Still Crunching The Data On The Surge: On “mixed results” and Iraqi Bonds-UPDATED
Lance on Sep 19 2007 at 6:57 am | Filed under: Domestic Politics, Economics, Foreign affairs, Lance's Page, Notes on the war
Economist Michael Greenstone has done an admirable, even handed and statistically rigorous analysis of available data on the trends in Iraq since the beginning of the surge. His most important conclusions:
- Civilian casualties have declined. The decline is not only substantial, but statistically significant. It has been not only true in Baghdad, but it has not led to an increase in the rest of Iraq. So much for the violence merely moving (though that has undoubtedly occurred to some degree.
- Despite an increase in active operations and with more troops involved coalition and Iraqi Security Force casualties have not increased, though total numbers have not declined. Some areas such as non fatal injuries have declined.
- Oil production has not increased and may have declined. Electricity may have become more widely available, though not in Baghdad.
- While Iraqi Security Forces have continued to increase, they have increased at a slower rate.
The analysis shows results which confirm the overall assessment of the data underlying Petraeus and Crockers testimony. No cooking of the books can be seen in this assessment. Number 4 above may also not be the negative it seems at first glance. Some of the decline in the growth rate may be due to increased vetting of Iraqi forces. A great deal of effort has been put into eliminating sectarian influence since the beginning of the surge. He doesn’t address this possibility, but it may be worth having him follow up. This assessment is hurried by in many hands, such as Kevin Drum’s, as “inconclusive.” Actually the word was mixed in the paper. They are mixed, and certainly are not conclusive, but the trends are overall encouraging and similar to what we have been hearing on the ground and from the military.
Of less interest to me is his analysis of the Iraqi bond market. Bond prices have a pretty good predictive record, so why my skepticism? First his results. They show a rapid rise in the yields of Iraqi bonds. Not only absolutely, but more analytically important, relative to other bonds. My rationale:
- James Hamilton argues that the reason for the spread may be an assessment that regardless of Iraq’s prospects overall, the bonds could be showing a reshuffling of the manner in which that prosperity may manifest itself. Iraq’s prospects may still be same or better, but the central government may be less likely to profit from those prospects.
- The difference seems huge when expressed as a 40% increase in the implied rate of default and when illustrated with this graph:
But it only means the implied default rate has moved from 6% per year to 8%. As an assessment it is negative on its face, but hardly disastrous. More importantly, essentially all risky bonds have seen a substantial increase in yields during this time period. - “But they also saw the increase in relative terms, even against emerging market bonds.” He did control for that, but I will gently and sympathetically fault his methodology and suggest there may be no way to correct for this. He even suggests he knows this is a problem. One of the issues with riskier debt holdings has been that spreads, not only between broad categories of risk such as developed market bonds and emerging debt have been abnormally tight, but within those broader asset classes as well. Iraq’s spreads have even been used in the past as an example of the absurdity of these tight spreads. Iraqi bonds (much like much subprime debt) never deserved the low yields/high prices they were garnering. Notice the big rise occurred at the same time as the credit markets began making much more individualized assessments of risk across the world of debt. So I see this as most likely being a reassertion of more appropriate risk spreads between and within broader aggregates of credit instruments. (Which is why Kevin Drums argument: “The Iraqi bonds suffered even compared to other risky government bonds” does not necessarily imply his “Bottom line: the decline in Iraqi bond prices appears to be genuinely related to events in Iraq, not just events in the global credit markets.“) To assess this would require an index of similar risky bonds and how they behaved, and since the relative risk of Iraqi debt is precisely the issue at hand I am not sure how he could have controlled for that in any satisfying way. I also should address another point of Kevin’s: “First, the price of Iraqi bonds began to plunge before the subprime meltdown.” Uh, not really. The effect he is describing began well before the middle of July, that is just when the flames raged most intensely. So, unless those issues are addressed, I see to much noise to feel confident that the markets perceptions of credit risk are related to the surge at all.
- Even assuming that issue was moot, the credit markets could be reflecting broader concerns about Iraq unconnected with the surge. He does a good job of testing that on some discrete factors, but the likelihood, for example, that the surge may be abandoned may account for it irrespective of the factors he tests. Combined with a more realistic appraisal of risk in general discussed in the previous paragraph the spread may be quite reasonable even if the surge is working spectacularly well.
- Finally, as readers here are aware I have been arguing for some time that risk has been mispriced for some time across and within markets generally. If generally market predictions are superior to most individual predictions, the vary volatility we have witnessed recently illustrates the markets have not been pricing risk efficiently and thus our confidence in their ability to assess relative risk should be somewhat tempered in any particular group of securities at any particular point in time, even if overall and over time they do a pretty good job.
I feel number three is the major factor and would appreciate someone with more time than me taking a more in depth look at how the credit disruptions of the last few months could or could not account for the change.
Other thoughts at Marginal Revolution, Freakonomics and from Mark Thoma.
Update: Mark Thoma has some thoughtful responses to James Hamilton’s post. I will reiterate my point on his use of the term “mixed.” I will also repeat that the data quality questions while real, do not alter that both on the ground observation, and through independent assessment of data, the evidence overwhelmingly points to success in reducing civilian casualties, and more importantly, day to day violence, at this point.
Finally, Mark brings up a point I have brought up before, that some of the decrease is likely due to the cleansing of many neighborhoods. That is undoubtedly true. However, a possible conclusion which I have not posted on yet has some bearing here. My own guess given what we do know about the extent of the still remaining opportunities available for sectarian violence, is that absent the US forces and the change in tactics violence would not have declined, and probably would have continued to escalate. What the cleansing accomplished has made it easier for the coalition forces to reduce the violence due to the greater ease in keeping rival groups apart. It does not necessarily follow that the violence was due to fall anyway. In fact, given the desires of the extremist groups carrying it out they had far more work to do on a far grander scale. On the ground reporting such as this also shows that mixed communities still exist as well, and sunni-shiite cooperation is starting to occur.
Sphere: Related Content7 Responses to “Still Crunching The Data On The Surge: On “mixed results” and Iraqi Bonds-UPDATED”
Trackback URI | Comments RSS
The View From the Bluff: Lance’s Economic and Finance Blog
I gave the survey a fair scan.
The calculus of the civ cas calculations aren’t 100% clear to me, but that’s my fault, not the paper’s. Don’t want to spend the time to mentally walk through that.
However, I note that the data sources are Iraqbodycount.org. Those are honest people, but as they’ve said themselves, their archives almost inevitably understate casualties – and they’ve also said that they specifically understate casualties *more* for recent data. (I could probably find that cite, if you forced me). The surge is, of course, recent data, so it’s suspect.
The second reason one has to suspect is this graph . The pentagon’s own civ casualty reporting doesn’t square much with the data points in here. Am I saying I trust the Pentagon, now that it’s got the most pessimistic data set? No, I don’t, although I read its institutional incentives as leading to underreporting, not overreporting. Who knows, maybe Admiral Fallon hates Petraeus so much, he’s willing to push those Pentagon numbers up.
On the other hand, maybe if we drew a regression line for the Pentagon, we’d still see a trend break from pre-surge. I’m not completely discounting the possibility that adding more US troops reduces civilian casualties. It’s certainly a valid idea
Does this ressurect Petraeus’ reputation, because an independent study confirms his overall conclusion? Not really. The monkey business with the statistics had a variety of origins and conclusions, such as the silly claim that “sectarian violence had declined in Baghdad by 75%”. I thought Petraeus’s actual civ casualty graph from the presentation was more realistic – and less promising – than I expected, frankly. But we still have data sources that disagree with each other, and casualty estimates that didn’t start diverging until the surge. That was my basis for suspiscion of Petraeus, and it stands.
Even if he is playing fast and loose with the numbers, he’s still a smart general driven to win the war of public opinion by any means neccesary. I understand it. It has to be fought on principle: the truth has to be out there. I still like the guy better than most people in the Bush Admin.
But – and this is important – it’s entirely possible that the sudden casualty divergences are all in good faith, Petraeus just disagrees with the old way of counting, and nobody knows what’s really the truth. It’s also possible that Petraeus and this paper are right, civilian casualties are down. I actually think they’re down as well: I don’t think they’re down as dramatically as Petraeus or this paper presents them.
The question is: so what? I mean, it’s great that only 1500 reported civilians are dying violently a month in Baghdad instead of 2000. (which means the real numbers are something like 4500 instead of 6000). Part of me really appreciates that – or at least understands why this leads surge proponents to think they’re doing the right thing. But they’re not. The reason is that the new down %25 (Petraeus estimates, not mine) numbers are still awful and unacceptable. Reducing the intensity of a civil war or an insurgency by 25% has nothing – literally nothing – to do with winning it!
The situation is still basically the same: senseless waste of life on a large scale, a fractionally smaller large scale. A large-scale withdrawal would still wind this war down more quickly than the surge, and save more lives in the long run than another decade of COIN combat.
History says that is wrong, but I do agree that the casualty count by itself is not the issue. I would not necessarily be negative about the surge even if they went up, but I am glad they have not and think it is a positive.
As for divergent statistics, my point is whatever statistics I have seen they go down. He actually does account for the undercount of more recent data, but all independent sources show, and believe based on less data intensive evidence, there has been a decline. As for the link you provide, I saw that. I draw nothing from it at this point without hearing an explanation. I have too many times been told something about the data (see the Lancet, the recent claims about sectarian death methodology, claims about the Jones Commission, etc) only to find it was being dealt with improperly to jump to any conclusion. I am keeping my ears open in the meantime.
I have seen no evidence to back that up despite it being claimed repeatedly in many places.
I have no doubt his numbers are wrong, they all are. The claim that needs to be refuted is that the numbers are disingenuous. As for the 75% claim, it doesn’t seem to be that far off, if the other data is correct. By the methodology they are using, and it is as reasonable as any, the data shows a decline along those lines. You can argue some of the violence that is attributed to other factors is in fact sectarian, but they wouldn’t argue that. That however would have been true before the surge as well. it makes sense the decline would be most dramatic in the case of sectarian murders, so I see no reason to doubt it overall whatever methodological issues one might have about the specific share.
Well, if you follow the link to the pentagon data, they have a graph in their report that matches that of Gen. Petraeus. Compare the actual pentagon data (see pg 16) with that from Gen. Petaeus.
The Gen. Petaeus graph shows “Ethno-Sectarian Deaths”; the pentagon shows “Sectarian Deaths and Incidents” and agrees with the General on the deaths.
The link you provided ignored pg 16 of the pentagon report and used pg 20, which shows “Average Daily Casualties in Iraq”, including all KIAs and WIAs, including friendly fire.
If your link is correct, then the pentagon report not only disagrees with the general, but with itself as well.
Another point on the graphs:
Both the Gen. Petraeus and the pentagon graphs indicate a peak of about 2250 deaths in Dec, dropping to about 750 in August. The Democracy Arsenal graph shows the “Gen Petraeus data” at 3000 and 1500, respectively. And Democracy Arsenal links to the Gen. Petraeus data that doesn’t match its claim!
Will the real Gen. Petraeus data please stand up!
It looks like I jumped to conclusions; Democracy Arsenal was referencing a different chart presented by Gen. Petraeus, one not contained in the pentagon report (as far as I have seen).
I was just going to point that out Don, but the data obviously doesn’t match. Rather than assume something I suggest we try and figure out why. Hopefully we will get an explanation we can test or examine.
I don’t need a fancy talk or graph to tell me how well we are doing in Iraq. I see it. I see it when a man throws a shoes at our (then) PotUS. I see it when i am in Iraq and see the progress we are making.
The reason why most of the underlings don’t see it is because they aren’t getting any good stories from the media. They (the media) is narrowly focused on the negative aspect of the war because that is what sells add space.
Have you met an active member who has been in the crap and said, nope; I’m just sitting there doing nothing, spending all you money for no reason. Nope, Me either. Try picking up an Air Force, Army or Marines Newspaper or go to their websites. Find the great thigns we are doing for them, and find out for yourself why they want us to stay as long as possible.
Alas, you wont. You are much to lazy.
Oh well. Good luck with the rest of your conspiracy theorys.