Imagine two doctors examining the same patient and reaching diametrically opposed conclusions. One says, “This person is sick because of an unhealthy lifestyle – the embargo has nothing to do with it; it explains less than ten percent of his ailments.” The other replies, “Wait a minute. You measured the wrong thing and calculated it incorrectly. Recalculate – and you'll find that the external factor you nearly wrote off could explain everything.”
This is precisely the kind of debate that flared up in academic economic literature surrounding one of the longest and most painful economic experiments of the 20th century – Cuba after 1959. And although it's a purely academic dispute, behind it lies a question important for anyone who wants to understand how the real economy works: how much can we trust the numbers if the numbers themselves depend on how we count them?
What's the Debate and Why Does It Matter
In 2021, economists Bastos, Geloso, and Bologna Pavlik published a study with a provocative conclusion: the American trade embargo against Cuba, in effect since the early 1960s, explains less than ten percent of the difference in living standards between the real Cuba and a hypothetical Cuba that did not follow the path of socialism. In other words, according to their calculations, the main cause of Cuban poverty is not external pressure but its internal economic system.
It sounds convincing. Such findings fit neatly into the conventional logic: “It's their own fault; external restrictions are just a convenient excuse.” But a group of other researchers set out to check what this conclusion was based on and discovered it hinges on two methodological decisions, each of which can be challenged. And challenged not ideologically, but mathematically.
Let's break this down in plain English – without formulas and jargon.
The First Problem: One Number Changes Everything
At the heart of the calculations by Bastos, Geloso, and Bologna Pavlik lies what is known as the elasticity of income to trade openness. It sounds intimidating, but the idea is simple: how much will a country's income grow if it starts trading more actively? Or conversely, how much will it fall if trade shrinks?
The researchers took the average value from a 2009 meta-analysis by economist Dueñas-Varu – a major review of 135 academic papers on the topic: 0.16. So, roughly speaking, if a country's trade decreases by one unit, its income falls by 0.16 units.
But here's the thing. The same meta-analysis by Dueñas-Varu showed that this number is highly dependent on how exactly “trade openness” is measured. If openness is measured by tariff rates, the elasticity is tiny, around 0.02. If it's measured by the trade share of GDP (exports plus imports, divided by the size of the economy) – the most standard and common method in the global literature – the average value is already 0.08, with the confidence interval ranging from 0.05 to 0.11.
In other words, Bastos, Geloso, and Bologna Pavlik used a number twice as large as the one that corresponds to their specific method of measuring trade. Not because they were dishonest – it was most likely a well-intentioned but flawed decision. But the consequences of this error were significant.
Why? Because in their calculation framework, it is this very elasticity that determines how much the embargo is to “blame” for Cuba's poverty. The embargo reduced Cuban trade – no one disputes that. The question is how much this reduction in trade impacted incomes. If you use an elasticity of 0.16 instead of 0.08, you overestimate trade's ability to influence income – and, paradoxically, this leads to the embargo's impact seeming smaller compared to other factors.
This is one of those cases where intuition can be misleading. It would seem that a larger elasticity should mean a greater effect from the embargo. But no: if you believe that trade, in principle, has a very powerful impact on growth, then the reduction in trade due to the embargo looks relatively small against the backdrop of everything else happening to the economy. The math here is tricky.
A Real-Life Analogy: Who's to Blame for the Fire?
Let me explain the second problem with an everyday example, because this is about something that seems technical at first glance but is actually a matter of common sense.
Imagine a fire breaks out in an old wooden house. Investigators determine that the house was built with violations – faulty wiring, decaying floor joists. And that someone dropped an unextinguished cigarette near a wall. The fire happened because both factors interacted: in a new house, the same cigarette wouldn't have caused such damage. In the old house, without the cigarette, the fire might not have happened right then.
Now the question is: how do you assign the “blame”? One could say, “The main thing is the dilapidated state of the house; without it, nothing would have happened.” One could say, “The main thing is the cigarette; without it, nothing would have happened.” Or one could say that both factors interact, and to attribute the damage entirely to just one of them is to distort the picture.
Bastos, Geloso, and Bologna Pavlik do exactly this: they openly state that the embargo “is not capable of generating Cuba's economic structure” – and therefore attribute the entire interaction effect between the embargo and other factors solely to those other factors. The embargo is only assigned its “pure,” isolated effect – which is indeed small.
But critics rightly point out: this is logically flawed. The embargo doesn't exist in a vacuum. It acted on a specific economy with specific institutions and structure. It was the interaction of these factors that led to the result we observe. To say that “the interaction belongs entirely to the other factors” is like saying the cigarette had nothing to do with it at all, because without the old house, it wouldn't have caused such damage.
The researchers who wrote a commentary on the work of Bastos, Geloso, and Bologna Pavlik reran the calculations with more appropriate input data. Here's what they found.
If you take an elasticity of 0.08 – the value that corresponds to the very method of measuring trade used by the original study's authors – the calculated effect of the embargo already changes. But the key shift occurs when we stop automatically attributing the entire interaction effect to “other factors” and start considering the embargo as a contributor to this interaction.
In this case, the picture changes fundamentally. The embargo could explain a substantial share of Cuba's economic lag after 1959. In some scenarios – where the interaction between the embargo and the planned economy was particularly destructive – it could explain the entire lag. Reality is likely somewhere between these extremes. But “less than ten percent” is definitely not the answer that emerges from a correct calculation.
A telling example is the so-called “Special Period” in Cuban history, which began after the collapse of the Soviet Union in the early 1990s. Cuba lost its main trading partner and subsidizer, and the economy collapsed catastrophically: by various estimates, GDP shrank by 30–35% in just a few years. The collapse of the USSR was not, in itself, a direct consequence of the U.S. embargo. But it was the embargo that prevented Cuba from diversifying its trade relations and quickly finding replacements for lost supplies. The vulnerability exposed by an external shock was largely created by the long-standing trade restriction. This is the interaction effect in action.
One might ask: so what? This is an academic debate about methodology concerning one specific country during a specific historical period. Why should we care?
I would answer like this: because this case is an almost perfect illustration of how a single technical decision in methodology can completely overturn a political and historical conclusion. The study by Bastos, Geloso, and Bologna Pavlik is not quackery. It is a serious academic work. But it contains two debatable assumptions, and it is these assumptions that determine the final conclusion. Change the assumptions, and the conclusion becomes the opposite.
This happens in economics all the time. Different researchers analyzing the same question – say, the impact of the minimum wage on employment, or the effect of tax reforms on growth, or the consequences of trade wars – arrive at radically different conclusions. Often, it's not because one of them is lying or bad at math, but because they chose different “technical” parameters: a different observation period, a different way of measuring a variable, a different assumption about who “owns” a particular interaction effect.
Money exists only because we believe in it. But economic conclusions also exist only because we believe in the correctness of the tools used to obtain them. The difference is that faith in money is hard to challenge – it is rooted in habit and culture. But faith in a specific methodology can and should be challenged.
What This Tells Us About How to Read Economic Research
When you see a headline like “Scientists have proven that X explains only 10% of problem Y,” it's worth asking a few simple questions.
- Where did the key numbers come from? As in our case: where did the 0.16 elasticity figure come from, and does it actually correspond to the context in which it is being applied?
- How did the authors divide responsibility between factors? If two factors interact, what assumption was made about who “owns” the interaction effect?
- What would change if these assumptions were different? If a conclusion falls apart with a minor change in parameters, that's a red flag. If the conclusion is robust under various reasonable assumptions, that's a sign of reliability.
These questions don't require a special education. They only require the habit of not taking numbers at face value just because they look precise.
Conclusion: The Modesty of Numbers and the Immodesty of Conclusions
The story of the Cuban embargo in academic literature is, in a sense, a story about the hubris of precision. When researchers get a specific number – “less than ten percent” – it takes on a life of its own. It seems irrefutable because it is specific. But behind this number lies a chain of assumptions, each of which can be challenged.
The authors of the commentary showed that by adjusting the elasticity and rethinking the distribution of the interaction effect, the “modest ten percent” turns into “all or nearly all.” This doesn't mean the latter are right and the former are wrong. It means the issue is far more complex than any single number would suggest.
Economics is not physics. There is no constant speed of light that can be measured once and for all. Here, there are people, institutions, historical accidents, and the interactions between them. And any attempt to reduce all of this to a single coefficient requires an honest admission: this is an assumption, not a fact.
The Cuban case is extremely complex. The country's trajectory after 1959 was influenced by many factors at once: the nature of its economic system, geopolitical isolation, dependence on Soviet subsidies and then their disappearance, the specifics of Caribbean geography, and much more. Correctly disentangling these influences is a challenge that the field of economics has been grappling with for decades.
But this is no reason to throw up our hands and say, “It's all too complicated, nothing is known.” It is a reason to demand more honesty from research regarding its assumptions – and more modesty regarding its conclusions.
Because sometimes the difference between “ten percent” and “everything” can be found in a single digit in the third decimal place.
