Friday, April 19, 2019

What do we do with €112 billion of annual savings?


Last week the CSO published the Q4 update of the (non-financial) Institutional Sector Accounts.  These are a great source of information on what is happening in the economy but are terribly difficult to navigate.

Here is a summary of the aggregated current account (Q1 to Q4 2018) by sector. Click to enlarge.


The starting point is the first estimate of nominal GDP for 2018 which is €318.5 billion.  Looking across by sector we can see where this is generated and the clear domination of the non-financial corporate sector in Ireland’s GDP figure (which in turn is dominated by foreign-owned MNCs). The final column gives the flows with the rest of the world. Figures in parenthesis are amounts paid by the relevant sector.

Deducting wages paid and adjusting for taxes paid and subsidies received on products and production gets us to Gross Operating Surplus/Mixed Income.  For the household sector, mixed income is a combination of the earnings of independent traders (the self-employed) and the rent that owner-occupied are imputed to pay themselves (this income is deducted as consumption later down the table so the bottom line is unchanged).  As with GDP, the main generator of GOS in the economy is the NFC sector.

Adding wages received, adjusting for taxes received and subsidies paid on products and production, and accounting for property income paid and received (mainly interest and dividends among others) gets is to Gross National Income.

The move from Gross National Income to Gross Disposable Income is done by adjusting for taxes and transfers.  Most of this are inter-sector flows with payments by one sector being receipts of another.  For example, income taxes paid by households and companies go to the government sector (some minor cross-border flows notwithstanding). 
GDI is a couple of billion lower than GNI because of some cross-border transfer flows.  The rest of the world received about €5 billion more under "Other Current Transfers" from Ireland than Ireland receives from abroad under this heading - €9.5 billion out versus €4.5 billion in. 

Some of this has to do with Ireland’s foreign-aid budget and other transfers.  A large part of it is made up of Ireland’s contribution to the EU budget but it should be noted that earlier in the table the €1.6 billion of "Subsides Paid" from the rest of the word mainly come from the EU and these make up the bulk of the €1.5 billion of "Subsidies Received" by the household sector (agriculture).

Anyway, by this point we have a total economy Gross Disposable Income of €248 billion, of which we use “only" €136 billion on consumption.  That leaves us with Gross Savings of €112 billion and the breakdown by sector can be seen in the bottom row. 

To see what we did with this we turn to the capital account.  Again click to enlarge.


The first panel of the table gives the change in net worth by taking into account capital taxes and transfers and consumption of fixed capital (depreciation on existing assets).

The second panel shows what happened to gross savings and it can be seen that we did €82.8 billion of gross capital formation on produced capital assets and had net purchases of €22.5 billion of non-produced assets (such as marketing assets and customer lists).  That left the economy in a net lending position of €6.6 billion for 2018 (with this €6.6 billion being borrowed by the rest of the world).

The continued deleveraging of the household sector is evident in its net lending position of €5.5 billion.  This will have been, in part, used to repay debt and the household sector has significantly reduced its outstanding debt over the past decade.  The government sector had close to a balanced position in 2018 so, unlike the household sector, did not have a surplus to reduce its debt.

The big figures are again in the NFC sector with €83 billion of Gross Savings fully offset by €62 billion of investment in produced capital assets (gross capital formation) and €22 billion of net acquisitions of non-produced assets giving a net borrowing position of €1 billion.
This relatively modest outcome at an aggregate level probably belies significant changes within the sector.  It is highly unlikely that the companies with the €83 billion of Gross Savings were the companies that invested €84 billion in assets.  The companies with the savings would have used that to reduce their debts (built up when acquiring assets, including intangible assets, in earlier years) while those acquiring assets in 2018 would have funded that by new borrowing of their own.  So while the accounts might show €112 billion of Gross Savings most of it is the result of MNC activities and is not ours to spend.

It is probably a little more than a coincidence that the numbers in the NFC sector were so close in 2018 giving a net outcome of "just" minus €1 billion.  And, it should be noted, that these are just the first estimates.  Things could be very different when the National Income and Expenditure results are published during the summer.  We saw this for the 2015 results though such changes are largely limited to the NFC sector.

The previous compositional issue is also true for the household sector, though on a smaller scale.  While the household sector had net lending of €5.5 billion, repayments against existing debt would have been much larger than this, possibly twice as large.  Those in the household sector who undertook investment (which is mainly on houses) could have funded this with new borrowing.  The balance of repayments on existing debt and new borrowing for investment gives the overall net lending position of €5.5 billion.

This is a Gross Saving that is ours to spend.  Eventually the deleveraging will stop.  Whether that leads to an increase in consumption or investment is hard to tell.  The vulnerable position of the government sector probably means that some caution in the household sector is warranted but whether this caution will persist remains to be seen.

Wednesday, March 6, 2019

Housing Costs in the SILC

Eurostat’s Statistics on Income and Living Conditions (EU-SILC) have lots of household-level data.  Here we will look at outcomes related to housing costs and look for the impact of the ongoing difficulties with housing in Ireland in the data. 

We’ll start with what Eurostat call “total housing costs” as a share of disposable income.

EU15 SILC Housing Costs to Disposable Income 2004-2017

In 2017, Ireland had the lowest “total housing costs” as a share of disposable income in the EU15.  In Ireland, the average share of housing costs was 16.1 per cent of disposable income in 2017 and this has been falling slightly in recent years, i.e. household disposable income is growing faster than housing costs – on average across all households. 

The picture isn’t much different if we look at the median instead of the average with Ireland again being towards the bottom.

EU15 SILC Median of the Housing Cost Burden 2004-2017

In 2017, in Ireland the median housing cost burden was 11.3 per cent of disposable income.  Half of households had a housing cost burden as a share of disposable income that was less than this.

Of course, we would like to know what is included in “total housing costs” used as the numerator in the above charts.  The details are provided here.  For all tenure types it includes the following if they are paid by the occupant:

  • Structural insurance,
  • Mandatory services and charges (sewage removal, refuse removal, etc.),
  • Regular maintenance and repairs,
  • Taxes, and
  • The cost of utilities (water, electricity, gas and heating).

For all tenure types housing costs are considered gross of any housing benefits (i.e. housing benefits should not be deducted from the total housing cost), then for each tenure type it also includes:

OWNERS: Mortgage interest payments, net of any tax relief

TENANTS at market price: Rent payments

TENANTS at reduced price: Rent payments

The most notable thing is that principal repayments on mortgages are not included as part of “total housing cost”.  In part, this reflects the element of choice involved.  If two borrowers take out identical mortgages except one is over 15 years and one is over 25 years it is not appropriate to consider that the household with the shorter term has higher housing costs – they have a higher savings rate.  Of course, the degree to which households have control over the pace of capital reduction on their mortgage may not make it a matter of choice at all.

There will also be differences across countries where borrowers in some countries tend not to make ongoing capital repayments which may be a factor in explaining why countries such as  The Netherlands, Sweden and Denmark are so high in the previous charts.  Capital repayments also increase the equity the household has in the property so aren’t a consumption expense.

Thus, it could be argued that this measure of “total housing costs” doesn’t give a full picture of housing costs but rather is the answer to a question that seeks to find something that is consistent across households and countries and not subject to change due to choices or preferences.  The size and length of mortgage payments will still be picked up through the interest component of “total housing costs”.

It would be useful if “total housing costs” as a share of disposable income was provided by tenure status but it does not seem to be available.  It is, however, provided in Purchasing Power Standard (PPS) units for “owners” and “tenants”.

EU15 SILC Owners Total Housing Costs in PPS 2004-2017

EU15 SILC Tenants Total Housing Costs in PPS 2004-2017

It would be useful if owners were broken down by those with a mortgage or loan and those with none, and if tenants were broken down by those paying the market price and those paying a reduced price.  This is particularly true for Ireland where there are more tenants paying rent at a reduced price then at the market price.

The chart for tenants above shows housing costs rising in recent years. This was around two per cent a year up to 2016 but accelerated to 4.5 per cent in 2017.  This will likely be dampened by the presence of tenants paying a reduced price (such as to local authorities and housing bodies) in the cohort so we could expect the increase for those paying at the market rate to be higher.

Eurostat also provide the share of rent in disposable income for tenants.  Again, it is for all tenants so both those paying a reduced price and the market price will be included.  As a share of the average disposable income of renters, Ireland has the third-lowest average rents in the EU15.  However, Ireland had been the lowest up to 2017 but lost this position due to the increase in rents highlighted above.

EU15 SILC Share of Rent in Disposable Income 2004-2017

Eurostat measure the housing cost overburden rate as being the share of population living in households where total housing costs (as defined above) exceeds 40 per cent of household disposable income.  In 2017, Ireland had the second lowest housing cost overburden rate in the EU15, with only Finland having a (slightly) lower rate.

EU15 SILC Housing Cost Overburden Rate 2004-2017

Ireland might have the next-to-lowest level for all households but as with most of these measures there is plenty going on under the surface.  Here is the 2017 breakdown of the housing cost overburden rate by tenure status.

EU15 SILC Housing Cost Overburden Rate by Tenure Status 2017 Table

Ireland’s overall rate might be 4.5 per cent but for tenants with rent at the market price the housing cost overburden rate is 21.5 per cent putting Ireland sixth-lowest in this category in the EU15.  Thus, just under one-fifth of tenants paying market rates had “total housing costs” in excess of 40 per cent of their disposable income.  As shown in the chart below tenants paying market price has always being the tenure status with the highest housing cost overburden rate in Ireland.

SILC Housing Cost Overburden Rate by Tenure Status in Ireland 2004 to 2017

It may also be surprising how little this is changed over period shown, though there is now a noticeable upward trend since 2013.  It has gone from 16.6 per cent  in 2013 to 21.5 per cent in 2017.

The picture is much the same if we reduce the threshold.  The next chart looks at the share of the population by tenure status where “total housing costs” exceed a lower threshold of 25 per cent of disposable income.  Again, tenants paying the market rate fare worst but here the increase is only visible after 2015 and at 60 per cent in 2017 is the highest in the series.

SILC Housing Cost Burden greater than 25pc of Disposable Income in Ireland 2004 to 2017

And, just for completeness here is a chart of the housing cost overburden rate (40 per cent of income threshold) for tenants renting at the market price across the EU15.  Ireland, has generally had one of the lower housing costs overburden rates for tenants paying market rates since 2004, but, though still near the bottom, Ireland’s relative position has been moving upwards in recent years.

EU15 SILC Tenants Housing Cost Burden more than 40pc of Disposable Income 2004-2017

Next, is the housing cost overburden rate for households that are at-risk-of-poverty, i.e. those with an equivalised disposable income of less than 60 per cent of the median.  Again, this is another instance where Ireland is near best-in-class in the EU15, with only Finland again having a slightly lower rate.

EU15 SILC Housing Cost Overburden Rate AROP Households 2004-2017

Ireland’s position in the above chart indicates that Irish households who are at-risk-of-poverty have to devote a lower share of their disposable income to housing costs than in other countries.  For example, in Denmark around 80 per cent of households who are at-risk-of-poverty have housing costs that are greater than 40 per cent of their disposable income.  And, the more income that has to go on housing costs the less there is for other items of consumption.

To explore this further, Eurostat also calculate an at-risk-of-poverty rate that deducts “total housing costs” from household disposable income.  This uses the standard at-risk-of-poverty threshold, i.e. 60 per cent of the national median equivalised disposable income, but compares it to household income after housing costs have been subtracted from it which can give an indication of what is available to spend on items other than housing.

EU15 SILC AROP after Housing Costs 2004-2017 2

Using this approach Ireland had the fourth-lowest at-risk-of-poverty rate in the EU15 in 2017.  Housing costs include rents at the market price but this has not resulted in any noticeable increase in this measure and, in fact, has declined slightly in the past few years.

As “total housing costs” used to calculate the above shares and rates excludes capital repayments on mortgages it may be that the calculations and ratios do not give a full insight into housing costs.  To this end, participants in the SILC are asked to assess the “financial burden” of their housing costs on the scale of give their view of whether it is:

  • a heavy financial burden,
  • a financial burden, or
  • not a financial burden.

Eurostat’s notes tell us that here a broader approach is to be taken to housing costs when this more subjective view of housing costs is being assessed:

With regard to the calculation of the financial burden of the total housing cost, the following methodological issues should be taken into consideration:

  • The objective is to assess the respondent feeling about the extent to which housing costs are a financial burden to the household.
  • Total mortgage repayment including instalment and interest is to be taken into account for owners and actual rent for renters. In addition, service charges (sewage removal, refuse removal, regular maintenance, repairs and other charges) are to be considered.

Here is the share of people living in households who consider the impact of their housing costs (including capital repayments on mortgages) to be a heavy financial burden.

EU15 SILC Financial Burden of the Total Housing Cost 2004-2017

This is likely closer to what we expect for Ireland. The share of people living in households who assessed that they were experiencing a heavy financial burden due to housing costs increased after 2007 and reached 43.3 per cent by 2013. It has since fallen back and was down to 28.4 per cent in 2017, though still the seventh highest in the EU15 – Ireland’s relative position was fourth highest in 2013.  However, as this is a subjective measure such cross-country comparisons may not be entirely valid.  Still, the individual trend for Ireland is revealing and the 2017 level is close to the levels seen from 2004 to 2007.

Monday, March 4, 2019

Reducing the legacy debts of the credit bubble

The information provided in the annual accounts of the banks allows us to observe developments in the remaining stock of debt from the loans that were issued during the credit bubble.  For example, here are Bank of Ireland’s mortgages (PDH and BTL) by year of origination with the data going back to 2011, the first year such information was provided.

BOI Mortgages Outstanding 2018

Between the end of 2011 and the end of 2018 BOI’s stock of mortgages decreased by 15 per cent – from €27.9 billion to €23.7 billion.  However, this headline figure masks what is happening within the loan book and new loans issued each year replace those which are repaid.  The reduction in loans originating in 2011 or earlier is much greater.

At the end of 2011, BOI had €20.0 billion of mortgages that originated between 2004 and 2008, the peak years of the credit bubble.  By the end of 2018, the amount of mortgages issued in that five-year period had reduced to €11.1 billion.  This is a reduction of €9 billion or 45 per cent over the seven years. 

And that is only since 2011.  If the figures were available for earlier years they would show that well over half of the mortgage debt issued by BOI between 2004 and 2008 no longer exists.

Of course, this may overstate the reduction in debt for individual households as many may have remortgaged because, for example, they moved house.  At the end 2011, BOI had 115,000 mortgage accounts that were issued between 2004 and 2008. By the end of 2018 this number had fallen to 85,500.

We can use the number of accounts to get an average outstanding balance for each year.

BOI Mortgages Average Balance 2018

This shows that from the end of 2011 to the end of 2018, the average balance on the remaining mortgages issued between 2004 and 2008 fell by between 24 and 30 per cent.  This is not a like-for-like comparison each year as the average is calculated using only the number of mortgages which remain on the bank's balance sheet; mortgages which are repaid in full or replaced due to remortgages are not included. 

The figures above, though, probably give a good indication of what is happened to mortgages that are being reduced with regular monthly repayments.  For example, at the end of 2011, the 24,000 mortgages BOI has which it had issued in 2007 had an average balance of just over €200,000.  By the end of 2018, the number of these mortgages had fallen to 19,000 and the average balance of those remaining was €148,000 – a reduction of 26 per cent (and that is since the end of 2011 not the point of origination).

Whichever way we look at it – remaining stock or average balance – it can be seen that the legacy debts of the credit bubble have been significantly reduced.  They will have a long tail but, for BOI mortgages at least, we are probably passed the half life.

Wednesday, February 27, 2019

Do we need another SSIA scheme?

IBEC have joined a number of recent calls for the introduction of an SSIA-type scheme for Irish households.  The various proposals are usually accompanied by claims that such a scheme should be counter-cyclical and can help prevent the economy “overheating”.  Given that the previous scheme ran from May 2001 to April 2007 corresponded to the build up of unsustainable levels of economic activity it may be hard to find support for such claims.  And it may be that such a scheme simply ends up being a transfer to those who can save.

To assess the impact, if any, of the previous SSIA scheme and whether we need one now let’s look at some outcomes for the household sector.  First, the gross savings rate.  This is essentially the share of gross disposable income that is not used for consumption (i.e. current expenditure).

Household Sector Gross Savings Rate 2001-2018

This offers some support for the thesis that such schemes can take money out of the economy.  The gross savings rate rose from around six per cent in 2001/02 to around ten per cent by 2005.  But then the savings fell in 2006/07 when overheating pressures reached their peak.  It is possible that the timing of contributions and maturities of the SSIAs played a role in these outcomes.

It is also worth noting that even without an SSIA-type scheme we have seen a significant rise in the gross savings rate in recent years.  During 2018, the savings rate averaged almost 12 per cent.  This compares to an average of around eight percent for the period 2003 to 2006.  And here is where we stand relative to most of the EU15 (Greece and Luxembourg are excluded):

EU15 Household Sector Gross Saving Rate

Ireland’s current gross savings rate is pretty much in the middle.  It is not clear that it needs to be higher.  And do we need to incentivise something that is already happening?

Of course, the current account only gives a partial impact of a sectors impact on the economy – we need to look at the capital account as well.  The bottom line for a sector is the amount of net lending/borrowing it does.  This is the final outcome after all earnings, taxes, transfers and spending have been accounted for by adding capital flows to the current flows that are used to get the savings rate.  And it is on the bottom line that we see where the action was:

Household Sector Net Lending-Borrowing 2001-2018

The 2001 to 2007 SSIA scheme might have been taking some money out in the current account but once we add in what was happening in the capital account we can see that the household sector was a net borrower and that this increased during the period.  It is very likely that some of the income that was considered “saving” in the current account only arose because of the significant borrowing shown above in the capital account. 

In 2001, for every €100 of disposable income that the household sector had it, had total spending (current plus capital) of around €110. By 2006, household spending was more than €120 for every €100 of income (with some of this fueled by maturing SSIAs).  The borrowing came to a shuddering halt in 2008 and for the past decade total spending has been around €95 for every €100 of income.  The Irish household sector has been a net lender and, in aggregate terms, this has averaged around €5 billion a year for the past decade or so.

This impact of this borrowing and lending can clearly be seen in the household sector balance sheet:

Household Sector Loans and Deposits 2002-2018 CB Data

There was a huge run-up in household sector loan liabilities up to 2008, reaching a peak of €204 billion.  Since then, the net lending has led to deleveraging with a loan liabilities reducing to €138 billion by the third quarter of 2018.  At the same time, household deposits have been increasing and during 2018 actually exceeded household loans for the first time since 2002. 

It has been a remarkable improvement in the aggregate balance sheet of the Irish household sector.  That this deleveraging, which ought to have been a drag on growth, coincided with a resurgent economy is even more remarkable.

And here is Ireland’s household sector net lending/borrowing relative to the rest of the EU15 (with only Luxembourg excluded in this instance).

EU15 Household Net Lending-Borrowing

As of 2018, the Irish household sector has the third highest net lending rate in the EU15.  Maybe we could look for a change in the composition of that lending.  This could happen as the debts of the credit bubble are paid off and households may switch to something like increased pension saving.

But, it is doubtful that it needs to be higher.  It might be that, given the risks we face, that we need other sectors to be doing a bit of saving (the government sector maybe); after a decade of deleveraging the household sector can probably afford to cut loose a bit. 

Tuesday, January 15, 2019

Why the “working poor” makes for an inappropriate policy target

The publication by the CSO of the results from the Survey on Income and Living Conditions (SILC) always generates plenty of reaction.  One focus following the release of the 2017 results has been the “working poor” such as the headline of this piece.

The CSO provide income and poverty rates by principle economic status and the 2017 outcomes are summarised here:

CSO SILC Income and Poverty by PES 2017

As can be seen for the “at work” category the at-risk-of-poverty rate (equivalised disposable income below 60 per cent of the national median) is given as 5.4 per cent.  When we add in measures of deprivation, we find that 1.4 per cent of those with a principle economic status of “at work” live in households deemed to be in consistent poverty.  Given how low these levels are relative to other categories, the “at work” group seem a peculiar group to focus on.

Here we provide five reasons why targeting the at-work at-risk-of-poverty rate may be inappropriate.

  1. Ireland already has the second-lowest AROP rate for employees in the EU15.
  2. The measure is as much a function of household type, especially the presence of children, as it is labour market outcomes.
  3. When it comes to labour market outcomes the most important factor is the amount of work with low AROP rates for households with high or very-high levels of work intensity.
  4. The link between low pay at the level of the individual and low income at the level of the household is weak.
  5. One-third of the “working poor” are self-employed who are excluded from most policy proposals.

1 Comparison across the EU15

Eurostat provide figures that allow us to compare the at-risk-of-poverty rates across the EU for people who are employed and one feature of this is how well Ireland does.  Here are the AROP rate for employees since 2009:

EU15 SILC Employees AROP 2009-2017

The 2017 figure for Ireland has not been provided to Eurostat yet but it seems likely that Ireland will have close to the second-lowest in-work at-risk-of-poverty rate for employees in the EU15.

2 The role of household type

The determination of whether of being at-risk-of-poverty is based on equivalised household income rather than than earnings of the employee on their own.  For example, you could have two employees both earning €30,000 – one could be deemed to be at-risk-of-poverty and one may not. How can that be if there are both earning the same amount? Type of households or, more particularly, children. 

If there are more people in the household then the available income has to be spread over more people thus reducing the equivalised, or per (weighted) person, income of the household. 

Here are the figures from Eurostat for the at-risk-of-poverty rates for people at work but living in two different types of households in Ireland:

  • Household with two or more adults with dependent children
  • Households with two or more adults without dependent children

SILC Eurostat In Work At-Risk-Of-Poverty Rate 2004-2017

It can be seen that, bar the peak crisis years of 2009 to 2011, the at-risk-of-poverty rates of workers in the households with children is about twice that of households without children.  It is not the labour market that drives in-work at-risk-of-poverty rates; it is household type.

3 The amount of work

This amount of work can be measured by household work-intensity: the amount of available time that someone is working.  If the working-age adults in a household have a high or very-high work intensity there is close to no chance of that household being at-risk-of poverty.

SILC Eurostat Work Intensity At-Risk-Of-Poverty Rate 2004-2017

In-work, at-risk-of poverty rates are highest for those households with low work intensity.  These are households where members of working age worked between 20 per cent and 45 per cent of their total potential during the previous 12 months.  Households composed only of children, of students aged less then 25 and/or people aged 60 or more are completely excluded from the work-intensity indicator calculation.

Again, relative to the rest of the EU15, Irish households with low work intensity have at-risk-of-poverty rates well below those of other countries.

EU15 SILC AROP Low Work Intensity Households 2003-2017

Over 60 per cent of Irish households who are classed as in-work and at-risk-of-poverty have either low or medium levels of work intensity.  It is not earnings that drives the in-work, at-risk-of-poverty rates; it is the amount of work.

Something, such as a refundable tax credit may have very little impact on at-risk-of-poverty households with children.  The at-risk-of-poverty threshold for a 2 adult plus 2 children household in 2017 was €29,000.  Even allowing for Child Benefit such a household close to that threshold which gets its income from work will have used almost all the available tax credits.  Making them refundable will make little difference to them.

The majority of households who are deemed be in-work and at-risk-of-poverty have low or medium work intensity.  Refundable tax credits in this instance would be a reflection of a low amount of work rather than low earnings.

4 Low pay and household income

The link between low pay and at-risk-of-poverty rates is weak.  Ireland has workers who are low paid but they are not in low-income households.

Low Pay and the Distribution of Income

The chart would suggest that something around six per cent of low-pay employees (below a threshold of €12.20 a hour in the analysis shown) are in households who are at-risk-of-poverty. Or, in other words, 94 per cent of low-pay employees are in households who are not at-risk-of-poverty.  Indeed, over half of low-pay employees are in households in the top half of the income distribution.  There are almost as many low-pay employees in households in the top decile as there are in the bottom decile.

Policies, such as refundable tax credits, that target the low paid seem likely to make overall inequality and at-risk-of-poverty rates worse as very little of the benefit would accrue to those at the bottom of the income distribution.  It is likely that part-time second earners would appreciate it but in most cases these already come from middle- to high-income households.

It is also not clear how a refundable tax credit would work in the case of the self-employed. It was noted at the press briefing for the SILC publication that around one-third of those deemed to be in-work and at-risk-of-poverty are self-employed. Refundable personal and PAYE tax credits would mean that one-third of the target group is excluded. And a large share of the resources used would go to people outside the target group – the low-paid in high-income households.

Saying that the “working poor” should not be a policy target doesn’t mean we should have policies that try to increase incomes. We should. But using the in-work at-risk-of-poverty rate as a benchmark for either the justification of certain policies or in judging the success of polices may not be appropriate.

We conclude with a comparison we have made before:

Ireland Sweden AROP by Work Intensity

For all levels of household work intensity Ireland has at-risk-of-poverty rates that are lower than Sweden’s, and significantly so in some cases, e.g. medium and low. Yet, the overall at-risk-of-poverty rates of the countries are very similar.

Printfriendly