Abstract
Selected Issues
Household Deleveraging in Ireland1
A. Introduction
1. Irish households have deleveraged steadily in recent years. After peaking during the financial crisis, household debt moderated to about 150 percent of disposable income and about 19 percent of total assets at end-2016—its lowest level in a decade.2 The deleveraging process was initially driven by a significant decline in nominal debt, as households increased their saving rate and reduced new borrowing, and more recently by the strong rebound of economic activity and the rapid increase in incomes and asset values. Still, despite the sharp deleveraging, Ireland’s household debt remained well above the euro area average and levels seen in Ireland, prior to the property-market boom.
Ireland: Household Debt1
(Seasonally-adjusted)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1 Household debt includes loans and other accounts payable.Sources: Haver, and IMF staff.Ireland: Household Debt1
(Seasonally-adjusted)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1 Household debt includes loans and other accounts payable.Sources: Haver, and IMF staff.Ireland: Household Debt1
(Seasonally-adjusted)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1 Household debt includes loans and other accounts payable.Sources: Haver, and IMF staff.
Household Debt to Financial Assets, 2016Q3
(Percent)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: European Comission and IMF staff calculations.Household Debt to Financial Assets, 2016Q3
(Percent)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: European Comission and IMF staff calculations.Household Debt to Financial Assets, 2016Q3
(Percent)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: European Comission and IMF staff calculations.
Household Debt, 2016Q3
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Household Debt, 2016Q3
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Household Debt, 2016Q3
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
2. Against this background, this paper takes stock of the recent deleveraging while seeking to identify some of its drivers. In particular, it looks at the recent dynamics of household debt, changes in its composition, and the pace of deleveraging compared to other European countries. It also provides an overview on the distribution of household debt and households in negative equity by age and income cohort, building on the recent findings of the ECB’s Household Finance and Consumption Survey. Finally, the paper aims to identify some of the determinants of household debt, drawing also on the experience of euro area peers, and explore their role in the deleveraging process in Ireland. This exercise—although subject to large uncertainties—may help gauge whether the end of household deleveraging in Ireland’s is near. The latter may have implications for economic activity, including bank operations and private consumption dynamics.
B. The Dynamics and Composition of Household Debt
3. Ireland’s household debt reached unprecedented levels during the financial crisis. In the run-up to the crisis, household financial liabilities climbed rapidly to a peak of nearly 240 percent of disposable income in late 2009 from 115 percent in early 2003. Similarly, the household leverage ratio, defined as debt to financial and housing assets, increased to a record high of nearly 30 percent from 15 percent in 2003. The surge in household debt took place in the face of strong economic activity, low unemployment, and easy credit conditions, and was largely driven by a sharp increase in mortgage borrowing. In turn, the sustained appreciation of house prices boosted household assets and confidence, fueling further borrowing for consumption. As households consumed a larger share of their disposable income, their saving rate in 2003–07 stood at an average 7½ percent, nearly half the rates seen in European peers.
Household Assets
(Billions of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland, and IMF staff.Household Assets
(Billions of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland, and IMF staff.Household Assets
(Billions of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland, and IMF staff.
Gross Household Saving Rate
(In percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.Gross Household Saving Rate
(In percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.Gross Household Saving Rate
(In percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.4. The extent of household debt accumulation in Ireland in the pre-crisis period was more pronounced than in peers. While many European countries experienced an increase in household debt in 2002–08, the rapid increase in Ireland was exceptional and similar in magnitude only to Denmark, which also experienced a period of rapid house price appreciation that peaked in 2007. Consequently, Irish household debt-to-disposable-income was among the highest in Europe when the global financial crisis unfolded and, combined with an overvaluation in house prices, it amplified household vulnerabilities, likely contributing to the prolonged adjustment process.3
The Change in Household Debt
(In percent of Disposable income, 2002-08)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, Eurostat, and IMF staff.The Change in Household Debt
(In percent of Disposable income, 2002-08)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, Eurostat, and IMF staff.The Change in Household Debt
(In percent of Disposable income, 2002-08)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, Eurostat, and IMF staff.5. The recent deleveraging was supported by both net debt repayments and a strong rebound of disposable income. The adverse shift in economic conditions in 2009, which was accompanied by rising unemployment and tightening lending standards, prompted households to begin reducing their nominal debt almost immediately. Nevertheless, household debt as a ratio of disposable income continued to increase in 2009 as disposable income contracted significantly. In 2010–11, the household debt ratio started to moderate, initially due to significant net debt repayments, and later—particularly from 2011 onwards—also due to the recovery of disposable income. In recent years, the moderation in household debt was mainly driven by the strong increase in disposable income as new borrowing picked up and net repayments slowed.
Contribution to the Change in Household Debt
(Percentage points of disposable income, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Contribution to the Change in Household Debt
(Percentage points of disposable income, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Contribution to the Change in Household Debt
(Percentage points of disposable income, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.6. The magnitude of household deleveraging has been high compared to international comparators. Irish household debt registered a decline of about 90 percentage points of disposable income from its peak in late 2009, beyond the levels seen is other euro area countries, including those that experienced a financial crisis. The sharp deleveraging, which can be seen as a correction to the rapid debt accumulation prior to the crisis and was accompanied by a significant level of defaults, was partly driven by the strong growth of disposable income in recent years, low level of new lending, and intensified supervisory efforts, which promoted “sustainable solutions” for distressed borrowers.
Household Debt Decline: From Peak to Trough
(Percentage of disposable income, 2009Q1-2016Q3)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Countries’ Central Banks, Statistical Offices, and IMF staff.Household Debt Decline: From Peak to Trough
(Percentage of disposable income, 2009Q1-2016Q3)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Countries’ Central Banks, Statistical Offices, and IMF staff.Household Debt Decline: From Peak to Trough
(Percentage of disposable income, 2009Q1-2016Q3)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Countries’ Central Banks, Statistical Offices, and IMF staff.7. The debt service burden declined significantly in recent years. The interest burden, which stood at more than 10 percent of household disposable income in 2007, fell significantly in recent years on the back of strong recovery of disposable income and lower interest rates. Despite high debt levels in Ireland, the debt service burden is similar to that in the euro area across most of age cohorts. Fasianos and others (2017) argues that this is mainly due to (i) high income levels among those that hold mortgage; (ii) long debt maturity;4 and (iii) significantly lower interest rates, which can be largely attributed to the prevalence of “tracker” rate mortgage, which are linked to the ECB rate with a full pass-through to mortgage interest rates.
Debt Service-to-Income Ratio, by Age Cohort
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: ECB’s Household Finance and Consumption.Debt Service-to-Income Ratio, by Age Cohort
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: ECB’s Household Finance and Consumption.Debt Service-to-Income Ratio, by Age Cohort
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: ECB’s Household Finance and Consumption.
Household Debt and Interest Payments
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ Total household loans, percent of 4Q gross disposable income.2/ 4-quarter interest payments excluding FISIM adjustment.Sources: Central Statistics Office, and the Central Bank of IrelandHousehold Debt and Interest Payments
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ Total household loans, percent of 4Q gross disposable income.2/ 4-quarter interest payments excluding FISIM adjustment.Sources: Central Statistics Office, and the Central Bank of IrelandHousehold Debt and Interest Payments
(Percent of gross disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ Total household loans, percent of 4Q gross disposable income.2/ 4-quarter interest payments excluding FISIM adjustment.Sources: Central Statistics Office, and the Central Bank of Ireland8. Despite the prolonged deleveraging, a high share of households, particularly in the younger cohorts, remained heavily indebted. The recent ECB’s Household Finance and Consumption Survey (2016) indicates that about 57 percent of Irish household held debt in 2013/14—well above the euro area average of 42 percent. At about 12 percent, the share of negative equity households was also among the highest in the euro area, largely reflecting younger cohorts who bought homes near the peak of the property market. Additionally, household debt among young Irish households is well above that in their euro area peers, reflecting in part high home ownership in Ireland. Indeed, Fasianos and others (2017) indicate that, in this group of borrowers, debt is more concentrated in property than in comparator countries, such as the UK, the US and the euro area. Lydon and McIndoe-Calder (2017) show that deleveraging proceeded at a faster pace among older households. In contrast, young households who saw large increases in their debt-to-income ratio from 2006 to 2010 deleveraged only modestly in recent years due to weak disposable income growth and slow amortization rates. They concluded that deleveraging for these young households still has some way to go.5
9. The composition of household debt holders changed significantly over time. Household debt mainly consists of loans from Monetary Financial Institutions (MFIs), and Other Financial Institutions (OFIs).6 A small portion of loans, about 2½ percent of total household loans, is provided by the general government, and the non-financial corporate (NFC) sector.7 The share of MFIs loans, which averaged about 80 percent of household total liabilities in the pre-crisis period, receded to 62 percent in 2016q3, reflecting securitization of loans, retained securitization, and sales of distressed loans by MFIs to non-MFIs, as well as new lending by non-MFIs.8 New lending by non-MFIs, also seen in some other European countries, may be driven by the tight credit conditions applied in the banking system and more stringent regulations (ECB, 2016).
Household Loans: Shares of MFIs and Non-MFIs
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Household Loans: Shares of MFIs and Non-MFIs
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Household Loans: Shares of MFIs and Non-MFIs
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.
Composition of Households Loans Holders, 2016Q3
(In percent of households total loans)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ “Other” includes general government, NFC sector, pension funds, insurance companies, and non-money market investment funds.Sources: ECB, and IMF staff.Composition of Households Loans Holders, 2016Q3
(In percent of households total loans)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ “Other” includes general government, NFC sector, pension funds, insurance companies, and non-money market investment funds.Sources: ECB, and IMF staff.Composition of Households Loans Holders, 2016Q3
(In percent of households total loans)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
1/ “Other” includes general government, NFC sector, pension funds, insurance companies, and non-money market investment funds.Sources: ECB, and IMF staff.
Contribution to the Change in the Household Debt
(In percent, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Contribution to the Change in the Household Debt
(In percent, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.Contribution to the Change in the Household Debt
(In percent, y/y)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Haver, and IMF staff.
Composition of Households Debt
(Billion of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland.Composition of Households Debt
(Billion of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland.Composition of Households Debt
(Billion of euros)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Sources: Central Bank of Ireland.C. Determinants of Household Debt
10. Cross-country data is employed to study the determinants of household debt and the role of institutions. We follow recent papers (e.g. Jappelli and others, 2008, Barnes and Young, 2003, and Albuquerque and others 2014), which link household debt to several factors, both structural and cyclical.9 More specifically, we regress household debt to disposable income on the following variables:
Demography. As implied in the life-cycle model (Modigliani and Brumberg, 1954) and existing empirical studies, households incur debt when they are young. This hypothesis is broadly consistent with the findings of the recent ECB’s Household Financial and Consumption survey and other empirical studies (e.g. Albuquerque and others, 2014).
House prices-to-disposable income ratio. Higher house prices require higher borrowing and allow greater access to finance due to higher value of collateral. Additionally, house price appreciation can work through the traditional wealth effect and lead to higher borrowing to boost consumption (Dynan and Kohn, 2007).
Unemployment rate. Unemployment has an ambiguous effect on household debt. On the one hand, higher unemployment may be associated with the lower future income and uncertainty, therefore discouraging household from borrowing (Mainal and others, 2016, Meng and others, 2013). On the other hand, higher unemployment could represent an income shock and reduction of household wealth, thus leading to an increase in the debt burden. Furthermore, unemployment is normally highly correlated with high level of household distress and limited debt repayment capacity. Indeed, using a micro-level data of euro area member states, Du Caju and others (2016) found that unemployment triggers over-indebtedness.
Home-ownership rate. Homeowners tend to have more debt than non-homeowners as many household do not have sufficient equity to purchase a home (Dynan and Kohn, 2007). Moreover, higher housing assets held by households would reduce borrowing constraints and raise available collateral for consumer loans.10
Interest rate on mortgages. Higher real interest rates on mortgages reduce the affordability of household debt and increase the opportunity cost of buying a home.
Institutional structure. Better information sharing among credit institutions and greater legal rights for both borrowers and creditors were found to be positively correlated with higher credit to households (Jappelli and others, 2008, Djankov and others, 2007). We use the World Bank’s Doing Business indicator of “getting credit,” which measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and the effectiveness of information sharing about borrower characteristics.
Public debt. Following the Ricardian equivalence argument, public debt might influence taxation and therefore household saving and debt. In the context of Ireland, Andritzky (2012) finds a strong and significant impact of the Ricardian effect, i.e., households tend to relax savings when the government cuts the fiscal deficit (and vice versa). In addition, a higher public debt-to-GDP ratio may induce banks to be regular holders of government securities, crowding out loans to households (Coletta and others, 2014).
Methodologies Applied to Study Household Debt
Recent studies applied different specifications and estimation methods to examine the determinants and sustainability of household debt. For example, Barnes and Young (2003) developed a calibrated partial equilibrium overlapping generations model to explain the increase in aggregate indebtedness in the United States. Tudula and Young (2005) applied a similar approach to the United Kingdom. Dynan and Kohn (2007) used micro-level data to assess the determinants of household debt in the US, using variables such as age, education, and homeownership. Colletta et al. 2014 applied a panel data of 32-countries to study the determinants of household debt, taking into account both demand and supply-side factors.
Other studies identified the “equilibrium” level of household debt and the short-run deviations. Philbrick and Gustafsson (2010) and Meng and others (2011), for instance, studied the determinants of household debt in Australia through a Vector Error Correction Model, while Mokhtar and others (2013) employed a similar methodology to study household debt in Malaysia. Similarly, Albuquerque and others (2014) used a panel dataset to estimate a time-varying equilibrium household debt-to-income ratio across US states by employing a Pooled Mean Group developed by Pesaran et al. (1999).
A somewhat different strand of the literature focused on deleveraging needs based on a sustainability analysis. Cuerpo et al. (2013), for instance, proposed a time-varying measure that relies on the notion of stationarity of household debt-to-asset ratio. While this modelling approach considers valuation effects on the asset side, it ignores the possibility that the sustainable debt ratio could also depend on factors other than assets, such as income expectations, uncertainty, or the cost and access to credit. Andritzky (2012) used “out-of-sample” prediction of household savings rates to assess the future trajectory of household debt.
11. Data and methodology: The sample covers the period 2003Q1–2016Q3 and includes eleven euro area countries where data for both household debt and the considered explanatory variables are available.11 For some countries, the sample is shorter due to lack of data availability. Data is taken from Haver (household debt-to-disposable income, public debt-to-GDP, and interest rates on mortgages), Eurostat (home ownership, and the share of population age 25–44), OECD (house price-to-income ratios), and the World Bank (“Getting Credit” indicator).12 The dependent variable is household debt as a share of gross disposable income. The regressions are estimated using an instrumental variable method to address a possible endogeneity bias (two stage least square), with country fixed effects.13 The specifications also include a time dummy to control for common euro area macroeconomic effects.
12. Table 1 presents summary statistics for Ireland and the rest of the countries in the sample. It shows that average household debt-to-disposable income was nearly double in Ireland compared to the rest of the countries in the sample. Moreover, the home ownership rate and the share of young population (ages 25–44) are on average higher in Ireland than in the rest of the sample, though in recent years it has shown a significant decline (see Annex). Real interest rates on mortgages were somewhat lower on average, reflecting in part the prevalence of tracker mortgages. The data also show that house price-to-income was, on average, higher in Ireland than in peers, despite the housing-market bust in the crisis, and that Ireland’s unemployment rate was much more volatile around its mean. The “getting credit” indicator suggests that access to credit in Ireland is, on average, better than that in other euro area members in the sample.14
Ireland: Summary Statistics, 2003Q1–2016Q3
Data are seasonally adjusted.
Ireland: Summary Statistics, 2003Q1–2016Q3
| Debt-to-disposable income/1 | Public debt as a share of GDP /1 | Real interest rate on mortgages | Home ownership | Population at age 25-44 | House price to income ratio | World Bank’s “Getting Credit” | Unemployment (diff. from long-term average) | |
|---|---|---|---|---|---|---|---|---|
| Full sample, excluding Ireland | ||||||||
| mean | 99.8 | 100.1 | 3.5 | 71.7 | 28.8 | 110.3 | 61.8 | 0.000 |
| min | 22.4 | 37.3 | 0.56 | 51.9 | 24.7 | 75.5 | 43.8 | −2.446 |
| max | 267.2 | 193.7 | 6.52 | 90.5 | 33.5 | 166.4 | 87.5 | 2.853 |
| Ireland | ||||||||
| mean | 186.9 | 61.9 | 3.1 | 74.4 | 30.8 | 119.2 | 84.9 | 0.000 |
| min | 112.7 | 24.7 | 2.1 | 68.6 | 28.9 | 81.8 | 70.0 | −5.292 |
| max | 237.2 | 117.9 | 5.0 | 81.8 | 32.1 | 165.4 | 87.5 | 7.137 |
Data are seasonally adjusted.
Ireland: Summary Statistics, 2003Q1–2016Q3
| Debt-to-disposable income/1 | Public debt as a share of GDP /1 | Real interest rate on mortgages | Home ownership | Population at age 25-44 | House price to income ratio | World Bank’s “Getting Credit” | Unemployment (diff. from long-term average) | |
|---|---|---|---|---|---|---|---|---|
| Full sample, excluding Ireland | ||||||||
| mean | 99.8 | 100.1 | 3.5 | 71.7 | 28.8 | 110.3 | 61.8 | 0.000 |
| min | 22.4 | 37.3 | 0.56 | 51.9 | 24.7 | 75.5 | 43.8 | −2.446 |
| max | 267.2 | 193.7 | 6.52 | 90.5 | 33.5 | 166.4 | 87.5 | 2.853 |
| Ireland | ||||||||
| mean | 186.9 | 61.9 | 3.1 | 74.4 | 30.8 | 119.2 | 84.9 | 0.000 |
| min | 112.7 | 24.7 | 2.1 | 68.6 | 28.9 | 81.8 | 70.0 | −5.292 |
| max | 237.2 | 117.9 | 5.0 | 81.8 | 32.1 | 165.4 | 87.5 | 7.137 |
Data are seasonally adjusted.
13. The estimation results are shown in Table 2. We present only the fixed effects specifications as Hausman tests indicate that the coefficients under the random effects estimations are not consistent. While the specifications have in general a relatively low explanatory power (0.18-0.35), they suggest that household indebtedness is higher in countries with a high share of young and high home ownership. As expected, a higher house price-to-income ratio has a positive effect on household debt while higher interest rates on mortgages and lower public debt contribute to lower household indebtedness. The results also show that high unemployment (relative to its long-run average) positively affects household debt ratios, possibly representing an income shock and/or a high level of distress, which is associated with limited debt repayment capacity. In addition, better information regarding borrower characteristics and stronger legal rights, which reduce the risk of default, are correlated with higher household debt. Finally, the pre-crisis dummy, which obtains a value of one in the pre-crisis period 2003Q1–2007Q4 and zero otherwise, shows that, on average, household debt was higher in the post-crisis period.
Ireland: Determinants of Household Debt-to-Disposable Income
Instrumental variable estimation, two-stage least square (GLS2LS), Fixed Effects 2003q1-2016q3
difference from the country’ long-term average.
Ireland: Determinants of Household Debt-to-Disposable Income
Instrumental variable estimation, two-stage least square (GLS2LS), Fixed Effects 2003q1-2016q3
| (1) | (2) | (3) | (4) | (5) | (6) | |
|---|---|---|---|---|---|---|
| House price to income | 0.491*** | 0.478*** | 0.456*** | 0.550*** | 0.494*** | |
| Real interest rate on mortgages | −0.791 | −2.677*** | − 0.427 | −4.379** | −4.106** | |
| Share of population at age 25–44 | 7.293*** | 8.187*** | ||||
| Public debt | − 0.076 | −0.272*** | −0.127** | −0.128*** | −0.028 | |
| Home ownership | 0.014 | 1.337*** | 0.143 | −0.055 | 0.696* | |
| Getting Credit | 0.837*** | 0.719*** | 0.682** | 0.597*** | 0.483*** | |
| Unemployment 1/ | 2.940*** | 2.326*** | 3.360*** | 2.6269*** | 2.605*** | 2.451*** |
| Pre-crisis | −5.448** | −12.621*** | −4.614*** | −3.538*** | −14.419** | −14.500*** |
| Time dummy | yes | no | yes | no | yes | yes |
| R2 (within) | 0.649 | 0.527 | 0. 623 | 0.619 | 0.612 | 0.639 |
| R2 (overall) | 0.311 | 0.347 | 0.214 | 0.327 | 0.204 | 0.177 |
| # obs. | 395 | 395 | 401 | 407 | 402 | 395 |
| Hausman test p-value H0: RE H1: FE | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| # countries | 11 | 11 | 11 | 11 | 11 | 11 |
difference from the country’ long-term average.
Ireland: Determinants of Household Debt-to-Disposable Income
Instrumental variable estimation, two-stage least square (GLS2LS), Fixed Effects 2003q1-2016q3
| (1) | (2) | (3) | (4) | (5) | (6) | |
|---|---|---|---|---|---|---|
| House price to income | 0.491*** | 0.478*** | 0.456*** | 0.550*** | 0.494*** | |
| Real interest rate on mortgages | −0.791 | −2.677*** | − 0.427 | −4.379** | −4.106** | |
| Share of population at age 25–44 | 7.293*** | 8.187*** | ||||
| Public debt | − 0.076 | −0.272*** | −0.127** | −0.128*** | −0.028 | |
| Home ownership | 0.014 | 1.337*** | 0.143 | −0.055 | 0.696* | |
| Getting Credit | 0.837*** | 0.719*** | 0.682** | 0.597*** | 0.483*** | |
| Unemployment 1/ | 2.940*** | 2.326*** | 3.360*** | 2.6269*** | 2.605*** | 2.451*** |
| Pre-crisis | −5.448** | −12.621*** | −4.614*** | −3.538*** | −14.419** | −14.500*** |
| Time dummy | yes | no | yes | no | yes | yes |
| R2 (within) | 0.649 | 0.527 | 0. 623 | 0.619 | 0.612 | 0.639 |
| R2 (overall) | 0.311 | 0.347 | 0.214 | 0.327 | 0.204 | 0.177 |
| # obs. | 395 | 395 | 401 | 407 | 402 | 395 |
| Hausman test p-value H0: RE H1: FE | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| # countries | 11 | 11 | 11 | 11 | 11 | 11 |
difference from the country’ long-term average.
14. While the model provides useful information regarding some of the determinants of household debt, some important caveats should be recognized. First, the estimations’ relatively low explanatory power suggests that there may be time-varying factors that affect the level of household debt that are not directly captured in the specifications above (e.g., sentiment about future economic activity). Second, the sample covers a highly volatile period that includes a boom-bust cycles in several countries, including Ireland, elevated public debt levels, and an unusually accommodative monetary policy stance. Furthermore, widespread uncertainty regarding future economic conditions in the post-crisis period, which may not have been fully captured in the analysis, may have also affected household saving-borrowing behavior. Third, the analysis, which is based on a macro-level data, ignores the distribution of debt across households, and possible asymmetries between borrowers and creditors. Lastly, while we use a two-stage instrumental variable methodology, the possibility of endogeneity cannot be ruled out, particularly given that some of the variable lags are used as instruments.
15. With these caveats in mind, we turn to assess the impact of the identified determinants on Ireland’s deleveraging in recent years. We use the model’s coefficients and the actual values of the identified determinants to estimate household debt in Ireland. This exercise replicates the hump shape pattern of the Irish debt ratio in the past decade, though it shows a relatively large unexplained gap between actual and estimated debt levels. More specifically, the estimated debt ratio increases from about 140 percent of disposable income in early 2005 to nearly 185 percent at end-2009, largely on the back of higher house prices, the rising share of young prior to the crisis, and favorable consumer sentiment and economic conditions in the euro area, as captured by the time dummies. From 2010 onwards, the estimated model explains more than half of the actual deleveraging. The moderation of the estimated debt ratio is largely driven by the decline in unemployment, which points to improved economic conditions and a lower level of distress (especially after 2013), as well as a decline in the share of homeownership and young, lower access to credit, and moderation of house prices (mainly up to 2014). At 2016Q3, estimated debt was just below 140 percent of disposable income, but since the actual debt ratio declined more rapidly, the unexplained gap narrowed to about 15 percentage points of disposable income from nearly 55 percentage points in late 2009. All else equal, continued reduction in the unexplained gap would be consistent with further deleveraging in the near term.
Ireland’s Household Debt: Actual and Estimated 1/
(Percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: IMF staff calculations.1/ Household debt includes loans and other account payable, seasonally adjusted.2/ Based on specification 6 in Table 2.Ireland’s Household Debt: Actual and Estimated 1/
(Percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: IMF staff calculations.1/ Household debt includes loans and other account payable, seasonally adjusted.2/ Based on specification 6 in Table 2.Ireland’s Household Debt: Actual and Estimated 1/
(Percent of disposable income)
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: IMF staff calculations.1/ Household debt includes loans and other account payable, seasonally adjusted.2/ Based on specification 6 in Table 2.D. Key Takeaways
16. Ireland has experienced a significant reduction in household debt in recent years. Following a significant accumulation of debt in the pre-crisis period, Irish households have endured a prolonged period of adjustment. This paper provides a short overview of the deleveraging dynamics and household indebtedness using both aggregated and more granular data, and explores some of the factors that may have supported the adjustment process. The analysis suggests that:
Household debt declined sharply in recent years, but remained above the euro area average and levels seen in Ireland prior to the property boom. Moreover, a significant share of households, particularly young and high income households, remain heavily indebted with negative equity.
The composition of household debt holders changed significantly over time. While the banking system still holds the lion’s share of household loans, the share of household loans held by nonbanks registered a twofold increase compared to the early 2000s and stabilized at just below 40 percent of total household loans in recent years. This increase largely reflects securitization of loans by MFIs, sale of distressed assets by MFIs to non-MFIs, and also new lending by non-MFIs.
Household debt-to-disposable income was found to be higher in countries with a high proportion of young, high home ownership rates, better access to credit, high unemployment (relative to the long-term average), and a high house price-to-income ratio. A lower public debt-to-GDP ratio and real interest rate on mortgages also contribute to higher household debt.
The estimated model—although subject to several caveats—suggests that the identified determinants contributed to more than half of the decline in household debt since 2010, while the unexplained gap narrowed sharply from 55 percentage points of disposable income in late-2009 to about 15 percentage points at 2016Q3. All else equal, continued reduction in the unexplained gap would be consistent with further deleveraging in the near term.
References
Albuquerque, B., U. Baumann, and G. Krustev, 2014, “Has US Household Deleveraging Ended? A Model-Based Estimate of Equilibrium Debt”, European Central Bank Working Paper No. 1643.
Andritzky, J, 2012, “Household Consumption, Wealth, and Saving”, Ireland 2012 Article IV Report: Selected Issues Papers, International Monetary Fund.
Barnes, S. and G. Young, 2003, “The Rise in US Household Debt: Assessing its Causes and Sustainability”, Bank of England Working Paper No. 206.
Banerji. A., C. Ebeke, J. Siminitz, and H. Scholermann. “Household Debt in Europe: When it is Bad for Growth?” IMF Working Paper (Forthcoming).
Clancy, D., M. Cussen, and R. Lydon, 2014, “Housing Market Developments and Household Consumption”, Economic Letters, Vol 2014, No. 9, Central Bank of Ireland.
Coletta, M., R. De Bonis, and S. Piermattei, 2014, “Household Debt: a Cross-country Analysis”, Bank of Italy, 2014–0989.
Connor, G., T. Flavin, and B. O’Kelly, 2012, “The US and the Irish Credit Crisis: Their Distinctive Differences and Common Features”, Journal of International Money and Finance, Vol. 31 (1), 60–79.
Cussen, M., and G. Phelan, 2010, “Irish Households: Assessing the Impact of Economic Crisis”, Central Bank of Ireland, Quarterly Bulletin 04.
Cussen M., O’Leary B. and D. Smith, 2012, “The Impact of the Financial Turmoil on Households: A Cross Country Comparison”, Irish Central Bank Quarterly Bulletin Article 02, April.
Djankov, S., C. McLiesh, and A. Shleifer, 2007, “Private Credit in 129 Countries”, Journal of Financial Economics 84 (2007).
Du Caju, P., F. Rycx, and I. Tojerow, 2016, “Unemployment Risk and Over-indebtedness: A Micro-econometric Perspective”, European Central Bank Working Paper, No. 1908.
Dynan, K. and D. Kohn, 2007, “The Rise in US Household Indebtedness: Causes and Consequences”, Board of Governors of the Federal Reserve System, Finance and Economics Discussion Series 37.
European Central Bank, 2016, “The Role of Euro Area Non-Monetary Financial Institutions in Financial Intermediation”, ECB Economic Bulletin, Issue 4/2016.
European Central Bank, 2016, Household Finance and Consumption Survey.
Fasianos, A. R. Lydon, and T. Mclndoe-Calder, 2017, “The Balancing Act: Household Indebtedness Over the Lifecycle”, Quarterly Bulletin 02, Central Bank of Ireland.
International Monetary Fund, 2012, “Dealing with Household Debt”, World Economic Outlook, April 2012, Chapter 3.
Philbrick, P. and L. Gustafsson, 2009, “Australian Household Debt - An Empirical Investigation into the Determinants of the Rise in the Debt-to-Income Ratio”, Department of Economics, Lund University.
Jappelli, T., M. Pagano, and M. Di Maggio, 2008, “Households’ Indebtedness and Financial Fragility”, Center for Studies in Economics and Finance, Working Paper No. 208.
Lydon, R. and T. McIndoe-Calder, 2017, The Great (De)Leveraging 2005-14. Research Technical Paper, Central Bank of Ireland, 05/RT/2017.
Maninal, S. A., N. Akila Mond Kassim, C. S. F. Ho, and J. Mohd Yusof, 2016, “Preliminary Investigation on the Determinants of Household Debt Burden”, Proceedings of the 1st AAGBS International Conference on Business Management 2014 (AiCoBM 2014).
Meng, X., N. Hoang, and M. Siriwardana, 2013, “The determinants of Australian household debt: A macro level study”, Journal of Asian Economics, Vol. 29, December 2013.
Modigliani, F., and R. Brumberg, 1954, “Utility analysis and the consumption function: An interpretation of cross-section data” In: Kurihara, K., editor. Post-Keynesian Economics. New Brunswick, NJ: Rutgers University Press.
Annex I. Determinants of Household Debt1
Household Net Financial Assets, Percent of GDP
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.Household Net Financial Assets, Percent of GDP
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.Household Net Financial Assets, Percent of GDP
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: Eurostat.
Getting Credit, Distance from the Frontier, 2003–06
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: World Bank’s Doing Business.Getting Credit, Distance from the Frontier, 2003–06
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: World Bank’s Doing Business.Getting Credit, Distance from the Frontier, 2003–06
Citation: IMF Staff Country Reports 2017, 172; 10.5089/9781484305546.002.A002
Source: World Bank’s Doing Business.Correlation Matrix
Correlation Matrix
| Household debt to disposable income | Home ownership rate | Real interest rate on mortgages | House price to income ratio | Share of population at age 25-44 | Getting Credit | Public Debt to GDP ratio | Unemployment (diff. from long-term average) | |
|---|---|---|---|---|---|---|---|---|
| Household debt to disposable income | 1 | |||||||
| Home ownership rate | 0.182 | 1 | ||||||
| Real interest rate on mortgages | −0.005 | −0.223 | 1 | |||||
| House price to income ratio | 0.127 | 0.368 | 0.173 | 1 | ||||
| Share of population at age 25-44 | 0.201 | 0.459 | 0.119 | 0.181 | 1 | |||
| Getting Credit | 0.389 | −0.311 | −0.001 | −0.170 | 0.104 | 1 | ||
| Public Debt to GDP ratio | −0.412 | −0.155 | −0.139 | −0.323 | −0.117 | −0.431 | 1 | |
| Unemployment (diff. from long-term average) | 0.308 | 0.142 | −0.417 | −0.197 | 0.146 | 0.090 | 0.257 | 1 |
Correlation Matrix
| Household debt to disposable income | Home ownership rate | Real interest rate on mortgages | House price to income ratio | Share of population at age 25-44 | Getting Credit | Public Debt to GDP ratio | Unemployment (diff. from long-term average) | |
|---|---|---|---|---|---|---|---|---|
| Household debt to disposable income | 1 | |||||||
| Home ownership rate | 0.182 | 1 | ||||||
| Real interest rate on mortgages | −0.005 | −0.223 | 1 | |||||
| House price to income ratio | 0.127 | 0.368 | 0.173 | 1 | ||||
| Share of population at age 25-44 | 0.201 | 0.459 | 0.119 | 0.181 | 1 | |||
| Getting Credit | 0.389 | −0.311 | −0.001 | −0.170 | 0.104 | 1 | ||
| Public Debt to GDP ratio | −0.412 | −0.155 | −0.139 | −0.323 | −0.117 | −0.431 | 1 | |
| Unemployment (diff. from long-term average) | 0.308 | 0.142 | −0.417 | −0.197 | 0.146 | 0.090 | 0.257 | 1 |
Prepared by Nir Klein. I would like to thank Tara McIndoe-Calder and the participants of the workshop at the Central Bank of Ireland for their useful comments and suggestions.
Household debt includes loans and other accounts payables, and is seasonally-adjusted.
Empirical evidence suggests that economic downturns tend to be more severe and prolonged when they are preceded by excessive increases in household debt (see for example IMF, 2012).
Connor and others (2012) show that the proportion of long maturity loans (over 30 years) increased from 10 percent to 35 percent between 2004 and 2007.
Banerji and others (forthcoming) use the ECB’s Household Finance and Consumption Survey to identify the debt level at which households are likely to become credit constrained. Their preliminary results, which are based on cross-country analysis, indicate that a significant proportion of Irish households (40-50 percent) in the bottom three quintiles have debt above the thresholds, and are likely to remain credit constrained.
OFIs include financial auxiliaries, captive financial institutions, money lenders, and financial vehicle corporations.
Household borrowing from the NFC sector and the government is less than 10 percent of non-MFIs lending.
Retained securitizations are securitizations, which are brought back onto the balance sheet of the bank and used as collateral with the ECB as part of monetary operations. Preliminary Central Bank of Ireland analysis of Q3 2016 data indicates retained securitizations represent a significant percentage of overall non-MFI lending to Irish households.
As some of the factors are purely cyclical, the analysis aims to explore which factors have contributed to household debt dynamics in recent years, rather than to identify the “equilibrium” or sustainable debt level across countries.
This indicator may also capture household wealth, which is largely affected by housing assets. Cross-country data on housing assets are not available.
The countries are Austria, Belgium, Finland, France, Germany, Greece, Italy, Ireland, the Netherlands, Spain, and Portugal.
Getting Credit is measured as the distance from the frontier. Higher value indicates that the distance from the frontier (100) is closer.
The variable lags as well as financial assets and employment are used as instruments.
Get credit is measured as distance from frontier. Higher value suggests that the distance from frontier is smaller.
