1. MACROWELFARISM?

The present state of scholarly studies about such a major issue as what income distribution should be is nothing short of a scandal of the first magnitude. On the one hand, a large number of philosophers go on discussing this issue along the lines drawn by John Rawls – reputed to be the best known or most important philosopher of the last century. On the other hand, the field called “optimum income taxation” goes on with the ethical assumption the indictment of which is precisely the motive and reason for Rawls's theory, as if Rawls had never existed.

This assumption is macrowelfarism, i.e. the application, to the issue of macrojustice, of “welfarism”, that is, taking welfare understood as happiness, satisfaction and the like as the ultimate moral reference notably for choosing distribution between persons, applied here as the maximization of a social welfare function which is an increasing function of individual utilities (the term welfarism is due to Hicks (Reference Hicks1959) in his criticism of such an ethical theory for similar macro applications). These utility functions aim at describing the noted welfare of the individuals rather than their judgments about society, i.e. in Pareto's (Reference Pareto1913) terms, their ophelimity rather than their utility. This model may be a roughly approximate description of a number of choices in issues of micro or mesojustice, notably when utility means lower suffering (e.g. in medical choices or “welfare” policies), for a number of choices in groups of people who sufficiently know one another (which permits empathy), or for the allocation of specific goods.Footnote ²

The macrowelfarist taxation literature treats problematically the two most revered scholars in the field, John Rawls (Reference Rawls1971) and Jim Mirrlees (Reference Mirrlees1971) – its very inspirer. As we shall see, both hold in fact identical ethical views (discard tastes and collectivize abilities), the difference being one in formulation.

This literature gets rid of Rawls in two strokes. First, it occasionally but rightfully criticizes Rawls's imprecision and the lack of Pareto efficiency of his “difference principle” (omitting, however, the possibility opened by the addition of 1974, as we shall see). Second, it hijacks Rawls's name by calling “Rawlsian” a maximin in utilities, which triggered Rawls's (Reference Rawls, Sen and Williams1982) uncharacteristically severe reaction that this is “a complete misunderstanding from the philosophical standpoint”.Footnote ³ Several issues of consistency make the macrowelfarist tax theory quite intriguing.

1. 1. This theory purports to base the choice of the best income tax on individuals' welfares. However, most of the time (this excludes Mirrlees (Reference Mirrlees, Arrow and Intriligator1986) and very few later papers) it considers individuals who have the same utility function (giving satisfaction as function of income or consumption and labour). But a nation's individuals are not clones with the same utility function. They are de facto sui generis, specific, different human persons with their own particular tastes. Hence, this function does not represent individuals' welfares (except, perhaps, for one of them). Therefore, this theory is actually not welfarist. In particular, the result is not Pareto efficient. Nor can this function describe individuals' own choices, a central assumption of the theory. This utility function determines the resulting income-tax schedule – along with the aggregation function which is not determined either. What is this function, what does it mean, how is it chosen? In fact, Mirrlees (Reference Mirrlees1971) does provide a reason for this model: the individuals' utility functions are the same because “differences in tastes raise different kinds of problems”. Then, however, the criterion does not refer to the actual “psychological well-being”, satisfaction, pleasure or happiness of individuals, which depend on their tastes and preferences. And all the noted consequences and questions remain. Yet, since individuals actually have different tastes, and differences in tastes influence the result (the tax) – as shown by Mirrlees (Reference Mirrlees, Arrow and Intriligator1986) if needed – discarding differences in tastes can only be done by discarding tastes, for the distribution in question. Hence the consistent implementation of Mirrlees's perceptive reason is simply to discard utility functions altogether, that is, Rawls's solution.Footnote ⁴

2. 2. As far as one can see, macrowelfarist scholars do not believe in their own scholarly moral ideal. All those I have asked have answered “no” to the following questions. Should someone with a joyful character pay a higher income tax as a form of compensation? Or because she accepts more easily the loss of income? Or because she enjoys her remaining euros more than others would (and one is egalitarian in utility)? Or should she pay a lower income tax because she would have enjoyed more spending the euros taken away (a higher cost in utility terms)? Or because she enjoys more her remaining euros (and one is utilitarian)? Should the European Union tax the industrious Swedes to subsidize the cheerful Neapolitans who presumably know better how to enjoy a euro's worth of consumption? Are these considerations relevant for the income tax? There seems to be an interesting split of personality between the scholar and the sincere citizen here.

3. 3. Macrowelfarist tax scholars often justify welfarism by their belief that Pareto efficiency implies it. However, the best-known theorem of economics is probably Pareto's or Debreu's which says that a perfect market from given resources provides a Pareto-efficient outcome. Hence, Pareto efficiency is secured by a resourcist policy distributing given resources and evaluating directly this action (from some consideration of equality, for instance), with a market implementing freedom of exchange and corrected for its “failures” if necessary. A number of social welfare functions, increasing functions of individual utilities, have their maximum at the resulting allocation. However, characterizing them can only be done by this consequence, hence tautologically. Pareto efficiency does not imply that the proper distribution is defined by comparisons of levels or variations of individuals' welfare, which is the characteristic of a welfarist ethic defined by a social welfare function.

4. One may add that macrowelfarist tax scholars speak of Pareto efficiency with individual welfares (ophelimity), not with “individual values”, as Kenneth Arrow puts it, or individual “utilities” in Pareto's sense, that is, the individuals' higher overall judgments of society. However, this latter conception is the relevant one if Pareto efficiency is justified by the fact that no other possible state is unanimously preferred (with the possible indifference of some people), hence in particular by democracy. As we will see, this conception leads, rather, to ELIE optimum taxation.

5. 4. The macrowelfarist doxa says that earned income is the tax base because individuals' abilities are not known, but Mirrlees (Reference Mirrlees1971) proposes that one can add labour duration, which provides the wage rate (the value of abilities), and remarks that we have a number of other possible sources of information about abilities. Moreover, 30% of the income tax base evades the tax in all countries (see for instance Slemrod Reference Slemrod2002, for the USA): this is a far cry from perfect information. Mirrlees is often right about knowing duration, and wage rates are also often directly observable or more or less inferred from occupation as a standard in a given labour market (in fact, 9/10 of labour is wage labour with an official pay sheet indicating wage rate and type of labour, in developed countries).

The first two points make it hardly surprising, therefore, that this income tax theory, which has been well-known during the last 38 years, does not seem to have received the beginning of an application yet (contrary to both the noted analogues in public utility pricing and important aspects of the ELIE scheme, as pointed out shortly). If Thomas Kuhn is right, however, it will remain in the academic world for a long time, because it is the material of heavy investments in courses, textbooks, doctorates, respectable hard-won reputations and the ensuing positions. We will still have editors of some major journals who will not even consider a nonwelfarist paper because they “do not know of a non-utilitarian possible referee”, and an official list of fields of economics that does not include normative economics (it has welfare economics which is only a subfield of normative economics).

2. MACROJUSTICE

On almost all the points of the proposal of Macrojustice, Erik Schokkaert's summary is better than the one I would have been able to provide. Hence, I will not mention these issues. On a very small number of points only, my formulation would differ – most of the time only slightly. A number of issues of MJ are further developed in the study Economic Macrojustice (Kolm Reference Kolm, Gamel and Lubrano2009, henceforth denoted as EM) to which the interested reader may want to refer.

Endogenous social choice is inescapable for properties that are unanimously wanted (scholars belong to society, and, at any rate, if one had another view, he would have to have an army and make a coup to impose it on the other people, including officials and former voters). This principle plus a few facts seem to lead to the ELIE structure with a rather fair degree of necessity. They provide, indeed, the following points.

• - The ethical distinction of the criterion of macrojustice from those of microjustice or mesojustice.

• - The predominance of labour income, especially in the intertemporal view in which capital is produced.

• - Nonwelfarism for macrojustice (see more in section 2.6 of EM).

• - The resulting issue of liberty as the material of macrojustice (more on this below).

• - The logical necessity of some kind of equality as the ideal structure of justice.

• - Social liberty, possibly necessarily desired, and its structure of non-rivalry.

• - Three proposals for defining equal total liberty, which give the same results (see a better presentation in section 4 of EM). (1) Social liberty (hence free exchange) from an equal allocation of income and leisure. (2) A condition that amounts to Maniquet's (Reference Maniquet1998) proposal that the budget lines in an income-leisure (or labour) space have a common point. (3) Equal liberty for non-identical domains of choice (since identical domains cannot respect social liberty or Pareto efficiency when individual productivities differ, without referring to individuals' preferences).

These properties imply the ELIE structure. For the coefficient of equalization or redistribution k (the equalization labour), which completes the determination of the transfers, MJ provides no figure because it depends on the society (on its degree of community and solidarity). Only methods for helping the determination of this coefficient are provided, in part 4 (with remarks about the levels that would decrease inequality as present-day national redistributions do, one to two days per week).

Moreover, judging a principle implies evaluating all its aspects, properties, meanings and consequences, and hence the 20 or so different equivalent important meanings of ELIE have to be considered (EM, section 5).

However, what arouses the interest of economists most is the possibility of basing the tax on wage rates. Fortunately, this is no longer an issue for speculation. We just have to observe, since it is applied in a whole country. A number of economists go on saying “you can't tax wage rates”. Perhaps one can't, but the French tax administration just does it, in the form of the tax exemption of overtime labour over a rather low official benchmark (35 hours a week, 218 days per year for executives whose daily hours are uncertain, and a specifically adjusted lower benchmark for people who choose to work part-time). Seeing this, the same economists propose, as a Pavlovian reflex, that firms will agree with employees to cheat by declaring as overtime some income that is not. The good or bad news is that this does not happen (except, perhaps, for very small firms), simply because falsifying the whole system of pay sheets is too complicated and risky, not possible without the tax administration being aware of it or informed about it.Footnote ⁵ Since the tax schedule also has a tax credit (and wage labour provides 9/10 of labour income), the whole tax structure makes large steps towards an ELIE structure.Footnote ⁶

When income distribution is optimal, public expenditures should be financed according to the principle which is prima facie distributionally neutral, namely benefit taxation. The necessary information is roughly that produced by the cost-benefit analysis used in well-managed, clear and justified public choices. This favours the optimization of public finance.Footnote ⁷ It also permits its efficient decentralization in various possible forms. Practical approximations are classical. If some expenditures are not dealt with in this way, one may agree that their financing follows some other of the classical principles of public finance. Paying “according to capacities” means, for earned income, according to capacities to earn, i.e. the wage rate. Paying “equally”, if it is not equal sharing, may be equally in labour. Then, both principles are the same and are the way ELIE finances the basic income kw (by taxes kw_i). Moreover, important branches of public expenditures implement issues in the fields of microjustice or mesojustice, and some may have to palliate imperfections in the realization of macrojustice (Chapter 25 of MJ).

The few eccentric (and likeable) very productive people who freely choose to work very little raise an issue of microjustice – yet their case usually arouses the interest of scholars more than the fate of the toiling masses seems to do. Since, by dropping out of collective production, they give up the advantage of a high wage offered to them by society, they are a priori not submitted to the corresponding tax liability. There are several possible solutions to this problem. In the French tax system, they can choose a part-time labour contract. But the simplest rule is: let them earn their sandwich, which requires little work (Rawls's view, for instance). Yet Philippe van Parijs eloquently argues that their “right to laziness” (Paul Lafargue) or to meditation should be subsidized by the general basic income. And if a person has a rare skill that is very useful to others, one should try to persuade her to somewhat help others with her (undeserved) gift of nature – with a high wage rate and for a little time. If this skill can save lives (medical research, defence, etc.), then draft – for a limited time and a high wage – may not be illegitimate. However, E. Schokkaert and E. Ooghe have made an important in-depth analysis of the case and effects of labours of low duration in general, and I certainly endorse their conclusions.

The dispersion of people's views about the minimum income, their evolution and their relation to the homogeneity of society is presented in previous works referred to in MJ. Moreover, this level kw and the degree of equalization k are mutually determined (given the average wage rate w) and part 4 of MJ deals essentially with the variety of people's opinions in this respect, the derivation of the optimum level from these views, their causes and their evolution (including the theory of their convergence in a theory of fair debate).

For people with low-paid skills, MJ shows that, de facto, ELIE can replace all aids while making everybody better off. For the other end of the spectrum of incomes, the tax treatment of capital or capital income raises the question, present for all tax reforms, of the treatment of wealth accumulated under previous distributive regimes.

I think that bequest is obviously an important distributive issue. Simply, the theory of optimum inheritance taxation is not in MJ. Considerations about it appear in other works (Kolm Reference Kolm1985, Reference Kolm, Kolm and Mercier Ythier2006). This theory can have similarity with ELIE and be related to it. Bequest is a gift of a very particular type (it is not a fully voluntary transfer since it is triggered by death), but, more generally, the justice and taxation aspects of gift giving can make good use of further analysis – as noted by E. Schokkaert – especially since this includes family influence on capacities and education.

I am also very careful, in theoretical analyses, to speak of the relevant wage rate w_i as, specifically, the value of given capacities. In particular, the part of training and formation, initial or later in life, that the person can be held responsible for, is a dimension of her labour.

As I have noted, on all points not mentioned here, I entirely endorse E. Schokkaert's perceptive and important views (let me just note that the size of the volume precluded the inclusion of further developments).

3. PARADIGMS OF JUSTICE

However, the right orientation of specific research needs to rest on a clear chart of the overall issue. The general question is very simple. The problem is macrojustice. The most important resource to be allocated is the human resource, by very far (particularly in an intertemporal view, in which capital is produced). A given capacity of an individual can be either put in the common pool for overall redistribution by transfers or compensation, or self-owned, allocated a priori to this individual who is entitled to its value and accountable for its weaknesses. Individual capacities divide into productive ones and eudemonistic ones (represented by utility functions), as a very sufficient approximation. The two limiting cases are classical liberalism for which all capacities are self-owned, and welfarism for which none is. Classical liberalism is the basic social ethics of modernity endorsed for instance by John Locke, Friedrich Hayek, Milton Friedman or Robert Nozick, or its “leftist” variant which advocates an equal sharing of non-human natural resources (the so-called “libertarianism” of Murray Rothbard or David Friedman, or the view of James Buchanan, are something else, not based on a priori rights). A third, intermediate, position is that of Rawls: eudemonistic capacities and utility functions are self-owned, whereas none of the productive capacities are (the former determine “goodness” and the allocation of the latter concerns fairness or justice). These are the three polar paradigms of distributive justice. Welfare and income are the material of the judgment of justice in the first and last case, respectively, but this can be with various criteria of distribution having various possible types and degrees of inequality aversion.Footnote ⁸

In this menu, what do people choose? It depends on the society, in particular on its size. Welfare, justice as fairness, and liberal freedom are the standard criteria within closely knit small societies, for nation-wide solidarities, and beyond this, respectively. Welfare is adopted when there is much altruism, in societies of people who know one another well with mutual empathy (it is common for intra-family distribution, for instance), but also when utility means lower suffering. At the other end of the spectrum, there are very few disinterested transfers at the level of the global economy. Rawls opposes transfers across “peoples” (say nations); he endorses “trade, not aid” in this respect. Yet our present topic is the third, intermediate level, national redistributions in societies in normal situations (not in the case of an overall disaster in which suffering is the common fate).

There, people do not appear to use or endorse welfarist judgments for macrojustice and the income tax (specific causes of suffering are the object of particular insurance and microjustice policies). However, they do not generally support any of the two alternative polar cases either. They commonly see some injustice in the very wide discrepancies between individual earning powers due to nature, family background and society's demand. But not fully so, in general. Legitimate self-ownership is often seen not to stop at eudemonistic capacities. People are often seen to have some entitlement to this “brute luck” of theirs. This seems to be the main reason that prevents more redistributive fiscal policies: still higher high taxes are often seen to be unjustly confiscatory of the products of individuals' efforts and capacities. Limiting disincentive effects is only added to this reason and is certainly not actually the only one (it is also clearly often a pretext for this policy). Similarly (and relatedly) almost nobody wants an inheritance tax of 100% and a full ban of bequest. In most countries, someone who finds a treasure is entitled to keep some part of it for herself, and this is commonly admitted. No one has proposed, it seems, that handsome people compensate those born with a less attractive face. And yet it is commonly acknowledged, in genuine communities of all kinds, that people endowed with high earning capacities by their genes, family influence or society's demand for skills should provide some solidaristic aid to people less fortunate in this respect. Opinion takes the choice to be between lower income inequality on the one hand, and lower redistribution in order to respect the liberal freedom of self-ownership more on the other. Note that both corresponding polar ideals of income equality and self-ownership ban welfarism (the former reveals that the corresponding end-value of justice is income, not welfare, because individual utility functions differ). There is, in the end, a demand for some redistribution, which can, however, be a priori at various possible levels.

4. THE SURPRISING RECENT HISTORY OF MACROJUSTICE

The extraordinary situation of present-day scholarship in the field of optimum income distribution is the result of a surprisingly confused history. As we have seen, the paradigm shift that Erik Schokkaert sees in Macrojustice is from the theory of optimum income taxation that dominates the institutions of economics and derives from Jim Mirrlees's article of Reference Maniquet1971, but this theory, which generally purports to be welfarist but in fact is not, is morally ambiguous. What is the difference, then? It could be: (1) deep, in the principles, ethical; or (2) factual, in the constraints, practical; or again (3) in the formulation of the problem.

The analysis can be based on three basic statements found in Mirrlees's article, which express, respectively, his views about the ethical role of individuals' utilities and earning capacities, and about information, for optimum income taxation:

1. 1. Mirrlees's First Moral Statement:

2. “Differences in tastes. . . raise rather different kinds of problems.”

3. 2. Mirrlees's Second Moral Statement:

   “The great desirability of . . . offsetting the unmerited favours that some of us receive from our genes and family advantages.”

4. 3. Mirrlees's Practical Statement:

   “[We] could . . . introduce a tax schedule that depends upon time worked as well as upon labour-income: with such a schedule, one can obtain the full optimum . . . We also have other means of estimating a man's skill-level.”

Labour income divided by time worked gives the wage rate, the value of skills. Hence, as we have seen, Mirrlees's Practical Statement is so true that it is simply applied at a country-wide level by an income tax based on the wage rate, in the form of an exemption of overtime labour above a rather low benchmark, without any of the impossibility or fraud that is predicted by a standard prejudice of the sub-culture of economists.

Note that Mirrlees's two moral statements constitute precisely Rawls's ethics, and that the bulk of his paper, before he presents his Practical Statement, is about the disincentive effect of a tax based on full earned income, which explains the incentive effects of inequality so central in Rawls's conception. This analogy is supposed to be surprising, since Rawls's aim is to criticize, and to provide a substitute to, the ethics that Mirrlees is commonly presumed to model.

In particular, discarding tastes as irrelevant for optimum income distribution is exactly Rawls's opinion. He notes (1982), for instance, the classical example of “expensive tastes”: you do not have to finance some other person's beverage because she only likes expensive wines. He could have added cheap tastes as well: you probably do not have to finance another person's beverage because she likes cheap beer and this produces utility at low cost, as demanded by utilitarianism.

Rawls concludes that individual utilities have to be dropped from the optimality principle of “social justice” (i.e. macrojustice, he says “macro” and “not micro”), and that, therefore, the material of this principle consists in individuals' means of free choice and action: basic liberties and “primary goods”. There is one economic primary good, income (or wealth). Rawls's prima facie ideal is an equality in primary goods. However, an equal redistribution of earned income in a large society with free labour would discourage anybody from working for an income. Rawls settles, therefore, for a maximin in primary goods, his “difference principle”.

However, Mirrlees's Practical Statement, and facts, suggest that Rawls is excessively incentive-pessimistic. Public transfers do not discourage labour if they are based not on the labour of the individuals but on their wage rate (earned income is labour multiplied by wage rate). The theory of such transfers (along with the proposal of exempting overtime labour) and the comparison of the effects of taking various bases, are presented in Kolm (Reference Kolm1974).

In 1974, also, a cultured economist who left us recently, Richard Musgrave, explained to Rawls the disincentive effect of people substituting untaxed leisure for taxed income and proposed, as a solution, to introduce leisure as another primary good. Rawls (Reference Rawls1974) accepted. Therefore, we now have two economic primary goods, income and leisure. The equality can be in each of these two goods. Budget balance requires that this is done by demanding from each individual the proceeds of the same labour (the complement to the equal leisure), with her specific productivity, and by equally sharing the total amount. This is Equal-Labour Income Equalization or ELIE. However, the basic right of free exchange, which enables people to freely buy consumption goods with their income, should also enable them to freely exchange labour for income from this basic equal allocation of income and leisure or labour, that is, to freely work and earn given the financial transfers of the ELIE scheme. This outcome is Pareto efficient if markets (possibly corrected for any “failure”) are, when labour is defined by all its dimensions chosen by the individual (duration, intensity, formation, etc.), because the tax or subsidy is not based on elastic items – the tax refers to a given labour. This extra part of individuals' incomes is not equalized, however, but it can a priori be smaller or larger, by choosing the equalized income and its labour base larger or smaller, as just discussed. The solution becomes classical liberalism when this “equalization labour” vanishes, and goes towards the Mirrlees–Rawls ethics when it becomes larger.

Nevertheless, in both Rawls's view and Mirrlees's model, individuals with more productive capacities tend to have a higher income. This is not for a reason of partial self-ownership, however, but only for the practical reason of limiting disincentives. Rawls does not explain this disincentive. Mirrlees does, and this result is due to the fact that the tax in his model is based on the full earned income, and therefore does not take advantage of the possibilities highlighted by this perceptive author himself. Finally, neither the level nor the structure of this correlation between productivity and income are those that would result from partial self-ownership with the optimum tax base.

The difference between Rawls's and Mirrlees's (Reference Rawls1971) moral views concerning the relevance of tastes for macrojustice seems to be tiny, but it appears to create the great divide in the theories of optimum distribution and taxation. For Rawls, tastes are irrelevant, whereas, for Mirrlees, “differences in tastes” are irrelevant. Mirrlees interprets his view by assuming that all individuals have the same utility function. This may seem natural, and any such function can do the job. His numerous followers almost always endorse this assumption. As we have seen, however, this raises diriment problems, but they are in fact fictitious since Mirrlees's reason necessarily implies discarding tastes and utilities altogether for this issue, that is, Rawls's solution.Footnote ⁹

Mirrlees ended choosing the opposite way, however, and, discarding his initial view, he presented, fifteen years later (Reference Mirrlees, Arrow and Intriligator1986), the theory with different individual utility functions (yet with a particular formulation of differences in tastes and analogy in utility functions for convenience). Then, individuals' differences in tastes become relevant. The tax authorities have to know all individual utilities, with interpersonally comparable intensities, an intractable problem of information (and meaning) compared to which knowing wage rates seems a mere trivia. The general result is that your marginal tax rate should depend, in some unspecified way, on comparisons of the pleasures that all people derive from each good. This cannot be known and, in fact, hardly makes sense. Fortunately, however, these “relative tastes” do not seem to be actually considered relevant for defining the just income tax. Nonetheless, this optimization model is particularly praiseworthy for representing particularly optimal issues, either distribution in a society of optimum individuals altruistic and caring much about others' welfare, or the cases in which the proper income distribution is particularly needed for the relief of widespread suffering. However, actual thoughtful common opinion, which may applaud the reference to individual welfare for allocating in a family or a hospital, would have to be politically bypassed if it is to be applied to standard national taxation. This would be vox dei but not vox populi.