Friday, March 24, 2006

Invading Iraq versus Containment

Was the decision by the Coalition to invade Iraq in 2001 unwise? Many will say it was because of the difficult situation the allied military forces have subsequently encountered. They will also argue that WMD, a prime reason for intervention, have not in fact been discovered. Finally, there is growing recognition of the huge financial and other costs of the war. Initial official guesstimates of the cost of the Iraq invasion were much lower than those actually incurred, see here.

The first two reasons for opposing the invasion need not be sensible if the information that gives rise to these critical judgments was invisible to decision-makers at the time of invasion. It otherwise involves ex post wisdom. The other reason, that the costs are large, is not compelling if the costs of taking alternative available courses of action exceed those of invasion. A sensible analysis of the decision to invade needs to account for these issues.

Recently several more careful studies of the invasion costs have been provided (here). Of these, the Davis, Murphy, Topel (DMT) study (here) is of interest because it addresses the concerns expressed above. Specifically:

(i) It assesses invasion costs ex ante based on data and ‘views of the war’ as they were known in 2003 when the invasion occurred. The DMT approach is intended to avoid the ex post wisdom of many recent critiques of the invasion’s costs. While it is puzzling that actual costs turned out to be markedly higher than were loosely claimed to be plausible at that time – the DMT argument is presumably that an insufficiently detailed study was not made then, although it could have been along the lines of DMT.

(ii) The DMT cost estimates for the invasion policy are compared to the counterfactual of pursuing a sustained containment policy (with substantial US military resources based around Iraq) of the type pursued 1991-2003. This amounts to considering net costs – costs net of those avoided by invading. Containment was costly in itself because the US maintained 28,000 troops, 30 ships, about 200 aircraft and so on in the region partly to put pressure on Iraq.

(iii) Alone among recent, careful studies, DMT criticize the notion that the invasion can be rejected as a mistake because of its ‘high’ costs. They argue it is opportunity cost that matter – the cost of invasion compared to the cost of the next most plausible policy option, containment.

The DMT study is detailed and lengthy and deserves more thorough analysis than I can give it. Nor do I wish to be identified as a supporter of the DMT conclusions. I think conceptually DMT gets it right and is worth serious study on this account. But I am unsure about specific assumptions attached to the events they envisage as possible and which they attach costs.

Main Points

Pre-invasion views of the cost of the war suggest to DMT a current present value of costs of the invasion at $100b-$870b (at a 2% discount rate) compared to the other major recent cost estimates I am aware of due to Bilmes and Stiglitz (S-B) of $1-$2 trillion. About half of the DMT costs are military occupation costs. (The difference between the upper limit here and that of S-B is largely explained by the fact that S-B account for the implied macroeconomic costs of $5-$10 per barrel increased oil prices and costs such as higher interest rates which Davis et al. do not – these comprise most of the extra trillion in the S-B estimates. Also S-B make no attempt to compute benefits of the invasion by considering costs avoided through it). The S-B cost estimates are criticized on other grounds by Becker-Posner.

Much of DMT is taken up with articulating such costs – costs of US lives lost, injury costs to US forces and so on. These estimates can all be qualified and discussed but it is beyond my scope to do so here.

DMT cost a continued containment program (the counterfactual) at around $300b which is a lot even if the policy was successful - the cost would be much higher were it not. Moreover, DMT claim the containment policy was flawed since it involved Iraqi collaboration with terrorism, difficulties of ensuring compliance with WMD inspections and even raised the prospect of a possible terrorist attack on the US. These add to the costs of containment. DMT suggest a total cost, accounting for these factors, of $350b-$700b which is not vastly different from the cost they compute for the invasion policy. My intuition suggests this figure seems high – the costs of policy failures more than doubling the costs of containment itself. And indeed one could argue that coming up with a high cost of failures almost amounts to presuming a prior cost justification for invasion.

In calculating these extra costs they argue that, even though WMDs were not uncovered after invasion, that this was not reasonably knowable in 2003. This seems reasonable. A recent report indicates:

‘it was not unrea­sonable to believe that Iraq retained stocks of chemical and biological weap­ons produced prior to the first Gulf War and that it would use them if the regime’s survival were threatened. The importance that senior Iraqi officials had ascribed to WMD, Baghdad’s repeated efforts to obstruct UN weapons inspections, the regime’s failure to resolve inconsistencies in its declarations to the United Nations (in part to foster ambiguity about the status of its WMD and thereby to reassure domestic supporters and deter domestic and foreign enemies), and low confidence in the ability of US intelligence to track WMD developments in Iraq all contributed to this impression’.
Shawcross (2004) makes similar points. Saddam was obsessed with WMD and ‘the very existence of a new global terrorist network made Iraq’s presumptive possession of WMD much more threatening’ (p. 69-70). The extra costs of containment are also assumed to reflect costly attempts to maintain credible threats to compel compliance with inspection and disarmament and to include possible costs of a limited war against Iraq.

Finally, and very controversially, DMT include as containment costs the likelihood of an enhanced terrorist attack on the US. They cost the 9/11 attack on the US at $50b assessed as three times earnings of lives lost plus cleanup and restoration costs. They guess the cost of a future attack, such as a 'dirty' nuclear attack, at $100b. They then claim that ridding Iraq of Saddam reduced the probability of such an attack by 4% per year yielding a present value saving of $80b. As DMT themselves note it is unclear that the prospect of a terror attack on the US has been reduced by the invasion – it may have increased - in which case this 'savings' would instead become an additional cost.

DMT assess the costs of intervention to the people in Iraq in two ways.

(i) Lost output. In material terms they claim the Saddam regime reduced Iraqi incomes by 75% mostly due to foregone oil production and a bloated military. Pre-war Iraq had 500,000 people in security and intelligence services and a further 800,000 in the armed forces. DMT claim, if US actions can recover half this loss over one generation, the Iraqi people will be materially better-off. The strong DMT claim is that ‘war is an economic blessing for Iraqis compared to the policy of containment’ (p.55).

(ii) Lost lives and suffering. Apart from torture and repression Saddam killed or caused the deaths of 500,000 of his own citizens. He killed 200,000 Kurds and forcibly relocated 1.5 million of them. DMT estimate that, under the containment policy introduced after 1991, Saddam killed another 200,000. Their estimate is that 10,000-30,000 Iraqis would have continued to die annually had containment continued. This suggests a higher death toll than some calculate has occurred since the invasion. They acknowledge that this might not continue to be true if full-scale civil war breaks out as a consequence of invasion but suggest that one cannot assume that the alternative to invasion would have been low loss of life. Some however have claimed that civil war is already upon Iraq in terms of accepted criteria for judging such situations (e.g. Juan Cole at Salon). Some on the left seem almost jubilant with the prospect of all-out civil war!

This extent of deaths is of major concern to the international community. The Iraq Body Count (an anti-war group) claim that over the three years of the war 35,000 civilians have been killed as a consequence of the war. The death count has risen steadily as the war has continued from 20/day in year 1 of the war to 36/day in the year to March 2006. With these figures the DMT claim is only correct if (i) the civil conflict does not deteriorate further into civil war, (ii) the assumed level of deaths that would have prevailed under the Saddam regime is correct and (iii) the scale of deaths estimated by Iraq Body Count is approximately accurate. Estimates of deaths in the conflict vary greatly - the Lancet study described here suggests deaths of around 100,000 to mid-2005 but the range attached to the figures is so wide (from less than 10,000 up to 194,000) that some commentators described them as 'dartboard'. The difficulty is that in a time of war, or during an oppressive regime such as Saddam's, accurate records of body counts are unsought by those in power. Incidentally, while media attention has focused on US military deaths of around 2,000 this figure is low compared to the Vietnam War where 60,000 were killed. Of course each of the 2000 lives lost is tragic but overall this is not of the dimensions of the earlier US conflict.

DMT is a large study with many strong assumptions many of which can be criticized as being arbitrary and some as unrealistic. It is however an instance of using cost-benefit analysis to provide an assessment of the case for US intervention to deal with ‘rogue states’ based on sound methodology. I think no-one with sense could believe specific conclusions but setting the issues out as DMT have forces us to consider, in a consistent way, the types of costs that arise in this type of conflict and how they should be assessed.

No comments: