Changing social citizenship through information technology
Karolina Sztandar-Sztanderska, University of Warsaw
Marianna Zieleńska, University of Warsaw
This article takes a closer look at the implementation of computer-integrated profiling tool by Public Employment Services (PES) in Poland. It was introduced in 2014 in order to categorise the unemployed according to the so-called “employment potential”. The paper focuses on the tool’s influence on social citizenship in the area of active labour policy (ALMP). It shows that profiling has profound consequences for social rights of the unemployed, since it blurs the criteria of distribution of ALMP, excludes large number of the unemployed from access to it and deprive them of a right of appeal.
At a more general level, this paper tells a story of decomposition of legally defined collective categories which is mediated by information technology. On the one hand, the PES reform represents a move away from target groups acknowledged in the law and distinguished by a single and objective characteristic (such as duration of unemployment or age). Instead, the access to services and benefits is determined on the basis of multiple non-transparent and seemingly technical criteria inscribed in the IT tool designed for profiling (for other examples of non-transparent criteria used by employment and welfare agencies, see Berkel 2011; Dubois 2009). What is more, this computer-integrated assessment tool has significant consequences for the unemployed as there is no possibility to appeal against its verdict and approximately one third of the unemployed, who are classified as lacking “employment potential” become formally excluded from most forms of active labour market programmes (Niklas, Sztandar-Sztanderska, & Szymielewicz 2015). On the other hand, the profiling technology has been introduced outside the mechanisms of democratic oversight. The operating principles of the profiling tool have not become a part of the regular legislative process, though this instrument is undoubtedly political in nature, since it affects distribution of public services. They have been neither publicly disclosed, nor consulted with different administrative bodies as well as social partners and civil society organizations as it is the case of regular legislative process.
We argue that information technology was the key component of the decomposition of legally defined collective categories in the area of access to ALMP, since it enabled two major changes under a pretence of technical improvement, scientification and standardisation. First of all, the usage of IT made possible the departure from rules of access to services based on target groups and defined by law. Instead, the mechanisms of selection, that aim to assess an individual (rather than define a target groups) are hidden in the algorithm, which was established by unknown experts. In fact, the policy discourse presenting profiling as a progressive, technical and scientific created a smokescreen to hide the change of social rights of the unemployed and justified keeping its operational rules secret. Second of all, introduction of computer-integrated profiling technology was supposed to promote standardisation of the selection processes performed by frontline staff in PES – in order to make it more transparent compared to their discretionary application of legally defined access criteria (for analysis of discretion in Polish PES, see Sztandar-Sztanderska 2016; 2009). Meanwhile, as we will show, the use of IT requires a great deal of discretionary judgements that street-level bureaucrats have to make about people – judgements that are not made against the background of any legal rules and are very much dependent on interaction between a frontline worker and an unemployed person.
The paper is organized as follows. Firstly, we introduce the context of the implementation of profiling in Poland. Secondly, by adopting the street-level perspective we show that processes of interpretation and judgement are inherent in the practice of profiling. Finally, we show how process of profiling is linked to major changes in terms of social citizenship.
2 Implementation of profiling technology in Poland
Polish policy-makers legitimized the introduction of the standardised and computer-integrated classification system through an expert narrative that presented the numerical knowledge produced by the tool as “objective” in contrast with “subjective” assessment of frontline officials (for examples of similar narratives justifying the computerised assessment tools, see Caswell, Marston, & Larsen 2010). The previous system of allocation of ALMP based on broad target groups defined in legal framework was criticised as non-transparent, leaving space for arbitrary decision-making as well as leading to territorially differentiated policies and creaming. In this context, new profiling tool was promoted by central authorities as a technical solution that guaranteed transparency, standardised process of identification of individual barriers to labour market entry, unified principles of allocation of resources and increased the overall efficiency of ALMP (Sztandar-Sztanderska, 2013; (Niklas et al. 2015).
The operating principles of the profiling tool have not been reflected in the law. The discourse referring to science, expertise, IT-enabled standardisation and efficiency – used by policy-makers – contributed to that, by depoliticizing profiling process. Legal acts include only basic information – mainly that the unemployed will be divided into three “profiles of assistance” and that people included in each category are potentially entitled to a different range of services and support. Whereas the mechanism according to which selection to profiles takes place is not presented in any of the acts, it is only inscribed it the profiling tool in a form of an algorithm. What is more, the Ministry of Labour and Social Policy defined the algorithm as “not public information” and was refusing to disclose it until the administrative court decided otherwise in 2016 (Niklas 2016).
Profiling involves ranking the unemployed depending on the calculated value of two variables called “distance from labour market” and “job readiness”. This scoring significantly affects the life situation and professional chances of the unemployed, since depending on a categorisation different ALMP can be accessed and different case management procedures apply. Frontline workers are theoretically allowed to change the automatic classification, but such a decision is recorded by the software and should be complemented by their justification typed in the electronic system.
Profiling classification instrument is composed of 24 questions. Each of questions has standardised answer choices. Every response has an assigned score. Eight questions are processed automatically, based on data already available in PES electronic system such as age, sex, education level and work experience, foreign language skills, duration of unemployment, disability and registered instances of non-compliance with PES requirements (such as refusals of suitable employment or no-shows for an appointed meeting). The rest of pre-defined answer choices have to be fill in the software by the client counsellors, who are responsible for conducting the interviews with profilees. The questionnaire includes open-ended questions that might provoke longer descriptive statements from the unemployed rather than simple answers included in the standardised grid.
What is important, client counsellors are advised to conceal the questionnaire and standardised answer choices from the unemployed. This ministerial recommendation is explained as a means of security against abuses on part of the unemployed. The unemployed are – from the very beginning of the reform – presented as a potential threat to the objectivity of the profiling result. According to policy-makers and frontline staff, this group is prone to provide false declarations in order to outsmart the system and obtain access to services to which they are formally not entitled to (MPiPS 2014). A similar “moral construction” of the unemployed as “welfare abusers” (Handler & Hasenfeld 1991, Brodkin 1993) appears in training materials. Frontline workers, are advised that they are allowed to change the automated classification suggested by the software, when they sense “the discrepancy between the attitude represented by the unemployed” and “his answers” (MPiPS 2014, pp. 47-48).
The assessment finishes with the assignment of one of three categorisations, that supposedly reflect level of “employment potential”. First profile is for those with the highest level of “employment potential”, second profile for those with its average level and third for those with its lowest level. The priority is given to people classified as the second “profile of assistance”. They are allowed to apply for the largest range of services. Those classified as first “profile of assistance” have more limited potential access, but may apply for attractive instruments such as subsidy for opening their own business. There is also a possibility to offer them services intended for the second profile, if PES decides that they need it and should make good use of it. Those classified as the third profile – approximately one third of the unemployed (MRPiPS 2016) – are not entitled to any of the regular instruments offered by PES. Instead, they may be referred to two types of facultative programmes or to outsourced services, if there are any in a given territorial unit. However, outsourcing is still very limited in scope and facultative programmes are hardly launched by PES (Herman-Pawłowska et al. n.d.). In consequence, there is no offer of support for most of the members of third profile.
3 Profiling in practice of PES
One of the main official goals of profiling tool was to standardise the assessments performed by street-level bureaucrats. At the same time, the Ministry of Labour and Social Policy left some leeway in decision making to client counsellors, by officially allowing them to change the automatically generated classification: such option was included in the software. This way frontline workers’ capacity of making decisions and judgments about individual cases was formally acknowledged, by leaving them space for discretionary judgements. Surprisingly enough, this possibility has been rarely used by staff performing profiling: although 49.3% of counsellors have at least once used the option to change the profile, in fact, cases in which the profile was altered are marginal. They represent only 0.58% of all profiling cases (MPiPS 2015b, p. 1). One of the reasons for this reluctance was the requirement to write justification of the decision to alter the profile, which is both time-consuming and may become subject to ministerial control, as all such changes are visible in the computer system. The software itself issues a warning highlighted in green, pointing out that a counsellor “should remember that all the changes of profiles are registered in the system” (MPiPS 2014, p. 48). However, it does not mean that the frontline staff lack any discretionary power in the profiling process. On the contrary, making judgements is an inevitable part of it, yet invisible from the perspective of IT. It is situated in the interaction between the unemployed and the SLB during the profiling interview: frontline workers make numerous decisions about how to conduct the interview and how to interpret given answers. What is more, this interaction is not bounded by any legal rules, since operating principles of profiling are not in the law, but in the algorithm.
3.1 Presenting profiling procedure and formulating questions
Case counsellors have large margin of manoeuvre regarding how exactly the interview will be conducted. The first step is making a decision on how to present the profiling process and its consequences to the unemployed person. As some of the unemployed we interviewed emphasised, they had not been aware what exactly was going on during profiling and even which profile was assigned to them:
Researcher: And were you forewarned beforehand that you will be profiled? At the previous meeting?
Unemployed: No, not really, I did not know (…). He did not say "You know, I'm going to profile you now", only he just caught me off guard [slight laughter]. That's what you say, by surprise.
Researcher: So he started asking questions?
Unemployed: Yes, he started to ask, I started to answer him, of course. […]
Researcher: And what profile did you get, have you been informed?
Unemployed: No, no, no, I do not know it, what profile I have. And there are some profiles, they have ...?
Researcher: - One, two, three.
Unemployed: Oh, one, two, three ... Something might have come up, but I do not know exactly. It means ... what are those profiles about, those first, second and third?.
Also the way case counsellors approach asking questions varies. They may simply read the questions aloud and refrain from any sort of additional interaction with profilees. They may also try to translate the questions from complex administrative language into a simpler version which is more connected to everyday life experience of the unemployed. This choice is extremely important taking into consideration the fact that questions are asked as open-ended rather than having a limited number of predefined responses. Even though there is a standardised grid of answer choices, the unemployed neither sees the possible answers, nor is informed about them. Apart from that, the questions are often ambiguous (for examples see box 1) and leave a lot of room for interpretation, thus it is probable that the person subjected to profiling will not think of some of the predefined answers when answering the question on his or her own
Question: “What are you able/ready to do in order to increase your chances for employment?”
· The grid includes, among others, the following options: I am ready/able to:
· Change the place of residence
· Commute to work more than 1.5 hours in one direction
· Take up part time work
· Take up work in flexible working hours
· Work in a multi-shift system
· Get a job below qualification
· Get a job even without a contract
· Go abroad
· Increase own job search activity
· Start a business
· I'm not ready to do anything
The problem of clarity of questions is raised also by the frontline staff. As mentioned by one of the experienced client counsellors, on every day basis she deals with this ambiguity by explaining the questions in her own words. She refers to the question about increasing employment chances (Box 1) as especially difficult:
“This is a question that poses a lot of problems to understand its intentions. We need to explain it a lot, really; we need to make clear what it is about (...) Here we need to steer [the conversation], because it really is a problem” [PES6]
Apart from that, some questions may not be asked at all, especially if frontline workers believe that they already know the answer. Such an attitude bares the risk of making assumptions about various matters on the basis of personal impressions about a given person or the conviction that the answer is self-evident. For example, the frontline worker may draw conclusions regarding person’s lack of motivation to find work on the basis of the information about the history of registration with local PES and the fact that he or she rejected various work offers; whereas the reason for rejecting offers might have been multiple and the situation of a person might have changed over time. Moreover, the unemployed have no possibility to neither correct nor take critical stance to the counsellor’s interpretation, since they do not see what is actually being ticked as an answer. There is also no appeal procedure from the attributed categorisation.
In practice, many of the choices made by street-level workers concerning the interview strategy may be related to the caseload and the time pressure resulting from it. The reform introducing profiling of the unemployed imposed new obligations on a frontline staff without improving the already difficult staffing situation in local PES (Niklas et. al). According to the data of the Ministry of Labour and Social Policy, in 2014, when the reform entered into force, on average there were as many as 234 unemployed per one “client counsellor”, i.e. a street-level worker responsible for activation process (MPiPS, 2015a, p. 26) . This indicator varies locally and some PES have even more serious staffing problems.
It is also worth mentioning, that the caseload was especially high at the beginning of the profiling process, when all the unemployed had to undergo the procedure and not only those newly registered. As shown by the qualitative research conducted shortly after introduction of profiling in one of the offices declaring staff shortages, the estimated time for one interview was approximately 7 minutes, with 60-70 persons processed on a daily basis [PES 3] (Niklas et al., 2015, p. 24). What is more, ministerial statistics from December 2014 based on the information from electronic system regarding all the PES agencies show even shorter average time of interview – 5.2 minutes which leaves approximately 20 seconds per a question. In 2015 majority of interviews (69%) took maximum 4 minutes, which gives maximum 15 seconds per question (MRPiPS 2016). This suggest that the so-called “objective diagnosis” of individual “employment potential” carries a risk of being extremely superficial and that some questions are actually not asked directly.
4 Interpretation of answers
Another space for discretion is connected with entering the data to the software. There are at least two situations in which frontline staff have actual decision-making power, because their role consists in translating complex realities and individual life-circumstances into predefined answer choices.
As already mentioned, some of the questions leave room for interpretation. In consequence, the answer given by the unemployed may simply not fit the grid of predefined answers. In that case the frontline worker has to make a decision what to do and how to inscribe the answer into the computer assisted questionnaire. One of the questions reported by frontline staff interviewees as generating this particular problem is about “the reasons making it difficult to take up work” (MLPS 2014a, p. 41). Even though there are 22 possible answer choices, the list is non-exhaustive: the predefined choices do not include reasons such as homelessness or criminal record that actually re-occur among answers given by the unemployed.
· Box 2
· Question: “Please indicated reasons that make it difficult for you to take up work”
· The grid includes, among others, the following options:
· lack of adequate education
· lack of work experience
· job offers without insurance (undeclared work)
· too high requirements of employers
· too much competition
· too low remuneration proposed by employers
· lack of job search skills and self-presentation
· care for children and / or a dependent person
· health restrictions
· lack of conviction that I need to work
· no causes
Counsellors admitted to three strategies of dealing with this situation. The first comes down to ticking the answer choice that is – according to the interpretation of the frontline worker – the closest to the issue mentioned by the unemployed person. It means that in case of homeless person, the counsellor chooses “too much competition” or “health restrictions” and “lack of job-seeking skills and self-presentation” as the reasons making it difficult to take up work. The interviewee explained that she chooses answer “too much competition”, if she suspects that the core problem is only the lack of formal place of residence (“employer does not want to hire persons without a residence address” [PES 3]) or she chooses “lack of job-seeking skills and self-presentation” when the problem is personal hygiene (a person “is dirty and stinks” [PES 3]). The second strategy is trying to find out whether other issues – this time mentioned in the questionnaire – are connected to the previously mentioned difficulty: in case of homelessness it may be “health restrictions” or “a lack of conviction about the necessity to take up a job” [PES 6]. Finally, the third strategy is to make sure that a person will eventually be included in the third profile as “distant from the labour market”, no matter the result of the automated scoring [PES 3]. In other words the fact of being homeless will not be translated into the grid, but will be a decisive factor in assigning the profile.
Another situation in which frontline workers act as translators of complex reality is when the unemployed person gives a conditional answer to a question. A good example of it is answering to the already mentioned question “What are you able/ready to do in order to increase your chances for employment?” (see, Box 1). When declaring the compromises they are willing to make in order to get a job, the unemployed persons are likely to perceive it as dependent on extra factors, e.g. one may agree to commute 3 hours a day or move abroad, but only if the pay is sufficiently high. In cases like those frontline staff have to make quick, arbitrary decisions, whether to classify these context-specific declarations signs of “readiness” to increase employment chances or not.
The frontline workers’ interpretation of answers given by the unemployed is also strongly influenced by the quality of communication between those two parties, i.e. how much will the unemployed be willing to say, what language skills will she or he possesses or how much effort will the counsellor make to explain the ambiguous and difficult questions. What is important, this process of communication will be dependent not only on the will and attitude of the profiler and the profilee, but also on contextual factors. The limited time for the interview and the caseload have already been mentioned as restraints. However, what seems to be equally important is also the organization of space. The research conducted so far (Niklas et al. 2015, p. 24; Herman-Pawłowska et al. n.d.) indicate that profiling interviews are often held in an open space or an office shared with other frontline workers without any guarantee of privacy or confidentiality, even though it is required by legal regulations. Such conditions are not favourable to an in-depth diagnostic conversation and may simply hinder the ability to concentrate in case of both – the frontline staff and the unemployed person – leading to errors in judgement.
5 Efficiency indicators and creaming practices
Assigning the profile does not determine
person’s access to any ALMP, but it may close admittance to some of the
instruments (especially in the case of third profile). In fact, profiling
actually opens another process of selection, this time to concrete labour
market programmes. Due to insufficient funding, street-level bureaucrats have
to make choices to whom among the potentially eligible unemployed give access
The recent reform of PES created an important incentive for creaming practices by linking performance indicators with financial rewards for PES agencies (Niklas et al., 2015; Herman-Pawłowska et al. n.d.). The agencies with higher efficiency receive additional money that can be distributed among staff as bonus. In result, ALMP are distributed to the “more promising” cases by the frontline staff (Herman-Pawłowska et al. n.d., p. 44).
Some creaming practices are linked directly to profiling itself. There is a legal possibility to assign ALMP reserved for the second profile also to the unemployed from the first profile (those who are diagnosed with the highest employment potential who do not need much support) in exceptional cases. This exception is used by local PES to boost performance indicators, since people from the first profile are assessed as more likely to find employment after activation measure than people from the second profile. One of the upper level managers called this practice “picking raisins” (which could be understood as cherry picking):
“I keep telling my employees: we have a limited group of people who are able to manage – those are the most active persons. The law allows us to help them in exceptional cases, yes? So, if you have a motivated client from the first profile, who needs an additional training, renewal of certain skills or something of that sort – give it to them. Why? Because we will have higher employment efficiency! When someone has one leg shorter, I’m not forcing them to take part in a race. I choose those with whom I have 100% certainty that they will attend the training or internship. (…) You cannot help everyone. This means that you have to pick raisins and invest in them. Only then we have 6 thousand happy people [those who we can actually help with our budgets – author] and we get bonuses, so we are happy too…and the minister receives high efficiency rates”. [PES 1].
Performance indicators have also influence on the frontline staff’s approach to the unemployed classified as third profile. As the least promising cases, not only are they excluded from the majority of ALMP, but also less time is devoted to them by client counsellors, e.g. they meet with unemployed classified to third profile less frequently (even as rarely as twice a year). Street-level workers tend to prioritize meetings with people from second profile instead (they meet even up to once a week) [PES 3]. Local PES staff seem to believe that this strategy is in the best interest of the unemployed assigned to both these groups. They are under impression that most of the unemployed who are categorised in third profile have registered with PES only to obtain free healthcare insurance and are neither “interested in office support” nor in finding a job [PES 1, PES 3]. According to one of the client counsellors it is about respecting her own time and time of the unemployed persons and focusing on those who are more promising in terms of activation:
“For me, personally, it is a big plus that persons with this third profile (...) maybe I will tell it briefly, I don’t have to take care of them. Meaning: they have their obligatory visit from time to time. I ask them if anything had changed, that maybe they are interested in something [that PES offers] (...) and if not, I appoint them another date. This visit is short (...) and I don’t waste time (...) Because a person might not be interested in taking up a job (...) for various reasons. It doesn’t matter if she doesn’t want to or if, if she is ill, for instance, and she can’t. It is like I don’t waste my time and I don’t waste hers (...) And I have time for these people who I can help.” [PES 3].
Even though devoting more attention to people from second profile does not automatically boost performance indicators, it may be considered as economizing strategy for distribution of scarce resources such as time and money. It seems especially important if we take into consideration the fact that the number of available ALMP is very limited and the frontline workers have to deal with a high caseload.
The analysis of the case of profiling has
shown that – despite using a standardised computer-integrated tool – case
counsellors have meaningful discretionary power influencing the final categorisation
of the unemployed. It starts with the presentation of profiling and the
information that is conveyed at that point, i.e. whether the unemployed is
informed about the importance and consequences of the whole process. It is also
situated in the moment of asking (or not) questions and interpreting the
answers that do not fit within the standardised grid. Additionally, the process
is influenced by working conditions that are not favourable to in-depth
interviewing and by performance indicators that may contribute to creaming
practices manifesting itself as investing in those who seem the most promising.
There is however, another important component of case counsellors’
decision-making power, that has not been emphasised so far – it is the lack of the legal framework thus, bounding discretion, regulating criteria of who belongs to
which profile of assistance. If we look at the definition of discretion as
defined by Lipsky (and subsequently by street-level bureaucracy scholars) it
should be analysed against the background of legal rules regulating a given
administrative field (2010, p. 14). It is centred around application of general law to particular
cases, that are often complex and inconclusive. In case of profiling such rules
are virtually non-existent and, thus, discretion is exercised against the
background of IT, rather than a legal framework.
This new type of IT-centred discretion is linked to the changing expectations set for street-level bureaucrats in profiling. Their task is no longer verification of formal entitlements based on belonging to certain target groups. Now it is about (computer-assisted) assessment of the individual and her or his suitability to different kinds of ALMP, based on criteria that are quite blurry and combine both certain structural, socio-demographic and labour market features (e.g. gender, place of living, work history, etc.) with the ambition to measure person’s motivation to work (e.g. answers connected to the question about what a person is willing to do to improve employment chances) (for the analysis of individualisation of social service provision, see van Berkel, Valkenbourg 2007; Borghi, Berkel 2007, Garsten et al. 2016).
The mechanism we observe through the case
of profiling is similar to what Vincent Dubois (2009) referred to as a growing
importance of interaction between
the street-level bureaucrats and the poor in the reformed welfare state. In his
article about the paradox of controller, he linked the central position of this
interaction for attribution of benefits to the changing logic of the
functioning of welfare services under the influence of managerialism. The
pressure to economise forces welfare state organizations to reduce costs and
increase control of individuals. Thus, street-level bureaucrats are no longer
tasked with corroboration of entitlements to benefits and services on the basis
of documents delivered by the unemployed; they are tasked with the assessment
of individuals and their deservingness according to the principle that everyone
is suspect of cheating, some are guilty but give prospects for resocialisation
and some are simply not irredeemable and should be left out.
Thus, this process of decomposition of legal frameworks regulating access to benefits and services of different groups changes the role of street-level bureaucrats. As suggested by Evelyn Brodkin (2013, p. 32), welfare state reforms driven by managerialism push street-level organisations (SLOs) further than mediating policy understood translating legal frameworks into everyday practice – they push SLOs into mediating politics:
“The work of SLOs should be understood as deeply political in part because their practices determine “who gets what, when, and how (…). But this is a different form of policymaking from that which occurs through legislative or other authorized channels: it is difficult to observe and has consequences that are difficult to assess and trace”.
Profiling certainly is a key step towards making decisions about “who gets what, when and how”, because it serves as a “sieve” separating those who – according to the algorithm – should not apply for particular services from the more deserving ones – an in that sense it is political in nature. Similarly to what Brodkin suggests, this political nature of profiling is also very difficult to observe, since it is hidden behind the veil of objectivity of science and technology.
7 Conclusions: profiling and changes of social citizenship
Profiling case shows how the use of IT enables major changes influencing social citizenship in the area of ALMP and contributes to decomposition of legally defined collective statuses. First of all, criteria of distribution of ALMP have been shifted from law to an algorithm. In result, there is no legal framework regulating the process of selection to profiles. It is the more important provided that the mechanism of selection has been profoundly altered from one based on target groups defined by single, objective socio-demographic or labour market characteristic, such as duration of unemployment or age, into one based on a multiple blurry criteria, combining objective characteristic with subjective assessment of personal motivations and attitudes.
Second of all, the process of selection itself is far from transparent, as it requires a lot of discretionary judgments made by case counsellors conducting profiling. What is more, those judgements are not bounded by any legal rules, what has always been at the core of discretion understood not as “the absence of principles or rules” but as “the space between them” (Evans 2004, p. 881 see also: Lipsky 2010). From what the analysis of the profiling process shows, the interaction between the unemployed and the case counsellor plays a key role in dividing the unemployed into categories, i.e. factor such as how a particular question is formulated, how it is understood by the unemployed and how it is interpreted by the counsellor are crucial for the results of profiling.
To conclude, the abovementioned changes have profound consequence for the people who undergo profiling. The introduction of IT has increased power asymmetry between the unemployed and the state and contributed to the administrative discrimination of, perhaps, the most vulnerable individuals. They do not know what kind of information about them is gathered and stored by PES (including what answers are actually ticked off), as they do not have access to the criteria determining the results of profiling. They do not have an administrative path of appeal as well. Finally, a large group of the unemployed – those classified as having the lowest employment potential – have been officially excluded from most of the ALMP offered by PES agencies. They have been also constructed as immoral and guilty of their difficult labour market situation. We will, therefore, argue that a specific variant profiling tool adopted in Poland has been a typical example of policy that is little effective in counteracting unemployment, but as phrased by Brodkin (1993, pp. 649-650): “has been more effective in vilifying” the unemployed as a group.
Berkel, R. van, & Valkenburg, B. (2007) (Eds). Making it personal: individualising activation services in the EU. Bristol, UK: The Policy Press.
Berkel, R. van. (2011). The Local- and Street-Level Production of Social Citizenship: The Case of Dutch Social Assistance. In S. Betzelt & S. Bothfeld (Eds.), Activation and Labour Market Reforms in Europe (pp. 195–217). London: Palgrave Macmillan. https://doi.org/10.1057/9780230307636_10.
Borghi V., & Berkel R. van, (2007). Individualised service provision in an era of activation and new governance. International Journal of Sociology and Social Policy, 27 (9/10), 413-424, https://doi.org/10.1108/01443330710822101.
Brodkin, E. (2013). Street-Level Organizations and the Welfare State. In E. Z. Brodkin & G. Marston (Eds.), Work and the Welfare State: Street-Level Organizations and Workfare Politics (pp. 17-34). Washington, D.C: Georgetown University Press.
Brodkin, E. Z. (1993). The Making of an Enemy: How Welfare Policies Construct the Poor. Law and Social Inquiry, 18(4), 647–670. https://doi.org/10.1111/j.1747-4469.1993.tb00754.x.
Caswell, D., Marston, G., & Larsen, J. E. (2010). Unemployed citizen or ‘at risk’ client? Classification systems and employment services in Denmark and Australia. Unemployed citizen or ‘at risk’ client?. Critical Social Policy, 30(3), 384–404. https://doi.org/10.1177/0261018310367674.
Dubois, V. (2009). Le paradoxe du contrôleur: Incertitude et contrainte institutionnelle dans le contrôle des assistés sociaux. Actes de la recherche en sciences sociales, 178(3), 28-49. https://doi.org/10.3917/arss.178.0028.
Evans, T. (2004). Street-Level Bureaucracy, Social Work and the (Exaggerated) Death of Discretion. British Journal of Social Work, 34(6), 871–895. https://doi.org/10.1093/bjsw/bch106,
Garsten, C., Jacobsson K., & Sztandar-Sztanderska K. (2016). Negotiating social citizenship at the street-level: local activation policies and individualization in Sweden and Poland. In: M. Heidenreich M. & D. Rice, Integrating Social and Employment Policies in Europe, (pp. 265–294). Edward Elgar Publishing. https://doi.org/10.4337/9781783474929.00019.
Handler, J. F., & Hasenfeld, Y. (1991). The moral construction of poverty: welfare reform in America. Sage Publications.
Herman-Pawłowska, K., Stronkowski, P., Bienias, S., Dybaś, M., Kolczyński, M., Kulawik-Dutkowska, J., & Skórska, P. (2015). Ocena skutków regulacji wybranych aspektów wdrażania Ustawy z dnia 14 marca 2014 r. o zmianie ustawy o promocji zatrudnienia i instytucjach rynku pracy oraz niektórych innych ustaw. IDEA Foundation.
Lipsky, M. (2010). Street Level Bureaucracy: Dilemmas of the Individual in Public Services. New York: Russell Sage Foundation.
MPiPS. (2014). Profilowanie pomocy dla osób bezrobotnych. Podręcznik dla pracowników powiatowych urzędów pracy.
MPiPS. (2015a). Informacja o stanie i strukturze zatrudnienia w wojewódzkich i powiatowych urzędach pracy w 2014 roku.,. Retrieved from: https://psz.praca.gov.pl/documents/10828/1383908/Stan%20i%20struktura%20zatrudnienia %20w%20PUP%20i%20WUP%20-%20Rok%202014.pdf/f09c5a72-d375-43b3-9798- 76f100476a38?t=1429004671000.
MPiPS. (2015b). List do dyrektorów urzędów pracy.
MRPiPS. (2016). Analiza rozwiązań wprowadzonych ustawą z dnia 14 marca 2014r. o zmienie ustawy o promocji zatrudnienia i instytucjach rynku pracy oraz niektórych innych ustaw (DZ. U. POZ. 598).
Niklas, J. (2016). Sąd administracyjny: Zasady profilowania bezrobotnych są informacją publiczną! | Fundacja Panoptykon. Retrieved June 25, 2018, from: https://panoptykon.org/wiadomosc/sad-administracyjny-zasady-profilowania-bezrobotnych-sa-informacja-publiczna.
Niklas, J., Sztandar-Sztanderska, K., & Szymielewicz, K. (2015). Profiling the Unemployed in Poland: Social and Political Implications of Algorithmic Decison Making. Warsaw: Fundacja Panoptykon. Retrieved from: https://panoptykon.org/biblio/profiling-unemployed-poland-social-and-political-implications-algorithmic-decison-making.
Sztandar‐Sztanderska, K. (2009). Activation of the unemployed in Poland: from policy design to policy implementation. International Journal of Sociology and Social Policy, 29(11/12), 624–636. https://doi.org/10.1108/01443330910999069.
Sztandar-Sztanderska, K. (2013). Nie zrzucajmy całej winy na nieefektywne urzędy pracyifikcyjnychbezrobotnych. Uwagi oniepożądanychkonsekwencjachnowelizacji ustawy o promocjizatrudnienia i instytucjach rynku pracy.
Sztandar-Sztanderska, K. (2016). Obywatel spotyka państwo: o urzędach pracy jako biurokracji pierwszego kontaktu. Warszawa: Wydawnictwo Naukowe Scholar.
Karolina Sztandar-Sztanderska, Ph.D.
University of Warsaw, Poland
Department of the Sociology of Work and Organizations
Marianna Zielenska, PhD.
University of Warsaw, Poland
Institute of Sociology
 The article was prepared in the frame of the ongoing project “Information technologies in public policy. Critical analysis of the profiling the unemployed in Poland” directed by Karolina Sztandar-Sztanderska and financed by the National Science Centre, Poland (2016/23/B/HS5/00889). We also use data gathered in the frame of previous 3 pilot projects: 2 research grants for young researchers – 1) “To measure, to weigh, to calculate, to classify. Methods of making citizens legible” carried out by K. Sztandar-Sztanderska (DSM 112900/16); 2) “Profiling the unemployed - social and political consequences of new categorization tools introduced by local labour offices” carried out by K. Sztandar-Sztanderska (DSM 110400/66) and M. Zieleńska (DSM 110 400/72) as well as the research “Profiling the Unemployed in Poland. Social and Political Implications of Algorithmic Decision Making” conducted by K. Sztandar-Sztanderska and J. Niklas for Panoptykon Foundation and funded by Media Democracy Fund, Ford Foundation and Open Society Foundations.