Causal inference
The group works on the identification and estimation of causal effects when assignment is not random. Much of the methodological output concerns instrumental variables: whether an instrument is valid and how that can be tested against the data rather than defended in an appendix, which LATE-type estimands remain identified once monotonicity is relaxed or limited, sharp bounds where point identification fails, and inference that stays valid when the instruments are many or weak.
A second strand concerns what happens after identification. Causal mechanisms separate the direct effect of an intervention from the part that runs through a channel, and the same idea carries into the synthetic control design. Effect heterogeneity and policy learning ask a different question again: once the budget binds and not everyone can be treated, who should be.
Key research areas
| Instrument validity and partial identification Testable implications of the LATE assumptions, sharp bounds on average effects under endogeneity and noncompliance, and bounds under sample selection. The common thread is that an assumption which the data can refute is worth more than one that it cannot. |
| Many and weak instruments Inference that remains valid when the number of instruments grows with the sample and the errors are heteroskedastic, including jackknife approaches. |
| Mechanisms, synthetic control and heterogeneity Separating direct effects from those running through a mechanism, the finite-sample behaviour of the estimators involved, subgroup effects with valid inference, and optimal allocation of an intervention under a capacity constraint. |
Who works on this
| Christian Møller Dahl Professor |
Applies machine learning to digitised historical records, and works on time series econometrics, quantile regression and causal inference. |
| Alexander Klein Professor |
Works on economic geography, agglomeration and long-run growth,and on spatial econometric methods for historical data. |
| Jørgen T. Lauridsen Professor |
Works on health econometrics and spatial econometrics, studying social and geographic inequality in health and health-care use. |
| Giovanni Mellace Professor |
Develops methods for identifying and estimating causal effects, much of it on instrumental variables, and applies them in empirical labour and health economics. |
| Volha Lazuka Associate Professor |
Studies the long-run effects of health and social policies on health, earnings and mortality across the life course. |
| Seetha Menon Associate Professor |
Studies how health shocks, job insecurity and domestic violence affect health, with a focus on gender differences |
| Torben Skov Dyg Johansen Assistant Professor |
Works on machine learning for digitising historical records and on causal methods for allocating an intervention when capacity is limited. |
| Riccardo Di Francesco Postdoctoral Researcher |
Develops machine learning methods for estimating and interpreting heterogeneous causal effects, with applied work on worker health. |
Grants in this area
| Project | Funder | Grant recipient |
|---|---|---|
| Causal Inference with Multiple IVs: Using Mendelian Randomization to Estimate the Causal Effects of Cannabis and Alcohol on Labour Productivity and of Creativity and Risk Taking on Entrepreneurship Develops instrumental variable methods for settings with several instruments, and applies them with genetic variants as instruments in the IPSYCH cohort. The two applications estimate the effects of cannabis and alcohol consumption on schooling, earnings and health, and the effects of creativity and risktolerance on entering entrepreneurship. |
DFF | Giovanni Mellace 4.77M DKK |
| Should We Tax Unhealthy Food Consumption? | DFF | Giovanni Mellace 1.9M DKK |
| Intended and Unintended Effects of the Danish Fat Tax (Påtænkte og utilsigtede virkninger af den danske fedtafgift) | Helsefonden | Giovanni Mellace 200,000 DKK |