From my casual observation, it appears that in countries where the unemployment rate is rather high, unemployed workers tend to rely more on government run employment offices to find jobs. I do not think there is causation in this correlation though. But this makes it more important to understand whether employment office are good at anything.
Jens Hainmueller, Barbara Hoffmann, Gerhard Krug and Katja Wolf discuss an experiment in Germany where the case load of some employment officers was halved, which would be equivalent to doubling the number of employment officers. Does this help in increasing the placement of the unemployed? The above four claim that yes: the unemployment rate is reduced and so is the number of people registered in employment offices. But these results are obvious, except maybe that they are statistically very significant. What matters more is that they are economically significant.
Is a 0.5%-point reduction in the unemployment rate economically significant? Probably yes. Is it worth the cost? Remember, you double the number of case workers and you increase taxes to pay them. In fact, some of the unemployed may have been hired as case workers. It is much less obvious that the experiment is positive look at it this way.
This is what should have been studied. I am too often frustrated by studied that just look at the statistical significance of an effect, ignoring the economic significance and especially how this fits in the rest of the economy. And this happens too often with labor economics papers. If there is one thing that distinguishes economists from other social scientists is that we tend to think more in a general equilibrium way, and this papers for sure does not do justice to economics.