The Aarhus Center for Econometrics (ACE) is a research center focusing on the development of econometric theory and methodology. ACE is funded by a Center of Excellence grant from the Danish National Research Foundation of DKK 36m with an additional DKK 23.5m in co-financing from Aarhus University. The center opened on March 1, 2025, for an initial period of 6 years.
The ACE logo is a combination of the letters A, C, and E, using the mathematical symbols for set intersection and inclusion. The intersection of sets consists of their common elements, and here it symbolizes our commitment to the common values and goals of the center. Set inclusion represents a part of a greater whole and symbolizes our values of inclusion and belonging.
ACE will pioneer the development and analysis of a new generation of modern and robust econometrics; specifically, methods that are robust to nonconformity with classical paradigms and assumptions. We will broaden the scope of robust methods to more advanced and relevant models that are used in empirical studies, such that researchers may realize the full potential of modern data. This new generation of econometrics will have wide-ranging impact on empirical practice and will bring a new and improved understanding of data and the world it describes. Three examples of ACE research pillars are given below.
Modern datasets can have thousands or more variables for each unit (individual, firm, etc). Such high-dimensional datasets allow detailed empirical analysis. However, classical econometric methods rely on many observed units and few variables per unit, so the number of parameters to estimate is small relative to the sample size. These methods do not apply in high-dimensional settings, where the number of parameters is large, which necessitates development of new techniques.
Most non-experimental data involve unobserved dependence among observations. If this is not taken into account, severe errors of inference are likely. It is commonly assumed that data fall naturally into clusters (sometimes the division can be inferred from the data or design), and that data are dependent only within clusters. Cluster-robust methods date back to the 1980s and have transformed empirical research in economics and related fields. However, their theoretical properties are largely unknown except in special cases derived recently. Much work remains to give an econometric foundation for many relevant models.
There is increasing awareness in economics that standard statistical methods can be very sensitive to outliers. With high-dimensional data or clustering, detecting outliers is even more challenging. Robust methods exist, but their properties are not well developed.
