DataScience@BI seminar with Vladislav Morozov
Deconvolution Estimation and Inference on the Distribution of Marginal Effects.
- Starts:12:00, 14 October 2025
- Ends:13:00, 14 October 2025
- Location:BI - campus Oslo, B3 inner area - next to meeting room B3i-108 or Zoom
- Enrolment deadline:13.10.2025 00:00
DataScience@bi seminar invites Vladislav Morozov, Assistant Professor (Junior Professor) in Econometrics, University of Bonn, to give a talk within the field of econometrics.
Abstract
This paper considers estimation of the distribution of heterogeneous marginal effects using panel data. The setting is general in three key respects: (i) unrestricted time-invariant unobserved heterogeneity in both form and dimension, (ii) nonstationary time-varying unobserved components,, and (iii) flexible time effects in the structural function. Within this framework, we identify the full distribution of marginal effects. Only two periods of data are required. We propose a conditional deconvolution-based nonparametric estimator and derive its convergence rates for both ordinary smooth and supersmooth settings. To enable inference, we further construct uniform confidence bands for the distribution, supported by new theoretical results on conditional deconvolution problems.
Key research areas
Econometrics: econometrics of heterogeneity, panel data econometrics Mathematical statistics
- Read more about the speaker
- Visit our website to read more about the Department of Data Science and Analytics
The DataScience@bi seminar is organized by the Department of Data Science and Analytics, BI Norwegian Business School.