Is unpredictability a virtue or a hindrance in the realm of national security? 

Since the election of Donald Trump, inquiring minds want to know. Some political commentators view volatility as an asset. This line of argument tends to invoke Richard Nixon’s “madman theory” of international relations, according to which adversaries are more cautious simply because they are worried about the impulsiveness of a U.S. president.

But Sandeep Baliga, a professor of managerial economics and decision sciences at the Kellogg School, takes a different view. In his research, Baliga applies game theory to study issues in international relations. Using this model, he has found that meticulous strategy, not volatility, is ultimately the best approach.

“National security always involves a complex set of games,” Baliga says. “It’s not just about acting tough or keeping others on their toes. You have to track the logical conclusions of your assumptions, outline those assumptions carefully, and deduce the conclusions in a rigorous way. That’s how you avoid mistakes.”

No Single Model

In zero-sum games like poker, it can pay to be unpredictable—even strategically random. If a player’s bluffing becomes apparent to the other players, his bluff will be called and he will not win.

But international relations is not a zero-sum game. In the language of game theorists, war leads inevitably to “surplus destruction,” which essentially means less to go around for everyone. Nations thus often have a clear incentive to trade with one another, and coordinate on peace efforts, rather than appropriate one another’s resources by engaging in a costly and uncertain war. 

Ultimately, an effective national security policy is one that accounts for these complexities and entertains a wide range of variables and outcomes. “There’s no single model of the world that works all the time,” Baliga says. “Acting tough might work in some cases, but it might backfire in others. So before deciding on a policy, leaders should first determine what kind of situation they face.”

Take the U.S. response to Al Qaeda and ISIS over the past two decades. As Baliga sees it, Al Qaeda was engaged in “asymmetric warfare” with the goal of provoking the U.S. into a larger military conflict, thereby winning broader support from moderate Muslims around the world. 

This example suggests that a national security strategy must account for the possibility that an aggressive military intervention might actually end up radicalizing a population rather than eliminating terrorism.

“In national security, predictability can definitely pay.” 

ISIS, on the other hand—comprised as it is of Sunni militias that emerged from Iraq’s disintegration—has been fighting a more traditional war. “At least in the beginning, they were acting more like an army, and it wasn’t in their strategic interest to provoke the U.S.,” Baliga says. 

But by radicalizing young people and inspiring terror attacks with its propaganda, ISIS may have miscalculated, virtually guaranteeing U.S. military involvement. And now that it is on its heels, ISIS might be adopting a strategy of provocation similar to the one that characterized Al Qaeda. 

From the U.S. perspective, Baliga points out, “these are two different games. One is a game of escalation; the other is a game of deterrence.” Recognizing this is critical, because each requires a different response. 

“Your optimal strategy should take into account what you know or don’t know about the game you’re playing,” he says. 

Who Wants a Madman?

So is there truth to the effectiveness of the “madman” theory of international relations? 

For the most part, Baliga’s answer is “no.” To explain why, he points to the work of Thomas Schelling, an economist and Nobel laureate who used game theory to study approaches to nuclear deterrence during the height of the Cold War. 

Schelling promoted the importance of reputation as a useful deterrent. But while Richard Nixon believed that a reputation for unpredictability was the most effective deterrent, Schelling knew that consistent behavior was more effective. If your opponents believe you will keep your word, then your word can shape their actions. 

As Baliga puts it: “in national security, predictability can definitely pay.” 

Schelling’s approach pioneered the “principal–agent” problem, where a “principal” finds herself dependent upon—and needing to incentivize—an “agent” whose interests may not align with her own. In a classic example, an employer might incentivize an employee to be productive by linking his salary to outputs such as sales. If the employee believes the employer will renege on paying him, or that her payment is random and independent of performance—“the madman theory of pay?” Baliga asks—there is no incentive for him to work.

The same dynamic holds when the principal is the United States and the agent is North Korea. In this case, the U.S. wants to influence North Korea’s decision-making by linking it to outputs such as the de-acceleration of North Korea’s nuclear arms program. If North Korea is left without a clear sense of how the U.S. might respond to aggression or appeasement, its leaders might choose a more reckless and destabilizing course of action. After all, Baliga points out, it has not gone unnoticed in North Korea that the end of Libya’s nuclear program led eventually to Muammar Gaddafi’s death—not his permanent acceptance into the international community. 

In other words, an agent is more likely to be swayed if he believes the principal will abide by any agreement the two parties reach. Therefore, it is important for a principal to cultivate a reputation for following through on policies designed to incentivize the actions of even unsavory agents.  

“If you are a truly mad leader, why would anyone change their behavior as a function of what you do? If they know you might do something crazy whether they do something you like or not, they might just say ‘the hell with it, I’ll do whatever I want.’ The ‘madman’ actually has to be clever, doing something crazy if you don’t do what he wants and being accommodating if you do. In that case, well, he’s no longer mad.”

The underlying logic for designing incentives is similar in the employer–employee and the U.S–North Korea versions of the principal–agent model, says Baliga. But there is one key difference: In a business transaction, a contract enforced by a powerful court system can make a principal behave, making a reputation for trustworthiness slightly less important. In international affairs, on the other hand, there is no comparable, efficient enforcement mechanism.

“So a reputation for trustworthiness is all the more important,” says Baliga.