This is like the Great Irish Bake-off but all about delevaging, Central Bank economists Reamonn Lydon and Tara McIndoe-Calder put together an excellent paper (05/RT/2017)on the topic, the full technical paper is here.

Our condensed and plain English version is below:

ABSTRACT
The authors drew on the 2013 household Finance and Consumption Survey (HFCS) to stimulate household balance sheets form 2005 to 2014 for the purposes of investigating household leveraging and deleveraging during this period. The paper shows that deleveraging has proceeded significantly faster with older households as opposed to younger ones. With younger borrowers, tracker mortgages have eased the debt repayment burden in the presence of large income shocks. All in all, income shocks are the main factor contributing to mortgage repayment problems.

INTRODUCTION
From the early 2000s through to the peak of the property boom in 2007, rapid increases in leverage ratios and repayment burdens far outstripped growth in disposable income, leaving households exceptionally vulnerable to the economic shock of 2008. One result of the crisis was the large increase in non-preforming mortgage loans.
The authors of this paper attempts to understand how different households are affected by the crisis, drawing on a range of data that they have cumulated in a dataset labelled HFCS-SIM. The HFCS-SIM is able to accounts for the heterogeneous nature of households and highlights how some borrowers can disproportionally benefit from the change in ECB policy rates.

The income component of HFCS-SIM allow the authors to better stimulate changes when shocks have differential impacts on households in different parts of the asset, debt and income distributions.

THE HFCS AND OTHER DATA USED IN THE ANALYSIS
The Irish Household Finance and Consumption Survey (HFCS) is the predominant dataset used in the authors’ analysis. It was conducted by the Irish Central Statistics Office between March and September of 2013 and covers income, employment, and more importantly, gross wealth and debt. Significant patterns in the data show that Household Main Residence represents the bulk of both assets and debts in all age groups. Mortgage debt accounts for the largest share of household debt and declines with age. In older age groups, residential investment property and business property also represents a significant chunk of assets.

Other datasets used by the authors to mainly cross check various data on loans, debts and incomes includes: Loan-Level Data (LLD, Central Bank of Ireland); Quarterly Financial Accounts (QFA, Central Bank of Ireland); Money and Banking Statistics (MBS, Central Bank of Ireland); Administrative Data on Earnings from Work (TAX, CSO); and Survey of Income and Living Standards (SILC, CSO).

CONSTRUCTION OF HFCS-SIM
This section details how the authors constructed each of the three component of their HFCS-SIM model: assets, liabilities and income.

In the model, gross assets consists of property (77.6%), other real assets and financial assets. Property is also split into the subsections of household main residence (47.8%), other residential property, holiday homes and other property.

For the household main resident, the authors used the price of the property at the time of purchase and the property’s location, type, and size to calculate a regression with fitted values. A cumulative distribution of HMR house values in the HFCS in 2013 and the simulated distribution in 2006 show a 55% peak to trough fall in house prices in Ireland reflective of the scale of the asset price shock. At the top of the distribution, the nominal euro value fall is around 200,000 and at the bottom end, the fall is around 70,000.

For the liabilities component of the HFCS-SIM, household debt consists of HMR mortgage debt (71.6%), other property mortgages, and non-collateralised debt. Outstanding balance on mortgage debt is calculated with regards to monthly payments, interest rates and terms remaining.

The authors also validated changes in net wealth to analyse leverage and debt distress trends over time for different facets of the population. Data reveals that from 2007-2013, across the middle income groups, the percentage change in total net wealth is around -45%. The bottom of the income distribution saw no change in net wealth during the same period due to significantly lower property ownership rates and the top of the income distribution saw a lesser -38% change in net wealth due to households in this bracket having a more diversified portfolio. For households with property assets, there is a net wealth loss across the board, but the bottom of the distribution saw exceptionally large losses at over 100% of initial property wealth. Younger households were also disproportionally hit, seeing almost all of their wealth wiped out. This is the result of large property price declines combined with high initial debt levels.

To analyse how debt repayment burden evolves over time, the authors also simulate changes in household incomes to go alongside the debt repayments series described above. The backbone of the income stimulation is a dataset on earnings from work from the year 2005 to 2014, which contains information on weeks of work and annual earnings for each individual in the HFCS data.

DELEVERAGING, INCOME SHOCKS AND REPAYMENT
This section describes changes in household indebtedness over time, focusing on property related debt, and the evolution of debt repayments relative to income. Overall, since 2007, Irish households have experienced a significant decline in their disposable income due to a combination of job losses, pay cuts and higher taxes. The HFCS-SIM highlights a disparity in interest rate pass through, and shows how it falls along age lines.
The authors plot the evolution of debt, disposable income and mortgage debt repayments for three groups: Households born during or after the 1970s, households born in the 1960s and households born before the 1960s. As expect, the youngest group have the sharpest increase in debt in the run up to the crisis, a 20% increase, which is more than twice the rate of income growth. Since 2009, older households have reduced their debt levels at a much faster rate, with debt falling by over 30% up to 2014. The main reason for this is the lower debt level and shorter mortgage terms for this group of borrowers.

The sharp fall in ECB policy rates in 2008 and 2009 led to a fall in debt repayments for all age groups, particularly, younger borrowers saw an almost 30% fall in their debt burden. The older ago groups exhibit lower pass through from changes in the ECB policy rate because fewer of them are on tracker mortgages. Since 2009 onwards, crisis-hit lenders change their approach to setting rates for tracker versus other variable rate mortgage loans, thus despite larger declines in disposable income between 2008 and 2013, borrowers with tracker loans benefited from significantly large payment reductions.

DEBT REPAYMENTS AND INCOME SHOCKS
The income shock that hit Irish households after 2008 led to widespread mortgage repayment problems. The authors use income time series and information on loan modifications to attempt to understand how households responded to the income shock.

Mortgage arrears statistics published by the Central Bank of Ireland show that by early 2014 one out of every eight loans had been modified in some way, and around one third of modified loans moved to IO or term extensions. In this section the authors compare the scale of the income shock, the debt service burden and other characteristics of borrowers that sought out loan modifications during the recession. On average, households that sought out a modification experienced greater negative income shocks between 2008 and 2013. The median debt to disposable income ratio for IO borrowers is 5.2, compared with a range of 2.5 to 3.1 for all other groups. The authors highlight the significant benefit to borrowers from modifying their loan terms. In 2013, the median IO borrower’s monthly repayment was almost €160 lower than a principal plus interest repayment arrangement. The reduction in repayments for borrowers who extended their loan terms was slightly smaller.

INCOME SHOCKS AND MISSED MORTGAGE DISTRESS
In this section, the authors attempt to illustrate how income shocks lead to debt distress, focusing on understanding outstanding missed payments or arrears.

The authors compare trends in average repayment rations and outstanding loan to value rations between stressed borrowers (households that have missed a mortgage repayment), and non-stressed borrowers (households that have not). The major difference is that households that have missed a mortgage repayment have a significantly higher repayment burden. This suggests that the amount of ex-ante financial headroom a household has is an important factor in determining their ability to cope with repayment shocks. After 2008, stressed borrowers saw larger income declines and a slower recovery. More generally, 23% of stressed borrowers lost their jobs between 2007 and 2012, compared to only 7% for non-stressed borrowers.

Qualifying the relationship between income shocks and mortgage distress, the authors found that households that experienced larger negative income shocks between 2008 and 2013 are significantly more likely to be in arrears. For households that experienced at least a 10% income drop, the likelihood of being in arrears is increased by between 4.4% and 6.6%. The authors also conclude that gross liquid assets to income ratio is a potential buffer against income shocks. Supported by the finding that households with more savings are less likely to be in arrears.

For households that entered the recession with an already high debt service burden, the marginal effects on the income hock variable almost doubled. The job loss and unemployment duration effects have also doubled in size. The authors believe that this is an important result for macro-prudential policy, as it suggests that policies aimed at reducing the number of households that have a high debt service burden can have very positive effects for the ability of indebted households to withstand negative income shocks.

CONCLUSION
The authors of this paper makes use of household information on asset values, debt and income in the 2013 Irish HFCS to simulate and re-construct the balance sheets of over 5,000 households from 2015 to 2014.
This simulation provided key insights into how different households were affected by the recession, both in terms of shocks to their income and asset price declines. Net wealth to disposable income have declined sharply from the 2007 peak, with older ago groups suffering the largest losses due to the high concentration of gross wealth in property assets. In terms of level of wealth, the heaviest levels of losses are concentrated at the bottom end of the loss distribution. This end is composed of mainly younger households, who started out with low net wealth positions. To illustrate, the median net wealth for households in the bottom quintile in 2006 is €1,700, and by 2013, these households have been driven into a negative equity position, with net wealth holdings of €-31,000.
The HFCS-SIM dataset also gives a sense of how leverage, measured as the debt to disposable income ratio, has evolved other time. On average, HMR mortgage debt fell from €190,000 in 2007 to €160,000 by 2013. However, once we take into account of the fall in income in the same period, a very different picture emerges. Younger households saw large increases in their leverage ratios from 2006 to 2010 (rising 75%), largely due to a fall in disposable income. The widespread use of mortgage modifications among this group also plays a significant role in slowing the rate of mortgage debt repayment. However, the prevalence of tracker mortgages amongst this group provided relief to the effects of the income shock. Those without tracker mortgages did not benefit to the same degree.

Finally, the authors also applied the HFCS-SIM to examine factors associated with mortgage repayment problems. Accounting for income shocks levels, households that have built up a financial buffer through income levels, employment shocks and debt-service ratios are less likely to miss a payment. All other things held constant, households in negative equity are also more likely to miss a payment.
All in all, this paper highlights the importance of accounting for distributional effects when examining wealth, leverage and debt distress from a policy perspective.

The applicability of the results to a non-Irish context depend heavily on comparability between institutionally similar environments elsewhere.