Reinhart and Rogoff have a series of working papers (here) analysing the key features of financial crises, using data over many centuries and from around the world. In their papers 2008a and 2008b (here and here) they highlight two common characteristics of banking crises: 1. they are usually followed by severe government deficits, but the relationship is not caused by bank bailout costs; the direct costs of bank bailouts are swamped by recession-related tax shortfalls and expenditure increases, 2. banking crises are sometimes followed by sovereign defaults, but again it is not due to direct costs of bank failures including bailout costs, since these are not typically a substantial part of the government budget crisis.
In another one of this series of papers, 2007 (here) they demonstrate the prevalence and wide ubiquity of sovereign debt crises. They stress that sovereign debt crises are not confined to less developed countries. The small number of sovereign debt crises in developed Western countries since WWII is a typical trough in the cycle, which by historical patterns is due to end soon.
Looking around the world and acknowledging the historical patterns shown in their paper, a hidden message of the paper (in my own subjective reading) is that the Euro zone is due for at least one sovereign debt default in the next decade. Will it be Ireland, and what can be done to prevent such a disaster?
Rossa White of Davy Stockbrokers argues (here) that quote “the risk of Ireland not being able to meet ongoing debt payments over the next few years is very low.” The credit default swap market seems to disagree. The most recent Ireland government bond CDS rate is 3.41% meaning that institutional investor are paying 3.41% to receive the shortfall on par value when and if Irish government bonds default. If the payment shortfall on default is 50% (a typical disastrous sovereign default) this is equivalent to a risk-neutral per-year default probability of 3.41%/.5 = 6.82%. Over five years this aggregates to a cumulative default probability of (1- (1-.0682)^5)=28.09%. Of course investors are not risk neutral so there is a substantial risk premium in the CDS rate, but still this indicates a non-negligible true probability of default reflected in market rates.
Ireland must make policy changes to set the probability of sovereign default close to zero. There is an over-emphasis on dealing with the problems in the banking industry as a solution to the current economic crisis. Going forward, dealing with public finances is at least as important, probably much more important.
The relative ordering of policy priorities is relevant since there is a finite supply of political capital to address the current crisis. If all of the political capital is spent on the least important problems, the most important ones will go unsolved. That increases the medium-term risk of a true disaster: a sovereign debt default. An article in the Irish Times by Stephen Collins, “The Real Danger is in Political Stalemate as the Roof Falls In,” is disquieting reading (here).
The two crucial intermediate policy goals are broadening and deepening the net tax take from households, and lowering the cost of public services. The proposed 10% levy on public service pensions is an admirable first start on the second of these goals. The recent political weakening of the government, in the face of continuing revelations about banking industry shenanigans, has endangered this important first step. Does the present government have enough political capital to face up to raising taxes on middle earners?
Another policy goal is to decrease economic policy uncertainty, so that businesses and households can plan and invest. Mervyn King stresses the importance of economic policy predictability and stability (wittily stated here). He quips that one of his policy goals as head of the Bank of England is to be extremely boring. Presumably this comes easy to him as a well-trained economist. There is also a wealth of evidence that the reason inflation lowers growth is that it creates planning uncertainty for businesses and consumers, see for example Huizinga 1993 (here). Irish businesses and households are not facing price turbulence but rather government policy turbulence, which can be even worse. The Irish government needs to make hard policy choices, state clearly that it intends to enact them fully, and then do so reliably.
21 replies on “Reinhart and Rogoff, and Ireland”
Gregory, could you address the point made by the Department of Finance and others that the CDS markets are driven by speculative investors who are essentially gambling on further increases in this rate rather than by investors trying to genuinely insure themselves and that the CDS market does not accurately reflect default risks. Are there any academics writing seriously on the CDS market? They are currently capturing a very large share of headlines and I am unclear from reading the coverage how seriously they should be taken as reflecting actual default risk. It also looks to me like such markets create pretty weird incentives. A basic google search I spent some time doing revealed mostly horror stories from news organisations (such as the CBS one below) about how they have been implicated in company collapses in the US though much of the coverage seemed one-sided and superficial and very few people in the articles made a strong defence of these instruments. Perhaps someone who believes these are worthwhile markets could make the case here.
A paper that looks at their costs and benefits is linked below. The paper discusses some of the benefits of such instruments such as increased opportunities for hedging and some of the costs including potential moral hazard and a very weak regulatory structure governing these markets.
http://www.cbsnews.com/video/watch/?id=4502673n
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=929747
@Liam
I think that http://baselinescenario.com/2009/02/20/dublin-and-vienna-calling/ suggests better than any of us the issues on CDS’s.
Whats important to remember is that there is an equiblibrium relatiomship between bond spreads and CDS’s . So, saying the CDS spreads are “wrong” is tantamount to saying that the bond spread is incorrect. Now, I am fundementally convinced that the DoF do not understand these instruments. Their statement was internally inconsistent and fundementally flawed. Like any derivative market there is a measure of both speculative and hedging trade in these. But, babies and bathwater…
Like most derivative markets they are very good at giving advance warning of events. See for example the papers below.
Finally, simon Johnson today says
“Even if you are convinced that the CDS market represents pure speculative pressure, i.e., unrelated to “fundamentals”, spreads at this level are still a call for action. In fact, in that case there is no excuse for not putting in place transparent and well-communicated fiscal policies with massive external financial support. That should scare any speculators away.
Of course, if you believe that the CDS market is completely uninformative, there is nothing to worry about. And there was, in retrospect, nothing to worry about when the same market pointed to growing dangers for UK mortgage lenders in fall 2007, US banks in 2007-2008, Iceland in fall 2008, and emerging markets right before the IMF started handing out big loans.”
I would rather err here on the side of caution.
On a technical note Greg, the model you give is a point approximation. The modelled recovery rate on Markit and Reuters is 40%. Extracting implied default rates from CDS curves is very hard – see for example
Cariboni, J., Schoutens, W.
Jumps in intensity models: investigating the performance of Ornstein-Uhlenbeck processes in credit risk modeling
(2008) Metrika, pp. 1-26.
Düllmann, K., Sosinska, A.
Credit default swap prices as risk indicators of listed German banks
(2007) Financial Markets and Portfolio Management, 21 (3), pp. 269-292.
Zhu, H.
An empirical comparison of credit spreads between the bond market and the credit default swap market
(2006) Journal of Financial Services Research, 29 (3), pp. 211-235.
Di Cesare, A.
Do market-based indicators anticipate rating agencies? Evidence for international banks
(2006) Economic Notes, 35 (1), pp. 121-150.
Norden, L., Weber, M.
Informational efficiency of credit default swap and stock markets: The impact of credit rating announcements
(2004) Journal of Banking and Finance, 28 (11), pp. 2813-2843
Thanks for the information Brian. I think the view expressed in the linked article that asking questions about the CDS market implies that one is ignoring the need for fundamental government action on the economy is one that has been floated a lot and I think this is coarse thinking. The reason I want to question the CDS market is not because I dont believe that the Irish government needs to make serious reforms but rather because I really want to know what is driving those markets and what potential independent risks they may place on markets.
We should be able to think about two questions at the same time: (1) how does the government make the maximum possible response to ensure financial stability and generate confidence in the economy? (2) what are the dangers of taking credit default swaps seriously? These are two separate questions and asking the second doesnt mean that you dont believe that the first should be in any way ignored or that any complacency should be allowed to come in.
@liam
What drives these markets is
a)greed
b) fear
c) spread against bond benchmark
d) see a)
e) see b)
There is a danger of NOT taking the message seriously, in all seriousness. What they are sayign is that a significant external constituency has said that the govt is not adressing your first q at all. I leave it to the macroheads here to come up with a plan on that. I think we have the broad outlines of what to do however.
Apologies for the double negatives in my previous comment.
Perhaps someone like Brian and/or another expert in finance could provide further clarity at least on the nature of the types of credit default swaps being held on Irish bonds. For example, is it known how many of the people who hold these instruments own Irish bonds?
The suggestion Brian is that they are currently being used far more as a speculative than a hedging tool. The papers you link to, as you have argued, mostly suggest that these instruments do have an informational value and therefore should be taken seriously as an early warning single. However, surely there must be some distortion generated by a group of investors gambling on the default of a particular bond, if that is what is happening. Does the existence of speculative holdings in these instruments create an incentive to spread rumours about the collapse of a country or company and, if so, how is this regulated? I think we can ask this question and still demand immediate Irish government reform.
In the simplest case credit default swaps are priced exactly by static (that is, no need for continuous trading) arbitrage. Suppose that riskless (eg German) bonds have a time-invariant one-coupon-per-year yield of x% and the risky Irish bond have a yield of x+y%. Suppose also that the Irish bond only defaults at the end of the year and if it does default it immediately gives a partial redemption of P the rest of the bond’s principal and future interest is lost due to default. Assume that this partial redemption amount P is known beforehand. The credit default swap can be created exactly by shorting the Irish bond and using the proceeds to purchase the German bond. Very unrealistic since real arbitrage of the CDS market using cash bonds has to face term structure shifts, uncertainty of the proportional redemption, coupon reinvestment, transactions costs, risk capital costs.
There are CDS – bond arbitrageurs still active, but the decline in credit market liquidity and drying up of bond arbitrage capital has led to larger gaps between risky-riskless bond yield differences and CDS spreads. Incidentally the difference between empirical CDS rates and empirical risky-riskless yield spreads can be negative the CDS rate is not always higher.
There is no reason to treat the CDS implied probability as a true probability there will always be a risk premium so the true probability is lower.
There is also I understand evidence for some delibrate manipulation of CDS spreads, but I do not know the empirical literature or business practice well enough to comment upon it.
@ Liam
As to the speculative/hedge percentage otc markets are hard to penetrate so the answer is we dont know. I dont care as the two markets together give information to each other. Speculation is not a bad thing, imho…. Call it gambling if you wish, I call it speculation.
As to whether these can “cause” in some way the event they predict, I dont know of papers on this nor do I know how one might test (except in some Twin share approach such as rd-shell).
What we do know is that there is always some insider trading . See
Meng, L., Gwilym, O.A., Varas, J.
Volatility transmission among the CDS, equity, and bond markets
(2009) Journal of Fixed Income, 18 (3), pp. 33-46.
who note no real issue versus
Acharya, V.V., Johnson, T.C.
Insider trading in credit derivatives
(2007) Journal of Financial Economics, 84 (1), pp. 110-141.
who say that there is limited importance from insiders.
@Greg : yes, again I am sure that in relativly new and profoundly underregulated markets there is a lot of murky stuff. But, I think we both agree that these instruments can give information.
Firstly Brian, congrats on your great showing in the Westbury today, yourself and Alan did very very well against an audience that would not normally be so open.
With that carrot out of the way, I have a bit of a devil’s advocate question for any that want to take it.
Nouriel Roubini on Bloomberg yesterday singled us out for particular mention when talking about sovereigns that he thinks might ‘crack’.
http://www.bloomberg.com/avp/avp.htm?N=av&T=Roubini%20Says%20Europe's%20Banking%20System%20Faces%20Growing%20Risks&clipSRC=mms://media2.bloomberg.com/cache/vcQ6.Dc03xzM.asf
Now, while I understand that Nouriel is the most dismal of scientists, the question that comes to mind is what are the chances that he will be once more proved right?
Are the chances so small that it is not something that we should even plan for, or is it another elephant trying to squeeze into the room?
“There is also a wealth of evidence that the reason inflation lowers growth is that it creates planning uncertainty for businesses and consumers…”
Wouldn’t this suggest that the problem was unstable inflation rather than inflation itself? I recall a Joseph Stiglitz paper suggesting that there was alarmingly little evidence that inflation (as opposed to volatility in inflation which would obviously create uncertainty as you say) below 10% actually damaged economic performance, and that if such evidence ever did emerge it would be as much a tribute to central bankers’ data mining capacities than anything else!
@James yes you are correct: if all prices increase in lockstep at a known fixed rate then in theory inflation does not cause planning uncertainty. Also my claim about policy turbulence having the same type of effect is a bit unsupported by evidence etc. but I believe it to be true but cannot give references. Yes you are correct about the Stiglitz paper but frankly I was not convinced by that paper and I do not believe that it gives a consensus view. Sometimes these counter-intuitive empirical papers just come down to vagaries of the data choice, or sample period, or estimation technique. The broad evidence indicates inflation hurts growth substantially; that is the consensus view.
Meanwhile, €10b or 3% of deposits went on holidays last week. Thats 5% of GNP. Any ideas on how the markets will respond to this?
@Brian: Two things: do you think that the conventionally modeled recovery rate of 40% is low for sovereigns? It does not matter when derivative pricing using the risk-neutral probability since only risk-neutral probability of default x recov rate is identified not the two components. It does matter when inferring the physical not risk neutral probability of default. Two: what is the new info on deposit outflows that you reference I was not aware of it?
@Liam Delaney Thanks for the link to the CBS news report on credit markets. I thought that they should have given at least a few seconds to the enormous increase in sub-prime mortgage generation by Fannie Mae. Fannie Mae and its American Dream Commitment to issue $2 trillion in sub-prime mortgages is less “sexy” as a media villain but they were a major player in the catastrophe. They make a good point in the broadcast about the importance of CDS positions in the failure of Wall Street investment banks.
yes, my general impression from looking for information on CDS is that they are demonised in the American media and you have to look far to find a balanced account of the costs and benefits of this market – the CBS news story is pretty much representative. Indeed it looks from recent news reports that at least some of these markets are going to be hit with heavy regulations in the US.
You certainly wont find an argument from me about the need to keep property markets in focus in the discussion about what has gone wrong in America and also not to forget the role of our particular breed of property insanity in putting Ireland in the current position. I am currently writing a review of Akerlof and Shiller’s new book “Animal Spirits”. Among the chapters include “The Current Financial Crisis: What is to be done?”; “Why do Real Estate Markets go through Cycles?” and “Why are Financial Prices and Corporate Investments so Volatile?”. I dont know whether its a good or bad thing that they make the answers seem so obvious.
@Greg
I think so, 40% is too low. Of course, as the recovery rate rises so to monotonically does the implied default : 350/(1-0.4) is 583 bp while a recovery rate of 75%gives us 1400bp.
As for the outflows : http://www.independent.ie/national-news/scandal-sees-836410bn-flood-out-of-country-1649052.html
Now, I know its the indo but there is a quote from finance there which hasnt been retracted. I was also aware thu that there were queries on this issue going around, so it wasnt a saturday makeeuppee.
Bad, no? 10b is 3% of the deposit base of the resident banks and what, 5% og GDP. Is this is a crisis yet?
I am an economic neophyte (is that some sort of weird oxymoron?) who considers himself to be reasonably intelligent and has been (like some of the contributors on here) seeking a sane set of arguments as to what might and should be done to get us out of the situation we are in (assuming we knew what that is/was….)
I have always considered the ‘instruments’ of various kinds as speculative; and take the view that the present situation is 90% speculative, but mis-read by non-specialists and, more unfortunately, by the media. My gut feeling is that a lot of the movements are over reactions to small levels of actual market activity, and the movements are made in many cases by ‘traders’ who are looking for shorter term gains to substitute for the lack of much else going in their usual places of activity.
Is what is being said in this thread an indication that our DoF mandarins are to be counted in the non-specialists? If so, what of your colleagues who work there, are they un-thinking?
Bye, Barry
Apropos – “Firstly Brian, congrats on your great showing in the Westbury today, yourself and Alan did very very well against an audience that would not normally be so open.”
Can some one point me to the papers please?? I’ve read the newspapers take but want to read the truth 😉
@barry: The CDS rate is a betting rate for a bond default, but it has a risk premium in it of unknown magnitude. Financial economists call it “rational inference” when one tries to infer information by looking at market prices and the information implicitly impounded in them (this requires that markets are rational and have good information). The Irish Times in today’s editorial (Tuesday) had it correct — the CDS implied default probability is a “crude measure” but it should not be ignored entirely. Brian Lucey pointed out that I could have used the standard assumption of 40% recovery rate rather than 50% which lowers the implied probability a bit. The CDS market has been widely abused in the USA but it can be a useful instrument if used properly.
@Gregory O’Connor thanks for that clarification.
The nub is ‘good information’ is it not? If the CDS rate is used as a hedge/gamble tool/instrument then isn’t there an inbuilt risk element of how much information the different participants in the market have?
I hate to use a gambling analogy but isn’t the movement of CDS akin to the movement in the ‘ring’ when certain ‘well connected’ people place bets??
The Irish Times editorial is a good pointer though, takes some of the ehat out of the general media comment.
Hi,
I stumbled upon this conversation from the website propertypin.com.
Coming from a background of modeling cds, pricing them as well as looking to discover arbitrage opportunities, I find that there seems to be allot of confusion regarding their intricacies as well the fact that many assumptions are made to make pricing easier (such as the recovery rate).
There is an inexorable link between underlying bond yields and the related CDS spread as well as the connected family of yield curves.
For example, Irish govt bonds are linked to several yield curves, such as benchmark curves, such as the German bund curve, the basic libor swap curve and similar subset of sovereign bonds, (i.e. yield curve constructed from bonds of similar European countries such as Spain, Italy and other Latin blocs that display more similarities than the total family of European countries.
But CDS curves also factor in other elements, such as
• Probability of technical default
• Probability of debt restructuring
• Liquidity costs, CDS are cheaper to enter than outright positions
• Perception of needs of upcoming debt issuance by each country.
• Direction of fiscal policy….
Amongst other interrelated elements…..
With regard to recovery rates:
The assumed recovery rate on sovereign bonds is 40% in Europe, as per markit, the market source in pricing information. I agree that this number is too low. But this is the number being used in the pricing models of the various i.banks offering a market in these derivatives. In the US, many states trade at assumed recovery rates of 80%. For example, both NJ and NY trade at 80%. It is impossible to countenance the argument that these offer a higher credit than Germany, but c’est la vie.
Also, as mentioned by a previous posted comment, recovery rates are hard to extract, impossible without leaving the spreads constant. So the market standard is to assume a constant recovery rate and display movements in credit risk movements entirely through the spread movement. Of course this is not the case….
Another element that clouds the discussion is the fact that quite a large number of the newer CDS that have been traded are exotic in nature, with types such as binary cds, cds with various different optionality and CDS’ linked to CDOs.
All in all, CDS are a useful instrument but one that currently, has too many unknowns in place. Also, at the moment, there is a very large speculative element driving spreads as certain investors are looking to place large bets. This has a very large effect on the spread movements, as the investment banks selling the protection are pricing their CDS so as to deter business without actually holding up their hands and saying so. This again, is another market perception ploy to reduce their risk without telling the market what their position or view point is.
Thoughts??
@Noel: Very interesting comments thanks for contributing. We needed a bit of Street Knowledge on some of these issues. I accept your point that the investment banks who normally would provide liquidity to this market have very stressed levels of risk capital and this can cause the CDS implied probabilities to be uninformative. That supports the Irish Dept of Finance which are stressing that same notion. Still it does seem a worrying sign that the demand for this Irish-sovereign-risk insurance is so high to give these high rates. Regarding the recovery rate, it does not matter in your derivatives pricing in many cases as you state since often only risk-neutral probability times recovery is empirically identified by the derivatives pricing model. However from the sake of the Irish government worrying about the probability of default the physical probability of default matters most, not the recovery rate afterwards.
Regarding all the complexities needed in pricing, how good are the simple approximations for getting risk-neutral default probability from the CDS given a assumed recovery rate? Brian Lucey seemed to side with you that my approximation was too simplistic?
“How good are the appoximations”
Not good at all unfortunately.
Below are the current numbers, I pulled these off Bloomberg quickly but the error would be minimal in small bps.
5yr Govt bond yield Spread over Germany Spread over Euribor Curve
Greece 4.91% 2.73% 2.11%
Ireland 4.67% 2.49% 1.88%
Austria 3.23% 1.04% 0.43%
Germany 2.18% 0.00% -0.61%
CDS spread on 5yr credit
Greece 2.70%
Ireland 3.88%
Austria 2.69%
Germany 0.92%
Simple rough risk covered yield
Greece 2.21%
Ireland 0.79%
Austria 0.54%
Germany 1.26%
5yr swap rate 2.79%
So, buying a bond and shorting the credit on the same maturity yields strange results.
Germany offers a yield almost twice as good as that of Ireland. What does this indicate?
Either the bonds are under or the cds are over or a combination of both. Personally, I believe that the CDS spreads are pushed further than what a risk neutral valuation would indicate. This would make me think that the market is looking to go long Irish credit in the CDS market to hopefully make money on the widening of the spreads. This would indicate a speculative stance as per the DoF.
Of course, we also need to look at the bond side. The bond market pricing may be somewhat stale. Banks with holdings in a particular bond or family of bonds may not want to sell a bond as it may trigger a mtm movement on the remainder of their portfolio. So in essence, the “skin” in the market may be rather thin at the moment, so perhaps our bond yield should be somewhat higher.
The only thing that should matter to the Irish govt is the cost of borrowing. As my rather simple tables above show, the bond yields are under where they should be for a risk neutral position if we were to take the CDS as an accurate measure of Irish credit risk. (am ignoring cds counterparty risks, CSA costs etc). This would lead me to believe that there is an element of truth to what the DOF claim.
I just hope, that it’s the cds that move in to meet the bonds rather than the other way around.
excuse the table formating, seems excel isn’t so good here.