IMF on Costs of Financial Stabilisation

The Irish Times lead story cites the IMF’s Global Financial Stability report as having the following sentence: “The United States, United Kingdom and Ireland face some of the largest potential costs of financial stabilisation (12 to 13 per cent of GDP) given the scale of mortgage defaults.” It turns out, however, that the IT was a little behind on a (fairly silly) controversy about this sentence.

It turns out that the IMF’s cost estimates are not new at all but actually first appeared on page 17 of this report released on March 6, which was written as a companion to this report on the outlook for public finances around the world.  The March 6 paper reports a cost figure of 13.9 percent of Irish GDP, which amounts to €24 billion.  Table 4 also reports the cost for the UK at 9.1 percent of GDP.

For this reason, there was a bit of a flap over this when the BBC reported the 12-13 percent figure, with the UK Treasury pointing out correctly that the sentence and its accompanying table were wrong.  The version of the report on the website no long contains the parenthetical “(12 to 13 per cent of GDP)” that the Irish Times had quoted and the table has been altered—I think Ireland may have been listed in the original Table 1.8 but we are not now.  In any case, you can find the original source of the calculations from the above link.

So, not the IMF’s finest hour.  However, beyond the silliness, it is clear that the IMF’s assessment of the likely costs of financial sector support measures to the Irish taxpayer does not fit well with the government’s current stance that “under extreme stress scenarios” BOI only need €3.5 billion in additional capital, while AIB only need €5 billion.

12 replies on “IMF on Costs of Financial Stabilisation”

Thanks Lorcan — That link still has the wrong 13.4% figure for the UK so I reckon it won’t be long before they take it down, perhaps replacing it with the shorter table in the report.

The figures are out of date, superceded already, but the confusion is an indicator that some have voted with their feet. The Bush admin on the economic side had difficulty recruiting and keeping personnel for some time before the bubble was acknowledged to have burst. Those who know are generally the first to leave a sinking ship. The media points on these issues have yet to be linked and brought to our attention as connected in this way. But we can’t have a new world order unless we are all kept in ignorance?
Do I recall that the Icelandic Finance/Prime minister actually found it impossible to believe what had happened? But he was a part time poet. Anyone got any comments on the book “Confessions of an economic hitman”? Or would that suggest that the mess was foreseeable to those involved?
Remember Iceland!

How do the IMF estimate these “financial stabilization costs”?

Well, they don’t say, exactly. The bottom line to this comment is: it seems to be based mainly on inferring likelihood and scale of bank defaults from credit default swaps for Irish banks (as of last November). There it is in a nutshell, take it or leave it.

So, where is the basis of the calculation set out? Curiously, this seems to be in a mysterious Chapter 5, which was not included in the published GSFR.

It’s pretty clear, however, that the estimates come from applying loss factors to the various elements of the gross financial support (lending etc) under the headings set out in Table 1 of the background paper. These include “Capital Injections”, “Purchase of Assets and Lending by Treasury”, “Central Bank Support Provided with Treasury Backing”, “Liquidity Provision and Other Support by Central Bank” and “Guarantees”. Only the first and last items are relevant for Ireland.

These are the gross amounts committed or guaranteed.

Regressing the estimated financial stabilization costs from (old) Table 1.8 of the GFSR on the individual elements of the “Headline Support” table does not give a perfect fit (RSQ=0.96).

This appears to reflect the use of country-specific loss conversion factors for the Guarantee item, on which Ireland is an outlier. The text states that this was based on the banks’ CDS in November:

“The expected cost of the (explicit) guarantees provided so far is not trivial, but the margin of uncertainty is large. Some indicative estimates can be obtained using standard financial derivative pricing models—in particular, by estimating the Expected Default Frequency Implied CDS (EICDS) spreads and applying them to the guaranteed amounts. EICDS can be regarded as indicative of the “insurance” premium for providing the
guarantees, and the approach—which takes into account market volatility and hence the probability of default of individual institutions—provides an approximate measure of the cost to government of providing this “insurance.” Based on November 2008 market data, outlays from contingent liabilities could be of the order of 2–6 percent of GDP (cumulative) for 2009–13 for the advanced G-20 countries, although higher outlays are possible in some countries (notably Ireland).”

There is some more about the method used in Box 3.6 of the GSFR.

Now I have found what I referred to as the mysterious Chapter 5. It is in the other paper mentioned by Karl, and sure enough it is as I supposed. Almost all of the estimated financial stabilization cost for Ireland comes from applying expected loss conversion factors from CDS-spreads to the Government guarantee`.

In calculating the cost of financial stability, the IMF does not seem to take into account the increased cost to the government of accessing funds.

The IMF predicts that our national debt effectively double in the two years to dec 2010. But, because of the perceived (actual) instability of our financial institutions can we reasonably attribute 100/120bps of our funding costs to the actions taken to protect the institutions? (the spread traded in the 50-100bps spread range before sept)

@Patrick Honohan: Impressive analysis in such a short period! As for the IMF methodology as you know that is what financial economists used to call “rational inference”. Assume that aggregate investors correctly price all financial securities, impounding all available information, and infer from the security prices what that underlying information must be. So it is really the CDS markets estimate rather than specifically due to IMF nuts-and-bolts analysis. Since we are all behavourists now, do we still believe in rational inference?

From a nuts-and-bolts perspective the estimate seems on the high end of plausible. With 90 billion book value of bad loan assets falling to 60% of book value gives .4 x 90 = 36 billion loss by the domestic Irish banking sector with current shareholders absorbing 12 billion of the loss and the remaining 24 billion paid for out of Irish government provided equity-like capital injection.

This is all wonderful stuff. But, would someone please explain how an insolvent (state, company or individual) will be able pay down their debts – debts which are legally binding. Are default or bankruptcy the alternative options?

National, corporate or individual income (nett of necessaries) must be +ve in order to pay both principle and interest. But we are currently in a Deflation Spiral – where money is being destroyed – so where is the surplus money to come from to pay down the debt – this is a very urgent matter!!!

Individuals may have contracted mortgages in good faith, so what happens to them when they have insufficient surplus money to pay their debt obligation – and their home (asset) is declining in value, whilst their debt is increasing. Oh I know, re-possess their home and watch whilst it is vandalized to zero value.

Presumably the financial institutions will be willing to either increase or extend credit to persons who will be unable to pay back their borrowings and we will all live happily ever after!

Could someone please explain the real financial, economic and social catastrophe we are headed into instead of conducting specious analyses of questionable financial data.

Many of the citizens of this state are headed for destitution. Hungry people WILL riot!!

Brian P

@Karl, that original link has changed, now reflects, as you suspected the shorter new table.

So just for divilment, here’s the original table for any that missed it..

@Brian P. The options with debt are either pay it or default. There’s no other way out. It just comes down to which is the less painful/harmful route to take.

Paying off the loan can be made easier by having inflation running at a decent clip, providing that loans are at fixed rates. Or by increasing the wealth of the indebted nation/company/individual. Inflation may be possible in the medium term, increased wealth seems even more aspirational.

So that leaves default. But we don’t talk about that, because that’s for when everything else has broken down. When our economic position becomes comparable to situation faced by Captain Hans Langsdorff of the Graf Spee in Montevideo in Dec 1939, then it will be discussed. But while hope/denial prevails, default will not be countenanced.

Two other thoughts occur to me about this market-based estimate (suspending, for the sake of discussion, our lack of confidence in rational inference per Greg)

First point is that these CDS spreads move a lot. The published IMF loss estimate was explicitly based on November spreads. If we used March 2009 spreads the estimate would be much higher–though exactly how much I don’t know as presumably the formula is quite nonlinear.

Second, I am quite curious to know how the IMF disentangle the sovereign credit risk from the bank default risk in working with the bank CDS spreads. So far I have not seen exactly how this is done. Seems to me that, given the guarantee, any methodology of this type can only provide a lower bound to the expected losses of the banks on the grounds that, if the losses are higher, that should not affect the CDS on guaranteed debt unless it affects the sovereign risk.

I’ll keep digging and report any further findings I may discover about the way these estimates were formed

Comments are closed.