The Determinants of CDS Spreads: Evidence from the Model Space

Pelster, Matthias; Vilsmeier, Johannes

doi:10.2139/ssrn.2802316

Cited by 4 publications

(7 citation statements)

References 136 publications

(133 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, instead of deriving the posterior model probabilities and the implied BIC model weights based on this marginal likelihood, we follow Burnham and Anderson (2002) and instead use model weights based on a smoothed AIC, which induces model weights δ i that are proportional to the probability of being the best Kullback and Leibler (1951) model in repeated samples (see also Pelster and Vilsmeier, 2017, for a discussion). Then,…”

Section: Multicollinearity Filtermentioning

confidence: 99%

See 1 more Smart Citation

On a Quest for Robustness: About Model Risk, Randomness and Discretion in Credit Risk Stress Tests

Siemsen

Vilsmeier

2018

SSRN Journal

Self Cite

View full text Add to dashboard Cite

Research QuestionFollowing the financial crisis, bank-internal and supervisory stress testing has gained importance in risk management and supervisory practice. As in any exercise with scarce data and highly collinear regressors, stress testing is susceptible to model uncertainty. Given the prominent role of stress tests, quantifying the extent of model uncertainty and developing tools to mitigate its effects on stress test results is of ample importance for both banks and supervisors. This paper takes a first step in trying to address both of these challenges. ContributionOur contributions are twofold: first, we quantify the impact of model uncertainty on credit risk stress test results using the German banking sector as an example. We define "model uncertainty" as the range of stress test results that are supported by equally plausible model frameworks. Second, we propose a stress testing framework that mitigates the impact of model uncertainty by considering a battery of different models rather then just one model, filtering the models for statistical, economic and "stress testing" plausibility and finally combining the surviving specifications into a single model. ResultsWe find that -depending on model specification -the dispersion in predicted increases of default probabilities during a stress horizon can be huge. In our application they lie in a range from −90 % to +7, 000 %. This leaves stress test results subject to significant model uncertainty. Our proposed framework eliminates "stress test implausible" predictions (relative to a second benchmark model) at both ends of the stressed forecast distribution. This leads in our application to the German banking sector to a reduction in the median stress effect, compared to the model without the "stress testing plausibility" filter, from −5.0 pp of CET1 capital to −2.5 pp.

show abstract

Section: Multicollinearity Filtermentioning

confidence: 99%

“…In an application of the general methodology proposed in this paper,Siemsen and Vilsmeier (2017) conduct a top-down stress test of the German residential mortgage market Pelster and Vilsmeier (2017). apply a similar framework to the market for CDS.…”

mentioning

confidence: 99%

On a Quest for Robustness: About Model Risk, Randomness and Discretion in Credit Risk Stress Tests

Siemsen

Vilsmeier

2018

SSRN Journal

Self Cite

View full text Add to dashboard Cite

show abstract

“…Burnham and Anderson (2002) provide a Bayesian justification for AIC (with a different prior over the models than the BIC approximation) and suggest the use of AIC-based weights as posterior model probabilities. The smoothed AIC approximation is used in the context of assessing the pricing determinants of credit default swaps in Pelster and Vilsmeier (2018) 59 . 58 Regular models are such that the sampling distribution of the maximum likelihood estimator is asymptotically normal around the true value with covariance matrix equal to the inverse expected Fisher information matrix.…”

Section: Approximations and Hybridsmentioning

confidence: 99%

Model Averaging and its Use in Economics

Steel¹

2017

Preprint

View full text Add to dashboard Cite

The method of model averaging has become an important tool to deal with model uncertainty, for example in situations where a large amount of different theories exist, as are common in economics. Model averaging is a natural and formal response to model uncertainty in a Bayesian framework, and most of the paper deals with Bayesian model averaging. The important role of the prior assumptions in these Bayesian procedures is highlighted. In addition, frequentist model averaging methods are also discussed. Numerical methods to implement these methods are explained, and I point the reader to some freely available computational resources. The main focus is on uncertainty regarding the choice of covariates in normal linear regression models, but the paper also covers other, more challenging, settings, with particular emphasis on sampling models commonly used in economics. Applications of model averaging in economics are reviewed and discussed in a wide range of areas, among which growth economics, production modelling, finance and forecasting macroeconomic quantities.

show abstract

“…In these cases, regressor coefficients may be non-unique, since variables catch up on effects of other variables. In such an empirical environment it is highly recommendable to analyze the entire model space rather than picking one model at random or by discretion and neglecting the information from all other models (for another application see Pelster and Vilsmeier, 2017). This is especially true in a stress testing context, where biased estimates can substantially affect stress test outcomes (see Gross and Poblacion, 2017).…”

Section: Pd Modelmentioning

confidence: 99%

“…To economize on computation time we restrict our attention to a subset of the 10,000 best models according to the adjusted R 2 using the leaps and bounds algorithm (see Furnival and Wilson, 1974). Pelster and Vilsmeier (2017) provide additional details.…”

Section: Pd Modelmentioning

confidence: 99%