University of Groningen
Methods for High-Dimensional Data in Econometrics
Koning, Nick
DOI:
10.33612/diss.168716962
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Koning, N. (2021). Methods for High-Dimensional Data in Econometrics. University of Groningen, SOM research school. https://doi.org/10.33612/diss.168716962
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
558408-L-sub02-st-Koning 558408-L-sub02-st-Koning 558408-L-sub02-st-Koning 558408-L-sub02-st-Koning Processed on: 31-3-2021 Processed on: 31-3-2021 Processed on: 31-3-2021
Processed on: 31-3-2021 PDF page: 1PDF page: 1PDF page: 1PDF page: 1
Propositions
accompanying the PhD thesis
Methods for High-Dimensional Data in Econometrics by Nick W. Koning
1. Employing regularization in the presence of a sum restriction on the parame-ters in a linear model should be done with caution, as this can interfere with the interpretation of the regularization and may also lead to more complica-ted computation (Chapter 2).
2. Mixing `0- and `1-regularization in unit-sum regression leads to substantially
sparser solutions than `1-regularization, and can guarantee a unique solution
that may not exist when only using `1-regularization (Chapter 2).
3. In a ‘many moment inequalities’ setting, ignoring all but the largest t statis-tic leads to a test with low power against alternatives with multiple violated moment inequalities. This can be improved by maximizing over additional t statistics corresponding to moment inequalities that are implied by the origi-nal set of inequalities (Chapter 3).
4. In a ‘many moment inequalities’ setting, a test on symmetry randomization may lead to considerable performance improvements compared to bootstrap methods in small samples (Chapter 3).
5. Compared to a quadratic statistic, a test based on a conic statistic requires a weaker existence condition at the cost of dependence of the alternative on the covariance structure of the parameter estimators (Chapter 4).
6. With a well-constructed cone, one can focus the power of a test based on a conic statistic towards a specific subset of the alternatives (Chapter 4). 7. The asymptotic analysis of tests, where the level converges to zero is
unde-rappreciated in the current literature.
8. The noise produced by an office as a function of the number of its number of occupants has a local maximum at 2.
9. “If you chase two rabbits, you will lose them both.” – Erasmus in Adagia.
// De magenta omlijning geeft de netto maat aan en zal niet zichtbaar zijn in het eindproduct // // Let op: Dit proef bestand is niet geschikt om correcties in te maken //