One main strand concerns how what happens to parents reaches their children. With Lorenzo Cappellari and Marco Ovidi, I have studied what a plant closure does to the children of the people who worked there. What proved decisive was the timing: if a parent loses their job while the child is an infant, it still shows up in the final examinations of compulsory school, whereas a job loss later in childhood leaves far less trace. Families that cannot borrow their way through are the ones who carry it. A temporary loss of income becomes a permanent difference in education.
The same authors, with Matteo Sandi, have put the question for a different kind of shock: what happens to children when a parent becomes the victim of a crime? This is the first study to establish a causal effect. It is violence, not theft, that reaches the children, and again early childhood is the vulnerable period. The route runs through the home: families become less stable, and fathers' behavior and attachment to work deteriorate. Crime carries a cost in the next generation that no victim statistic records.
To see what schools can do about this, I studied trials. With Jens Dietrichson, Mette Friis-Hansen and Vibeke Myrup Jensen, I ran an experiment across 81 schools and more than 2,000 pupils, in which the weakest readers in the first year of school were given fifteen-minute tutoring sessions three or four times a week for ten weeks, delivered by the schools' own staff. Decoding improved markedly, and the more interesting result was that classmates who received no tutoring improved as well. The arithmetic of scaling the program up therefore looks better than previously assumed, because the gain is larger than the one the participants themselves receive.
Institutions do not always work the way they appear to. Conscription falls at the point in life when young men are choosing between education and work, and men who serve score worse on skills tests than men who do not. But because service is allocated by lottery, the two can be told apart, and with Stéphanie Vincent Lyk-Jensen and Anders Rosdahl I have shown that the picture then reverses: service itself improves literacy and numeracy. The effect only becomes visible about ten years later, because service changes what the men do afterward; those with middling and low test scores in particular go on to vocational training rather than higher education.
Something similar holds for taxation. With Gauthier Lanot, I have measured how much taxable income responds to tax changes across forty years and six tax regimes. Ordinary employees barely respond; the self-employed respond appreciably. The difference is not that employees are more sluggish or more dutiful, but that their income is reported by their employer, leaving nothing to move. How much people respond to tax therefore depends on what they can move. That explains why calculations from individual data and from aggregate data have disagreed for so long.
The last stretch of the life course has occupied me for twenty years. With Nabanita Datta Gupta, Malene Kallestrup-Lamb and Peder J. Pedersen, I have produced the Danish analyses for an international collaboration on social security and retirement, in which researchers in a dozen countries have measured the same things the same way since the 1990s, so that differences between countries can be attributed to policy rather than to accounting. The conclusion after two decades is unambiguous: political choices shape the financial incentive to keep working at older ages, and those incentives move when people actually stop. The share of men over 65 in work fell through most of the twentieth century even as life expectancy rose. That trend was not inevitable.
How long people live, by contrast, is more unequally distributed than usually assumed. Inequality in life expectancy is normally measured on a single indicator, most often income or education. With Claus Thustrup Kreiner and Benjamin Ly Serena, I have instead combined education, income, occupation, and the intelligence test taken at the draft board, following the 1942-44 cohorts from age 40. Measured that way, the top lives almost 25 years longer than the bottom, and the social gradient is between 50 and 150 percent steeper than a single indicator shows.
The foundation is the Danish administrative registers, which follow the whole population year by year and link individuals to parents, siblings, children, and employers. This makes it possible to ask questions that require both long time series and large numbers, such as how a child's education is faring fifteen years after a parent lost their job.
Registers alone do not settle what causes what. So the work rests on situations where something was allocated by chance: the conscription lottery, plant closures that hit some employees and not others, and trials designed from scratch, as in the classroom reading experiments. Where none of that is available, affected families are compared with closely matched families who were not affected. The difference between the obvious comparison and the one that holds up is often the whole result. With conscription, the two point in opposite directions.
I am a professor in VIVE's Health and Later Life department in Copenhagen. My early work was in the United Kingdom, on how the tax and transfer system affects employment among lone mothers and other low-income families. Since the late 1990s, the research has been based in Denmark. I have been affiliated with Aarhus University and was professor at SFI, The Danish National Center for Social Research, until SFI became part of VIVE in 2017. The work has appeared in journals including the Review of Economics and Statistics, the Journal of the European Economic Association, and the Journal of Labor Economics.
