From closed form to black box? a comparison of debiased vs automatic debiased machine learning in the sharing economy

Simple item page
dc.contributor.MentorSosa Escudero, Walter
dc.creator.AutorQuispe Rojas, Anzony
dc.date.accessioned2025-10-17T14:26:55Z
dc.date.available2025-10-17T14:26:55Z
dc.date.issued2025-10
dc.descriptionFil: Quispe Rojas, Anzony. Universidad de San Andrés. Departamento de Economía; Argentina.
dc.description.abstractEstimation of causal parameters in the context of high dimensional parameters (đť‘ť > đť‘›) can be challenging. Debiased Machine Learning (DML) method solves this issue adding machine learning models in the estimation of parameter controlling regularization and overfitting bias using Neyman Orthogonal scores and cross fitting. However, this method needs to derive the proper equation function for the specific low dimensional parameter: policy variable, continues variable, etc. Automatic debiased machine learning (ADML) skip this step by approximating the explicit form using available data. This paper (i) briefly describe the statistical differences between DML and ADML, (ii) illustrates the asymptotic and finite-sample properties of both methods in two synthetic settings: regular conditions and high omitted variable probability conditions and (iii) applies both methods to Airbnb listings from Ciudad de Buenos Aires, Argentina, to estimate the effect of Superhost status on listing prices, and contrast this result with other cities (New York City, Mexico DF, and Rio de Janeiro). I find that both estimations (DML and ADML) are the same in terms of sign and statistical significance with very low discrepancies in the magnitude.
dc.formatapplication/pdf
dc.identifier.urihttps://repositorio.udesa.edu.ar/handle/10908/25840
dc.languageeng
dc.publisherUniversidad de San Andrés. Departamento de Economía
dc.rightsinfo:eu-repo/semantics/openAccess
dc.rightshttps://creativecommons.org/licenses/by-nc-nd/4.0/
dc.titleFrom closed form to black box? a comparison of debiased vs automatic debiased machine learning in the sharing economy
dc.typeTesis
dc.typeinfo:eu-repo/semantics/masterThesis
dc.typeinfo:ar-repo/semantics/tesis de maestrĂ­a
dc.typeinfo:eu-repo/semantics/updatedVersion
Files
Original bundle
Thumbnail Image
Name:
[P][W] T. M. Eco. Quispe Rojas, Anzony.pdf
Size:
4.02 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Tesis de MaestrĂ­a en EconomĂ­a