Iraqi government bonds

by Daniel on September 18, 2007

Via Marginal Revolution and the Freakonomics blog, Michael Greenstone of MIT’s analysis of developments in Iraq since the beginning of the surge. Most of it is pretty unobjectionable stuff (if not terribly related to economics), but the bit that caught my eye was the use of the secondary market price of Iraqi government bonds (they’re tanking) as an indicator of whether the surge is working.

IRAQI EMPLOYEES CAMPAIGN UPDATE: sorry to interrupt – the post is continued below the fold. The following update is mainly aimed at our British readers – once more could I prevail on the goodwill of other CT authors to keep this one at the top for the rest of UK daytime? Thanks.

Frank Dobson, my local MP, has replied to me about the Iraqi employees. His letter is mainly concerned with administrative arrangements for the speaker meeting on the 9th – I had asked him to book a room for us, but Dan had already sorted one out. He is, however, in firm sympathy with the campaign and is utterly sound on the issue. If any readers who wrote letters the last time we asked have received replies from their MPs, then do please say so in the comments here (or if you have a blog, post them there). We’re trying to keep the list of MP replies up to date; Dan’s apparently having quite a good response to the mailing we sent out but it’s important to keep the blog campaign running too. Thanks very much.

As longtime readers will know, I have a lot of prejudices against this sort of analysis. First, because my hackles rise like a grumpy porcupine when someone says “What do the data have to say about this question?” when they mean “What do I have to say about this[1] data?”. And second, because I have specific and serious issues about any attempt to read broad conclusions about the world off securities prices.

There is of course a lot of information in bond prices; they’re very informative indeed about how much it costs to buy bonds. Any further conclusions are extrapolation, and an extrapolation has to be justified with separate reasons for believing it to be valid. And there are at least two problems with the extrapolation from “Iraqi government bond prices have fallen” to “therefore things are going worse in Iraq”.

First, there have rather been one or two other things going on in the bond market over the last couple of months, as a casual reader of the newspapers might have noticed. The first implicit assumption you have to make when you’re extrapolating from market prices is that you’re dealing with a reasonably efficient market, at or near equilibrium. If this condition isn’t satisfied (and in my opinion in the current case it isn’t even close), then all you’re really picking up from the price movements is information about internal conditions in the relevant market.

Second, I don’t think that even in an efficient market there can be a one-to-one mapping between the prospects of the Iraqi state and the credit spread on their bonds. It’s true that if Iraq becomes a successful and healthy democracy the bonds are much more likely to be repaid in full, and if it descends into anarchy they’re much more likely to be completely worthless. However, that doesn’t entitle one to assume a linear relationship of the sort that we would need here. For example, if I was minded to I might argue that the opposite relationship could hold locally – that as bonds went down, they were telling us that the Iraqi state was becoming strong enough to get rid of the Americans and negotiate a bond restructuring, while rising bond prices meant that Iraq would continue being an American protectorate for the foreseeable future, and thus the bonds would benefit from an implicit US guarantee.

As I say, the rest of Greenstone’s paper is pretty good. But in a way, that makes it more irritating that the “financial markets say …” bit is given such prominence. If you’ve taken the trouble to do a whole lot of good analysis yourself, why on earth would you want to then bow down to the household god of the last few developments in a semi-related securities market?

[1] yes, this data.



Barry 09.18.07 at 2:07 pm

“First, there have rather been one or two other things going on in the bond market over the last couple of months, as a casual reader of the newspapers might have noticed. ”

Daniel, he uses the spread between Iraqi government bonds and some other high-risk bonds.

From page 22 of the PDF: “Prices and Yields of Iraqi Bonds. With this background, I now turn to a comparison of how the difference or spread between the yields of the Iraqi state bonds and three alternative bonds changes after the Surge began. Each of the alternative bonds is intended to serve as a control to avoid confounding the impact of the Surge with other drivers of default risk or shifts in the yield curve.”

He then lists the ‘controls’.


dsquared 09.18.07 at 2:08 pm

I know, but how convincing is this? All sorts of strange spreads have opened up between securities which had previously been excellent “controls” for one another.


RWB 09.18.07 at 2:22 pm

I haven’t read the paper in question (but I will! it sounds fascinating), but this statement intrigued me:

“Second, I don’t think that even in an efficient market there can be a one-to-one mapping between the prospects of the Iraqi state and the credit spread on their bonds. It’s true that if Iraq becomes a successful and healthy democracy the bonds are much more likely to be repaid in full, and if it descends into anarchy they’re much more likely to be completely worthless.”

Obviously there can’t be a “one-to-one” mapping, but there can be a “stochastic mapping,” if you will. As you state, there appear to be a distribution of outcomes for this bond, from default to full repayment. The current yield is the market’s estimation of where the bulge in that distribution curve lies. That price has to say something about what the investors believe about volatility and, since this bond is backed by the Iraqi government, the prospects of the government’s continued success, at least on economic issues.

The problem with using the yield (or the current price of any asset) to predict the future is that the yield tells us what investors feel right now. But as we can see, any piece of news at any moment can move the yield. So even if you can make the leap and say that the yield on Iraqi bonds is implying a lack of confidence in the surge on the part of investors, so what? That all might change tomorrow.


aaron 09.18.07 at 2:47 pm

Problem is that it’s not likely and indicator of the effectiveness of the surge, it’s an indicator of the effectiveness of the Iraqi government. Two very different things.

If Iraqs current governement tanks, is that necessarily a bad thing. I personally wouldn’t mind seeing them give that whole constitution thing another go.


abb1 09.18.07 at 3:10 pm

…extrapolation from “Iraqi government bond prices have fallen” to “therefore things are going worse in Iraq”

Maybe it should be “therefore things will be going worse”. Because typically when your interest rates are going up fast it’s not likely that you are on the path to great stability and prosperity.


matt wilbert 09.18.07 at 3:16 pm

He does use controls, but they are (in my view) prima-facie inappropriate to control for the market conditions of the past two months because they are all significantly higher quality (as judged by their yields which are 3.5-8% lower prior to the surge) so not as subject to the recent issues in the market for lesser credits. The spike in the spreads in the July-August timeframe is very similar to the spike in high-yield spreads generally.

It is clear from the paper (pg 19) that even with knowledge of the unusual market conditions Greenstone thinks his chosen controls are appropriate, but I can’t understand why he thinks that–he doesn’t explain why he thinks so, he just asserts it.


dsquared 09.18.07 at 3:35 pm

RWB: no. It can’t be simply asserted that the relationship between the value of the bonds and Iraqi government success has a stable and known sign. As I noted above, one can argue just as convincingly that bondholders benefit from continuing chaos which keeps the Americans there and suffer from improvement which might lead to the Iraqi government having enough stability to declare a moratorium. Also, there are plenty of risk factors in Iraqi bond prices which aren’t related to risk on the Iraqi economy and which I do not think the Lehman EMBI (which is mainly an index of Latam and Central European bonds) proxies for terribly well.

You can actually see from the charts at the end of the piece that apparently, the financial market suddenly decided the surge wasn’t working over the space of about a week around the 15th of July, which week superimposes more or less perfectly on your favourite chart of the subprime mortgage crisis.


aaron 09.18.07 at 5:01 pm

Popular opinion overseas and US politics are also likely factors.

So are rising interest rates in general (are they, I haven’t looked), since it makes it more likely that the US would be less willing finance using debt and a tax hike beyond expiration of previous cuts probably isn’t likely. When the war started, people were practically lending money for free, taking it was a no brainer.


MattF 09.18.07 at 6:24 pm

I’ve always thought that the plural of ‘data’ is ‘dati’.


Warren Terra 09.18.07 at 7:59 pm

I’ve always thought that the plural of ‘data’ is ‘dati’.

Please clarify whether this is intended as humor?

(In a related vein, I’ve always thought that ‘the plural form of anecdote is data‘ was cute, if untrue).


Justin 09.18.07 at 8:06 pm

I think the mistake you make here is that you make the assumption that markets work, and then deconstruct from there. But markets are hardly perfect – the price indicates the market’s aggregate beliefs. And the idea that the market’s aggregate belief would be that Iraqi bonds are worth more under a functioning Iraq than a dysfunctional one seems reasonable.

The real question is that these are the market’s beliefs – they can be led by “experts” and followed by the crowd ,or they can be random guesses. But there doesn’t seem to be any reason to believe that the experts could peg this one any better than the rest of us – so their prediction via bond prices isn’t worth much – and that’s the real criticism against using bond prices.


Justin 09.18.07 at 8:08 pm

Nevermind on my previous post – I skipped too many paragraphs when I made the jump.


MattF 09.18.07 at 8:15 pm

Yes, it’s humor– although I did take the precaution of googling ‘dati’ just to make sure that it’s not a word.


bi 09.18.07 at 8:23 pm

MattF, Warren Terra:

Of course, the plural has to be… “datae”! :B :B :B :B :B :B :B :B

Well, certainly the plural of “virus” is not “viri”. Or “virii”. (Maybe “vīrūs” or “vīrora”, though I’ll stick with “viruses”. But “virus” is a mass noun in Latin and a countable noun in English, the opposite situation of that for ‘singular’ “data”.)

And for frig’s sake, a bloke who participates in a jihad is not a “jihadi”. Why people can’t just say “jihadist” or “mujahidin” just boggles my mind.


bi 09.18.07 at 8:30 pm



Felix 09.18.07 at 9:19 pm

D —

You may or may not be aware that what he’s using IS restructured debt: the US govt basically imposed a whopping great big haircut on all Iraq’s Saddam-era creditors, and the bonds now floating around the capital markets are the end result of that restructuring. At no time since the 2003 invasion has Iraq been anywhere near the point at which it could actually issue new bonds. And frankly if the secondary-market price of a bond is not equal to the price at which that credit can issue new paper, then a lot of the assumptions which go into these kind of papers have to get thrown out the window.


John Quiggin 09.18.07 at 9:43 pm

Whether or not this particular application is valid, the general point that the market value of government bonds is likely to decline when the government looks likely to fall is neither new nor controversial. Obviously markets are going to react to publicly available information like this, whether or not they do so optimally.

The silly claim to which this stuff was linked is that, therefore, markets in events like terrorism futures, where most of the relevant info is private, are going to do well.


David 09.19.07 at 3:10 am

Parallel may be that going into Iraq as the US did was not unlike some of the common subprime mortgage arrangements that were being made by homebuyers or home refinancers at about the same time — easily affordable payments early on, maybe even with some negative amortization for the first couple of years. It’s tempting during that early period to rub hands gleefully, open a bottle, and say, “mission accomplished.” But eventually the piper comes calling to be paid, the payments begin to adjust, and reality sets in.

Comments on this entry are closed.