Believe it or not, the conflict in Libya didn’t end with “Benghazi.”

U.N. negotiators have been working to bring the warring factions in the Libyan civil war to the negotiating table. They hope to end the fighting, form a national unity government and provide a clear alternative to the Islamic State. One proposal would establish a temporary government – the Libyan Government of National Accord – in which competing factions would share power.

But not all power-sharing agreements are created equal. In particular, not all types of power-sharing agreements are equally effective at keeping the peace after a civil war.

In a forthcoming article, we and our co-authors — Scott Gates, Håvard Strand and Kaare W. Strøm — examine this relationship between power-sharing institutions and the rise or recurrence of civil wars. Using new data on 180 countries over 35 years, we look not only whether power-sharing institutions can prevent a civil war from recurring, but also whether they can prevent civil wars from breaking out in the first place. This work builds upon and goes further than several earlier important studies on this question.

Here’s what works best: power-sharing arrangements that check the power of governments to abuse ordinary citizens. The current Libya proposals don’t do this. That means civil war is more likely to return.

Three types of power-sharing

We put power-sharing institutions into three categories: constraining, inclusive and dispersive. Each type has drastically different rules for how factions share power.

1. “Constraining” power-sharing arrangements.

These limit the power of political officeholders. This protects vulnerable groups and individuals, and civil society more broadly, against repression. These arrangements include institutions like a supreme court with judicial review powers, and constitutional provisions to protect religious freedom. These types of institutions that can stop whoever’s in government from usurping power through such repressive steps as taking control of newspapers or limiting the rights of minority factions.

By limiting government repression, constraining institutions make life better for ordinary citizens. More satisfied citizens are less likely to follow leaders who want to violently overthrow the government. In other words, the odds of a future civil war are lower when more constraining institutions are in place.

2. “Inclusive” power-sharing arrangements.

By contrast, “inclusive” power-sharing puts the representatives of each group into specific offices or gives those groups a voice in specific decisions. That’s what’s currently proposed for Libya. This type of power-sharing exists, for example, in Lebanon, where the president must be Christian, the prime minister must be Sunni Muslim and the speaker of parliament must be a Shia Muslim.

This type of inclusion is designed to ensure that no group fears being cut out of power entirely. In Libya, the proposed unity cabinet would include representatives from different regions and political factions. On top of that, negotiators are pushing for rules that ensure that different groups must come to a consensus before certain major decisions are made.

It’s true that ensuring that each group’s voice will be heard in the future government can convince warring factions to stop fighting. But these institutions are geared toward elites. They tend to coopt the leaders of opposition factions into government institutions. But that doesn’t prevent them from using that power to repress ordinary citizens.

In fact, countries with “inclusive” institutions are prone to abuse human rights. Those citizens are then ready to take up arms to improve their lives — and do. “Inclusive” power-sharing arrangements are more likely, on average, to break down into civil war.

3. “Dispersive” power-sharing arrangements.

Dispersive power-sharing faces a similar problem. It distributes power among various social and political groups in a pre-defined, often geographic pattern. They often result in federal systems, where power is shared among various states, provinces, ethnic groups, or tribes. This could happen in Libya, where most of the important political factions are tied to specific regions.

But once again, this approach gives power to elites – and leaves ordinary citizens vulnerable to abuse by the newly powerful. As a result, national leaders may try to sway disaffected opposition leaders in certain regions in order to focus human rights abuses on other groups. As a result, these arrangements, on average, do not reduce the risk that civil war breaks out.

Peace now or peace in the future?

Our findings suggest that negotiators in Libya face something of a Catch-22. On the one hand, inclusive and dispersive institutions dole out power to specific elites, which is an excellent way to convince them to stop fighting and join a government. Unfortunately, that isn’t a reliable path to stable, long-term peace. Constraining institutions do help secure long-term peace. But the elites at the negotiating table aren’t especially interested in having their power constrained – and so these constraints make elites less inclined to join a unity government that has them.

In other words, the best deal for tomorrow may be difficult to reach today.

Since stopping the fighting now could save tens of thousands of lives, negotiators may well decide to settle for a less-than-perfect power-sharing arrangement in order to stop the killing. Establishing a competent state is a vital first step — but will not be enough if that state lacks effective mechanisms to constrain its power.

Libya is more likely to be stable in the long-term if elites consent to constraining institutions. Any other form of sharing power would be, at best, an uneasy political truce, or, at worst, a ticking time bomb.

If the proposed unity government can shepherd Libya through its first election, there will be losers as well as winners. If the rights of those who loser in this process are not protected, they may return to the fight. Constraining institutions do the best job of protecting those rights and are Libya’s best chance for staving off future bloodshed.

Benjamin Graham is assistant professor in the School of International Relations at the University of Southern California. 
Yonatan Lupu is assistant professor of political science and international affairs at the George Washington University.