Taking a Chance on War

The American Economic Review, Vol. 97, No. 4, September 2007

When President George W. Bush unveiled his National Security Strategy in the fall of 2002, he declared that the United States would "actively work to bring the hope of democracy… to every corner of the world." He was just the latest in a line of American leaders to call for the spread of democracy. It’s a belief that has driven the United States to war time and again, not only because it’s a moral imperative, but because it can be a pragmatic thing to do as well. After all, ever since the German philosopher Immanuel Kant espoused his idea of "perpetual peace" in the 18th century, many social scientists have argued that countries with certain characteristics — namely, democracy — will refrain from attacking each other. Politicians take this a step further: If the United States helps other countries embrace democracy, then the probability of war will decrease.

There is no consensus, however, about why "democratic peace" occurs. Two economists have analyzed a mathematical model of conflict that tries to answer that question. In an article titled "Political Bias and War" in the September 2007 issue of the American Economic Review, Matthew Jackson of Stanford University and Massimo Morelli of Ohio State University argue that conflict is often driven by divergent incentives of leaders compared with society as a whole, which they term "political bias." Autocratic Iraq invaded autocratic Kuwait in 1990, they would argue, partly because what was good for Saddam Hussein was not what was good for Iraqis.

To understand the model, consider the case of two fictional countries, Narnia and the smaller Neverland. The leaders of each country, Aslan and Peter, respectively, are assumed to control a fraction of the economic activity in their countries. If Narnia defeats Neverland in a war, Peter loses part of his wealth and Aslan gains a share of Neverland’s wealth. Narnia’s political system is biased if the spoils of victory that go to Aslan are greater than the losses Aslan would suffer if Narnia were defeated. In other words, political bias occurs when the incentives of a leader diverge from those of his people. Positive bias means a leader gets more of the spoils of victory than the losses from defeat, relative to his own subjects: Heads, I win more; tails, we both lose.

Jackson and Morelli’s analysis begins by asking not just why leaders would choose war over peace, but why leaders would choose war over a negotiated solution. To see why this is important, suppose Aslan and Peter are fighting over $100. Aslan is stronger, so if it comes down to it, he has a 90 percent chance of getting the money. But fighting is hard and costs each of them supplies and troops equivalent to $5 in value. Suppose Peter offers Aslan a deal: Aslan gets $90 and Peter $10. That sounds better than fighting to Aslan — he figures a 90 percent chance at $100 is worth about $90. If he fights, he has to pay an additional $5, so Peter’s offer is a good one. In fact, under reasonable assumptions, and not just in the case of Narnia and Neverland, there is always a negotiated solution that both sides prefer to war if the leaders are unbiased. Why, then, does war occur? The authors show that if the political bias is large enough relative to the costs of war, conflict follows.

Jackson and Morelli address a second possible cause of war, known as a commitment problem. Suppose that after Peter gives Aslan $90, Aslan uses the money to buy weapons that increase Aslan’s probability of winning a fight to 96 percent. Aslan might then come looking for more concessions — not such a great deal for Peter. Anticipating this possibility, perhaps Peter wouldn’t make the deal in the first place, and the two might just choose to fight at the outset. Therefore, two countries with an inability to make firm commitments would always fight each other, as long as the costs of war are not too great, right? Jackson and Morelli demonstrate, however, that this is not the case. Whether the inability to commit to abiding by the terms of a settlement leads to war depends on how wealth translates into power. If the probabilities of victory are proportional to each side’s wealth, for instance, unbiased leaders still won’t ever fight each other, and biased leaders sometimes will.

Jackson and Morelli’s propositions make the case that political bias is a key cause of war. If democracies are less biased, their research offers a novel explanation as to why democracies don’t fight each other. Democratic peace is not directly related to checks and balances, liberal values, transparency of the political process, or even representative institutions. Rather, it’s explained by the extent to which the fortunes of rulers and ruled rise or fall together.

It’s certainly plausible that positive political bias, when it exists, is a factor that encourages conflict. But does bias actually exist that often? And does the model explain democratic peace? It is worth remembering that bias, as defined by Jackson and Morelli, has nothing whatsoever to do with economic inequality. If a leader controls 99 percent of the economy and stands to gain 99 percent of the spoils of war, this constitutes a case of no bias.

Even in autocracies, leaders often appear to share more in losses than in spoils, making these cases of negative political bias rather than the positive bias that favors war in the model. If a leader stands to be deposed from office or have their treasury raided if a war is lost, these would be cases of negative bias. Suppose a king extracts 50 percent of the economic activity of a territory in taxes and stands to lose this territory in a war. Is there any reason to believe the king would be able to extract more from the economy of a conquered territory? If not, as has probably been the case since the rise of nationalism in the 19th century, then such cases are also not instances of positive political bias. When Saddam Hussein invaded Kuwait, when Czar Nicholas II entered the First World War, and even when Leonid Brezhnev sent Soviet troops into Afghanistan, these autocrats risked the very existence of their regimes. Positive political bias probably had nothing to do with it.

Perhaps if a natural resource exists in one country that could be expropriated by the leadership of another country, as in the case of Iraq’s attacking Kuwait, this would constitute a case of positive bias if the Iraqi side stood to lose no oil wealth of its own. Or perhaps if a king were able to enslave the subjects of a conquered territory, but not those of his own, this too would result in political bias. Alternatively, bias might result from the ability of the leadership of a large state to pilfer the coffers of a smaller state without really risking similar treatment if the war went badly. On the whole, and especially in the modern world, however, positive bias is probably rare. Although it might help explain conflict in specific historical periods, it seems unlikely to explain current differences between autocratic and democratic foreign policies. But the foreign-policy choices of Neverland have never been more clear.