Measuring the Effect of the Embargo on Cuba's Economy
Claims that the U.S. embargo has had a small effect on Cuban development don't stand up to scrutiny.
By Francisco Rodríguez
The extent to which the U.S. embargo of Cuba, which began to be imposed in 1960 and has been significantly strengthened over the past six decades, plays a primary role in explaining the country’s economic underperformance is at the center of a long-standing and contentious policy debate. Critics of the embargo highlight its pervasive effects: it does not just block trade with the U.S.; it also creates significant, often binding, deterrents for many third countries to interact with Cuba. Others claim that socialist economic policies are the main culprit for Cuba’s underperformance and argue that the embargo plays a secondary role.
A recent paper by Bastos, João Pedro, Vincent Geloso, and Jamie Bologna Pavlik claims that the embargo has played at most a minor role. The paper, provocatively titled The Forsaken Road: Reassessing Living Standards Following the Cuban Revolution and the American Embargo, uses a method known as synthetic controls to create a counterfactual scenario that resembles how Cuba would have evolved in the absence of the events that took place around 1959. This counterfactual Cuba behaves similarly to Cuba prior to 1959, but then begins to diverge significantly, showing a much higher rate of GDP growth than Cuba. This result suggests that if Cuba had not experienced the major changes to its economy and political system after 1959, it would have grown much more rapidly than it did.
The authors then take another step, which is to try to disentangle how much of this underperformance is due to the embargo and how much to socialist economic policies. The authors’ idea for how to do this is to consider how much Cuban trade changed in the aftermath of the embargo. The rationale is straightforward: if the embargo hurts the Cuban economy by reducing trade below what it would have otherwise been, then we should be able to estimate the magnitude of that effect by appealing to existing estimates in the literature on the effects of trade on economic growth.
Once Bastos, Geloso, and Bologna Pavlik carry out this calculation, they arrive at a surprising result. The embargo can explain, at most, under optimistic assumptions, 8% of the difference between Cuba’s realized GDP and the GDP it would have had if it had behaved according to the synthetic control group.
Research Shortcomings
When I first read this paper, I found the reported results quite surprising. The authors recognize that Cuba faced a huge trade shock as a result of the embargo. In 1958, Cuba’s trade with the US represented 36% of its GDP, and this was nearly two-thirds of its total trade. Yet they still come up with an effect of a second-order magnitude.
There is, in fact, a vast literature linking trade and growth, and some of the most prominent papers in that literature suggest that trade has a very significant quantitative effect on growth. So how exactly do the authors find that losing access to trade amounting to more than a third of GDP has such a minor effect on income?
This puzzle led me to look more closely at the authors’ results. What I found, in a nutshell, is that their calculations are severely biased against finding an effect of the embargo for three reasons. First, they select a contribution to the literature that estimates a relatively low elasticity of income with respect to trade. Second, they focus on short-term rather than long-term effects. Third, they choose a decomposition that attributes to the embargo only part of the effects that could reasonably be attributed to it.
Bastos, Geloso, and Bologna Pavlik base their calculations on an estimate from a 2003 paper by Yanikkaya, which reports a semi-elasticity of growth with respect to trade of 0.018. That is, a one percentage point increase in trade—say from 55% to 56% of GDP—leads to a 0.018 percentage point increase in the growth rate.
Yanikkaya’s is one of hundreds of papers that have tried to estimate the effect of openness on growth, so the first question one is led to ask is how the authors selected that particular estimate. The authors do not provide a clear explanation. They claim that the estimate is similar to that of another recent paper by Raghutla, but there is a problem: they misread Raghutla’s estimate, which is 0.186, not 0.018, more than ten times larger. In any case, the estimates are not directly comparable: Raghutla estimates a long-run elasticity, while Yanikkaya estimates a short-run semi-elasticity. As a result, Yanikkaya’s specification captures a temporary, transitory effect on growth rather than a permanent one.
So how exactly do the authors find that losing access to trade amounting to more than a third of GDP has such a minor effect on income?
The authors also cite a recent meta-analysis by Heimberger (2022) and claim that their chosen elasticity is higher than those found in the rest of the literature surveyed in that paper. However, Heimberger provides no support for their claim: his analysis focuses on standardized coefficients, which do not provide information about the magnitude of absolute effects.
The other problems with Bastos, Geloso, and Bologna Pavlik’s calculations have to do with the horizon they adopt and their treatment of interaction effects. The authors assume a 12-year horizon for their calculations, which is particularly arbitrary given that they claim to be interested in understanding why Cuba is poor today.
They also adopt a decomposition of growth effects that attributes to socialist policies all of the interaction between policies and trade. To see why this matters, suppose that lifting the embargo doubled GDP, and that ending socialist policies also doubled GDP. If both changes were implemented simultaneously, GDP would quadruple. Yet under the Bastos et al. decomposition, two-thirds of the total effect would be attributed to socialist policies, and only one-third to the embargo.
Earlier Research
What is the result of addressing these problems? That is the question I explore in my recent working paper.
To do so, I chose to investigate estimates from some of the most well-known papers exploring the relationship between trade and growth. I do not claim to have conducted a systematic survey of the literature. There is a very large body of work on this topic, and comparing effect sizes across papers can be difficult, particularly when one is dealing with hundreds or possibly thousands of studies.
But neither am I cherry-picking. The four papers I use are extremely well known, highly cited, and widely accepted as seminal contributions to the field. More importantly, Bastos, Geloso, and Bologna Pavlik are explicitly claiming to establish an upper bound on the effect of the Cuban embargo. To dispute that claim, it is sufficient to show that their calculation yields much higher values when one substitutes their elasticity with those found in some of the most prominent contributions in the literature.
The papers I consider are Sala-i-Martin, Doppelhofer, and Miller (AER, 2004), Frankel and Romer (AER, 1999), Alcalá and Ciccone (QJE, 2004), and Feyrer (AEJ, 2019). The first three have accumulated thousands of citations, while the fourth, though more recent, is widely recognized for addressing key identification issues in the literature.
The results I obtain using these alternative elasticities suggest that the embargo is associated with a much wider and higher range of effects. In their intermediate scenario, Bastos et al. estimate that the embargo can explain only 6.5% of Cuba’s underperformance. In contrast, my estimates vary widely across the four papers, ranging from 34% to 124%. In fact, two of the papers (Alcalá and Ciccone and Feyrer) imply effects greater than 100% of the difference between Cuba and the synthetic control group. This would suggest that, in the absence of the embargo, Cuba would not have underperformed relative to this group, but might in fact have outperformed it.
Bastos et al. also refer to another exercise in their paper, in which they compare Cuban growth with that of formerly socialist economies after the collapse of the Soviet Union. Interestingly, they find that these economies outperform Cuba, but the difference is not statistically significant. Yet to me, it is not clear that the most salient difference between Cuba and former socialist economies is that they maintained different policies. It could also be that the countries of Eastern Europe did not have to contend with a punitive trade embargo from their most natural trading partners.
Think about it the following way: imagine that after the collapse of the Soviet Union, the countries of Western Europe had decided to ban all trade with former Soviet bloc countries, including by applying secondary sanctions to penalize anyone who traded with them. Would countries like Poland, Hungary, and the Czech Republic have experienced the same growth that they did if they had had to face such measures?
Understanding the effect that the embargo has played in Cuba’s economic underperformance relative to other factors (such as the choice of economic institutions) remains an important and relevant research question. Bastos et al. have put forward an interesting idea, which is to assess the embargo’s effects through its impact on the integration of the target economy into global trade.
Regrettably, their work falls short, primarily because of their choice of an elasticity estimate that is neither representative nor a reasonable upper bound of the values found in the literature. Nevertheless, this remains an important area for further research, in which future contributions may allow us to better understand the effects of these competing forces.
Francisco Rodríguez is a Venezuelan economist, who is a Senior Research fellow at the Center for Economic and Policy Research and a Faculty Affiliate at the Josef Korbel School at the University of Denver. He is also a member of the Bologna Initiative for Sanctions Relief.
Section: (all-articles) Photo: Canva