What works – study into the effects of four different social welfare schemes

This study includes three treatment conditions, one control condition and a reference group. In the control condition, claimants are subject to the rules and regulation of the Dutch status quo social welfare regime. It is instructive to briefly introduce the status quo regime. Social welfare in the Netherlands is a non-contributory transfer programme that provides monthly income support to poor households identified based on a means, and work test. The regulations are extensive and complex, which is why we focus on the most important rules and regulations. The regime foresees a monthly transfer payment of maximum 992 EUR for a single-person household. On top of that welfare claimants may be eligible for child, housing and healthcare allowances. The policy is designed as a temporary safety net aiming to deliver income support until recipients can provide for their own income again (mostly by finding employment). Accordingly, welfare recipients have to comply with certain rules, such as writing application letters, accepting job offers or following skill-development programmes. Recipients that fail to comply with the rules may be sanctioned by freezing or cutting their monthly payment. Exemptions apply to those who are incapacitated for work or face severe health problems. Local governments (municipalities and regional councils) are charged with the task of executing the scheme. In the three treatment conditions the rules and regulations are varied. The three interventions tested are: (1) claimants are exempted from the usual obligations to maintain their benefit, such as applying for jobs or joining reintegration programmes; (2) claimants receive extra support through tailor-made supervision and intensive mediation from the welfare agency (=Municipality of Utrecht); (3) claimants are allowed to keep additionally earned income to a larger extent than under current regulations (current: keep 25% of additionally earned income up to max. 202 EUR/month for max. six months; intervention: keep 50% of additionally earned income up to max. 202 EUR/month for max. 16 months). Additionally, we use administrative data to follow a reference group of randomly selected claimants who were not invited to participate in the trial.

2018-06-01

2019-12-31

Our primary outcome indicators concern labour market outcomes, such as differences in the uptake of paid work (full time and part time) and quality of reemployment.

The uptake of paid work and reemployment quality (earnings, type of contract) will be measured using administrative data from Statistics Netherlands.

Our secondary outcome indicators are (1) job search behaviour, (2) societal participation and social activation, (3) participants’ health and well-being, (4) participants’ satisfaction with the scheme, and (5) recipients’ financial situation.

Job search behaviour is measured through self-reported survey data (e.g. number of applications). Societal participation and social activation is measured through self-reported survey data (e.g. hours of volunteer work). Participants’ health and well-being is measured using medical data from Statistics Netherlands (e.g. drug use) and self-reported survey data (e.g. Mental Health Inventory 5). Participants’ satisfaction with the scheme is measured through self-reported survey data. Participants’ financial situation is measured using administrative data and self-reported survey data.

The experiment is a cooperation between Utrecht University School of Economics and the Municipality of Utrecht. Participating welfare claimants are randomly assigned to the four different conditions described above. The target population for the experiment consists of all welfare claimants in Utrecht, unless participating is a risk or not possible. Claimants were invited to participate during a large-scale information and enrollment campaign. Participation had to be confirmed by signing an informed consent sheet. Participation is voluntary, which means that participants can decide to drop out at any time. In total, 752 claimants have signed up to take part in the study, which makes up for around 10% of the targeted population. Data is collected using administrative data as well as surveys. Data collection takes place at three points of time: Mar – May 2018 (baseline), Jan – Feb 2019 (midline) and Aug – Sep 2019 (endline).

Stratified randomization. Variables to form strata were claimants’ distance to the labour market as diagnosed by the welfare agency (four categories ranging from small to large) and living situation (two categories: living alone and living together). Per stratum units were assigned to each of the four groups with equal probabilities. Misfits were randomly allocated within each stratum. Randomization was performed using the user-written RANDTREAT module in STATA version 14.

Individuals

No

0

752 welfare claimants

Treatment 1: 189 claimants; Treatment 2: 188 claimants; Treatment 3: 187 claimants; Control: 188 claimants.

The goal was to assign at least 176 individuals to each group. The minimum sample size was chosen based on a power test, which showed that a sample of 352 (control + treatment condition) is sufficient to detect effect sizes of 0.3 SD (two-tailed) with an alpha of 5% and a power of 80%.

Pre-analysis plan What works study