Determinants of mortgage arrears in Europe: evidence from household microdata

Abstract

In purely economic terms, mortgage arrears can pose a risk to the stability of banks and limit households’ future access to credit. Moreover, arrears have social ramifications: they reduce aspects of households’ well-being and health, and addressing such negative effects can lead to a requirement for higher social spending by governments. It is therefore important to identify the drivers of arrears and design policies to reduce them. We use a European household data-set to analyse what drives arrears. Controlling for household characteristics such as age and education, we find that affordability problems, such as unemployment, low income and high mortgage payments, matter. Households facing the dual trigger of affordability problems and negative equity are more likely to go into longer- term arrears. We also find that households in Cyprus and Greece are particularly prone to miss mortgage payments, while those in the United Kingdom and Belgium are very unlikely to do so. Generally, arrears tend to be higher in poor countries and where investors’ rights are poorly protected.

Keywords:

• arrears
• affordability
• negative equity

Acknowledgements

The views expressed in this paper are those of the authors and do not necessarily represent those of the Swiss National Bank. The paper is based on data from Eurostat, Survey on Income and Living Conditions, 2004–2011, and the responsibility for all conclusions drawn from the data lies entirely with the authors. The same applies for the data analysis discussed in Footnote 4, which uses the Eurosystem survey of the Household Finance and Consumption Network. We thank Brian O'Connell for his help with the data and the Editor, two referees, Michael Ehrmann, Stefan Gerlach, seminar participants at the Bank for International Settlements, the Economic and Social Research Institute, Dublin, the Nederlandsche Bank, the Swiss National Bank, conference participants at the 2014 Bank of Canada-Cleveland Fed-Atlanta Fed-SNB Conference in Ottawa and workshop participants at the 2014 Empirical Monetary Economics Workshop in Paris for useful comments and discussions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Data made available to us by Eurostat cover Bulgaria, Estonia, the Czech Republic, Denmark, Finland, Iceland, Latvia, Lithuania, Malta, the Netherlands, Norway, Romania, Slovakia, Sweden and Switzerland. We do not include these countries because of too few observations or missing explanatory variables.

2. This is the question from the 2004 Irish questionnaire. Formulations vary slightly over time and between countries and languages.

3. It should be noted that the EU-SILC questionnaire included the general question on arrears already in 2004; the distinction between one-time and repeated arrears yielded positive answers only from 2008 onwards, when the question was added “Thinking about mortgage payments, how many times have you been in arrears in the last 12 months?”

4. The EU-SILC data can be compared with data from the Eurosystem survey by the Household Finance and Consumption Network (HFCN). Of this survey, only the first wave was available at the time of writing, whereas we can use eight waves of EU-SILC. The HFCN survey contains data on mortgage maturity at origination, mortgage value at origination and negative equity. If we calculate country averages for these variables and compare them with the country averages from the EU-SILC data, we obtain correlations of 0.62, 0.97 and 0.02, respectively. The low correlation of the HFCN and the EU-SILC measures of negative equity does not necessarily imply that the EU-SILC measure is inferior. Our proxy assumes that households are likely to be in negative equity if they have outstanding debt and if property prices have fallen a lot. The HFCN survey gives information on the current value of a household's property and the outstanding mortgage. However, it is not clear how households evaluate the current value of their property. Presumably, there is considerable measurement error in both measures.

5. For a mathematical representation of this model, see Gerlach-Kristen and Lyons (2015).

6. Because of collinearity, we concentrated on one interaction between negative equity and affordability. The significance was strongest when we used unemployment.

7. We also considered the correlations of the country fixed effects with national Gini coefficients and the fraction of Catholic population, but found no link.