Time to default in credit scoring using survival analysis: a benchmark study

Abstract

We investigate the performance of various survival analysis techniques applied to ten actual credit data sets from Belgian and UK financial institutions. In the comparison we consider classical survival analysis techniques, namely the accelerated failure time models and Cox proportional hazards regression models, as well as Cox proportional hazards regression models with splines in the hazard function. Mixture cure models for single and multiple events were more recently introduced in the credit risk context. The performance of these models is evaluated using both a statistical evaluation and an economic approach through the use of annuity theory. It is found that spline-based methods and the single event mixture cure model perform well in the credit risk context.

Computational approaches and data analytics in financial services: A literature review

“…Finally, it is worth noting that while most of the above approaches mostly follow a static approach providing risk estimates for a fixed time period, another line of research has adopted models that incorporate dynamic characteristics. Typical examples include survival and hazard models that consider time-varying variables and enable the modeling of the time to default (Bellotti & Crook, 2009aCrook & Bellotti, 2010;Dirick, Claeskens, & Baesens, 2017;Serrano-Cinca, Guti errez-Nieto, & L opez-Palacios, 2015), whereas credit rating migration (i.e., the dynamics of credit ratings) is commonly analyzed through Markov models (Baena-Mirabete & Puig, 2017;D'Amico, Janssen, & Manca, 2016;Quirini & Vannucci, 2014).…”

Survival Regression Models for Single Events and Competing Risks Based on Pseudo-Observations

“…The application of the competing-risks methodology to credit-risk assessment is quite a recent idea (c.f. Watkins et al, 2014); it is more common to use singleevent models (see Dirick et al, 2017). In the single-event approach, only time to default is considered, whereas credits that do not default until data-gathering are censored observations.…”

Fuzzy Relational Model to Establish Credit Worthiness of Sacco Members in Kenya

“…The approach makes the model developer at least the only one knowledgeable about the operation of the system which consequently shuns off third party users. This can further be explained by the steepness of the learning curve required to learn the operations of system, a fact that has been pointed out as a major limitation of current computational models [6]. The problem even becomes magnified in cases where such models must be used by less formal creditors like Sacco's which have members that may be less knowledgeable of the computational complexity that comes with jargon linked to the various methods.…”