« Rational Exuberance and the Greenspan Fed | Main | An Interview With David Card »

Wednesday, December 06, 2006

Glaeser: The Political Economy of Warfare

This paper constructs a model of warfare where leaders benefit from war even though the population as a whole loses. One of the conclusions is that "While the logic of the wag-the-dog section of the model has been used to explain American intervention in Iraq, if the war was meant to increase popularity, then this seems to have been a profound miscalculation. It is hard to think that the change in the probability of staying in power (which in retrospect was probably negative) times the modest benefits of being president was even in expectation large enough to justify the difficulties inherent in leading the country through an unpopular war." Profound miscalculations? Here's the abstract, introduction, and conclusion to the paper:

The Political Economy of Warfare, by Edward L. Glaeser NBER WP 12738, December 2006: Abstract Warfare is enormously destructive, and yet countries regularly initiate armed conflict against one another. Even more surprisingly, wars are often quite popular with citizens who stand to gain little materially and may lose much more. This paper presents a model of warfare as the result of domestic political calculations. When incumbents have an edge in fighting wars, they may start wars even if those wars run counter to their country's interests. Challengers are particularly likely to urge aggression when they are unlikely to come into power and when the gains from coming to power are large. Leaders who start wars will naturally try to create hatred by emphasizing the threat and despicable character of the rival country. Wars will be more common in dictatorships than in democracies both because dictators have stronger incentives to stay in power and because they have greater control over the media.

I. Introduction

Why do countries, both democracies and dictatorships, engage in massively self-destructive wars? As Mansfield and Snyder (2005) make quite clear, wars are often pursued by democracies and are often enthusiastically supported by the population as a whole. Given the enormous amount of destruction caused by armed conflict, it is remarkable that countries enthusiastically enter into wars when negotiated settlement is presumably always an option (Fearon, 1994).

This paper follows Marx (1859), Fearon (1994) and Hess and Orphanides (1995) and presents a model of warfare where leaders benefit from conflict even though the population as a whole looses.[1] Warfare creates domestic political advantages, both for insecure incumbents like Napoleon III and for long-shot challengers, like Islamic extremists in the Middle East, even though it is costly to the nation as a whole. Self-destructive wars can be seen as an agency problem where politicians hurt the nations but increase their probability of political success. This problem becomes more severe if the population can be falsely persuaded that another country is a threat.

The model in this paper considers three different cases. First, I assume that ability at warfare is known only for current leaders.[2] As a result, current leaders with above average skills in making war will have an incentive to start wars because, once these wars begin, voters won’t want to switch horses in mid-stream for a leader of unproven ability. While the model relies on asymmetric knowledge of incumbents and challengers, any assumption that wars increase the costs of switching leaders can generate similar results. In this “wag-the-dog” version of the model, leaders will be more likely to support wars when their war-making ability is higher, when the war is less costly to the nation, and when the benefits of office are higher. Popular incumbents are less likely to start wars.

While this version of the model can explain why incumbents might favor warfare, it cannot explain why wars would ever be popular. In the second variant of the model, war can beneficial because another country poses a future threat.[3] In this model, both challengers and incumbents may be too pacific relative to the needs of the nation because they bear some costs of war and few future benefits. There is a strategic complementarity between incumbent and challenger decisions about war. If a challenger is bellicose then an incumbent may declare war first to reduce the challenger’s electoral support.

This second version of the model can’t explain why countries would ever start wars that have higher costs than benefits. To address the existence of extremely popular yet costly wars, I allow the possibility that political leaders can send false messages incorrectly depicting another country as a threat. This model follows Glaeser (2005) and assumes that people can be misled by political entrepreneurs who vilify outsiders as a means of increasing support for their own policies. Wars become popular because the population has been convinced that the outside nation is a threat, but since the people have been falsely informed, the wars are ultimately costly to the nation.

In this version of the model, incumbents can again be moved to war by the threat of a bellicose incumbent. Outsiders who are unlikely to come into power otherwise support war as a strategy of political desperation. They may gain politically but are unlikely to pay the costs since they are unlikely to be elected. This result helps us to understand why out-groups in countries like Saudi Arabia and Egypt are today the strongest voices for an incredibly costly conflict with the U.S. and Israel. This version of the model also suggests that information costs are important. Trust in the government also makes war more likely. Dictatorial control over information may also help understand a dictatorial penchant for welfare.

While there is some debate about the facts (Mansfield and Snyder, 2005), most political science work following Bueno-de-Mesquita (1981) finds a robust negative effect of democracy on war (Russett and Oneal, 2001, Rousseau, 2005). This model predicts this relationship both because dictators have a greater ability to mislead the public due to their control over the media and because the gains from staying in power are much higher for a dictator than for a leader in a democracy. As I discuss in the case studies, there are many examples of nineteenth century leaders like Napoleon III and Bismarck who were willing to go to war to stay in power, but that there is little evidence of any 20th century American or British incumbents going to war for political purposes.

The first case study looks at nineteenth century Europe where both Napoleon III and Bismarck used wars for domestic political purposes (Mansfield and Snyder, 2005). Napoleon III’s foreign adventures rallied support for his wobbly regime and for understandable reasons. Given the unrest he faced in 1870, even the Franco-Prussian war doesn’t seem like an irrational gamble. Bismarck also undertook three wars to create both Prussian hegemony within Germany and to silence the critics of the Hohenzollern monarchy within Prussia. These wars helped ensure Hohenzollern authority in Germany until 1918.

I then turn to entry into World War I—a collectively disastrous act for the European powers. Austro-Hungarian action against Serbia was certainly motivated by internal politics, as they were fighting to suppress an ethnic uprising, and to reaffirm the primacy of the Hapsburg regime. German and Russian belligerency also had much to do with internal politics. The Kaiser and his ministers had regularly justified large army budgets (which funded a force used to eliminate internal opposition) by emphasizing foreign threats from England and Russia. The Czar similarly tried to build support by pushing Russian interests in the Balkans. In 1914, both Romanovs and Hapsburgs faced tremendous internal unrest, and it is unsurprising that both leaders sought popular support through a war that could be depicted as defense against outside aggression.

Finally, I turn to American wars between 1896 and 1975. During this period, there is no solid evidence on incumbents acting in a particularly bellicose way to further their own electoral chances. Entry into World War I and World War II seem to have been decisions made out of conviction with little or no political motivation. Cuba, Korea and Vietnam fit a pattern of bellicose outsiders raising the costs for the incumbent. Nationalistic outsiders pushed war with Spain in 1898 and McKinley was forced to act before he lost control of his own foreign policy. In 1950, Truman faced a drumbeat of Republican hostility to his supposed weakness in Asia. Likewise, in Vietnam, the Johnson tapes show a leader who acted aggressively in part because he feared Republican opponents would use any weakness towards Russia against him.

The phenomenon of incumbents using war to shore up their power is quite real, but the case studies and the model suggest that is primarily a feature of dictatorships rather than democracies. If anything, the tendency in democracies has often been for incumbents to be warier of conflict than opposition groups or the public at large. Both in democracies and in dictatorships, the strongest voices for conflict are often outsiders that have little chance of having to lead a war torn country, but that stand to gain politically by pushing warfare.

II. Aggressive Wars and Political Survival—Wagging the Dog


VIII. Conclusion

Does this model help us to understand conflict in the Middle East today? While the logic of the wag-the-dog section of the model has been used to explain American intervention in Iraq, if the war was meant to increase popularity, then this seems to have been a profound miscalculation. It is hard to think that the change in the probability of staying in power (which in retrospect was probably negative) times the modest benefits of being president was even in expectation large enough to justify the difficulties inherent in leading the country through an unpopular war.

The one aspect of the model that is supported by recent events in the U.S. is that the supporters of the war, both in and out of the administration, worked hard to vilify Saddam Hussein. The fact that some allegations appear to be incorrect ex post is consistent both with the model and with almost every other example of wartime leadership. As Arthur Ponsonby said in 1928, “when war is declared, truth is the first casualty.” The model does better at explaining the political patterns in the Middle East.

While previous Middle Eastern leaders, notably Nasser and Saddam Hussein, led their countries into wars, quite possibly to quiet domestic discontent, the current leadership of Islamic countries has not pushed towards war with Israel. Instead, it has been generally allied with the U.S. Egypt, Jordan, and Saudi Arabia and has frequently cooperated with the U.S. While the leadership of these countries has generally sought to avoid conflict (after all, any conflict would be immensely costly to them), dissident groups within these countries have been far more bellicose. These groups have worked to demonize the U.S. and Israel. For these political actors, who have little chance of coming into power, supporting conflict increases political support and there is little chance of having to actually lead the country in war.

Can the threat posed by these anti-U.S. and anti-Israel Arab leaders be reduced? The model suggests three parameters that might be changed and that might matter. First, decreasing the returns to leading these countries, through checks on the power of the executive, should reduce the incentives to use military means to stay in power. Second, a greater rotation of leadership, where the current opposition has a higher probability of leading in the future, should reduce the attractiveness of promoting highly self-destructive policies. Third, policies that increase the costs of spreading hatred against the U.S. and Israel, will also make war less popular and war-mongering less attractive.

1 While the core idea in this paper is in Fearon (1994) and in the work of earlier analysts such as Karl Marx, this paper differs from most of its predecessors in providing a political micro-foundation for the electoral gains from combat. Hess and Orphanides (1995) is a prominent exception as it shows war resulting from well-described electoral competition.

2 This is a primary difference between this model and Hess and Orphanides (1995). In that paper, leaders go to war to show their war-making ability. In this model, leaders go to war so that their war-making ability becomes relevant to re-election.

3 Wars can serve a country’s interest because countries lack the ability to commit to long run contracts, where one country promises never to exploit another in the future. If one country thinks that the balance of favor is increasingly favoring a neighbor then pre-emptive war may be attractive. Although, in a world with complete contracts both nations would happily commit never to start a conflict.

    Posted by on Wednesday, December 6, 2006 at 11:43 AM in Academic Papers, Economics, Iraq and Afghanistan, Politics | Permalink  TrackBack (0)  Comments (50)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Glaeser: The Political Economy of Warfare:


    Feed You can follow this conversation by subscribing to the comment feed for this post.