Friday, April 5, 2024

On the Price and Quantity of Housing

Housing problems are complex but usually can be distilled back to a shortage of housing relative to demand. For those looking, it can be difficult to find somewhere to live, particularly for renters.  More housing makes it easier to find somewhere to live, though sometimes this is in dispute. 

Earlier in the week we had a opinion piece that focused on prospective homeowners with a broad focus on price and a limited view on supply.

On Price

The piece sets out findings of research undertaken by the Bank of England:

For every 1 per cent increase in lending rates, the researchers of Threadneedle Street found a commensurate 3 per cent decrease in house prices, and vice versa.


Another significant determiner of house prices is population growth. For every 1 per cent increase in the natural growth of households since 1990, the UK researchers found a 2 per cent increase in house prices. The same 1-to-2 per cent ratio also applies to immigration – people coming to work here.


And the third factor: wages. Each 1 per cent rise in incomes results in a 2 per cent rise in house prices.


Finally, housing supply itself. A 1 per cent increase in stock (some 20,000 houses) should theoretically reduce prices by 2 per cent.

The precise source isn’t provided but we will just taken them as given in the piece.  We will use the most recent intercensal period, 2016 to 2022, to do a crude approximation of the impact of these estimates if applied to Ireland.

Central Bank data shows that the interest rates on new business mortgages averaged 3.50 per cent in 2016. For 2022, this has declined to 2.68 per cent.  This gives a decline of 0.82 percentage points (which seems more relevant than the 30 per cent reduction)

The population data from the Census shows an 8.1 per cent increase over the same period.  The annual data from the Employment Costs Survey shows the yearly earnings of a full-time employee rising from €45,640 in 2016 to €54,189 in 2022, an increase of 18.7 per cent.

And again using the Census, the housing stock increased from 2,003,645 in 2016 to 2,112,121 in 2022, an increase of 5.4 per cent.

Applying the Bank of England’s estimates gives the following impacts on prices:

  • Mortgage Rates: –0.8 x 3 = +2.4
  • Population: +8.1 x 2 = +16.2
  • Annual Earnings: +18.7 x 2 = +37.4
  • Housing Stock: +5.4 x –2 = –10.8

It is all very crude but the sum of those impacts is +50.6 per cent.  The CSO’s residential property price index went from 107.5 in 2016 to 164.0 in 2022, a rise of 52.6 per cent.  I don’t know, it’s like this supply and demand stuff works, or something.

However, the negative impact on prices of the increase in the housing stock is dismissed in the piece because prices haven’t fallen as the housing stock has increased and “are at their highest level since the Celtic Tiger.” 

But the price effect of the increase in the housing stock is there! And, as done here, can be shown using the estimates provided in the piece.  Using those shows that house prices in 2022 would have been 10-11 per cent higher if there had been no increase in the housing stock since 2016.

The increase in the housing stock reduced house prices relative to what they would have been if that increase hadn’t occurred.  This is how supply and demand models work: relative prices, ceteris paribus etc.

The piece is right that interest rates have been a key driver of house prices in recent decades.  We examined that here which showed that even though price to income ratios have increased in Ireland, indicative initial mortgage payment to income ratios are actually lower than they were in the 1970s and 1980s.  This is the final chart from that post:

As was pointed out in the previous post it is with the deposit required that the most significant increases have been seen.  Higher nominal house prices (due to lower interest rates) have seen the amount required for a 10 per cent deposit rise from around a quarter of the average annual household income in the 1970s and 1980s to almost a half of household income now.

Ireland does not stand out as having unusually high mortgage payment burden. Here is OECD data on the median mortgage burden as a share of disposable income for owners with mortgages (with data for all bar five OECD member countries).

For newer and possibly larger mortgage burdens we can look at the housing cost overburden rate.  This is the share of households where housing costs consume more than 40 per cent of disposable income.  In this instance OECD data for both owners with mortgages and private tenants are provided.

Ireland is hard to locate in Panel A because Ireland has one of the lowest housing cost overburden rates for owners with mortgages in the OECD.  In Panel B it can be seen that Ireland has one of the highest housing cost overburden rates for private tenants in the OECD.  Whose plight is in need of easing?

On Quantity

The piece takes a very limited view of quantity: it’s new houses for first-time buyers or nothing, and is best represented by this paragraph.

In Dublin city last year, 94 per cent of all new housing was apartments, 98 per cent of which were for rent. First-time buyers there bought just 75 new houses. In Cork city just 3.5 per cent of all new housing was sold with first-time buyers buying 17 new houses. In 2017, over 80 per cent of all new scheme houses (what the CSO calls housing estates) was sold on the market, and last year that was 52 per cent. Individual buyers have been sidelined and forgotten by successive governments.

The figures for the Dublin City local authority area are correct but they are limited.  Per the CSO, there were stamp duty filings for 1,975 first-time buyer transactions in Dublin City Council’s (DCC) area in 2023. That’s a long way from 75.

The difference, of course, is explained by the most frequent type of property that is transacted: existing ones. Dublin City has a housing stock of around 250,000 units.  That most of the supply for first-time buyers comes from this should not be a surprise.

We can get insight into the composition of the housing stock in DCC’s jurisdiction from the Census.

The area has 150,000 houses – 60 per cent of the stock.  If 50,000 apartments are added, houses would still make up 50 per cent of the stock.  3,200 apartments were completed in the DCC area last year.

We can start to see if “individual buyers have been sidelined and forgotten” by looking at the national tenure status figures from the Census.

From Census 2016 to Census there was an increase of almost 140,000 in the number of households in Ireland.  Of this increase, 63,400 were owner-occupier households.

Somewhat unusually the next largest increase was for the “not stated” category.  Then comes the increase for tenants of local authorities and approved housing bodies which increased by just over 23,000.  The fourth largest increase was for households renting from a private household which made up 15 per cent of the total increase in households over the period.

The supplementary data to the CSO’s residential property price index gives details of the number of transactions. Between Census 2016 and Census 2022, stamp duty filings were made for 80,000 FTB transactions.  The figures show that there were 17,500 FTB transactions in 2023.

The breakdown shows that there were 5,000 FTB purchases of new properties and 12,500 FTB purchases of existing properties.  For the whole of Dublin, i.e. County Dublin, there were 5,000 FTB transactions last year.

On top of purchases, there will also be first-time owners due to single or one-off houses that are built for the occupier and will not have a stamp duty transaction to record.  5,500 single houses were completed in 2023.

On Homeownership Rates

Homeownership rates are used to highlight the plight of prospective buyers.

As the proportion of new housing for sale plummets, despite increasing overall supply, our home-ownership rates follow suit, and are now below the European average at just 66 per cent. Thirty years ago, 81 per cent of households owned their own home. That is a staggering drop in a short space of time, but the Government is remarkably silent on (or oblivious to) the issue.

Ireland’s homeownership rate has declined since 1991 but the Ireland of 2022 is very different to the Ireland of 1991.

In the 1991 Census, 98.4 per cent of the population of 3.5 million was born on the islands of Ireland and Great Britain.  The equivalent figures for the 2022 Census were 85.4 per cent of a population of 5.1 million.  The number of people living here who were not born in Ireland or Britain went from 55,000 in 1991 to 730,000 in 2022.

It is true that the overall homeownership rate declined from 80 per cent in 1991 to 69 per cent in 2022 (when households with a tenure status of ‘not stated’ are excluded).  However, what the various census results show is that this decline occurred from 20o2 to 2011, during which the rate declined from 80 per cent to 71 per cent.  The homeownership rate for All Households has been relatively stable since 2011.

The above chart also shows some outturns for homeownership by nationality – specifically the homeownership rate for all households and for those  headed by an Irish or UK national. 

As discussed, this was essentially the entire population in 1991, when the homeownership rate was 80 per cent.  The first time a breakdown of homeownership by nationality was provided was with the 2006 Census. 

Thus far, no results for homeownership by nationality have been published by CSO for Census 2022.  Going back to Census 2016, the homeownership rate for households headed by an Irish or UK national was 75 per cent.

Going from 80 per cent in 1991 to 75 per cent in 2016 is a decline, but not a staggering one.  The significant change in Ireland has been in the proportion of the population who were born somewhere else. 

In line with the experience elsewhere, immigrants have lower homeownership rates. Of the 150,000 households in Census 2016 not headed by an Irish or UK national, just 22,800 were owner-occupiers, giving a home-ownership rate for this cohort of 16 per cent.  Nearly three-quarters of such households are private tenants.

Finally, on homeownership rates, it should be noted that they are based on households that actually exist.  If there is not enough housing people may stay in owner-occupied households longer than they want to.  We previously noted that in the recent Census, Ireland was short around 70,000 households headed by under 35s.

Between 2011 and 2022 the number of people aged between 20 and 34 declined by 11 per cent.  However, the number of households headed by an under 35 declined by 30 per cent.  Many of these people are living in owner-occupier households but they are not the owners; they are living with their parents.

There has not been enough analysis of the impact of current policies. Some 68 per cent of Irish 25-29 year-olds live with their parents; in Finland this is 5.7 per cent.  We should be discussing that, or how a policy obsession with urban hyper-density housing and homes unsuitable for families has created unsustainable commuter sprawl under a Green Minister for Transport.

There is no doubt that there are more people living with their parents for longer, but a figure of 68 per cent is outlandish.  For a start, around a quarter of the population aged 25 to 29 are non-Irish citizens (69,000 out of 296,000 in the latest Census). It seems unlikely that a large share of those are living with their parents.

The figure comes from the SILC, and the CSO have set out the differences in figures for young adults living at home in this useful note.  In the 2022 Census, 33 per cent of young adults aged 25 to 29 were enumerated at their parent’s house on Census night.  This is still a high figure.  It is much higher than it was in 2011 (24 per cent) and far, far higher than it is in Finland.

If the share of 18 to 35 year olds living with their parents had stayed at the level it was in 2011, then almost 100,000 more young adults would have left the family home.  We might ask how does Finland do it.

They are not doing it through homeownership. Recent figures from Statistics Finland show that “eighty-one per cent of household-dwelling units of persons aged under 30 lived in rented dwellings.”  A household-dwelling unit is formed by all persons living permanently at the same address.  Separate figures show that they also tend to live alone.

Here are the population distributions for Ireland and Finland by type of household.

Over 20 per cent of Ireland’s population lives in households of three or more adults. These are the households of young adults living with their parents or young renters sharing a semi-d. In Finland less than four per cent of the population live in households with three or more adults. 

On the other hand, in Finland, nearly 16 per cent of the population are people younger than 65 living alone.  In Ireland, this group is six per cent of the population. 

Ireland has lots of dwellings that are suitable for families.  Over 85 per cent of our housing stock are houses, with 1.8 million houses in a total stock of 2.1 million dwellings.  What we don’t have, if we are emulate Finland,  are sufficient dwellings for households of one or two people, especially renters.

Tuesday, March 26, 2024

The Recent Decline in the Inequality of Market Income in Ireland

One feature  of how the inequality of disposable income is determined in Ireland has been the recent decline the in the inequality of market income. Market income is typically defined as including wages, rents, dividends, interest and typically certain pension income such as from private and occupational pensions.

Here is the OECD’s estimate of the gini coefficient for market income in Ireland since 2012 (chart crime incoming):

In the OECD’s dataset, the figure for the latest year, 2021, corresponds to the household survey undertaken for SILC2022.  The reference year for income in the SILC is now the previous calendar year and this is the year the OECD use in their database.  The 2022 outturn for Ireland will be based on the survey done for SILC2023. It should also be noted that in SILC2023 the CSO revised their SILC results for 2020 to 2022 (due to the population findings of the Census) though the impact of the revision on the main measures of income inequality was modest.

The recent reduction in Ireland of the inequality of market means Ireland no longer has the highest gini coefficient for market income in the OECD.  The latest estimates show that France, Italy, Finland and the U.S. among others have higher gini coefficients for market income, with Ireland eighth highest.

It should also be noted that although the inequality of market income has been declining in Ireland for the past decade it has been lower previously – with lower gini coefficients estimated for the late 1990s/early 2000s.

The chart below has the recent outturns for Ireland and a selected set of OECD countries, showing Ireland losing its outlier status (and is also a chart having the vertical axis more appropriately start at zero).

We can start to get some insight into the nature of the fall in the inequality of market income in Ireland by looking at income shares.  These can be determined using the information in the ESRI’s PILSReP Database

Here are the quintile shares in the SILCs for selected years since 2013:

It can be seen that the market income share of the top 20 percent declined from 57 percent in 2013 to 50 percent in 2021.  The offsetting share increase can be seen for the bottom three quintiles. The share of market income going to the bottom 60 percent increased from 16.5 percent in 2013 to 23.4 percent in 2019.

We can get more granular detail on this change by looking at the Lorenz curves – a plot of cumulative shares by the cumulative population.  Here are the Lorenz curves for the distribution of market income in the 2013 and 2021 SILCs (again using figures from the ESRI’s PILSReP spreadsheet):

The most significant change happened at the bottom where there was a large reduction in the number of zeroes (i.e. non-recipients of market income).  In the 2013 distribution of market income, a cumulative share of one percent was not reached until the 33rd percentile.  By 2021, this was reached by the 24th percentile (with the 33rd percentile having a cumulative share of 4.1 percent).

The 2013 Lorenz curve remained at zero until the 22nd percentile; the 2021 distribution was zero until the 13th percentile.  The recovery and expansion of employment significantly reduced the number of zeroes in the distribution of market income.

The associated reduction in unemployment also means that the impact of transfers on income inequality has also fallen.  We can see this if we return to OECD data and look at their gini coefficients for market income, gross income (market income plus transfers) and disposable income (gross income minus taxes and social contributions).

For 2012, the OECD put the impact of transfers on Ireland’s gini coefficient at 0.20 points, reducing the gini coefficient from 0.58 for market income to 0.38 for gross income.  The impact of taxes further reduced the gini coefficient by 0.07 points, giving the gini coefficient of 0.31 for disposable income.

For 2021, the impact of transfers reduced to 0.15 points (due to there being fewer recipients of unemployment related transfers).  Transfers reduced the gini coefficient for market income from 0.51 to 0.36 for gross income.  The impact of taxes was largely unchanged again reducing the gini coefficient by 0.07 points, giving the gini coefficient for disposable income of 0.29.

Thus, while there was a reduction of .07 in the gini coefficient for market income, this translated into a reduction of “only” 0.02 in the gini coefficient for disposable income, due to the reduced impact of transfers. 

We can use the OECD to examine the impact of transfers and taxes on income inequality in all OECD member states in 2021 (or the latest available year in the dataset).

For 2021, the combined impact in Ireland of transfers and taxes on income inequality was the fourth highest among OECD countries.  For transfers, Ireland was eighth highest, with transfers in Finland, France, Greece, Belgium, Austria, Poland, Italy and Czechia all having a bigger impact on inequality. 

For taxes and social contributions, Ireland continues to stand out, with the 0.07 point reduction in Ireland having the largest impact on income inequality in the OECD.

Eurostat also provide measures of market income inequality.  One of these is an S80/S20 quintile share ratio for gross market income.  The outturns of this quintile share ratio for Ireland have shown a remarkable reduction in recent years; falling from almost 100 in 2013 to around 15 now.

The huge swings in this measure, at least for Ireland, indicate that it is sensitive to small changes at the bottom of the distribution.  It was the relatively high number of zeroes in the distribution of market income in Ireland that gave rise to the exceptionally high outturns up to 2013 for this measure.

One can imagine the top 20 per cent/bottom 20 percent ratio being something like 50/0.5 in 2013 (giving an answer of c.100) and this moving to something like 50/3.5 in recent years (giving an answer of c.15).

This is a relatively small change in the share of the bottom 20 percent (from 0.5 percent to 3.5 percent) but it has a marked impact on the outturn for the quintile share ratio.  And even with this reduction, Ireland remains the highest in the EU for this measure.

We could calculate a similar quintile share ratio using the shares of market income from the ESRI estimates in the table provided earlier.  But the answer would be very very different, i.e. over 100 for recent years (50.3/0.4 = 127, for 2022), rather than the 15 shown in the Eurostat results for 2022.

The reason for the difference is due to the income definitions used.  The CSO (and the OECD) only include private and occupational pensions in their measure of market income.  For their measure of gross market income, Eurostat include all old-age transfer payments.  This includes state pension benefits such as means-tested and non-contributory pension benefits and all survivors pension benefits.  This is based on the concept of pension income being a part of lifetime earnings.  Outside of private and occupation pensions, the CSO (and the OECD) count pension income as a social transfer, i.e., not part of market income.

Ireland’s gross market income quintile share ratio is high because of the relatively high number of non-recipients of market income and the relatively low number of pensioners.  It has been found that excluding zeroes, Ireland “lies around mid-table in terms of market income inequality” in the EU. 

It is also the case the estimates of market income inequality have tracked the estimates of quasi-jobless households (households with very-low work intensity where less than one-fifth of the available time is worked).  For a considerable period, Ireland had the highest share in the EU for population aged under 60 living in households with very-low work intensity. This has not been the case since 2018.  The lower level of quasi-jobless households is linked to the lower level of market income inequality – both are estimated using the SILC.

Eurostat also provide a number of gini coefficients.  All of these are after income taxes and social contributions have been deducted but then vary in the amount of social transfers that are included.  Eurostat provide three gini-coefficients:

  • Disposable income before all social transfers
  • Disposable income including pension transfers
  • Disposable income including all transfers

The final one is just the standard disposable income while the first is akin to market income after taxes.  Here are the estimated gini coefficients across the EU15 for disposable income before all social transfers: 

By this measure of market income inequality, Ireland has had, in recent years, one of the lowest levels in the EU15.  This is after taxes and social contributions have been deducted and again we note that in Ireland these have the biggest impact on the gini coefficient across both the OECD and EU.

Taxes alone are unlikely to account for Ireland’s relative position here, which is at odds to what we have seen for other measures of market income inequality, such as the quintile share ratio.  The difference in results suggests the definitions used are important. The quintile share ratio is determined using gross market income

As discussed above this includes all pension income.  The previous chart excludes all pension income.  This would suggest that pension income has a significant impact on Ireland’s relative position using the quintile share ratio for gross market income. 

We can see this if we look at the gini coefficient after adding in pension income.

So, after taxes and before all transfers, Ireland has a relatively low gini coefficient across the EU15.  Once pensions are added in, Ireland has one of the highest gini coefficients in the EU15 (with the higher UK estimate in the above chart going back to 2018 when the UK last provided data to Eurostat).  Eurostat quintile share ratio for gross market income includes all pension income which may go some way to explaining Ireland’s relatively high outturn for that measure (due to Ireland having a lower old-age dependency ratio).

This highlights that the impact of pensions is key to the transition from the distribution of market income across the population to that of disposable income.  We don’t get much insight into this from the OECD’s income inequality estimates. 

However, we can examine it if we look at the three gini coefficients provided by Eurostat at the same time.  Here they are for Ireland since 2012:

In line with earlier results we see a decline in the gini coefficients from Eurostat for those income measures most closer aligned with market income.  The gini coefficient for disposable income before all transfers declined from 53.5 in 2012 (Eurostat uses the 0-100 scale) to 47.0 in 2022.  The gini coefficient for disposable income including pension transfers declined from 46.1 in 2012 to 38.6 in 2022. 

The difference between those gini coefficients reflects the impact of pension transfers on income inequality.  We can see that this has increased over the past decade, from an impact of 7.4 in 2012 t0 one of 8.5 in 2022.  This is likely explained by the increasing share of pensioners in the Irish population.

The final gini coefficient is that for disposable income.  This has also declined, albeit more modestly, from 30.4 in 2012 to 27.9 in 2022.  The different to this gini gives the impact on income inequality of all transfers other than pensions.  We can see that this has declined from 15.7 in 2012 to 10.6 in 2022.  Again, this is likely driven by the reduction in unemployment and the reduced need for unemployment-related income supports.

We can compare the impact of pensions and other transfers across the EU in 2022.

In line with the OECD estimates, the estimates from Eurostat show that Ireland does not stand out for the impact of transfers on income inequality.  In 2022, the impact of transfers on income inequality in Ireland was the ninth highest in the EU27.

However, what the Eurostat estimates allow us to see is the composition of that impact which can be broken down into that due to pensions and that due to transfers other than pensions. 

And using this we can see that the impact of pensions on income inequality in Ireland is the lowest in the EU27, due likely to the low share of pensioners in Ireland’s population and also possibly the flat-rated benefits under the State pension schemes.  And, on the other the impact on income inequality of transfers other than pensions in Ireland is the highest in the EU27.  This will mainly reflect the impact of working-age transfer payments.

And, all of this is just how we get to the bottom line: disposable income.  This is the income people have available to spend.  The rest are somewhat notional figures.  So, to conclude here are the gini coefficients for disposable income across the EU15.

In terms of the EU15, Ireland has moved from upper-middle to lower-middle for the inequality of disposable income.

Tuesday, January 9, 2024

Ireland’s growing population of young adults

Using the results of Census 2022, the CSO estimated that Ireland’s population in April 2023 of young adults aged 25 to 34 was 645,000.  If we go back five years to 2018, using Census 2016 as the benchmark, the CSO estimate that the population then of young adults aged 20 to 29 (the same cohort) was 575,000.

Between 2018 and 2023 the population of this cohort grew by 60,000, or a little over 10 per cent.

Using the CSO’s estimates of the population by single year of age we can get the population of this group back to 1998 (when the youngest, now 25, would have been born).  After 1998, migration is the main driver of changes in the population of this cohort.

Population of Cohort Aged 25 to 34 in 2023 1998-2023

We can see that from 1998, the population of the cohort increased, likely as the parents of children moved to Ireland with them.  There was then a fall after the 2008 crash.  For the last ten years the size of the group has been increasing but this time, because the group is older, it is likely due to the autonomous decisions of young adults in their twenties.

We can see this pattern if we look at the estimated annual change in the population of this cohort.

Population of Cohort Aged 25 to 34 in 2023 Annual Change 1998-2023

In the years immediately preceding the pandemic, the size of this group was growing by an average of 10,000 per year.  Since then there has been volatility in the series due to COVID and, more recently, the war in Ukraine.

Of course, these changes are the net outcome of inward and outward migration.  We don’t have published estimates of migration by single year of age but the CSO do provide migration flows by age group.  Here is what they show for people aged 15 to 24 and for those aged 25 to 44.

Migration Flows of Population Aged 15-24 1987-2023

Migration Flows of Population Aged 25-44 1987-2023

In both cases, we that the estimated net flows in the past ten years have generally been positive.  Over the period since 1987 there is, though, a noticeable change in the age profile of migration.

In the late 80s and early 90s, migration, which was mainly emigration, was larger in the younger age category.  In 1989, negative net migration of 15 to 24 year olds was more than twice as large as that of 25 to 44 year olds.  By the turn of the millennium there was very little emigration of 25 to 44 years olds, while immigration of this group had become much stronger.

In the last 15 years, migration flows – both directions –  in the 25 to 44 age group have been larger than those in the 15 to 24 age group.  In the five years, pre-Covid, net migration of 25 to 44 year olds averaged +20,000 a year.  The impact of Covid and the war in Ukraine make identifying the underlying trends for the last couple of years difficult.

To conclude here are two additional snapshots of Irish migration flows:

  1. Migration flows of Irish citizens (available from 2006)
  2. Migration flows with Australia (available from 2008)

Migration Flows of Irish Citizens 2006-2023

Migration Flows with Australia 2008-2023

The estimated net migration of Irish citizens was positive from 2017 to 2021 and was negative in each of the last two years: –2.200 in 2022 and –900 in 2023.  And of the past eight years, Ireland has had one year (2022, -800) of negative net migration with Australia.

Finally, we use the Census results to assess how Ireland’s population of young adults has changed in recent years.  From Census 2016 we have the population by nationality for the 20 to 29 age group.  We roll that forward six years for Census 2022 and get the population by citizenship for the 26 to 35 age group.

Population of Young Adults by Citizenship 2016 and 2022

There was a 76,500 increase in this cohort in the inter-censal period.  Census 2016 recorded 554,000 people aged 20 to 29, and six years later, in Census 2022, there was 630,500 people aged 26 to 35

By country, the most significant change is for India, which shows an increase of 19,000 over the period.  No other country had a five-figure increase. The next largest increases were Brazil, Romania and Italy.  There were modest declines for Lithuania and the UK, with Poland showing the largest decline.

Over the period we can see that the number for Ireland increased slightly over the period – unsurprising as these are the benchmark for the estimated migration flows shown above.

Tuesday, January 2, 2024

Just how expensive has housing become?

Nominal House Prices

Back in October, the CSO produced a publication to mark 50 years of Ireland and the EU.  Included in this was a series of annual average house prices from 1970 to 2019.  The values in the table are charted below and they show the average annual house price rising from €6,700 in 1970 (when converted to euro) to €295,700 in 2019.

The chart takes all the values as published by the CSO with the additional values for 2020 to 2022 taken from the CSO’s databank. The series closely follows other nominal house price series for Ireland (such as Fred’s) except for 2010 to 2015, with the chart above understating the post-2008 fall in house prices (which continued to 2012) shown by other indices – including the CSO’s own Residential Property Price Index

The average for 2023 is likely to be around €370,000 and this estimated value is also shown above.  Here are the CSO’s notes on the original table:

Note that 1970 – 2009 data is based on mortgage data whereas the 2010 – 2019 CSO data is based on Stamp Duty returns from Revenue. Figures for 1970 to 1977 are new prices only while all others are for new and second-hand properties in Euro or Euro equivalent.

The methodological change in 2010 and the low level of transactions, impacting the composition, likely explain the larger nominal fall of around 60 per cent) shown elsewhere.  We will proceed with the values as given in the table.  Our primary interest is a comparison of current prices to what they were in the 1970s, 80s and 90s rather than to what they were in the depths of the post-2008 recession.

In nominal terms, house prices are far higher than they were in the 1970s, 80s and 90s.  In simple nominal terms, at €370,000 houses prices now are almost ten times higher than the average price from 1970 to 1995 (€38,700).

Indeed, the latest estimates are a record high for the series exceeding the previous peak of €350,000 from 2007.  The chart also shows that in the last 50 years, Ireland has had just one period of nominal house price declines: 2008 to 2012.

But these numbers are of their time. Just how does €6,700 from 1970 compare to €370,000 in 2023?  The price of everything has changed in the interim. Just how can we tell which year had more expensive housing?

Inflation-Adjusted House Prices

We can try to put the prices in terms of other products. The CSO collect national average prices for a range of items. Since 1983, they have been publishing a national average prices for 567ml of draught stout in a licensed premises.  Using this, and some additional sources to extend the series back to 1970, we can put the average house price in any given year in terms of the number of pints that could be bought that year for the same sum.

In 1970, the average price of a pint was 21c and with an average house price of €6,700 the equivalent price of a house was 31,900 pints.  For 2023, the average price of a pint is put at €5.48 and our estimated average house price of €370,000 puts 2023 house prices as equivalent to 67,500 pints of stout.  Pint glasses mightn’t make good lenses but using them points to average house prices being about twice as high than they were 25 to 50 years years ago and not the ten times higher as the unadjusted nominal house prices suggest.

Using pints, we can see that houses were at their most expensive in 2006/7 – equivalent to over 90,000 pints.  And the pints equivalent for 2023 is pretty much where it was back in 2002.  Nominal house prices are well up on where they were in 2002 (€213,000 versus €370,000) but so too is the price of a pint (€3.21 versus €5.48).  The current level is about twice as high as the 34,700 it averaged from 1970 to 1995.

Now maybe recent house prices are flattered by putting them in pints equivalent – due maybe to the impact of Excise increases. The CSO’s 50 years in the EU publication also gives us some other prices to use.

For example, a white sliced pan was 15c in 1973 and is €1.66 now.  Taking the same approach as above shows that average house prices have gone from the equivalent of 60,100 sliced pans in 1973 to an estimated 222,400 now.

Perhaps for reasons for interest, it is harder to track down the full series for the national average price of a sliced pan.  The pattern is as before and again we see that the peak level was seen back in 2006 – when the average house price was equivalent to over 300,000 sliced pans.

Real House Prices

While messing about with pints and sliced pans is useful for showing that nominal prices now are not the same thing are nominal prices from the past, it is whimsical at best.  A much better approach would be to take a much broader set of prices such as the Consumer Price Index (CPI).

Figures from the CSO, show that prices as measured by the CPI in 2023 are around 1,420 per cent higher than they were in 1970.  This means that spending €6,700 in 1970 (the nominal average house price) would be equivalent to spending 6,700 x 15.2 =101,900 if faced with the 2023 prices in the CPI.

We can use this to convert the nominal house price series into real prices using a constant price deflator.  We can choose any year as the base. We will use 2023, thus the 2023 nominal and real figures are the same.  The real figures for all other years are found by determining what the nominal house price for that year would be equivalent to if faced with the 2023 prices in the CPI.

The general pattern is roughly in line with what we got using pints and sliced pans.  As expected using pints flatters current house prices: the price of a pint is around 26 times higher than it was in 1970 compared to a 15 times increase for the CPI.  Sliced pans did the opposite as their price is around 11 times higher.

Anyway, using the CPI-deflated house prices we can see that Ireland had two sustained periods of real house price declines: the first from 1978 through to 1986 and the second from 2007 to 2012.  Again 2006/07 shows as the peak.

Thinking in constant price terms isn’t always that intuitive and the numbers can be a bit messy.  A better way to present real house prices is as an index: select a base year – typically set to 100 – and get values for all other years relative to that.

The shape of the charts are exactly the same.  There was an extraordinary run-up in real houses from 1996 to 2006 when they more than trebled.  When CPI adjusted, house prices in Ireland now are just over three times higher than what they averaged from 1970 to 1995.  Why is this?

There are lots of reasons.  One reason is that buying a capital good such as housing is not the same as buying consumption goods.  Income may play a different role in the demand for housing compared to the demand for individual goods for consumption.

House Price to Income Ratios

To see whether house prices have become more expensive we can check how they have changed relative to income.  Again, we can turn to some long-term series provided by the CSO.  One such  is the historical series published for the Average Industrial Wage (AIW) which goes all the way back to 1938. 

We just need it from 1970 and this shows that the AIW now (€856 per week) is 22 times higher than what it was in 1970 (€23.20 per week).  One simple thing we can do is divide the average nominal house price for each year by the annual Average Industrial Wage for that year. This gives:

In 1970, average house prices were around 5.5 times the annual average industrial wage.  By the depths of the recession of the 1980s this has fallen to 3.6 in 1987.  It remained around four up to 1994 and then exploded, with average house prices reaching almost 11 times the annual industrial wage in 2006 and 2007.  It has remained on a rollercoaster and recent figures put average house prices at over eight times the AIW.

  • 1970: €23.28 x 52 = €1,210           ;            €6,700 / €1,210 = 5.5
  • 1990: €286 x 52 = €14,866           ;            €63,900 / €14,866 = 4.3
  • 2006: €601 x 52 = €31,262           ;            €339,500 / €31,262 = 10.9
  • 2023: €856 x 52 = €44,512           ;             €370,000 / €44,512 = 8.3

Relative to consumer prices, we saw that house prices in 2023 were around three times higher than what they averaged from 1970 to 1995.  Relative to the average industrial wage, we see that house prices are around two times higher than what they averaged over the same period.

Before moving away from income we note two things:

  1. The share of workers earning the AIW has declined
  2. Housing is bought by households, not earners.

The AIW is useful because we have a long-run series for it.  However, the weekly earnings of “production, transport, craft and other manual workers” in industry sectors (NACE B to E) may not be as representative of earnings as they previously were.  Workers in many services sectors will earn less than the AIW.  The share of workers earning more may have also have increased.

We also note that recent decades have seen a significant shift to more dual-income households.  Thus a measure of household income may be more appropriate than the individual-level earnings data.

To get household income we turn to the national accounts. Definitions matter but here we will just note that we are using Gross Household Disposable Income. In national accounts, ‘gross’  means it is before depreciation and ‘disposable’ means after taxes and transfers.  Data from 1995 on is available from the CSO and we will use the nominal growth rates in Stuart (2017) to extend the series back to 1970.  We also need the number of households and these are taken from the Census and interpolated for the intra-Census years.  This gives us figures for disposable income per household.

For 2022, CSO data puts gross household disposable income at €138.2 billion.  With the Census reporting 1,841,152 households for the same year that gives us a figure of just over €75,000 for disposable income per household.  With an average house price of €357,000 in 2022, the ratio of house prices to household disposable income is €357,000 / €75,050 = 4.76.

We can do this for all years since 1970 to get the following:

Again, we see the peak was reached in 2006, when average house prices were almost six times average household disposable income.  It is likely to come in around 4.6 times for 2023, which is around where it was in 2002.

For our long-term comparison, this ratio of price to income averaged 2.6 from 1970 to 1995.  The means the ratio is now 1.75 times than what it was over that period.  The increase is lower than what was shown using individual earnings but not hugely so.

Mortgages: Rates, Repayments and Deposits

Before concluding that housing is far more expensive than it was 25 years ago we must look at how most households pay for housing: mortgages.  This introduces interest rates which are hugely important for determining mortgage payments.  Consider two 25-year mortgages for €100,000 which only differ by the interest rate charged. Let the first have a rate of 4 per cent and the second a rate of 12 per cent. What will be monthly repayments be?

  • €100,000 25-year mortgage at 4% interest = €527.84
  • €100,000 25-year mortgage at 12% interest = €1,053.22

The difference in the interest rate leads to a monthly mortgage payment that is almost twice as high.  Interest rates matter!  And in Ireland mortgage interest rates have varied a lot.

The CSO have long-term data on average mortgage interest rates – see Table 10 of this 2003 publication marking 30 years of Ireland’s EU membership.  Data for recent years are taken from the interest rate statistics of the Central Bank of Ireland.  They give the following series of annual averages:

From 1970 to 1995, mortgage interest rates averaged 11.4 per cent. They are now around 3.6 per cent (though rising).

We can get indicative monthly mortgage payments with three parameters:

  1. The initial amount borrowed
  2. The length of the mortgage
  3. The interest rate charged

To make a comparison we will assume that the initial amount borrowed is 90 per cent of the average house price in each year.  We will use a term of 25 years in all cases and apply the interest rate for each year as shown above.  With these, we can get an indicative monthly mortgage payment and we will put that as a share of monthly average household disposable income. And this shows:

From 1970 to 1995, the indicative monthly mortgage payments averaged 29 per cent of household disposable income.  This is higher than the current estimate which is around 25 per cent of household disposable income.

The above shows that the peak for mortgage payments to income was 41 per cent back in 1982, exceeding the local maximum of 37 per cent from 2007.  As we have seen 2007 corresponds to the peak of real house prices. 1982 corresponds to the peak of mortgage interest rates which averaged 16 per cent over the year (which is also shown in this blog post from the Governor of the Central Bank).

It should be pointed out that this is very much a point-in-time assessment of mortgage payments to income.  Of the three parameters used, only one remains unchanged: the initial amount borrowed.  Key for determining the payment-to-income-ratio are the interest rate charged and the household’s disposable income.  Both of these will change over time for borrowers.  Interest rates trended down through the 80s and 90s while household income has trended up.  For individual borrowers this will reduce the ratio of the monthly mortgage payment to household income.

The point-in-time assessment is useful, though, for the initial payment burdens faced by prospective purchasers.  And as we have seen, this burden is now slightly lower than what it averaged over the period 1970 to 1995.

And this is before increased life expectancy and longer duration mortgages are taken into account.  The above analysis assumed a 25-year term for all mortgages.  In the last 20 years, mortgages with terms of 35 years have become prevalent.

Although the age of first-time buyers has increased this has been somewhat offset by increased life expectancy.  In the early 1970s, an Irish male aged 25 had a life expectancy of a further 46 years (i.e. to 71).  The detailed results based on Census 2016 give an Irish male aged 25 a life expectancy of a further 55 years (i.e. to 80).  It is likely to be now another year or two higher again.

It is not all a one-way street from lower interest rates though.  Present-day borrowers may benefit from lower interest rates but higher prices means they are required to have a bigger deposit. 

From 1970 to 1995, a ten percent deposit based on the average house price was 26 per cent of average annual household disposable income.  In 2023, a ten percent deposit is equivalent to 46 percent of average annual household disposable income.  Prospective buyers may also need even higher deposits due to the Central Bank’s macro-prudential rules on mortgage lending.  And that is what the evidence would appear to bear out.

The overview of new lending from the 2019 Household Credit Market Report shows that the deposits of FTBs in Dublin were just under 90 per cent of household income and they were 72 percent for Non-Dublin FTBs.  Higher deposits means it is harder to access those lower interest rates. And we haven’t even mentioned ever-increasing private rents which further add to the difficulty of gathering that deposit.


Here is a summary of the indicators presented:

By most metrics housing is more expensive than it was 25 to 50 years ago. Nominal prices are almost ten times higher compared to what they averaged from 1970 to 1995.  When adjusted for the general increase in the price level, real house prices are around three times higher now than the 1970-1995 average.  Using income we see that house prices are under two times higher than that average.

When looking at mortgages we see that the measures are lower. Mortgage rates now are around one-third of their level from 1970 and 1995 and though, house prices have increased, an indicative monthly mortgage repayment is now a lower share of average household income.  Moving in the other direction is the deposit required which is now almost twice as large as a share of average household income compared to what it was in the 70s, 80s and early 90s.

Compared to 2003, most of the indicators are actually little different.  Inflation-adjusted house prices and price-to-income ratios did rise, fall and rise again over the period but are back close to 2003 levels. Mortgage interest rates and the payment-to-income ratio are also close to 2003 levels.

The ratio of a 10 per cent deposit to income is also similar but the lending environment is much different. Back in 2003, borrowers were more likely to be lent more than 90 per cent of the purchase price and factors like parental guarantees were much more widespread. The Central Bank’s macro-prudential lending rules now restrict these with a greater requirement for borrowers to have the deposit to hand.

Wednesday, December 20, 2023

The Ongoing Trickle of Repossessions

The latest mortgage arrears statistics from the Central Bank provide an update for the end of September (Q3) 2023. They show there continues to be a trickle of repossessions.

PDH Repossessions to Q3 2023 - AREA

During the third quarter of 2023, there were ten court-ordered repossessions and ten abandonments/voluntary surrenders of primary dwelling houses.  During 2023, court-ordered repossessions have been occurring at a rate of around one per week.

We are now also provided with a more complete breakdown of repossession activity.  The number of repossessions shown above can be broken down into those carried out by banks and non-banks. Non-banks include regulated lenders such as Pepper and Start and also unregulated loan owners.  The chart below shows a split out of the maroon area in the opening chart.

PDH Repossessions by Banks and Non Banks to Q3 2023

In the last 15 years, there has been a total of 3,400 court-ordered PDH repossessions. Of these around 70 per cent have been undertaken for banks. However, the banks stalled their execution of repossession orders in early 2020 (likely pandemic related) and there has been no increase since.  In the year to the end of September, there were 57 court-ordered repossessions and 42 (nearly 75 per cent) were carried out for non-banks.

Figures on legal proceedings suggest that the banks have pretty much ceased using legal proceedings to obtain repossessions orders.  Here is the number of accounts with legal proceedings in progress.

PDH Legal Proceedings by Banks and Non Banks to Q3 2023

In early 2016, the banks had almost 12,000 PDH mortgage accounts with legal proceedings in progress. The last figures show that this had fallen to just 1,100 by the end of September. This is 0.2 per cent (1 in 500) of the 594,000 PDH mortgage accounts held by the banks.

Some of the reduction, of course, came about as the non-performing loans were sold to non-bank entities but the fall in the overall total is very clear. And it can be seen that the level of non-banks has been stable in recent years. In Q1 2019, non-banks had 4,400 PDH accounts with ongoing legal proceedings. The latest figure for these entities is 4,300.

That is not to say there has been an improvement in the mortgage accounts held by non-banks.  The arrears on these accounts is still very high.

PDH Mortgage Accounts with Non Bank Entities Q3 2023

Of the 710,000 or so PDH mortgage accounts in the latest update, almost 115,000 (16 per cent) were held  by non-bank entities.  And of those almost a quarter (27,300) were in some form of arrears.

The arrears ranged from relatively small (less than 90 days arrears) to incredibly large (over 10 years arrears).  It is bizarre that we have to produce figures for such significant arrears.  Non-bank entities had 21,000 PDH mortgage accounts that were more than 90 days in arrears.

Included in that are 4,500 accounts that are more than ten years in arrears.  The total balance outstanding on these loans is €1.3 billion, giving an average balance of €280,000.  There have been €833 million of missed payments on these giving an average level of arrears of just under €185,000.   That is an incredible amount of missing payments.

It is worth noting again that arrears is not a good measure of current loan distress. It does not tell us when the payments were missed.  An account that had two years of missed payments a decade ago but has had every payment made since will be counted as being two years in arrears.

Also, the measurement of arrears in terms of days past due is impacted by the repayment required.  If an account has had €6,000 of historical missed payments and the required monthly payment is €1,000 then that account will be 180 days (six months) in arrears.  If the required monthly payment rises to €1,200 – due to. say, interest rate increases – and there is no new arrears, then the €6,000 of historical arrears becomes the equivalent of 150 days (five months) of arrears. 

This measurement issue won’t impact a count of the total number of accounts in arrears.  There has been no rise in the total number of accounts in arrears.

PDH Mortgage Accounts in Arrears Q3 2023

As shown above, there remains around 29,000 PDH mortgage accounts which are more than 90 days in arrears.  Of these 21,000 are held by non-bank entities.  The banks have 8,000 such accounts but this represents just 1.3 per cent of the total PDH mortgages they hold (594,000).  The banks do not have a problem with a long-term mortgage arrears.

In some cases this will have occurred with the borrower getting back in track either with or without a cure or modification. In other cases it is because the banks have simply sold the loans.  There were plenty of claims that these sales would lead to a surge in repossessions. All we have seen so far is a trickle.