What makes checks and balances function in some presidential systems, but not others? With President Trump’s 100th day in office fast approaching, the constitutional crisis that many critics feared has yet to materialize. Rhetorically, Trump continues to claim vast powers. In practice, when both the courts and Congress have pushed back, the administration has demurred.

Contrast this with the behavior seen this month in presidents Recep Tayyip Erdogan in Turkey and Nicolás Maduro in Venezuela. In the past month, both have ridden roughshod over democratic institutions, trying to dramatically expand their powers.

As a scholar who has extensively researched separation of powers and checks and balances in Latin America, let me outline some lessons the United States can draw from constitutional crises in this part of the world.

Presidential systems are often more problematic than parliamentary systems

A recent Pew Research Center survey shows that more than three-quarters of Americans agree with the statement that it would be “too risky to give presidents more power.” Apparently U.S. citizens understand what political scientists have long debated: The idea that presidential systems — particularly strong ones — are inherently problematic. For decades, political scientists have argued over whether presidentialism causes democracies to break down.

Defending the claim that presidential systems are inherently perilous, sociologist Juan Linz contended that separate elections for the executive and legislature create confusion — and sow the seeds for conflict — over which branch bears the mantle of democratic legitimacy.

At the same time, scholars have long recognized that not all presidencies have the same powers. Nor are they all equally prone to instability.

My research on contemporary Latin America provides new evidence that granting extensive constitutional powers to the president — particularly when the government is divided — makes all three branches of government less stable. Political scientists call a president with strong constitutional power and weak partisan support — i.e., facing a legislature dominated by opposition parties — the “difficult combination.” I looked into how such a system influences the way presidents are ousted, as happened recently when Brazil’s legislature removed Dilma Rousseff.

So what makes a presidential system dangerously prone to crisis?

My research is based on a systematic examination of legislative attempts to remove sitting presidents — either via impeachment or forced resignation — across 474 administration-years in 18 Latin American countries between 1985 and 2008. Here is what I found: under divided government, on average, one in 20 presidents will face an attempted ouster.

But if the president is constitutionally strong, the probability of this sort of constitutional crisis under divided government jumps to one in five.

The link between granting the president broad constitutional powers and presidential fragility is straightforward. Increasing the president’s constitutional powers allows them to make policies unilaterally, and sometimes recklessly — and gives them no incentives to compromise with political opponents. As the stakes of holding the presidency rise, presidents are more willing to engage in risky behavior to remain in power. What’s more, their opponents are more motivated to exploit opportunities to remove them.

Brazil’s Dilma Rousseff — who was impeached by a coalition of opposition legislators — is the latest case in point. Analysts have pinned her downfall on her go-it-alone attempt to making policy and carrying it out, neglecting to build consensus and alliances.

But concentrating constitutional power in the president can also destabilize other institutions, as well. Strong presidents are the most vulnerable when they lose their own parties’ support. And so, my research finds, that’s when they are more likely to act preemptively to attack against both legislative supremacy and judicial independence.

Few presidents facing divided government go so far as to shut down the legislature. But when a president’s constitutional power increases from moderate to high, the likelihood of such a “self-coup” approximately doubles, something that recently happened in Venezuela, where embattled Maduro had his Supreme Court lackeys strip the opposition legislature of its authority.

The same shift in presidential power more than quadruples the threat to courts, whether government is divided.

Consider Peru’s former president Alberto Fujimori, who defended suspending the constitution and closing down congress by claiming “if I hadn’t taken those measures, they would have deposed me.” In Venezuela, before former president Hugo Chávez took office, the country faced rumors of a coup. Shortly after being sworn in, Chávez began to remake both the legislature and the judiciary, which were dominated by his opponents. In Ecuador, the current outgoing president, Rafael Correa, came into office fully aware that most of his predecessors had been prematurely ousted — and so convened a constituent assembly that allowed him to seize control over both branches of government.

Is a similar crisis coming to the United States?

What does Latin America teach us about whether the United States will continue to avoid a constitutional crisis? Part of the answer depends on how we think about the extent of presidential power in the United States. On the one hand, the U.S. president is quite constrained, compared with counterparts in Latin America. The formal constitutional powers of the U.S. president are reactive: The president can veto, not pass, laws unilaterally. Given this, the evidence from Latin America suggests that whether or not President Trump faces unified or divided government would not markedly affect the prospects for presidential impeachment, let alone a power grab of the other branches of government.

On the other hand, critics on the left and right have warned of (or applauded) the expansion of presidential power in the United States, particularly since World War II. Some attribute that to the growth of the federal bureaucracy, others to an increasingly dysfunctional Congress. Either way, many argue that U.S. presidents’ ability to shape policy via executive orders or other unilateral actions has increased substantially.

If that is true, then there are at least two ways that the difficult combination that underlies constitutional crises in Latin America might eventually emerge here.

Most obviously, the president, like many of his predecessors, might suffer a major defeat in midterm elections, resulting in divided government. Confrontations between Trump and a Democratic Congress could then conceivably escalate into a constitutional crisis. Consider that when a Republican Congress refused to take action on any of President Barack Obama’s proposals, the Democrat increasingly turned to executive actions — and that’s when impeachment threats, however remote, began.

There’s an alternative scenario, as well: The president could lose the support of his own party. Imagine a scandal that leads to a huge dip in the president’s popularity, which suddenly makes it too electorally costly for his own party to continue to support him. That’s what happened recently in South Korea, when President Park Geun-hye was impeached by her own party.

But here’s the point to keep in mind. Constitutional crises may be triggered by any number of particular causes, but what matters more deeply is how institutions are constructed. Because partisan support in presidential systems is inherently fragile, limiting the president’s constitutional powers helps downgrade the potential for the sorts of constitutional crises that often beset presidential systems in other parts of the world.

Gretchen Helmke is a professor and the chair of political science at the University of Rochester, and author of Institutions on the Edge: The Origins and Consequences of Inter-Branch Crises in Latin America (Cambridge University Press, 2017). She is a co-director of Bright Line Watch, where this post also appears.