Thursday, December 12, 2013

12/12/2013: Measuring the Mortgages Crisis in Ireland

As the readers of this blog would know, I rarely comment on articles in Irish press, and rarer yet on articles in the Irish Times. So here is a rare occasion, not because of the article itself, but because of what it suggests about our national treatment of statistics.

Let's start from the top. The New York Times published an article on Irish crisis today. Here's the link:

Irish Times - in some ways correctly took the New York Times article to task:

Let me take up one point in the two articles. Original version of the NYT article cited - quoting from the IT response - that "most startling is the assertion that two-thirds of homeowners have not paid their mortgage “on time for the last two years”".

IT correctly points that this is not true, saying that "The bank’s most recent arrears figures suggest 18.5 per cent of mortgage holders are in arrears of some sort or other.
They also indicate that 22 per cent of those currently in arrears are behind in their payments for at least two years or more."

The NYT published correction to their original claim. Story ends there.

But from the point of view of reality, it does not. This is pivotal to our narrative about the crisis.

Mortgages arrears have many meanings in the economy. But in the social context and in relation to mapping out the extent of the crisis here's what matters: Mortgages arrears are a signifier of the extent of the crisis. In this, they are neither the only indictor, nor are they a relative indicator. Let me explain.
  1. Assume we want to identify the extent of the crisis as it impacted the households holdings of property.
  2. Assume we have official data to do so only.
From (1) and (2), identifying the crisis extent is simple and yet hard. 

Take an analogy of identifying the crisis in the macroeconomy. That would be GDP. Or rather, the size of the crisis = the gap between the GDP at pre-crisis peak to GDP at crisis-period trough. One thing that does not matter to this analysis is where the GDP is today (post-trough). Should in the future the GDP hit a new trough and should the drivers for this be consistent with the drivers for the original crisis, then that new trough becomes the crisis metric. Should GDP recover to pre-crisis highs (as it will one day), the magnitude of the crisis will not be zero, it will still be GDP pre-crisis less GDP at trough.

Variants on the above are possible by looking at various GDP metrics and/or pre-crisis and trough metrics (trend, potential etc). But the essence of analysis is the same: GDP pre-crisis - GDP at trough = Crisis Impact.

Now, back to the original issue: How shall be measure the impact of the crisis when it comes to mortgages?

The IT comments can suggest (and usually the media obliges to take this as given) that Arrears Current = Crisis Impact. But are they?

My view is that they are not. Let's compute the total impact:
  1. Peak of arrears (we are yet to reach that) = part of impact
  2. Restructured mortgages that are not in arrears = part of impact for two reasons: (a) they face high probability of going back into arrears (just under 50:50 chance currently and declining slowly); and (b) restructured mortgages are no longer the original pre-crisis mortgages, so the mere fact of restructuring them is a sign of the crisis impact
  3. Repossessed homes = direct impact; and
  4. Voluntary surrenders = direct impact.
What do we know from official sources: Total mortgages outstanding: 915,746 per CBofI (composed of 768,136 principal residences-linked mortgages and 147,610 BTLs), of these:
  1. Total mortgages in arrears: 181,946 per CBofI (composed of 141,520 principle residences and 40,426 BTLs)
  2. Restructured, not in arrears: 56,186 (composed of 43,034 principal residences and 13,152 BTLs)
  3. Repossessed homes - we have no numbers for aggregates repossessed - neither the CBofI, nor Department of Finance report these on any regular basis. But in Q3 2013 we had 1,566 properties in repossession (1,050 residences and 516 BTLs). These are properties held in possession by the banks. We do not know how many they have sold since the beginning of the crisis.
  4. Voluntary surrenders - we have no data on these from any official source, but the properties that are surrendered and are still in the possession of the banks are aggregated into (3) above.
So, with incomplete information on (3) and (4), to-date, the extent of the crisis is for all types of properties:

181,946 in arrears + 56,186 restructured, not in arrears + 1,566 repossessed and surrendered = 239,698 accounts or, ca 26% of all accounts outstanding.

And the number is growing...

This is not 2/3rds as claimed originally in the NYT, not even 1/3rd, and it is certainly not the percentage of mortgages in trouble over 2 years... and the above 26% include BTLs too... But the true extent of the crisis is that 26 out of 100 mortgages in the country have been directly impacted by it. And the crisis has not peaked, yet.

But here's what this tells us about our psychology when it comes to measuring the extent of the crisis: we commonly interpret arrears alone (and often only arrears in excess of 90 days) as the metric of the crisis. This is an error - an error based on the implicit anchoring of the idea of the crisis to the news and thus, to relative position in time. This is simply wrong. The crisis of WW2 is measured by the absolute level of destruction wrecked at the peak, cumulatively, not by where the losses were in 1955 or in 1948.

Actually, should you be interested, I track the evolution of the above metric (I call it mortgages in default, defaulted or at risk of default) in my regular posts on CBofI quarterly data. The latest was provided here:

And for the conclusion: I recall in 2007 CEO of AIB at the time stating in a meeting with analysts that "Irish people do not default on mortgages. They never do." I replied: "Never is a very precise term. Is there any uncertainty around this claim?" and he retorted: "None." Back to that 26% figure, then?..
Post a Comment