An Evaluation Of The Survival Model's Contribution To Thrift Institution Distress Prediction

A growing body of literature suggests that survival-time methodologies can improve ex-ante and ex-post assessments of firm survival in financial institutions. Yet none of these studies (1) compare the prediction accuracy of survival models with the more commonly used conditional probability methods,...

Full description

Saved in:
Bibliographic Details
Published in:Journal of managerial issues 2001-12, Vol.13 (4), p.401-417
Main Authors: Catanach, Anthony H., Perry, Susan E.
Format: Article
Language:English
Subjects:
Accounting
Accounting policies
Accuracy
Bank failures
Banking regulation
Bankruptcy
Classification
Cost control
Discriminant analysis
Event history analysis
Fees
Financial institutions
Financial performance
Generally accepted accounting principles
Industrial regulation
Insurance regulation
Logistics
Losses
Modeling
Parametric models
Research design
Research methodology
Savings & loan associations
Savings and loan associations
Statistical methods
Studies
Thrift
Time
Time series
Variables
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A growing body of literature suggests that survival-time methodologies can improve ex-ante and ex-post assessments of firm survival in financial institutions. Yet none of these studies (1) compare the prediction accuracy of survival models with the more commonly used conditional probability methods, (2) evaluate predictive ability using multiple measures of distress, or (3) consider the effect of misclassification costs. This study compares the performance of fully parametric survival-time models with that of probit models for variable identification and distress prediction in the savings and loan industry. The empirical results reveal that survival-time models can identify distressed institutions with greater accuracy than probit models. Moreover, these findings are robust to alternate definitions of distress. This study's results suggest that survival-time models may be a valuable supervisory tool for discovering signs of financial institution distress, particularly if a broad definition of distress is warranted or misclassification costs are a concern. Consequently, regulators may wish to complement their use of traditional probit models with survival methodologies to improve their early identification of distressed financial institutions, a function critical to the maintenance of stability in the banking system.
ISSN:1045-3695
2328-7470