The Impact of COVID-19 Income Supports on Earnings

[Disclaimer: This blog represents my personal views and not those of the Central Bank of Ireland or the European System of Central Banks.]

Last week, the CSO released the latest in its Insights from Real Time Administrative Sources series on the “Impact of COVID-19 Income Supports on Employees”.  Using administrative data on employee earnings and COVID-19 support scheme payments (PUP and T/EWSS), it tracks the individual gross weekly earnings of matched cohorts to Q4 2020 (‘matched cohort’ means the CSO is comparing the (median) earnings of same group of individuals year-on-year).* 

For policy makers deliberating the future of COVID supports (PUP and WSS), the information in the release provides several useful insights.** I highlight three aspects in this post: (i) the strong earnings growth of non-recipients of supports throughout 2020; (ii) the diverse earnings experience of different demographic and income groups; and (iii) the different earnings growth experienced by PUP versus T/EWSS recipients, which, for the first time, is shown separately in this release.  Some more details on each below.

For non-recipients, earnings grew strongly throughout 2020
One of the most striking observations is the strong growth in earnings for non-recipients of either PUP or T/EWSS, as shown in Chart 1 below.

Note: A recipient is defined as an employee who received at least one payment of either PUP or WSS during the quarter. All the comparisons are of growth in the median. The median is the mid-point in the distribution of the level of earnings, i.e. half of earners in a given group earn more, half earn less.

The fourth quarter, with year-on-year earnings growth of 7.1%, is particularly strong for non-recipients (as we already knew from the previous release to Q3, the group of all recipients experienced large falls in earnings in 2020). Year-on-year growth for the bottom 20% per cent of (non-recipient) earners was over 21% in Q4 (see Figure 1.5 in the release). The release notes that this increase is mainly among younger age groups, and workers in the wholesale & retail and health sectors. Some of this likely reflects the steep slope of the age-earnings profile for younger workers, and also perhaps longer hours worked during the pandemic. However, the CSO also notes that it reflects individuals changing jobs during the pandemic, which could point to more permanent effects on the level of earnings.

But it is not just at the bottom where we see earnings growth for non-recipients. We see growth right across the distribution. For example, the top-20% of (non-recipient) earners registered growth of 4.7% in Q4 2020. In fact, the combination of strong growth with the size of the non-recipient group (75% of employees in Q4), puts growth for the median of all workers (i.e. non-recipients plus recipients) in positive territory in Q4 (+4.6%). This is a remarkable outturn for the year that was in it. As highlighted in this piece with Tara McIndoe-Calder in March, the fact that average household incomes actually grew during 2020 is one of the reasons for the large jump in savings we saw during the year.

The earnings growth of recipients differs significantly across demographic groups
Reflecting the concentration of shut-downs in face-to-face services and some retail sectors, we know that the pandemic has affected the employment of some groups more than others, notably females, lower paid and younger workers (see, for example, this recent speech by Deputy Governor Sharon Donnery).  However, when we look at earnings growth, and including supports, the same groups tend to be the least affected during 2020 (see Table 1 below, which is based on Table 1.2 in the release).  The fact that some groups fare less badly reflects, in part, the level of supports relative to pre-pandemic earnings. We know, for example, that younger and female workers generally work fewer hours during the week (i.e. part-time), and tend to work in lower-paying sectors.  So, these are almost like ‘reference effects’. Note that the CSO analysis is of gross earnings, i.e. before income tax, and there may be future tax liabilities for some recipients where supports were not taxed at source.

One thing we do not see in the Q4 data – at least for growth at the median – is much evidence of an impact from the change in payment rates from mid-October onwards, when payments were more closely aligned to pre-COVID gross earnings. In my paper with Brian Cahill (CSO) from February, when we only had data to Q3 2020, we thought this could change income for some groups. But we do not see it in this release. As the payment could have both risen for some (the top rate increased to €350) and fallen for others, it could be washed out in the median, and we would need to dig down to the individual data to see it.  The Christmas Bonus, when there was additional PUP payment in December, could also be a factor.

Very different earnings growth by type of support and demographic group
The release provides, for the first time, a breakout of earnings growth (plus supports) for the two different types of supports, PUP and T/EWSS***. Whilst I focus on earnings here, the update on numbers receiving various supports released last week is also relevant.  

In Q2 and Q3, earnings for PUP (only) recipients fell by around 5% (for the median), less than half of the fall for T/EWSS recipients (Chart 2, below). So, even with employer top-ups, the gross earnings of (all) wage subsidy recipients fall a lot during this period (at the median). The one positive is that by Q4 both groups’ earnings were back at around Q4 2019 levels.   

Focusing on all recipients misses some of the variation within demographic or regional groups (Charts 3-5). For some workers – like males, older workers and workers in Dublin – wage subsidy recipients fare ‘better’ in terms of earnings growth, than PUP recipients in these groups. I say ‘better’, but in all quarters, earnings plus wage subsidies are still significantly lower than pre-COVID earnings for these sub-groups.

In contrast – for females, younger workers and workers outside of Dublin – PUP recipients fare better than wage subsidy recipients, and, in almost all quarters, earnings plus PUP are higher than pre-COVID earnings (although this is less so the case for the non-Dublin group).  For wage subsidy recipients in these sub-groups, earnings in Q2 and Q3 are below pre-COVID levels (except for under-25s), before recovering somewhat in Q4.

The timing of different health restrictions, and specifically which sectors were closed and when, likely explains some of the differences we observe between certain groups. For example, the closure of construction, with higher earnings on average and more male workers, could explain some of the gender differences we observe (this is also related to the ‘reference effect’ I mentioned above).  Related to this, we also need to be careful about comparing medians across individual quarters, as the composition of matched cohorts varies from quarter-to-quarter, as shown in Table 1.9 in the release.

With the above caveats in mind – plus the fact that this is aggregate data, albeit medians for specific groups – these patterns point to at least two important considerations when considering the future for these supports as the economy re-opens.

First, and most obviously, not all workers will be affected equally by changes to PUP or wage subsidies.  Taking PUP as an example, changes to the scheme as different sectors reopen could impact younger, female and, to a lesser extent, workers outside Dublin more.

Second, and related to the first point, for some recipient groups median earnings increase during 2020, notably PUP recipients. Therefore, whatever changes are made, and at what speed, could have a bearing on work incentives and labour supply.  I am not saying that workers on PUP do not want to return to work, but rather we should pay close attention to the literature and evidence on how unemployment benefits affect unemployment (see, for example, this recent paper on youth unemployment duration and unemployment benefits by Maynooth economists Aedin Doris, Donal O’Neill and Olive Sweetman). Furthermore, labour supply is a multi-dimensional issue, of which pay is just one. In the current context, other factors like the spread of the virus, migration (and how the pandemic affects cross-border flow of workers), caring responsibilities and health concerns also play a role.

Notes: (*) Average weekly earnings are estimated by summing all earnings during the quarter and dividing by weeks worked in the quarter. (**) Revenue pays the wage subsidy (T/EWSS) directly to qualifying employers who then pay employee wages. So, formally, it is a support for employers. For convenience, I refer to it broadly as an ‘income support’ in this post. (***) The release also provides data on the earnings growth of a third group: employees who receive both PUP and T/EWSS during a quarter (but not necessarily at the same time). This group accounts for around one-in-five recipients in Q4. For a clear comparison between the experience of recipients on the two types of supports, I omit this third group here.

I thank, without implicating, Brian Cahill (CSO) for answering my many queries about the data in the release.

Occupational Pension Funds in Ireland: What do we know?

A guest post by Kenneth Devine (Central Bank of Ireland) on new occupational pension fund data highlighting household exposure, concentrated asset holdings and the impact of COVID-19.  [Disclaimer: This blog represents the author’s views and not those of the Central Bank of Ireland]

Pensions are the primary source of income to households in retirement. The volatility and economic shock associated with COVID-19 have compounded pre-existing issues for pension systems. These include aging populations, the low interest rate environment and the prevailing low yields on safe assets (OECD, 2020).

In a recent Behind the Data publication, Ciarán Nevin, David Mulleady and I ask the question – What do we know about occupational pension funds in Ireland?  Our note highlights the role of occupational pension funds as a household asset, outlines the breakdown of financial assets, and examines the impact of the pandemic on these holdings. An overview of the key findings can be seen in Figure 1 below.

Figure 1: Overview of key findings

While previous work by the OECD (2014) provided a comprehensive review of the Irish pension system, its analysis of occupational pension funds was constrained by a lack of data. New Central Bank of Ireland statistics covering occupational pension funds help to fill this gap by providing a better understanding of the structure and asset holdings of the sector.

We show that, in June 2020, Irish occupational pension funds had assets of €118 billion, accounting for 30 per cent of household financial assets. This is the second largest household financial asset behind currency and deposits. Household sector housing assets accounted for €542 billion in the same period.

According to the Pensions Authority’s 2019 annual report, the Irish sector consists of over 75,000 active occupational pension funds, representing almost half a million active members. This represents over 90 per cent of total euro area pension funds by number. The size, and role, of occupational pensions varies across euro area countries (Curos et al., 2020), with total assets of the pension fund sector amounting to €3 trillion at September 2020.

We have seen a transition away from Defined Benefit (DB) funds in recent years (fall of 50 per cent in number of active schemes since end-2009). For Defined Contribution (DC) pension funds, the member’s income in retirement is dependent on asset performance. Therefore, the switch from DB to DC pension funds has shifted investment risk from the corporate sector to households (Brown, 2016). Households, and their retirement income, are now increasingly exposed to financial market shocks.

The Behind the Data piece outlines that Irish pension funds primarily invest in investment funds shares and unit-linked insurance products. Combined, these two instruments account for three quarters of the sector’s balance sheet. However, structural differences in asset holdings exist across DB/DC pension funds. While the larger DB pension funds are seen to directly invest in hundreds of diverse assets, smaller DC pension funds tend to predominantly hold a limited number of investments.

Figure 2: Impact of COVID-19 on pension fund asset prices

As can be seen in Figure 2, at the onset of the COVID-19 pandemic the total value of pension fund assets fell by 6.5 percent (€7.9 billion). These asset values largely recovered across Q2 and Q3 2020 to sit at €118 billion. The movements were predominantly caused by financial market price gains and losses as the pandemic, and global policy responses, evolved. At Q3 2020, asset values were 1.8 per cent below pre-pandemic levels.

Going forward, the Central Bank will publish Pension Fund Statistics information releases on a quarterly basis. The next steps in developing this dataset will include an investigation into asset breakdowns by their sector and geography, to further explore these household investment exposures.

Researchers interested in hearing more about the data can contact Kenneth Devine.

Household wealth in Ireland: results from the 2018 HFCS survey

Guest post by David Horan (Central Bank of Ireland). Disclaimer: this blog post represents the author’s views and not those of the Central Bank of Ireland or the European System of Central Banks.

The Central Bank has published a Research Technical Paper on household wealth in Ireland: “Household wealth: what is it, who has it, and why it matters” (Horan, Lydon & McIndoe-Calder).

The paper uses data from the latest wave of the Household Finance and Consumption Survey (HFCS 2018) to track changes in the financial position of Irish households between 2013 (the last survey) and 2018. The CSO carries out the survey in Ireland. It is part of a wider cross-country project examining household wealth, income and consumption, coordinated by the ECB.

While carried out prior to the outbreak of COVID-19, the HFCS survey provides insights into issues relevant to the assessment of the economic impact of the pandemic on Irish households. For example, the data highlight the improved financial position and resilience of households prior to the COVID-19 crisis, than was the case leading into 2008. Moreover, these data highlight distributional considerations and differences between households that align with the asymmetrical effects the COVID-19 induced economic shock has had on households, including along age, employment sector and housing tenure status dimensions.

Key developments between 2013 and 2018
When comparing changes between 2013 and 2018, it is important to consider the economic context under which both surveys were conducted. In many respects, 2013 reflects the low point of the recession following the financial crisis; by 2018 the economic recovery was well underway. With this in mind, it may come as little surprise that the financial position of many households in Ireland improved considerably between waves.

We observe that household net wealth grew by over €76,000 for the median household – or by 74 per cent – to €179,200 between 2013 and 2018.  House price growth and declining mortgage debt were the primary drivers of this development.

Net wealth increased across the entire wealth distribution, while inequality, as measured by the gini coefficient, fell between waves. Key to this was the decline in negative equity, which fell from 33 per cent of mortgaged households in 2013 to 4 per cent in 2018. Median gross household income surpassed its previous peak in 2007, reaching €47,700 in 2018. Combining household wealth and income, we find the two are closely linked and that relatively higher income households also tend to be wealthier households — although the relationship is not one-for-one. 

Economic implications
Compared to 2013, households were more resilient in 2018, with debt to asset and debt to income ratios falling significantly between survey waves. These improvements are particularly pronounced for those between 30 and 49 years of age. The debt service burden – the cost of servicing debt repayments to (gross) income – has also fallen since 2013, primarily due to rising incomes. Net liquid assets – the sum of liquid assets less non-collateralised debt – are a commonly used financial buffer metric. The proportion of Irish households with net liquid assets increased to 72.6 per cent in 2018, while the median value of these financial resources increased from €2,000 to €3,000 2018.

To better understand the resilience of indebted households to negative shocks, Table 1 shows the proportion of households by debt-service bucket in 2018, where savings account for at least three mortgage payments. Over two thirds (67.8%) of lower debt service households – that is households with mortgage repayments less than 5% of their gross income – have savings at least three times that of their regular mortgage repayments. Looking at households with the highest debt service ratio (>40% of income), we find that 42% of these households have savings of at least three times that of their regular mortgage repayment.

Turning to household spending patterns, we find that the gross income share that households regularly spend on goods, services and housing varies substantially by income. The average household spends about 80 per cent of their income. Those in the bottom income quantile, on average, report spending more than their income on regular expenses. For the 13% of households that report having expenses greater than their income, typical strategies employed to bridge the gap include using savings, especially for middle income households; getting help from friends and family, especially for lower income households; and using credit cards and overdrafts.

In line with other data sources, homeownership rates have fallen while the share of those renting accommodation has risen. Over 60 per cent of recent owner-occupier home-buyers were under the age of 40 at time of purchase. Almost 30 per cent of recent buyers report receiving an inheritance or gift within three years of their house purchase with a median value of €25,000 (self-reported). The prevalence of inheritance was lower for older borrowers, however the amounts they received tended to be higher.

House price developments play a key role in changes in the net wealth position of Irish households. The ‘collateral channel’ argues that wealthier households have easier credit access. We do find that households are less credit constrained in 2018, which holds true for younger and older households, and for homeowners and renters. Although we cannot rule-out improvements on the supply-side as a driver of this development.

In the mid-2000s, housing equity was used by many households to fund both consumer spending (often on durables) and investment (often in more housing). This peaked in 2006/07, when the value of housing equity withdrawal for the household sector was equivalent to some 10 per cent of income. When house prices fell sharply, this had real effects on spending and investment. In the paper we show that, despite housing wealth in 2018 exceeding previous highs, the household sector as a whole continues to inject as opposed to withdraw equity. In 2018, injections were running at around 10 per cent of income (Figure 1). This reasons for relatively large ‘injections’ include the continued repayment of long-lived, large debts from the early-/mid-2000s, and a far lower level of top-up borrowing relative to the past.

The paper highlights several potential areas of future research using the HFCS data*. For example, we provide useful insights into how households can withstand unexpected income shocks and the financial resilience of households, which are particularly relevant in light of the COVID-19 crisis.

In many respects, we can see that households are better placed going into 2020 than they were leading into the last crisis in 2008. Given the healthier position of many household balance sheets in 2018. Our work indicates that, if house prices and/or incomes falls, we would not expect household debt to drag on spending in the same way it did going in to 2008. Incomes developments are therefore likely to be the primary determinant of consumer spending when the public health threat from COVID-19 recedes. Understanding the distribution of income shocks within the context of household wealth and income position will be important going forward.  

(*) Datasets for research and analysis are available from both the CSO (the HFCS RMF) and the ECB. The ECB dataset also includes cross-country data for most countries.