HERMAN DELEECK
CENTRE FOR SOCIAL POLICY
Marjolijn De Wilde & Sarah Marchal
Weighing up work willingness in social
assistance: a balancing act on multiple
levels
WORKING PAPER
NO. 18.08
March 2018
University of Antwerp
Herman Deleeck Centre for Social Policy
centrumvoorsociaalbeleid.be
1 CSB Working Paper No. 18/08
Weighing up work willingness in social assistance: a balancing act on multiple
levels
Marjolijn De Wilde & Sarah Marchal
Working Paper No. 18/08
March 2018
ABSTRACT
In all European countries, social assistance receipt is conditional upon the willingness to
work. Yet despite the harsh consequences of losing social assistance, we know surprisingly
little about how social assistance agencies and social workers implement this policy in
day-to-day practice. In this paper, we focus on three important questions regarding the
implementation of work willingness as a condition for continued social assistance benefit
receipt. First, how does the actual implementation of the work willingness condition take
place in light of specific client characteristics, circumstances and behaviour? Second, is the
interpretation of such behaviour similar across case managers and municipalities, or does
the combination of vague work willingness legislation and a decentralized organisation
lead to variation in implementation? Third, can such variation be seen as the express
objective of decentralization and personalized work willingness assessments? We build on
an innovative and purpose-designed factorial survey of social workers in Belgium. We
identified the determinants of 582 social workers’ sanction decisions upon a job refusal,
clustered in 89 municipalities, on almost 5000 experimentally varied client cases. These
unique data allow to distinguish between the effects of individual client characteristics,
characteristics of the social workers assessing the individual cases and the characteristics
of the local welfare agency and municipality in which she operates. Moreover, we assess
how characteristics within and between these levels interact. In line with the literature,
we find substantial variation in sanctions related to work unwillingness at the client level,
that can be explained by individual client characteristics. Variation between municipalities
is relatively limited, and can be fully explained by municipality characteristics. Surprisingly,
we find the largest variation at the social worker level. Whereas some of this variation is
random, a substantial part can be explained by the characteristics of the social worker.
This finding raises concerns about the unintended consequences of the large discretion
awarded to social workers within contemporary social assistance schemes.
Keywords: Social policy implementation; discretion; decentralization; social assistance;
willingness to work; activation; factorial survey; vignette study
2 CSB Working Paper No. 18/08
1 Introduction
In all European countries, social assistance receipt is conditional upon the willingness to
work. Legislation allows or even mandates to withdraw or reduce benefits from beneficiaries
who fail to demonstrate a motivation to work, for instance by refusing a reasonable job offer
(Marchal & van Mechelen, 2017; MISSOC, 2015). Yet despite the harsh consequences of
losing social assistance, we know surprisingly little about how social assistance agencies and
social workers implement this policy in day-to-day practice (Pavetti, Derr, & Hesketh, 2003;
Scott, 1997).
This lack of research is unsurprising as work willingness is not an objective criterion that can
be implemented and investigated in a straightforward manner (Heidenreich & Aurich-
Beerheide, 2014; van Berkel, 2010a, 2010b; Watkins-Hayes, 2009). Although some
guidelines, such as suitable job offer specifications can be listed in legislation, it is impossible
to completely prescribe behavioural conditions. Its implementation necessarily hinges on
decisions of social assistance beneficiary’s case managers, who are trained to translate
legislation to individual cases. In addition, social assistance is notoriously decentralized, both
for historical reasons and to facilitate taking account of local circumstances (Sabatinelli,
2010). More recently, scholars and policy makers alike have started to see decentralization
as an incubator for innovative policies (Kazepov, 2010). In sum, it is fair to expect that the
specification and the assessment of behaviour that can be considered as work
(un)willingness are to a large extent subject to both the organisation and the social worker
that is following the client.
In this paper, we focus on three important questions regarding the implementation of work
willingness in social assistance legislation that have so far remained under investigated. First,
how does the actual implementation of the work willingness condition take place in light of
specific client characteristics, circumstances and behaviour? Second, is the interpretation of
such behaviour similar across case managers and municipalities, or does the combination of
vague work willingness legislation and a decentralized organisation lead to variation in
implementation? Third, can such variation be seen as the express objective of
decentralization and personalized work willingness assessments? To open up this black box
of social assistance activation policies’ implementation, we bring together academic
3 CSB Working Paper No. 18/08
research streams on conditionality, decentralization, discretion and professionalization (De
Wilde & Goos, 2017; Priem, Walters, & Li, 2011; Rice, 2012).
This paper is the first study to quantitatively assess these questions while expressly taking
account of the different relevant implementation levels. Past research on the topic generally
focused on only one or two levels of decision-making. Examples include descriptions of
national or local legislation (e.g., Eleveld, 2016), studies on large-scale register data (e.g.,
Soss, Fording, & Schram, 2011) and evidence in qualitative case studies or in case law on the
specific considerations concerning particular cases (e.g. Brodkin, 2011).
A detailed examination of the determinants of work conditionality assessments at different
levels is thus far lacking.
We build on an innovative and purpose-designed factorial survey of social workers in
Belgium. We identified the determinants of 582 social workers’ sanction decisions upon a
job refusal, clustered in 89 municipalities, on almost 5000 experimentally varied client cases.
These unique data allow to distinguish between the effects of individual client
characteristics, characteristics of the social workers assessing the individual cases and the
characteristics of the local welfare agency and municipality in which she operates.
Moreover, we can assess how characteristics within and between these levels interact.
Belgium is a particularly interesting case for an analysis of work willingness implementation
as work conditionality is a key factor in federal social assistance legislation. Actual
implementation is left to local social assistance agencies and to the professionals handling
client cases.
In the following section, we identify for each of these levels likely determinants of the
implementation of a work willingness related sanction. Next, we describe the Belgian social
assistance scheme. We then present the data and the multi-level method we use to identify
the determinants of sanctioning implementation at the client, agency and social worker
level, and their interactions. After the presentation and discussion of the results, we
conclude.
4 CSB Working Paper No. 18/08
2 Literature review
It is an open question which considerations at the client, the municipality and the case
manager level, may impact on the implementation of work unwillingness sanctions. Social
assistance legislation is generally vague on work willingness, so that actual assessments can
take individual client characteristics and circumstances into account. Certain characteristics
are by most people and in most cases considered as reasons to sanction clients (van
Oorschot, Meuleman, Roosma, & Reeskens, 2017), which is what we focus on in the first
section. Yet the different levels at which work willingness is assessed may give rise to
variation in actual implementation. In the second section, we review the legislator’s express
intent and acceptance of variation by decentralising policy implementation. The last section
focusses on the inevitable variation due to the human case managers applying legislation in
real-world cases.
2.1 The assessment of client characteristics
Client characteristics that should or should not lead to sanctions are in some countries to a
certain extent detailed in legislation. However, most of the social assistance legislation is
aimed at a personalized assessment of a client’s individual situation. It can be expected that
such an assessment is influenced by the same considerations that apply for the society at
large when thinking about solidarity, as a common personal and professional understanding
(Jasso, 2006; Keiser, Mueser, & Choi, 2004; B. J. Taylor, 2006; Wallander, 2012).
Such an understanding would be guided by the assessment of ‘deservingness’ (van Oorschot,
2000). We expect that the need to demonstrate work willingness or to be exempt originates
from the traditional deservingness criteria, namely reciprocity, attitude, control, need and
identity (De Wilde, 2017; van Oorschot et al., 2017).
There is only little quantitative evidence on how deservingness characteristics are weighted
against activity-related infringements and possible exemptions for work willingness in
treatment reality. Studies based on administrative data only provide insight into which
groups are more often sanctioned. These studies show that sanctioned recipients are often
foreign, young, never married, poorly educated, parents, sick, caring for a sick child,
