What do bankrupcty prediction models tell us about banking regulation? Evidence from statistical and learning approaches - Archive ouverte HAL
Preprints, Working Papers, ... Year : 2021

What do bankrupcty prediction models tell us about banking regulation? Evidence from statistical and learning approaches

Pierre Durand
  • Function : Author
  • PersonId : 1264973
Gaëtan Le Quang

Abstract

Prudential regulation is supposed to strengthen financial stability and banks' resilience to new economic shocks. We tackle this issue by evaluating the impact of leverage, capital, and liquidity ratios on banks default probability. To this aim, we use logistic regression, random forest classification, and artificial neural networks applied on the United-States and European samples over the 2000-2018 period. Our results are based on 4707 banks in the US and 3529 banks in Europe, among which 454 and 205 defaults respectively. We show that, in the US sample, capital and equity ratios have strong negative impact on default probability. Liquidity ratio has a positive effect which can be justified by the low returns associated with liquid assets. Overall, our investigation suggests that fewer prudential rules and higher leverage ratio should reinforce the banking system's resilience. Because of the lack of official failed banks list in Europe, our findings on this sample are more delicate to interpret.
Fichier principal
Vignette du fichier
WP_EcoX_2021-2.pdf (3.02 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04159762 , version 1 (12-07-2023)

Identifiers

  • HAL Id : hal-04159762 , version 1

Cite

Pierre Durand, Gaëtan Le Quang. What do bankrupcty prediction models tell us about banking regulation? Evidence from statistical and learning approaches. 2021. ⟨hal-04159762⟩
13 View
23 Download

Share

More