«MichaeL ehrMann MichaeL ZiegeLMeyer January 2014 Household Risk Management and Actual Mortgage Choice in the Euro Area1 Michael Ehrmann2 and Michael ...»
HouseHold Risk ManageMent
and actual MoRtgage cHoice
in tHe euRo aRea
MichaeL ehrMann MichaeL ZiegeLMeyer
Household Risk Management and Actual Mortgage Choice in the Euro Area1
Michael Ehrmann2 and Michael Ziegelmeyer3
Mortgages constitute the largest part of household debt. An essential choice when taking out a
mortgage is between fixed-interest-rate mortgages (FRMs) and adjustable-interest-rate mortgages (ARMs). However, so far, no comprehensive cross-country study has analyzed what determines household demand for mortgage types, a task that this paper takes up using new data for the euro area. Our results support the hypothesis of Campbell and Cocco (2003) that the decision is best described as one of household risk management: income volatility reduces the take-out of ARMs, while increasing duration and relative size of the mortgages increase it. Controlling for other supply factors through country fixed effects, loan pricing also matters, as expected, with ARMs becoming more attractive when yield spreads rise. The paper also conducts a simulation exercise to identify how the easing of monetary policy during the financial crisis affected mortgage holders. It shows that the resulting reduction in mortgage rates produced a substantial decline in debt burdens among mortgage-holding households, especially in countries where households have higher debt burdens and a larger share of ARMs, as well as for some disadvantaged groups of households, such as those with low income.
JEL-codes: D12, E43, E52, G21 Keywords: mortgage choice, fixed-rate mortgage, adjustable-rate mortgage, household finance, monetary policy.
This paper uses data from the Eurosystem Household Finance and Consumption Survey. This paper should not be reported as representing the views of the Banque centrale du Luxembourg (BCL), the Bank of Canada, the Eurosystem or the Eurosystem Household Finance and Consumption Network (HFCN). The views expressed are those of the authors and need not be shared by other research staff or policymakers at the BCL, the Bank of Canada, the Eurosystem or the Eurosystem Household Finance and Consumption Network. We thank Martin Brown, Dimitris Christelis, Joao Cocco, Pirmin Fessler, Dimitris Georgarakos, Paolo Guarda, Christos Koulovatianos, Jean-Pierre Schoder, an anonymous referee of the European Central Bank (ECB)Working Paper Series, participants at the European Economic Association Congress 2013, the Luxembourg Workshop on Household Finance and Consumption, the European Conference on Household Finance 2013, and at seminars at the ECB, BCL and the HFCN for useful comments.
Bank of Canada, email@example.com.
Banque centrale du Luxembourg, firstname.lastname@example.org.
Die Schulden von Haushalten bestehen zum größten Teil aus Hypothekenkrediten. Eine der wichtigsten Entscheidungen, die Haushalte bei der Kreditaufnahme zu treffen haben, ist die Wahl zwischen einem variablen oder einem festen Zinssatz. Ziel dieses Artikels ist es zu untersuchen, welche Entscheidungs- und Einflussfaktoren Haushalte in der europäischen Währungsunion (EWU) über die letzten Jahrzehnte bei dieser Wahl berücksichtigt haben. Die Basis dafür bilden Daten aus einer neuen gemeinsamen Erhebung zu den Finanzen privater Haushalte in 15 Ländern der EWU, der “Eurosystem Household Finance and Consumption Survey (HFCS)”. Dieser Datensatz enthält detaillierte Informationen zur Hypothekenwahl und einer Vielfalt von Haushaltscharakteristika von fast 9.000 Haushalten.
Die Auswertung der Daten zeigt, dass Hypotheken auf das selbst genutzte Wohneigentum 63% der Haushaltsverschuldung in der EWU ausmachen. Dabei variiert der Anteil variabel verzinster Hypotheken stark: zum einen über die Länder der EWU (von 13% in Frankreich bis zu 85% in Luxemburg und Portugal), zum anderen über die Zeit und verschiedene Haushaltscharakteristika. So ist zum Beispiel der Anteil variabel verzinster Kredite für Haushalte mit niedrigem Bildungsstand am größten und steigt mit dem Einkommen an.
Auf der Nachfrageseite sind die Entscheidungsfaktoren zur Hypothekenwahl konsistent mit den von Campbell und Cocco (2003) dargelegten Hypothesen. Die Entscheidung der Haushalte entspricht in diesem Modell dem Abwägen und Managen verschiedener Risiken. Die Untersuchung zeigt, dass Haushalte mit stärker schwankendem Einkommen seltener variable Zinsen wählen, da sie mit festen Zinsen eine Art Versicherung gegen variable Zinszahlungen (Zahlungsrisiko) erhalten können. Dagegen werden Hypotheken mit längeren Laufzeiten oder größere Hypotheken (relativ zum Einkommen oder dem Wert der Sicherheit) eher mit einer variablen Verzinsung abgeschlossen, weil für solche Hypotheken der Risikoaufschlag für feste Zinsen ansteigt. Schließlich spielt auch noch die Preisdifferenz zwischen variablen und fest verzinsten Verträgen eine Rolle im Entscheidungskalkül der Haushalte. Wenn die Spanne zwischen lang- und kurzfristigen Zinsen ansteigt, werden variabel verzinste Verträge (die nahe an den kurzfristigen Zinsen sind) attraktiver, wohingegen fest verzinste Verträge (die näher an den langfristigen Zinsen sind) an Attraktivität zumindest in der Anfangsperiode einbüßen. Da die Angebotsseite nicht Gegenstand der Untersuchung ist, kontrolliert dieser Artikel für diese durch länderspezifische fixe Effekte.
Der letzte Teil des Artikels umfasst eine Simulationsstudie, die darauf fokussiert inwieweit die Lockerung der Geldpolitik im Verlauf der Finanzkrise die Schuldenlasten der Hypothekenbesitzer beeinflusst hat, und inwieweit sich diese Effekte über verschiedene Haushaltstypen und Länder hinweg unterscheiden. Es zeigt sich deutlich, dass Haushalte in verschiedenen Ländern unterschiedlich stark von der geldpolitischen Lockerung profitieren. Zu einer stärkeren Reduktion der Zinslast kommt es vor allem in Ländern mit höher verschuldeten Haushalten und einem hohen Anteil von variabel verzinsten Verträgen. Ferner wurde vor allem benachteiligten Haushaltsgruppen mit zum Beispiel geringen Einkommen geholfen. Weitere Verteilungswirkungen dieser Politik, z.B. auf Sparer und Finanzinstitutionen, sind nicht Gegenstand der Untersuchung.
1. Introduction In several advanced economies, household debt increased substantially in the years leading up to the global financial crisis. Increasing levels of debt were often accompanied by concurrent booms in house prices and stock markets, leaving debt-to-asset ratios comparatively low, but exposing many households to a substantial risk if asset prices were to fall. Similarly, the servicing of the debt seemed possible in a benign economic environment, but was at risk of becoming unsustainable if conditions were to deteriorate. Of course, this risk materialized during the crisis: stock markets and house prices fell sharply, triggering a substantial drop in household wealth and an often unsustainable increase in the debt-to-asset ratios. Furthermore, many households also experienced a drop in income, or at least greater income uncertainty resulting from higher unemployment in these economies, making it more difficult or impossible for them to meet their debtservicing obligations.
While the need for household deleveraging has been identified by several policy institutions (see, e.g., European Central Bank 2012 or International Monetary Fund 2012), an important question is how such deleveraging can be brought about without endangering macroeconomic and financial stability. One factor that might have eased the adjustment burden is the large decline in interest rates that was brought about because of the accommodative monetary policies in several economies. To assess the relief that this brought to households, it is important to understand how the effects were distributed. An important dimension of this assessment is the type of mortgage contract that households have – first, because the bulk of household debt tends to be mortgage debt, a decrease in debt-service payments on mortgages will have a measurable impact on the financial situation of mortgage holders; and second, because there are typically two main types of mortgage contracts, namely, adjustable-rate mortgages (ARMs) and fixed-rate mortgages (FRMs), which led to vastly different effects of the decline in interest rates on household debt burdens. Households with an ARM could benefit more directly from reduced interest rates (even if after some delay), whereas those with an FRM could not benefit, or would have to revert to an often costly refinancing of their mortgage.
In light of these factors, this paper addresses two questions. First, it studies the determinants of mortgage choice in the euro area over previous decades. Campbell and Cocco (2003) developed a related theory, which stipulates that under borrowing constraints and income risks, ARMs are relatively less attractive to riskaverse households with a large mortgage, risky income, high default costs or low moving probability. This paper is the first to put the findings by Campbell and Cocco to an empirical test for the euro area as a whole.
The analysis focuses on the demand side, as does the model of Campbell and Cocco. However, it is important to note that the household decisions might be constrained by the type of mortgages offered. For instance, European Commission (2011) finds that there is a dominant mortgage type in most euro area countries, although this seems to have become less common since the 1990s.
The second part of this paper conducts simulation exercises to identify which mortgage holders saw particularly large declines in their debt burden following the substantial easing of monetary policy during the financial crisis.
This paper uses recently collected micro data from the Eurosystem Household Finance and Consumption Survey (HFCS), which covers information on mortgage choice for nearly 9,000 households in the euro area, and contains detailed information on mortgage contracts as well as other household characteristics. Using these data, the paper first demonstrates (section 3) that mortgages on the household main residence constitute an important part of household balance sheets, making up 63% of the total debt of euro area households. The data also show considerable variation in mortgage choice across euro area countries (from 13% ARMs in France to 85% in Luxembourg and Portugal), but also with regard to the year of mortgage take-out and household characteristics. For instance, in the euro area, the share of ARMs is highest for loweducation households and rises with income. By identifying the point in time when a given mortgage was taken out, we can exploit time-series variation in our data even if the survey contains only one cross-section.
Our estimates of the determinants of household decisions to take up an ARM or FRM are well in line with the hypothesis of Campbell and Cocco (2003), according to which households base their decision on a riskmanagement approach. While we observe household characteristics only at the time of the survey rather than at the time of mortgage take-out, we are able to match the macroeconomic conditions to the point in time when the mortgage choice was actually made. Accordingly, the results with regard to household characteristics should be interpreted as descriptive rather than causal. Still, in line with our hypothesis, we find that higher income volatility is related to a reduced propensity to take out ARMs (which is in line with the notion that FRMs are perceived as insurance against adjustable interest rates), whereas, for mortgages with longer maturities and larger relative size, relatively more ARMs are chosen by households (for such mortgages, the insurance premium of FRMs is bound to be very large, thus making FRMs relatively expensive compared with ARMs). In addition, households also consider the pricing of mortgages: if the spread between long-term and short-term interest rates rises, adjustable rates (which should be closer to short-term rates) become more attractive than fixed rates (which are proxied by long-term rates), at least in the initial periods.
In the second step, this paper focuses on the effects of the monetary policy easing during the financial crisis across mortgage holders. While other distributional effects of monetary policy, e.g., on savers versus borrowers, or on the financial sector versus the household sector, would be interesting to examine, they are beyond the scope of this study. This paper shows that the monetary easing led to particularly large reductions in the debt burden of households in several euro area countries (mainly countries with households that have higher debt burdens and countries with a larger share of ARMs, such as Cyprus, Luxembourg, the Netherlands, Portugal and Spain), as well as for a number of disadvantaged household groups, such as those with low income.
The paper proceeds as follows: section 2 provides an overview of the related literature. The data are explained in section 3. Section 4 reports the findings on the determinants of mortgage choice in the euro area. Section 5 studies the effects of the easing of monetary policy on mortgage holders, and section 6 concludes.