The income distribution and the Irish mortgage market

How concentrated are mortgage originations among those on higher incomes? Has this pattern changed through expansion and contraction in the Irish housing market? Combining for the first time information on the incomes at origination of a large sample of Irish mortgage holders with survey information on the population income distribution in each year, my colleague Reamonn Lydon and I address these issues over the period 1994 to 2014 in an Economic Letter released recently by the Central Bank.

In the work we highlight that the income profile of borrowers entering the First Time Buyer, Mover-Purchaser (also referred to as Second and Subsequent Buyer), and Buy to Let markets is markedly different.

The first chart below shows the evolution of the share of new First Time Buyer mortgage originations going to each population quintile between 1994 and 2014 (income distribution data were not available to us for 2015 at the time of carrying out the work). A number of patterns are evident:

  • The share of those in the top income quintile fell from over 40 per cent in 1994 to around 12 per cent by 2008.
  • The share of those in the middle income quintile rose from 15 per cent to over 40 per cent over the same period.
  • There has been a slight reversal of this pattern since the financial crisis; however, the share of originations going to the middle quintile is still well ahead of the top quintile.
  • Those in the bottom 40 per cent of population incomes have generally accounted for less than ten per cent of mortgage originations in a given year.

ftb

Next we examine the Mover-Purchaser or SSB market, and find that:

  • There was a similar convergence in the market shares of the 5th and 3rd quintiles over the Celtic Tiger period.
  • The reallocation towards the top income quintile in this market has been much sharper since 2008, with the market share standing at above 60 per cent for 2014.

ssb

The findings suggest that the crisis has been associated with some significant structural shifts in mortgage market participation. In the case of the SSB market, it is possible that the role of negative equity in impeding mover-purchases has been much more prevalent in recent years outside of the top quintile of the population income distribution. In the case of First Time Buyers, where the changes have been relatively less pronounced, the shifting age profile, where borrowers are entering the market later in life, may also explain the shift towards higher-income purchasers. Our research does not attempt to definitively quantify the role of supply side (such as bank lending policies) and demand side factors in explaining these changing patterns.

Other related research was also released in the Bank’s recent Quarterly Bulletin: The balancing act: household indebtedness over the lifecycle, by Apostolos Fasianos, Reamonn Lydon and Tara McIndoe-Calder. Finally, another related piece came out as a Research Technical Paper on the Great Irish (De)Leveraging by Reamonn Lydon and Tara McIndoe-Calder.

Household formation among young adults

A guest post from Reamonn Lydon and Apostolos Fasianos of the Irish Economic Analysis division here at the Bank.

The overall population increased by 4% between 2008 and 2016. At the same time, the number of young adults aged between 20 and 34 fell by a quarter– from 1.15 million to 860 thousand (see Table 1. See also the excellent study by Glynn, Kelly and MacEinrí (2015) on migration patterns for this group).

Table 1 CSO population estimates (Table PEA01)
 (`000s) 2008 2016 % change
Age <20 1209 1327.4 9.8%
20-24 373.6 226.8 -39.3%
25-34 777.8 633.6 -18.5%
35-44 662.2 733.2 10.7%
45-64 978.9 1127.2 15.1%
65+ 483.7 625.5 29.3%
4485.2 4673.7 4.2%

The large decline in the 20-34 population means that housing demand will be lower than the past.  However, there have also been significant changes in the household formation patterns of this group which are relevant when it comes to thinking about housing demand in the future.  As the figure below shows, just before the property crash just over 30% of young adults lived with parents, but by 2016 this had risen 37%.  Taking into account the population drop, this is around an additional 25,000 young adults versus the situation in 2006, and just under 320,000 in total living with parents in 2016.

F1Census data for 2016 is not yet available to calculate the latest figures, so we have drawn on the QNHS and Household Finance and Consumption Survey (HFCS, 2013) to try and complete the picture to 2016.  The HFCS is particularly useful as it allows us compare Ireland with other countries (Figure 2).  Ireland looks similar to both the EU and US (although the US data is for 18-34 year olds living with relatives, not just parents), but is somewhat higher than the UK. We know, however, that UK third level students are more likely than their Irish counterparts to live away from home.  Southern European countries, with relatively high rates of youth unemployment – such as Spain, Portugal and Italy tend to have a higher proportion of young adults living with their parents.

rea2

What do these figures mean for housing demand?

The answer depends on the extent to which you believe the shift towards more young adults living at home is a cyclical or a structural change. Certainly, there is a slow-moving cyclical part to it – the proportion rose as the employment prospects for this group worsened and young people stayed on in education (Conefrey, 2011).  The CSO also reported a sharp drop in the proportion of 19-24 year-olds in shared accommodation (i.e. renting), from 22 to 18% between 2006 and 2011.  So there may be a jump in demand in the short term, because not only do the delayed entrants want to enter the market after a (cyclical) delay, but those who are younger will now start forming households at a younger age.  There is already some evidence of this in the 2016 QNHS, which shows the percentage of 20-24 year olds living at home falling for the first time in almost 10 years, from 70 to 68%.

However, there might also be structural changes to consider. For example, if the easy credit of the bubble years meant that buyers got on the housing ladder at a younger age than previously, and this has since been reversed, then the ‘pent-up’ demand might not be so large.  We know that the average age of FTBs has risen in recent years, having fallen during the boom.  In this case, young adults could continue to form households at a rate similar to what we are now seeing.

In all likelihood, the shifts we have witnessed are a mix of cyclical and structural changes. However, how much is cyclical does matter. As Table 1 showed, there were just over 860,000 20-34 year-olds in 2016. Ignoring immigration flows which could increase the size of this age cohort further, each 1% fall in the proportion living at home means an additional 8,600 individuals looking to rent or buy. This is a large figure in the context of current estimates for annual housing demand, which range from 20,000 and 40,000 units.

Mortgage term as a credit condition

The aim of this post is to introduce the topic of mortgage term to maturity in an Irish setting. I will outline how mortgage terms were highly pro-cyclical during the pre-2008 expansion, and that they were used by credit-hungry borrowers to make high-leverage strategies more affordable on a monthly basis. It is well established that credit conditions, as measured by Loan to Value ratios (LTV), Loan to Income ratios (LTI) and Debt Service Ratios (DSR, ratio of monthly repayments to net income) reached unsustainable levels in Ireland in the run-up to the 2008 crash.[1]  By contrast terms have received much less attention. Previous work by McCarthy and McQuinn (2013) on Irish credit conditions is the only piece known to me which contains an analysis of mortgage terms. They show that median terms increased rapidly in the 2000-2008 period, and did not decrease after the housing market crash. Further, they show that terms were higher for First Time Buyers, and that longer terms are correlated with an easing in other credit conditions. Today’s post concerns recent analysis carried out by my colleague Edward Gaffney which looks at the distribution of mortgage terms for loans originated between 1997 and 2014.

Why should mortgage terms be of interest to us? Firstly, they can have a huge impact on mortgage affordability. Identical loans, for identical houses, to people with identical incomes, at identical interest rates, can have widely varying monthly repayments driven by varying terms. As a simple example, a loan for €250,000 at an annual rate of 4.5 per cent has a monthly repayment of €1,581 when taken out over 20 years, which falls to €1,183 when that loan is extended to 35 years. A saving not to be sniffed at!

Secondly, they appear to have associations with risk-taking behaviour. Research from Central Bank colleagues[2] has shown that, controlling for a range of factors associated with higher credit risk, mortgages with longer terms are more likely to default, even though a longer term allows for a lower monthly mortgage repayment, all else equal!

Finally, terms also have a big impact on banks’ profitability, as longer term loans will have higher lifetime interest income for the bank. Further, longer terms allow banks to make larger loans while continuing to respect any Debt Service to Income rules that may be in place.

So what can we say about mortgage terms in Ireland? In all that follows, I will focus on the First Time Buyer (FTB) segment of the mortgage market. Our first figure (Fig 1) provides clear evidence from Edward’s work that mortgage terms lengthened significantly during the boom phase in Ireland. Of the 1997 cohort still outstanding in 2014, 60 per cent were originated with terms of 20 years or less, with 90 per cent having terms under 25 years. At the turn of the millennium, the 35 year mortgage was close to non-existent. However, its proliferation through the period of rapid credit growth was quite remarkable, moving to a market share of roughly 50 per cent by 2006-07, with a further 5 per cent of the market taking terms between 35 and 40 years.

While Kelly, McCann and O’Toole (2015) report that credit conditions tightened considerably in the aftermath of the financial crisis, Edward’s chart shows that “credit tightening” in mortgage terms has been much less stark, with 60 per cent of mortgages originated in 2014 still having terms above 30 years.

Fig 1: The distribution of originated mortgage terms per year, First Time Buyer segment 1997 to 2014.

term_line

This evidence of a structural shift towards longer terms begs a number of questions relating to the role of term as a credit condition. Figure 2a from Edward’s work provides conclusive evidence that, during the boom phase, longer terms were associated with higher leverage. In 2007, those taking out 40 year mortgages had an average LTV of over 90, with LTVs of 85, 72, 60 and 50 as we move in five-yearly intervals down to 20 year mortgages. While these LTVs have converged since the crisis, the rank ordering persists. Figure 2b completes the picture by showing that borrowers on different terms have in fact been accessing similarly valued houses, which implies of course that those with longer terms were taking out larger average loans. Taken together, these figures strongly suggest that term was being used by FTBs to mount the property ladder at as high a point as possible for a given down-payment amount, using higher originating LTVs to access valuable housing with small down-payments and large loans, while easing the monthly repayment burden of this high-leverage strategy via longer terms.

Figure 2: Average LTV and property value for mortgages originating at different terms.

(a)    LTV (b)   Average Property Value
 term_ltv  term_val

Our next piece of evidence links borrowers’ incomes to mortgage terms. Figure 3 reports clear differences in originating Loan to Income ratios (LTI) across term groups. Those on 40 year mortgages in 2007 were accessing loans with an average LTI of 5, while the equivalent number was under 3 for those with 20 year mortgages, again providing strong evidence that mortgage terms were used as part of a broad “credit conditions package” by credit-hungry borrowers. Figure 3(b) confirms that this highly-indebted strategy was in fact more common among high-income borrowers in the run-up to 2008, with median incomes falling as terms shorten. The one exception to this rule is the 40-year mortgage, which appears to have been popular among lower-income households with extremely high LTIs during its short existence.

Figure 3: Average LTI and income for mortgages originating at different terms.

(a)    Loan to Income ratios (b)   Incomes (median)
 term_lti  term_inc

So where did this “credit condition package” leave borrowers on a monthly basis? It is unclear from the above whether long terms acted to offset the affordability difficulties brought on by high-leverage strategies. To do this, Edward calculates a monthly repayment to gross income (RTI) ratio for each loan in the data, applying an indicative opening interest rate. The evidence is conclusive: despite the fact that longer terms mechanically improve mortgage affordability by lowering repayments all other things equal, it was still the case up to 2008 that borrowers with longer terms were taking out such large loans that their RTIs were in fact higher than borrowers with shorter terms. This is likely part of the explanation for the finding of Kelly, O’Malley and O’Toole (2015) that longer-term mortgages have higher default probabilities, even after controlling for a range of explanatory factors.

More posts on the mortgage market in Ireland to follow over the coming months.

[1] McCarthy and McQuinn (2013) and Kelly, McCann and O’Toole (2015)

[2] Kelly, O’Malley and O’Toole (2015)