Deep Uncertainty Scenarios in Latin America
Instead of 80% or 20% scenarios, let's think about some low-probability, high-impact events.
For those interested in both the mathematics and the philosophy of dealing with probabilities and risk, I recommend "The Art of Uncertainty" by David Spiegelhalter. I read a copy borrowed from the library in a week, and now I feel like I need to own a copy because I’ll be returning to the concepts within so often. That’s how good and useful it is.
As just one part of the book, Spiegelhalter dedicates a chapter to “Deep Uncertainty,” issues where it can be quite difficult or impossible to assign quantitative assessments of probabilities and impacts because they are at the edges of the scenarios or fall completely outside of the scenarios considered. The Black Swan concept falls within this framework for those familiar with its definition in Taleb’s book, but is only one type of deep uncertainty.
Since I recently finished the book and outlined some likely 80% scenarios last week, today’s newsletter focuses on some deep uncertainty questions and low-probability, high-impact scenarios in Latin America.
For those who think and write about Latin America, one way to consider this is to model elections and always include a scenario where the election doesn’t occur. There are plenty of quantitative ways to predict the odds of the election results1 as well as their potential impacts in a normal sense. However, attempting to map out the probabilities and the impacts if Chile cancelled its presidential elections or Argentina cancelled its mid-terms later this year becomes a more ludicrous exercise. That’s not even trying to guess at why that improbable scenario occurs (an undemocratic power grab, a military coup, a natural disaster, some legal technicality?). The specifics of this type of event can’t be predicted ahead of time, but we should all recognize and prepare for the general possibility that some unknown event could surprise us, even if it’s not in the basic scenarios we model.
Two relatively recent events demonstrate the challenges of this sort of uncertainty. The obvious one is Covid, which had an outsized impact on every country, even for those who had modeled pandemic issues prior to 2020. The second event is the 2019 protest wave and specifically the protests in Chile. The size and impact of those protests were significantly larger than any typical protest prediction models would have captured prior to that October. Protests are an event that have a fat-tail distribution, meaning that some, like Chile, fall outside of the protest scenarios analysts wrote out prior to 2019’s protest wave occurring.
One response to deal with this deep uncertainty is to engage in “think outside the box” efforts to consider edge scenarios that aren’t in a normal prediction framework. With that prompt, here are four deep uncertainty issues that I think about from time to time, but don’t often come up in my writing because the scenarios are fairly remote and there isn’t a reason to write about them on a weekly basis.
Interstate Conflict - In the past 100 years, there has only been one major interstate war in the hemisphere - the Chaco War between Bolivia and Paraguay in the 1930s. There have been maybe two mid-sized conflicts (Peru vs Ecuador in 1941 and Argentina vs the UK in 1982). There have also been a bunch of minor skirmishes, such as those between El Salvador and Honduras or later clashes between Ecuador and Peru, which hardly qualify as serious wars2. The occasional threats of war, including between Venezuela and its neighbors, have not materialized.
Because of that history, we minimize the odds of an interstate war in Latin America and almost certainly underestimate the potential impact. Yet, many Latin American countries build up their militaries and purchase weapons systems as if they are preparing for the interstate war that never arrives.
Latin America’s role in a US vs China or China vs Taiwan conflict - Talk to analysts focused on Asia right now, and they will highlight the small but significant potential for a regional conflict that goes global and draws the US in. Most of these scenarios revolve around China attempting to conquer Taiwan. Whether you think the odds of this sort of conflict are 5% or 40% in the next five years, they aren’t close enough to zero3. If that conflict were to occur, the impact on Latin America would be enormous. Countries will be forced to take sides. Economic warfare will definitely hit the region, and proxy kinetic battles are possible.
Weapons development - If Mexico or Brazil chose today to start a program to develop nuclear weapons in the next three years, how close could they get without the world knowing it? No, I don’t think either country has a nuclear weapons program. But if you spend time researching either country’s peaceful and military nuclear research4 and thinking through the potential incentives to build a nuclear weapon, you’ll come back a lot more uncertain than the initial gut reaction most people have to that question being absurd. Move away from the nuclear question and consider other types of WMD, missiles, drones, cyber, etc, that other countries or even non-state actors could develop or obtain, and there are plenty of ways for surprises to hit.
Democratic backsliding becomes authoritarian consolidation at a regional level - You could argue this is more of a gray rhino than a black swan, given that everyone has been writing about the democratic backsliding trend for decades. That said, though we all see and write about parts of the big trend, not much is written about the implications if it worsens, hits a tipping point, and consolidates across a majority of countries in the hemisphere. If democratic backsliding leads to one or several authoritarian alliances and an outnumbering of democratic countries, it represents a significant shift in how the region operates. It also greatly impacts the other scenarios above. Interstate war and arms races become more likely with larger numbers of undemocratic regimes forming parts of global blocs of power.
What do we do with these deep uncertainty scenarios?
Spend too much time thinking about scary and unlikely scenarios, and you’ll waste time that should be spent thinking through the far more plausible scenarios with many past precedents5. It’s still worth the effort to consider these long-shot scenarios at both the regional and country levels. You’ll never get the exact details of an unlikely scenario correct, but running the exercise helps prepare for the almost inevitable surprise “nobody expected” that does hit at some point in the future.
Doesn’t mean we don’t get those models wrong!
This is a question that inspires “what about [name a random conflict from the past century]?” We could discuss for days which past conflicts I should include on this list and how to categorize them.
With the caveat that I’m not an Asia expert, I’m at the lower end of the scale for a potential China invasion of Taiwan, maybe 10-15% odds in the coming five years. That said, it’s the sort of scenario that can escalate quickly once it begins.
Scary note from an IAEA alert in March 2025 about a Mexican nuclear research and waste facility: “The video surveillance and environmental radiation monitoring systems that CADER has were disabled and there was no way to know the status inside the Center.”
such as a future Argentina default.