The vulnerability of indebted households during the crisis:

evidence from the euro area

“The Bank of Italy’s Analysis of Household Finances”Rome, December 4th 2015

L.Bartiloro, V.Michelangeli and C.Rampazzi

• The financial crisis has shown that households’ financial vulnerability plays a pivotal role for financial stability.

• We analyse the characteristics that are correlated with vulnerability for the euro area households.

• We focus a standard indicator of financial vulnerability and we also compare our findings with those obtained using with the other indicators.

• Policy implications of our results

Outline

Related literature:• Household vulnerability in a country over time (IMF, 2011, 2012, 2013;

ECB 2013b; Magri and Pico, 2012; Michelangeli and Pietrunti, 2014)• Household vulnerability in a country with different indicators of financial

fragility or over-indebtedness (Bartiloro and Rampazzi, 2013; D’Alessio et al., 2013, among others).

• Household indebtedness in the euro area (Bover et al., 2013) • Household mortgage choice in the euro area controlling for macro and

financial variables (Ehrmann and Ziegelmeyer, 2013)

In this paper:• Focus on household vulnerability in the euro area countries (Financial

stability) • We also evaluate how the main mortgage characteristics are related to

household vulnerability (Policy implications)

What’s new

Household vulnerability

• Standard indicator of vulnerability: DSR=Debt service payments/Income

• Low income households may find it difficult to face other general expenses and to accumulate savings in order to offset unexpected negative economic shocks

• A household is vulnerable if1) its DSR ≥ 40 per cent and2) its income is below the median of the population

4

Household Finance and Consumption Survey

• Fist harmonized survey on households’ wealth, debt, income and consumption in the euro area

• It is voluntary conducted by national central banks of the euro area member states and coordinated by the ECB

• Data are so far available for just one wave and they mostly refer to year 2010.

• The total sample of the first edition consists of about 62,000 households and covers 15 euro area countries

• We excluded Finland from our sample because of lack of information on debt service and Slovenia because of the very limited sample size. Therefore, our euro area aggregate includes 13 countries.

5

Indebted households: a comparison in the euro area

Vulnerable households: a comparison in the euro area

Share of debt held by vulnerable households

8

Main regression:

Where:sex: gender of the head of the householdHMR: dummy variable equal to 1 if the household owns her main residenceage: age class of the head of the household (<35, 35-44, 45-54, 55-64, >=65) educ: education level of the head of the household (low, middle, high) Nchild: n. of dependent children (0, 1, 2, or greater than 2)Nempl: n. of hh members in employment (0, 1, 2, or greater than 2)work: work status (employee, self-employed, unemployed, retired, other)finass: quintile of financial assets based on each country distributionmortg: dummy equal to 1 if the household has only mortgage debt conscred: dummy equal to 1 if the household has only consumer creditYj : controls (Model1: country dummies, Model 2: macro variables, Model 3:

banking variables, Model 4: macro and banking) Lji :variables related to mortgages on the home main residence (LTV, number

of mortgage loans, refinancing, mortgage length, type of interest rate)

The model

Odds ratios

• We estimate the odds of being vulnerable as a function of socio-demographics variables and mortgage characteristics.

• Odds of being vulnerable=

• For any explanatory variable, an estimated coefficient higher than 1 implies that the probability of being vulnerable is higher for the category defined by the explanatory variable with respect to the baseline household; while a coefficient smaller than 1 implies that the probability of being vulnerable is lower than for the baseline household.

10

Baseline household

• the head of the household is aged between 35 and 44• employee• medium level of education • two members in employment• no dependent children• first quintile of financial assets • both a mortgage and loans for consumption purposes.

11

Benchmark logistic regressions

12

1.150.821.351.071.040.761.48 ***0.781.291.63 **2.12 ***3.85 ***1.92 ***1.072.57 ***1.250.721.300.69 **0.46 ***0.42 ***0.37 ***0.74 *0.23 ***0.55 ***0.76 ***1.02

no0.19 ***20,603

Odds ratioModel 2

1.140.801.331.031.020.811.61 ***0.801.341.65 ***2.02 ***3.80 ***1.92 ***1.082.53 ***1.360.671.360.72 *0.48 ***0.43 ***0.37 ***0.73 *0.22 ***

1.01 **1.00 ***1.03 **no

0.03 ***20,603

Odds ratioModel 3

1.160.841.321.071.040.751.45 **0.761.281.63 **2.05 ***3.71 ***1.92 ***1.072.68 ***1.260.751.330.70 **0.46 ***0.43 ***0.36 ***0.74 *0.24 ***0.65 **0.87 *1.04 **1.01 **1.00 ***1.00no

0.05 ***20,603

Odds ratioModel 4

Sex: female 1.15Tenure of house: non owner 0.88Age <35 1.34Age 45-54 1.06Age 55-64 1.04Age ≥65 0.72Education: Low (0-2 ISCED) 1.41 **Education: High (5-6 ISCED) 0.75N° of children: 1 1.27N° of children: 2 1.62 **N° of children: 3+ 2.11 ***N° of empl. members: 0 3.95 ***N° of empl. members: 1 1.93 ***N° of empl. members: 3+ 1.05Working status: Self-employed 2.67 ***Working status: Unemployed 1.18Working status: Retired 0.75Working status: Other 1.25Financial asset quintile: 2° 0.72 *Financial asset quintile: 3° 0.47 ***Financial asset quintile: 4° 0.43 ***Financial asset quintile: 5° 0.36 ***Only mortgage debt 0.73 *Only non-mortgage debt 0.23 ***InflationGDP growth rateUnemployment rateBank concentrationBank deposits/GDPGrowth rate of bank loansCountry fixed effects yesConstant 0.03 ***No. of observations

Model 1Odds ratio

20,603

Benchmark logistic regressions only hh with mortgage

13

10% ≥ LTV <30% 1.83 ***30% ≥ LTV <50% 1.76 **50% ≥ LTV <80% 2.43 ***LTV ≥80% 2.92 ***N° of mortgage loans 0.98Refinancing 1.16Rate on HMR mortgage: fixed 1.07HMR mortgage lenght 0.99Only mortgage debt 0.74 *InflationGDP growth rateUnemployment rateBank concentrationBank deposits/GDPGrowth rate of bank loansCountry fixed effects yesConstant 0.02 ***No. of observations 8,583

Model 1Odds ratio

1.81 **1.69 **2.28 ***2.84 ***0.991.240.870.990.760.750.871.04 **

no0.03 ***

8,583

Model 2Odds ratio

1.82 **1.69 **2.29 ***2.77 ***0.961.280.800.990.74 *

1.011.00 ***1.04 *

no0.01 ***

8,583

Model 3Odds ratio

1.86 ***1.80 **2.51 ***3.12 ***0.991.210.940.990.750.851.001.08 ***1.02 ***1.00 ***1.00

no0.01 ***

8,583

Model 4Odds ratio

All the results with respect to the demographics are confirmed

Other indicators of vulnerability

• debt to income ≥3: long-run ability of repaying accumulated debt given the future stream of income

• net wealth < 0: long-run ability of repaying accumulated debt given the accumulated savings

• income - debt payments < food expenses: short-run ability to face expected expenses

• financial assets < 2 months of income: short-run ability to face

unexpected expenses

14

Other indicators of vulnerability (Cont.)

 memo:hh withany debt

DSR>=40% & income<median

Debt/income>=3 Net wealth<0

Income-debt payments<food

expenses

Liquid asset< 2 months of

income

N° indicators >Euro area

meanAustria 35.6 2.8 9.2 14.8 2.9 36.8 1Belgium 44.8 6.0 15.0 6.0 8.0 44.3 2Cyprus 65.4 13.6 31.5 4.4 13.6 56.8 4France 46.9 3.3 12.4 8.3 1.2 47.5 0Germany 47.4 2.2 11.3 15.7 2.3 49.0 1Greece 36.6 6.7 12.2 6.9 5.5 72.4 3Italy 25.2 4.9 11.0 5.7 5.1 50.5 2Luxembourg 58.3 5.1 20.6 6.5 3.2 44.1 2Malta 34.1 1.2 10.5 2.4 3.2 20.8 0Netherland 65.7 7.8 35.2 17.7 8.7 42.9 4Portugal 37.7 10.2 28.3 6.8 9.1 53.8 4Slovakia 26.8 6.7 11.4 4.4 6.0 70.7 3Spain 50.0 10.9 24.2 6.9 6.6 57.0 4

EURO AREA 43.4 5.0 16.0 10.9 4.1 49.9                 

           

15

… but the main results are confirmed: being self-employed, a reduction in the n. of employed members, lower financial assets increase the odds of vulnerability in a significant way.

Heterogeneity across euro area countries ….

• Combine the different indicators of vulnerability

• Weights obtained using the principal component analysis (First component)

• We obtain a continuous variable (OLS regression)

• Provide a synthetic representation of vulnerability

Building a unique indicator of vulnerability

Vulnerability index (PCA weights): OLS regression

17

Sex: female 0.07 * 0.08 ** 0.07 ** 0.08 **Tenure of house: non owner -0.08 * -0.08 ** -0.09 ** -0.09 **Age <35 0.19 *** 0.19 *** 0.18 *** 0.18 ***Age 45-54 -0.01 -0.02 -0.02 -0.01Age 55-64 -0.11 * -0.11 * -0.10 * -0.11 *Age ≥65 -0.19 ** -0.16 * -0.13 -0.16 *Education: Low (0-2 ISCED) 0.10 ** 0.12 *** 0.15 *** 0.11 **Education: High (5-6 ISCED) -0.05 -0.03 -0.03 -0.04N° of children: 1 0.03 0.03 0.04 0.03N° of children: 2 0.11 ** 0.09 * 0.10 * 0.10 *N° of children: 3+ 0.18 ** 0.16 * 0.14 0.15 *N° of empl. members: 0 0.55 *** 0.53 *** 0.52 *** 0.51 ***N° of empl. members: 1 0.25 *** 0.25 *** 0.25 *** 0.25 ***N° of empl. members: 3+ -0.08 -0.06 -0.06 -0.07Working status: Self-employed 0.37 *** 0.35 *** 0.35 *** 0.37 ***Working status: Unemployed 0.09 0.13 0.15 0.13Working status: Retired -0.19 ** -0.21 *** -0.23 *** -0.19 **Working status: Other 0.00 0.06 0.08 0.06Financial asset quintile: 2° -0.18 *** -0.20 *** -0.19 *** -0.20 ***Financial asset quintile: 3° -0.32 *** -0.33 *** -0.32 *** -0.34 ***Financial asset quintile: 4° -0.35 *** -0.35 *** -0.35 *** -0.36 ***Financial asset quintile: 5° -0.40 *** -0.37 *** -0.39 *** -0.41 ***Only mortgage debt -0.18 *** -0.17 *** -0.17 *** -0.17 ***Only non-mortgage debt -0.60 *** -0.62 *** -0.62 *** -0.60 ***Inflation -0.32 *** -0.30 ***GDP growth rate -0.12 *** -0.12 ***Unemployment rate 0.00 0.01 **Bank concentration 0.005 *** 0.004 ***Bank deposits/GDP 0.000 *** 0.001 ***Growth rate of bank loans 0.003 -0.013 **Country fixed effects yes no no noConstant 0.23 *** 1.06 *** 0.06 0.76 ***No. of observations

Model 1 Model 2 Model 3 Model 4

20,597 20,597 20,597 20,597

Coef. Coef. Coef. Coef.

Vulnerability index (PCA weights): OLS regression - only hh with mortgage

18

10% ≥ LTV <30% 0.18 *** 0.18 *** 0.19 *** 0.19 ***30% ≥ LTV <50% 0.28 *** 0.28 *** 0.29 *** 0.29 ***50% ≥ LTV <80% 0.70 *** 0.69 *** 0.71 *** 0.72 ***LTV ≥80% 1.00 *** 1.01 *** 1.02 *** 1.03 ***N° of mortgage loans 0.05 0.05 0.05 0.05Refinancing 0.15 0.17 * 0.18 ** 0.16 *Rate on HMR mortgage: fixed -0.04 -0.15 *** -0.17 *** -0.12 **HMR mortgage lenght 0.00 0.00 0.00 0.00Only mortgage debt -0.13 ** -0.12 * -0.13 ** -0.12 **Inflation -0.14 -0.13GDP growth rate -0.07 -0.01Unemployment rate 0.01 * 0.03 ***Bank concentration 0.00 ** 0.01 ***Bank deposits/GDP 0.00 *** 0.00 ***Growth rate of bank loans 0.02 * 0.00Country fixed effects yes no no noConstant -0.47 ** -0.10 -0.60 *** -0.67 *No. of observations 8,582 8,582 8,582 8,582

Model 1 Model 2 Model 3 Model 4

All the results with respect to the demographics are confirmed

Conclusion• Large heterogeneity in the euro area with respect to the share of

indebted households and of the debt at risk.• Some common aspects relevant for financial stability:1) Strong correlation between self-employment and vulnerability,

which may suggest the implementation of some policy initiatives to support the liquidity of self-employers (financial education, efficient mortgage insurance market)

2) Higher LTV is associated with higher vulnerability, while no effect has been detected for the number of mortgages or their duration.

Main policy implication: this analysis provides support for the introduction of limits to LTV ratios for macroprudential purposes.• Our benchmark indicator is a good measure of financial distress in

the household sector. It is therefore crucial that macro-prudential authorities work on a unique, correct and exhaustive way of using the DSR in order to identify vulnerable households.

Thank you 19