Is the Surge Working? Ask the Data, Not the Politicians

One of the most important political questions of the day is whether the troop surge in Baghdad is working. If you ask politicians, the answer you get to that question is very predictable. Republicans say yes, Democrats say no.

What do the data have to say about this question?

Michael Greenstone, an M.I.T. professor, good friend, and one of the best young economists in the world, has just released an incredibly thorough and thoughtful analysis of the impact of the surge.

He finds the answer is mixed.

On at least one dimension, there is strong evidence of progress. The data clearly suggest that deaths of civilians in Baghdad have fallen, and there is no evidence that the crackdown in Baghdad has shifted violence to the rest of the country.

Coalition troop fatalities have been stable since the surge, which in some ways signifies progress since they were on a steady upward trend prior to the surge.

The surge does not seem to have helped in other dimensions such as the amount of oil produced or hours of electricity in Baghdad.

The most interesting part of Greenstone’s paper is his analysis of the pricing of Iraqi government debt. The Iraq government has issued bonds in the past. These entitle the owner of the bond to a stream of payments over a set period of time, but only if the government does not default on the loan. If Iraq completely implodes, it is highly unlikely that these bonds will be paid off. How much someone would pay for the rights to that stream of payments depends on their estimate of the probability that Iraq will implode.

The bond data, unlike the other sources he examines, tell a clear story: the financial markets say the surge is not working. Since the surge started, the market’s estimate of the likelihood of default by the Iraqi government has increased by 40 percent.

I have a few thoughts after reading Greenstone’s work:

1. This paper shows how good economic analysis can contribute in a fundamental way to public policy. Anyone who reads Greenstone’s article will recognize that it is careful and thorough. It is even-handed and apolitical. It combines state-of-the-art data analysis techniques with economic logic (e.g., using market prices to draw conclusions about how things are going).

2. Top economists like Greenstone virtually never write papers like this. The simple reason is that this sort of work is not rewarded in our profession. Academic economists are judged by the papers they publish in peer-reviewed journals. The lag time between submission to these journals and publication is often two or more years. By that time, no one will be interested in the surge, so editors won’t want to publish the paper. Consequently, good economists don’t think it is worth their while to do topical work like this.

3. Instead, this sort of analysis tends to be done by bad economists, or economists on the payroll of special interests. These reports might appear informative, but instead are often pure propaganda. To outsiders, it is difficult to determine what is careful analysis like Greenstone has produced versus the usual junk.

4. The internet can potentially solve both problems (2) and (3) above, leading to an increased supply of good, timely analysis. If people like Greenstone can immediately get their findings into the public debate through the internet, it gives a real purpose (not just an academic one) to doing the work. In addition, there are now online peer-reviewed academic journals that have greatly sped the time from submission to publication, potentially increasing the academic payoff to someone like Greenstone. With many respected economists now blogging, there is also a vehicle for these folks to weigh in on the quality of policy-related economic writings — like I am doing in this blog post.

Robert Stein

Analyzing the Surge in terms of data is precisely how Petraeus and Crocker erred in their presentation to Congress. The short-term weighing of costs and benefits omits the more salient price of staying this course for the rest of Bush's term--more than 1000 American lives, $100 billion and an irreparable breach in our political life:


I'm not sure that the bond market is a good indicator of actual probability. It's an indicator of what people perceive as the probability.

I don't have any opinion on the matter, but what people *think* is happening and what is *actually* happening are often painfully unrelated.


a) The way to measure the surge is NOT with data. The justification for the surge was that it would allow for political progress, therefore the standard of review should be whether the surge has helped the political process.

b) Though it is anecdotal, the fact that the President's friend/colleague/advisor Oilman Ray Hunt just signed a deal with Kurdistan to develop their oil shows that the insiders don't think that the IRaqi gov't will be around for long

Zack Turner

It is clear that the progress or decline in Iraq is mixed. Complex political and military situations are so difficult to simplify clearly, it is a shame that because of this problem they are condensed into short politically beneficial rhetoric. Like you began with Steven. Republicans say yes and democrats say no. Sometimes I wish they could be authentic and disclose the truth that it is not yet a complete yes and not a complete no. Too bad we have a president who considers himself omniscient.

Colin McFaul

The internet can potentially solve both problems (2) and (3) above, leading to an increased supply of good, timely analysis.

The first thing I thought of when reading this sentence was, which a lot of physicists use to discuss ideas in a format that's more formal than a blog, but not so formal as an actual journal.


Any suggestions as to why the financial markets have lost faith in the Iraqi government, even though the data show reductions in violence?

Herman Hudson

They keep telling the same type of lies that got us into this war. I just wonder when are the American peole going to wake-up, and put a stop to this madness.
No doubt in my mind that if Dem. had got us into this war with the results that they have now the Rep. would have thrown the Dem. and their president out on their head long ago.

Peter S Magnusson

I have seen commentary that violence is down in Baghdad partly because of rapid ethnic segregation. Is that taken into account?

David R.

Thanks for the 58 pages of discussion, data and analysis. I would like to see links to more research papers on more topics.


Isn't there a bit of a disconnect in linking the performance of Iraq bonds to the effectiveness of the troop surge? I haven't read the paper, but it seems like that would assume that investors in Iraq bonds have total knowledge of whether or not the troop surge is working.

If anything, it might be a great indicator of what people think is happening, but I would think that the entire point of Freakonomics is that just because we think we know, we may have no idea.


My only concern with this would be the source and quality of the data. For example, you note: "The data clearly suggest that deaths of civilians in Baghdad have fallen..." But what deaths do those numbers include? After all, the Bush administration has decided not to include deaths from car bombings in their casualty counts.

Is this an actual drop, or just a drop created by redefining the data set?


The data clearly suggest that deaths of civilians in Baghdad have fallen, and there is no evidence that the crackdown in Baghdad has shifted violence to the rest of the country.

Really? No evidence?

Did he look the Associated Press numbers?


Prof of Economics

If many members of congress state, "we need to get out of Iraq now", that could increase the expected probability of default (and increase the risk premium on the bonds); yet that is not an indicator of how well the surge is doing. So I have no idea why you think bond prices are an indicator of the surge working.


A similar effort to disseminate information to the public is currently being undertaken by the NIH (see PLOS1 ). A quick way to produce peer-reviewed work that others can comment and contribute to. It'll be interesting to see if scientific print publication will go the same way as other print media.


I like these comments (especially when they are on a blog about looking at the data). I would recommend that those people who have questions about where the casualties came from take time and actually look at the data. I took 3 minutes, downloaded the report, went to the bottom (the appendix is where they usually talk about sources of information), and found on page 37:

"X. Civilian Fatalities from
Our primary source for data on civilian fatalities comes from the (IBC) website. The principal researchers are Hamit Dardagan and John Sloboda. It is important to note upfront that the website is partisan—that is, anti-war—but their methodology is conservative and seeks to provide an accurate lower bound for the number of civilian deaths reported.
The data were downloaded from Documentation of the IBC's methodology is available at and it is briefly summarized here."

It goes on to explain in detail how IBC collects data. So next time you have a question about something like this, take 5 minutes and check the source to see if it says something about that.

On a side note, it is ironic that caia posts a link to a First, the site says it uses 2 sources for data, the AP and IBC (the same site that this report used). Then the author goes on to make up his own analysis of the data to show that the surge is not working because the numbers are wrong. By saying this, he hits on one of my biggest pet peeves, manipulating data. Anyone who looks at the graph would say that the surge is reducing civilian casualties, but he tries to argue that since casualties have been higher overall this year then this time last year it is evidence that the surge isn't working, even though the graph shows an increase in casualties over this period of time last year but a decrease in casualties since January.

This shows what happens when citizen journalists try to do complex analysis of data, they get it wrong or distort it. If you want to see real analysis, read the Greenstone paper.


Bruce in CA

Asking whether the Surge is working without referring to its originally stated goals is like trying to evaluate a field goal kick, without knowing where the field goals are.

The stated purpose of the Surge was to facilitate a political solution in Iraq. Virtually no progress has been made in that area. It's a failure.

Michael Broshi

There was a similar post on the Marginal Revolution blog. One interesting comment made there was that even if the price of bonds reflects the likelihood that Iraq will "implode," the drop in prices does not necessarily mean the surge has not worked--perhaps they would have dropped even more precipitously without the surge.


That Iraq's bond prices are falling is in line with most of the commentary I've read, as well as Petraeus's testimony to Congress: military, the surge is seeing success and decreased deaths, but the Iraqi government is not making the political gains that it needs to.

As for deaths, Matt and caia are misleading at best. The NYT's own report says that overall, deaths and carbombings are down 50% year-over-year. Two casualty counts are kept: one for total deaths, and one for deaths due to sectarian violence. Deaths due to car bombs which do not appear to be aimed at a particular sect are counted in the first category, but not the second. Likewise, for the front of the head/back of the head thing--realistically, though, how are you going to count it if a dead body is found in the street? It's unrealistic to ascribe them all to sectarian murders, or all to criminal violence; some distinction must be made, and the M.O. for the two different types of murderers seems to be about as close as one can get.



He isn't taking into effect the goals of the surge, just whether or not it did some good. From the article: "The central challenge for this analysis is to determine the causal impact of the Surge on the functioning of the Iraqi state."

According to your argument, (hypothetically) if the surge was to stop all civilian and military deaths but the political situation was still unstable, it would still be a failure.