In academic research, various proposals, related to basic income or negative income taxation, are put forward to remedy such inactivity traps. Obviously, other approaches to the incentive problem for low-wage earners are possible, such as (a) topping up low-skilled workers’ purchasing power by selective tax credits, or (b) increasing their net pay by lowering personal social security contributions on low earnings, or (c) supporting sufficiently high minimum wages for low-skilled workers by selectively subsidising employers. These alternative instruments reflect not only technical differences, but also more fundamental differences in approach. Therefore, it is useful first to assess alternative instruments from a normative vantage point, that is, by examining the conceptions of distributive justice underpinning their use, before assessing them with more explicit reference to the particular problems created by tax and benefit systems in economies plagued with involuntary unemployment.

In this paper we focus on this normative issue. In section 1, we start with a broader question, namely why the problem of inactivity traps has gained heightened attention in debates on social policy over the last years. We present the idea of an active welfare state, which adopts increased participation as a central goal of social policy. In our view, this idea should also be based upon a responsibility-sensitive egalitarian conception of social justice. The active welfare state and responsibility-sensitive egalitarianism will act as the general background for our discussion of the relationship between a government’s normative conceptions and its eventual choice of instruments.

In section 2, we explain why defining a government as responsibility-sensitive egalitarian does not suffice to determine which distributive instruments it is supposed to choose. Making specific choices depends on the collective decision rule which is used and on normative positions taken by the government that do not reduce to responsibility-sensitive egalitarianism. We look at each of these in turn. In section 3 we present and discuss a collective decision rule that is based on the work of John Roemer. We proceed in section 4 by discussing the impact of a government’s normative stance with regard to its conception of personal responsibility. In section 5 and the remainder of the paper, we emphasise the impact of the government’s conception of individual well-being on the eventual outcome of using this decision rule.

While we keep the level of formal analysis to a strict minimum, we point out that a formal model can be constructed allowing for a more general analysis of the issue discussed in sections 4 and 5 (Vandenbroucke, 1999a, chapter 3). In sections 6 and 7, we will refer to some of that model’s results insofar as they indicate how optimal policy design, with regard to the use of instruments such as labour subsidies or a basic income, is influenced by the government’s conception of individual well-being.

The primary aim of such a model is to make clear how differences in a government’s normative conception lead to differences in policy choices. Such a highly simplified representation of reality is valuable, if only because it shows that, other things (e.g. the economic environment) being equal, the way a government publicly values an individual’s well-being has a profound impact on the choice of instruments. In section 8 we briefly discuss whether one can go further and identify criteria for choosing between the different optimal policies as established by this model. Yet we also point out that such a discussion may overlook some essential elements in current debates on social policy. This point is developed more fully in section 9, where we return to the idea of the active welfare state and indicate in which respect the model falls short of capturing some of the active welfare state’s essential constitutive elements. We provide a short summary and conclusion in section 10.

Combating inactivity traps is not the only —indeed, not even the most important— hallmark of a renewed social policy. To safeguard the welfare state, one must alter its working methods, for example by gearing social policy intervention more towards preventive actions, by complementing social spending with social investment in human and social capital, etc. This new kind of welfare state can be characterised as an active welfare state (Vandenbroucke, 1999b), to express both the change of methods ("an intelligently active state") and the change of goals ("a state of active people", i.e. a state combining social protection with citizens’ active participation in society).

Arguably, the potential sustainability problem of the current welfare state has served as an impetus to the debate on the re-orientation of social policy. However, this problem in itself cannot be taken as an ethical underpinning of the plea for increased participation. The fundamental argument should start from the recognition that participating in society can be regarded as an individual ‘functioning’, in the terminology of Amartya Sen. If one advocates genuine equality of opportunity, active participation in society should be included in any individual’s option set.

Cohen states that pursuing equality of opportunity is tantamount to removing "obstacles to opportunity from which some people suffer and others don’t." (1999, p. 354). He distinguishes socialist equality of opportunity from right-liberal and left-liberal alternatives in that "it seeks to correct for all unchosen disadvantages, disadvantages, that is, for which the agent cannot herself reasonably be held responsible, whether they be disadvantages that reflect social misfortune or disadvantages that reflect natural misfortune." We believe this maxim should apply to a social-democratic, that is, an egalitarian active welfare state. Hence, the egalitarian active welfare state, so conceived, is bound to address questions of individual responsibility. Thinking about distributive justice on egalitarian lines compels us to distinguish personal ‘luck’ from personal ‘choice’.

It is not really surprising that the notion of individual responsibility emerged on the social agenda at the same time as the call for a new kind of activating social policy. Some of the new social risks are more predictable than their traditional counterparts, and more easily linked with personal behaviour. For example, being long-term unemployed is closely associated with having a lower level of educational achievement. Achievement in education obviously depends on circumstances, such as family background and innate talents, but also on personal effort and choice. Hence, socio-economic developments in themselves already explain why questions regarding one’s personal responsibility are showing up more than they used to do previously in discussions about welfare. But, as we have just indicated, these questions also arise if one limits the discussion —as we do in this paper— to more abstract thinking about normative issues. In this paper we will confine our discussion to the case where government is committed to the idea of responsibility-sensitive egalitarianism.

When phrased in such a general manner, this responsibility-sensitive conception of equality provides little guidance to a government in the actual design of its distributive policies. More specific substantive questions have to be answered if one wants to understand why and to what extent a responsibility-sensitive egalitarian government would actually prefer particular instruments to achieve particular goals. What is the government’s actual conception of personal responsibility, that is, for which individual characteristics does the government take the citizens to be personally responsible? What is the government’s conception of the individual well-being of citizens, that is, what constitutes an advantage according to this government? In addition, there is an operational question to be answered: how will an egalitarian government actually choose between alternative policies? These questions are examined in the following sections. We shall handle the operational question first.

Admittedly, our broad phrasing of what responsibility-sensitive egalitarianism amounts to, does not refer to equal option sets, but rather to equal access to advantage, where ‘advantage’ refers to the government’s evaluation of an individual’s well-being (whereby we take it that well-being can be measured in an interpersonally comparable way). To explain the relationship between the general idea of "equal option sets" and "equal access to advantage", we propose to focus immediately on the issue of participation in the labour market. A group of people has identical option sets in the labour market when everybody in that group has access to everybody else’s job, for the same salary, in similar circumstances, etc. Calling this situation ‘equal access to all jobs’, one thus assumes that it holds if the only reason why individual A has a different job to individual B is because she genuinely chooses to have a different job. There are a variety of reasons why this situation is not the same as ‘equal access to advantage’. For example, people’s well-being in distinct jobs may be evaluated by government in the same way, so that two individuals may have equal access to advantage even when their option sets are not identical. Conversely, different people may not be able to convert identical jobs into the same level of personal advantage. Nevertheless, if the government values the advantage of a job independently of the person who holds the job, then ‘equal access to all jobs’ is a sufficient condition for ‘equal access to advantage’. Throughout we will assume that the government does indeed value advantage in such a way. We will use in our numerical examples the (highly unreal) situation of equal access to all jobs (hence, equality) as a useful counterfactual, serving as a benchmark for judgements on justice.

Consider then a situation in which only two jobs are on offer, fully described by the following data:

job F: full-time, salary 10;
job P: part-time, salary 5;
zero option Z: not working, income 0.

Call this option set I. Consider then option set II:

job F: full-time, salary 9;
job P: part-time, salary 6;
zero option Z: not working, income 3.

Let us further assume that alternative policies can give all people access to all the options in option set I (by using a policy I), or to all the options in option set II (by choosing a policy II). Hence, both policies admit equality as a result. Note that, while both admit equality, they are clearly different. For example, policy II is naturally associated with the establishment of an unconditional basic income, whereas policy I rules it out. We now want to indicate how a responsibility-sensitive egalitarian government would actually decide which option set is best for the collectivity.

It should be noted that this "natural way to proceed" builds into the conceptual framework of responsibility-sensitive egalitarianism an additional idea. That is, when making choices between policies, one should be impartial between the individuals with whom one is concerned, and simply aggregate the advantage those individuals choose to obtain. This feature captures an essential aspect of a methodology developed by John Roemer (1994, 1996a, p. 279, 1996b) to implement responsibility-sensitive egalitarian justice.

Let us provide a more detailed account of this implementation procedure. Consider a world in which individuals differ from each other with regard to their economic productivity and with regard to their preferences over income and time for non-market activity (i.e. their propensity to work). Accordingly, one can subdivide the population into ‘productivity types’ (people having a certain level of productivity in common) and ‘preference tranches’ (people having in common a certain preference ordering). Let us represent citizens by indexed capital letters, the letter indicating a preference tranche and the subscript indicating a productivity level. So, A₁ is a person belonging to preference tranche A and productivity type 1. Productivity type 1 comprises a range of citizens {A₁, B₁, C₁,...}, preference tranche A consists of citizens {A₁, A₂, A₃,...}.

Next, we state explicitly which individual characteristics are assumed to be the subject of personal responsibility and which are not. Except for the following section, we will assume in this paper that implementing responsibility-sensitive egalitarian justice requires choosing an allocation rule that satisfies the principle "no individual should be put at a disadvantage relative to his fellows because of his lower productivity level; if such a disadvantage is the result of his preferences, we do not consider that unjust." (Any model simplifies reality, of course. One may also consider productivity as a metaphor for productive talents and preferences for working time as representing economic choice).

Roemer then proposes to reason as follows. Ideally, what the responsibility-sensitive egalitarian would like to do is choose the policy which equalises advantage across productivity types, but allows it to differ within the productivity types according to the choices people make concerning their working time. Or put another way, two people with different productivity levels that have the same propensity to work should end up with the same advantage, whereas two people with the same productivity level but different propensities to work should not have the same advantage. Of course, full equality is not always feasible. In such cases the responsibility-sensitive egalitarian has to satisfy herself with an allocation rule that maximins advantage across productivity types for each preference tranche. In other words, she looks for the policy that yields, for each tranche of preferences, maximal advantage for the worst-off productivity type. In terms of our notation, what the egalitarian government would like to do is to maximise the advantage of the worst-off productivity type within tranche {A₁, A₂, A₃,...}, do the same within preference tranche B={B₁, B₂, B₃,...}, etc. Now, as such a series of maximisations cannot be performed simultaneously, Roemer proposes to choose the policy that maximises a weighted average of the minimum advantages across productivity types, where the weight assigned to a given preference tranche is its frequency in the population.

Hence, if for example the ‘productivity 1 type’-individuals are the worst-off in terms of advantage in each of the preference tranches A, B, C,...,N, Roemer proposes a method which maximises a weighted average of the advantage of A₁, B₁, C₁, ..., N₁. In his own words, this position can be described as one where, when looking at a preference tranche, it is Rawlsian; among tranches, it is utilitarian, in giving equal consideration to each tranche.(4) Roemer justifies this method by appealing to the fact that egalitarians who hold people responsible for their preferences must, for that reason, be impartial between variations in individual preferences in the population. Responsibility-sensitive egalitarians have to count "each preference for one".

We now state two basic assumptions that serve to simplify the setting. First, we assume that it is indeed the case that the worst-off citizens belong to a single productivity type (e.g. type 1, as in our examples above). This is of course not a trivial assumption; it depends not only on empirical facts, but also on the way one measures individual advantage. Second, it is assumed that productivity levels and preferences are distributed independently from each other. Within every productivity type one finds the same distribution of preferences; within every preference tranche one finds the same distribution of productivity types.

Under these assumptions, Roemer’s procedure is straightforward and proceeds as we have indicated in our examples above. First identify the (unique) worst-off productivity type. (5) Then select the option set that maximises average advantage for the worst-off group. Since by assumption the distribution of preferences within these worst-off groups corresponds to the population-wide distribution of preferences, impartiality between the preferences of the worst-off is equivalent to impartiality between the preferences of all individuals. (6)

At this point, we stress two things. First, a collective choice rule such as the one just presented is needed to guide choices between states of affairs, i.e. to turn from a situation in which a responsibility-sensitive egalitarian could only distinguish equal states from unequal states, into one where actual policy choices (between option sets) can be made. This need for guidance appears both in situations where different allocation rules can achieve equality (as in option sets I and II), as well as in situations where only maximin can be satisfied (where some individuals would enjoy better option sets, but where the policy that maximises the position of the worst-off is chosen). Second, when calling this collective choice rule "impartial", we actually refer to the impartial way of aggregating interindividual differences in ‘responsible traits’. Clearly, the very fact of using an objective evaluation of personal well-being, "independently of the person who holds the job", will introduce a bias in favour of those individual preferences fitting in most with the publicly affirmed normative valuation.

The responsibility-sensitive egalitarian way of choosing between policies can be contrasted with a case in which an egalitarian government does not consider individuals responsible for their preferences (next to considering them not responsible for their productivity).(7) The profound difference between these two conceptions of responsibility can be illustrated starting from our option set example. Whereas both option sets depict an equal state if one holds people responsible for their propensity to work, neither of them does so if the government thinks people are not responsible for their preferences. Indeed, since the government then thinks people do not genuinely choose the number of hours they work, equality would only be obtained in our simple setting if the reward is the same for everybody.

More generally, if we still take the maximin rule as expressing egalitarianism, such a government would not look at averages across preference tranches. It would choose a policy that maximises advantage for the worst-off individual across both preference tranches and productivity types. Clearly, such a government would opt for a policy that differs from the one chosen by its responsibility-sensitive counterpart.

Since such a government holds people responsible for fewer characteristics, it will in general redistribute more (referring to the example, it would prefer an option set in which one receives the same ‘salary’, whether one chooses job F, job P, or the zero option). This does of course not imply that it is ‘more egalitarian’ than a responsibility-sensitive government. They simply have a different conception of what equality (and compensation) implies.

Before we proceed, we emphasise that the conception of individual responsibility and the conception of advantage are two distinct normative issues that each bear on the ultimate choice of instruments. The first bears on how the government aggregates ‘something’, whereas the second determines what it aggregates. In fact, the conception of advantage would also be important if the government is responsibility- insensitive or even non-egalitarian if its aggregation method is merely to add up advantages. Yet, it is clear that both questions are pertinent when linking responsibility-sensitive egalitarianism with government decision-making.

To keep our exposition brief, we will in the remainder of the paper focus on the impact of the conception of advantage on the choices of a responsibility-sensitive egalitarian government. (We refer the reader to an appendix for an additional discussion about the influence of the conception of personal responsibility.) Hence, we now return to the setting in which people are held responsible for their preferences but not for their productivity.

Some have claimed that responsibility-sensitive egalitarianism entails a principle of natural reward. Thus, according to Fleurbaey (1995, p. 685), an ethic of responsibility conveys the ideal that a society "should let the agents exercise their responsibility and bear the consequences of such exercise, without trying to distort their outcomes in a particular way and with particular incentives. If there is some ‘natural reward scheme’, it should, according to this view, operate as freely as possible." The differentials in individuals’ responsible characteristics should operate fully. One can formulate various axioms specifying a principle of natural reward. Such an axiom could be constituted by the following requirement: if all individuals are identical with regard to the traits for which the government does not hold them responsible, there must be no difference between the pre- and post-transfer distribution of resources in society. Hence, in the setting we discuss, this axiom demands that there be no redistribution when all individuals have the same level of productivity.

In Vandenbroucke (1999a) it is argued that responsibility-sensitive egalitarianism by itself does not lead to the imposition of natural reward, but we will not develop this issue here. We do want to show a related point however. On the basis of what has been said previously, we want to highlight the interconnection between principles of rewards and metrics of advantage. This interconnection is most visible when, starting from given specific assumptions, one establishes a highly simplified optimisation model. Given specific assumptions, it can be shown that with the use of a Roemerian "impartial" collective choice rule, principles of reward and metrics of advantage completely determine each other. Or, stated differently, if decision making by a responsibility-sensitive (egalitarian) government is based on this collective choice rule, the actual government’s conception of individual well-being will drive the choice between various ways of framing the option sets (and hence between their associated principles of reward).(8)

This can be illustrated by some simple calculations on the basis of our earlier example. Recall that the two option sets allowed for equality, and recall the three-step procedure we indicated for choosing between them. First we determine the choices people would effectively make in each of the option sets. Then, we assess the level of advantage associated with these choices. Finally, we average advantage over all individuals, counting each individual for one in the weighting procedure, after which the two option sets can be compared. It can already be seen that ‘valuing advantage’, which comes in at the second step, will thus be quite crucial in determining the eventual outcome.

To illustrate this, we need assumptions about people’s effective choices. Hence, we assume the following responses:

Given option set I,

50% of the population choose job F;
50% of the population choose job P;
no one chooses the zero-option.

Given option set II,

40% of the population choose job F;
55% of the population choose job P;
5% of the population choose the zero-option.

Consider next some alternative public systems of metric to calculate the individual advantage associated with particular choices. In each alternative system we count the advantage of income as being equal to 1 per unit of income. The difference between the metrics is based on different conceptions of the burden of paid labour. More precisely, the government thinks that, when one does not take into account the monetary reward, paid work is on balance a burden for people. Work may be a mixed blessing: paid work brings some benefits ("developing human capital", "structuring one’s life",…) and some burdens ("having less time for the family", "less leisure", etc.). We will use ‘the burden of working’ as shorthand for the non-monetary advantages and disadvantages of labour market participation, where we assume that on balance the former are outweighed by the latter. We measure the burden of working in a particular job as a disadvantage expressed in numbers (9):

Metric M1: disadvantage job F=0; disadvantage job P=0;
Metric M2: disadvantage job F=5.5; disadvantage job P=2.5;
Metric M3: disadvantage job F=7; disadvantage job P=3;

Metric M1 implies that the government measures advantage only in terms of the income people have. Metric M2 implies that government conceives individual well-being as being negatively influenced by working and metric M3 associates an even higher disadvantage with the burden of working.

Clearly, if government measures people’s well-being only in terms of money, option set I is preferable as it yields a higher average advantage in comparison to option set II. But if we increasingly value time for non-market activity (i.e. if we attach growing disadvantage to working time), there comes a point at which option set II is preferable (e.g. given metric M3), because policy I’s strong work incentives lead people to overwork on advantage-maximising standards.

Broadening the perspective to a case where more option sets are available, the example suggests that to each metric corresponds a best option set. Since each metric describes a definition of advantage, we may summarise this by saying that, given the Roemerian collective choice rule, to each governmental conception of individual well-being, i.e. to each public metric of individual advantage, corresponds a collective choice of the ‘best’ option set, and hence, a ‘best’ policy, (and a ‘best’ principle of reward). This implies that no further philosophical arguments are required within the framework of this model in order to explain differences in policy choices. For example, choosing option set I, characterised by the absence of a basic income, is not justified by appealing to a principle of "reciprocity" (no productive contribution, no income). Conversely, if in the above setting the government chooses option set II, this is not because it appeals to deeper philosophical foundations justifying an unconditional basic income. Of course, this last remark should be put into the right perspective. It should by no means be considered as a proof of the superfluous character of such foundations per se. For instance, White (1997) appeals to a principle of reciprocity to reject basic income; following White we may supplement the principle of equal access to advantage by a principle of reciprocity. The approach presented here does not rely on such a principle; in a sense, it boils down to "advantage-maximizing" among equal-access-to-advantage structures.

Figure 1: an optimal policy track

Finally, figure 3 displays the result for an intermediate case in which there is a constraint of the type s < s* (covering for example a setting in which a government pays a limited wage subsidy, which benefits low-paid workers relatively more, but which is universal qua technique, to avoid training disincentives). The outcome is a mixture of the unconstrained case considered in figure 1 and the regime without labour subsidies as captured by figure 2.

Figure 3: an optimal policy track with constrained subsidy level

It can be proven that this will indeed always be the case. In the neighbourhood of the optimal policy track —which depicts an optimal relationship between s and t—, an increase in the subsidy leads to a lower level of B (which still is, given the government’s objective, an optimal level). Two conclusions can thus be drawn as regards the relationship between basic income and the responsibility-sensitive egalitarian government depicted in this paper:

An alternative statement of conclusion (b) is that a lower basic income level is associated with a government valuing more highly the non-monetary rewards associated with participation in economic activity. Referring to our discussion in the previous section, conclusion (b) implies that in general a rise in s will not be exclusively financed by a rise in t. What underlies this result is that a change in the metric of advantage entails a change in the overall reward structure for paid labour, which can be achieved by shifting the balance between income taxation and the unconditional transfer B.

Finally, we point out that conclusion (b) also holds when the model operates with an additional constraint on the level of s (12). A fortiori, it also holds for the restricted model without labour subsidies (i.e. where s is restricted to the value zero, as in figure 2). Again, the underlying reason for this result is that one seeks to increase the net reward for working when the burden of work is regarded as less important.

Now, some may ask whether there exists something like an optimum optimorum on the optimal policy track. That is, one could ask whether the different optimal policies are not merely different per se, but could differ in offering greater or lesser justice. In other words, one could take the findings of the model as a point of departure and then ask whether a policy favouring labour subsidies over basic income offers greater or lesser justice than its opposite. Within the framework of the model, this requires one to focus on choices between different metrics of advantage.

In this respect, we first point at a negative result that follows from conclusion (c) identified above. Conclusion (c) is important with respect to the operational value of the real freedom inclusion rule discussed by van der Veen (1997, 1998) in the context of the broader debate on basic income. Van Parijs (1995) defines real freedom as the ability to do what one might want to do (irrespective from what one actually does). Van der Veen (1997, 1998) suggests an ordering of option sets on the basis of this concept: "a person’s real freedom is said to improve from one regime to another if and only if his choice set unambiguously expands […] If someone’s choice sets in two regimes contain non-overlapping income-leisure combinations, it then follows that the extent of his real freedom in this regimes cannot be compared." (1997, p. 276-277) Now, the irreducible conflict between s and B when the metric of advantage changes, results in any pair of alternative option sets (defined by t,s, and B) not being comparable on the basis of the real freedom metric. The real freedom inclusion rule has consequently no operational value as a means to help someone choose between alternative definitions of advantage.(13)

An alternative approach starts from the observation that metrics of advantage and principles of reward determine each other, given the adoption of a Roemerian "impartial" collective choice procedure and specific assumptions of the model under review. Now, if one accepts the argument stated earlier that a principle of natural reward is intimately linked to responsibility-sensitive egalitarianism, then of course one can use such an (axiomatic) principle to identify its associated advantage metric, which we may then call the ‘neutral conception of advantage’. This would evidently amount to the identification of a specific point on the optimal policy track, given the parameter values in the complete model.

While we have indicated above that one may raise fundamental doubts about the fact that responsibility-sensitive egalitarianism necessarily entails such a principle of natural reward, it is still true that someone could wish to rely on it to select a specific policy. When this is the case, the following result becomes highly relevant (see Vandenbroucke, 1999a). If government opts for a neutral conception of advantage —by adhering to the axiom that there should be no redistribution when all individuals have the same productivity level—, and if it can achieve equality by using earned income taxes and labour subsidies, then it will not cash out a basic income. In other words, it follows from a model such as the one outlined above that a neutral government that can achieve equality (as its optimal policy), will use earned income taxation exclusively to fund labour subsidies.(14),(15)

Indeed, one should recall that the ethical justification of the active welfare state considers participation in society as one of the social bases of respect and self-respect, thereby making clear that one should not merely look at participation on the labour market. For evidently, the social bases of respect and self-respect can also be strengthened through other, non-market activities such as taking care of a child, a parent, a friend,…, voluntary social or cultural work, etc.

One could claim that such activities are nevertheless subsumed in the model, more specifically through the individual advantage associated with time for non-market activity. Still, this is a rough and indirect manner to treat them. It is a rough manner, since it does not allow to distinguish between various forms of non-market activities —e.g. ‘pure leisure’ versus voluntary social work, versus taking refresher courses, versus…— which could be valued differently by the government. Moreover, such valuations would then again be dependent on normative choices (relating to the question as to what extent such activities are considered as bases for respect and self-respect), which in this instance could be captured in principle by some multidimensional metric of advantage. It is an indirect manner, since the advantage of time for non-market activity appears in the model only as the mirror image of the disadvantage associated with the burden of paid labour (which turns the model into a kind of simple dichotomous model).

More generally, the model lacks an explicit valuation of participation in social activity, be it in or out of the formal labour market. Indeed, also with regard to participation on the labour market there is no direct reference to its capacity as a social basis of respect and self-respect. In short, it would be erroneous to consider the model as one that describes how different conceptions of participation would influence policy design. Putting things into the right perspective, it is a model which is suited to capture the influence of public valuations of some specific qualitative aspects of paid labour (such as stress, or its opportunity cost in terms of time for other activities), but it does not address the importance of participation per se.

Finally, the model is also very much of a simplification because it has no intertemporal dimension and therefore can only incorporate a limited set of distributive instruments. The broader conception of participation in society requires one to think about the most appropriate combination of work and non-market activities over the entire life cycle (e.g. should we not design policies such that, as compared to the current situation, people work less between ages 25 and 40, and more between ages 55 en 65?). A "relaxed labour market" would allow people to leave and re-enter the labour market easily (for example to care for someone, to update their qualifications, or simply to replenish their reserves; see Vandenbroucke, 1999b). The model we presented cannot capture this conception of a relaxed labour market. Assuming that it could capture it, and neglecting the other problems we have just discussed, it is not at all clear whether a government granting a high value to an individual’s time for non-market activity would still opt for a high basic income, or whether it would instead favour other policies such as attaching to every job a right to full-time or part-time sabbatical leave, which acts as a more direct (qualitative) supplement to the monetary reward of paid labour.

Hence, one should avoid extensive optimum optimorum-debates within the confines of a limited model. We nevertheless think it is worthwhile to convey the basic message that a discussion about how to get rid of inactivity traps is not merely a matter of choosing the most appropriate instrument given economic circumstances. We have to make a complex political choice on at least two issues: a conception of responsibility, and, as we stressed in this paper, a conception of individual well-being. As we have mentioned in the previous section, if a richer model were to be constructed, it would ultimately need to address essentially the same kind of issues in a multidimensional way.

Hence, even if one constructs a model that directly addresses the broader idea of participation and/or allows for more instruments, one still faces the issue that the way in which one measures the quality of life or the level of advantage, in particular that of the worst-off, co-determines policy choices with regard to distributive instruments. "Being a responsibility-sensitive egalitarian" is no sufficient basis for guiding policy choices. Implementing policies presupposes that one makes choices concerning the metric used to evaluate the quality of life; a choice that can only emerge from democratic debate in general and from the political process in particular.

(a’) Other things being equal, a government that considers people responsible for their preferences concerning time for non-market activity will propose a higher earned-income tax rate and labour subsidy rate than a government that does not hold people responsible for their preferences (16).

(b’) For a given conception of advantage, and given a set of instruments and constraints, a government that considers people responsible for their preferences will propose a lower basic income level than a government that neither holds people responsible for their productivity nor for their preferences for working.

Obviously, the choice of instruments is not exclusively driven by normative conceptions, but by the economic environment as well. To illustrate this, we reproduce here a simple formal expression for the optimal earned income tax rate in a regime without labour subsidies. The optimal t in this case depends on two variables, the policy parameter a , and a factor J that captures pre-tax inequality:

COHEN, Gerald, A. (1989), "On the Currency of Egalitarian Justice", Ethics, 99, pp. 906-944.

COHEN, Gerald, A. (1993), "Equality of What? On Welfare, Goods, and Capabilities", in NUSSBAUM, Martha and SEN, Amartya K., The Quality of Life, Clarendon Press, Oxford, pp. 9-29.

COHEN, Gerald, A. (1999), "Socialism and Equality of Opportunity", in ROSEN, Michael, WOLFF, Jonathan, and MCKINNON, Catriona, (eds.), Political Thought, Oxford University Press, Oxford, pp. 354-358.

FLEURBAEY, Marc (1995), "Equality and responsibility", European Economic Review, 39, pp. 683-689.

FLEURBAEY, Marc (1998), "Equality among responsible individuals", in LASLIER, Jean-François, FLEURBAEY, Marc, GRAVEL, Nicolas and TRANNOY, Alain (eds.), Freedom in Economics, New perspectives in normative analysis, Routledge, London, pp. 206-234.

LE GRAND, Julian (1991), Equity and Choice. An Essay in Economics and Applied Philosophy, Harper Collins Academic.

RAWLS, John (1971), A Theory of Justice, Oxford University Press, Oxford.

ROEMER, John. E. (1994), "Equality of resources", Part II in Egalitarian Perspectives. Essays in Philosophical economics, Cambridge, Cambridge University Press, pp. 115-196.

ROEMER, John. E. (1996a), Theories of Distributive Justice, Harvard University Press, Cambridge Mass.

ROEMER, John. E. (1996b), "On a Mechanism for Implementing Egalitarianism with Responsibility", in FARINA, Francesco, HAHN, Frank & VANNUCCI, Stefano (eds.), Ethics, Rationality, and Economic Behaviour, Clarendon Press, Oxford, pp. 142-158.

VAN DE GAER, Dirk, (1993), Equality of Opportunity and Investment in Human Capital, PhD-Dissertation, K.U. Leuven.

VAN DE GAER, Dirk, MARTINEZ, Michel and SCHOKKAERT, Erik (1998), Measuring Intergenerational Mobility and Equality of Opportunity, Center for Economic Studies, Discussion Paper Series DPS 98.10, Leuven, April.

VANDENBROUCKE, Frank (1999a), Social Justice and Individual Ethics in an Open Society: Equality, Responsibility, and Incentives, PhD-Dissertation, Oxford, forthcoming in the series "Studies in Economic Ethics and Philosophy" ed. by P. Koslowski, Springer-Verlag, Berlin (17).

VANDENBROUCKE, Frank (1999b), The Active Welfare State: A European Ambition, Stichting Dr. J.M. Den Uyl-Lezing, Amsterdam.

VAN DER VEEN, (1997), Robert J. "Debate: Real Freedom and Basic Income: Comment on Brian Barry", The Journal of Political Philosophy, 5, pp. 274-286.

VAN DER VEEN (1998), Robert J., "Real Freedom versus Reciprocity: Competing Views on the Justice of Unconditional Basic Income’, Political Studies, 46, 1 (March).

VAN PARIJS, Philippe (1995), Real Freedom for All. What (if anything) can justify capitalism?, Clarendon Press, Oxford.

VAN PARIJS, Philippe (1997), "Reciprocity and the Justification of an Unconditional Basic Income. Reply to Stuart White", Political Studies, 45, 327-330.

WHITE, Stuart (1997), "Liberal Equality, Exploitation, and the Case for an Unconditional Basic Income", Political Studies, 45, 312-326.

WHITE, Stuart (1999), "The egalitarian earnings subsidy scheme", The British Journal of Political Science, 29, pp. 601-622.

1. We are well aware that behind this phrase lies the difficult debate as to what "responsibility" means, i.e. whether one is responsible for a certain characteristic because one is actually "in control" of it, or whether one is held responsible for that characteristic on the basis of a mutual agreement.

2. This succinct statement is inspired by Cohen (1993, p. 28; 1989, p. 916).

3. This first step is the "natural way" for economists to proceed: they traditionally assume that the real value of an option set lies in the best use that can be made of it, which is under their usual assumptions the use that is actually made. This can be criticised from a freedom perspective, i.e. it can be said to neglect the importance of flexibility for individuals (the importance of having options which are not actually taken up, but which might be taken up given some change in preferences).

4. The expression "utilitarian" may be somewhat misleading in this context since we are not comparing utilities, but a publicly affirmed objective notion of advantage. Nevertheless, it clearly reveals the similarity with the utilitarian conception of impartiality.

5. Note that the simple example deals with a case in which two equal states are compared. Therefore, in this example everybody belongs to the worst-off (or best-off) type, implying that we will have take into account the effective choices of the entire population to calculate average advantage. This is notably the way we proceed in section 4 when choosing between the two option sets of the example (as can be seen in more detail in footnote 10).

6. Moreover, if these basic assumptions hold, Roemer’s approach yields the same results as those obtained through a slightly different implementation procedure proposed by Van de gaer (1993) and Van de gaer, Martinez, and Schokkaert (1998). On this, see Vandenbroucke (1999a).

7. At the risk of introducing some confusion in terminology, such a government could be labelled "Rawlsian", since Rawls in his mainstream exposition rules out individual responsibility for "work effort" (Rawls, 1971, p. 312).

8. Of course, as we mentioned at the end of section 4, the conception of advantage will influence the choice between policies as soon as ‘advantage’ is an argument in the social objective function of a government.

9. This implies that we allow for inclusion of "time for non-market activity" in the Rawlsian bundle of social primary goods.

10. For instance, for the entry combining option set I with metric M2, the average advantage is found using the formula 0.5 x (10-5.5) + 0.5 x (5-2.5) = 3.5. Or, given option set II and metric M1, using (0.4 x 9) + (0.55 x 6) + (0.05 x 3) = 7.05, etcetera.

11. It can be seen on the figure that the track eventually stops at the point where s=t=1. This feature follows from the assumption that t can be at most 1, and from a labour supply constraint in the complete model ensuring that an individual cannot work more than a certain maximum number of hours. The latter requirement is fulfilled if the subsidy is not set higher than the income tax rate.

12. Cf. Vandenbroucke (1999a).

13. Referring to the appendix, neither can it help to choose between different conceptions of personal responsibility.

14. As the model in Vandenbroucke (199a) allows for another revenue source, viz. capital income taxation, and for spending on education, the result as identified there states that, under the conditions identified in the main text, there will be no positive basic income unless there is a positive residual left after having funded educational expenditures generated by the capital income tax. In other words, if this residual were negative, B<0, and one ends up with a uniform poll tax to co-fund education.

15. Moreover, given the specialised setting identified in the main text, it can be shown that this result is essentially the Egalitarian Earnings Subsidy Scheme as proposed by White (1999). Recalling our remark at the end of section 4, White’s underlying idea of reciprocity (no paid work, no subsidy) is thus vindicated in this setting, without having to appeal to other philosophical foundations than those presumed in the maximin rule and ‘neutrality’.