Tải bản đầy đủ - 0trang
Quantifying the effects of public action on the unemployed: disputes between experts and the rethinking of labour market policies in France (1980-2000)
of public assistance with a control sample (Gaudillère 2006), had been used for several
decades in the USA and in a number of European countries. However, they were late
appearing in France (Labrousse 2010), not coming into use until the second half of the 2000s
in order to evaluate the introduction of the French income support benefit known as the
revenu de solidarité active (RSA) (Gomel and Serverin 2009) and in order to evaluate the
performance of the Agence nationale pour l’emploi (the French national employment agency,
or ANPE) compared to private agencies in the competition to find placements for job seekers.
The speeches given by advocates of these methods described them as an obvious advance,
one that marked the introduction in France of evidence-based policy (Duflo and Banerjee
2011). Nevertheless, questions continue to be asked about the ‘reliability’ and ‘robustness’ of
these methods (Perez 2000; Allègre 2008, L’Horty and Petit 2010). Their scientific nature and
value, far from being established facts, are the object of disputes between various actors.
While experimental methods were adopted late in France, experts in the ministries in charge
of these policies, particularly at the Ministry of Labour, had been developing original survey
methods since the 1980s. Thus reviewing the political and bureaucratic issues at stake in these
surveys is a way of understanding how and by whom they were imported and diffused, and
for what purposes. The use of random surveys in the second half of the 2000s cannot be
reduced to a mere technical change or a process of scientific rationalisation applied to the
evaluation of public policies. Rather, the introduction of such methods led to a real change in
the conception and role of evaluation by removing it from the purview of the social sciences
and relocating it firmly within the sphere of the so-called ‘hard’ sciences. This process was
part of the battles between groups of experts and researchers and of the changes in state
intervention in labour market policy, whose role henceforth was to be less to correct
inequalities in the labour market than to help to make it function more flexibly.
1. When the social sciences seized control of labour market policy evaluation
1.1. Measuring the effects of labour market policies: the innovation of sample surveys
Throughout the 1970s and until the mid-1980s, assessments of labour market policies
were dependent on management and administrative practices and knowledge. They were
based on data put together by government departments (e.g. the flows of state-subsidised
contracts) and on so-called ‘operational’ surveys that sought to describe the conditions under
which policies were implemented. It was not until the second half of the 1980s that the
Ministry of Labour’s department of studies and statistics (SES), which in 1993 became the
Directorate of Research, Studies and Statistics (Direction de l’animation de la recherche, des
études et des statistiques/DARES), began to conduct surveys using samples of unemployed
individuals in state-subsidised jobs, on the one hand, and, on the other, samples of firms
making use of such jobs. Compared with the administrative data that had been available until
then, these sample surveys increased the volume of information that could be used by
government departments in order to measure the various effects of labour market policies. For
example, the surveys of company directors were to be used to measure the deadweight effects
(when an unemployed individual is hired on a state-subsidised contract when a worker would
have been hired in any event) or substitution effects (when a state-subsidised worker is hired
7 Quantifying the effects of public action on the unemployed ...
rather than one on a standard contract) produced by these policies. Similarly, a major
innovation involved the introduction of panel surveys carried out using representative samples
of unemployed people and/or beneficiaries of labour market policies. Using this method, it
became possible to objectify several types of effects that could be observed along the labour
market trajectories of the unemployed: return to short or long-term employment, as well as
the quality of any job found, access to training and so on. Statistical analysis of the panel data
gave rise to typologies of labour market trajectories and to descriptions of these trajectories
based on the social characteristics of the unemployed individuals surveyed. The officials
responsible for the surveys usually used multiple correspondence analyses and classifications
in order to draw up these typologies. The social characteristics of beneficiaries of labour
market policies were taken into account in order to ascertain whether the subsidies were
actually being allocated to the ‘target’ populations – young people with few qualifications, the
long-term unemployed etc. – and to measure the counter-selective effect or, to put it another
way, their ability to invert the employment queue by favouring the most disadvantaged. Thus
these surveys were used to promote a socio-economic approach, since they were concerned
with the principles animating the actors’ trajectories and took account of the various resources
(educational, social, etc.) available to them and of the labour market segments in which they
had to find employment.
These innovations were an attempt to respond to political and administrative injunctions
concerning labour market policy evaluation. Statisticians and experts at DARES did not
appropriate them uncritically or without seeking some degree of leeway for themselves.
Firstly, they were critical of what they regarded as the overreliance among politicians and
civil servants on the return-to-work rate as the sole indicator for measuring the success of
labour market policies (DARES 1996: 303 and 317). They insisted on the need to vary the
evaluation criteria by taking account of indicators of earnings levels, job quality and social
integration. They were also cautious about and, in some cases, doubtful of the feasibility and
even the value of a ceteris paribus measure of the effects of public action (Gélot and Simonin
1996; Aucouturier, 1994). In panel surveys, after all, the beneficiaries of labour market
policies are compared with unemployed individuals who have not received assistance in order
to isolate the effects of state assistance, on the assumption of ‘all other things being equal’.
This reluctance to adopt experimental methods was linked to the technical challenges they
pose and the fact that, because of their academic and professional backgrounds, some of the
officials in charge of the surveys lacked proficiency in econometrics (cf. below). They also
advanced a moral and political argument, namely that a conscious decision not to help all the
unemployed individuals potentially eligible for assistance would contravene the republican
principle of equality (Gélot and Simonin, 1996). Finally, some of them took the view that the
legitimacy of state intervention in the labour market was not an issue, since the state is not an
exogenous variable but rather one actor among others in the labour market (Aucouturier,
This attitude to experimental methods went more or less unchallenged in both
administrative and academic circles. Among university economists, the development of the
new neoclassical microeconomics of labour markets was in its early stages and was still based
mainly on theoretical approaches (Perrot 1992). The first econometric studies using empirical
data did not become available until the early 1990s. Moreover, unlike in the USA, labour
economics based on the institutionalist paradigm still had a structural role in academic circles
(Gazier 1992). Among the economists at INSEE, the French national statistical service,
macroeconomics remained dominant and the emphasis was on analysis of economic and
budgetary policies, with labour market policy still a marginal concern.
1.2. Institutionalization of evaluation of public policies and opposition to the neo-liberal
approach of the French Ministry of Finance
On their introduction, these statistical surveys became a resource in the bureaucratic
struggles of the late 1980s and early 1990s. For government research analysts, they were a
means of responding to injunctions from government and the ministries of state that were
seeking to institutionalise public policy evaluation as part of a broader programme for
‘modernising’ the state (Spenlehauer 1998). Analysts at DARES championed an academic
approach to evaluation and regularly opposed the ‘operational’ departments of ANPE, the
French national employment agency, and the Ministry of Labour, which were in the habit of
using the administrative evaluation methods mentioned previously. Thus the new survey
methods enabled these research analysts to legitimise and build up their institutional position
within the bureaucratic space.
These surveys were to be used above all by senior civil servants in the Ministry of
Labour to oppose the neoliberal policies advocated by the Ministry of Finance. As
unemployment in France increased, reaching 3 million in 1993, labour market policies
became the object of much debate within government and the civil service. Keynesian
demand-side policies had been discredited since 1983. The economists in the Ministry of
Finance, who had been converted to neoliberalism (Jobert and Théret 1994), were advocating
measures designed to combat labour market rigidities by making the regulations on dismissal
protection more flexible and reducing labour costs for unskilled jobs. In political terms, their
advice was reflected in a series of measures to reduce social security contributions that were
introduced in 1993, when a right-wing government was in power. For their part, senior civil
servants in the Ministry of Labour advocated labour market policies intended to support the
most disadvantaged groups among the unemployed in order to compensate for the labour
market’s selective effects (inequalities of access to employment) and ensure that social
measures were put in place for these groups. They embodied the ‘left hand’ of the state
(Bourdieu, 1998), which took the view that economic objectives should be linked to measures
to improve social justice and ensure that individual well-being was taken into account in
developing policies (Mathiot 2001).
As these institutional struggles raged, senior civil servants in both ministries drew on
statistical and economic appraisals in order to give legitimacy to their respective ‘doctrines’.
The economists in the Ministry of Finance’s forecasting department encouraged the
development of micro-econometric studies and used them to criticise the measures introduced
by the Ministry of Labour, particularly subsidised jobs. They noted, in particular, that the
‘net’ effects of these policies on the back-to-work rate in the market sector were in fact weak
(Tresmontant and Ermakoff 1990). Conversely, drawing on their own surveys, civil servants
in the Ministry of Labour showed that, despite everything, these measures were succeeding in
7 Quantifying the effects of public action on the unemployed ...
reducing unemployment – which was an important argument in a context in which this
problem was highly politicised – and also preventing unemployed individuals with few
qualifications who were no longer able to find ‘normal’ jobs from becoming socially excluded
(DARES 1996). The data from these surveys were also used to support the Ministry’s
proposals on access for the unemployed to training or less precarious subsidised jobs. Among
most of the senior officials and analysts at DARES, this defence of the social management of
unemployment went hand in hand with the promotion both of Keynesian macroeconomic
policies and the sharing out of working time (DARES 1996 : 256).
1.3. Statisticians and experts defending heterodox knowledge
These surveys were after all designed by experts who, by virtue of their career
trajectories and position within their academic field, championed a certain view of statistical
surveys and labour market policies. They were sociologists or economists who subscribed to
non-mainstream schools in economics (Lebaron 2000), notably Keynesian, Marxist or
institutionalist approaches. They collaborated frequently with their counterparts in the
Observatoire franỗais des conjonctures ộconomiques (French Economic Observatory, or
OFCE), a Keynesian economic forecasting institute, and in the Séminaire d’Economie du
Travail (a CNRS research centre at the University of Paris 1), a bastion of institutionalist
economists. There were also INSEE statisticians working at DARES who had been trained in
the sociological approaches in the 1970s, when Pierre Bourdieu and his team were teaching
that discipline there, and who had cut their professional teeth in the INSEE departments
responsible for gathering social statistics. Despite differences in their career trajectories, these
experts belonged to the same groups within their respective academic and administrative
spheres and had contributed to the process, which had begun in the 1970s, of making the
social sciences part of the body of knowledge on which the state could draw (Bezès et al.
2005). They took a position against neoliberal economic policies. Moreover, while they
championed a strongly empirical approach to evaluation and were for the most part skilled in
the use of descriptive statistics, they were not econometricians.
This background, and in particular the fact that the methods used in labour market
policy evaluation had been developed by experts in the social sciences, explains France’s
relative distinctiveness in this regard. In the United States, for example, experimental surveys
and, more broadly, microeconometric methods of policy evaluation had predominated since
the 1960s and in the UK and the Nordic countries since the 1970s/80s (Fougère 2000). Thus
the introduction of these statistical surveys as a means of evaluating state action involved a
series of problematisations and choices that reflected certain notions of what state action
should be. The political issues at stake in policy evaluations were all the more significant
since they were part of the broader institutional struggles whose ultimate goal was to define
the forms that state intervention should take (Bourdieu and Christin 1990).
2. The imposition of econometrics and neoliberalism in the first decade of the 21st
2.1.Experimental methods to measure the ‘pure’ effect of public action on the back-towork rate
In the first decade of the 21st century, the evaluations carried out by DARES underwent
some significant changes. The types of survey used became less diverse and the surveys that
were carried out were designed to isolate the net effects of public action and to measure the
‘back-to-work’ rate, an indicator that had been used but also criticised in DARES studies
carried out in the 1990s. The methodological effort was focused principally on the techniques
used to select a control sample that could be used to analyse the effects of the measures
ceteris paribus (Even 2002). Experimental methods were then imported and developed under
the umbrella of a partnership between DARES, ANPE, INSEE’s Centre for Research in
Economics and Statistics (Centre de recherche en économie et statistique/CREST) and the
Paris School of Economics (l’École d’économie de Paris). The aim was to evaluate the effects
of the increased support for unemployed people put in place in 2007 by ANPE and the private
placement companies funded by the National Interprofessional Union for Employment in
Industry and Commerce (Union nationale interprofessionnelle pour l’emploi dans l’industrie
et le commerce/UNEDIC) (DARES et al. 2009 ; Behagel, Crépon and Gurgand 2012).
The experimental method (i.e. random or randomised evaluation) involves the random
selection of a group of users who will receive aid and support in order to compare their
trajectories with those of users not receiving the aid and support provided to the other group.
Thus the survey method is still based on the monitoring of a panel of unemployed individuals,
but the link between quantification and access to support is reversed. The aim is no longer to
survey unemployed people, who have already received aid and support, i.e. to adapt the
methodology to the population surveyed, but rather to adapt the allocation of the population to
the survey protocol. This type of survey is supposed to ensure the best possible degree of
comparability between the population that receives support and the control population, since
the random allocation neutralises selection biases among the individuals surveyed (what
econometricians call ‘unobserved heterogeneity’), thereby enabling analysts to evaluate,
ceteris paribus, the effect of the measures taken by the ANPE and the private placement
agencies on the back-to-work rate. The technical and administrative legitimacy accorded to
this method was put before the public at several events, notably at an international conference
entitled ‘Expérimentations pour les politiques publiques de l’emploi et de la formation
professionnelle’/The Use of Experimental Methods in the Evaluation of Public Employment
and Vocational Training Policies’, organised by DARES in May 2008. The main foreign
economists internationally recognised as promoters of these methods, such as Abhijit
Banerjee, professor at MIT, and Esther Duflo, holder of the chair in ‘knowledge against
poverty’ at the Collège de France, were invited to take part. Alongside them were the French
economists who had imported their work: Marc Gurgand, CNRS research director, Franỗois
Bourguignon, director of studies at the EHESS (Ecole des hautes etudes en sciences sociales),
both members of the Paris School of Economics, Denis Fougère, CNRS research director, and
Bruno Crépon, a senior civil servant at INSEE, both members of CREST (the INSEE research
It might be thought that this use of randomised evaluation essentially constitutes a
development of the survey methods put in place at the end of the 1980s. This is not the case.
7 Quantifying the effects of public action on the unemployed ...
This technical change is in fact linked to changes in the discipline of economics and in
government thinking on labour market policy.
2.2.The import of experimental surveys into France by a new generation of economists
In order to understand the situation at the turn of the century, it is necessary to consider
the changes taking place contemporaneously in the discipline of economics. Focusing the
investigations on the ceteris paribus effect of public measures was, after all, a way of
adapting the evaluation methods to the changes that had been taking place in economics since
the second half of the 1990s, in which the various groups in the dominant school were seeking
to bring economics closer to the experimental sciences by making the discipline an empirical
science capable of identifying causal relationships (Lordon 1997). Distancing themselves
from the ‘Washington consensus’ (Dezalay and Garth 1998) and a highly theoretical approach
to economics, some of the new generation of economists were adopting a highly pragmatic
approach. They did not proclaim themselves followers of the neoclassical school, while at the
same time retaining the individual rational actor paradigm, and also distanced themselves
from the systematic discrediting of the state. They emphasised empiricism and the need to
respond to questions about the efficiency of state action (Labrousse 2010). This was reflected
in the exponential use of microeconomic data and econometric panel techniques, specifically
in the area of labour economics (Hsiao 2006). This shift in the discipline was also a factor in
the changes that were taking place in the hierarchies within economics at the international
level. Empirical studies, which had long been considered less worthy than theoretical studies,
were now regarded particularly highly (Han Kim et al. 2006). Since the canons of French
economics were derived largely from those in the English-speaking countries (Lebaron 2000:
131-132), these changes in the discipline found an echo in the upper echelons of French
economics. It was the French labour economists closest to these dominant groups who
imported these methods in the early 2000s. Thus the effort expended on developing these
methods cannot be explained solely by changes in the demand from politicians for evaluation,
but was also linked to specific issues within the academic discipline.
In the 1980s, panel surveys were carried out by research analysts trained in social
sciences. In the 2000s, those developing these methods had different profiles. Members of the
most prestigious research centres (INSEE research centres, Paris School of Economics), these
economists had several characteristics in common: they had built up their positions in the
academic sphere rather than in the administration and, above all, were oriented more to
mathematics than to the social sciences, both in their training and in their approach to
economics. Moreover, although they defined themselves primarily by the primacy they gave
to empiricism and to econometric methods, they championed more liberal labour market
policies, in which the unemployed were regarded above all as homines economici responding
to economic incentives and no longer as individuals belonging to social groups whose
situation was the result of labour market dysfunctions. After all, the imposition of
econometrics and randomised evaluation went hand in hand with a shift in the dominant
theories in labour economics: these economists highlighted the mismatch between the skills
supplied and those demanded in the labour market, on the one hand, and, on the other, the
inadequacy of earnings from work (L’Horty 2006). Unemployment was explained principally
by the behaviour of unemployed individuals rather than by the macroeconomic context or
agents’ socio-demographic characteristics.
From the beginning of the 21st century onwards, these economists embarked on the task
of critiquing the DARES surveys in order better to legitimise the experimental methods that
had been used for a long time in the English-speaking and Nordic countries. To that end, they
began by emphasising the fact that ‘when it comes to unemployment, nothing has been
seriously tried because nothing has been seriously evaluated’ (Cahuc and Zylberberg 2005:
13), thereby indicating that the DARES surveys were incapable of measuring the ‘net effects’
of public action. In contrast, they emphasised France’s isolation as the only developed
country not to use experimental surveys, which were presented as the gold standard for
evaluation. They made extensive efforts to diffuse and popularise these approaches within
academia and, above all, within the civil service, through ministry journals (Fougère 2000;
Brodaty et al. 2007) and government reports (Pisani-Ferry 2000). This discourse on the
performance of public action and comparison with other countries gained all the more traction
since it very much chimed with more general ideas around the need for reform and for the
state to demonstrate its effectiveness (Bezès 2009; Bruno and Didier 2012). Finally, they were
able to take advantage of their academic qualifications, unlike the former DARES research
analysts, and of the symbolic capital of the economists advocating these methods, such as
Esther Duflo and James Heckman, professor of econometrics at the University of Chicago and
Nobel Prize laureate in economics in the year 2000. These different registers of legitimation
enabled them to discredit the studies carried out by the DARES experts and hence to
marginalise some of them within their own directorate. They were all the more successful
since they were able to find intermediaries within the civil service.
The downgrading of heterodox approaches in both academia and the civil service,
notably at the Paris Graduate School for Economics, Statistics and Finance (l’École nationale
de la statistique et de l’administration économique/ENSAE) in the 1990s (Lebaron 2000),
was reflected in a depletion of the breeding ground for the economists and sociologists who
had carried out the earlier surveys and led to the recruitment of a generation of young
econometricians. Moreover, from the early 2000s onwards, the recruitment channels for
senior posts within government agencies such as DARES were restructured in favour of
mainstream economists from the Ministry of Economics and Finance (Penissat 2009). The
INSEE officials who arrived at DARES in the 2000s brought about a change in the networks
of researchers with whom DARES collaborated. They built up alliances with econometricians
from INSEE and EEP, who had hitherto played little part in the networks of researchers
funded by DARES. The arrival of these new senior officials created a favourable environment
for the adoption of new survey methods and new economic paradigms.
2.3.Rethinking anti-unemployment policies and the uses of econometric evaluations
The political and institutional configuration at the turn of the century made this
endeavour possible. Thus the labour economists at the Paris School of Economics (Marc
Gurgand, Luc Behagel, Pierre Cahuc, etc.) and INSEE (Bruno Crépon, Roland Rathelot,
Thomas Le Barbanchon, etc.) placed their expertise at the service of institutions enmeshed in
highly competitive configurations or positioned as outsiders within the administrative
7 Quantifying the effects of public action on the unemployed ...
bureaucracy. This was the case when the RSA was introduced. The High Commission on
Active Solidarity against Poverty, headed by Martin Hirsch36, which was a subordinate and
peripheral organisation compared with the Ministries of Health or Labour, was able to take
advantage of the experimentation and its results to extend the RSA to the whole of France
(Gomel and Serverin 2009). A similar process occurred in the case of the evaluation of the
increased support programme for job seekers. The randomised surveys were put to use when
competition was introduced between the ANPE and private placement agencies. The
offensive launched in the 2000s by employers within UNEDIC, the organisation managed by
the social partners responsible for the administration of unemployment benefit, and the
development of a policy combining increased support for the unemployed and stricter
monitoring of their job seeking put the ANPE under pressure (Barbier 2007). In particular, the
management of UNEDIC redefined its relationship with the ANPE by imposing
contractualisation, with future budgets being granted on the basis of the ANPE’S performance
in increasing the ‘back-to-work’ rate. In 2005, this contractualisation principle was further
reinforced by the UNEDIC management’s decision to experiment with competition between
the ANPE and private placement agencies in specific segments of the labour market. In the
course of this experiment, the ANPE and UNEDIC formed alliances with economists from
CREST and the EEP, who were seeking funding and opportunities to introduce randomisation
techniques. As a result, the reform launched in 2007, which involved partial privatisation of
placement services for job seekers, acquired academic legitimacy, with the experimental
surveys being presented as an ‘objective’, impartial means of deciding between the private
and public organisations (Cahuc and Kramarz 2004 : 48).
As these survey techniques were diffused, they were used to support new ways of
conceptualising and planning labour market policies. After the episode of the 35-hour week,
which was introduced between 1998 and 200237, a series of neoliberal prescriptions were put
forward by certain elected politicians, senior civil servants and employers’ representatives.
The activation principle became established as the dominant doctrine underlying social
expenditure. The debate shifted from the problem of the distribution of work and inequalities
of access to employment towards the question of labour market fluidity, approached from a
neoliberal perspective. The emphasis was less on providing targeted assistance for the most
disadvantaged among the unemployed than on putting in place positive (financial rewards for
promoting employment) and negative (stricter monitoring of job seeking) incentives to
encourage the unemployed back into work. As the institution responsible for matching supply
and demand in the labour market, the ANPE’s role and the actions it took emerged as one of
the issues at stake in the debate on public action and the evaluation thereof (Pisany-Ferri
The choice of questions and variables to be used for evaluation was linked less to the
survey tools themselves than to these experts’ predispositions and the ways in which they
conceived the labour market and state action. Since the beginning of the 21st century, they
On Martin Hirsch’s specific role in the importation of these methods, cf. Elisa Chelle, 2012.
Between 1997 and 2002, the socialist Jospin government introduced a major reform, encouraged in part by
some experts at DARES, with the aim of combating unemployment: it involved reducing the statutory weekly
working time from 39 to 35 hours.
had, after all, been promoting so-called ‘activation’ policies, which sought to strengthen the
monitoring of job seeking behaviour and to encourage labour market mobility. Drawing their
inspiration from the US workfare model, these policies involving the devolvement of
responsibility to the individual job seeker. For advocates of these reforms, the state’s function
was to ensure fluidity of adjustment between supply and demand in the labour market rather
than to remedy inequalities or alter the way the labour market functions.
The introduction of randomisation methods was, on the one hand, part of a drive to
evaluate public action by objectifying its effect, all others things being equal, and, on the
other hand, to focus on increasing the ‘back-to-work’ rate, regardless of the form of
employment found. Although it was an innovation in France, there was a certain degree of
continuity between this use of experimental surveys and the methods developed from the end
of the 1980s onwards. Thus it was more the profiles of the actors involved and the uses to
which these surveys were put that changed. This observation brings us back to our initial
assertion, namely that the symbolic power of statistics and their uses in public action are
closely connected with the social predispositions of the actors who produce them and the
power relationships specific to their sphere of operations.
The change observed in the evaluation of labour market policies reflects changes in the
academic discipline, with a generation of heterodox sociologists and economists being
downgraded in status and influence in favour of econometricians. This has led to the
marginalisation of the social sciences and the imposition of mathematical formalisation and
methods based on the medical experimentation model. It also reflects shifts in the power
relationships within the bureaucratic space – the diffusion of neoliberal prescriptions, changes
in the profiles of senior officials in the Ministry of Labour and the establishment of
competition between public employment service institutions and private providers. In other
words, the change in statistical tools and, above all, the analytical uses to which they are put
has less to do with technical ‘progress’ than the struggles between various fractions within the
bureaucratic and academic elites and their rival interests and bodies of knowledge. Several
research studies have highlighted the growing influence of consultancy companies (Power,
1999) to the detriment of academic researchers; here, however, the reverse model can be
observed: it is a grouping within the economics discipline that has sought to bring about
change in the upper echelons of the state by promoting new policy prescriptions and new
survey and evaluation methods. Thus these experimental methods are as much a change in the
scientific and statistical techniques used by the ministries of state as a reconfiguration of the
technologies of government.
7 Quantifying the effects of public action on the unemployed ...
Allègre, G. (2008). L’expérimentation sociale des incitations financières lemploi :
questions mộthodologiques et leỗons des expộriences nord-amộricaines. Document de travail
de l’OFCE, n°2008-22.
Aucouturier A.-L. (1994). Panels et évaluation des politiques de l’emploi. Cahiers Travail et
Emploi. Paris, Ministère du Travail, de lEmploi et de la Formation Professionnelle, La
Barbier J.-C. (2007). Réformer le service public de l’emploi : une fenêtre d’opportunité.
Regards croisés sur l’économie, 2, 235-253.
Behagel L., Crépon B., Gurgand M. (2012). Private and public provision of counseling to jobseekers: Evidence from a large controlled experiment. IZA, Discussion Paper, n°6518.
Bezès P. (2009). Rộinventer lEtat. Les rộformes de ladministration franỗaise (1962-2008).
Paris : PUF, Lien social.
Bezès P., Chauvière M., Chevallier J., DE Montricher N., Ocqueteau F. (2005). L’État à
l’épreuve des sciences sociales. La fonction recherche dans les administrations sous la Vème
République. Paris : La Découverte, collection Recherches.
Bourdieu P. (1998). Contre-feux. Propos pour servir à la résistance contre l’invasion néolibérale, Paris : Liber-Raison d’agir.
Bourdieu P. (1993). Esprits d’État. Genèse et structure du champ bureaucratique. Actes de la
Recherches en Sciences Sociales, 96-97, 49-62.
Bourdieu P. et CHRISTIN R. (1990). La construction du marché. Le champ administratif et la
production de la "politique du logement". Actes de la Recherche en Sciences Sociales, 81, 6585.
Brodaty T., CREPON B. et FOUGERE D. (2007). Les méthodes micro-économétriques
d’évaluation et leurs applications aux politiques actives de l’emploi. Économie et Prévision,
Bruno I. et DIDIER E. (2013). Benchmarking. L’Etat sous pression statistique. Paris : La
Cahuc P. et Kramarz F. (2004). De la précarité à la mobilité : vers une sécurité sociale
professionnelle, Rapport au ministre de l’Economie, des Finances et de l’Industrie, et au
ministre de l’Emploi, du Travail et de la Cohésion Sociale. La documentation franỗaise.
Cahuc P. et Zylberberg A. (2005). Le chụmage, fatalité ou nécessité ?. Paris : Flammarion,
Chelle E. (2012). Expérimentation sociale : la tentation américaine. Informations sociales,
DARES, UNEDIC et ANPE (2009). L’évaluation des expérimentations d’accompagnement
renforcé des demandeurs d’emploi conduites par l’ANPE et l’Unedic en 2007. Rapport du
comité de pilotage.
Desrosières A. (1997). Refléter ou instituer : l’invention des indicateurs statistiques. In
Dupoirier E. et Parodi J.-L. (Ed.), Les indicateurs socio-politiques aujourd’hui, Paris :
Dezalay Y. et Garth B. (1998). Le "Washington consensus". Contribution à une sociologie de
l’hégémonie du néo-libéralisme ». Actes de la Recherche en Sciences Sociales, 121-122, 3-22.
Duflo E. et Banerjee A. V. (2011). Poor Economics: A Radical Rethinking of the Way to
Fight Global Poverty, U.S.: Public Affairs.
Ermakoff I. et Tresmontant R. (1990). L’impact de la politique d’emploi : une estimation
quantitative de certains dispositifs. Économie et Prévision, 92-93, 111-125.
Even K. (2002). Un outil rénové pour évaluer les politiques de l’emploi et de la formation
professionnelle : le panel des bénéficiaires. Premières Informations-Premières Synthèses,
Fougère D. (2000). Expérimenter pour évaluer les politiques d’aide à l’emploi : les exemples
anglo-saxons et nord-européens. Revue Franỗaise des Affaires Sociales, 54, 111-144.
Gaudilliốre J.-P. (2006). La mộdecine et les sciences. Paris : La Découverte.
Gazier B. (1992). Economie du travail et de l’emploi. Paris : Daloz.
Gélot D. et Simonin B. (1996). L’évaluation de la politique de l’emploi, un bilan partiel des
pratiques depuis la fin des années 1970. In DARES, 40 ans de politique demploi. La
documentation franỗaise, 277-322.
Gomel B. et Serverin E. (2009). Expérimenter pour décider ? Le RSA en débat. Document de
travail du CEE, n°119.
Han Kim E. H., Morse A., Zingales L. (2006). What Has Mattered To Economics since
1970 ». Journal of Economic Perspectives, 20 (4), 189-202.
Hsiao C. (2006). Panel Data Analysis. Advantages and Challenges. Working Paper 06.49,
Institute of Economic Policy Research (IEPR), University of Southern California, May 2006.
Jobert B. et Théret B. (1994). France : la consécration républicaine du néo-libéralisme. In
Jobert B. (Ed.), Le tournant néo-libéral en Europe. Idées et recettes dans les pratiques
gouvernementales. Paris : L’Harmattan, 21-85.
Labrousse A. (2010). Nouvelle économie du développement et essais cliniques randomisés :
une mise en perspective d’un outil de preuve et de gouvernement. Revue de la régulation [En
ligne], 7 | 1er semestre / Spring 2010, mis en ligne le 29 novembre 2010.
Lebaron F. (2000). La croyance économique. Les économistes entre science et politique.
Paris : Seuil, Liber.
L’Horty Y. et Petit P. (2010). Évaluation aléatoire et expérimentations sociales. Document de
travail du CEE, n° 135.
L’Horty Y. (2006). Les nouvelles politiques d’emploi. Paris : La Découverte, Repères.