New DWP study on sanctions may not be all it seems

Today the Department for Work and Pensions (DWP) has released analysis on the impact of universal credit (UC) sanctions on subsequent employment. It concludes that getting sanctioned causes people to stop claiming UC, but reduces the likelihood that they get into paid work. This comment investigates the robustness of this analysis, and briefly summarises what we know about sanctions from other research.

Benefit claimants who are out of work or on low earnings, and who do not have disabilities or certain kinds of caring responsibilities, are required to search for work in order to be eligible for benefits - a policy known as ‘conditionality’. Such claimants must meet with work coaches at a Jobcentre to prove that they are looking for work. To enforce conditionality, the government has a ‘sanctions’ regime, whereby if the work coach deems the claimant’s work search insufficient or if they fail to do other tasks, their benefits can be reduced or stopped entirely for a period. A sanction for a single claimant is £12 per day and in extreme circumstances can last up to six months. In November 2022 (the latest data) around 120,000 individuals on UC were being sanctioned, a figure that has risen steadily since the pandemic when the conditionality regime was essentially suspended.

The DWP study uses administrative data to estimate the impact of sanctions on future employment and benefit claiming of those who were sanctioned. The basic approach is to take claimants who are subject to conditionality, and compare the outcomes of those who get a sanction to those who do not. Following this strategy, they show that those who are sanctioned are more likely to stop claiming UC, but less likely to enter paid work (and therefore those who are sanctioned are more likely to be left out-of-work without benefit support).[1]

For this sort of methodology to reliably indicate the impact of sanctions, it needs to be the case that those who are sanctioned and those who are not sanctioned would have had the same future employment and benefit outcomes as one another, had no sanctions been applied.[2] 

However, this assumption is rather unlikely to be true. Sanctions are not imposed randomly: they are often imposed precisely because the work coach believes that the claimant’s work search has been insufficient. Clearly, work search efforts are going to influence one’s ability to get a job - and not engaging in work search activity might reflect underlying reasons that make, or at least are associated with, such work search being less likely to be fruitful. So people who get sanctioned would, on average, very likely have had a lower chance of re-entering work even if no sanction had been applied. But in DWP’s analysis, this effect will be attributed as the impact of sanctions themselves. It is therefore not surprising that DWP find that sanctions appear to reduce the likelihood of working.

Indeed, it is clear from the analysis that those that get sanctioned look quite different from those who do not; for example, DWP show that the prior earnings of those who end up getting sanctioned are about half of those that do not get sanctioned. For differences which DWP observe in their data - such as prior earnings, age and sex - they can control for them. But the presence of these sorts of large differences suggests that there are other, unobserved, ways that the sanctioned differ to those who do not get sanctioned (e.g. educational qualifications or caring or other non-work-related priorities). Any differences along these lines which influence employment outcomes will - incorrectly - be attributed to a sanction effect, whereas in reality it is the impact of these wider factors.

Taken together, this suggests that the report’s estimate of the effect of sanctions on future employment outcomes is likely to be mixing together any genuine impact of sanctions, the impact of differences in claimant work search between the sanctioned and non-sanctioned, and the impact of other ways that the sanctioned and non-sanctioned differ. We should therefore not put much store by these results.

The bigger risk with sanctions, which generally remove a large fraction of household income (for a period) from people with already low incomes, is that they create a lot of hardship. This is to be balanced against the intended purpose of conditionality, which is to promote work-related activities. On that point, wider research tends to be relatively clear that imposing sanctions increases, rather than reduces, employment, sometimes by large margins. This is a common finding across research in Europe, seen in one paper in Denmark (Svarer, 2011), two in the Netherlands (van den Berg et al., 2004 and Abbring et al., 2005), two in Germany (van den Berg et al., 2021 and Boockmann et al., 2014), and one in Switzerland (Lalive et al., 2005).[3] The benefit and sanctions regimes in these countries are of course not identical to that in the UK, but it is very notable that the same direction of effect is seen across a number of different countries and studies. It is also worth noting that the direct impact on those who are sanctioned is not the only way in which sanctions policy affects employment. The existence of sanctions helps enforce conditionality, and so encourages those subject to conditionality to search for work and to get into work even if they are never sanctioned themselves. A successful conditionality regime could primarily impact employment by ensuring that people do comply with work search conditions, and find work as a result - since these people would not be sanctioned, analysis such as DWP’s would not pick up this effect (a point noted in the DWP report). Indeed, recent IFS research finds a substantial impact of conditionality on employment, albeit with significant limitations and unintended consequences too. The key question for policymakers is really around whether these impacts justify the downsides of conditionality, rather than whether the effect of sanctions on employment is actually counterproductive.

Footnotes