November 18, 2007
Iraqi Bonds and the Effects of the Surge
Iraqi Bonds and the Effects of the Surge-Becker
Michael Greenstone of MIT had the excellent idea of using financial data to provide information on what global investors believe about the viability of this government. In an unpublished article revised on September 18th of this year, he uses data on movements over time in the yields on Iraqi bonds issued in 2006 by the present Iraq government to assess what investors believed about the prospects for success of the "surge" in American military personnel that began in mid-February, 2007. He shows that yields remained at about 9.6 percent from the beginning of the surge in February to midsummer, then climbed to 11.5 percent in August, and was at that level when he finished his study. What can one make out of this analysis?
The bonds were issued with a 5.8 percent annual yield, which seems too low, given the high risks of a default by this government, and that high yield bonds (junk bonds) typically provide considerably higher yields. So not surprisingly, these Iraqi bonds have been priced in the marketplace at a considerable discount in order to yield to buyers returns that are much higher than 6 percent if the government manages to pay both the interest and principal (due in 2028).
I agree with Greenstone and Posner that prices of these bonds offer a valuable way to determine expectations about the stability of the Iraqi government held by the savvy investors in the international bond market who are placing substantial financial resources at risk. This does not mean that these investors are never wrong, or do not change their views as the evidence unfolds, but rather that bonds prices offers relevant information about the assessments of Iraq's future by persons who have an important financial stake in whether they are right or not.
The additional evidence available since Greenstone's September study provides a more optimistic assessment than at that time of how the surge is going. Both American military and civilian casualties are way down during the past two-three months-Greenstone also refers to such data up until the end of August- and civilian life in Baghdad has returned to a semblance of normality for the first times in a few years. Explosions, mortar attacks in Baghdad, and bombings all declined by a lot in recent months. As a result, Iraq bond prices have also rallied significantly, so that their yields are down to about 10.5 percent recently. This is still about 10 percent above yields in February, but the additional premium has been reduced from almost 2 to a little less than 1 percentage point.
Several commentators have emphasized that the steep decline in Iraqi bond prices and corresponding increase in yields that began in August coincided with the beginning of the credit crunch, which significantly raised interest rates on all bonds. The crunch especially raised rates on riskier bonds, so it is not surprising that highly risky bonds like those issued by the government of Iraq should have fallen greatly in value starting in August. To correct for this, it would be valuable to compare Iraq bonds with bonds of comparable risk. Some traders have suggested comparisons with bonds issued by the government of Lebanon since that government's stability is also highly uncertain. Apparently, Iraqi bonds have outperformed Lebanon's since the beginning of the surge.
This would suggest that the surge may be having a positive effect on investor's expectations about the viability of the Iraqi government, at least relative to the viability of other governments with questionable stability in that most unstable region. However, Greenstone in his paper, and in some updated calculations of his that he sent me, prefers to use not a single country's data (a single country could be biased by choice of country), but an index of emerging market bonds. The gap between Iraqi yields and emerging market yields did decline noticeably from September, but the gap is still much larger than in February.
Another consideration relates to the ability of bond prices to accurately reflect what is happening to risk. Suppose investors and others believe that the surge has greatly improved the average financial and other prospects of Iraq, including the stability of its government. However, it is plausible that the riskiness of these outcomes has increased because of growing uncertainty about America's commitment to continue its involvement, and the greater chaos that might follow if the surge failed. On this view, the surge could have improved by a lot the average outcome expected for Iraq in the future, while at the same time it would have increased the likelihood that the government would fail, and that bonds would go into default. Since the best that bondholders can do is receive interest and principal, while the worse is default on either or both these classes of payments, any increased risk produced by the surge would lower bond prices and raise yields even if expected outcomes greatly improved.
A more appropriate way to assess the effect of the surge on expectations about the economy, and presumably indirectly about the stability of the government, would be to examine changes in valuations on the nascent Iraq stock exchange. Greenstone discusses trying to do that, but gave up because few stocks are traded, and trades are highly irregular. Since the market is very thin, it is hard to reach any strong conclusions, but I have the impression from a few reports that prices of stocks on the Iraq stock exchange have risen in recent months.
Whatever the final conclusions about the evidence on the political future of Iraq provided by its bonds, financial markets are an underutilized source of information about the expectations of investors about political outcomes. To be sure, financial expectations can be very wrong. For example, Eugene Lerner has shown that the Confederate currency did not depreciate very rapidly (relative to the growth of the money supply) until only a few months before the end of the Civil War, even though historians are unanimous that the South had effectively lost the war long before that. Still, I generally would have more confidence in the accuracy of the expectations of persons with a serious financial stake in outcomes than in the forecasts of most others who express their views on future political outcomes.
Is the Bond Market the Best Predictor of the Outcome of a War?--Posner
An article in the business section of the New York Times last Sunday (November 11) by economist Austan Goolsbee, summarized an academic paper by an MIT economist named Michael Greenstone that uses Iraqi government bond prices to estimate the bond market's response to the Bush Administration's "surge." Greenstone's paper, available from the Social Science Research Network, is dated September 18, which is two months ago, and perhaps developments since would alter his conclusions. To estimate default risk from the bond's current trading price, and in particular to estimate the effect of the surge on the default risk, is not straightforward. For example, Greenstone adjusts for the probability that a Democrat will be elected President next year (regarded as increasing the likelihood that Iraq will default on the bonds); that probability may have changed since September 18. The worldwide credit crunch has worsened since then, and that might have an effect, independent of the surge, on the price of the Iraqi bonds.
The bonds ($3 billion worth issued in January 2006), which mature in 2028 and until then pay 2.9 percent on their face value twice a year, so almost 6 percent per annum, are trading at a steep discount (currently about a 40 percent discount, which jacks up the yield to almost 10 percent--almost $6 for a bond that costs $60). This means that purchasers of the bonds (which are actively traded) are demanding compensation for bearing a substantial risk of default. The most interesting conclusion in Greenstone's study is that, after correction for other factors, the surge is correlated with a 40 percent increase in the bond market's estimate of default.
It seems unlikely that the surge itself would increase the risk of default, though it might, by enabling both the Sunnis and the Shiites to rest and augment their forces for the eventual showdown, taking advantage of a kind of truce imposed by the additional American troops. More likely, the bond traders see the surge as a desperate last gamble by the United States; as a preclude to U.S. withdrawal and specifically as a sign that the United States will withdraw soon after the next Presidential election, whoever wins the election; as a political gimmick; and as a failure in the aim of the surge of promoting progress toward a political settlement that will enable Iraq to be a functioning nation when we leave. If, as the bond traders fear, Iraq is likely to be divided well before 2028 into three separate nations (Kurdish, Sunni, and Shiite), a default is likely.
There are two general questions that Greenstone's interesting study raises. The first is the relation between default risk and U.S. failure. For comparison, consider another recent study, by Kim Oosterlinck and Marc Weidenmeir, this one of the price of Confederate bonds in the Amsterdam market during the American Civil War. Initially the bond prices traded at a discount that indicated that the Confederacy had a 42 percent chance of winning the war and therefore presumably of repaying the bonds; but with the crushing twin defeats of the Confederacy at Gettysburg and Vicksburg in the summer of 1863, the bond market quickly downgraded the Confederacy's chances of winning the war to only 15 percent, and the estimate kept falling till the end of the war. The difference between that case (and other examples of using bond prices to predict the course of a war) and the Iraq case is that the risk of the Confederacy's defaulting on its bonds depended essentially on just one event, namely whether the Confederacy--the issuer of the bonds--lost the war. In the case of Iraq, the relation of the risk of default to the outcome of the war is obscure. No one thinks that the United States can actually be defeated by Sunni insurgents, al Qaeda in Mesopotamia, Shiite militias, Iranian infiltrators, or any other armed groups in Iraq or the surrounding areas. At the same time, few believe that the United States can win the war in the sense of eliminating widespread violence and coupling withdrawal with handing over control of the country to a functioning government, as the United States was able to do several years after the end of World War II in Germany and Japan. Moreover, if we wanted to avoid a default, we could do so simply by buying up the bonds (at the current discount, they would cost only $1.8 billion, provided they were bought up surreptitiously so as not to force up the price significantly) and then forgiving the debt. And finally, one can imagine a scenario in which American policy is an utter failure, but the utter failure actually reduces default risk. Suppose we pull out of Iraq and Iran takes over. Iran might decide to pay off the bonds in full (the amount of money is small) in order to increase its credit standing.
So really the only (though major) significance of Greenstone's bond market study, so far as our situation in Iraq is concerned, is that it is evidence that the surge, while it has reduced the number of deaths in Iraq, has not increased the viability of the Iraqi state, but instead has revealed (possibly even contributed to the prospect revealed) that the attainment of viability is increasingly unlikely.
The second general question raised by Greenstone's paper is whether financial markets are better predictors of the outcome of wars and other political crises than experts are, including the experts who staff intelligence agencies. One might think that experts would be better predictors because they had specialized knowledge that bond traders would lack and that experts who work for intelligence services would have not only expert knowledge but knowledge that bond traders could not obtain by consulting experts, because it would be classified knowledge. A careful historical study (and access to classified information, at least for recent crises) would be required to answer the question which predictor is better. But it would not be surprising if the financial markets turned out to do better than the experts, including national security personnel. Financial markets aggregate the opinions of a vast number of investors, and those investors who at least think that they have real insight tend to be the ones who determine the prices in those markets.
Friedrich Hayek's great legacy to economics was to show that the price system can aggregate vast amounts of information much more efficiently than a centralized bureaucracy can do. And intelligence agencies are centralized bureaucracies. The innumerable mistakes that the United States has made in Iraq suggest that our government does not have good means of obtaining and evaluating information concerning that country, possibly because of a combination of bureaucratic inefficiency and the vastness of the quantity of relevant data. The people who trade Iraq government bonds do so not because they are told to study Iraq or paid a salary to do so or have an academic or journalistic interest in the country, but because they hope to make money. Presumably therefore they are self-selected for knowing a lot about Iraq—and for thinking they know enough to put their money where their mouth is. They may be right.