 So in the paper, I ask for the role of case workers in unemployment insurance. Finding a new job when being unemployed is not easy and that's why most countries offer individual support through case workers. This support usually works through personal meetings that take place on a regular basis. And in these meetings, case workers support and check on the job search process of unemployed workers. In the paper, I seek to quantify the economic value of these meetings. In the first step, I ask how successful these meetings are in improving job finding. So what is the effect of a case worker meeting on the average length of unemployment spells? Second, I ask for the role of the individual case workers. So as an individual job seeker, does it make a difference whether you're assigned to case worker A or case worker B and how can we measure the extent of this difference? So to answer these questions, I need a lot of data on case workers and their assigned job seekers. What I use is administrative data from the Swiss Unemployment Insurance Registers and these data are very well suited for my purposes because they report the universe of interactions that take place between case workers and job seekers. Now even with such exhaustive data, it's not trivial to estimate the causal effect of a case worker meeting on the aduration of unemployment. And the main challenge is that it's not random how much a job seeker meets his or her case worker. For example, we observe that job seekers who face more difficulties also tend to have more case worker meetings. And as a result, the relationship between the number of meetings and the actual duration of unemployment is unlikely to be causal. The way how I address this challenge is by exploiting the incidence of unplanned case worker absences. More precisely, I use incidences where case workers plan to come to work and scheduled meetings, but had to cancel them on a short notice. And as a result of these cancellations, some job seekers unexpectedly see their meeting frequency reduced. I exploit these cancellations to estimate how much unemployment spades get longer when a meeting had to be cancelled. And this then indirectly informs us about the value of a case worker meeting that takes place. So the first question was how much does a case worker meeting on average affect the duration of unemployment spades? So in my data, I observe that the average job seeker has about three meetings during the first six months of unemployment. And what I find is that when one of these three meetings is lost due to a case worker absence, the unemployment spell gets 12 days longer on average. This corresponds to an effect of about 5%. So this means that clearly these case worker meetings, they do matter for the duration of unemployment. What we can also try to do now is to calculate how much the welfare state saves in terms of benefit payments and gains in terms of additional tax revenues if one of these additional meetings takes place. This calculation is very much back of the envelope. It's based on some assumptions, but still it gives us a little bit of a sense how much these case worker meetings matter in monetary terms. What I find when doing this calculation is that one meeting gains roughly 1,800 Swiss francs on average. Again, that's quite a simple calculation, but it shows that indeed these case worker meetings matter and it's quite remarkable because the average case worker meeting only lasts about 40 minutes. Now the second question was how important is the individual case worker? And my results show that they are quite important because I find that the effect of these absences is very heterogeneous. So if some of these case workers are absences, then this prolongs the duration of unemployment even more than these 12 days, up to about 30 days on average. But if some of some other case workers are absences, we do not find any effect on the duration of unemployment. What this suggests is that some of the case workers are very successful in bringing people back into employment while others do not have such a large impact. What I also find is that the differences in the impact of case workers cannot be explained by differential use of certain program assignments. For example, I do not find that more successful case workers are more likely to place their job seekers into certain types of training programs or that they use benefit sanctions more or less often, for example. Instead, the results suggest that these differences are really due more to the nature of the personal interaction between the job seeker and the case worker, which unfortunately we cannot observe in the data. The findings establish that public human resources can have a large economic impact. In the context of unemployment, a natural conjecture is that it can be helpful to invest in these human resources to improve the outcomes of unemployed individuals. In many countries, for example, case workers have a large case load. They are responsible for many job seekers at the same time. Reducing this case load and thereby allowing case workers to have more time for a job seeker might be an investment that is likely to pay off. One may also argue based on my findings that case workers are largely underpaid given the economic value they generate. More generally, my findings suggest that any program or intervention can only be as successful as the persons who provide it. This might also relate to other contexts where a lot of personal interaction takes place, such as mentoring programs or counseling programs. The paper establishes that case workers matter, but it does not answer the question of what makes a good case worker. So that's clearly a question that needs to be addressed in the future. Colleagues of mine from Sweden have done a first step in a new yet unpublished study where they relate the success of a casework to a lot of characteristics that they observe in the data, such as cognitive ability, educational background or the case worker's own experience with being unemployed. And surprisingly, they find that none of these characteristics is able to explain the success of a case worker. So we have to say that we do not yet know what makes a case worker successful. From a policy perspective, the most pressing questions are probably what can you do to attract the right case workers into the profession? What is, for example, the role of pay in this regard? At the same time, we also want to understand what policymakers can do to empower their existing case workers. How can they, for example, invest into the skills of case workers and what are the skills that matter? What are, for example, the importance of technical versus social skills? These are questions which hopefully research will be able to address in the future.