for remaining nontaxed “luxury” goods included in the questionnaire from one-third to one-half of the respondents believed that a tax existed when it did not.
*59 The ignorance as to rates of tax was even more serious. For the cigarette tax, only 14 per cent of the respondents were accurate within a 10 per cent rate range. Approximately equal numbers of respondents estimated the tax to be higher and lower than it actually was. With the tax on sugar, an even smaller per cent made accurate estimates as to rate, and, this rate being considerably lower, there was a general tendency toward overestimation.
Schmölders also asked the question: How much additional income do you think you would get if all excise taxes should be removed? Including the turnover tax, the appropriately computed answer was estimated at some 10 per cent of family income on the average. His findings were that lower-income families tended consistently to overestimate the weight of the taxes, while upper-income families tended to underestimate them. This finding is consistent with that of Wagstaff, reported earlier, with respect to the personal income tax.
In 1954, Robert Ferber attempted to determine the awareness of American consumers of the excise-tax reductions introduced earlier in that year. He found that not more than 30 per cent of the respondents were aware of the tax reductions in any case, and, for some goods, this figure was as low as 16 per cent. Consistent with Schmölders’ results, Ferber found that a significant share of respondents could not distinguish taxed and nontaxed products.
It is clear that much more research is required before we can really know much at all about individual information on burdens of indirect taxes. For specific excises, where the real incidence is reasonably predictable, such information is surely limited severely. For the more important indirect sources of revenue, such as general sales taxes, turnover taxes, value-added taxes, and corporation income taxes, where the incidence is in dispute even among experts, the ignorance of the taxpayer-voter must be great indeed.
General Tax Awareness
In an extensive British survey, completed in 1965, individuals were asked to estimate total amounts of taxes paid (direct and indirect) and these estimates were compared with reasonably computed totals. Somewhat in contrast to Schmölders’ results, there seemed to have been general underestimation. Those who overestimated tax liabilities seem have been relatively concentrated in lower-income groups in the sample, independently confirming one of Wagstaff’s results. Respondents were also asked to estimate the proportion of income paid in taxes for the country as a whole. Here the results indicate reasonably accurate estimates if only averages are employed, with low-income groups again exhibiting higher tax awareness. Dispersion about the averages suggests, however, that respondents were not at all well informed as to average levels of tax.
Tax Estimation and Tax Awareness
Individuals make fiscal choices not on the basis of how accurately they estimate tax costs, but rather on their consciousness or awareness of these costs along with their predictions as to magnitudes. The ignorance of the individual concerning his tax liability does not measure “consciousness” at all accurately. The more informed the taxpayer is, the more “tax conscious” he is likely to be. Hence, ignorance becomes, in some rough sense, a measure of “unconsciousness.” To the extent that an individual is genuinely unaware of the existence of a tax, he will tend to be more acquiescent to its imposition, even if, when it is brought to his attention, he tends to overestimate its impact.
For our purposes, fiscal institutions should be classified, if possible, in terms of their net impact on individual choice behavior. Tax institutions can affect behavior in two ways: First, the institution can affect the transmission to the individual of an awareness of public services costs. Different institutions will generate different results in this respect. Secondly, once the individual is fully aware that the tax exists (as he must be when confronted with a questionnaire asking for his estimates of liability), the form of the tax may affect the degree as well as the direction of error in estimation. These two separate effects of taxes should be distinguished.
The studies to which reference has been made are largely directed to the second of these effects. Enrick’s, Wagstaff’s, and some of Schmölders’ work is aimed at finding out how accurately the individual can estimate his tax liability and what is the direction of his errors. Much less work has been done on finding out whether or not the individual is even aware that a tax exists. Here Wagstaff’s finding that employees subject to withholding tend systematically to underestimate their gross incomes suggests that the institution of withholding, as such, reduces income-tax awareness. This can be valid despite the accompanying finding that, when questioned concerning the amount of income tax withheld, there was some tendency toward overestimation, especially among lower-income groups. If, from such data, we should try to make some predictions as to the net impact of withholding on fiscal choice behavior, some comparison of the relative strengths of these offsetting effects might be required.
Similar conclusions follow for Schmölders’ rudimentary findings on indirect taxes. As suggested, individuals are not even conscious that such taxes exist in many instances, and, on the other hand, many think that taxes exist which do not. However, when taxes that do exist are brought to their attention, certain groups tend to overestimate the extent of their own liability.
Benefit Estimation and Awareness
Information on the individual’s estimation or awareness of his tax liabilities is meager; but, by comparison, it is plentiful. Almost no empirical work has been directed at determining the individual’s estimation or awareness of the value of the benefits that he secures from the availability of publicly provided goods and services.
Logically, we can think of classifying public goods and services in two categories similar to the direct tax-indirect tax classification on the other side of the fiscal account. But if we try to apply such a classification on the spending side, problems immediately arise. What is a
direct public good or service? On the tax side, directness implies that the levy is upon the person who is expected to bear the final incidence. Defined in this manner, there is no necessary connection between
generality in taxation; a direct tax could be highly selective or discriminatory. However, the standard institutions of direct taxation also tend to be those of
general taxation, while those of indirect taxation tend necessarily to embody discriminatory features. This leads to the commonly accepted link between indirectness and nongenerality. When we consider the public expenditure or benefit side, however, the situation is almost reversed. A
direct public service, one that involves the most direct linkage between the individual recipient and the fisc, must, almost by definition, be discriminatory or selective. On the other hand, the most
indirect public goods or services, those that are the most remote from the individual’s private economic situation, tend to be the most
general. Unless this asymmetry between the two sides of the account is kept in mind, confusion is likely to arise when discussing direct and indirect benefits from public goods.
Direct benefits are those which come closest to providing individuals with money payments, freely convertible into goods and services as desired. It follows that we should expect persons to be more aware of these benefits than those from public goods that yield value only in quite specific and nonindividualized forms. For example, the benefits that the individual secures from national defense spending are surely indirect; he cannot convert Polaris submarine patrols into anything which he can privately enjoy, even to the extent of placing a roughly comparable value upon them. Hence, we should predict that he will be less conscious of such benefits than he will be of those which he secures under various social welfare programs.
Empirical testing of benefit awareness under the several types of public programs would be helpful, but the relevance of such data for the purposes of this study should not be overemphasized. The individual’s awareness is, for our purposes, relevant only to the extent that his choice behavior may be modified predictably. On the tax side, it is reasonable to assume that, other things equal, the institution that reduces tax consciousness or tax awareness tends to modify individual choice in the direction of greater public outlay. It is difficult, however, for reasons to be discussed, to apply symmetrical predictions on the spending side. Can it be maintained that, other things equal, the public spending program that generates the lowest “awareness” on the part of the individual beneficiaries tends to bias individual choice in the direction of lower tax rates? The difficulty here lies in the fact that the provision of public service benefits is the
raison d’être of the whole fiscal structure. Analysis is meaningful which suggests that differing tax institutions, all of which draw generalized purchasing power from the individual, may exert differing effects on his choice of public goods, generically defined. One cannot, however, simply reverse this statement and say that analysis of the effects of differing budgetary compositions on levels of taxation is equally meaningful. The point is that differing budgetary compositions represent differing fiscal “purchases,” so to speak. Again an analogy with markets will be helpful. One can say that the
manner of paying for goods, oranges or apples, may affect the quantity purchased. One cannot say, in reverse, that whether or not oranges or apples are purchased affects total outlay. Of course it will. But wholly different choices are involved in this case, and comparison becomes essentially irrelevant.
It follows that the
institutions of providing public goods and services are relevant to our purposes only to the extent that choices are comparable; that is, only to the extent that the
same public-goods mix is selected under the varying institutions. While it may be presumed that an individual beneficiary is less aware of a remote and indirect public good, such as defense, than he is of a direct welfare payment, his choice, at the margin of adjustment, need not be affected. This is because, in any sort of “political equilibrium,” the various budgetary items will tend to be extended so as to yield roughly equi-marginal benefits. At this point, the representative taxpayer-beneficiary should value marginal extensions of the various public services in some roughly comparable manner. This “equilibrium” may or may not be “optimal” in some other sense, but this does not concern us here. Our task is simply that of determining what effects, if any, structural institutions of the fiscal system exert on the final adjustment. How can the institutions through which public goods and services are made available to the individuals affect their choice calculus in the collective decision-making process?
On the benefit side, the question then reduces to: Are there institutional differences that may affect the
directness or the
indirectness and hence the awareness of the individual beneficiary for
comparable public services? If the issue is posed in this way, the relatively limited range for institutional variability on the spending side can be more readily appreciated. National defense may be financed through any number of different tax institutions, but how many spending institutions can be utilized to provide the citizen with national defense? How can institutional variations here affect the individual’s awareness of benefits? For genuinely general public goods, such as defense, there are few structural changes that can be possible. For public goods and services that are somewhat less remote from the individual, there are some institutional variations possible, and, with these, research hypotheses can be formulated and tested.
Consider collective outlay on elementary education. What variations are there in the institutions through which this might be provided, and how might these variations affect the willingness of citizens to bear the necessary tax costs? Suppose that a local community proposes two alternative schemes: In the first, all collective outlay takes the standard form of direct operation of a free public school system. In the second scheme, all families with school-age children will be provided with tuition grants or vouchers that they may use in paying for privately provided but qualified educational services. These are two distinct institutional arrangements for providing comparable services. In each case, assume that the same tax institution is to be employed, say, local property taxation. Will the community choose to finance precisely the same outlay under each scheme? It seems evident that there may be differences in individual choice behavior under the two schemes. For those families that are direct beneficiaries, the voucher plan tends to make the outlay more direct perhaps, and these families would probably choose to spend more collectively under this than under the alternative institution. For those citizens who are taxpayers, but who are not direct beneficiaries, the results might be reversed; the voucher plan might well make the benefits seem more remote, because concentrated directly on specific beneficiaries. To my knowledge, no research has been directed specifically at finding empirical answers to such questions. Such research would be difficult because observed institutional variations here are much more narrow than comparable variations on the tax side.
The elementary education example is discussed here because it does suggest that institutional hypotheses may be developed and tested for those goods and services that are comparable, such as public outlay for medical care, for housing, for social services generally. Different institutions for providing roughly comparable benefits can have different effects here, and where such differences in institutions are observed to exist, empirical research that asks the right questions can yield highly valuable results.
Choice in Welfare: The Institute of Economic Affairs Surveys. British surveys conducted in 1963 and 1965 indirectly provide some evidence of benefit awareness of individuals under several existing welfare programs.
*63 The emphasis of these surveys was upon securing information on individual preferences for alternative institutional arrangements, as such, and not upon examining the possibly differing results under such alternatives, which is the information directly relevant for our purposes. However, certain by-product results of this survey suggest levels of benefit evaluation at least in rough opportunity-cost terms.
In the 1963 survey, public attitudes on the provision of medical care, education, and pensions were ascertained. With respect to medical care, respondents tended to underestimate seriously costs of providing standard quantity units, for example, a week’s hospitalization. The lack of cost awareness was even more emphatically shown in the fact that some 36 per cent of the respondents thought that NHS contributions were sufficient to finance fully the National Health Service when, in actuality, such contributions provide less than one-fifth of the total revenues. Similar, although less startling, results were found with respect to public attitudes on the costs and benefits of public education. Some 18 per cent of the respondents explicitly considered it unnecessary to pay for education “either directly or through rates or taxes.” By contrast with their answers in medical services, however, the respondents tended to overestimate the costs of public educational services that were provided, and also the costs of providing educational services privately. With respect to the public pensions schemes in Great Britain, some 35 per cent of the respondents genuinely considered the operation one of “insurance,” with the pensioner having accumulated sufficient reserves to meet the full costs of benefit payments. In fact, both the employer and the employee contributions account for only some 10 per cent of total revenues for the program.
The 1965 survey attempted to determine respondents’ awareness of the total value of social benefits received from all government welfare programs. Here, as in the estimate of taxes, there was general underestimation, and the range of estimation was wide. No attempt was made in 1965, as there was in 1963, to secure benefit estimates for particular programs.
For purposes of this study, the most interesting result of both British surveys is the revealed failure of individuals to make any effective translation of public service benefits into tax-costs.
The Relevance of the “Economic-Individualistic” Model
A second broad area for empirical research involves tests of the possible explanatory range of the central “economic” hypothesis of our individual collective choice models. Quite independently of the information content problem, to what extent do individuals choose among collective alternatives on the basis of criteria that may be externally measured? Even if an individual should be accurately informed as to both his own tax liabilities and the value of the public service benefits that he receives, both in total and in per-unit terms, will he choose in such a manner that an economist can make some rough predictions? Unless this question can be answered affirmatively, little progress can be made toward developing a “scientific” theory of individual behavior in fiscal choice.
As suggested in Chapter 12, casual everyday observation of politics tends to corroborate the central hypothesis here in numerous ways. Elections are contested, and campaigns are waged, at least in part, on issues that can be reduced to direct and measurable economic content. Behavior of the electorate is surely influenced, to a degree, by predictions as to measurable gains and losses. It remains useful, nonetheless, to extend research well beyond this range of casual observation, and to test the validity of the central hypothesis more specifically if possible. And the construction of applicable and relevant subsidiary hypotheses is not nearly so easy here as the behavior of politicians makes it appear. The analyst must be able, independently, to identify the effects of the fiscal variables examined upon the circumstances of the individual participant before he can develop any testable propositions concerning individual response. Such identification is often difficult. Specific conclusions must be made concerning the incidence of both taxes and public benefits, and the real or actual incidence must be distinguished from the apparent. To the individual chooser, it is the apparent incidence of taxes and benefits that affects his choice.
Schmölders’ work has already been mentioned several times. He has not been concerned with testing hypotheses in the direct sense, but some of his subsidiary findings can be adduced at the outset to provide corroboration of the “economic” motivation in the fiscal preferences of individuals. In a survey of individual attitudes on the desirability of public or governmental support for private industry, he found that those persons from the industries most likely to secure grants were the most likely to respond favorably. Similarly, he found that governmental employees made up the group most likely to say that public services were worth more than the tax-costs of providing them.
*64 These particular results are, of course, presented along with many others concerning general public attitudes toward the fisc, and it should be emphasized here that Schmölders does not consider directly the testing of the economic motivation hypothesis.
Michigan Studies. The problem of identifying individual interest can be resolved with ease only for some of the cruder tests. Individual utility maximization cannot always be related directly to income, wealth, or economic position. Unless, however, some such direct relationship is assumed, specific testing of the utility-maximization hypothesis becomes difficult. By and large, individual utility may be related to income level for purposes of examining individual responses to tax alternatives. If high-income receivers are found to express a relative preference for sales taxes over income taxes, whereas low-income receivers are found to express the opposing preference, some support for the explanatory value of the economic model is provided. This result was forthcoming from the study of tax preferences conducted in Michigan in 1959 by Elizabeth David.
*65 She also found, not unexpectedly, that property owners tended to be less favorably inclined than were renters toward the use of the property tax as a revenue-raising device. Generally speaking, David’s survey corroborates the hypothesis that the economic positions of individual citizens are an important determinant of their attitudes on fiscal alternatives.
In a survey conducted by the Survey Research Center of the University of Michigan in 1960 and 1961, and reported by Eva Mueller,
*66 the rule played by economic self-interest in determining respondent attitudes was shown to be a significant, although not necessarily a dominating, explanatory factor. Public programs aimed at providing benefits to low-income groups (aid to the poor, the unemployed, hospital and medical care, public works) tended to be viewed more favorably by the lower-income members of the sample. Aid to small business and highway outlays tended to be more heavily favored by the higher-income groups. All groups, regardless of income level, seemed to favor expenditures on education and on aid to the aged. This finding, along with certain other features of the survey, suggested that economic self-interest, in any narrow sense, fails as a self-contained, all inclusive explanatory hypothesis.
One of the more interesting findings of the Survey Research Center project was the attitude expressed by higher-income individuals with respect to relative expansions in public spending programs of all sorts. On the basis of a crude and unsophisticated version of the self-interest hypothesis, higher-income groups should be predicted to be less favorably inclined toward additional public goods and services than the lower-income groups. However, the data collected here suggested that, on the average, the two groups that view expansions in spending programs most favorably are those at the two extreme ends of the income scale. This result was, for example, quite clear with respect to educational spending.
Does this sort of evidence contradict or refute the utility-maximization hypothesis more generally considered? No refutation seems indicated when it is recognized that the higher-income individual may place a higher evaluation on the “spillover” effects of public spending than persons in lower-income groups, and, also, that the pattern of tax incidence may not be so progressive as the relevant ratio of income and price elasticities of the public services provided. In other words, the degree of what has been called “Samuelsonian publicness” in any particular public service may increase substantially as income levels increase. This would allow a sophisticated form of the hypothesis to remain valid in “explaining” the survey data.
Wilson-Banfield Studies. Attitudes toward fiscal alternatives as expressed by respondents to interviews or questionnaires can be extremely helpful in filling out the many gaps in our knowledge. It is commonplace, however, that such expressed attitudes do not necessarily enable us to predict behavior accurately. When confronted with genuine choice, individuals may not respond in the way that expressed attitudes might indicate. Somewhat more conclusive tests of the central behavioral hypothesis may, therefore, be carried out if actual choice behavior can be observed. Such tests are, of course, difficult to conduct since individuals do not normally choose directly among fiscal alternatives. Nevertheless some testing is possible where data on voting referenda can be secured, and where numerous observations can be found.
Municipal referenda on spending programs can provide the basis for such tests, and experiments utilizing these data have been completed by James Q. Wilson and Edward Banfield of Harvard University.
*67 Their work is noteworthy in that it sets out explicitly to test a hypothesis: Voters act as if they are trying to maximize their net family incomes. This hypothesis leads to the prediction that the lowest income groups would vote in favor of most public spending programs that are proposed by municipal governments. The upper-income groups should be expected to oppose most proposals for spending, whereas middle-income groups should be expected to be highly selective.
Results tend to confirm the behavioral hypothesis for low-income groups. However, upper-income groups also tend to vote heavily in favor of many spending programs that provide them with little direct benefits. Opposition to municipal spending programs tends to be concentrated in the middle-income ranges.
These findings corroborate the attitudinal surveys reported by Mueller. To an extent at least, the crude income-maximization hypothesis seems contradicted by the data. Wilson and Banfield “explain” the behavior of the upper-income groups by the importance of “altruistic” motives, or by “public-regardingness.” A more sophisticated version of the utility-maximization hypothesis would, also, “explain” the same data. In one sense, the presence of “altruism” as a motive is the same thing as including “redistribution” as a “good” in the individual’s utility function. This construction suggests both the strength and the weakness of the economic model. Properly stretched, the model can be helpful in “explaining” almost any observed behavior. But precisely to the extent it does, it becomes useless as a predictive hypothesis.
The Gillespie Study. The Wilson-Banfield and the Mueller findings become less damaging, even to the crude form of the income-maximization hypothesis, when these are examined in juxtaposition with the research results of W. Irwin Gillespie.
*69 His work represents an attempt to measure, empirically, the net incidence of the over-all fiscal structure, federal and state-local, including the expenditure or benefit side as well as the tax side. In his standard model, which incorporates perhaps the most acceptable set of assumptions as to benefit and tax incidence, Gillespie found that, on balance, federal tax-spending patterns favor low-income receivers as expected, treat a fairly wide range of middle-income receivers neutrally, and impose net burdens on high-income receivers. For state-local systems, by comparison, he found that low-income receivers are again favored. But the two remaining income groups are treated differently than they are under the federal system. Here the middle-income receivers are subjected to net fiscal burdens while the high-income receivers are treated neutrally.
This pattern for state-local fiscal incidence strongly implies that opposition to extensions in public spending programs at these levels would be concentrated among middle-income receivers even if a crude income-maximization hypothesis is accepted. In this respect the Wilson-Banfield results, which are exclusively drawn from state-local data, tend to corroborate rather than to refute the hypothesis with respect to the behavior of both the low-income and the middle-income groups. And, since high-income groups are subjected, on balance, to neither net burdens nor net benefits at the state-local level, their behavior in supporting spending programs provides considerably weaker refutation of the hypothesis than might have been supposed independently of the Gillespie findings.
Gillespie results should be helpful to future researchers in framing hypotheses to be selected and tested. It would seem appropriate to examine voting data in legislative assemblies at the state-local and at the federal level separately to determine whether or not opposition to spending programs by high-income groups tends to be more prevalent at the federal level.
The Davis Studies. If voter choices in referenda on spending or taxing programs are not directly available, the central hypothesis of economic model may be tested by reference to comparative data from various fiscal jurisdictions characterized by differing economic circumstances. This is the approach that was taken by Otto A. Davis of Carnegie Tech in his research on local public expenditures. The first study concentrated on public school outlay in the Pittsburgh area.
*70 Implicit in this approach is the assumption that elected public officials, in this case members of local school boards, act in accordance with the preferences of individual citizens. With this assumption, the results emerging from the deliberations and decisions of representative assemblies can be taken to reflect accurately the preferences of citizens and these results can be used directly to test hypotheses about the choice behavior of individuals.
Davis looked first at the problem of explaining the per pupil outlay on public schools in the various districts of his area. There are, of course, the orthodox or standard explanatory variables here which can be expected to account for a major share of the interdistrict variations in spending. These are the familiar variables for per capita incomes, population density, property values, level of education. Davis’ main emphasis was on the question as to whether or not additional variables designed to reflect the economic-individualistic model of choice behavior could add to the total explanation of variability. Specifically, he sought to predict the directions of influence exerted on the outcomes by several additional variables of this sort.
Some of these variables were: (1) value of industrial property, (2) per cent of voters owning property, (3) per cent of total population in schools, (4) per cent of school children in public schools. He predicted independently that the first variable, value for industrial property, should positively affect public school outlay since this provides a source of revenue not wholly borne by local residents of the districts. Secondly, he predicted that the larger the per cent of property owners in the jurisdiction, the lower the outlay since most local revenues are raised through the tax on real property. He further predicted that there would be a negative relationship between public school outlay per pupil and the per cent population in school and the per cent of children in public schools.
The second study, undertaken jointly with James L. Barr,
*71 attempted to determine whether or not the economic positions of the median voters in Pennsylvania counties, defined in terms of property holdings, provide some of the explanation of variations in local spending levels. Specifically, the study examined the hypothesis that the ratio between the median’s voter’s property holdings and the assessed valuation of all property in the local jurisdiction was inversely related to per capita spending levels. Statistical results indicated that the directions of effects were those predicted, and that the hypothesis was of explanatory value. Other variables must, of course, be added for any satisfactorily “complete” explanation of intercounty expenditure variance.
Davis’ results, in both studies, while admittedly inconclusive, do not contradict his central hypotheses. Significant corroboration should not, of course, have been expected. The data upon which the empiricist must draw remain crude indeed, and many elements influence the final outcomes of collective decision processes.
Much more research, perhaps especially of the sort undertaken by Wilson-Banfield and by Davis and Barr, is required before the genuine predictive power of the utility-maximizing model of individual choice behavior in politics can be determined. No one would, I presume, claim for this model or hypothesis exclusive domain in a fully developed theory of fiscal-collective choice. The limited studies that have been completed support the view that the model can provide important explanatory assistance, but that errors can be easily made if this alone is relied on for positive prediction.
The Influence of Fiscal Institutions. The several areas of research that are useful for our purposes are all closely related. Studies designed to fill in the gaps in our knowledge about individual fiscal consciousness or studies designed to test broadly the applicability of utility-maximizing models of individual behavior are, in one sense, antecedent to the research that is more directly related to the theoretical analysis of preceding chapters. Here the emphasis was on making predictions about the effects of fiscal institutions on individual choice behavior, and through this, on collective outcomes. The most direct research is that which aims at testing these predictions themselves, not their underlying presuppositions.
As in the case of testing the validity of the economic model generally, casual observation and introspection should not be overlooked merely because these are unexciting research tools. To an extent, these are the most useful tests that can be employed; they are cheap to conduct and they are probably more convincing than more complicated methods. One of the institutional predictions made in Chapter 5 concerned the effects of an old- or an existing-tax institution. The hypothesis here is so widely known and accepted that it would warrant the effort of only the most pedestrian of doctoral candidates to check it out and corroborate it. The general prediction concerning the relative impact of direct and indirect taxation on individual behavior is almost equally evident from ordinary experience. This is perhaps best exemplified in the persistence of the corporation income tax in the federal revenue structure despite its opposition by almost all those who discuss “tax principles.” Note that when, in 1963, proposals were made to reduce both the corporate income tax rate and personal income tax rates, the 1964 legislation, as finally approved, resulted in a larger-than-proposed cut in personal rates but a smaller-than-proposed cut in the corporate tax rate. There could hardly be stronger corroboration of the hypothesis that the corporation tax, because it is not directly sensed by the individual voters, tends to generate a greater degree of acquiescence than the more direct personal income tax.
One of the most important institutional elements of the political decision structure with respect to its effects on individual behavior in making fiscal choices is the apparent separation of decisions on taxes and on public spending programs, decisions that must, in some underlying real sense, remain interdependent. The discussion of this problem in Chapter 7 suggested that the predictable results of a budgetary process that allows fragmentation of decision are ex ante deficits between approved tax revenues and approved spending, deficits which must under some balanced-budget rule be adjusted ex post. As the analysis suggested, this tendency toward ex ante deficits can be predicted, independently of any predictions as to whether, in the final adjustment, total public outlay will be less than, equal to, or greater than that which might be produced under alternative decision processes where the two sides of the fiscal account are simultaneously considered.
The Fragmentation of Choice: Survey Research Center Results. The general prediction is amenable to testing in several ways. One indirect test is provided in the Survey Research Center’s attitudinal study reported by Mueller, and mentioned previously. Respondents were asked whether or not more, the same, or less public spending should be undertaken under each of several program categories. No mention was made of the means through which the additional necessary funds for financing the programs might be secured. The results, reproduced from Table II in Mueller’s paper, are shown in the second column of Table 13.1. The same set of questions were then asked with the additional proviso “even if taxes must be raised.” The change in response is striking; the per cent indicating that more should be spent under the tax-bridge situation is shown in the right-hand column of Table 13.1.
Although the limited value of such attitudinal surveys for genuine choice behavior should always be stressed, these results seem to suggest, quite clearly, that individuals do not, in general, translate meaningfully between the two sides of the fiscal account. Insofar as the decision process allows them to think in terms of approving spending programs independently of tax-costs, a deficit will tend to be produced. Note that, on the basis of the responses shown in Table 13.1, a majority of respondents would favor additional spending on the first five programs, but that no single program could have secured majority approval if the accompanying tax increases should be required. Note, also, that the ordering of preferences for the various programs is changed in the two situations.
|Program||Per Cent Respondents Who Think More Should Be Spent||Per Cent Respondents Who Think More Should Be Spent, Even If More Taxes Required|
|Help for older people||70||34|
|Help for needy||60||26|
|Hospital and medical care||54||25|
|Source: Table II, p. 215, in Eva Mueller, “Public Attitudes Toward Fiscal Programs.”|
In the 1963 British survey previously cited, respondents were asked whether or not programs should be extended, but only under the tax-bridge situation. Here a majority of respondents (51 per cent) indicated a preference for expanded outlay on education, while 41 and 43 per cent respectively favored expansions in outlay on public health and public pensions.
*72 Somewhat interestingly, these percentages were reduced to 41 (education), 32 (health), and 34 (pensions) in the 1965 survey.
Reid’s Study of Veterans’ Bonus Legislation. More definitive tests of the hypothesis that separation of tax and expenditure decisions exerts important influences on behavior can be made by an examination of democratic choice itself. Given a multiplicity of governmental units, comparative analysis becomes possible, provided only that the institutional arrangements under study differ significantly among separate units.
An excellent opportunity to test the hypothesis was provided by the actions of the separate states in enacting legislation granting cash bonuses to veterans of World War II and the Korean War. This was one of the objectives of a University of Virginia study completed by John J. Reid in 1961.
*74 Twenty-one states enacted bonus legislation after World War II and twenty states after the Korean War, these being essentially the same groups. In twenty-six states, bonuses were proposed but were not finally enacted. Since the exogenous factors that might influence such a spendings decision would seem roughly comparable in all states, this divergence in results should enable some predictions to be made as to the effects of institutional differences if such are found to have existed.
Bonus legislation was presented for public referenda in twenty-seven states, a feature that allowed a direct examination of individual choice behavior in the voting booth. In fifteen of these twenty-seven states, the question presented to the electorate was framed in such a way as to impress upon the voter the necessary accompanying tax-costs of the bonus. In some cases, approval was tied explicitly to the approval of a corresponding tax levy; in other cases, voters were merely asked to indicate their preferences for the tax increases that would be necessary. In the remaining twelve states, the question as to the bonus was presented without reference to tax-costs.
The results that Reid observed tend strongly to corroborate the hypothesis that spending programs secure approval more readily when the tax-bridge is not present. In each of the twelve states where the bonuses were proposed without note of tax-costs, the referendum was favorable. Of the remaining fifteen states, bonus proposals were rejected in six, or 40 per cent, of the total.
Although not directly testable, Reid also found that the action on bonuses in state legislatures (in states where no referendum was required) tended to be influenced significantly by the degree to which the necessary tax-costs were tied into the positive approval of the bonuses.
Birdsall’s Finding. Similar corroboration of the hypothesis can be found in William C. Birdsall’s study.
*75 He noted that, in November, 1955, a referendum proposal was submitted to the electorate in New York state calling for the authorization of a $750 million bond issue for highway development. The state’s legislature had, prior to the referendum, enacted a fuels tax increase that was scheduled to go into effect
only upon favorable adoption of the referendum proposal. The proposal was then defeated, 54 per cent to 46 per cent. Only one year later, in November, 1956, a similar proposal calling for the authorization of a $500 million bond issue for the same purposes was passed, with a 66 per cent yes vote. The second proposal differed from the first only in that, in 1956, the referendum decision was in no way tied to an increase in the fuels tax.
Birdsall’s finding in this one instance cannot, of course, be taken for more than one isolated corroboration of the hypothesis. However, when this is added to the Survey Research Center material on fiscal attitudes and to Reid’s study of veterans’ bonuses, the corroboration of the hypothesis becomes more impressive. These various strands of evidence, coupled with the introspective plausibility of the hypothesis, provide perhaps ample grounds for making firm predictions concerning the direction of effects exerted by the fragmentation of the budgetary process.
Political Institutions and Fiscal Institutions
It is not easy to draw a sharp distinction between institutions labeled as “fiscal” and those that might be called “legal,” “political,” or “constitutional.” Such institutions as earmarking, annual budgeting, the separation of revenue and expenditure decisions, etc. are “political” at the same time they are strictly “fiscal.” But there exist, beyond these, institutions or rules that are essentially political without being limited to fiscal choices. I refer here to the rules for reaching collective decisions. There are numerous ways of attaining outcomes in democratic politics, and constitutions exist which specify such variations.
Once it is recognized that political or collective decision processes are not means of arriving at “truth judgments,” at “right” answers, but are, instead, simply the processes through which individual choices as to alternative outcomes are combined for producing collective results, it becomes obvious that different rules can generate differences in outcomes. In another work, the theoretical background for an analysis of varying rules for political choice has been presented.
*76 As with the specifically fiscal institutions already discussed, however, little empirical research has been directed at determining the influence of the several possible rules.
One area where empirical research seems possible is in the approval of school bond issues by the electorate of the various school districts in the United States. Here we have a multiplicity of units making roughly the same decision and we should be able to compare results. And, because of differing state constitutional and legislative provisions as to the rules under which local school districts make decisions, some influence of these rules on final outcomes should be noted. In a University of Virginia study, begun in 1963, John Robert Cooper tried to predict the influence exerted by a few of the basic rules for decision on such bond issues. In many jurisdictions, simple majority approval in a referendum is required. In many others, a qualified majority of those voting is required, with the percentage “yes” vote varying over a rather wide range. In still other jurisdictions, the majority, simple or qualified, must be computed on the basis of all registered voters. Still other districts operate under a property-ownership qualification for voting in bond elections.
Cooper arrayed the differing requirements in terms of apparent restrictiveness as follows: Simple majority, majority with a property qualification, special majority greater than 51 per cent, special majority with property qualification, and, finally, a majority of all eligible voters. He then hypothesized that the proportion of bond issues approved would tend to fall as one moves down this array. The data, which were limited to issues presented for referendum votes in 1961, only partially support the general hypothesis. More bond issues tend to be approved in jurisdictions where only a simple majority is required than in those where a qualified majority is required. As with some of the other findings reported in this chapter, this result seems almost intuitively obvious to anyone who adopts a broadly general democratic decision-making model and who imputes any rationality to voter participation. It is useful, nonetheless, to have corroborating evidence from the real world. Cooper’s results, in this respect, are in full agreement with previously completed and more limited surveys.
Predictions contained in the hypothesis concerning the effects of property qualification were not supported by the data. Accepting some version of the crude income-maximization hypothesis, and recognizing that taxes on real property must provide funds for servicing and amortizing debt, the limitation of suffrage to property owners in bond referenda would be predicted to reduce the proportion of approvals. Cooper’s study fails to support this prediction. Broadly speaking, the percentage of bond issues approved remains approximately the same with and without the property qualification, other aspects of the referenda remaining the same. In part this seemingly paradoxical result is explained by the fact that, in the sample drawn, those jurisdictions subjected to property qualifications tend also to be those where average income levels are relatively higher. If the Wilson-Banfield findings, mentioned earlier, are applied here, this tendency toward a high percentage approval even with the property qualification becomes understandable.
The Several States
Hypotheses concerning the effects of fiscal-political institutions on choice behavior can be tested only in those situations where different institutions can be observed to operate under circumstances that are, in other respects, broadly similar. A multiplicity of jurisdictions becomes a necessary condition for research. For this reason, most of the research reported in this chapter, and indeed most of the research that seems possible, relies on the availability of data from the separate state-local fiscal systems. Data on choice behavior under the institutions of the federal or central government are available, but these can reveal little or nothing about the effects of different institutions of fiscal choice simply because no controls, no contrasting institutions, exist with which results might be compared. This fact severely limits empirical research on fiscal process.
Researchers should, nevertheless, consider themselves to be fortunate that data on state-local systems are readily available in substantially complete form, especially for those years when censuses of government are completed. With these data researchers should be able, with the aid of sophisticated techniques, to isolate the effects of some of the most important institutional variables.
One approach suggested is that of trying to “explain” the variations in state-local expenditures over the several states. Here the standard and familiar explanatory variables—income, population, property values, urbanization—have been shown to explain a substantial proportion of the interstate variation. To test the relevance of the institutional approach, the researcher needs to add to these familiar variables those which represent the apparently important fiscal-political institutions and try to predict the directions of influence.
A preliminary research study along these lines has been partially completed by Jack Forbes of the University of Virginia. In an attempt to isolate the influence of political-institutional variables on fiscal choice, Forbes chose as his basic dependent variable
per capita state-local expenditure as a percentage of income. He then tried to explain the variance in this value among the separate states. Note that the task here is not that of explaining the variance in levels of spending among the states, the problem that has been often examined. Forbes sought to explain what might be called the variance in the propensity to spend publicly out of income in the several states. To a large extent, the major explanatory variable for differential spending levels, that of per capita income, is eliminated here. The level of per capita income remains as an explanatory variable for Forbes’ purposes only to the extent that the income elasticities of demand for public goods diverge from unitary.
What influence on this propensity to spend publicly from income is exerted by the several possible political-institutional variables? Initially, in his study, Forbes examined the prospects of isolating the effects of some thirty separate variables. Despite the availability of
Census of Governments data, however, particular variables reflecting institutional differentials are not readily converted into statistically usable forms. As a result of his refinements, and recognizing data limitations, Forbes finally settled on the following list of independent variables for inclusion in his multiple regression.
1—number of governments per capita.
2—degree of state-local reliance on revenue from the federal government.
3—degree of local government reliance on revenue from the state.
4—degree of local expenditure autonomy.
5—degree of local tax autonomy.
6—degree of reliance on indirect taxation in the state-local system.
The predicted signs for the several coefficients in the regression equation were as follows. It was predicted that the influence of X
4, and X
5 on the dependent variable, per capita state-local spending as a per cent of per capita personal income, would be negative. That is to say, a larger number of government units, a higher degree of local tax and expenditure autonomy, would, other things equal, reduce the propensity to spend publicly from income dollars. These predictions as to the negative influence were based on the notion that, to the extent that governmental-fiscal decisions are “brought closer to the people,” more rational collective choices would be made, and that the direction of this influence would be toward reduced spending. A contrary prediction as to signs could have been made here if the important consideration was estimated to have been the economics of scale in public spending programs. It was predicted that the signs of the coefficients for X
3, and X
6 would be positive. The prediction for X
2 and X
3 requires little explanation; surely state-local spending would be expected to increase with increased reliance on external revenue sources, at either level. The positive sign predicted for X
6 stems from the notion that indirect taxation tends to generate a fiscal illusion and serves thereby to conceal from the taxpayer-voter the real weight of tax, and hence causes him to support a somewhat higher level of public spending.
The predictions as to signs were not supported in the case of X
1, the number of governmental units per capita. The predictions for X
2 and X
3 were supported by the data. Predictions for X
4, local spending autonomy, were supported, but not those of X
5, local tax autonomy. In the case of X
6, the degree of reliance on indirect taxation, the prediction was not supported by the data. The results suggest that state-local spending from income dollars tends to increase with the level of direct taxation, rather than decrease as the hypothesis predicted.
Using only the institutional variables listed above, and using 1957 data, Forbes was able to “explain” some 41 per cent of the variation in the dependent variable, per capita state-local spending as a per cent of per capita incomes. To these he added three additional, and more familiar, independent variables, those for per capita personal incomes, for degree of urbanization, and for population density. With these added, the explanatory potential of the regression increased to 45 per cent. Independently considered, the three noninstitutional variables “explain” 30 per cent of the variation, indicating of course some auto-correlation among the institutional and noninstitutional variables.
The Forbes study remains incomplete and its detailed results will not be presented in this summary treatment. The study is important for present purposes, however, in that it suggests both the relatively unexplored territory for useful statistical investigation of the influences of many institutional variables on the results of fiscal choice and the fact that such investigations will encounter serious data collection, methodological problems. The data are not such that the influences of the specific institutions can be readily isolated and examined. But the attempts can be made. And the refutation of each hypothesis adds something to the stock of our understanding.
The research reported in this chapter varies widely, both in rigor and in relevance for this study. The results should be considered primarily as suggestions for further work by the competent technicians that increasingly come to take their places in the ranks of professional scholars. The model of fiscal choice behavior that this book elaborates, that of democratic decision-making where individual preferences do count, has never been widely adopted by specialists. For this reason, the data that are available have rarely been examined in terms of the various hypotheses that emerge from the model. Few of these hypotheses have been formulated, and still fewer have been put to even rudimentary tests. The work reported on here is not that which characterizes a single-purpose, methodologically fixed, workshop. Much additional development must take place before this stage is attained. There must be additional effort devoted to deriving relevant, and potential testable, propositions. Research at the empirical level will continue to be limited by the availability of suitable data. Those which do exist are largely fortuitous, since collection has never proceeded with the appropriate questions in mind. The main conclusion seems obvious and elementary. The range and scope for potentially productive research embodying both the development of imaginative hypotheses and the testing of these seem almost unlimited.
Journal of Political Economy, LXIX (June, 1961), 213-25.
National Tax Journal, XVI (June, 1963), 169-73, and in “A Further Study of Income Tax Consciousness,”
National Tax Journal, XVII (September, 1964), 319-21.
Southern Economic Journal, XXXII (July, 1965), 73-80.
National Tax Journal, XII (December, 1959), 340-45.
Das Irrationale in der öffentlichen Finanzwirtschaft (Hamburg: Rowohlt, 1960), with special reference to the part mentioned on pages 84-86. The methods and procedures of the survey, along with more extensive results, are reported in
Steuern und Staatsausgaben in der öffentlichen Meinung der Bundesrepublik (Köln: Westdeutscher Verlag, 1960).
National Tax Journal, XVIII (September, 1965), 258-67.
Finanzarchiv, Band 20 (1959), 23-34.
National Tax Journal, VII (December, 1954), 355-58.
Education and the Southern Economy, Supplement to
Southern Economic Journal, XXXII (July, 1965), 15-34. Stubblebine does not, however, claim to examine empirical evidence to test his hypotheses.
The New Society (No. 43), 25 July 1963, pp. 14-16. Also,
Choice in Welfare, 1965.
Das Irrationale in der öffentlichen Finanzwirtschaft.
Quarterly Journal of Economics, LXXVII (May, 1963), 210-35.
The Public Economy of Urban Communities, ed. J. Margolis (Resources for the Future, 1965), pp. 74-91. Essentially the same results are reported in James Q. Wilson and Edward C. Banfield, “Public-Regardingness as a Value Premise in Voting Behavior,”
American Political Science Review, LVIII (December, 1964), 876-87.
Essays in Fiscal Federalism, ed. R. A. Musgrave (Washington: Brookings Institution, 1965), pp. 122-86.
The Public Economy of Urban Communities, ed. J. Margolis (Resources for the Future, 1965), pp. 92-111.
The Calculus of Consent (Ann Arbor: University of Michigan Press, 1962).