Or why you should take courses that are not for credit and do presentations in class.
In the spring of 2015, I followed a course in microeconometrics at the Department of Economics at the University of Oslo to learn some more econometrics. As part of the course, the students had to present published papers employing the methods from the course. I volunteered to do two such presentations, that both ended up as papers of my own!
For my first presentation, which was on the synthetic control method (SCM), I asked the instructor for that lecture, Tarjei Havnes, if I in stead could present an application of my own. I did not know much about the SCM from before, but in the lecture I noticed that it would fit very well to a recent sickness absence reform in a Norwegian region I had head about from Knut Røed, a colleague at the Frisch Centre. The fact that the seminar schedule was on a few weeks lag from the lectures gave me some time to implement a basic analysis of the reform and put together a presentation. The subsequent positive feedback in the seminar motivated me to develop it into a proper paper, which is now just published in the Journal of Health Economics. The paper if of course much extended since that seminar, but the core remained the same.
The reform in question was a program undertaken by the Norwegian region of Hedmark in 2013. It was aimed at strictly enforcing a requirement that people on long-term sick leave be partly back at work unless explicitly defined as an exception. I found that the reform reduced sickness absenteeism by 12% in the reform region compared to a comparison unit created by a weighted average of similar regions. Thus, making use of the partial work capacity of temporary disabled workers has the potential to reduce long-term absenteeism and bring down social security costs.
A key graph is below, showing how actual absenteeism in Hedmark (solid line) after the reform diverged from the estimate of absenteeism in the absence of reform (dashed line):
Fig. 1. Trends in the sickness absence rate in Hedmark and the synthetic control region. Note: The dotted line at the fourth quarter of 2011 indicates the final quarter of the matching period. The dashed line at the second quarter of 2013 indicates the period in which the activation program was introduced.
The effect is driven by both increased part-time presence of temporary disabled workers and accelerated recovery. Musculoskeletal disorders was the diagnosis group declining the most. I conclude that such an activation strategy represents an alternative to traditional attempts at welfare reform involving stricter screening or reductions in generosity, and may be more compatible with already existing legislation and obligations, as well as easier to find support for across political priorities.
The paper is freely available for a month here.
Pingback: New paper: “Distributional effects of welfare reform for young adults: An unconditional quantile regression approach” | Ø-blog