No full text
Unpublished conference/Abstract (Scientific congresses and symposiums)
Variable selection in proportional hazards cure model with time-varying covariates, application to bank failures
Beretta, Alessandro; Heuchenne, Cédric
20171st International Conference on Econometrics and Statistics (EcoSta 2017)
 

Files


Full Text
No document available.

Send to



Details



Keywords :
Finance; Bank failures; Survival Analysis; Mixture Cure model; Time-varying covariates
Abstract :
[en] In the last three decades, as a consequence of failures and corporate actions, the number of commercial banks in the United States has shrunk by two thirds. Empirical evidence in the analysis of bank failures suggests the existence of banks which are not susceptible to default. For this reason, we use a semi-parametric proportional hazards cure model with time-varying covariates to study their effects either on the probability that a bank is susceptible to default and on the survival time of failed institutions. We propose a penalized maximum likelihood method for the selection of the most significant variables, using a Smoothly Clipped Absolute Deviation (SCAD) penalty. A simulation study shows that this procedure performs reasonably well. We apply this methodology to a quite large sample of United States commercial banks insured by the Federal Deposit Insurance Corporation (FDIC) and with more than 50 million dollars of total assets during the last quarter of 2002. More in detail, we use bank-specific covariates observed on a quarterly basis until the end of 2015, that we use as proxies for capital adequacy, asset quality, earnings, management efficiency, liquidity, cost structure and size.
Disciplines :
Quantitative methods in economics & management
Author, co-author :
Beretta, Alessandro ;  Université de Liège > HEC Liège : UER > Statistique appliquée à la gestion et à l'économie
Heuchenne, Cédric ;  Université de Liège > HEC Liège : UER > Statistique appliquée à la gestion et à l'économie
Language :
English
Title :
Variable selection in proportional hazards cure model with time-varying covariates, application to bank failures
Publication date :
June 2017
Event name :
1st International Conference on Econometrics and Statistics (EcoSta 2017)
Event organizer :
Working Group on Computational and Methodological Statistics (CMStatistics), network of Computational and Financial Econometrics (CFENetwork) and Hong Kong University of Science and Technology (HKUST) Business School.
Event place :
Hong Kong, China
Event date :
from 15-06-2017 to 17-06-2017
Audience :
International
Available on ORBi :
since 23 June 2017

Statistics


Number of views
107 (8 by ULiège)
Number of downloads
0 (0 by ULiège)

Bibliography


Similar publications



Contact ORBi