Webbläsaren som du använder stöds inte av denna webbplats. Alla versioner av Internet Explorer stöds inte längre, av oss eller Microsoft (läs mer här: * https://www.microsoft.com/en-us/microsoft-365/windows/end-of-ie-support).

Var god och använd en modern webbläsare för att ta del av denna webbplats, som t.ex. nyaste versioner av Edge, Chrome, Firefox eller Safari osv.

Estimating Heterogeneous Panel Data Models


Summary, in English

This thesis deals with the development and application of new estimation methods of heterogenous panel data models.

In Paper I, a new estimator for heterogeneous panel data models with random interactive effects is proposed. The heterogeneity in this paper is viewed as heterogeneity over time, which is modeled by having the slope coefficients exhibit multiple structural breaks. The suggested estimator is suitable when the number of time periods, T, is fixed, and only the number of cross-sectional units, N, is large. To estimate the multiple structural breaks, I suggest minimizing a penalized objective function that induces structural breaks.

In Paper II, I discuss the estimation of what I call "Coefficient-by-Coefficient" breaks. Existing econometric methods take an all-or-nothing approach when estimating structural breaks, in the sense that either all parameters shift together or not. However, we typically do not know which parameters are shifting and when. To address this, I suggest a penalized estimator that allows for the estimation of breaks in each component of the slope vector, providing further insight into what is breaking and when. In the same paper, I propose two estimators: one that accounts for homogeneous breaks and one for heterogeneous breaks. Heterogeneous breaks are breaks that vary across different groups. Hence, the considered heterogeneity is very general in the sense that the slope coefficient changes over time, but also over cross-sectional units.

Paper III is concerned with the robustness of pooled estimators to random breaks in panel data models. The main point of this paper is to showcase that the least square estimator is not necessarily consistent under random breakpoints.

In Paper IV we discuss the CCE estimator of Pesaran (2006). In this paper, we show that this estimator is more useful than commonly appreciated, in that it enables consistent and asymptotically normal estimation of interactive effects models with heterogeneous slope coefficients when only the number of cross-sectional units, N, is large.








Media-Tryck, Lund University, Sweden


  • Economics


  • Heterogeneity
  • Interactive effects
  • Latent groups
  • Panel data
  • Penalized estimators
  • Structural breaks
  • Lasso




  • ISBN: 978-91-8104-036-4
  • ISBN: 978-91-8104-035-7


23 maj 2024






  • Rickard Sandberg (Associate Professor)