Mark Weisbrot and Jeffrey Sachs recently published a report about the potential effects of US sanctions on the humanitarian crisis in Venezuela. The authors established the following causal relationship: “The sanctions have inflicted, and increasingly inflict, very serious harm to human life and health, including an estimated more than 40,000 deaths from 2017–2018.”
According to the authors, the mechanism through which sanctions have led to the deaths of 40,000 Venezuelans is petroleum production. Their argument is summarized in two causal statements:
- The sanctions of August 2017 caused the collapse of Venezuelan oil production.
- This collapse reduced the access to foreign currency, which caused the humanitarian crisis.
This influential report has been utilized by many politicians and commentarists to argue against the use of economic pressures against the Venezuelan regime, and is presented as evidence that the humanitarian crisis in Venezuela is due to the economic war against the country rather than the economic policies of the Venezuelan government.
Given both the analytic nature of the report as well as the political importance of its conclusions, it is worth evaluating the methodology and assumptions used by the authors to substantiate the two causal statements, and compare them against academic standards with the goal of establishing causal relationships between social phenomena in an empirical way.
Brief notes about causality
The primary problem in detecting causal relationships with past data is that, by nature, the “what would have happened” with a variable if it had behaved differently, is never observable. For example, if someone eats a hamburger and later suffers a heart attack, it is difficult to establish that the hamburger caused the heart attack because we can’t know what their health would have been if they hadn’t eaten the hamburger – it is perfectly possible that they would have equally suffered a heart attack. This “what would have happened”, impossible to observe, is what social scientists define as the “potential outcome.”
If, in fact, the effect of a hamburger on a particular person’s health cannot be established, experimental studies with many people can estimate the average effect a high-calorie diet has on health. Experimental assignment (at random), is important due to the “selection bias.” If people with health problems tend to worsen their eating habits for emotional reasons, the average association between diet and sickness will not capture the causal relationship between diet and health, rather, it will capture a value that confuses this effect with the emotional mechanism that ties health and diet. It is for this reason that economists usually state “correlation does not imply causation.”
Relevance of these notes on the question of sanction effects
If one wishes to study the average effect of sanctions on sanctioned countries, an important issue to consider is that the allocation of sanctions does not occur on an experimental basis, but rather that these are due to political and economic events in the countries that receive and assign them. For example, if the imposition of sanctions is associated with authoritarian incidents or violations of human rights, and these usually occur during economic crises, the relationship between sanctions and crises will confuse the economic effect of sanctions with the fact that sanctions are also the consequence of economic crises through political mechanisms.
For this reason, economists try to “control” for factors that confuse causal relationships, as well as “implement” changes in the variable that measure the cause so that they are not responses to changes in the variable that measure the consequence. When this cannot be done, you simply cannot interpret any causal association.
However, the question that Weisbrot and Sachs ask is even more complex and difficult to answer. Their question is not about the average effect of the sanctions, but about the specific effect of the August 2017 sanctions on Venezuela. This is much more difficult than estimating the average effect of the sanctions because, as we mentioned earlier, Venezuela’s potential economic situation in the absence of sanctions cannot be seen.
Sanctions as a consequence of an accelerating crisis
Transferring these ideas to the approach of Weisbrot and Sachs, fundamental doubts arise as to whether they capture any causal effect.
Financial sanctions imposed on Venezuela were not assigned randomly. The trigger for the sanctions was the same that sought the creation of the Lima Group, also in August 2017. The Lima Group and the US financial sanctions were geopolitical responses to the already devastating and growing economic and humanitarian crisis that the country was experiencing at that time, accompanied by flagrant violations of human and civil rights of Venezuelans through the repression of the 2017 protests, and by the entry into effect of the Asamblea Nacional Constituyente (ANC) (National Constituent Assembly) as a supra-constitutional body.
It is perfectly possible that living conditions would have worsened in the period following the sanctions even without them. If this cannot be controlled for in some form, there is simply no way to approach a causal effect, and anything presented as such would be misleading. It is not only possible that things would have worsened in the absence of sanctions, but that they were on the cusp of doing so at the moment in which the sanctions were approved (and in part was what caused them). The logic of showing economic collapse pre-sanctions (as Hausmann and Muci do in Americas Quarterly) is to suggest that this dynamic would not be unforeseen in the absence of sanctions.
Sanctions were not an isolated event
In certain spheres it has been suggested that the econometric method of “difference in differences” could be used to address this question. This is not correct, because said method requires multiple treated and controlled units (countries sanctioned and not sanctioned), and seeks to estimate an average treatment effect (average effect of sanctions). In this case, we have only one unit treated (Venezuela), whose “potential result” can never be observed. For this reason, the question could only be addressed as a case study, through the use of comparative methods more prevalent in Political Science than in Economics, which do not seek to approximate a statistical measure of a causal effect.
An alternative statistical method that could be used in these cases is that of “synthetic control” for cases of comparative study. However, for this method to be useful, two things must happen that do not in this case: 1) there must be other units (countries) experiencing similar patterns in the consequence measuring variable (economic and social variables) before treatment (sanctions), and 2) the only relevant event in the unit of interest (Venezuela) at the moment of treatment (sanctions) is the treatment.
On one hand, there is no country in the world that has experienced a decline comparable to that of Venezuela in recent years. On the other hand, the sanctions were not an isolated event. There were multiple events that could have contributed to the worsening economic conditions around the period in which sanctions were approved. For example, the entry into effect of the Asamblea Nacional Constituyente (ANC), which led to the approval of the sanctions and the creation of the Lima Group, could have produced a degree of legal uncertainty that generated further deterioration.
Do we know what the effect the ANC had on the living conditions in the country? No. Any estimate of the economic effect of the ANC would be misleading. Well, it’s exactly the same for sanctions.
Is Colombia a good reference for Venezuelan oil production?
The only evidence that Weisbrot and Sachs show to assert that sanctions reduced oil production is the following figure. Adapted from the analysis of Francisco Rodríguez in WOLA, the figure shows Venezuelan oil production in contrast with that of Colombia, and notes the financial sanctions of 2017 as the only relevant event in August 2017.
The figure gives the following lecture: Venezuela’s production had been falling before the sanctions, but that was not the fault of oil production management, as Colombia’s production had also been falling due to the reduction of oil prices. As Colombia and Venezuela produce high-cost heavy oil, it makes sense that its production reduces in periods of low prices. However, once the sanctions were in place, we see that Colombia’s production was stable while that of Venezuela fell. Not only is chavismo (chavism, chavezism) excused from the fall of production pre-sanctions, but the sanctions are the culprit of the following drop.
The entire previous discussion suggests that presenting the result of this exercise as a causal effect of oil production sanctions is misleading. Throwing out a number casually and saying “this could be the effect of sanctions” is irresponsible when the circumstances suggest that the real effect could very well be zero. However, it is worth considering Weisbrot and Sach’s argument in the sense that they also seek to excuse the regimen for the fall in production before the sanctions.
The assumption of Weisbrot and Sachs is that Colombian oil production is a good “counterfactual” of Venezuelan oil production. A counterfactual is a combination of untreated units (unsanctioned) that tries to approximate the “potential result” of the treated units in the absence of treatment (oil production in Venezuela if the sanctions had not have occurred).
The principal reason that they present is that a large portion of Colombian oil production is based on heavy crudes with similar costs to those of Venezuela, which are rendered commercially inviable with relatively low prices. However, production in Colombia fell in the pre-sanction period for idiosyncratic judicial and social reasons [1]. According to data from EIA (U.S. Energy Information Administration), Canada, our main competitor in extra heavy crudes, increased their production by 8.8% between January of 2015 and August of 2017. This is not due to the fact that Canada is an exceptional example of producers with high marginal costs. The United States, with high marginal costs due to their concentration in extraction of crude oil and shale gas, increased its production during this period by 5.3%. Brasil, another neighbor with high marginal costs of offshore oil extraction, increased its production at an impressive 29.5% during the same period [2]. If we look at OPEP production as a whole, it increased 8%. World oil production increased 3.1%.
In the following figure we see the rate of Venezuelan, Colombian and Canadian production alongside the figures of oil prices. As you can see, the Canadian figures vary with the price of oil; that is to say, that its level of production is affected by price. However, we see that Colombian and Venezuelan production went down without apparent correlation to the price level. Weisbrot and Sachs make note of the moment in which oil went below $30 per barrel price, but do not note the series. By doing this they do not show that oil price was only below $30 for a couple of months at the beginning of 2016. Prices increased the rest of 2016 while Colombian and Venezuelan production fell.
Pre-Sanction Dynamics vs. Post Sanction Dynamics
If we use world production as a reference, we see that there is a clear downward trend of the Venezuelan production prior to the sanctions, relative to the events of the oil market in aggregate. An analysis based on OPEC (Organization of the Petroleum Exporting Countries) production or a weighted average number of countries with high marginal costs would produce similar results.
Furthermore, Weisbrot and Sachs show no evidence that production declines within Venezuela have come from extra heavy oil production fields, which would be the most vulnerable to price declines for commercial reasons. If the fall in Venezuelan production were a consequence of low prices, one would expect the contraction to be observed only in extra-heavy crude with marginal costs close to those prices. Put another way, the production of medium and light crudes, with low marginal costs, would offer a “placebo”: If production also falls in these fields, it is unlikely that the reason for the decrease would be low prices.
Unfortunately there is no high frequency information about the production by field or type of field. However, the IPD consultant reported that during the first quarter of 2016 (that of the lowest prices) there were decreases in both the Orinoco Petroleum Belt (extra-heavy crude oil handled by the joint ventures) and in other producing regions of the country, concentrated medium and light crudes and handled mostly by Petróleos de Venezuela, S.A PDVSA (Petroleum of Venezuela). According to the consultant, the reasons for the fall in production were “challenges in drilling, compression problems of natural gas, difficulties in the maintenance of wells due to restriction of field services, and theft” [3]. That is, when the problem of low prices and commercial viability of extra heavy crude was more threatening, production also fell in the medium and light fields, and the reasons mentioned to explain the fall were strictly operative and managerial.
The conclusion of this analysis is that Colombia is not a good counterfactual for Venezuelan oil production, which should not have fallen in the period prior to the sanctions: The Colombian decrease was due to idiosyncratic reasons. Other producers with high marginal costs did not reduce their production, and the loss of Venezuelan production was due to managerial reasons, not commercial reasons. Using Colombia as a reference is to select a case to excuse Chavismo from responsibility for the fall in oil production before the sanctions, and to attribute the sanctions for any subsequent fall.
Pre-Sanction Dynamics vs. Post Sanction Dynamics
If we use world production as a reference, we see that there is a clear downward trend of the Venezuelan production prior to the sanctions, relative to the events of the oil market in aggregate. An analysis based on OPEC (Organization of the Petroleum Exporting Countries) production or a weighted average number of countries with high marginal costs would produce similar results.
Establishing the difference between pre-sanction and post-sanction dynamics can be useful to give a range of values to the losses suffered that definitely have nothing to do with sanctions. Just as the future can not cause the past, the sanctions did not cause the downward trend in oil production or the Venezuelan crisis in general because it originated much earlier. Using world production and the trend of pre-sanction production as a reference, we can estimate the losses of oil production in Venezuela between 2015 and 2018, as well as the portion of these due to reasons prior to the sanctions.
Here we see the accounting of the Venezuelan monthly production losses compared to the world benchmark. As can be clearly seen, the bulk of all the losses incurred by the Republic of 2015 onwards (including the period following the sanctions) is explained by dynamic pressure.
The summation of these losses throughout the period suggests that between 2015 and 2018 production of 635 million barrels of oil was stopped. 76% of the total production losses between 2015 and 2018 are due to trends prior to the sanctions.
Reiterating the above, saying that the losses associated with subsequent deteriorations are a consequence of the sanctions would at the very least be misleading. In addition to the above, a fundamental factor is that in November 2017, Nicolás Maduro appointed Manuel Quevedo president of PDVSA, establishing military control over the Venezuelan oil industry. There is countless anecdotal evidence of how the appointment of Quevedo and the military disruption to the management of PDVSA and the oil industry generated huge additional bottlenecks to the already affected oil production during 2018 [4].
Given that many relevant events occurred in the same period, it is impossible to impute causal effects to any of them. However, in the figures shown above, it can be seen that the trend towards falling production does not accelerate significantly between August and November (the months between the sanctions and the appointment of Quevedo to the presidency of PDVSA), but later. This factor, added to the aforementioned anecdotal evidence, and to the fact that the sanctions of August 2017 did not prevent oil trade relations between the US and Venezuela, lead us to speculate that the main cause of any acceleration in the fall in production is associated with the military’s management of the oil industry.
But beyond any speculation, points previously mentioned refer to the fact that the bulk of losses incurred by the country due to the fall in oil production is definitely due to pre-sanction trends. If Weisbrot and Sachs really wanted to understand the determinants of the collapse in oil production and the economic and humanitarian crisis in Venezuela, they should focus on studying possible causes prior to sanctions.
Given that many relevant events occurred in the same period, it is impossible to impute causal effects to any of them. However, in the figures shown above, it can be seen that the trend towards falling production does not accelerate significantly between August and November (the months between the sanctions and the appointment of Quevedo to the presidency of PDVSA), but later. This factor, added to the aforementioned anecdotal evidence, and to the fact that the sanctions of August 2017 did not prevent oil trade relations between the US and Venezuela, lead us to speculate that the main cause of any acceleration in the fall in production is associated with the military’s management of the oil industry.
But beyond any speculation, points previously mentioned refer to the fact that the bulk of losses incurred by the country due to the fall in oil production is definitely due to pre-sanction trends. If Weisbrot and Sachs really wanted to understand the determinants of the collapse in oil production and the economic and humanitarian crisis in Venezuela, they should focus on studying possible causes prior to sanctions.
Did the sanctions cause the death of 40,000 Venezuelans?
The most controversial claim in the article by Weisbrot and Sachs is that as sanctions led to the fall of oil production, the corresponding fall in income led to the death of 40,000 Venezuelans.
The most controversial claim in the article by Weisbrot and Sachs is that as sanctions led to the fall of oil production, the corresponding fall in income led to the death of 40,000 Venezuelans.
The authors say that the calculation of the 40,000 deaths is made with a rule of 3 based on estimates of general mortality of Encuesta Nacional de Condiciones de Vida ENCOVI (Living Conditions Survey) and an internal report of the United Nations. However, these estimates are not reported. Since we do not have the factors they used to make their calculations, we cannot replicate them – something that is standard for the academy.
As we mentioned above, to say that the fall of the Venezuelan oil production is a consequence of the sanctions is deceptive, and the article by Weisbrot and Sachs does not prove it. Consequently, in the absence of a causal mechanism, it is concluded that they do not prove any effect of the sanctions on mortality in Venezuela. And even if the sanctions had an effect on oil production, the authors do not show any statistical evidence that measures the effect of oil events on social variables, separating their effect from possible alternative explanations.
This is a very long way of saying that they aren’t doing anything of what they should be doing to prove that sanctions killed 40,000 people. They simply state the figure of 40,000 deceased and impute it to the sanctions by default. To say that any social phenomenon cost the lives of 40,000 people without any evidence is really bold – even more so for two development economists who present these results under an aura of academic credibility.
As we stated above about oil production, any study on the impact of sanctions on mortality in Venezuela should consider the trends at the time of them. The following figure, for example, shows the change in infant mortality [5] between the year 2000 and 2017 for all the countries in the region. We see that Venezuela increased its infant mortality rate by 76% between 2012 and 2017 (before the sanctions), and that this increase was accelerating at the time of the sanctions. Infant mortality was reduced in all other countries of the region. This same dynamic of aggressive deterioration is shown by Hausmann and Muci (citing Santos y Bustos) in relation to food and medicine imports in Venezuela, whose downturn was already accelerating at the time of the sanctions.
By way of closure
Weisbrot and Sachs do not prove that the sanctions affected oil production. They do not even try to analytically justify their claim that they caused 40,000 deaths in Venezuela. Given its methodological fragility, the analysis presented by the authors offers no insight about the causal effect of the sanctions on the Venezuelan economy. Being both PhDs in Economics, from two of the best universities in the world, the fact that Weisbrot and Sachs demonstrate such bold causal affirmations without the minimum necessary evidence is disgraceful.
The analysis elaborated in this rebuttal shows that the debate Venezuelan society needs is less about the effect of sanctions, and more about the causes of the horrendous crisis that the country is going through. If the authors were willing to contemplate any cause prior to the sanctions, they would not lack plausible explanations.
***
1: Here is a 2016 discussion about the domestic reasons for the fall in Colombian production.
2: The following references compare oil production costs in Venezuela with Canada, the US, Brazil, and other producers with high marginal cost.
3: See citation here.
4: The following Reuters report provides substantial anecdotal evidence on the effect of the military leadership of the Venezuelan oil industry.
5: Infant mortality rate: Number of infants dying up to one year after birth for every 1,000 infants born. Source: World Development Indicators of the World Bank.
