Invitation to PhD defense: Partial Identification

On Monday 24 March 2014 Lukáš Lafférs will hold a trial lecture on a prescribed topic and defend his thesis for the PhD degree at NHH.

13.03.2014 - Ed.

The thesis analyzes situations in which the economic model together with the data do not uniquely determine the quantity of interest. In such situations there may exist a set of values of the parameter that are consistent with both the economic model and with the data. Then the quantity of interest is only partially identified.

The model leads to sharp conclusions based on strong assumptions, but these assumptions often cannot be justified on economic grounds. The partial identification literature helps to free the analysis from such assumptions that are made only due to technical convenience and that are not derived from economic theory. This ultimately leads to a more credible inference. The partial identification literature addresses many important economic problems including identification of treatment effects, missing data, interval data or censored data. It also provides a way to do sensitivity analysis or handle multiple equilibria in game-theoretic applications.

The thesis consists of four single-authored papers.

The first chapter presents a novel identification framework that enriches the class of models with partially identified parameters. The second chapter compares various methods for statistical inference for partially identified scalar parameter. A Monte Carlo simulation study leads to some practical recommendations for choosing the right method in economically relevant scenarios.

The third chapter demonstrates the usefulness of the presented identification method from the first chapter in an application of the effect of parental schooling on child schooling.

The presented method provides a way to conduct a sensitivity analysis for the identifying assumptions and missing data. The fourth chapter is a note that uses the developed method to point out that conditional and unconditional identifying assumptions are often confused in the applied literature.

Prescribed topic for the trial lecture:
"The roles of nonparametric identification and economic theory for applied work in economics".

Time of the trial lecture:
10.15 - 11.00 in Karl Borch's Auditorium, NHH

Title of the thesis:
"Essays in Partial Identification"

Time and place for the defense:
12.15 - in Karl Borch's Auditorium, NHH

Members of the evaluation committee:
Associate Professor Christian N. Brinch, Department of Economics, BI (Norwegian Business School)
Associate Professor Adam Rosen, Department of Economics University College London
Professor Erik Ø Sørensen, Department of Department of Economics (NHH)

Supervisor:
Professor Gernot Doppelhofer Department of Economics (NHH)

The trial lecture and thesis defense will be open to the public. Copies of the thesis will be available from: [email protected]