Do tax expenditures create framing effects? Volunteer firefighters, property tax exemptions,...
I. INTRODUCTION
During the summer of 2003, the Michigan legislature decreed that the state's tax expenditure budget shall henceforth be denoted as the governor's report on "tax credits, deductions, and exemptions." (1) This name change was not an inadvertent or technical adjustment
Of course, in political discourse, labels matter--as politicians have understood from time immemorial. "No taxation without representation" was a better slogan than "We want to be protected by the British Empire but don't want to pay for it." In recent years, however, scholars working under the rubric of behavioral economics have taken the age-old insight that labels matter and have given it a new salience. These scholars have demonstrated that "framing effects" may occur when individuals--often reasonably sophisticated and otherwise rational individuals--make and frequently maintain substantively inconsistent choices depending upon the manner in which the choices are framed.
The notion of framing effects helps to explain the paradox of tax expenditure analysis. While tax expenditure analysis has been enormously successful as a procedural program, it has largely been unsuccessful in substantive terms, failing to curb legislatures' use of tax systems for expenditure-type programs. (3) An important explanation for that paradox is that, for many people, tax expenditures create framing effects. For these individuals, it is not the substance, but the label, that controls. For individuals succumbing to framing effects, labels obscure the similarities between policies that are substantively and procedurally identical but differently-named.
Tax expenditures can be designed (and sometimes are designed) to be economically and procedurally equivalent to comparable direct outlays. Nevertheless, some, perhaps many, people apparently persist in believing that tax expenditures and direct outlays are different, even when they are economically similar. Some otherwise rational persons view $100 granted through the tax system as different from $100 awarded through an equivalent direct outlay because, for them, the label attached to public largesse matters. For these individuals, policies unacceptable when framed as direct expenditures become supportable when labeled as tax subsidies, even though the economic substance of the policies is the same.
The framing effect phenomenon is not the only reason why tax expenditure budgets fail to curb legislatures' (now disclosed) penchant for tax-based largesse. The phenomenon, however, is an important cause of the paradox that, as tax expenditure budgets have proliferated, they have had little, if any, discernible effect on legislative behavior.
To test the proposition that tax expenditures engender framing effects, I have surveyed individuals' perceptions about property tax exemptions for volunteer firefighters. In the face of lagging recruitment and retention of volunteer firefighters, communities throughout the country have in recent years granted tax exemptions to firefighters. Tax expenditure analysis suggests that these communities delude themselves by maintaining the fiction that "volunteer" firefighters are unpaid when, in substance, there is no difference between abating firefighters' property tax obligations and paying them cash instead. The literature on framing effects suggests a less cynical conclusion: Many people feel differently about compensation for a community's firefighters when it is labeled as tax relief rather than as wages. They view cash compensation, but not comparable tax expenditures, as impairing firefighters' volunteer status. Labels matter, obscuring for some the similarity of substantively identical economic policies.
Part II of this paper provides an overview of tax expenditure analysis, its history, its limitations, and its implications. It notes the procedural success and the substantive failure of the tax expenditure movement. Tax expenditure advocates have, over the last two generations, had remarkable success, at both the federal and state levels, implementing their major procedural prescription: the promulgation of tax expenditure budgets to quantify and disclose tax-based subsidies. In contrast, as a substantive matter, those budgets have failed to bring about the chief result intended by the tax expenditure school, namely, fewer tax expenditures.
Part III explores the concept of framing effects. Such effects occur when individuals (including reasonably competent decision makers) make and persist in substantively inconsistent choices depending on the manner in which those choices are framed. The literature on framing effects suggests an explanation for the paradox that widespread disclosure of tax expenditures has had no impact on the quantum of such expenditures enacted: many people, fully informed about tax expenditures, still perceive such tax-based expenditures as different from identical direct outlays. For these people, the labels disguise the substantive similarity of policies denominated differently.
Part IV discusses the problem many municipalities face in recruiting and retaining volunteer firefighters and the predominant response to date, adopting tax expenditures. In particular, property tax abatements for volunteer firefighters, with no apparent controversy, have become widespread.
With this background, Part V tests empirically the proposition that tax expenditures produce framing effects. Specifically, it examines the results of surveys in which comparable groups were asked to assess the implications of financial assistance to volunteer firefighters for such firefighters' volunteer status. One group was asked whether volunteer status is impaired by outright payments to such firefighters. A second group was asked whether volunteer status is jeopardized by procedurally and economically equivalent property tax exemptions. A third group was presented with both the tax and the direct payment alternatives and was asked whether either or both approaches impair firefighters' status as volunteers.
The results confirm the hypothesis that tax expenditures create framing effects. Specifically, the respondents who were asked whether payments to firefighters impair such firefighters' volunteer status consistently answered "yes" at significantly higher rates than did the respondents asked whether comparable property tax abatements impair firefighters' volunteer status. For a statistically significant segment of the sampled population, the label attached to a subsidy (payment versus tax abatement) mattered to an assessment of firefighters' volunteer standing, even when these differently labeled forms of compensation are, in substance, the same.
Part VI identifies the implications of this result for tax policy and for tax expenditure analysis and its emphasis on disclosure as a device to limit tax subsidies. For those who persist in viewing economically equivalent programs as different when framed as tax or as direct expenditure programs, disclosure is irrelevant. The essence of a framing effect is its persistence in the face of disclosure, such as that embodied in tax expenditure budgets.
The Michigan legislature's decision to jettison the "tax expenditure" label for that state's tax expenditure budget may ultimately prove to be an isolated success by the opponents of tax expenditure analysis. Alternatively, it could, over time, reflect the first retreat in an ebb from the tax expenditure movement's high tide. Either way, it is appropriate to assess the paradox of the tax expenditure movement. This paradox--procedural success coupled with substantive failure--in important measure reflects the framing effect phenomenon.
II. AN OVERVIEW OF TAX EXPENDITURE ANALYSIS
Few academic doctrines have achieved the success of tax expenditure analysis. The father of tax expenditure analysis, Stanley Surrey, a professor of law at Harvard and Assistant Secretary of the Treasury for tax policy, classified tax provisions as either normative features of the tax law or as expenditure provisions. (4) Normative provisions are those necessary to implement the agreed upon tax base. Under an income tax, for example, the normative features of the tax are those that define and measure income. (5) Expenditure provisions, in contrast, exempt from taxation to subsidize and thus deviate from the normative structure of the tax. (6) According to this analysis, the deduction for home mortgage interest is a quintessential tax expenditure since that deduction does not measure the taxpayer's income but, rather, subvents his home ownership. (7)
The term "tax expenditure" derives from Professor Surrey's observation that subsidizing deviations from a normative tax are economically equivalent to comparable direct outlays. (8) Suppose that, as a result of the deduction for home mortgage interest, a homeowner reduces his federal income tax liability by $1000. That result, Surrey argued, is economically identical to denying the taxpayer the interest deduction and, instead, sending him a check for $1000 to reward his homeownership. (9) Either way, the taxpayer's net payment to the government is reduced by $1000, regardless of whether that reduction is achieved by a $1000 abatement of tax liability or by full taxation coupled with a cash payment of $1000 from the public fisc. (10)
Professor Surrey and his many followers raised both substantive and procedural concerns about tax expenditures, particularly in the context of the federal income tax. Advocates of tax expenditure analysis have argued with great force that income tax deductions and exclusions (unlike tax credits) have a regressive distributional impact, which Surrey powerfully denoted as the "upside-down effect." (11) Consider three homeowners in marginal federal income tax brackets of 30%, 20%, and 10%, respectively, each taking a home mortgage interest deduction of $1000. The highest bracket taxpayer receives a subsidy of $300 (his $1000 deduction multiplied by his marginal tax bracket of 30%). Similarly, the 20% bracket taxpayer receives a $200 subsidy for his homeownership and the 10% bracket taxpayer receives a $100 subsidy.
This pattern, a function of the progressive rate structure, provides assistance in an "upside-down" fashion; it sends the greatest aid ($300) to the taxpayer in the highest bracket, who, by virtue of his higher income, presumably needs such aid least of the three. Conversely, the taxpayer with the smallest income receives the least assistance from the mortgage interest deduction because the deduction is worth less to him because of his low marginal rate.
Procedurally, the argument continues, this regressive distribution of public largesse reflects a failure of legislative scrutiny. (12) Legislators, unaware of the expenditure implications of tax subsidies, do not review tax subsidies with the same care they give to direct budgetary outlays. No legislature, Professor Surrey and his supporters contended, would enact a direct spending program that conveys its benefits in upside-down fashion, nor would a legislature vote for such a tax subsidy if the subsidy's true nature were disclosed. (13) Hence, there is a need for "tax expenditure budgets" that identify and quantify proposed and actual tax expenditures. The truth, as revealed by tax expenditure budgets, will lead legislatures to recoil from the practice of channeling public expenditures through the tax system. (14)
As a procedural program, tax expenditure analysis has enjoyed remarkable success. The primary procedural prescription of the Surrey school--tax expenditure budgets to identify and quantify proposed and enacted tax expenditures--has become a central feature of federal and state budgetary processes. Besides the federal tax expenditure budgets, (15) over thirty states now promulgate such budgets as part of their fiscal processes. (16)
Contrary to the expectations of Professor Surrey and his supporters, tax expenditure budgets have, however, failed to curb the legislative appetite for tax-based expenditures. Victor Thuronyi's 1988 observation that "Congress has made little progress in replacing tax expenditures with direct spending programs" (17) remains true today. More recently, Professor Roin has noted that despite greater "political scrutiny and oversight" of federal tax expenditures, "the introduction of the tax expenditure budget did not markedly succeed in lowering the growth of tax expenditures." (18) While both of these commentators were writing about the federal government, their conclusion is also true of the states. Contrary to the prediction of the Surrey school, tax expenditure budgets and the information they provide have not caused state legislatures to reduce tax expenditures. (19)
Given the profusion and institutionalization of tax expenditure budgets, it is no longer credible to claim that tax expenditures reflect inadequate information. Rather, the fact that disclosure has not led to legislative restraint suggests, at least in part, that tax expenditures engender framing effects. Some people simply do not believe that tax and direct expenditures are the same, even when such expenditures are economically and procedurally identical. For these individuals, labels matter. Indeed, for these individuals, the labels attached to public largesse obscure the similarities of otherwise identical policies.
Despite its influence on tax policy discourse and on federal and state budgetary processes, Professor Surrey's analysis has received criticism from many quarters. (20) One aspect of Professor Surrey's argument is particularly important for the framing effect issue. Professor Surrey contended that a dollar is a dollar whether awarded as direct spending or through a tax expenditure program. (21) This argument has obvious intuitive appeal but is incomplete because ignores such concerns as the permanence of the program dispensing the dollar, the criteria for determining who is eligible to receive the dollar, and the existence of limits on the total number of dollars that can be received by any recipient. (22)
Consider, for example, a housing subsidy granted through a classic tax expenditure, embedded in a permanent tax code. Assume further that this subsidy is available to all homeowner taxpayers and is uncapped as to dollar amount since the taxpayers receive a subsidy equal to a percentage of their housing outlays, which they, themselves, determine. Contrast this prototypical tax incentive with a classic direct outlay subsidy for housing--a fixed dollar voucher subject to the annual appropriation process, available to a predetermined number of homeowners selected annually by the executive agency administering the program and limited to a maximum annual amount per selected homeowner.
Under the tax program, which roughly corresponds to the current mortgage interest deduction, (23) a taxpayer deciding to purchase a home may view the tax subsidy as more secure and more valuable than the direct outlay assistance because the tax subsidy is part of a permanent tax code. The taxpayer can qualify for the subsidy simply by purchasing his or her house, and the subsidy will automatically increase as the taxpayer's borrowing increases. In contrast, the direct subsidy for housing seems less reliable because the legislature must authorize the subsidy each year, the homeowner must pass administrative muster annually to receive the subsidy, and, once the taxpayer reaches the maximum annual amount, there is no further subvention of the homeowner's housing expenditures.
The point is that a dollar is not always a dollar. In this example, even if the dollar amounts are the same, a classic tax expenditure may be more secure and thus more valuable to the taxpayer than a direct subsidy program.
Upon reconsideration, however, the choice between the direct and the tax subsidy may look less stark. While, as a legal matter, the legislature must authorize the direct expenditure subsidy annually, in practice that authorization may be routine. The housing agency, as a matter of policy or practice, may consistently renew the subsidy once a homeowner has been selected for the program. Homeowners' housing expenditures may leave them well short of the program's maximum assistance.
Moreover, not all tax expenditures and direct outlays take classic forms. Entitlement-type spending, for example, can partake of some or all of the characteristics typically associated with tax-based programs. Suppose, for example, that, to impart greater political security to homeowners, the legislature formulates the direct housing subsidy as a permanent entitlement that is not subject to annual review, administrative discretion, or quantitative limit. Under this permanent entitlement program, all homeowners in the state, upon the presentation of documentation that they paid their mortgages, will receive a check for a percentage of their mortgage outlays. Comparing this entitlement-type subsidy with the tax-based alternative, the two programs are essentially identical as to permanence, eligibility, and unlimited quantity. Accordingly, in this setting, there is more force to Professor Surrey's argument that a dollar from either program is equivalent since the programs are the same in terms of their permanence, open eligibility, and uncapped quantities.
Consider, too, a so-called "rifle-shot" tax transition provision, a one-time tax benefit for a particular taxpayer. While such one-time public largesse comes through the tax system, that largesse looks less like a classic tax expenditure--embedded in a permanent tax code, available to all eligible, and unlimited in amount--and more like a direct appropriation, targeted once to a particular taxpayer.
In assessing whether tax expenditures create framing effects, it is important to control for these differences and thereby isolate the framing effect phenomenon from these other features. Those deciding to undertake the training necessary to become volunteer firefighters may incur the opportunity costs of that training more comfortably if they are confident that those costs will be reimbursed by a permanent property tax abatement, that is, one beyond the city council's annual review. In contrast, a city councilman may prefer to reimburse volunteer firefighters through the annual appropriation process that allows him to bestow largesse each year upon an important political constituency by affirmatively voting for that year's appropriation. In these cases, a preference for either the tax or the direct expenditure does not reflect the framing effect phenomenon but, rather, stems from an assessment of the comparative merits of permanent subsidies relative to those reviewed annually.
In contrast, if we control for the relevant features that might distinguish tax from direct expenditures, any perceived difference in the two would reflect a framing effect, in which the label creates the perception of difference when, substantively and procedurally, there is none.
In short, there are characteristics of tax expenditures and direct spending programs that can plausibly lead to preferences for one over the other, depending on the particular features of the program. To isolate the framing effect phenomenon we must control for these features.
III. FORMAL STUDIES AND FRAMING EFFECTS
In recent years, legal scholars have embraced the notion of framing effects. (24) A seminal demonstration of framing effects was a now-classic and much emulated experiment in which Professors Tversky and Kahneman asked two comparable but separate groups to decide between two alternative policies in the face of an impending epidemic. (25) If neither policy were implemented, all respondents were told, the epidemic would inevitably cause 600 deaths. Of the two possible responses to the epidemic, one had a single certain outcome and would save 200 of the 600 who would otherwise die. In contrast, the second response presented to the respondents had two possible outcomes, a one-third possibility that all 600 deaths would be avoided or a two-thirds possibility that none of the 600 deaths would be avoided. Thus, the first alternative, if pursued, would with certainty save a portion of those who would otherwise perish from the imminent epidemic while the second policy would result in either all or none of the 600 surviving. (26)
When the alternatives were framed for one group in terms of those saved ("200 people will be saved"), 72% of those presented with this labeling opted for the more predictable policy that would with certainty save 200 of those who would otherwise die from the epidemic. When the alternatives were expressed to the second, otherwise comparable, group in terms of those dying ("400 people will die"), a similar majority of 78% selected the risky policy that might save everyone or might save no one. (27)
Thus, the labels mattered to a large percentage of the Tversky/Kahneman respondents. A policy favored when framed in terms of survival ("200 people will be saved") was rejected by a comparable group when cast in terms of death ("400 people will die")--even though, as a substantive matter, it was the same policy with the same projected outcome. (28)
In their formal experiment, Professors Tversky and Kahneman informed neither group of the way in which the choice was framed for the other set of respondents. After obtaining their results, Professors Tversky and Kahneman informally presented some members of their sample with the alternatives labeled both ways (i.e., expressed once in terms of survival rates and once in terms of death rates). Many of these individuals persisted in deciding differently depending upon the labels, favoring the more certain outcome when it was expressed as 200 lives definitely saved but rejecting that outcome when it was framed as 400 lives predictably lost. When Professors Tversky and Kahneman confronted these respondents with the substantive inconsistency of their choices, many remained firm in their selections. This led Professors Tversky and Kahneman to conclude that "[i]n the persistence of their appeal, framing effects resemble visual illusions more than computational errors." (29)
The demonstration of framing effects has now become something of a cottage industry. The format pioneered by Professors Tversky and Kahneman--substantively identical but differently framed alternatives presented separately to each of two comparable groups--has become a research mainstay. For example, in an experiment measuring framing effects in a financial context, Professors Fetherstonhaugh and Ross asked two groups of respondents about their choices between alternative retirement ages when those ages are associated with different levels of annual social security payments. (30) Professors Fetherstonhaugh and Ross presented to both groups a choice between retiring at age sixty-five and receiving $10,000 yearly from social security or retiring at age sixty-eight and receiving $12,500 annually. To one set of respondents, Professors Fetherstonhaugh and Ross described the extra $2500 at age sixty-eight affirmatively, as "a credit of $2,500 a year for delaying retirement" from sixty-five to sixty-eight. To the other group of respondents, they characterized the same $2500 difference negatively, as "a penalty of $2,500 a year for retiring early" at age sixty-five rather than at age sixty-eight. (31)
For a significant proportion of respondents, this labeling mattered. When the extra annual payment was described to one group as a credit of $2500 yearly for late retirement, a majority (62%) of that group elected age sixty-five as the preferred retirement date. When the $2500 annual difference was described to the second set as a penalty for retiring earlier, however, a majority (57%) of that group opted for age sixty-eight and the higher annual payments associated with that later retirement age. Thus, 19% of this sample manifested the framing effect phenomenon; (32) comparable groups' preferences shifted between economically identical choices depending upon the characterization of the same $2500 as an annual penalty for early retirement or as an annual credit for delayed retirement.
When Professors Fetherstonhaugh and Ross stratified their sample by age and by income, they found that, in their choices of a retirement age, older individuals (age forty-seven or higher) and less affluent persons (those earning under $60,000 per year) were more sensitive to the label attached to the extra $2500 payable annually at age sixty-eight than were younger and more affluent individuals. (33)
Professors Fetherstonhaugh and Ross repeated their experiment using age sixty-two as the earlier retirement date with annual social security payments of $10,000 and age sixty-five as the delayed retirement date with the larger annual pension of $12,500. (34) In this setting, the framing effect did not materialize. In particular, when the extra $2500 yearly was described as a "credit" for postponing retirement from age sixty-two to age sixty-five, 49% of respondents preferred retirement at age sixty-two. When the extra $2500 was characterized for the second group as a "penalty" for retiring at the younger age, essentially the same proportion (51%) opted for the earlier retirement age. (35) Thus, framing differently the choice between age sixty-two and age sixty-five made no discernible difference to the alternatives selected.
Several inferences can be drawn from the foregoing. First, framing effects do not always occur when they might be expected. For a significant portion of Professors Fetherstonhaugh and Ross's sample, the labeling (credit or penalty) attached to the extra $2500 per year affected the choice between retirement at age sixty-five and at age sixty-eight, but not the choice between retirement at age sixty-two and at age sixty-five. Second, even when framing effects occur, many people do not succumb to those effects. Among respondents, roughly 40% favored retirement at age sixty-eight irrespective of the framing while approximately the same percentage elected retirement at age sixty-two regardless of such framing. It was a "swing vote" of roughly 20% of respondents that manifested the framing effect phenomenon. Their responses depended on the label attached to the extra annual payment.
In Professors Tversky and Kahneman's experiment, smaller, but still noticeable, percentages of the respondents did not succumb to the framing effect. (36) However labeled, roughly 25% of the respondents favored the policy with the single certain outcome while approximately the same proportion supported the riskier approach. (37)
Third, when framing effects occur, they can occur in numbers that are potentially decisive. In the Tversky/Kahneman sample, roughly half of the respondents were sensitive to the manner in which the alternatives were labeled. In the Fetherstonhaugh/Ross experiment, the framing effect was less pronounced but still potentially important. Think of the Fetherstonhaugh/Ross sample as a voting public, deciding by majority vote whether to set normal retirement age for the community at sixty-five or at sixty-eight. In this context, the framing effect would decisively affect the outcome by turning a majority vote for one alternative into a majority for the other, depending upon the label--credit or penalty--attached to the annual pension differential.
Indeed, the percentage of individuals succumbing to a framing effect could be even smaller than the affected percentage of the Fetherstonhaugh/Ross sample and still be decisive. Whether the individual manifesting a framing effect is perceived as the median voter of a political process (38) or as a marginal participant in a market, that individual may exercise substantial influence on the outcomes reached by political or market processes.
Finally, framing effects may occur in different populations at different rates. In the Fetherstonhaugh/Ross sample, for example, the framing effect, when present, occurred in a more pronounced fashion among older and poorer respondents. (39)
IV. FIREFIGHTERS AND TAX EXEMPTIONS
Against this background of framing effects and their impact on formal studies, consider the increasingly widespread practice of abating the property taxes of volunteer firefighters, a practice that reflects both difficulty in staffing volunteer fire departments and reluctance to shift to paid firefighting forces.
The contemporary challenges of recruiting and retaining volunteer firefighters (including ambulance and EMT volunteers) are widespread and widely-noted. (40) In terms of causation, it is useful to separate the broader social trends underlying these difficulties from the problems peculiarly associated with attracting and holding volunteer firefighters. As a general matter, traditional fraternal and volunteer institutions are in secular decline. We might designate this part of the problem as the "Bowling Alone" phenomenon, after Robert D. Putnam's influential study. (41) From this perspective, it is not surprising that volunteer fire departments today have trouble attracting and retaining members. Most other voluntary institutions are facing similar problems.
However, there is more to the membership difficulties of volunteer fire departments than the Bowling Alone problem. Unique to such departments are the increasingly onerous training requirements that have been imposed on firefighters, including volunteers, over the last generation. In an earlier, simpler age, fighting a fire meant standing outside and spraying water onto the conflagration. Like other aspects of our society, firefighting has become more complicated in terms of equipment and tactics. (42) Complex equipment and tactics require training and practice which, in turn, require the firefighters' time and effort. (43) Consequently, being a volunteer firefighter today involves a greater commitment than it did a generation ago. (44)
As the opportunity costs of being a volunteer firefighter have increased, the job has become no less demanding physically. As aging Baby Boomers fall from the ranks of volunteer fire departments because of the increased time involved in firefighting and physical constraints, their replacements must be drawn from the smaller ranks of the post-Baby Boom generation. (45)
These concerns were voiced by firefighter Bob Klink, who found himself the only volunteer firefighter at his station in a rural section of Washington largely populated by retirees, and of chief Thomas Lowe, the fire chief for the district encompassing Klink's station. (46) Reflecting on the training required today to be a volunteer firefighter, Klink observed that, "(y)ears ago, if you had a six-pack, you could have joined." (47) This is no longer the case. As Chief Lowe stated, today, "[y]ou can't just put ... gear on a person and say, 'I've got a fireman.'" (48) The recruitment problem for Mr. Klink's station is compounded by the older nature of the population being served and the demanding physical requirements for firefighting.
So the question becomes: Why don't communities simply shift to paid firefighting forces? To an extent they are, but the change is usually difficult and is often resisted. (49)
Many factors explain the reluctance to pay firefighters. Existing volunteers, who enjoy their current status, resist being replaced by paid professionals. Taxpayers, while unwilling to volunteer, may also be unwilling to pay. There is powerful, indeed iconic, imagery underpinning the appeal of the citizen-volunteer, a modern-day Cincinnatus who drops his civilian role in the face of emergency, fights fire, and then returns to his regular job. That imagery is reinforced by the instincts of some voter-taxpayers that professionalized government agencies develop their own agendas and the political clout to make taxpayers pay for those agendas. (50)
Whatever the causes (the foregoing is admittedly suggestive and not exhaustive), strong support remains for keeping fire departments as volunteer institutions despite the difficulties of attracting and retaining volunteer firefighters. It is here that the story of volunteer firefighting intersects with tax expenditures and framing effects. It has become an increasingly widespread practice to encourage volunteer firefighting by abating the property taxes of volunteer firefighters. (51)
Tax expenditure analysis indicates that such property tax abatements are an exercise in self-delusion and that, once the economic equivalence of tax relief and cash payments are revealed, voters and decision makers will understand that "volunteer" firefighters whose property taxes are reduced for their service receive remuneration substantively identical to cash compensation. In contrast, the literature on framing effects suggests that disclosure will not end the delusion, that labeling largesse for volunteer firefighters as tax relief obscures, for a significant number of individuals, the similarity of such tax relief to equivalent cash payments.
There might be, as I suggest above, sensible procedural reasons for a volunteer firefighter to prefer tax abatement to a cash payment. The tax abatement, if embedded in a permanent tax code, may be more secure politically than a cash payment, which must be appropriated annually. When deciding whether to undergo or continue the training necessary to become firefighters, individuals might more comfortably commit their time, energy, and out-of-pocket expenses to the enterprise if they know that the financial reward will take the more politically secure form of a permanent property tax abatement.
Moreover, the chief of a fire department may prefer tax abatement as a recruitment and retention device if it is structured in classic tax form (that is, if it is available to all comers). This will allow the chief to recruit and retain as many volunteers as he can secure without bumping against an appropriations ceiling. In contrast, an appropriation in its prototypical form will be fixed in amount, thereby limiting the largesse the chief can distribute.
If we control for the relevant procedural factors, however, there is no substantive difference between giving a firefighter a check for $500 or abating $500 of his property tax liability. Either way, the firefighter benefits from his service to the extent of $500.
There are plausible reasons for concluding that a subsidy for a firefighter does not impair his volunteer status (e.g., the subsidy is nominal, in the nature of an honorarium). There are also plausible reasons for deciding that a subsidy for a firefighter negates his volunteer standing; any compensation, no matter how small, may be incompatible with the volunteer ethic. However, it is irrational to conclude that a cash payment converts a firefighter into a paid employee while deciding that an otherwise identical tax subsidy is consistent with continued volunteer standing.
Nevertheless, the framing effect literature suggests that, even if we control for the factors that might properly distinguish direct payments from tax assistance, some--perhaps many--individuals will perceive otherwise equivalent tax abatements and cash payments as different. Thus, to borrow an apt phrase from Professors Tversky and Kahneman, tax abatements for volunteer firefighters are "visual illusions" that disguise for many individuals the reality that "volunteers" receiving tax relief are receiving the economic equivalent of cash. (52)
V. MEASURING FOR FRAMING EFFECTS
To test this proposition, I developed three questionnaires. (53) Each questionnaire started with the same premise: Elmtown has a volunteer fire department and wants to keep that department as a volunteer institution, rather than shift to paid firefighters. The community, however, has serious problems recruiting and retaining volunteer firefighters.
In one questionnaire, the proposed response was a permanent, annual property tax reduction for ten firefighters. (54) In a second questionnaire, the proposed response was a permanent, annual payment for ten firefighters. (55) These two questionnaires followed the widely-emulated two-groups/two-questionnaires format pioneered by Professors Tversky and Kahneman, with one questionnaire presenting one set of respondents the choice framed one way and the other questionnaire offering a second, comparable set of respondents the substantively identical choice labeled differently. Neither group was informed of the manner in which the choice was framed for the other.
In a refinement of the standard two-groups/two-questionnaires format, however, I also developed a third questionnaire for administration to a third, separate set of respondents. (56) This third questionnaire presented to those respondents both possibilities for ten firefighters (i.e., a permanent annual property tax reduction and a permanent annual cash payment). Thus, this third questionnaire disclosed to a third group of participants both of the alternatives for framing the substantive choices (as tax reductions or as cash payments) in order to assess the framing effect internal to that group.
Each of the three questionnaires asked the respondent whether, in the respondent's judgment, the proposed tax reduction or payment impairs the firefighter's status as a volunteer. Each respondent was asked this question for three levels of annual largesse: $500, $2000, and $5000.
Consequently, the questionnaires were designed to control for characteristics that might otherwise influence the choice between a tax reduction and a cash payment. Both the tax and the payment alternatives were presented as permanent programs, capped in the amount each firefighter can receive annually (whether $500, $2000, or $5000) and limited to ten firefighters. Thus, the only difference between the property tax abatement and the cash payment is how the assistance is framed--as tax relief or as a cash payment.
To explore possible demographic influences, the questionnaires asked the respondents to provide their gender and their ages. Respondents were also given an opportunity to give comments, in narrative form.
On September 9-10, 2004, I administered the first two of these questionnaires in New York City to first-year students at the Benjamin N. Cardozo School of Law of Yeshiva University. One hundred and three first-year students were given the questionnaire that asked whether cash payments to firefighters impair the firefighters' volunteer status. (57) A comparable group of one hundred and ten first year students was given the questionnaire that asked whether equivalent property tax reductions for firefighters jeopardize their volunteer standing. (58) The results are presented in Exhibit D.
A framing effect emerged at all three compensation levels. (59) The framing effect is statistically significant (60) and becomes more pronounced as the compensation levels rise. (61) Of the group asked whether a firefighter receiving a $500 payment is still a volunteer, a majority (56%) concluded that such a firefighter retains volunteer standing. However, as to the second group asked about a $500 tax reduction, a substantially larger majority (71%) concluded that that economically and procedurally equivalent tax reduction is compatible with volunteer status. Thus, comparing the responses of these two similar sets of respondents, a statistically significant swing vote of 15% was influenced by the manner in which the economic largesse was labeled. This swing vote concluded in the first group that $500 cash payments negate volunteer status but decided in the second group that equivalent tax expenditures do not--even though the two forms of assistance are procedurally and substantively identical.
At the level of $2000 of annual subvention, a slightly greater framing effect appeared than at the $500 level. In the first group, only a minority (44%) indicated that a $2000 payment squares with firefighters' volunteer status. However, among the second set of respondents, a substantial majority (61%) deemed a comparable tax exemption of $2000 consistent with volunteer standing. Thus, at the $2000 level, the swing vote manifesting the framing effect grew to 17%. That swing vote turned a 56% majority which perceived a $2000 payment to negate firefighters' volunteer status into an even larger majority (61%) which saw an equivalent $2000 tax subsidy as consistent with such status, even though the two types of compensation are substantively and procedurally the same.
At the $5000 level, only 30% of respondents viewed a payment to firefighters as consistent with volunteer status. Thus, by more than two-to-one, respondents in the first group concluded that a $5000 payment to a firefighter impaired his or her volunteer standing. In contrast, at the $5000 level, respondents in the second group were evenly divided, with 49% viewing a $5000 property tax abatement as consistent with firefighters' volunteer status. At this level, the swing vote influenced by the label of the largesse grew to 19%. (62)
For those looking for evidence of rational choices, the good news is that roughly 80% of respondents did not succumb to the framing effect and were unaffected by the manner in which firefighters were to be compensated. (63) At the $2000 level, for example, roughly 40% of the respondents consistently concluded that annual compensation transforms a firefighter into a professional, whether that compensation takes the form of a payment or of tax relief. Another 40% of these respondents opined that $2000 yearly does not convert a firefighter into a professional, whether that largesse is bestowed as a direct payment or through the tax system.
However, approximately 20% of respondents fell victim to the hypothesized framing effect. Labeling largesse for volunteer firefighters as tax reductions caused approximately 20% more of those given this tax-based alternative to view the subsidy as compatible with volunteer status as compared to the members of a comparable group asked to assess the impact on volunteer standing of an equivalent cash payment. A swing vote of this magnitude is evidently enough to affect the outcome in a majoritarian electoral process. Indeed, in this example, the swing vote subject to the framing effect would control in such a process. The proposition that $2000 cash payments are consistent with volunteer status was defeated with only 44% support while the counterproposition--that $2000 property tax abatements are compatible with volunteer standing--passed comfortably in a comparable group with 61% assent. The framing effect made the difference in these two outcomes among otherwise similar sets of respondents confronting policies which were procedurally and substantively similar but labeled differently.
Organizing the data by gender reveals that the female student/respondents succumbed to the framing effect at rates significantly higher than did their male counterparts. (64) Indeed, at the $2000 and $5000 compensation levels, the framing effect was largely (though not exclusively) a female phenomenon. (65) While males succumbed to the framing effect at the $2000 level, the effect at that level was significantly less pronounced than for the female respondents. Among males, an 11% swing vote viewed a $2000 tax reduction as compatible with volunteer status but $2000 cash payments as not. This result fell short of statistical significance. (66) However, among female respondents, almost one-in-four reflected the framing effect (i.e., viewed a $2000 tax reduction but not an equivalent $2000 payment as consistent with firefighters' volunteer standing). Thus, female respondents manifested the framing effect at statistically significant rates, (67) viewing a $2000 payment as impairing a firefighter's volunteer status but perceiving a procedurally and substantively identical $2000 tax subsidy as consistent with such status.
This gender gap persisted at the $5000 compensation level. At that level of subsidy, the swing vote manifesting the framing effect amounted to 13% of the male respondents but almost double that percentage of the female respondents. Again, the framing effect was not statistically significant for the male respondents (68) but was statistically significant for their female classmates. (69)
I can only speculate as the cause of this gender divide. It could be a statistical fluke, though the significance tests performed on the data suggest that this is not likely. More plausible is the possibility that more male students studied subjects in college such as business, math, and economics that enabled them to see the equivalence of subsidies, however labeled.
Also suggestive are some of the comments on these questionnaires, comments which tend to reinforce the impressions suggested by the data. (70) A frequently articulated theme was that property tax reductions are "not payments," but are instead "incentives," "benefits," or "gifts." One respondent wrote, "A reduction in property taxes is not a weekly salary." Labels matter. Indeed, labels obscure.
On September 14, 2004, I administered the third questionnaire (71) in New York City to sixty-five students in my introductory federal income tax class at Cardozo. (72) The results are presented in Exhibit F and suggest, depending upon one's perspective, the limited efficacy of more complete disclosure or the persistence of framing effects in the face of more complete information. (73)
Of this third sample of respondents (presented with both the tax and the payment alternatives), 57% indicated that a $500 payment to a firefighter is consistent with volunteer status while 66% simultaneously viewed an equivalent $500 tax exemption as compatible with firefighters' volunteer standing. Thus, at the $500 level, nine percent of respondents presented simultaneously with both the tax and the payment options succumbed to the framing effect, concluding that substantially and procedurally identical subsidies had differing impacts on firefighters' voluntary standing.
By comparison, 56% of the group presented only the $500 payment option saw a firefighter receiving such a payment as remaining a volunteer while 71% of the sample given only the tax option thought that the firefighter granted the $500 property tax exemption maintains his volunteer status. Thus, at the $500 level, 15% of respondents subject to the traditional two-groups/two-questionnaires methodology perceived differently labeled but otherwise identical forms of compensation as impacting volunteer status.
These results pose the proverbial question whether the glass is half-empty or half-full. Better disclosure, namely, giving respondents both the tax and the payment alternatives together, reduced the framing effect to nine percent of the sample (i.e., the percentage of the third group which declared a $500 tax reduction but not an equivalent $500 payment as compatible with volunteer standing). At this level, the framing effect was not statistically significant for the third group. (74) However, even in the face of the better disclosure made to this group (i.e., the existence of both the tax and the direct payment possibilities), the framing effect was persistent: approximately 10% of the respondents in the third group, informed of both the tax and the payment alternatives, concluded that a $500 payment impairs a firefighter's volunteer status while an equivalent $500 tax abatement does not. (75) These respondents, alerted by the questionnaire to the possibility of both tax and cash payments, nevertheless persisted in viewing the two forms of subsidy as different in their implications for the recipients' volunteer standing. Labels matter even in the face of more complete disclosure.
The results at the $2000 level were similar. Thirty-four percent of the sample presented with both the tax and payment alternatives deemed a payment consistent with volunteer status while 43% of this group declared an equivalent tax exemption compatible with volunteer standing. The comparable figures under the traditional, two group methodology were 44% and 61%, respectively. (76) For the third group, at the $2000 level, the framing effect was not statistically significant. (77)
However, at the $5000 level, the framing effect detected using the single questionnaire approached the framing effect measured by the traditional, two-groups/two-questionnaires methodology. Of the respondents asked simultaneously the consequences of a $5000 payment and a $5000 tax reduction, 15% viewed the payment as compatible with volunteer status while 31% saw the equivalent tax reduction as consistent with volunteer standing. The resulting swing vote--16% perceiving a cash payment but not an identical tax reduction as impairing volunteer status--is almost the same as the 19% swing vote measured under the two-groups/two-questionnaire technique. At the $5000 compensation level, the glass is considerably less than half full. In the third group of respondents (presented the $5000 subsidy framed both as a tax reduction and as a cash payment), almost the same percentage viewed the tax and the payment alternatives differently as did the percentage measured by the traditional methodology under which respondents were not informed of the other alternative. At the $5000 level, the framing effect was statistically significant for the third group, (78) despite the better information furnished to it.
VI. IMPLICATIONS
In light of this data, it was astute of those advocating compensation for volunteer firefighters to have chosen property tax abatement, rather than equivalent cash grants, as the preferred means for achieving that compensation. The survey results suggest that a critical segment of the voting public--those subject to the framing effect phenomenon--views tax relief, but not similar cash payments, as consistent with firefighters' volunteer status. Insofar as these voters also value volunteer firefighting departments, they will support tax relief as compatible with volunteer standing but will oppose economically and procedurally identical cash payments as converting firefighters into paid professionals. Labels matter.
It is possible that my sample of law students--all college graduates--understates the extent to which the general voting public manifests the framing effect phenomenon. If so, it is even more understandable that advocates of compensation for firefighters push for tax relief, rather than equivalent cash payments, since an even larger segment of the voting population sees a difference in identical but differently-framed programs. Depending on how largesse for firefighters is labeled, many view cash, but not equivalent tax reductions, as transforming volunteers into paid employees. This data also suggests that some firefighters who cherish their volunteer status likely feel more comfortable with tax relief than with cash subsidies, perceiving the latter, but not the former, as impairing their volunteer status--despite the comparability of these differently-labeled subsidies.
Particularly suggestive are the data indicating that, as the hypothesized levels of compensation increase, the framing effect becomes more pronounced. As larger cash payments make firefighters look more professional to more people, the perception gap grows between such cash payments and the equivalent tax expenditures. At the $500 compensation level, the framing effect under the traditional two-groups/two-questionnaires methodology influenced 15% of the sampled population which viewed a tax abatement, but not an identical cash payment, as compatible with volunteer status. (79) At the $2000 compensation level, using the conventional two-group methodology, the swing vote manifesting the framing effect increased to 17% while, at the $5000 level, 19% of comparable respondents viewed a tax reduction (but not an equivalent cash payment) as consistent with volunteer standing. (80) Similarly, for the group given the third questionnaire (with both the tax and the direct payment alternatives), the 16% of the sample manifesting the framing effect at the $5000 level was higher than the 9% of respondents viewing tax reductions and cash payments differently at the $500 and $2000 levels. (81)
In practice, tax abatements for volunteer firefighters tend to be granted today at the lower end of the scale. To the extent that increases will be sought in the future, however, the proponents of compensating volunteer firefighters have chosen wisely in pursuing tax relief rather than cash payments. As the level of relief climbs, the perception gap widens, making tax-based compensation the more viable political choice in an environment that values firefighters' volunteer status.
In light of the foregoing, it is not surprising that the strategy of abating volunteer firefighters' property taxes is being emulated. Some states now grant such firefighters tax credits against their state income tax liabilities. (82) My survey results suggest that among the other reasons for adopting such credits is the portion of the public subject to the framing effect. That portion of the public views tax credits (but not equivalent cash payments from the state treasury) as consistent with volunteer status. For a significant segment of the population, channeling largesse through the tax system obscures the reality that that tax-based largesse is substantively and procedurally identical to an equivalent cash payment.
Also revealing is the rapid growth of state income tax deductions for organ donors. (83) My results suggest why the proponents of compensating organ donors for their costs, like those who favor compensation for volunteer firefighters, have chosen the route of tax-based assistance. A cash payment to an organ donor raises the specter that the donor is being paid for a body part, which many people find objectionable. As a result of the framing effect, however, a significant portion of the populace views a tax credit to an organ donor as different from an economically and procedurally equivalent cash payment to that donor. Labels matter.
More generally, groups seeking public largesse prefer tax relief, as a matter of political strategy, because, for a critical segment of the public, public subsidy framed as tax relief is different from, and less objectionable than, equivalent cash payments.
The implications of my survey results are sobering for tax expenditure analysis, with its heavy emphasis upon disclosure through tax expenditure budgets. If, as these results suggest, tax expenditures create framing effects for a significant part of the population, it is unsurprising that tax expenditure budgets and the information they convey have failed to curb tax expenditures. The essence of a framing effect is that disclosure does not matter; "framing effects resemble visual illusions more than computational errors." (84)
Thus, tax expenditure budgets and the data they contain have not inhibited the adoption of tax expenditures because such expenditures create, for important sections of the public, framing effects. Labels matter. Labeling largesse as tax relief matters for a critical portion of the voting public that persists in viewing tax-based relief as different from substantively and procedurally equivalent direct outlays.
Consider in this context Professor Stephen Diamond's observation about Progressive opposition to tax exemptions: "Progressives assumed that disagreement was caused by inadequate information and that obtaining facts and publicizing them would lead to action." (85) That observation is equally true of the proponents of tax expenditures, grounded in a Progressive-like belief that better information will cause legislatures to curb tax subsidies.
Today, in a sadder, if not wiser, age, we understand that information is often a necessary, but rarely sufficient, condition for sound public policy. The most common criticism of the Progressives' managerialist perspective is that they ignored interests and interest groups. Public policy in a democracy responds, not just to information, but to constituencies and their desires.
Framing effects suggest an additional reason why better information does not necessarily make for better public policy. "Visual illusion" cannot be cured by more information like tax expenditure budgets, and an important segment of the population persists in the visual illusion that tax expenditures are different from comparable direct outlays.
There are additional reasons for the persistence of tax expenditures in the face of the now-extensive disclosure mandated by federal and state tax expenditure budgets. As a matter of marginal transactions costs, it is relatively efficient to communicate government policies and programs through the tax system. (86) Professors Weisbach and Nussim have recently argued that, as an issue of "institutional design," it is often best to integrate certain expenditure-type programs with the core functions of the tax system. (87) As noted above, such considerations as permanence, eligibility, and limits may lead to a preference for tax expenditures over direct spending programs. The legislative committees that design tax subsidies and the administrative agencies that implement such subsidies are less susceptible to political capture than are the committees and agencies that oversee direct expenditure programs. (88)
Along with these benign explanations for the persistence of tax expenditures, however, the results of my questionnaires indicate that, for some people and in some circumstances, tax expenditures create framing effects, leading to an acceptance of tax expenditures by persons who would oppose identical programs labeled as direct outlays.
Thus, the Michigan opponents of tax expenditure analysis are right that labels matter, but they are wrong that the label "tax expenditure" carries the weight they fear it does. For an important segment of the population, tax-based subsidies (or, at least, property tax abatements for volunteer firefighters) are different from direct outlays.
Finally, my results indicate the usefulness of alternative ways of measuring the framing effect phenomenon. Conventionally, framing effect studies today use the two-group/two questionnaires paradigm pioneered by Professors Tversky and Kahneman, that is, administering to two separate but comparable groups separate questionnaires, not informing either group of the way the choice is framed for the other. (89) In exploring the relationships between differently labeled subsidies and perceptions of firefighters' volunteer standing, it was useful to push beyond this now conventional paradigm by administering a third questionnaire that disclosed to a single group the choice framed both ways. In this manner, the third questionnaire identified the framing effect internal to the third sample group based on disclosure to the members of that group of the two, differently-framed alternatives. In effect, the third questionnaire did rigorously what Professors Tversky and Kahneman did less formally when they interviewed some of their respondents to determine the efficacy of disclosing to them the inconsistency of their choices.
In my research, the third questionnaire produced some provocative results. (90) At the $500 and $2000 levels, decision making improved somewhat from better information. In particular, simultaneous disclosure to the third group of both the payment and the tax alternatives reduced the framing effect below statistical significance. However, the benefits of disclosure diminished at the $5000 level. At that level, the percentage of the third group (16%) viewing identical tax and spending programs differently was statistically significant and approached the percentage (19%) similarly succumbing to the framing effect as measured by the conventional two-groups/two-questionnaires methodology. Even when the framing effect was less pronounced under the third questionnaire, it was quite noticeable, involving roughly 10% of respondents, more than enough to make the difference in a close election.
I suspect that other framing effect studies might benefit from this third questionnaire technique.
VII. CONCLUSION
The increasingly common adoption of property tax exemptions for volunteer firefighters helps to explore the paradox of tax expenditure analysis: Despite greater understanding that tax expenditures and direct outlays may be substantively and procedurally equivalent, this understanding has had no apparent impact on public policy. Tax expenditures continue to proliferate because they create framing effects. For a significant portion of the public, the label--tax relief versus direct outlay--matters. For these individuals, policies unacceptable when framed as direct expenditures become desirable when labeled as tax subsidies, even though the policies are procedurally and substantively equivalent. For these individuals, labels matter; indeed, labels obscure the comparability of comparable, but differently-framed, policies.
EXHIBIT A
(TAX ABATEMENT ALTERNATIVE)
Elmtown has a volunteer fire department and wants to keep the department as a volunteer institution, rather than shift to paid firefighters. However, Elmtown has serious problems recruiting new volunteer firefighters and retaining such volunteers.
To address this problem, it has been proposed that, on a permanent basis, Elmtown reduce the property taxes of ten (10) firefighters. Some opponents of this proposal claim that reducing property taxes impairs the firefighters' volunteer status, turning them into paid professionals. Proponents of the proposal disagree. Please tell us what you think by circling your answers below.
1) If Elmtown permanently reduces $500 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
2) If Elmtown permanently reduces $2,000 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
3) If Elmtown permanently reduces $5,000 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
Your background information
Gender:
Male
Female
Age: ______
Thanks for your time. Please put any additional comments here:
EXHIBIT B
(CASH PAYMENT ALTERNATIVE)
Elmtown has a volunteer fire department and wants to keep the department as a volunteer institution, rather than shift to paid firefighters. However, Elmtown has serious problems recruiting new volunteer firefighters and retaining such volunteers.
To address this problem, it has been proposed that, on a permanent basis, Elmtown pay ten (10) firefighters. Some opponents of this proposal claim that these payments would impair the firefighters' volunteer status, turning them into paid professionals. Proponents of the proposal disagree. Please tell us what you think by circling your answers below.
1) If Elmtown permanently pays $500 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
2) If Elmtown permanently pays $2,000 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
3) If Elmtown permanently pays $5,000 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
Your background information
Gender:
Male
Female
Age: ______
Thanks for your time. Please put any additional comments here:
EXHIBIT C
(CASH PAYMENT AND TAX ALTERNATIVES)
Elmtown has a volunteer fire department and wants to keep the department as a volunteer institution, rather than shift to paid firefighters. However, Elmtown has serious problems recruiting new volunteer firefighters and retaining such volunteers.
To address this problem, it has been proposed that, on a permanent basis, Elmtown either pay ten (10) firefighters or reduce their property taxes. Some opponents of this proposal claim that these proposals would impair the firefighters' volunteer status, turning them into paid professionals. Proponents of these proposals disagree. Please tell us what you think by circling your answers below.
1) If Elmtown permanently pays $500 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
2) If Elmtown permanently reduces $500 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
3) If Elmtown permanently pays $2,000 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
4) If Elmtown permanently reduces $2,000 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
5) If Elmtown permanently pays $5,000 every year to ten (10) firefighters, are these firefighters still volunteers?
Yes
No
6) If Elmtown permanently reduces $5,000 of property taxes every year for ten (10) firefighters, are these firefighters still volunteers?
Yes
No
Your background information
Gender: Male Female
Age: ______
Thanks for your time. Please put any additional comments here:
EXHIBIT D (RESULTS OF SEPTEMBER 9 AND 10, 2004; TWO-GROUPS/TWO
QUESTIONNAIRE METHODOLOGY)
RAW NUMBER
PERCENTAGE ANSWERING
ANSWERING "YES"/TOTAL
QUESTION "YES" ANSWERING
Are firefighters ... as a 56% 58/103
volunteers if payment?
they receive ... as a tax 71% 78/110 (91)
$500 ... reduction?
Are firefighters ... as a 44% 45/103
volunteers if payment?
they receive ... as a tax 61% 67/110 (92)
$2,000 ... reduction?
Are firefighters ... as a 30% 31/103
volunteers if payment?
they receive ... as a tax 49% 54/110 (93)
$5,000 ... reduction?
EXHIBIT E (RESULTS OF SEPTEMBER 9 AND 10, 2004; TWO-GROUPS/TWO-
QUESTIONNAIRE METHODOLOGY; BY GENDER)
RAW NUMBER
ANSWERING
PERCENTAGE "YES"/TOTAL
QUESTION ANSWERING "YES" ANSWERING
... as a Males 56% 30/54
Are payment?
firefighters ... as a tax 69% 37/54 (94)
volunteers if reduction?
they receive ... as a Females 57% 28/49
$500 ... payment?
... as a tax 73% 41/56 (95)
reduction?
Are ... as a Males 43% 23/54
firefighters payment?
volunteers if ... as a tax 54% 29/54 (96)
they receive reduction?
$2,000 ... ... as a Females 45% 22/49
payment?
... as a tax 68% 38/56 (97)
reduction?
Are ... as a Males 31% 17/54
firefighters payment?
volunteers if ... as a tax 44% 24/54 (98)
they receive reduction?
$5,000 ... ... as a Females 29% 14/49
payment?
... as a tax 54% 30/56 (99)
reduction?
EXHIBIT F (RESULTS OF SEPTEMBER 14, 2004; THIRD QUESTIONNAIRE GIVEN TO
THIRD GROUP)
RAW NUMBER
PERCENTAGE ANSWERING
QUESTION ANSWERING "YES"/TOTAL
"YES" ANSWERING
Are firefighters ... as a 57% 37/65
volunteers if payment?
they receive ... as a tax 66% 43/65 (100)
$500 ... reduction?
Are firefighters ... as a 34% 22/65
volunteers if payment?
they receive ... as a tax 43% 28/65 (101)
$2,000 ... reduction?
Are firefighters ... as a 15% 10/65
volunteers if payment?
they receive ... as a tax 31% 20/65 (102)
$5,000 ... reduction?
(1) MICH. COMP. LAWS ANN. [section] 21.271 (West 2003).
(2) Robert Kleine, Michigan House OKs Name Change for Expenditure Report, 2003 ST. TAX TODAY 118-19 (stating that "[c]ritics say that the concept is biased toward government spending rather than toward taxpayers because it appears to assume that money belongs to the government unless the tax laws specifically permit taxpayers to keep it"). Robert Kleine, Michigan House Would Change Reference to "Tax Expenditure" Reporting, 2003 ST. TAX TODAY 118-20 (stating that "[o]n the surface the bill seems insignificant, but it underscores the strong antitax stance of the current Republican-dominated Legislature and could be the first step to doing away with the state tax expenditure report").
(3) A qualified success for the tax expenditure movement has been the increased tendency of Congress over the years to enact tax expenditures as credits (rather than as deductions and exclusions) and Congress's related tendency to phase-out tax expenditures as the taxpayer's income rises. These tendencies in significant measure reflect the distributional teachings of tax expenditure analysis (i.e., that deductions and exclusions are regressive in impact). Hence, Congress resorts to credits and phase-outs to channel tax subsidies to lower- and middle-bracket taxpayers. See, e.g., David A. Weisbach & Jacob Nussim, The Integration of Tax and Spending Programs, 113 YALE L.J. 955, 979 (2004) (stating that "Congress now tends to use credits rather than deductions"). These victories for the tax expenditure movement are limited, however, falling short of that movement's major substantive goal: the elimination of tax expenditures altogether.
(4) The literature on tax expenditure analysis is now voluminous. For an introduction to this literature, see WILLIAM D. ANDREWS, BASIC FEDERAL INCOME TAXATION 401-14 (5th ed. 1999); STANLEY S. SURREY & PAUL R. MCDANIEL, TAX EXPENDITURES 3 (1985); Edward A. Zelinsky, James Madison and Public Choice at Gucci Gulch: A Procedural Defense of Tax Expenditures and Tax Institutions, 102 YALE L.J. 1165, 1168-71 (1993).
(5) See SURREY & MCDANIEL, supra note 4, at 3.
(6) Id.
(7) Id.
(8) Id.
(9) Id.
(10) Id.
(11) Id.
(12) Id.
(13) Id.
(14) Moreover, the Surrey critique continued, tax writers and administrators lack substantive expertise in the areas at which tax expenditures are targeted. Id. To continue the example of the mortgage interest deduction, if that deduction is seen for what it is--a housing subsidy--it should be reviewed and administered, not by tax writers and agencies, but by legislative committees and executive departments with expertise in housing. Tax expenditure budgets, by highlighting the reality that the mortgage interest deduction is a housing program, should facilitate review and administration of that program by legislators and bureaucrats with competence in housing. For a critique of this argument, see Zelinsky, supra note 4.
(15) See, e.g., 2 U.S.C. [section] 639(c)(3) (2004).
(16) Herman P. Ayayo, Tax Expenditures: Useful Economic Concept or Budgetary Dinosaur?, 93 TAX NOTES 1152, 1153 (2001); see also Frank Shafroth, "The Tax Doctor": The Strange State of Tax Expenditures, 2004 ST. TAX TODAY 119-5 (stating that "[m]ore than 30 states and the federal government issue annual tax expenditure reports").
(17) Victor Thuronyi, Tax Expenditures: A Reassessment, 1988 DUKE L.J. 1155, 1171.
(18) Julie Roin, Truth in Government: Beyond the Tax Expenditure Budget, 54 HASTINGS L.J. 603, 623-24 (2003).
(19) See, e.g., Shafroth, supra note 16, at 119-5 (noting that "[s]tates and local governments tend to be heavy users of tax expenditures in the belief that they are critical tools for economic development, but without a full appreciation of their import").
(20) Among other criticisms of tax expenditure analysis is the argument that the distinction between normative and expenditure provisions is unsound. Boris I. Bittker, Accounting for Federal "Tax Subsidies" in the National Budget, 22 NAT'L TAX J. 244 (1969). Scholars following Professor Bittker have argued further that particular provisions of the federal income tax, classified by the Surrey school as tax expenditures, are plausibly characterized instead as normative provisions. I have argued, for example, that the Code's treatment of qualified plans is best understood as a normative, rather than an expenditure, provision of the Code. See Edward A. Zelinsky, The Defined Contribution Paradigm, 114 YALE L.J. 451 (2004).
(21) See SURREY & MCDANIEL, supra note 4.
(22) See Edward A. Zelinsky, Are Tax "Benefits" Constitutionally Equivalent to Direct Expenditures?, 112 HARV. L. REV. 379, 400-09 (1998) (discussing these issues in greater detail).
(23) The current mortgage interest deduction is limited to the interest generated by the first $1,000,000 of acquisition debt, a limit that affects relatively few taxpayers. See I.R.C. [section] 163(h)(3)(B)(ii).
(24) See, e.g., Edward J. McCaffery, Cognitive Theory and Tax, 41 UCLA L. REV. 1861, 1863, 1905 (1994); Cass R. Sunstein & Richard H. Thaler, Libertarian Paternalism Is Not an Oxymoron, 70 U. CHI. L. REV. 1159, 1179 (2003).
(25) See Amos Tversky & Daniel Kahneman, Rational Choice and the Framing of Decisions, in DECISION MAKING: DESCRIPTIVE, NORMATIVE, AND PRESCRIPTIVE INTERACTIONS 167, 175 (David E. Bell et al. eds., 1988).
(26) Id. Both alternatives had an expected value of 200 lives saved. Under the first choice, there is a certainty of saving two hundred persons. Under the second policy, the possibilities of saving everyone or no one lead to the same expected value (i.e., two hundred persons saved). (1/3 * 600 + 2/3 * 0 = 200 + 0 = 200.)
(27) Id.
(28) Id.
(29) Id. A critical, typically unstated, premise of these framing effect studies is that the two groups questioned are for all relevant purposes the same and thus would produce similar outcomes if the questions were framed the same way for both.
(30) David Fetherstonhaugh & Lee Ross, Framing Effects and Income Flow Preferences in Decisions About Social Security, in BEHAVIORAL DIMENSIONS OF RETIREMENT ECONOMICS 187, 187-88 (Henry J. Aaron ed., 1999).
(31) Id. at 197.
(32) Thirty-eight percent of respondents in one group preferred retirement at age sixty-eight using the "credit" or gain nomenclature while 57% of respondents in the second group preferred that retirement age using the rubric of "penalty" or loss. Id. at 198.
(33) Id. at 198-99.
(34) Id. at 199-200.
(35) Id. at 199-201.
(36) See Tversky & Kahneman, supra note 25, at 175.
(37) Id.
(38) For an excellent summary and application of the median voter hypothesis, see WILLIAM A. FISCHEL, THE HOMEVOTER HYPOTHESIS: HOW HOME VALUES INFLUENCE LOCAL GOVERNMENT TAXATION, SCHOOL FINANCE, AND LAND-USE POLICIES 87-92 (2001).
(39) See Fetherstonhaugh & Ross, supra note 30, at 198-99.
(40) See, e.g., ROBERT D. PUTNAM, BOWLING ALONE 130-31 (2000) (stating that "the nationwide ratio of volunteers to professional firefighters fell by a quarter between 1983 and 1997").
(41) See id. at 127-33; see also Peter Dobkin Hall, Is Tax Exemption Intrinsic or Contingent?, in PROPERTY-TAX EXEMPTION FOR CHARITIES 276 (Evelyn Brody, ed., 2002) (noting that, in a study of New Haven, Connecticut, the long-term decline of "fraternal/sororal and other mutual-benefit and membership organizations" and their replacement by professional "human service, health, education, and arts and culture providers, most of which received substantial government subsidies").
(42) My thinking on this issue was assisted greatly by the comments of Professor John Sciliano who himself has extensive experience as a volunteer firefighter.
(43) See, e.g., Letter from Robert Andrews, U.S. Representative (D-N.J. 1st), to House of Representatives of 5/02/01 (supporting H.R. 1651, the Volunteer Firefighters' Protection Act of 2001 and stating that "[t]hese volunteers meet the same training and performance standards as paid, career firefighters"); see also Volunteer Firefighters' Protection Act of 2003, H.R. 3375, 108th Cong. (2003). Oklahoma's income tax credit for volunteer firefighters requires, inter alia, annual certification of training for "chemical, biological, radiological, and nuclear response(s)." See Kenneth L. Hunt, Oklahoma Offers New Income Tax Credit for Firefighters, 2004 ST. TAX TODAY 176-16.
(44) See, e.g., CONN. AGENCIES REGS. [section] 7-3231-80 (2004) (adopting the National Fire Protection Association's Professional Qualifications Standard 1001).
(45) The aging of the Baby Boomers is central to Professor Putnam's analysis of the decline of volunteer fire fighting forces. See PUTNAM, supra note 40, at 130-31.
(46) Lynn Steinberg, Buddy, Can You Spare Some Time?: The Only Volunteer Firefighter in Diamond Point is One Lonely Guy, SEATTLE POST-INTELLIGENCER, Mar. 28, 1996, at F1.
(47) Id.
(48) Id.
(49) See, e.g., Richard Lezin Jones, "Volunteer" Versus "Firefighter", N.Y. TIMES, Dec. 10, 2003, at B1 (describing "a roiling dispute that has raged [in Stamford, Connecticut] for the last year between paid and unpaid firefighters").
(50) For a classic statement of this position, see WILLIAM A. NISKANEN, JR., BUREAUCRACY AND REPRESENTATIVE GOVERNMENT (1971).
(51) See, e.g., ALASKA STAT. [section] 29.45.050(r) (Michie 2004); CONN. GEN. STAT. [section] 12-81w (2003); N.Y. REAL PROP. TAX LAW [section] 466-c, 466-d (Consol. 2004). For a discussion of the federal income tax consequences of these property tax abatement programs, see Thomas J. Monks, Income Tax Consequences of Connecticut's Emergency Volunteer Municipal Property Tax Reduction Program, CONN. BAR ASS'N TAX NEWSL., Mar. 24, 2004, at 17.
(52) See Tversky & Kahneman, supra note 25, at 175.
(53) See infra Exhibits A, B, and C.
(54) See infra Exhibit A.
(55) See infra Exhibit B.
(56) See infra Exhibit C.
(57) See infra Exhibit B.
(58) See infra Exhibit A.
(59) See infra Exhibit D.
(60) See infra Exhibits D, E, and F.
(61) See infra Exhibit D.
(62) Id. This 19% is coincidentally the same percentage which Professors Fetherstonhaugh and Ross found succumbing to a framing effect in their research. See Fetherstonhaugh & Ross, supra note 30, at 198.
(63) See infra Exhibit D.
(64) See infra Exhibit E (organizing the data by gender). Age did not prove a useful predictor, largely because my respondents (law students) were clustered within a relatively narrow range of ages.
(65) Id. At the $500 level, there was slightly less difference between the responses of the male students and the answers given by their female colleagues. Id.
(66) See infra note 96.
(67) See infra note 97.
(68) See infra note 98.
(69) See infra note 99.
(70) Some of these comments were made in the trial run I conducted in the summer of 2004.
(71) See infra Exhibit C.
(72) As this was the beginning of the semester, the students had not yet been introduced to tax expenditure analysis.
(73) See infra Exhibit F.
(74) See infra note 100.
(75) See infra Exhibit F.
(76) See infra Exhibits D and E.
(77) See infra note 101.
(78) See infra note 102.
(79) See infra Exhibit D.
(80) Id.
(81) See infra Exhibit F.
(82) See, e.g., Hunt, supra note 43, at 176-16; see also DEL. CODE ANN. tit. 30, [section] 1113 (2004) (providing a credit against state income tax liability for the actual expenses of volunteer firefighters and other service responders). This credit was recently increased to a maximum of $400 annually. In at least one state, teachers are now seeking similar tax credits. From this perspective, it is both troubling and unsurprising that other groups have started to seek similar property tax reductions for themselves. See, e.g., Dave Wasson, Teachers' Property Tax Break Moving Through Both Florida Chambers, 2004 ST. TAX TODAY 58-7.
(83) See, e.g., Steven S. Woo, Wisconsin Leads Way as States Consider Organ Donor Deductions, 2004 ST. TAX TODAY 173-1.
(84) Tversky & Kahneman, supra note 25, at 175.
(85) Stephen Diamond, Efficiency and Benevolence: Philanthropic Tax Exemptions in 19th-Century America, in PROPERTY-TAX EXEMPTION FOR CHARITIES: MAPPING THE BATTLEFIELD 115, 139 (Evelyn Brody ed., 2002).
(86) See Edward A. Zelinsky, Efficiency and Income Taxes: The Rehabilitation of Tax Incentives, 64 TEX. L. REV. 973 (1986); Edward A. Zelinsky, The Tax Reform Act of 1986: A Response to Professor Yorio and His Vision of the Future of the Internal Revenue Code, 55 FORDHAM L. REV. 885, 892-94 (1987).
(87) Weisbach & Nussim, supra note 3, at 957.
(88) For a development of this argument, see Zelinsky, supra note 4.
(89) See Tversky & Kahneman, supra note 25.
(90) See infra Exhibit F.
(91) These proportions are statistically significant with 97-98.5% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.216 -2.216
p-value 0.0267 0.01334
(92) These proportions are statistically significant with 98-99.5% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.515 -2.515
p-value 0.01189 0.00595
(93) These proportions are statistically significant with 99.5-99.8% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.82 -2.82
p-value 0.00467 0.00234
(94) These proportions are not statistically significant. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.388 -1.388
p-value 0.1651 0.0826
(95) These proportions are statistically significant with 91.5-95.9% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.7309 -1.7309
p-value 0.0835 0.04174
(96) These proportions are not statistically significant. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.1555 -1.1555
p-value 0.2479 0.1239
(97) These proportions are statistically significant with 98-99% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.372 -2.372
p-value 0.01771 0.00885
(98) These proportions are not statistically significant. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.388 -1.388
p-value 0.1651 0.0826
(99) These proportions are statistically significant with 99-99.5% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.5902 -2.5902
p-value 0.0096 0.0048
(100) These proportions are not statistically significant. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.0817 -1.0817
p-value 0.2794 0.1397
(101) These proportions are not statistically significant. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -1.0817 -1.0817
p-value 0.2794 0.1397
(102) These proportions are statistically significant with 96-98% confidence. Specifically:
Test 1: P(a) [not equal to] P(b) Test 2: P(a) < P(b)
t-stat -2.081 -2.081
p-value 0.0374 0.0187
Edward A. Zelinsky*
* Professor of Law, Benjamin N. Cardozo School of Law of Yeshiva University. The author wishes to thank his Cardozo colleagues, Professors Lester Brickman, Dan Crane, Mitchell Engler, Peter Goodrich, Arthur Jacobson, Melanie Leslie, Stewart Sterk, and Paul Shupack, and Professor John Sciliano for sharing his reflections on his experience as a volunteer firefighter. The author also acknowledges the beneficial research assistance of Robin Grossman, a member of the Cardozo Law School class of 2005 and the statistical assistance of Christopher Ferraro, a member of the Yale College class of 2006.
In addition, make sure to read these articles:
- CalPERS Names Darryl Watson Member Services Division Chief; Fund Reorganizes Regional...
- Free Cost Savings Calculator from NVFC Foundation Offered to Fire Departments.
- A First for Rhode Island Firefighters - Bladder Cancer Screening is Underway in Five...
- Message processing in realistic recruitment practices.
- Message processing in realistic recruitment practices
- The influence of ethical attitudes on taxpayer compliance.
- Cognitive Errors, Individual Differences, and Paternalism
- Expected utility theory vs. prospect theory: implications for strategic decision makers.
- The Outcome Saliency Effect on Negotiated Transfer Prices
- Decision Aids for Generating Analytical Review Alternatives: The Impact of Goal Framing...
Email
Digg It
del.icio.us
Related Resources
