Macroeconomic and macro-financial factors as leading indicators of non-performing loans

Purpose Large or increasing stocks of non-performing loans in the banking sector constitute threats to financial stability. This paper considers to which extent various macroeconomic and macro-financial factors may serve as leading indicators for the dynamics of the ratio of non-performing loans to total loans. Design/methodology/approach The paper estimates panel data models for all EU countries and two groups of EU countries using quarterly data over approximately 20 years. Findings The estimations show that many macroeconomic and macro-financial variables are leading indicators for non-performing loans in the EU countries, even years ahead. Higher GDP growth, lower inflation and lower debt are robust leading indicators of a lower ratio of non-performing loans in the future. The current account balance and real house prices are important indicators for the Western European group but not for the Central and Eastern European group. Research limitations/implications The estimations are carried out for panels of EU countries and the effects may hence be seen as averages for the countries in the particular panel and may not apply for individual countries. Practical implications National and international authorities have brought in systems to detect and address imbalances and emerging problems in the financial sectors. Many of the measures operate with long lags, and so it is important to assess whether various macroeconomic and macro-financial variables may serve as leading indicators for future developments of non-performing loans. Originality/value The main contribution of the paper is that it estimates models meant expressly for predicting non-performing loans several years ahead. The results are thus of practical use for national and international authorities which typically have access to measures that operate with a long delay. The analysis also includes more macroeconomic and macro-financial variables as leading indicators than have typically been used in earlier studies.