Journal of Economic North-Holland
13 (1992) 521-543
Tax evasion and inequity F.A. Cowell * London School of Economics, London, UK Received
24, 1992; accepted
A number of concepts of inequity are examined in a model that incorporates tax-enforcement parameters, personal resources, personal attributes, and perceived inequity. The rBle of risk aversion and its interaction with attitudes toward inequity, interaction with other tax evaders, and the means that each person has of influencing inequity are considered as determinants of individual tax-compliance decisions.
1. Introduction Although taxation and tax policy do not enjoy the reputation of being terribly fascinating subjects for the layman, there are some issues in this area which nearly always arouse considerable interest. Fairness and cheating are perhaps two of the themes which do make a substantial impact on the consciousness of Joe Public. The saga of the community charge/poll tax in the United Kingdom is a particularly fascinating example of the way in which both these themes can grip the popular imagination. Writers on tax affairs have commented on the dangers of demoralisation of those who pay but perceive the system as unfair (see Karzon 1983). However, in terms of the economic analysis of taxation there appears to be a surprising gap. Despite the interest of welfare economists in issues of social justice, fairness and equity, and although economic theory has been applied to tax evasion, the area of overlap between the two has until recently been relatively neglected. Correspondence to: F.A. &well, London London WC2A 2AE, UK. * I am grateful to Jim Gordon, participants journal for useful comments.
0 1992 - Elsevier
in the IAREP
and to two referees
B.V. All rights reserved
Street, of this
F.A. CoweN / Tax emsion and inequity
In fact it could be argued that conventional economic analysis takes a rather narrow-minded view of tax evaders. Typically economic models of tax evasion pose the problem in an individualistic fashion, either as a game against nature or, in more sophisticated versions of the recent economics literature, as a game against the tax authority. However, the issue of inequity, according to virtually any interpretation one may give to the term, is essentially a matter of social interaction rather than being individualistic in nature. Although some economists have examined situations in which the influence of the prevailing social climate upon each taxpayer’s willingness to engage in evasion, on the whole the economics literature has not tried to model the way in which social phenomena influence individual actions, though Benjamini and Maital (19851, Bordignon (19911, Falkinger (19881, Gordon (1989) and Schlicht (1985) are notable exceptions. In particular, it neglects the possibility that people’s participation in evasion may be driven by their perceptions of injustices or inequities in the tax system, or inequity in the system of economic rewards on which the tax system has been based. In this paper I shall examine the structure of social relationships within an economic analysis of tax evasion, and examine the r61e of a person’s perceptions of those relationships in influencing his actions. I do this by extending the standard economic model of tax evasion to incorporate interrelationships amongst taxpayers (evaders), and the areas of common concern to those individuals. The purpose of the paper is to clarity the way in which this structure affects the workings of the model of the ‘shadow economy’, and to compare the results with those from the more limited individualistic approach that is conventionally presented in the economic literature. Of course it is quite easy to highlight the limitations of the oversimplified ‘tax-evader-as-gambler’ model and refer in vague terms to the importance of including in one’s model factors arising from the social context. A more challenging task is to try to identify the specific economic mechanisms which link individual actions in complying or failing to comply with tax laws to particular aspects of injustice which may characterise the community to which the taxpayers belong. Section 2 considers the case for incorporating inequity into an economic model; the basics of this model are then set out in section 3. Sections 4 to 6 discuss three possible interpretations of inequity within the economic model. Section 7 concludes.
F.A. CoweN / Tax wasion and inequity
2. Inequity, uncertainty and tax compliance Connecting the topics of inequity and the economics of tax evasion poses two contrasting questions: What are the grounds for introducing considerations of inequity into an explanatory model of tax evasion? And, if perceptions of inequity - or indeed other social phenomena or personal characteristics - have a powerful impact upon the phenomenon of non-compliance, then why should one persist with the optimisation-under-uncertainty approach to the problem? In response to the first question - why include inequi~? - let us consider two general points. First, there is a good deal of both circumstantial and specific evidence that values derived from the community are important in determining people’s compliance or non-compliance with tax laws. Evasion behaviour seems to differ markedly between groups within the population (Groenland and Van Veldhoven 1983; Witte and Woodbu~ 1985) as well as between countries. Obviously some of these differences can be accounted for by differences in opportunities for evasion - for example the self-employed as against employees but it is also the case that individual taxpayers appear to take into account the ‘climate’ within the community or the groups to which they belong. This is borne out in investigations by direct questioning of admitted tax evaders (see for instance Wallschutzky 1984); whether other members of the community are known to evade appears to exert a strong influence on individual decisions, for example (Porcano 1988). Even though some studies which have been able to check self-reported behaviour against people’s actions have cast doubt upon the reliability of self-reports as a sole source of evidence (Elffers et al. 1987; and Hessing et al. 19881, the case is reinforced by the evidence of ‘climate’ from more broadly-based surveys of taxpayers’ attitudes (see for example Striimpel 1969; Song and Yarbrough 1973). Moreover, in studies of the control of evasion, it appears that social mechanisms or institutions that appeal to civic responsibility or to conscience may actually be more effective than legal sanctions in ensuring tax compliance (see Schwartz and Orleans 1967). This variety of evidence for the influence of social climate prompts the thought that various notions of social justice, fairness, or equity, are likely to be integral parts of this climate. Therefore some representation of
F.A. CoweN / Tax ecmion and inequity
inequity in taxation should be taken into account in analysing compliance behaviour. Second, tax perceptions - as distinct from specific tax rates appear to be important in determining evasion behaviour. For example there is evidence to suggest that the structure of taxes has an influence on people’s willingness to evade, over and above any income or wealth effects that are induced by the tax rates (Baldry 1986); the perception that the government may be using tax proceeds for inappropriate purposes also plays a role in predisposing people to evade taxes (cf. Vogel 1974); furthermore the perceived share in publiclysupplied goods financed by the taxes may be important as well as the perception of tax burden (Becker et al. 1986; Mackscheidt 1984). So, in forming decisions about whether, or how much, to comply with tax laws people seem to take into account what the tax system ‘looks like’ over and above simple considerations about the tax system’s implication for their own pockets. However, what if one were to move from the general topics of social climate and tax perceptions to pose directly the specific question of whether social or economic inequity specifically affects evasion behaviour? There is clear evidence from questionnaire studies for the explicit influence of various aspects of perceived inequity (Yankelovich et al. 1984; Hite 19901, although when some other social factors are included in the study the particular ‘inequity variable’ chosen may not appear to be so important. However, despite the importance of the tax structure on the evasion decision, as noted above, the evidence from experimental studies on the impact of perceived inequity is ambiguous. Spicer and Becker (1980) found that evasion was high for those who perceived substantial inequities in the tax system, whilst Webley et al. (1985) and Webley et al. (1988) found the opposite; Kaplan and Reekers (1985) found that beliefs about fairness of the tax system were less important than beliefs about the morality of tax cheating (see also Kaplan et al. 1986). Does the fact that some of the experimental studies find that such perceived inequity is unimportant as a determinant of evasion - in contrast to the evidence from attitude surveys - undermine the case for incorporating inequity in the tax evasion model? It could be the case that laboratory experiments pick up the ‘true’ - perhaps neglible - r6le of inequity, whilst the interview responses attributing evasion behaviour to inequity are just a socially acceptable RXpost rationalisa-
F.A. Cowell / Tax evasion and inequity
tion of base motives (in which case perhaps we might as well stick to the basic economic model); but an alternative explanation may be sought in the methodology of tax experiments. It is well known that a substantial burden rests on the nature and conduct of the experiment: people may modify their behaviour in certain ways simply because they are aware that it is an experiment and some key results may be highly sensitive to the particular experimental design (for example contrast the results of Webley and Halstead (19861, Webley et al. (1985), and those of Friedland et al. (1978) on the issue of the relative deterrent effectiveness audit probability and fines). Since inequity is essentially a social phenomenon - involving the tax treatment, behaviour or economic position of others - the context of the decisions seems to be crucial. It is interesting to note, for example, that when people are asked in experiments to assume that they themselves are the evador they respond differently as to the factors causing evasion from the way they would respond if it were a third party doing the evasion (see Hite 1987; Thurman 1988). So it could be that participants in experiments do not - perhaps cannot - perceive themselves to be in the same kind of environment as if perceived inequity and the tax-compliance/tax-evasion decision really involved their income and their neighbours. The upshot of this suggests that one concur with Mason and Calvin’s (1984) point that the effect of perceived equity in evasion is probably not simple and direct. Nevertheless, there is a good case for incorporating inequity in a tax-compliance model; but if considerations of equity or inequity are to be included in the model then clearly it is important what one actually means by the terms. ‘Inequity’ itself is conventionally interpreted in a variety of ways - as an unfair ‘exchange’ between citizens and State, as unwarranted disparities in tax assessments, or as perceived income inequality, for example. It is not clear that it means the same thing to different researchers, nor that the perceptions of equity (however defined) will be uniform across the members of a particular group. In this paper I shall treat inequity within the context of on economic model, but will draw on concepts that are relevant to both psychologists’ and economists’ approach to the subject. Finally, let us consider the second question, of why - if one is to build a model in which inequity is taken seriously - it is still worth working within the standard economic approach of taxpayer optimisa-
F.A. Coweli / Tax ermion and inequity
tion under uncertainty. It has of course become a commonplace to note that the choice between tax compliance and tax evasion is far more complex than just another form of gambling (Spicer 1975). However, even if tax evasion is not a gamble - in the sense that people’s decisions on whether to comply and how much income to conceal are not determined purely by the financial returns - it is still a risky business. If it were not, then either the detection of non-compliance would be immediate and certain, so that the problem would not actually arise in practice, or alternatively non-compliance would never be punished, in which case (whatever the law might state formally) everyone would know that tax payment in effect amounted to a voluntary contribution. In either case we would not really have very much to talk about; so it seems that people’s reaction to risk is bound to be a fundamental component of any comprehensive model of tax evasion. Furthermore, we might expect that evasion behaviour will be influenced by the interaction between people’s perceptions of inequity and their attitudes to the risk that is inevitably borne when a person engages in tax evasion.
3. Inequity and the evasion decision: A basic model With this thought in mind let me introduce a framework for analysis of the role of inequity in tax compliance. I shall employ a fairly standard economic model, modified to include alternative versions of ‘inequity’ that are related to the concepts discussed in the previous section. Although it is an economic model it can be used to illustrate the way in which issues of inequity that are of interest to psychologists and others interact with other components of the mechanism of economic decision making. The model has four basic components, the taxpayer’s opportunities, the tax-enforcement system, social factors, and taxpayer motivation.
The taxpayer’s opportunities. These encompass the resources available to the individual taxpayer and the range of economic choices that he faces. I shall adopt a very simple, standard model of both of these: we assume an atemporal world in which there is only the one tax and either one or two economic goods. Let y denote a person’s income, which is liable to tax at a proportional rate t. Suppose that the person
F.A. Cow& / Tax ellasion and inequity
actually reports to the tax authority an income x := y - e where e is the amount of evasion (concealed income). Because the model is atemporal we may identify disposable income with consumption. The only scope for choice that the individual has is the amount of income reported to or concealed from the government. However, this feature of the model is not perhaps as restrictive as first it might seem: as we shall see, under certain specifications of the model, the person may be able to affect inequity directly through these choices. The tax and enforcement system. The tax authority determines the tax rate t and also has the power to investigate and punish tax fraud. This power means that there is a given probability of detection and conviction p. Those found evading tax are subjected to a fine; for simplicity let us suppose that can be represented as a simple surcharge s on the tax that has been underpaid. I shall refer to the triple rr := (p, s, t) as the tax-enforcement parameters. These two components of the problem mean that the penalty on detected evasion is proportional to the amount of evasion, and that the disposable income (consumption) that is available to the tax evader is: c:=[l-t]y+ret,
where r is the ‘rate of return’ implicit in the tax-enforcement ters r; this rate of return is given by
(see, for example, plays an important
Cowell 19901. As we shall see, this rate of return role in the decision framework of the tax evader.
Social factors. I shall assume that these are summarised by an index of inequity i which can be modelled in a number of ways so as to incorporate alternative concepts of inequity within the economic model. The index i may for example be a function of the taxes imposed, of the incomes of other individuals in the community, and of the supply of publicly provided goods and services. For the moment
F.A. Coweli / Tax ec,asion and inequity
we shall not investigate the determinants of inequity i in further detail. However, it is worth noting at this point that the working of the model depends crucially on whether the opportunities of the individual include the possibili~ of altering perceived inequity through actions of his own. It is also worth noting that one person’s perception of a particular type of inequity may in practice differ from another (see taxpayer motivation, below): for the purposes of this paper these differences are not important if all persons’ perceptions may be assumed to respond in the same direction to the economic and social factors generating the inequity. For example two taxpayers may differ in their perceptions of the level of public good provision: this does not matter as long as both perceive the supply of public goods to have increased when there is an actual increase in the supply of public goods. Taxpayer motiuafion. Individual taxpayers may differ from one another in a variety of ways that will influence their disposition to tax evasion and, if they evade, by how much. I shall suppose that we may conveniently categorise individual taxpayers by a pair of variables ( y, a), where y is income and a is a list of exogenously given attributes, which may also incorporate information that is used by the tax authority in assessing tax. These attributes can be taken to include a number of aspects of tax evasion, their attitude to risk taking, and their attitudes to inequity. They are thus of central importance in the following key assumptions about the way a potential tax evader’s decisions are driven. These assumptions are founded on extreme economic orthodoxy:
can be represented by a von Neumann-Morgenstern utility function that is increasing and concave in consumption, and decreasing in inequity. Expected
utility: The taxpayer’s preferences
What this means is that the person evaluates the uncertain prospect generated by the possibility of investigation and punishment of tax evasion in terms of the disposable income (consumption) that he would get under each of the possible events that could occur (there are only two events in our simple model). This evaluation is the (subjective) expectation of the payoff (utility) generated under each possible event, and the payoff depends on perceived inequity as well
F.A. CoweN / Tax ermion
as disposable income; the assumption of concavity of the utility function means that the person is risk-averse. The person can influence the payoffs by his choice ex-ante of evasion e. We also need to assume something about the way in which the tax authority’s actions are linked to the decision problem of the taxpayer. Accurate tax and enforcement knows what the tax-enforcement
Although this assumption might appear fairly innocuous to many economists, I do not claim it to be realistic. People’s attitudes to risk and hence their awareness of the relevant detection probabilities appear to be strongly influenced by the context in which tax payment and tax evasion takes place. For example their attitudes may depend on their perceptions of the nature of the tax payment - whether it is perceived as reduced income or as a loss - see Chang et al. (1987). There is also experimental evidence that the probability of audit is more effective a deterrent for taxpayers who have already been audited than for others (see Spicer and Hero 1985; Webley 1987). Furthermore in some experimental studies of tax evasion it appears that some subjects were willing to accept unfair risks in order to carry out evasion; they still took the chance of being ‘caught’ even though the expected rate of return was negative, and in some cases exceeded - 100% (see, for example, Webley et al. 1991). This suggests that subjects could not work out accurately either the probabilities of various payoffs, or the implications of those probabilities (see Cowell 1991). In the present case these considerations are of secondary importance if individuals consistently follow erroneous beliefs about the probability that their tax evasion will be punished; it is not the purpose of this paper to investigate policy questions such as changes in audit probability where the issue of correct perceptions would become crucial (see, for example, Sheffrin and Triest 1990). So we may take these ‘awareness factors’ as given, and accept that some ( y, a> types may in practice under- or over-estimate the probability p. If people systematically overestimate p, then it may be the case that some taxpayers will be frightened into absolute compliance, even though Br > 0. Taking these two assumptions and noting that the tax-enforcement parameters are incorporated in the specification of disposable income
F.A. Cowell / Tux el,asion and inequity
(consumption) function as
c in (1) above
Bu(c, i; a),
where u is a standard concave cardinal utility function that is increasing in consumption (disposable income), and decreasing in inequity, and B denotes the expectations operator, given the probabilistic structure (2). Notice that the specification (3) is rich enough to encompass the ideas that a single evader’s actions respond to financial incentives (as set out in the tax-enforcement parameters r above), that these actions are dependent on personal characteristics (including tastes, attitudes to risk, and so on, included in a, and as well as income y>, and that they are influenced by social structure (the index of inequity i). Evasion behaviour according to this model can then be described as follows. The individual maximises (3) subject to (1) and (2). The first-order conditions for this can be summarised as
where subscripts here and elsewhere derivatives ‘. Expression (4) implicitly function for evasion:
’ For example: tlu(c, i; a) u, (c. i; a) is shorthand
i)‘u(c, i; a) i; U) is shorthand
refer to the appropriate defines the individual’s
the basis for a number
F.A. Cowell / Tax ec>asionand inequity
conclusions and for empirical estimation of the effects on evasion behaviour of tax-enforcement parameters of income (which enters through the consumption term in expression (4)) and of other taxpayer characteristics (Clotfelter 1983). To do this of course requires specification - explicitly or implicitly - of a particular functional form for preferences as represented by the function U, as well as a more detailed specification of the tax-enforcement parameters. Let us first examine the broad conclusions that can be derived from this without committing oneself to any one special case of this basic model. One immediate, well-known conclusion from this model is that any (y, a> type for whom risk aversion is not infinite, and who has the opportunity to evade, will without exception conceal some income if the expected rate of return to evasion 8r is positive. This is true irrespective of the role of inequity within the utility-maximisation model: it is an immediate consequence of the accurate perceptions assumption being invoked along with expected utility maximisation and the fact that penalties for evasion are proportional to evasion e. This conclusion - that taxpayers only comply fully if forced into doing so by draconian tax enforcement that ensures 8r I 0 - is one of the reasons why some investigators of the tax-evasion problem have found this model to unsatisfactory. It is clearly at odds with models that focus primarily on factors such as personality variables rather than risk aversion. This basic model can be used to examine the way in which evasion changes in response to perceived inequity, given that the optimisation-under-uncertainty condition (4) is satisfied. To do this we need to introduce a number of alternative specifications of perceptions which may be incorporated into the solution function (5).
4. Evasion and the exchange relationship The perception that one is not getting a fair deal from the government in return for one’s tax dollars has long been acknowledged both in the intuitive understanding of people’s motives for tax evasion and in the literature on fiscal psychology. Wallschutzky (1984, 1988) found that the ‘exchange relationship hypothesis’ was important as a possible cause of evasion. If taxpayers see the exchange of their taxes for public goods or other social services as ‘unequal’ - in the sense that
F.A. Cowell / Tax wasion
the exchange is unfairly loaded in favour of is argued that they may in effect withdraw tionship by evading taxes so as to offset inequity; alternatively, if perceived inequity tage he may actually reduce evasion (see, Becker 1980). In order to incorporate this notion let us
the government - then it from this exchange relaor reduce this apparent is to the person’s advanfor example, Spicer and model inequity
where i is increasing in the tax rate t, and decreasing in the person’s perceived share in public goods g. In order to make this specification operational within a model of taxpayer’s evasion decisions one has of course to specify how a person’s share in the overall supply of public goods and services G is determined: one may write this as some function g(G, a). However, one also needs to specify how G itself is perceived by the individual to be affected by his own evasion. It is clear that given the simple proportional tax structure outlined in section 3. we have
where F is the distribution tion and
of the ( y, a) types in the popula-
represents aggregate evasion activity. In relating the response of this aggregate to the activity of any one individual I shall make the standard assumption for a large economy: Self-perceived insignificance: The individual acts as though the evasion activity of others is unaffected by marginal changes in his own evasion activity. This implies that each tax evader acts as though his anti-social behaviour leaves the total supply of public goods unchanged. Natu-
F.A. Cowell / Tax evasion and inequity
rally G will be affected by the collection of evasion decisions over all ( y, a> types, but that is beside the point. It is perceptions that count, and this assumption means that each person is ‘small’ relative to the community, so that (according to his own perception) no action of his can directly affect the magnitude of overall perceived inequity i. Note that this is in contrast to the assumption that Falkinger (1988) uses in his standard case that each tax evader assumes that others change their evasion in exactly the same way as he himself does. Using this assumption we can get the effect of an exogenous change in perceived inequity, induced perhaps by a change in G on the person’s chosen evasion activity; if we differentiate the first-order condition (eq. (4)), we obtain: e*(y, a,
ii +> i;
To make further progress we need to introduce two restrictions on the structure of the person’s preferences that are common in the literature. First, define a person’s absolute risk aversion as the ratio -u,,(c, i; a>/u,(c, i; a), which will always be non-negative (Arrow 1970; Pratt 1964). Then we shall assume: Decreasing or constant absolute risk aversion. For any given values of risk aversion is a constant or decreasing function of disposable income.
i and a absolute
What this means in formal terms is that the ‘normalised curvature’ of the function u with respect to c is non-increasing for all values of c. The economic meaning of this is that the preferences represented by the function u have the property that as a person with given attributes in a given social context will never spend less of his resources on risky assets as he gets richer. Valuation of public output: For given tax-enforcement parameters r, and for any (y, a) type the marginal valuation of public goods in terms of the person’s own consumption depends only on the perceived share of publicly supplied output, g. This means that the personalised ‘price’ that a person pay for the goods and services provided by the government
is willing to is indepen-
F.A.CoweN / Taxrl~asion and inequity
dent of his private consumption (Cowell and Gordon from eq. (7) a simple result immediately follows. 2
Theorem. If people’s valuation of publicly supplied goods depends only on their personal attributes a, and their share in publicly supplied goods, then the evasion of a (y, a)-type person always increases with perceked share in publicly supplied goods. What this means is that the standard economic model of tax evasion would predict that an exogeneous increase in exchange inequity actually causes each person of any (y, a> type to want to evade less. The reason for this is the interaction between each person’s valuation of government output, and his attitude to risk. As the perceived share in the publicly supplied goods falls (inequity rises), so the person’s perceived ‘full income’ (his private disposable income and the value of the publicly supplied goods he receives) falls: this reduction in personal well-being makes him less willing to take risks. So evasion by members of all groups decreases. However, this result is not restricted to one particular narrow interpretation of inequity. First, notice that it can be extended to cases where both t and g change simultaneously: if both tax rates and perceived shares in public output are increased - for example in an expansion of total public sector output - then evasion activity of any (y, al-type at first rises and then falls (Cowell and Gordon 1988). Secondly, notice that the result does not make any special use of the specification (7) of inequity in the exchange relationship: in fact the result follows from an arbitrary exogenous change in perceived inequity. The corollary of this is quite interesting. Suppose that inequity actually came from a different source, such as the perception of injustice in the economic system. Then for example, inequity might be of the form: i = Q(F),
where F is again the distribution function of taxpayer types. This formulation encompasses situations where ‘inequity’ is interpreted as inequality of the distribution of pre-tax income or of disposable income: @ would then be a functional that satisfies the conventional properties of an inequality measure. Now consider a change in F that ’ The proof of this result
from the author
F.A. Cowell / Tax eL,asion and inequity
involves a widening of the income gaps between rich and poor, whether through increased opportunity for the wealthy to evade or some other route. This change in income distribution will affect everyone’s perceptions in the same direction, and again we will find that the increase in perceived inequity actually induces a reduction in evasion activity for all taxpayer types (Cowell 1990). These conclusions can be summarised formally as the proposition that evasion rises with inequity if and only if ‘inequity’ appears as a good in the individual’s utility function. This may appear to be counterintuitive. For instance, one might have expected that if ‘inequity’ incorporates concepts of rigidity of assessment (Striimpel 1969) or the reasonability of government demands, then evasion activity of individuals would go on rising as the government attempts to expand its share of total output, raising G and t simultaneously. Might one meet this objection with the suggestion of dropping the assumption that inequity enters negatively into the individual’s utility function? One could perhaps imagine that there are some types of person whose attributes a are such that they just happen to like inequity. However, this modification would not really get one very far since the direction of the response for an individual with given attributes a is independent of his income and his evasion activity, and would be unrelated, for example, to whether the particular individual loses by, or profits from the perceived inequity. The apparently curious performance of the expected-utility model extended to include a simplified form of the exchange relationship raises two issues: whether the model might usefully be extended to incorporate more sophisticated interaction between the individual and the community than is captured by the tiny impact of his evasion on public expenditure, and whether one might find that the conventional model would yield more compelling results were it to incorporate a more sophisticated form of ‘personalised’ inequity. I examine these in the next two sections.
5. Social conscience,
equity and the evasion threshold
The first of these two issues raises the possibility that some persons’ evasion behaviour is influenced by morals or other forms of social consensus. From questionnaire studies we know that social norms
F.A. &well / Tux er,asion and inequity
influence evasion behaviour (Song and Yarbrough 1973) and experimental evidence indicates that some people choose not to evade at all, apparently on moral grounds (see Baldry 1986, 1987). This remark takes us back to the objection often raised against the basic expected utility model, introduced in section 3: that the model implies that everyone - except for the infinitely risk averse - evades. Of course we could assume that there is a fixed group of ( y, a)-types who never evade tax under any circumstance: we would not be able to distinguish whether they happen to be honest or are just extremely averse to risk, but that does not matter. However, there is also evidence to suggest that the probability of audit has its principal impact on whether or not the person evades at all rather than on the specific amounts of income concealed (Spicer and Thomas 19821, so that the non-evaders appear to form a group whose choices should be susceptible of explanation within a behavioural model. This phenomenon could be explained in part by ‘technological’ influences such as the presence of substantial fixed information costs to evasion (finding an effective way to cover your tracks) or perceived transaction costs (actually changing your mode of work to increase evasion opportunity) each of which will induce a form of inertia that will inhibit an amoral tax-payer from engaging in evasion activity. However, there is evidence that social factors, including inequity, also appear to play a role here: for instance, the number of evaders personally known to a taxpayer is often cited as an important factor apparently predisposing people to evasion (see for example Porcano 1988; Schmijlders 1970; Spicer and Lundstedt 1976; Vogel 1974). The interpretation of this evidence is ambiguous: the reported awareness of evaders of other peoples’ tax evasion could be self-fulfilling prophecy (I assume that everyone else is doing what I am doing for other reasons), or it could be one of direct motivation (if he is doing it, then so should I>. However, in terms of the relationship between evasion and perceived inequity, this apparent ambiguity may not matter very much. Consider two roles for the perception of others’ evasion activity that correspond roughly to these two types of interpretation. First, tax evasion can be interpreted as defection in a game, the game of providing resources by common consent for the needs of government (Gaertner 1987; Cowell 1990: chapter 3): the extent of tax evasion by others indicates the extent to which the defection is unfairly taking
F.A. Cowell / Tax wasion and inequity
place. Alternatively one may take the view that there is a mechanism of perceived community sanction: alongside the legal penalties in the form of the surcharge s imposed by the tax authority there may also be the penalty of public shame or embarrassment that exposure as a tax evader would invoke. This social stigma imposes a threshold; whether the potential tax evader crosses this threshold is influenced by the interaction between a person’s evaluation of the consequences of his own decisions and how he thinks that such decisions would appear to others. Once again individual perceptions have a crucial role to play in the workings of the economic model: in this case it is the tax evader’s perception of other people’s perception of his own actions. This evaluation will in turn depend upon whether he and they are engaging in rule-breaking. Hence one should consider adapting the basic individualistic model of section 3 to take account of what is going on in the rest of the community. To see what this involves for the way in which perceived inequity affects evasion let us again use the notation developed for the basic model: so e*(y, a, r, i) represents the amount of evasion (concealed income) for a ( y, al-type chosen by an individual under uncertainty in the light of tax-enforcement parameters z- and perceived inequity; inequity is now represented as
where E is in turn determined by everyone else’s e” as given by (8), and proxies the perception of how other people are getting away with evasion, and E is a participation variable that takes the value 1 if the person is himself evading and 0 otherwise. The function 41,is non-decreasing in E and has the property that t,HE, 0) > JI(E, 1) for high values of E, and I,!J( E, 0) I J/(E, 1) for low values of E. The interaction between tax evaders’ actions in this model provides an account of the part played by inequity in tax compliance: if word gets around that others are defecting from compliance, then the increase in perceived inequity can lower the evasion threshold for other tax evaders, who then switch to participation. This process will typically generate epidemics of tax evasion (see Benjamini and Maital 1985; Cowell 1990). However, within the group of people who are already predisposed to evasion, the conclusions derived in section 4
remain: within the utility-maximisation model under conventional assumptions further increases in inequity will reduce evasion. 6. Individual
action and inequity
In the previous section we considered the view that the notion of ‘inequity’ in the basic model is overly narrow in that is too selfish; now we consider whether it is too narrow in that is too impersonal. There are two issues involved. Firstly, should the formulation of inequity be ‘personalised’ in that it be made conditional on each persons attributes? Secondly, if the individual’s own actions can directly and substantially affect perceived inequity within the economic model, would we then obtain a different story from that in section 4? The first issue can quickly be disposed of: the theorem in section 4 above and its corollaries hold irrespective of the (y, al-type to which a particular taxpayer belongs, so that modifying expression (6) or (10) to incorporate a dependence on personal attributes a will not alter the main conclusions. This is not to say that an individual’s characteristics or perceptions would be unimportant: they could strongly influence the magnitude of an individual’s behavioural response to the tax-enforcement parameters r, for example. However, if personal attributes a are incorporated into the model of inequity purely to allow the functional form of relationship (6) (or relationship (10)) to differ across taxpayers, then the qzza~~tat~L1~nature of the results linking inequity and evasion will remain unchanged. Further adaptation of the model is required in order to arrive at more interesting results. The second issue introduces a new component in the modelling of individual tax-compliance behaviour. The previous specifications of inequity have ruled out by assumption any perceived impact of the person’s own evasion upon the magnitude of i. By contrast consider the situation in which inequity is essentially a personalised problem related to the disparity, as perceived by any particular taxpayer, between his own effective economic position and the economic position of the group to which he aspires. To capture this idea we may imagine the situation in which, for each person of the ( y, a&class we may write i =jfy)
where L(y) denotes the mean of the population
(12) group consisting of
F.A. Cowell / Tax ecasion and inequity
those people who are observed to have an income greater than his own, and 8c is his own expected disposable income, which can be inferred from (1) and (2) above. The use of j(y) is a crude, but effective way of defining an appropriate reference group which may influence economic behaviour. The formal argument here is not restricted to a model with this specific way of representing inequity. The main result would still hold if j were replaced by the effective disposable income in the reference group, if (12) were to be replaced by a more complicated function of its two main components (possibly depending on a), or if 8c were to be replaced by [l - 01~’ + 0~” where c’, c” are the values of c corresponding to r in (2) and 0 < 0 < l/[l + s]: the model can be adapted to account for misperceptions of the probability of detection upon the evaluation of the expected gains from evasion and the evaluation of inequity. Now consider the way in which this modified model of personalised inequity responds to changes in the distribution of income. Clearly if j( y)/y were to rise everywhere income inequality would rise. It is evident that such an increase in income inequality would increase perceived inequity i. However, by contrast to the model discussed in section 4, if the person is not already concealing all his own the individual can do something about the ‘inequity’ that he experiences by taking action himself. Through increasing his own evasion activity, thus increasing expected disposable income, 8c, he will directly reduce inequity, irrespective of any further feedback effects upon other people’s evasion or upon publicly supplied goods. If we modify the basic model to incorporate (12) then, instead of the first-order condition (4) we will find, since %/de = -ZC(r>t that g(uJc,
i; a)r) -8(r)8(ui(c,
for anyone whose evasion is positive, but less than total income y. Now let us consider the impact of an exogenous change in the environment tht increases perceived inequity i. As we have seen, this can be modelled by an increase in j, so, differentiating (13) with respect to J and simplifying, we have:
= - t[8(u,,(c,
8(r)8(uij(c, c, i; a)r)+
i; a)) Z(r)2ZY(ui,(c,
i; a))] ’
F.A. Cowell / Tax evasion and inequity
The first term in the numerator of the right hand side of (14) is the same as before (see eq. (9) above), and under the usual assumptions it must be negative; but the second term is clearly positive because the strict concavity of cardinal utility requires that the term UJC, i; a> be negative (this condition also ensures that the denominator is positive). If the person’s sensitivity to inequity (measured by -uii(c, i; a) /u~(c, i; a)) is sufficiently large this second term will dominate. Hence, if the taxpayer can do something directly about his own feelings of inequity - as in this adaptation of the basic expected utility model - as well as feeling sufficiently sensitive to inequity, then greater inequality of income will induce greater evasion. The analysis shows clearly that one has to specify fairly carefully what one means by the inequity or injustice that is often cited as a motive for evading taxes. It is not enough just to incorporate into the individual’s utility function some undesirable entity that depends on the tax rates, the shape of the tax schedule or on income inequality. How the person perceives his own role in modifying that entity is of central importance. In the model examined in this section the person’s actions do indeed operate directly upon perceived inequity rather than through the indirect route of reducing the aggregate of publicly supplied goods, as in the simplified model of the exchange relationship.
7. Conclusions In an empirical study of the factors predisposing individuals to evasion Hite (1990) commented on ‘the importance of measuring the participant’s level of sensitivity to the targeted inequity and of measuring the relevant a priori variables that could influence the subject’s perceptions’. A complementary point might be made about the specification of inequity within economic models of tax evasion. The role of economic analysis must surely be to offer a framework within which social and psychological factors that may influence economic choices can coherently be modelled. In the case of tax evasion we have seen that there is an apparent puzzle. Within the conventional economic model, if inequity is perceived to increase then, irrespective of the exact specification - as some function of tax rates and shares in public expenditure, as a conventional inequality of
F.A. Coweli / Tax erwsion and inequity
reported personal incomes, or as the inequality of estimated ‘true’ incomes - evasion activity will actually move in the ‘wrong’ direction. In fact this result illustrates the operation of a general phenomenon, that of an ‘externality’ generated by others in the community upon an individual taxpayer’s optimal decision in the face of risk: If this externality is unfavourabie (inequity is a ‘bad thing’) then everybody feels worse off and takes less risk: they evade less. One of getting around this result is to invoke some special nonstandard assumptions about the structure of the utility function. However, a more fruitful approach is to re-examine the way in which the inequity is modelled in the light of survey and experimental evidence: in particular the role of the community, personal attributes and individual actions. Whether it is in the form of a participation decision (as in section 5) or in the modification of disposable income gaps that are perceived to be unfair (as in section 61, it appears that the crucial question that a model of the tax evasion decision must address is: Can the person affect inequity directly by his own actions?
References Arrow, K.J., 1970. Essays in the theory of risk-bearing. Amsterdam: North-Holland Baldry, J.C., 1986. Tax evasion is not a gamble. Economics Letters 22, 333-335. Baldry, J.C., 1987. Income tax evasion and the tax schedule: Some experimental results. Public Finance 42, 357-383. Becker, W., H-J. Biichner and S. Sleeking, 1987. The impact of public expenditures on tax evasion: An experimental approach. Journal of Public Economics 34, 243-252. Benjamini, Y. and S. Maital, 1985. ‘Optimal tax evasion and optimal tax evasion policy: Behavioral aspects’. In: W. Gaertner and A. Wenig (Eds.), The economics of the shadow economy (pp. 245-264). Berlin: Springer. Bordignon, M., 1991. A fairness approach to income tax evasion. Mimeo, Catholic University Largo Gemelli, Milan. Chang, O.H., D.R. Nichols and J.J. Schultz, 1987. Taxpayer attitudes toward tax audit risk. Journal of Economic Psychology 8, 299-309. Clotfelter, C.T., 1983. Tax evasion and tax rates. Review of Economics and Statistics 65, 363-373. Cowell, F.A., 1990. Cheating the government. Cambridge, MA: The MIT Press. CowelI, F.A., 1991. Tax-evasion experiments: An economist’s view’. In: P. Webley, H.S.J. Robben, H. Elffers and D.J. Hessing, Tax evasion: An experimental approach (pp. 123-127). Cambridge: Cambridge Universi~ Press. CowelI, F.A. and J. Gordon, 1988. Unwillingness to pay: Tax evasion and public good provision. Journal of Public Economics 36, 305-321. Elffers, E., R.H. Weigel and D.J. Messing, 1987. The consequences of different strategies for measuring tax evasion behaviour. Journal of Economic Psychology 8, 311-337.
F.A. Cowrll / Tux ausion and inequity
Falkinger, J., 1988. Tax evasion and equity: A theoretical analysis. Public Finance 43, 388-395. Friedland, N., S. Maital and A. Rutenberg, 1978. A simulation study of tax evasion. Journal of Public Economics 8, 107-l 16. Gaertner, W., 1987. ‘Untergrundwirtschaft, Steuerhinterziehung und Moral’. In: H. Hesse (Ed.), Wirtschaftswissenschaft und Ethik (pp. 109-130). Berlin: Duncker and Humblot. Gordon, J.P.F., 1989. Individual morality and reputation costs as deterrents to tax evasion. European Economic Review 33, 797-805. Groenland, E.A.G. and G.M. van Veldhoven, 1983. Tax evasion behaviour - A psychological framework. Journal of Economic Psychology 3, 129-144. Hessing, D.J. and H. Elffers, 1985. ‘Economic man or social man?: A social orientation model for individual behaviour in social dilemmas’. In: H. Brandstaetter and E. Kirchler (Eds.), Economic psychology (pp. 195-203). Linz: Trauner. Hessing, D.J., H. Elffers and R.M. Weigel, 1988. Exploring the limits of self-reports and reasoned action: An investigation of the psychology of tax evasion behavior. Journal of Personality and Social Psychology 54, 405-413. Hite, P.A.. 1987. An application of attribution theory in taxpayer noncompliance research. Public Finance 42, 105-l 18. Hite, P.A., 1990. An experimental investigation of the effect of tax shelters on tax-payer noncompliance. Public Finance 4.5, 90-108. Kaplan, SE. and P.M.J. Reekers, 1985. A study of tax evasion judgements. National Tax Journal 38, 97-102. Kaplan, SE., P.M.J. Reekers and K.D. Reynolds, 1986. An application of attribution and equity theories to tax evasion behavior. Journal of Economic Psychology 7, 461-476. Karzon, A.U., 1983. International tax evasion: Spawned in the United States and nurtured by secrecy havens. Vanderbilt Journal of Transnational Law 16, 757-832. Kinsey, K.A., 1984. Survey data on tax compliance: A compendium and a review. Working Paper 84-1, American Bar Foundation, Chicago. Mackscheidt, K., 1984. ‘Konsolidierung durch Erhohung von Steuern und Abgaben?’ In: H.H. von Arnim and K. Littmann fEds.1, Finanzpolitik im Umbruch: zur Konsolidierung Gffentlither Haushalte (pp. 145-161). Berlin: Duncker and Humblot. Mason, R. and L.D. Calvin, 1984. Public confidence and admitted tax evasion. National Tax Journal 37, 489-496. Porcano, T.M., 1984. Distributive justice and tax policy. The Accounting Review 59, 619-636. Porcano, T.M., 1988. Correlates of tax evasion. Journal of Economic Psychology 9, 47-67. Pratt, J.W., 1964. Risk-aversion in the small and in the large. Econometrica 32, 122-136. Schlicht, E., 1985. ‘The shadow economy and morals: A note’. In: W. Gaertner and A. Wenig fEds.1, The economics of the shadow economy (pp. 265-271). Berlin: Springer. Schmolders, G., 1970. ‘Der Beitrag der Schattenwirtschaft’. In: Wandlungen in Wirtschaft und Gesellschaft (pp. 371-379). Tubingen: J.C.B. Mohr. Schwartz, R.D. and S. Orleans, 1967. On legal sanctions. Chicago Law Review 34, 274-300. Sheffrin, SM. and R.K. Triest, 1990. Can brute deterrence backfire? Perceptions and attitudes in taxpayer compliance. Mimeo, University of California-Davis. Song, Y. and T.E. Yarbrough, 1973. Tax ethics and taxpayer attitude: A survey. Public Administration Review 38, 4422452. Spicer, M.W., 1975. New approaches to the problem of tax evasion. British Tax Review, 152- 154. Spicer, M.W. and L.A. Becker, 1980. Fiscal inequity and tax evasion - An experimental approach. National Tax Journal 33, 171-175. Spicer, M.W. and R.E. Hero, 1985. Tax evasion and heuristics. Journal of Public Economics 26, 263-267.
F.A. Cowell / Tax evasion and inequity
Spicer, M.W. and S.B. Lundstedt, 1976. Understanding tax evasion. Public Finance 31, 2955305. Spicer, M.W. and J.E. Thomas, 1982. Audit probabilities and the tax evasion decision: An experimental approach. Journal of Economic Psychology 2, 241-245. Striimpel, B., 1969. ‘The contribution of survey research to public finance’. In: A.T. Peacock, Quantitative analysis in public finance (pp. 13-22). New York: Praeger. Thibaut, J., N. Friedland and L. Walker, 1974. Compliance with rules: Some social determinants. Journal of Personality and Social Psychology 30, 792-801. Thurman, Q., 1988. Tax-payer noncompliance and attribution theory: An experimental vignette approach. Public Finance 43, 147-156. Vogel, J., 1974. Taxation and public opinion in Sweden: An interpretation of recent survey data. National Tax Journal 27, 4999513. Wallschutzky, LG., 1984. Possible causes of tax evasion. Journal of Economic Psychology 5, 371-384. Wallschutzky, I.G., 1988. The effects of tax reform on tax evasion. Sydney: Australian Tax Research Foundation. Webley, P., 1987. Audit probabilities and tax evasion in a business simulation. Economics Letters 25, 267-270. Webley, P. and S. Halstead, 1986. Tax evasion on the micro: Significant simulations or expedient experiments? The Journal of Interdisciplinary Economics 1, 87-100. Webley, P., 1. Morris and F. Amstutz, 1985. Tax evasion during a small business simulation. Economic Psychology, Proceedings of the Tenth IAREP Annual Colloquium, Linz. Webley, P., H.S.J. Robben and I. Morris, 1988. Social comparison, attitudes and tax evasion in a shop simulation. Social Behaviour 3, 219-228. Webley, P., H. Robben, H. Elffers and K. Hessing, 1991. Tax evasion: An experimental approach. Cambridge: Cambridge University Press. Witte, A.D. and D.F. Woodbury, 1985. The effect of tax laws and tax administration on tax compliance: The case of the US individual income tax. National Tax Journal 38, l-13. Yankelovich, Skelly and White, Inc., 1984. Survey of tax-payer attitudes. Washington, DC: IRS.