Impact of COVID-19 on the Debt Sustainability of Irish Households Q3 2020

Restrictions to prevent the spread of the COVID-19 virus have led to business closures and loss of work in some sectors, particularly in businesses providing face-to-face services. This has necessitated the introduction of substantial State supports for firms and workers which have helped mitigate the short-term impact on incomes and thus supported households sustain their debt burdens through 2020.

Building on earlier analysis of individual earnings from Central Statistics Office (CSO) in a November Labour Market Bulletin and December 2020 Frontier analysis of COVID-19 income support schemes, this Frontier Publication shows changes in household incomes up to Q3 2020.  It shows how incomes have evolved during 2020, both across the distribution and by characteristics of the household and the household reference person.

A key question for financial stability and household resilience is how the COVID-19 income shocks are affecting debt-sustainability metrics. By combining administrative income data with survey data on household balance sheets, we show how debt-to-income and debt-service ratios (the proportion of gross income that goes towards debt servicing or repayments) changed in 2020. Debt-sustainability metrics weakened (increased) in Q2 2020, even when COVID-19 supports such as the Pandemic Unemployment Payment and Wage Subsidies (TWSS/EWSS) are included. They strengthened (declined) again in Q3 2020 as many, but not all, workers returned to work.  In almost all cases, these debt metrics were at, or were very close to, pre-COVID-19 levels by Q3 2020.  For the most indebted households, COVID-19 supports played a crucial role in limiting sharp increases in debt-service ratios and were therefore important for debt sustainability.

This Frontier Publication is the product of a joint research collaboration between the Central Statistics Office and the Central Bank of Ireland. The analysis is an experimental piece where new methods are used, new data sources are combined and analysed to provide timely policy relevant information and insight to policy makers on the wider economic effects of COVID-19.  Recent history tells us that large income shocks for indebted households can have large negative consequences for macro outcomes like financial stability and domestic demand (consumption), as well as heterogenous effects when certain groups of the population are more negatively affected than others.  With this in mind, the primary application of this research is to household debt sustainability.

Data Sources

The Household Finance and Consumption Survey (HFCS) is the official source of household debt and wealth statistics in Ireland. This publication uses data from the CSO’s HFCS 2018 combined with administrative data sources of income and COVID-19 income supports for the period Q1 2019 to Q3 2020. This allows us to estimate changes in gross household income and debt ratios by distribution and household characteristics. Income and debt developments are estimated both with and without selected COVID-19 income supports.  This gives a sense of the mitigating role of these supports.  However, following the approach in the CSO’s publication Impact of selected COVID-19 income support on employees (December 2020), the counterfactual ‘without supports’ makes no attempt to estimate likely incomes in case of receipt of ‘traditional’ out-of-work supports like Jobseeker's Allowance or Jobseeker's Benefit. 

The administrative income data matched to HFCS 2018 microdata include: employee earnings, from the Revenue Commissioners' (Revenue) PAYE Modernisation (PMOD) payroll data; public pension payments, unemployment payments and regular social transfers, from Department of Social Protection (DSP) payment data; the Pandemic Unemployment Payment (PUP), from DSP data; and the Temporary Wage Subsidy Scheme (TWSS) and the Employment Wage Subsidy Scheme (EWSS), from Revenue. Where no administrative income data is available the HFCS 2018 data is utilised instead. Further information on the data sources used in this publication can be found in the Background Notes.

 

The key findings from this analysis are as follows:

• In Q2 2020, when COVID-19 restrictions were tightest in 2020, median gross household income fell by 1.7% in the year. This measure of gross income includes COVID-19 income supports, specifically the Pandemic Unemployment Payment (PUP) and Wage Subsidies (TWSS/EWSS).
• In Q3 2020, when restrictions were eased, median income with supports increased by 3.0% in the year.
• Without COVID-19 income supports, median income would have fallen an estimated 19.6% and 5.7% in the year to Q2 and Q3 2020 respectively. This assumes no other replacement income such as pre-existing supports like Jobseeker’s Allowance and Benefit.
• The COVID-19 labour market shock affected the lower half of the income distribution most significantly. Household income, without supports, in the bottom half of the distribution would have fallen by between 18% and 30%, in the year to Q2 2020, compared to falls of between 7% and 18% in the top half of the distribution in Q2.
• Households most negatively impacted by the crisis benefited the most from the COVID-19 income supports, relative to their pre-COVID-19 income. Household income, with COVID-19 income supports, fell by between 0.1% and 4.2% in the year to Q2 2020 in the bottom half of the household income distribution, but would have fallen by between 18% and 30% in the year to Q2 without supports.
• The estimated ‘gap’ between income growth with and without supports was between 14 and 26 percentage points for the bottom half of the income distribution in Q2 2020, compared to a ‘gap’ of between 5 and 16 percentage points for the top half of the distribution.   
• For households with debt, the debt-to-income ratios for ‘all debt’ increased slightly in Q2 2020, from 60.7% to 60.9% of gross income for the median household, before falling again in Q3. Without COVID-19 income supports, the debt-to-income ratio would have increased 12.3 percentage points to 73.0% in Q2 and fallen to 68.0% in Q3.
• Debt-to-income ratios increased by larger amounts for more indebted households. For example, the debt-to-income ratios of the 90th percentile increased from 342.0% to 376.1% of income from Q1 to Q2 2020. Without COVID-19 income supports, the increase would have been considerably larger, bringing the ratio up to 552.3% of income.
• Debt-service ratios, which measure the amount of gross income used to service or repay debt, followed a similar pattern, increasing marginally to 11.7% of gross income (for all debt) in Q2 2020 for the median household, before falling slightly in Q3. Without COVID-19 income supports the debt-service ratios for the median household would have been 13.6% of income in Q2 and 12.3% in Q3.
• Debt-to-income ratios of Household Main Residence (HMR) mortgage debt increased to 152.9 in Q2 2020 and fell to 150.9 in Q3 for the median household with this debt. Without support the ratios would have been 171.0 in Q2 and 156.6 in Q3.
• Debt-to-income ratios of mortgage debt for the most indebted households (90th percentile) increased from 390.0 to 432.7% of income, with supports, in Q2 2020. Without the COVID-19 income supports, the debt-to-income ratio would have been 665.0% for the 90th percentile in Q2, 6.65 times the values of the mortgage.
• Debt-service levels of owner-occupier mortgage debt increased by 0.6 percentage points to 14.8% for the median household in Q2 2020, before falling slightly in Q3. Without COVID-19 support income this debt service ratio would have increased 2.9 percentage points to 17.1% of gross household in the Q2. Please note that this analysis does not account for mortgage payment breaks.

Household Income

This section summarises the changes in household incomes during 2020. The income concept is Total gross household income, as set out in the background notes. We show year-on-year changes in each of Q1, Q2 and Q3 2020. All data is nominal, i.e. there is no adjustment for changes in price levels that could affect purchasing power. We first show changes across the deciles of household income, both with and without selected COVID-19 income supports, including the Pandemic Unemployment Payment (PUP), The Temporary Wage Subsidy Scheme (TWSS) and the Employment Wage Subsidy Scheme (EWSS). This is followed by a comparison of changes according to the characteristics of the household and household reference person.

Year-on-year changes in gross household income by decile

Figure 1.1 plots the annual change in gross household income in the second quarter of 2020 for ten percentiles of the income distribution. The green line shows the change in income without COVID-19 income supports, the blue line including supports.

The without supports line in Figure 1 shows that the COVID-19 labour market shock was heavily concentrated amongst lower income households. Without supports, household incomes in the bottom half of the distribution would have fallen by between 18% and 30%, in the year to Q2 2020. In contrast, higher income households would have seen their income without supports fall by ‘only’ 12.9% and 7.7% for the 80^th and 90^th percentiles respectively. This observation is consistent with the concentration of the shock in face-to-face services sectors, where workers tend to be both younger and lower earners, on average. 

The gap between the green and blue lines (those without and with supports respectively) illustrates both the scale of the mitigating effects of the COVID-19 income supports and how those most negatively impacted by the crisis benefited the most from the supports, relative to their pre-COVID-19 income. Whilst households in the lower half of the income distribution still experienced falls of between 0.1% and 4.2%, with supports, the slope of the distribution (blue line) is much flatter than the without supports (green) line. It is also notable that higher-income households benefited quite significantly from COVID-19 supports. For example, in the top half of the distribution, the ‘gap’ between household income growth with and without supports ranges from 5 to 16 percentage points. The gap in the bottom half of the distribution was between 14 and 26 percentage points. Therefore, whilst much of the labour shock was concentrated in the bottom half of the household income distribution, it is not the case that higher income households were completely unaffected. See figure 1.1 and table 1.1.

Further analysis of COVID-19 income support recipients showed that in Q2 2020, 22.2% of support payments were made to persons in the upper half of the personal earnings distribution. When people were combined into households and when household income was equivalised to account for household size, the proportion of COVID-19 income support recipients in the upper half of the distribution was 50.7% in Q2 2020. See table 2.6 in Background Notes.

Figure 1.2 shows the year-on-year changes for Q3 2020.  The data shows that the return to work in the third quarter for many boosted incomes.  Even with this improvement, COVID-19 income supports were still required to raise annual income growth into non-negative territory, for all bar the 20^th and 70^th percentiles of households. See figure 1.2 and table 1.1.

Year-on-year changes in median gross household income by household characteristic

Table 1.2 shows the year-on-year changes in median gross household income, by characteristic of the household and household reference person (see the Background Notes for a description of the household reference person).  To give a sense of how household groups might impact the aggregate picture, the first two columns of the table show the percentage of households in the population with a given characteristic, and the percentage of households in each group that have owner-occupier (i.e. Household Main Residence) mortgage debt. 

Across all households, there was a relatively consistent pattern of year-on-year income declines in Q2 2020, followed by positive changes in the year to Q3. There was, however, considerable variation across characteristics.  Single-adult households with children under 18 years experienced an increase in income of 5.7% (with supports) in Q2 2020. This compared to falls of between 0.2% and 5.6% in other household types. This partly reflects the larger role that COVID-19 income supports play for this group, relative to pre-COVID-19 income levels. This increased income in the year to Q2 was followed by a further increase in income in Q3. This pattern was also apparent for households in rented accommodation and households where the education level of the reference person was Secondary or lower.

Pre-empting the debt-sustainability analysis, the final rows of Table 1.2 show median income changes by housing tenure status. Owner-occupiers with a mortgage experienced a fall in income in Q2 2020 (-2.6%) before returning to annual growth in Q3 (+0.9%). However, COVID-19 supports were playing a key role, increasing income growth by between 9 and 18 percentage points in Q3 and Q2 respectively, relative to the case without supports. Rented household were most significantly supported by COVID-19 income supports, relative to their pre-COVID income, with a “gap” of 31 percentage points between income with and without supports in Q2 2020. This gap was 12.5 percentage points in Q3 2020. See table 1.2.

Debt Sustainability

Unsustainable levels of household debt have the potential for widespread, and long-lasting negative effects on households, the financial system and the economy.  The high levels of long-term mortgage arrears that remain after the financial crisis are one reminder of this.  We also know that high levels of household indebtedness in the face of income shocks drags on domestic demand via lower consumption.

The analysis in this section is in two parts. First, we show gross debt-to-income and debt-service ratios for households with any debt, both across the distribution and by household characteristics. All debt ratios are conditional on having debt.  As mortgage debt tends to be much larger than non-mortgage (i.e. consumer) debt, debt ratios higher-up the distribution in this section will reflect this.  Secondly, we repeat this analysis, focusing only on owner-occupier (HMR) mortgage debt.

All Debt

Figure 1.3 shows the median debt-to-income ratio, conditional on households having any debt, 52% of HFCS households. The distribution of debt can have a long ‘right-tail’, that is households above the median with significantly higher debt levels than the median. This is illustrated in Table 1.3 where the 80^th and 90^th percentiles of the distribution are presented. 

Prior to the onset of COVID-19, the debt-to-income ratio was in steady decline, reflecting the general pattern of household deleveraging that has been happening over the last decade. For example, between 2013 and 2018, the median debt ratio for all debt declined from 100 to 64 .  In Q2 2020, the median ratio with COVID-19 income supports increased by around 0.2 percentage points from 60.7 in Q1 2020 to 60.9 in Q2 2020, before declining again to 60.7 in Q3.  As with the income statistics, COVID-19 supports mitigate the increase in the debt ratio. Without supports the median ratio would have increased by 12.3 percentage points to 73.0% in Q2 2020, and would have fallen to 68.0 in Q3. See table 1.3 and figure 1.3.

Table 1.3 shows the median debt-to-income ratio (all debt) across the distribution and by household characteristics and those of the household reference person.

Focusing on the most indebted households – that is, the 80^th and 90^th percentiles – even with COVID-19 supports, there were sharp increases in the debt-to-income ratio in Q2 2020, before falling back in Q3 to be above pre-COVID-19 (Q1 2020) levels. The second set of figures, which shows debt-to-income in the absence of supports, once again highlight the mitigating effects of the COVID-19 income supports even for the highest income households.  For the 90^th percentile, the debt-to-income ratio would have increased by 209 percentage points to a ratio of 552.3 – that is 5.52 times gross income – in Q2 2020, without the supports. This is compared to the actual increase of 34 percentage points that occurred with the supports. 

Looking across the median debt-to-income ratio by characteristic of the household, supports were particularly beneficial to households with a mortgage (reflecting the fact that these households were higher up the ‘all debt’ distribution), households with 2 or more adults with children under 18 years, households with the reference person aged 30 to 65 or with and education level of Third Level. These groups overlap with the owner-occupied households with mortgage debt group. See table 1.3.

The debt-to-income ratio is a measure of leverage relative to income. To also understand debt sustainability, we look at the debt-service ratio, which tells us the proportion of gross household income that goes towards servicing or repaying debts. We take debt repayments as reported by households in the HFCS in 2018. As the data methodology in the Background Notes shows, the distribution of these debt levels and repayments in the 2018 HFCS was very similar to that of the population in mid-2020, when benchmarked against the Central Credit Register.

Figure 1.4 shows the changes in the median debt-service ratio of all debt, with and without COVID-19 income supports. Table 1.4 shows the year-on-year changes across the distribution and by household characteristics and those of the reference person. Much like the debt-to-income ratio up to 2020, the debt-service ratio was declining year-on-year before the onset of COVID-19. In Q1 2020, and conditional on having debt, 11.2% of gross household income for the median household was used to repay or service debt. For comparison, in 2013, the figure for the median household was 15% (from HFCS 2013) and was 12.2% in 2018. 

In Q2 2020, the debt-service ratio including COVID-19 supports increased by just 0.5% to 11.7%, before falling back again in Q3 to 11.6%. The increases were smaller than the headline income changes for all households shown in Table 1.2. One reason for this is that the income falls for indebted households in Q2 2020 were smaller than for other groups. Without COVID-19 supports, the debt-service ratio rose to 13.6% in Q2, before falling back to 12.3% in Q3 2020.

By definition, debt-service ratios at the 80th and 90th percentiles were higher, fluctuating between 21.7% to 22.5% (80th) and 30.1% to 32.9% (90th) in Q3 and Q2 2020. Without supports, the peak ratio – in Q2 for the 90th percentile – was 48.6%. Whilst there is no set definition of a ‘sustainable’ debt-service ratio, macroprudential policy research often cites 30% as a threshold. It is also worth noting that these are gross debt service ratios and would be higher if calculated with net income at households’ disposal. See table 1.4 and figure 1.4.

Household Main Residence (owner-occupier) mortgage debt

This section focuses on debt-ratios for households with Household Main Residence (HMR) mortgage debt only, just over one-quarter (26%) of households in HFCS 2018.  Figure 1.5 shows the HMR mortgage debt-to-income ratio of all households with this type of liability.  The pattern here is similar to that for all debt, albeit at higher levels of both debt-to-income and debt-service.  Levels are higher because mortgage debt is typically larger than non-mortgage, or consumer debt. For example, the conditional median value of mortgage debt in HFCS 2018 is around €125,000, compared to €1,000 to €5,000 for non-mortgage debt, depending on the type of debt.

The greater burden of mortgage debt relative to income also contributes to a larger change in the ratios in Q2 and Q3 2020.  For example, and focusing again on the most indebted households, with COVID-19 income supports, the increase in debt-to-income ratio was between 16 and 43 percentage points in Q2 (80th and 90th percentiles respectively), before falling back again in Q3.  When we exclude supports, the 90th percentile debt-to-income ratio was 665 in Q2 2020. That means debt levels over six and a half times those of income.  It is notable that the 90th percentile HMR debt-to-income ratio in 2013 was lower than this, at 525. However, with supports, a ratio of 433 in Q2 2020 was below the 2013 level.

When assessing the median HMR debt-to-income ratio by household characteristics, supports were particularly important to households where the reference person had an education level of Second level or lower and in single parent households with children under 18 years. See table 1.5.

Figure 1.6 and Table 1.6 show the HMR debt-service statistics. In Q2 2020, debt-service on HMR rose by 0.6 percentage points to 14.8%, before falling again to 14.4% in Q3.  The 90^th percentile of debt service, including COVID-19 income supports, increased to 36.2 in Q2 from 33.6% in Q1 2020.  Without supports, gross debt service for the 90^th percentile would have peaked at 58.6% in Q2 and remained elevated at 39.9% in Q3.

While debt-service ratios on HMR were estimated to have risen in Q2 2020, it must be noted that up to 10% of mortgage holders received mortgage payment breaks during the COVID-19 crisis. Higher debt-service would not affect household in receipt of these payment breaks. We do not account for these mortgage breaks in this analysis.

Looking across household characteristics and those of the reference person, the mitigating effects of supports was largest for households with non-degree level education levels and single parent households.  This is broadly similar to cross-sectional pattern we saw for the mitigating effects of supports on income in Table 1.2. See tables 1.6 and figure 1.6.