Such reasoning, based on the narrow rationality postulate, presents two advantages for anyone tackling our research problem. First, it can be empirically checked by examining RMI exit rates. Second, when it gives rise to or fuels the idea that welfare protection levels are too high because they dissuade people from trying to find work or accepting work offered, it may well have ideological and policy effects. [2] If we look more closely, however, we see that the model, while not irrelevant (if only because vaguely tautological), is far from satisfactory. Our analysis of over 20,000 RMI recipient files shows that if we remain at this first level of rationality, a high rate of individuals behave “aberrantly”. A great number of those whose interest would indicate getting off the RMI, and who have the personal resources to do so, stay on, while just as high a rate of people whose interest would indicate staying on and who do not a priori have the resources to get out of the inactivity trap choose to go to work. Moreover, in all cases expected financial gain plays a lesser role than available opportunities and individual resources.

This observation obviously does not mean that these actors are not rational. Rather it invites us to effect a “reduction” of the coarse rationality assumed by the model. We use the term “reduction” in the chemical or culinary sense–an operation which identifies the active components in complex molecules or “simmers away” excess water in sauces so that only ingredient juices and flavors remain– with the idea that here that operation will enable us to perceive and describe actors’ fine, highly contextualized rationality. In practice this has involved exchanging quantitative for qualitative instruments : we questioned approximately one hundred RMI recipients or former recipients about how they understood their strategies and choices. On the basis of the interviews it was clear that the relevant model was no longer “narrow rationality” but “good reasons”. Still, the good reasons collected appear so diverse and reversible that we may well ask whether what sociological study gains in interpretative capacity through use of this model it does not lose in explanatory potential.

In even less favorable employment situations, where RMI recipients obtain only half- or three-quarter-time jobs, not only do they gain nothing by going off welfare, but they lose a considerable amount and, given the precariousness of their jobs, are in danger of needing to go back on (Padieu, 1997). A study conducted by the Institut National de la Statistique et des Études Économiques (INSEE) (Afsa and Guillemot, 1999) shows that 44.8% of jobs occupied by persons just off the RMI are part-time jobs of fixed duration. As O. Benoit-Guilbot (1990b) notes, it is as if employers, not confident in the capabilities of former RMI recipients, used the unstable jobs they have to offer as a kind of probation-period ordeal through which former recipients must pass. There are studies showing that more often than not these jobs do lead to stable employment–thus to significant marginal gains– rather than renewed unemployment (Florens et al., 1990). Nonetheless, from the actor’s point of view, the risk of renewed unemployment is there, and going back into the welfare system involves a whole set of costs in itself, such as renewed administrative procedures, liquidity problems while waiting for the response to one’s new application, and the danger of having that application turned down after taking a job that does not become stable employment. In game theory language, it would be in these individuals’ best interest to adopt a “maximin” strategy aimed less at winning big than simply not losing (Delvaux and Cappi, 1990).

Low income differential on the one hand and the financial uncertainties involved in reapplying for the RMI on the other may well lead the individual to become encysted in welfare and the financial minimum it guarantees. The inactivity trapdoor leads to a chamber in which the RMI recipient is caught by means of a pawl device for, over time, individuals lose their skills–what economists call the “degeneration of human capital”. Likewise, employers view a long period on RMI rolls as a negative indicator and are less likely to hire long-term unemployed job candidates. For these reasons unemployment can be called a reverse waiting line : the last to become unemployed have a better chance of being served – ie, finding a job– than the first in line.

Taking into account these two factors–expected financial gain and length of time on RMI rolls– we can use the inactivity trap model to distinguish four groups.

TABLE I.
Inactivity trap model

“Disincentive” factors combine and reinforce each other in Group 2; inactivity trap mechanisms should operate fully in this group : individuals may expect little gain from obtaining employment and are disqualified for having been a long time on the RMI. [3] Conversely, Group 3 comprises individuals who are more likely than all others to benefit financially from employment and who are most likely to be hired; their self-interest would seem to indicate getting off the RMI. It is on these two groups that the inactivity trap model should be tested, together with the particular “actors’ rationality” model on which it is based, as these situations are the most clearly identifiable, homogeneous, and decisive. Do individuals in Group 3 find jobs at a sharply higher rate than individuals in Group 2 ?

2. To determine the effects of income differential between RMI and employment, we used Claudine Padieu’s method (1997) of calculating income differences per household unit (not including rent and assimilable). For each type of household, we calculated the percentage of financial gain that employment would bring over and above the sum effectively received in the way of welfare minimums and allocations of various sorts. To do this we had to consider the various types of welfare aid, as well as number and age of children in household. We worked with different types of scenarios, comparing welfare minimums received with minimum wage and three-quarters minimum wage jobs. [7]

Table II is of crucial importance because it shows that while the probability of going off the RMI increases with expected income differential, the progression is neither strong nor unequivocal. The differences in exit rate by expected gain are slight, +6.4% for minimum wage to +4.3% for three- quarters minimum wage; above all, they are highly irregular, with jumps between differential categories. At three-quarters minimum wage, more persons who stand little to gain go off the RMI than persons who would increase their income by 50%; at minimum wage, persons who stand to gain between twice and 2.2 times as much are less likely to go off than persons who stand to gain between 1.8 and twice as much. As for general increase in exits by expected gain, the differences involved seem to us too slight to be convincing. We can therefore immediately reject the hypothesis that actors behave only in accordance with a narrow “economic costs/benefits” calculation. This would seem to condemn the narrow utilitarian rationality model, together with the simplistic ideological clichés that sometimes accompany it–not to mention the fact that many factors in addition to expected financial gain may intervene in choice-making. Moreover, the table is based on an abstraction : it is as if all individuals had been offered jobs and then decided either to take them or stay on the RMI, whereas this is obviously not the case. In reality, we cannot know all job offers received by all members of the sample. We do, however, know the rates of exiting the RMI by the various employment zones of which the Gironde department is comprised, and as it turns out, they are not correlated to job offers. We can therefore conclude that variation in exit rates is not merely an effect of job offers.

TABLE II.
RMI exit rate by income differential

TABLE III.
RMI exit rate by employment zone within the Gironde

Direction Régionale du Travail, de l’Emploi et de la Formation Professionnelle [Regional bureau of labor, employment, and occupational training] for Aquitaine region.

With these results in hand, we cannot accept the conclusions proposed by Laroque and Salanié (2000) on the basis of a 1997 INSEE employment survey, specifically their “non-employment” breakdown. They affirm that of the 3.25 million unemployed persons aged 25 to 49 throughout France, 57% are cases of “voluntary non-employment” explained by how little they would stand to gain from having a job relative to the welfare payments they receive by remaining unoccupied. 20% of cases, on the other hand, are said to be a matter of “classic non-employment” given what would be their low productivity compared to cost of paying them minimum wage. And 23% are said to be cases of “Keynesian” or “frictional” unemployment. The narrow rationality model is thus used to separate “wheat” from “chaff” poor persons in a time of ample job supply–an age-old social policy problem. The authors add that lowering what employers have to pay to the state on a minimum wage job would create 500,000 jobs, whereas increasing the minimum wage would cost 290,000 jobs. In his critique of this analysis, Sterdyniak (2000) particularly denounces the imprecision of their calculation method, showing that many job-holders agree to work for very low gain and that their number would be even higher if part-time work had been taken into account, whereas Laroque and Salanié’s projection makes no distinction between job seekers and unoccupied persons, pays no attention whatsoever to employer hiring behavior, does not count part-time work, and so forth. We would add that the expression “voluntary non-employment” is particularly shocking given that a study by Rioux (2000) indicates that three-fourths of RMI recipients look as actively for work as other unemployed persons and that most of them stress the financial constraints this entails : transportation, telephone bills, clothing costs, etc. 3. According to Laroque and Salanié, only 20% of unemployment can be put down to individuals’ low skill level. In fact, remaining a relatively long time on the RMI seems to weigh much heavier on exit probability. In general –and the trend is perfectly regular– exit rate decreases with “seniority”: at the time of the survey, it was 18.2% for persons who came on in 1989,29.8% for persons who came on in 1993, and 50.1% for those who came on in 1997. “Seniority” is thus an essential variable and must be included in the general model for predicting exit. If we distinguish between two groups in terms of length of time on the RMI–above or below four years– and income differential on the basis of three-fourths minimum wage (the most likely employment situation), we get the following results:

Like Table IV, Figure I, representing exit rate by length of time on RMI and expected income differential, shows that expected gain is less decisive than length of time and has such irregular influence that we cannot derive a general trend. When we combine length of time on the RMI with income differential, we observe that the inactivity trap model, while not entirely invalidated, has only limited explanatory capability. First, length of time on the RMI is much more decisive than income differential, with a total difference of 18 percentage points for the first as opposed to only 7 for the second. Also, individual’s social and personal resources count for more than expected benefit alone. Furthermore, more than 23% of individuals who don’t stand to gain much and who a priori don’t have much in the way of personal resources (Group 2), nonetheless go off the RMI. It is probably the case that twice as many people optimizing expected gain and resources (Group 3) leave the welfare system, but length of time weighs more than expected income differential. At this level of analysis and taking into account the fact that the material at our disposal was fairly cursory–by definition it included no mention of undeclared income– we can say that the results do not validate the simple utilitarian approach holding that individuals don’t go off the RMI because they don’t stand to gain financially by doing so. Moreover, if we examine the hypothesis of an income differential based on full minimum wage, thus significantly increasing expected gain, the differences in exit rates between Groups 2 and 3 are even slighter : 21.9% and 41.7% respectively.

TABLE IV.
Exit rate by length of time on RMI and income differential (average rate: 35.8%)

In order not to narrow the general inactivity trap model down to mere utilitarianism, we have included other items among predictive factors for going off or staying on the RMI. Data available in the CAF files enable us to examine five such variables. [8] Regression analysis shows that their combined influence is slight–0.136 of variance– and that among them, the role of expected gain is even slighter (see Appendix).

Because the sample could not be representative if the data were too coarse, we gave preference to the most clearly “aberrant” cases, those which are of greatest interest to sociologists because furthest removed from the utilitarian narrow rationality notion. Fifty-two of the interviews were with individuals who in utilitarian terms “should have” gotten off the rolls but hadn’t; 40 with individuals who “should have” stayed on the rolls but hadn’t. Another 20 interviews were with persons in the first case who did go off, 16 with persons in the second who did not–the two ways of conforming to inactivity trap model predictions.

2. One feature of the inactivity trap model that makes it seem particularly convincing is that it is based on a conception of the group under consideration as unified, homogeneous. Though with social services statistics (those furnished by the CAF) we can construct a few simple subcategories (based on length of time on RMI, family situation, sex, age, and so forth), those statistics constitute a kind of primitive administrative categorization incapable of accounting for the extreme diversity of the individuals on the RMI. As soon as we consider these individuals, notions of actors’ resources and actors’ interest literally blow apart into a mosaic of individual cases and stories. A survey by the CREDOC (Centre de Recherche pour l’Étude et l’Observation de Conditions de Vie) underlines the strong heterogeneity of RMI recipients; the only point they have in common is that many have occupational itineraries marked by unemployment and distance from the job market (Croutte, Iliakopoulos, and Legros, 1991). For all these people, the RMI is part of different situations and histories, and though their desire to improve those situations is not thereby affected, the rationality of individual actors seems singularly circumscribed and contextualized, quite removed from the narrow rationality notion that shapes many ideological discussions about poor people’s will to get out of poverty. At the risk of establishing a kind of catalog–worse, an incomplete one– let us evoke the patchwork made up of the one hundred or so persons interviewed, aware that it too is only an on-the-spot sketch that freezes situations at the moment of the interview. Indeed, the interview technique itself unavoidably engenders self-selection, since many respondents did not wish to meet sociologists, whom they could not help perceiving as agents of social control.

A segment of RMI recipients is made up of individuals and families from traditionally poor, dependent social groups, whose parents and grandparents already belonged to the “fourth world” of the odd seasonal job, rural or urban poverty pockets, children taken from parents and placed in custody of the social services, and so forth. Others, on the contrary, are literal unemployment accidents : laid-off managers, middle-class women alone after divorce, skilled workers laid off in company restructurings, and so forth. The two groups have nothing in common but the RMI, and they perceive the RMI quite differently : for the first group it’s considered a serious bit of good fortune while for the second it is both a catastrophe and a lifeboat.

We can also identify a group of RMI recipients who think of themselves first and foremost as having nearly exhausted their unemployment benefits. These persons may be seen–and see themselves– as workers deprived of employment, pure victims of the job market crisis. They are laid-off manual and office workers whose identity is entirely crystallized around work. In contrast, there are young people for whom the RMI is part of a way of getting started in adult life rather than any catastrophic exit from working life. They are prolonging their student lifestyle while looking for a job they really want; for them the RMI constitutes a kind of post-education financial aid. There are also recipients who aim to fulfil an artistic vocation, many in the visual arts or music, for whom the RMI is part of a bohemian life, functioning a bit like the special unemployment benefits for “intermittent” workers in the French entertainment industries.

At one end of the chain, then, are destitute, isolated individuals for whom the RMI is the last source of survival, while at the other we find persons for whom welfare benefits are part of a way of life or a life choice that preexists the benefit. The Tziganes [term for Rom population long present in France] offer a good illustration of the second case. For them the RMI is perceived as either a “gift” or a “debt” paid to a community whose members can thereby enjoy a greater degree of choice. Some have taken up nomadic life again; others have used the RMI to settle, but none has become “integrated into normal life” (Reynier, 1992).

RMI recipients are also an extremely heterogeneous group in terms of capabilities for acceding to employment. It would be misguided to think that if there were jobs for all, every RMI recipient could obtain one, for the simple reason that some are extremely impoverished, isolated, ill. Others, shattered by poverty, solitude, and unemployment, have sunk into a chronic depression that makes any occupational activity seem too much for them–and they may well be right. In many cases the RMI enables people to get through the wait for handicap benefits. Other individuals, on the contrary, present themselves as entrepreneurs in their own cause. They are very busy : looking for work, legal or not, getting occupational training, and generally constructing their lives, either in shadow or the light of day. For them the RMI is a resource among others, and above all a source of social protection and “legality”.

There are many other possible ways of distinguishing RMI recipients. We could contrast urban recipients with rural ones, or isolated ones with those who live in poor neighborhoods that both confine and protect them, transforming their individual stigma into a collective one. We could also contrast those who find job opportunities near them with those who have to move away; skilled with unskilled; those with family and those without, etc.

In sum, the RMI and other welfare programs are general, universal systems that meet up with particular individuals who for their part are indissociable from particular situations and histories. Given this fact, the rationality of behavior aimed at leaving or remaining in welfare benefit systems must be indexed in terms of those situations, histories, and projects. An overly general model cannot account very well for such variables. It is necessary to do away with the notion of a single type of RMI recipient and try to understand the different varieties of reasoning that individual actors follow, how those types of reasoning are themselves inscribed in social service systems, and how they are used. It is true that these individuals are generally weak and dependent, but they are not passive; more exactly, passivity is also a matter of strategy when no other seems possible. We need to be careful not to blame the victim, not to criticize some for their excessive passivity and others for being exceptionally talented at getting social benefits and under-the-table work.

It is always possible to propose typologies that confer a degree of order on complex, even confused-looking behaviors. After underlining recipients’ heterogeneity, the afore-cited CREDOC survey, for example, distinguishes between “marginal users”, “heavily assisted beneficiaries” and three other groups, said to use the RMI for, respectively, health, housing, and occupational integration purposes. A survey by the CERC (Conseil de l’Emploi, des Revenus et de la Cohésion Sociale) distinguishes three groups : those with occupational or employment assets but no social ties (40%), those with social ties but no occupational or employment assets (45%), and those with neither (CERC, 1991). Such typologies are not without their uses, especially for the social services, but they don’t really tell us anything about who RMI recipients are. The intellectual satisfaction to be had from this type of exercise would lead in our case to reifying abstract and above all arbitrarily defined sets of individuals, sets that, in any case, cannot account for individuals’ rationality. Either we must choose a typology so fine as to be meaningless, or a typology so simplistic as to go no further than contrasting men aged 25 to 34 married but childless with 70% expected gain from going off the RMI to all other recipients. It is diversity of RMI population, just as much as rate of “aberrant” cases in terms of the trap model, that invites us to distance ourselves from blueprint-like “narrow rationality” and its requirement that too many “other things” be “kept equal”. The analysis must thus be “reduced” a degree.

Overall, we risk little by saying that individuals with links to their families and various social networks get on better than those who remain apart and alone. The role of networks is not unequivocal, however. Everything depends on what sociability networks an individual belongs to and what use he or she makes of them. It seems obvious that when networks themselves are defined by poverty, separateness, or a particular lifestyle, they may well integrate individual members through strong ties that actually lock them into that position. The first effect such networks have is to preserve the individual recipient, protecting him or her from the stigma attaching to the RMI. If all the people one has relations with are in situations comparable to one’s own, if relatives and friends are all on the RMI, the individual is protected. “Why should I go off when we’re all in the same boat, all the same ?” Housing construction and government policy has had the effect of isolating just such groups of “welfare cases” within extended housing developments; individuals end up together in the same housing project, and as we have observed, poverty pockets are associated with greater probability of staying on the RMI (Table V in the Appendix). But if these networks are so strong, this is because they offer considerable survival resources. They develop collective skills for using the social services. It is in such settings that individual recipients speak of their “rights” and of how familiar and skilful they have become in using the social services since first being taken in charge by them. We also know that such networks increase the number of services exchanged among neighbors – babysitting, various kinds of helping hands, other expressions of local solidarity– and that unless, of course, they involve delinquency or the various types of trafficking that young people in housing projects can get mixed up in, they help provide access to essentially legal resources. Ethnic solidarity can play the same role of integrating individuals in a situation of marginality, organizing life on the margins, making bearable conditions that in solitude would not be.

Moreover, without networks and sociability, there is really no means of getting out of the bad situation; relations open up a way out. Often it is through membership in local and ethnic groups, or friends, or neighbors that individuals ultimately find employment. Being part of an extended family may push an individual to go on the RMI. But if the family has some resources, if someone in the family knows someone who knows someone with a job to offer, etc., it can also help him or her go off. Interviews with RMI recipients of immigrant origin clearly point up the ambiguous nature of networks : they protect and confine, but are also able to provide a helping hand, creating ties with the “ethnic” economy. As for Granovetter’s “weak ties” (1974, 2000), they are usually too few or too weak to be truly effective. Individuals who go off the RMI seem both integrated into a network of strong ties and connected to more open contexts; they’ve got one foot in and one foot out. This is what we hear from “bohemian” recipients, who find resources within their immediate circle but also have enough relations and “social capital” to get out of it when they wish to.

The ambiguity of getting paid “under the table” is exemplary here. In fact, working for payment under the table may confine individuals to their status of RMI recipient by making it “livable” and even, in some cases, “comfortable”, as long as they’re not found out. Individuals in this case don’t really stand to gain from getting legal work. But is it reasonable to believe in the first place that RMI recipients could live off of what they get in the way of official welfare aid ? In this sense the demand for under-the-table work seems inexhaustible–like the supply. This type of activity keeps those who do it from going off the RMI sooner. But it should be kept in mind that illegal work brings individuals into extended networks that may well lead to legal employment, temporary work, limited contract jobs and even sometimes open-ended contract jobs. It even sometimes happens that a mix of RMI benefits, illegal work, and part-time legal work enables the individual to become self-employed, set up a business. In fact, while working illegally confines people to the RMI, it also sets ajar numerous doors to legal work.

Social work seems marked by the same type of paradox. According to some, it locks people into dependency. Administrative procedures for obtaining welfare are so complicated that people cannot do without the social worker’s expertise; they set up case files for applicants and generally put some order into situations of marginality, while often closing their eyes to little illegalities that improve daily existence. The “good social worker” [most often referred to as female], seconded by actors with strategic abilities, can arrange quite acceptable survival situations for people. In some cases, social workers perceive individuals as having only a slim chance of getting off the RMI through employment and don’t really push them to try, thus working to transform recipient’s social problems into individual’s psychological ones. This criticism of social control as effected through “the finer feelings” and leading to recipients’ internalization of stigma is not unfounded. On the other hand, it is sometimes the ill will demonstrated by the social services that locks “clients” into the system–those who don’t want to have to face being humiliated at social service counters, taking “phony” training programs, attending meetings of RMI recipients organized by the Commission Locale d’Insertion (CLI) [Local social-integration commission]; services that force them to identify with a group of “fallen”. The various CLI judged harshly by RMI recipients have reputations that seem to be based on facts that make them deserve such judgments, and in general, the Agence Nationale Pour l’Emploi (ANPE) is criticized for its indifference and inefficiency. Though these are only matters of judgement and reputation, they run through so many respondents’ discourse that we would be mistaken to consider them insignificant. It should be noted here that in 1998, one out of six unemployed recipients had a job six months later. In two out of three cases where that job was obtained through the ANPE it was subsidized by the state in one way or another or else it amounted to paid training. On the other hand, when the RMI recipient found the job himself or herself, one out of four came with an open-ended contract and the same proportion with a limited-time contract, and only one out of four was a state-subsidized job (Rioux, 2000). Recipients’ ambivalence with regard to the effects of social work thus do not seem unfounded, if only because social services are never as efficient as personal resources–on condition that an individual have them, of course, which is not the case for all RMI recipients.

On the other hand, social work can help individuals go off welfare through the psychological support it provides, and above all because most recipients do not obtain employment directly. Having a state-subsidized job, a Contrat-Emploi-Solidarité (CES) [a type of state-subsidized employment], or one of a variety of part-time jobs may confine individuals to sub-networks, but it may also give them a leg up, on condition that such employment is not experienced as “a shit job” and source of continual humiliation. The atmosphere of ambivalence within which these programs function should not lead us to forget their positive role. Paradoxically, social work helps individuals get off the RMI when it is perceived as something repugnant, when recipients don’t want to have anything to do with social workers, don’t want to owe anything to anyone, can’t stand telling the story of their lives to strangers, don’t accept a relation in which they are assisted but also feel strapped up in a status they find unacceptable. Fully half of the individuals interviewed who went off the RMI underline how vehemently they refused to identify with the world of social welfare “cases” to which the social services unavoidably confined them. We could pursue this type of reasoning for many other social factors. Having any kind of educational degree, for example, is unquestionably of ambivalent value because while it is an objective positive resource it may also be a sub-jective handicap involving fear of failure or of seeing a long training period come to nothing.

We could say that those most apt to make it out of their bad situation both have educational degrees and are ready to accept a drop in status from the level suggested by their education. Likewise, having children in the household plays an ambiguous role because it makes it impossible to accept just any job in terms of hours or location yet constitutes a push to accept just any job because being employed will restore one’s image in the eyes of family, particularly children.

In the end, the numerous reversible factors just underlined make sense only in terms of particular situations, specific life stories, and singular combinations of situations and life stories. Individuals seem to move along the razor’s edge; the slightest breeze, sudden setback or opportunity, throws them to one side or the other. The dividing line can thus only be drawn by means of subtle variables and even subtler variable combinations. This remark may be disappointing for social scientists who like solid causalities and inarguable correlation coefficients, but it has the advantage of toppling a great number of clichés and stereotypes about any “specific population” of RMI recipients or welfare recipients in general.

One family of reasons does not appear at all in the general CAF file data and came out only in the interviews : the fact that there are individuals who cannot work. The omnipresent themes of depression, illness, handicap dominated in a great many interviews. We on the outside of recipient experience cannot on the one hand affirm that unemployment and poverty weaken an individual subject’s capacities and then be surprised to see that individuals are indeed destroyed by these situations. It matters little whether physical and/or mental disease is cause or consequence of getting caught in social welfare programs; what seems unquestionable, however, is that a segment of recipients are simply incapable of taking on the “job” of working. For these persons, whatever the overall employment situation or specific expected employment-linked advantage, the RMI functions not as a transitory measure but as subsistence income, very low-level survival income. It also seems that in such cases, in situations where the most pressing need is some dental care, or a place in yet another drug addiction treatment program, or learning the basics of “presenting oneself” in public, social workers have given up the official fiction of an “integration contract”. [10] Here the RMI is relaying defective social programs : recipients take refuge in the assistance made available to them while waiting for other welfare aid applied for to kick in; they seem definitively “out of the running”. We do of course need to know how many recipients are in this case. The founding concept of the RMI clearly does not apply to them since in reality they are waiting to retire, have a handicap recognized by the Commission Technique d’Orientation et de Reclassement Professionnel (COTOREP), [the commission charged with examining files of handicapped persons and fixing special pension rates], or are in some other “exceptional” case. In sum, they are individuals whose self-interest may indicate going off the RMI but who simply cannot.

Another good reason not to go off is the cost of trying to obtain employment–not the same thing as the “expected financial gain” that economists calculate. Many individuals stress how expensive it is to commit oneself to finding work. Here the inactivity trap model does function, on condition that we know and take into account the hidden costs of access to employment, costs that the basic model does not and cannot bring to light with such cursory indicators as “relative expected gain”. The simplest calculation shows that income differential between the RMI and other welfare benefits on the one hand and expected earnings on the other melts away as soon as we take into account expenses for transportation, clothing and “making a good impression”, and child-care. Single women with children are most affected by these kinds of constraints because if they get a job they will have to give up taking care of their children and are often offered working hours incompatible with family life. Without a partner and often without family to relay them, it is “objectively” not in their interest to choose work, or else for a gain so absurdly small that it would in no way compensate for the combined costs of occupational investment plus family disinvestment. If we take into account car maintenance, and/or transportation, and childcare costs, plus the financial effect of losing certain welfare benefits, it is simply not rational to go to work–though this is not at all to say the RMI is too high. To these calculations should be added what may be called status-shift costs. In effect, since most available jobs are not permanent, actors tend to assume they will have to go back on the RMI, and this implies a long series of efforts and humiliations : putting together a new application, standing in line at social service counters, signing up at the ANPE, telling one’s life story to several social workers (again), waiting several weeks or months before new services are delivered, and so forth (Dubois, 1999; Rivard and Thalineau, 1999). Who wouldn’t hesitate before undertaking in earnest to look for a job (with no guarantees of finding one), a job whose drawbacks could in the long term prove greater than its advantages ? In this connection we might mention bankrupt craftspersons, legally obliged to start paying off their debts as soon as they find work. They go on the RMI to protect themselves from this, while readily finding demand for their skills “under the table”.

The inactivity trap model is based more on the notions of employment and income than work content. With narrow rationality, it is as if there were nothing but jobs, wages, and abstract “work” without any social or subjective content. Real individuals are not focused exclusively on getting offered a job; they are also concerned about what the work will be. And often the work offered seems unacceptable : the “shit jobs” that no one wants, exhausting, unstable, “vile”, the “dirty work”, unpaid overtime, activities with no job security or career prospects. Cashier work, for example, where one has to remain available for the four hours between one’s two half-shifts–and isn’t paid for the time. The fact that there is little work to be had does not mean this kind of work has disappeared; on the contrary, it is likely to be the norm for work offered to the most fragile and least skilled. Often it does not seem acceptable to lose one’s life, as it were, in order to earn one’s living. Income differential cannot justify all in the eyes of persons reluctant to accept as their fate a kind of work where the dominant norms and models are themselves a constant affirmation that this is unacceptable work. Individuals with educational degrees who dread losing social status are wounded in their sense of pride when offered such work, and this is understandable when we know that working in unskilled jobs could put a definitive blot on their CVs. This reaction can also represent a long-term rational choice (Iribarne, 1990). The main objection that may be made against econometric models is that they reflect no knowledge of either job-seekers or work content, whereas sociologists reason in terms of socially defined individuals and precise offers of work, not just jobs.

Reluctance to accept just any job is especially strong given that, thanks to under-the-table work, many RMI recipients are in fact not unoccupied. Once again, no one seriously believes that most recipients live exclusively on welfare subsidies. Between the poorest, who have nothing other than welfare aid, and those for whom that aid is merely an income supplement to underthetable earnings, there exists a whole range of cases of “getting by” on chance opportunities : services rendered, overall resourcefulness, one-shot jobs. Of the ten recipients who admitted to working and getting paid under the table, eight chose not to work legally, whereas the statistical model constructed from the files indicates that it would be in these persons’ utilitarian interests to get regular wage-paying jobs. In the hundred or so interviews we conducted it was often said–of others, very rarely of oneself– that working under the table offers recipients a threefold advantage. The first is financial. Some say the RMI is a mere legal façade, that it is very much in their interest not to chose legal work, even supposing such work could be found (the idea being that the only reason there is under-the-table work is that it is under-the-table). Even if these cases are far from the rule, why give up monthly income of more than 8,000 francs (1,220 € ) plus the RMI and various other welfare benefits for a minimum wage of 5,500 francs [838 € ] net ? The second reason is social. Underthetable work implies being fully socialized in networks for it–in this it differs from legal employment. To work under the table one has to be integrated into such networks and keep up membership. Lastly and most importantly, individuals’ discourse on working under the table almost always turns on assertions of freedom and skill. Such work enables one to turn down “really lousy job” offers; reputations get built; occupational skills get maintained or developed, and so on. We are not dismissing the collective cost of black market labor, nor the exploitation and insecurity associated with it, when we say that such work may enable people to escape certain aspects of the above-board wage-earner’s condition and that it sometimes hooks up with old dreams of worker autonomy and free choice of labor.

It is time to mention those who stay on the RMI because they’re used to it or because it’s a way of continuing to make headway on a personal project. The “culture of poverty” theme is ideologically highly suspect because it can be a scholarly way of designating what is assumed to be an acquired inclination for dependency and inactivity. [11] Without accepting such extremes, we must still conclude that RMI recipients living in districts of concentrated unemployment and large numbers of RMI and other welfare recipients, may well feel that this situation is normal : “All my [girl] friends are unemployed and on the RMI.” In these cases the actors have learned to survive, to get by on their wits, to know their rights, to negotiate them with social workers, to be taken in charge, and in doing so to become, paradoxically, more autonomous. This is where all the little varieties of fraud that enable one to get the best out of the system come in; lying about whether one lives in a couple, about rent figures and housing status, about resources from the Centre Communal d’Aide Sociale (CCAS) [municipal rather than departmental-level welfare assistance], and so forth. Such highly rational use of the RMI is optimized among those whose life project involves living on the margins, such as artists, or those who are part of a culture in itself marginal, such as the Tziganes. The former may well live with a wage-earning partner, receive financial and other assistance from family, and collect their envelopes for work done under the table. Members of the latter group tend to think of the RMI as “good fortune” that need have no effect on their way of life. Here again, we on the outside can swell with indignation at the RMI being diverted from its stated vocation. Still, neither of these strategies was created by the RMI and nothing suggests that an end to unemployment would weaken artistic vocations or the long-enduring Tzigane communities.

We are not going to deny the general proposition that those who continue receiving the RMI have an interest in doing so, but once again, that proposition is so general that it tells us very little, since the “interests” in question are so multiple and in some cases so contradictory that they ultimately “explain” extremely diverse, and even opposed, behaviors–that of the single mother who could only go to work if she abandoned a major part of her family role, but also of the bohemian who has chosen to live “dangerously”.

The discourse of recipients who have obtained work, often only temporary or limited-time contract jobs, sounds like one of redemption. In their solitude, humiliation, poverty, and despair, they say, they had truly touched bottom. If in that situation they declare that they were ready to “do anything” to get out, this was because they were doing quite everything possible to get back into employment–training programs, state-subsidized jobs, temporary work– and devoting considerable energy to seeking work, using all available resources of any kind. They had abandoned self-pride and accepted losing social status in order to become better integrated at a lower level; they could “no longer bear being out of work”. In all these salvation accounts, reestablished contact with others plays a crucial role, because these persons could no longer bear what other people might think of them, especially family and children. In young couples the situation can also have dire effects on desire to have children. They’d met or run into a friend, relative, social worker, colleague in a CES or odd job. In sum, they went off the RMI because the stigma of inactivity was unbearable.

It should be noted that these actors’ discourse is not at all focused on the immediate economic interest of getting work; they hardly mention financial earnings or material advantage. On the other hand, all insist on the recovered sense of dignity that comes from no longer owing anyone anything. All say they’ve escaped the narrow sociability of poor persons and café pals, that they’ve become “like everybody else”, that the rhythm of their lives is once again set by work. And even if they feel exploited and tired in the evening, it’s another kind of fatigue than the crushing feelings of lonely apathy induced by inactivity. These observations confirm the results of a recent study (Afsa and Guillemot, 1999) showing that despite low remuneration (never much higher than minimum wage), 82% of former RMI recipients say they feel better since getting back into employment. For them, work is an essential vector of integration, dignity, and autonomy, and in this it is priceless. But they speak much less of the work, the activity in itself, than the sociability that comes with it : colleagues, recovered pride, what may be called a work ethic that enables them to “look themselves in the face again” because now others–children, family, neighbors, colleagues– can look them in the face. Above all, with work comes rights; one is no longer in the position of a beggar, of the beneficiary who can never repay the debt because there is no countergift for the help received beyond demonstrating a certain will to change one’s situation.

We should not be surprised to see how important the role of such moral concerns is : these are people who went off the RMI when it was not “objectively” in their interest to do so; all they had were their subjective imperatives. We should, however, be impressed by the power of those imperatives when we consider current clichés about the decline of “work ideology”, the trend toward letting oneself be assisted, the irreversibility of marginal status. This moral force, this capacity be a subject, is rooted in individual histories and personalities that cannot be satisfactorily accounted for in terms of narrow rationality and expected financial gain. We may, however, identify three central characteristics. The first is that being an RMI recipient seems particularly unbearable to people who have already held long-term stable employment, who have forged themselves an identity of “person who works”. Even when they hit bottom, they cannot really accept to forego work. The second has to do with the fact that the RMI’s perceived role can vary by life period. For young people, students, bohemians, the RMI is connected to a particular life period and assimilated to a kind of student financial aid package that prolongs the moratorium of youth. This representation ceases to be when one reaches a certain age, decides to leave home, start a family, when one falls in love. Finally, while all these individuals speak of courage and energy, they also speak of chance, being in the right place at the right time, coming upon a “good social worker”, getting a CES where one isn’t “treated like shit”, and so forth.

In some cases, then, the meaning of the RMI becomes completely transformed–or rather, it comes to correspond to the original philosophy behind the policy. People use their time on the RMI to acquire training, look full time for work, set up a business without risking catastrophe, thanks to the financial aid. But this success would seem due more to the individual’s qualities–one of which is knowing how to grasp the opportunities offered by the program– than to the program itself. The RMI is what recipient-actors make of it.

We cannot declare that those who remain on the RMI when they “should” get off are utilitarians while those who get off when they “should” stay on are ethical. Rationality and ethics are not necessarily distinct levels of discourse and don’t have to be motivation types. [12] Good reasons are both multiple and “swing”; they partake of both getting off the RMI and staying on.

Our figures so thoroughly invalidate the narrow instrumental rationality model that it would be fair to put the result down to the cursoriness of information on individual recipients. By working on a richer, though non-represen-tative, subsample, we can push our reasoning further. Indeed, the information collected through interviews is so rich that we slide imperceptibly toward a completely different representation of action, that of behavior determined essentially by individual’s situation and attributes. Here the actor is less determined by any aim of satisfying his or her own interest as by objective causes that enable him or her to self-project towards satisfying one or another interest. But here, too, no factor is really determinant, except for belonging to a subgroup of destitute individuals so little able to choose for themselves that it is hard to speak of action. As we have seen, the most notable phenomenon is cause reversibility. The same social causes, understood in their relation to each individual’s particular situation, can have quite opposite effects, or at any rate, quite different ones. The fact of being without children or educational degree, for example, can lead an individual to go off the RMI or encourage him or her to stay on. Causes that move people to go off are also causes that move them to stay on.

In fact, if we start with opposite anthropological postulates from opposite theoretical worlds, the determinist conception of action ends up running into precisely the same difficulties as narrow instrumental rationalism. Just as in RCI, anthropology’s version of utilitarianism is postulated a posteriori, so “habitus” is deduced from statistically regular behaviors interpreted, for our case, thus : individuals remain on the RMI because it’s in their interest to do so or because they have been programmed to. From both these perspectives, there is really no point in asking individuals what they think.

2. At the end of this “reduction” in the sense explained above, it thus seems reasonable to choose the subjective rationality model and conceive of action in terms of “good reasons” (Boudon, 1986,1995). [13] Reason, with a capital r, in fact takes the form of multiple subjective reasons defined by actors as so many instances of social and cultural utility. There are many advantages to this choice. It is based on economical, elegant reasoning that puts forward no strong hypotheses on blindly determined actors’ rationality. It mobilizes a kind of common-sense verstehen sociology in that the researcher can readily follow and share in the “good reasons” actors cite. In all the interviews conducted for this study, we as researchers could readily put ourselves “in the place” of the individuals we met and see for each case that, if we had those particular cultural orientations and definitions of the situation, we would probably do the same thing–out of neither cynicism or moralism. Moreover, when sociologists have more detailed information than actors, namely about the nature of the job market, they readily understand that the actor’s point of view is rational given his or her vantage point and the perception it gives him or her of his or her chances. They understand both the “bohemians” who use the RMI to realize a life-project and the Tziganes who use it to consolidate their lifestyle and culture, as well as the energy of individuals whose strongest desire is to escape the fate of being an RMI recipient, and the despair of others, fully aware that it is impossible for them to go off. They understand, too, the mixed nature of social work, which may both stigmatize recipients and mobilize considerable amounts of will in them. In sum, from the perspective of subjective rationalism, one man’s or woman’s dynamism is just as rational as another’s apathy.

The strength of this perspective is that it proposes a verstehen hermeneutics with strong descriptive power that duly takes into account the developed thinking of actors rather than smashing or invalidating their explanations with an interpretive grid. It also has the non-negligeable advantage of suggesting reasonable social policies, ones that take into account the diversity of individuals’ situations and the different types of logic they act upon. However, even supposing we went so far as to propose a typology of reasoning and experience, the explanatory capability of the good reasons model remains slim, because we can never know what kind of tie exists between the good reasons actors cite and their actual behavior. And since good reasons are convincing and plausible in any case, this approach threatens to create a “Pangloss effect” [14], just the sort of effect that proponents of the “good reasons” model regularly denounce in the rival model, holding as it does that behaviors are programmed by the habitus.

3. In order for subjective rationality to be more than the product of a verstehen perspective applied to what individuals say, we would have to go beyond description and bringing to light cognitive processes. In fact, we would need to identify and distinguish between several types of subjective reality, pertaining to several features in actors’ situations. When defining a situation in terms of personal resources, we would have to connect those resources to their social distribution structure and the objective possibilities individuals have. Likewise, we would need to be able to take into account the characteristics of an individual’s integration and identity. Individuals do not simply follow their best interests; they put themselves on the line–their image of themselves, their identities, the conflict between self-image and the image others have of them–as is attested to by their experience of encounters with social workers. To the good reasons of instrumental rationality we would have to add, and often oppose, good ethical reasons, those that call into play the subject’s relation to himself or herself. How are representations of one’s own resources constructed ? How is an image of one’s own resources grasped ? How might identities, the ways individuals internalize or reject stigmata, in ternalize or abandon aspirations and the realizing of life projects, be modelled ? Our distinction between two orders of good reasons is related to the fact that we have as yet learned nothing to suggest that there is strong congruence between objective resources and identities. If there were such congruence, the inactivity trap model would not produce so many “aberrant” cases. Above all, we would not understand why some of those “cases” go off the RMI without benefiting economically from doing so.

Yet even if we could combine these two dimensions, we would surely not obtain a satisfactory predictive and explanatory model, because the fact remains that in comparable situations some “decide” to go off the RMI and others don’t, while still others let themselves be carried by the system. We are thus confronted with a kind of “subject’s viewpoint” that poses real problem for determinist and “good reasons” sociologists alike. Why do some individuals mobilize considerable energy to go off welfare while others don’t ? The point of view constructed by the subject would itself generate good reasons, each of which is just as coherent and convincing as the next, but the subject, with his or her capabilities to act, is there “before” the good reasons he or she may cite. Why do some actors seem “destroyed” by the RMI while others resist ? What do some follow a kind of cursory adaptation logic, while others get a grip on themselves ? [15] The “personality” construct to which many refer is only another name for the problem, a translation of it into another language, not a solution. Because RMI recipients are often at the very bottom, they are, paradoxically, put through an ordeal as subjects, as individuals endowed, in most cases, with the capability of constructing themselves as authors of their own lives. From this point, “good reasons” sociology has to become moral sociology, a sociology of good ethical reasons “to be or not to be” a subject. Our gradual reduction of rationality opens onto the sociological problem of freedom. It is because the RMI often has the effect of breaking down freedom that this question once again arises. Above and beyond the resources that certain individuals have at their disposal, above and beyond their interpretations of their situations, some are as it were invaded by a sense of their roles as RMI recipient, whereas others resist those roles and turn them on their head. In this sense, they are like activists, even if first and foremost activists in the cause of their selves. This does not mean we should abandon rational action for who knows what arbitrary mysteries of the human soul. But we do have to remember that in each case a subject is interpreting a situation and interpreting the resources he or she disposes of, even if it is true that he or she cannot take just any action.

These themes are not readily admitted by sociologists; the fundamental sociological critique of humanism and Sartrian philosophy make us see them first and foremost as metaphysical questions. But it is too easy to refuse to consider them on these grounds. In fact, they are first and foremost methodological questions, and “reducing” rationality is first and foremost a matter of method. We have yet to construct a method capable of grasping the subject as subject. This study has not done so, nor even sketched a way of doing so. Rather, having progressively “reduced” the hypothesis here, we call for developing such a method. [16]

Interest in the diversity of itineraries and contexts for action, the subtlety of individuals’ “calculations”, and the very possibility of action that individuals have should not lead us to dissolve sociological study in the technique of analyzing more or less ineffable and irreducible “cases”. It remains essential to look for regularities, if not “laws”. What we need to do is reason on the basis of associated factors, mechanisms that shape contexts for opportunities, the ways in which aspirations are developed and acted on–rather than in terms of individuals. This is because singular contexts can be described as multiple combinations of various “causes”, but above all because individuals “circulate” and “evolve”. Itineraries and trajectories give meaning to situations and contexts that appear more like tests or trials to which actors must respond than situations involving sets of causes. Regularity is more to be found in these tests than in situations perceived as aggregations of “causes” for action. Sociologists should then ask how these tests or trials are constructed and how the individuals who must confront them are constituted.

Our “reduction” of the instrumental rationality model remains dependent on the methodological conditions of the exercise. We moved from quantitative analysis of over 20,000 RMI recipient files to interviews with around a hundred recipients because the data available were too meager to allow for more elaborate statistical analysis. The move from instrumental rationality to the universe of good reasons is as much a methodological necessity as a theoretical requirement. Still, whatever the models and methods, it seems unquestionable that financial gain expected in going from welfare to employment does not govern all decisions; it is only one variable among others, and not at all the most influential one. The space of rationality thus opens up onto that of good reasons and subjective rationality. But what we gain in interpretative, hermeneutic capability we lose in explanatory capacity, since recompositions of subjective rationality seem infinitely diverse. Constructing types would be more likely to give researchers a feeling of security than to actually improve our ability to explain. What must be explained is subject’s own capacity for action. Or, if one prefers such terminology, we must develop a sociology of good ethical reasons.

This conclusion, or absence of firm conclusion, at least shows that a social welfare policy such as the RMI cannot be fully grasped or evaluated as economic policy, if only because recipients are so heterogeneous. Effective economic policy for decreasing unemployment would dictate the ultimate abolition of the RMI to ensure that employment-dependent gain always win out. This would mean making the poor poorer so they could ultimately stop being poor–or so that they could become poor working people instead. But the benefits thus accrued might well engender considerable costs, not to mention the political and moral meaning of such a decision.

Need we be satisfied, however, with the RMI as it is ? Our study seems to suggest that welfare systems are more apt to protect individuals excluded from the legal job market than to help those same individuals become independent. But given that our study shows utilitarianism is not the rule, we would still have to devise welfare systems that don’t penalize persons who want to go off the RMI by making them pay the costs of moving from welfare into employment and undergo the risks incurred by trying to go off. It seems obvious that we should further develop sliding-scale unemployment benefits (Belorgey, 2000; Godino, 1999). Furthermore, individuals who go off the RMI are not only facing the difficulties of the job market; they also take risks that nothing but the weight of the welfare system itself can explain. In saying this we are not calling into question the soundness or utility of these policies. Let us underline two major obstacles to getting off the RMI and into work. First, the complexity of existing welfare programs and the procedures they involve. This gives rise to extremely high “conversion” costs and accentuates individuals’ dependence, since they can’t get a handle on welfare programs without the help of specialized intermediaries – ie, social workers. The complexity facilitates rule-breaking, slippage in applying policies and granting aid –defects that adroit individuals can use to their advantage. Finally, as this study shows, it would be simplistic to reason in terms of an opposition between unemployment and employment. Individuals don’t simply choose to have a job and income; they also choose the content of the work. And as we know, RMI recipients are often offered the most precarious, unpleasant, devalued jobs. They are in the front line when it comes to the worsening working conditions that characterize many jobs available to the least skilled today (Paugam, 2000). Getting back into work brings with it the problem of working conditions, a problem somewhat neglected in times of chronic unemployment. Wage-paid employment is not all it takes to construct a social actor; the content of the work plays a major role, if not a greater one. If we want to help individuals get back into work, that work has to be acceptable. If it is not, why should actors choose to quit the ranks of one subproletariat only to join another ?

Translation: Amy Jacobs

Previously published: RFS, 2001, 42, 3

TABLE V.
Logistic regression on variables

It is useful to look at the relative contribution of each variable to goodnessoffit between model and empirical data. The regression is therefore constructed by introducing the variables one by one in ascending order of their “predictive capacity” according to the model (this is based on “maximum likelihood”). In the table, they are presented in descending order of contribution to the model. The R² column shows how the increase in overall R² varies when another variable is introduced. Each dependent variable’s contribution in explaining the independent variable (in this case exit from the RMI) is determined by the partial correlation coefficient R, which varies from–1 to +1. Given R values and increase in R², it should be noted that the “income differential” variable, which is at the core of the utilitarian reasoning behind inactivity traps, contributes least. When the “expected income differential” variable is added to the preceding ones, R² only increases by 0.001, from 0.135 to 0.136. As was the case for the data in Table IV, income differential counts for less than the aggregation of variables concerning individual’s personal characteristics, while these variables themselves, or what we know of them here, weigh very little on the set of “causes”.

Logistic regression analysis using the supplementary variables of sex and nationality shows neither of them playing a significant role. Age, on the other hand, is a relevant element in likelihood of going off the RMI. The fact of being in a couple without children increases likelihood, while that of living in a poverty pocket confines one to welfare. This last point supports the hypothesis that there are indeed welfare ghettoes. The inactivity trap model can be amended, but it can never be fully convincing, namely because the financial gain individuals expect is not decisive for going off the RMI. There is nothing original in this anti-utilitarian-rationality conclusion, but it’s worth stressing, if only to help dismantle the widespread notion that poor people don’t work because it’s not in their interest to do so.