Why Morgan Kelly is spectacularly wrong on the cost of Mortgage Defaults

Discussion in 'Housing and mortgage arrears - policy issues' started by Brendan Burgess, Aug 19, 2011.

  1. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    Just like any other economist, Morgan Kelly is sometimes right and he is sometimes wrong. He is spectacularly wrong on the cost of mortgage defaults.

    While giving the Hubert Butler lecture in Kilkenny, in early August, Professor Morgan Kelly said:
    With these comments, Professor Kelly implied that the state will need to provide an additional €5 billion or more to account for these losses. This is part of the reason why he says that the national debt will be €250 billion by the end of 2015. This information was based on a press release by the Irish Brokers Association, which itself, was based on anecdotes.

    Professor Kelly is wrong on so many counts, it beggars belief

    • It is likely that the total number of mortgages in excess of €1m is probably around 4,000 and not 10,000.
    • The total number of interest-only mortgages is probably less than 2,000. Only Bank of Scotland did full-term interest only mortgages as a standard product. Some others were interest-only for a few years and then switched to capital and interest.
    • These loans were given to professionals such as doctors, dentists, solicitors, barristers and accountants. While most of these have suffered a decline in their income, the vast majority can well afford a cup of coffee and 90% of them are making their repayments in full and on time each month.
    • But worst of all, Professor Kelly made no distinction between mortgages given out by the covered banks such as AIB and Bank or Ireland and those given out by the foreign owned banks such as Bank of Scotland and Ulster Bank for which the taxpayer has no liability. None of the covered banks did full-term interest only loans as standard, so there are probably fewer than 100 of these – not 10,000.

    I checked with the Central Bank and the Central Statistics Office and they do not know how many mortgages in excess of €1m are on the books of the covered banks, although I understand that the CSO may be able to produce this information in the future as part of their house price index work.

    The 10,000 figure is a huge exaggeration. Rather than rely on anecdote and broker estimates, I checked directly with sources in each of the covered institutions - Bank of Ireland, ICS, AIB, PTSB and the EBS and the absolute maximum number of €1m mortgages on the books today is 2,500 and this includes principal private residences and investment properties. This represents around 3% of the total mortgage book of €98 billion.

    How many of these will default? We know that around 90% of these are paying the full interest and capital repayments on these mortgages at the moment. Morgan Kelly expects a Michael Davitt type figure to lead a mortgage strike and that 25% of borrowers will default. I wouldn’t expect that doctors, dentists, solicitors and accountants will form the bedrock of this campaign, but I will use Kelly’s 25% default figure anyway.

    Estimated number of €1m mortgages issued by covered banks|2,500
    Estimated default rate|25%
    Estimated number of defaults|625
    Average loss per mortgage|€500k
    Potential cost to the taxpayer|€300 million
    Cost implied by Professor Kelly|€5 billion +
    So the taxpayer has a potential loss of €300m, not the €5 billion which Morgan Kelly implied. So he is out by a factor of around 20. And these losses have been fully provided for in the recent stress tests and recapitalisations so there will be no further cost to the taxpayer.

    This is not the first time that Morgan Kelly has made an error of this magnitude. In November last year he spread panic when he said
    This hit the headlines internationally with papers such as the Guardian claiming that “a new wave of toxic [mortgage] debt could sink the country entirely”
    But only a few months earlier, in June 2010, he computed the total cost of the bank bailout to be €106 billion. This included a figure of only €8 billion for Irish mortgages which represented a 20% default rate. By November he was forecasting 200,000 defaults, which would be a 25% default rate. I think that this 25% default rate is too high, but even if this forecast is correct, it will cost the taxpayer around €10 billion . How can this be “ a good deal more traumatic than the bank bailout”?

    Yesterday, he said that the cost of debt forgiveness for home owners would be “only” €6 billion. Let’s say that is €4 billion for the covered banks. The covered banks have €23 billion in buy-to-lets. Even if half of these default, the cost will be €6 billion. So the total cost of mortgage defaults will be around €10 billion. Let's assume I have made a mistake somewhere in my calculations and double the cost to €20 billion.

    Anglo alone will cost around €35 billion. In the most extreme circumstances outlined by Professor Kelly, the losses on mortgages will account for 10% of the total cost of the bailout. It was outrageous to characterise this as being a good deal more traumatic than the bank bailout.

    Professor Kelly’s forecast on house prices has been spectacularly right. But his forecasts on the cost of mortgage defaults are spectacularly wrong. It is important that policymakers, the media and the public don’t attribute to him an infallibility which he has not earned.
     
  2. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    All my estimates here are excessively conservative.

    I have included the losses for the Bank of Ireland and ICS although their losses are likely to be borne by their shareholders and not the taxpayer.

    I have included owner occupied homes and buy to lets, although in Kilkenny he appeared to be talking about owner occupied loans only, although now he claims he was talking only about buy-to-lets.

    I have included interest only and repayment mortgages, although this difference is not as important as he suggests.

    In totalling the 2,500 figure I increased the estimates from two of the lenders as they were way below the market and I felt that there may have been a classification error.

    I assume a loss of €500k on each mortgage. This assumes that all these properties were bought at the peak and so have fallen by 50%. And that the owners have repaid no capital.
     
  3. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    Why is Professor Kelly so obsessed with interest-only mortgages?

    Interest-only mortgages have big advantages as well as disadvantages. The default rates are much lower. It is much easier to pay just the interest instead of the interest and the capital. In fact, when repayment mortgages get into difficulty, they are often rescheduled to interest-only.

    Of course, if a borrower took out a 25 year repayment mortgage in 2006 at the peak of the market, they would have reduced the principal by 17% by now.

    PTSB did have a Professional Investor package which gave interest-only mortgages, subject to review. These are now being reviewed and many of these borrowers will not be able to pay the capital. But while that is bad for the borrower, it is good for the lender. They have the option of leaving them on interest-only or taking them off the tracker.

    Bank of Scotland did jumbo mortgages full term interest only. But they are of no concern to the Irish taxpayer.
     
  4. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    In his defence yesterday, he was contradicting himself again.

    In Kilkenny

    Yesterday:

    So which did they buy? Very large properties or shoebox apartments?

    In Kilkenny, he was very clear - "10,000 properties of between €1m and €2m each"
     
  5. Sunny

    Sunny Frequent Poster

    Posts:
    3,145
    Good to see him throwing in econometric calculations and the concept of tail risk. I look forward to him publishing his data and models for examination....
     
  6. SeamusCoffey

    SeamusCoffey Frequent Poster

    Posts:
    29
    This could potentially be great data. I have looked at the issue of the 10,000 mortgages myself and while I believe they don't exist I'm not sure we can be 100% certain that they don't.

    Today Kelly has claimed that he was referring to investment properties. Your figures are that there is no more than 2,500 of these million+ mortgages in total.

    Where did you get these numbers? Are they as much as a guess as the original figure used by Kelly from the Irish Brokers Association?
     
  7. Sunny

    Sunny Frequent Poster

    Posts:
    3,145
    I am not sure why these mortgages have gotten the attention they have. Is Morgan Kelly suggesting that the banks have put all these mortgages in a dark cupboard and forgotten all about them or something? I am not sure why he is implying that this group of mortgages contain hidden losses that we know nothing about.
     
  8. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    Thanks Séamus

    Here is my process and why it took two weeks to get the data.



    1) I was astonished by the 10,000 figure so I contacted the CSO, the Central Bank and the Irish Banking Federation, none of which had the figures.
    2) I contacted Morgan Kelly who sent me the Irish Times article which he quoted
    3) I contacted people in the 4 institutions ( counting Bank of Ireland and ICS as one)
    4) I contacted two of the non covered institutions just to get an additional sense of perspective on the figures and the market.
    5) I spoke to three mortgage brokers to get their views on how frequent these were.



    When I added the total for the 4 covered institutions they came to less than 2500. The figures for one of the institutions felt too low compared to the information I had got from the other 3 covered and 2 non covered, so I upped the figure to 2500 to be on the safe side.

    So, I think it's far more likely that my estimate is more accurate than Morgan Kelly's figure which he got from a newspaper report.

    But let's assume that lenders all conspired to lie to me or misunderstood what I was asking and fed me the wrong figures and that Morgan Kelly is indeed correct and that there are 10,000 interest only mortgages in excess of €1m.

    The three brokers to whom I spoke suggested that Bank of Scotland could have anywhere between 35% and 50% of this market. Add in Ulster Bank, KBC, National Irish Bank and Leeds, and it's safe to assume that the non covered banks have around 50% of the jumbo interest only market.

    So the covered banks have 5,000 of these loans (and not the 2,500 I have estimated)

    They are nealy all cheap trackers. They are all interest only (by definition) So someone who borrowed €1m has to pay around €2,500 a month in repayments.

    They are investment properties so they are getting rental income. In most cases the rent would be covering the income, so there is no problem, apart from negative equity.

    Here is the revised table

    Estimated number of €1m mortgages issued by covered banks|5,000
    Estimated default rate|25%
    Estimated number of defaults|1250
    Average loss per mortgage|€500k
    Potential cost to the taxpayer|€600 million
    Cost implied by Professor Kelly|€5 billion +
     
  9. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    Hi Sunny

    This is why he said he was concerned

    This is how the Indo reported it the day after the speech

     
  10. Sunny

    Sunny Frequent Poster

    Posts:
    3,145
    Thanks Brendan.

    I struggle to see the logic of the argument.
     
  11. Mrs Vimes

    Mrs Vimes Frequent Poster

    Posts:
    872
    The Independent has figures from Stamp Duty receipts here

    I think Morgan Kelly's figures are way out of line
     
  12. kaplan

    kaplan Frequent Poster

    Posts:
    306
    Brendan you might want to consider the central bank PCAR loan loss estimates for buy to let and commercial real estate properties - which include the professional investor sub-class borrowers:
    BTL book 23.2bn base case losses per CB were 2.2bn, blackrock €4bn - stress case losses were CB €3.3bn and Blackrock €6bn
    CRE book 40.4bn base case per CB €7.1bn and stress case €8.9bn.

    So total base case losses per CB three year horizon are €9.3bn, stress losses are €10.4bn. If Kelly refers to professional investor types who would have both BTL and CRE exposures then his loss estimate may be on the money.

    As loss estimates apply to just four lenders they would have to be increased to take account of other lenders exposures - BOSI, Anglo, Ulster, NIB, ACC, Kbc and of course INBS.
     
  13. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    Hi Kaplan

    At the time of the speech he appeared to be referring to owner occupied homes. He has since claimed that he was referring to buy to lets. He was definitely not referring at any stage to commercial real estate.

    He has made three forecasts of the cost of mortgage defaults.

    1) In April 2010, he estimated that the total cost of mortgage defaults would be €8 billion or 10-% of the mortgage book of AIB, BoI and EBS. ( He left out PTSB as they were thought to be ok).

    This forecast may turn out to be correct.

    2) In November 2010, he said "If you thought the bank bailout was bad, wait until mortgage defaults hit home… This time the bad loans will be mortgages, and the foreign creditor who cannot be repaid is the ECB. In consequence, the second act promises to be a good deal more traumatic than the first."

    Given that he was forecasting that the bailout would cost €106 billion, presumably this suggests that mortgage defaults would cost more than this.

    Either 1) is way wrong or 2) is way wrong. It is clear to me, that 2) is way wrong.

    3) Despite a forecast that total defaults would be €8 billion, he now says that mortgages in excess of €1m will cost the taxpayer more than €5 billion. This is wrong as well. Completely wrong as it's based on incorrect information and wrong analysis.

    4) Since I started this thread he has said that debt forgiveness would cost €5 - 6 billion "10% of the €56 billion home loan book". The home loan book is €115 billion so he is wrong again here. Now it's possible that it will cost €5 billion, but he would then be accidentally right.

    His analysis is sloppy which is unacceptable for someone whom the media listen to.
     
  14. onq

    onq Former user

    Posts:
    4,380
    Thanks for debunking Kelly's estimates about the mortgage defaults last November Brendan. Reading that piece sent me into a tailspin that I didn't pull out of for a week.
     
  15. hastalavista

    hastalavista Frequent Poster

    Posts:
    448
  16. Brendan Burgess

    Brendan Burgess Founder

    Posts:
    31,144
    It's here

    I will need a fresh head tomorrow to read it. This line struck me as interesting

    or to put it another way...

    or as I said...

     
  17. Duke of Marmalade

    Duke of Marmalade Frequent Poster

    Posts:
    1,547
    This is an absolutely pathetic retraction from MK. Besides completely revising his figures he now admits:
    Tail between his legs but couched in academic gobbledygook so that it is unlikely to be seen for what it is, an absolute retraction. I note he still hasn't recognised Brendan's point that the BoS were the main culprit and they are not a covered institution.
     
  18. dewdrop

    dewdrop Frequent Poster

    Posts:
    1,296
    I wonder how many people can understand this analysis?