Information sharing, credit booms, and financial stability

by | Jun 3, 2016 | Papers | 4 comments

Samuel Guérineau and Florian Léon

FERDI Working Paper Number 159, July 2016

Global financial crisis has shown the vulnerability of financial systems. It has consequently stressed the need for improving the management of financial vulnerability. The study seeks to improve the understanding of financial vulnerability in low-income countries to provide efficient tools for financial stability. Specifically, it studies the determinants of financial vulnerabilities, measured by annual change of the ratio of non-performing loans to loans, for a sample of 87 developing countries including 25 low-income countries.

Credit Growth and Financial Vulnerability

The financial stability issue is less studied in low-income countries (LICs), insofar as the risks they face are lower than in high and middle income economies. This is due to the smaller size and less interconnected nature of their financial system. However, a better understanding of financial fragility mechanisms in low-income economies remains crucial as they could still suffer sharp increases in non-performing loans and banking crises (see Leaven and Valencia, 2008). Confidence in the banking system in LICs is weak and a banking crisis may permanently hinder the development of banking services demand. As such, improving the understanding of financial vulnerability in LICs to provide efficient tools for financial stability is important. The dynamics of non-performing loans (NPLs), notably the NPL variations, is analyzed to identify the determinants of financial vulnerabilities. Also, it is examined whether the impact of credit growth on financial vulnerability depends on the existence of information sharing on credit.

Findings

The main results from the analysis are as follows:

  • Credit growth is the main driver of financial vulnerability but its effect is mitigated by the presence of credit information sharing.
  • This mitigation effect also exists in low-income countries, even if the credit growth direct effect is smaller.
  • These results are not significantly different for Sub-Saharan African countries. These results are also robust to alternative measures of financial vulnerabilities.

Policy Implications

First, a particular attention should be paid to the NPL variations. Second, the credit growth is a key variable to conduct macro-prudential policies in low and middle-income countries. Third, current efforts to develop information sharing schemes should be strengthened, since the latter allow a credit expansion without excessive increase in the overall credit risk. These results also suggest that the use of credit by sector may provide additional information on the relevant indicators to conduct macro-prudential policies. Early information on the rise of financial vulnerability might be extracted from the sectoral credit growth rate or concentrations of loans.

Downloads

4 Comments

  1. Michael GOUJON

    Bonjour,
    J’ai lu avec intérêt votre papier, car il est rare que la question de la stabilité financière soit traitée en intégrant les pays à faible revenu. J’ai une question sur la mesure d’instabilité financière. Vous utilisez la variation des créances douteuses, ce qui me semble nouveau. Pourquoi ne pas utiliser les indicateurs standard, notamment ceux proposés dans les bases de données financières de la Banque mondiale ou du FMI (Z-score, ratio equity sur assets, etc..)

    In English
    Hello,
    I read with interest your paper, because it is rare that the issue of financial stability is treated by integrating low-income countries. I have a question about the extent of financial instability. You use the change in bad debts, which seems unconventional to me. Why not use the standard indicators instead, including those proposed in the financial databases of the World Bank or IMF (Z -score, ratio of equity assets, etc…)?

    Reply
  2. Florian Léon

    Bonjour Michael,
    merci pour ce commentaire très pertinent. En effet nous avons fait le choix de créer une nouvelle mesure d’instabilité financière plutôt que d’utiliser les indicateurs usuels car chacun des indicateurs usuels souffrent de lacunes. L’indicateur alternatif le plus pertinent pour notre étude aurait sans doute été le Z-score. Le Z-score est un indicateur microéconomique capturant les risques de défaillances d’une banque (distance au défaut). Si cet indicateur est pertinent pour des études microéconomiques, il pose des problèmes quant à son utilisation dans des études macroéconomiques. En particulier, rien n’assure que le Z-score moyen soit pertinent pour mesurer le risque de défaillance du système bancaire (ex : contagion des défauts).
    Nous aurions aussi pu utiliser d’autres mesures comme le niveau des NPLs (et non ses variations), le ratio equity sur capital ou encore l’occurrence des crises bancaires. Mais chacun souffre de limites. Les pays peuvent s’accommoder pendant une longue période d’un niveau élevé de NPLs. Le ratio equity sur capital capture non seulement une situation de fragilité mais également une réaction des banques et/ou du régulateur. Enfin l’occurrence d’une crise est une situation extrême alors que notre objectif est d’identifier les situations de tension (au-delà des crises). Face à ces constats, nous avons décidé de créer une mesure qui nous paraît plus en phase avec notre objectif initial.

    Merci encore pour l’intérêt porté à cette étude

    In English
    Hello Michael,
    Thank you for this very relevant comment. Indeed we have chosen to create a new measure of financial instability rather than using the usual indicators as each of the usual indicators suffer from shortcomings. The most relevant alternative indicator for our study might have been the Z-score. The Z-score is a microeconomic indicator capturing the risks of failure of a bank (distance to fault). If this indicator is relevant to microeconomic studies, it poses problems for its use in macroeconomic studies. In particular, there is no guarantee that the average Z-score is relevant to measure the risk of default of the banking system (eg contagion defects). We could also have used other measures such as the level of NPLs (not its variations), the equity capital ratio or the occurrence of banking crises. But each suffers from limitations. Countries can live for a long time with a high level of NPLs. The ratio of equity capital captures not only a situation of fragility but also a reaction of banks and/or regulator. Finally the occurrence of a crisis is an extreme situation when our goal is to identify situations of tension (beyond crisis). Given these facts, we decided to create a measure that seems more in line with our initial objective.

    Thank you again for your interest in this study.

    Reply
  3. Laurent Weill

    That’s a very interesting study with a nice dataset on information sharing.
    What about considering separately public and private information sharing in the estimations? Former empirical literature on information sharing considers both forms in a separate way (of course when data availability allows it).

    Reply
    • Florian Léon

      Hello Laurent,

      thank you for your relevant remark. In an updated version, we consider this issue. For doing so, we consider a model with two dummies: one for countries with a public credit registry and one for those with a private credit bureau. In a nutshell, our baseline findings are not altered by the type of credit information sharing mechanisms. Both depth and coverage remain negative and statistically significant when we consider only PCR or PCB. It should be noted that the presence of a PCR or a PCB (dummy) does not affect financial stability. Put differently, it is the level of development of each credit information sharing infrastructure that matters rather than its presence or its organization (public and private).

      Reply

Submit a Comment

Your email address will not be published. Required fields are marked *