Mapping financial vulnerability in CESEE: understanding risk-bearing capacities of households is key in times of crisis

A crisis of the real economy – like the current crisis caused by the coronavirus pandemic – and the countermeasures taken by countries worldwide can lead to a severe financial crisis if debtors turn out to be unable to pay back their debt. The support debtors need and the costs involved in providing it directly depends on the financial buffer households have and their general risk-bearing capacity. It is crucial to understand both aspects to be able to anticipate potential problems and prepare for mitigating their impact. Policies designed to mitigate the effects of income losses could benefit greatly from better knowledge of the exact nature of the nonlinearities involved. We analyze newly available microdata on households’ balance sheets to examine financial vulnerability in Central, Eastern and Southeastern European (CESEE) countries and Austria. As Austrian banks have a high and increasing exposure in the region, households’ risk-bearing capacities in CESEE are an important factor in determining credit risks of the banking sector in Austria. The Household Finance and Consumption Survey (HFCS) allows us to study the general indebtedness of households as well as borrower-level vulnerability in eight CESEE countries and compare them to Austria. While the share of households owning their homes is comparably large in these countries, the share of households holding mortgage debt is not particularly large. Uncollateralized debt levels, by contrast, vary greatly across the region, and some of the countries show rather high levels of loan-to-value ratios, which point to more generous credit standards in mortgage lending. The debt service-to-income ratio >40% vulnerability measure points toward households in Croatia, Lithuania, Slovenia and Hungary being particularly vulnerable. Subtracting the assets of vulnerable households from their debt reveals that the levels of potential losses for banks are generally low. The highest loss given default estimates are obtained for Slovenia, Hungary and Lithuania. Furthermore, we use a machine learning approach to reweight the data, thereby decomposing the observed differences between CESEE and Austria into one part that can be explained by observable household characteristics and a remainder, which might be linked to banks’ different treatment of similar clients in different countries. The different directions of the effects of the reweighting approach across countries indicate that there is no typical household structure that suggests a high level of vulnerability as different types of households are vulnerable across countries. One important lesson from this crisis is to make sure that better data are available to policymakers (e.g. registers covering the loans of households to the necessary degree) so that research does not have to rely on survey data alone to analyze households’ risk-bearing capacities and, hence, we are better prepared for the next crisis.