Abstract: Bahar and Hausmann (2025a) claim that increases in Venezuelan oil revenues lead to higher emigration flows using data on irregular migration flows into the United States. We show that their results are driven by a flawed empirical specification that appears to be the result of a coding error. Instead of using year-on-year variations, Bahar and Hausmann used a complex differencing operator with no meaningful economic interpretation. Once we correct this mistake, their central result disappears. Furthermore, to the extent that there is any positive correlation between Venezuelan migration and oil prices, it appears to reflect the influence of a key omitted variable: economic conditions in the United States, which were also highly correlated with global oil prices during the period of study.

In a recently published working paper and accompanying op-ed, Dany Bahar and Ricardo Hausmann argue that Venezuelan oil revenues are positively associated with out-migration of Venezuelans.¹ They reach this conclusion based on their analysis of time series data on monthly US border encounters involving Venezuelan migrants and Venezuelan oil revenues and prices. They claim to find a positive relationship between migration and both oil revenues and prices and interpret this result as indicating that greater oil revenues, by stabilizing the Venezuelan government, lead more people to decide to flee the country.

The policy implications of this conclusion are, in fact, far-reaching. If depriving the Venezuelan economy of oil revenues leads to less — not more — migration, then, Bahar and Hausmann argue, “there is no trade-off between actions in favor of weakening Maduro and migration risks.”² Lower oil revenues caused by sanctions, they argue, will raise the likelihood of regime change perceived by people living in Venezuela and lead them to remain at home in the expectation that such change might materialize. Furthermore, making the economy poorer might also lower out-migration by making people less capable of migrating. In the words of Bahar and Hausmann, “migration is a costly investment, and lower income might make it less affordable.”³

The normative implications, however, of the authors’ claims are concerning. First, it appears morally problematic to advocate for intentionally damaging the Venezuelan economy on the premise that further impoverishing Venezuelans will lessen migration by trapping them inside a collapsed economy. Furthermore, it is unclear why Venezuelans, who saw Nicolás Maduro’s grip on power strengthen during a prior episode of punitive economic sanctions, would now expect lower oil revenues to drive him from power. In fact, the empirical literature on the effectiveness of sanctions generally finds that these are relatively inefficient in generating regime change,⁴ while Idrobo⁵ has recently used administrative data to show that sanctions are more likely to lead opposition supporters to emigrate from Venezuela, thus further entrenching the power of the regime.

The case of Venezuela is also a peculiar one to present as an illustration of the Bahar-Hausmann hypothesis that higher oil revenues lead to more emigration. Venezuela, a country with traditionally low emigration rates, experienced the largest exodus ever documented in the history of the Western Hemisphere during the late 2010s — a period which coincided with a 93 percent collapse in oil revenues and a 71 percent decline in per capita gross domestic product (see Figure 1). But if Bahar and Hausmann were correct, then one would expect that emigration from Venezuela would have fallen, not risen, when oil revenues collapsed. It is unclear how Bahar and Hausmann account for these facts.

Figure 1: Venezuelan Per Capita Income and Emigration, 2010–2024

The findings of Bahar and Hausmann are puzzling in light of the existing empirical evidence on economic sanctions, migration, and Venezuela’s economic collapse. A large body of literature has documented the significant negative effects of sanctions on living conditions in target countries.⁶ In particular, there is strong evidence that sanctions lead to recessions and are associated with greater probability of growth collapses.⁷ At the same time, prior analyses have found that sanctions directly contribute to increased emigration and that recessions tend to lead to higher near-term out-migration rates.⁸ The claim by Bahar and Hausmann that sanctions contribute to lowering emigration rates appears inconsistent with these strands of the literature.

Bahar and Hausmann⁹ refer to the so-called migration hump hypothesis — which posits that below a certain income threshold, lower incomes might reduce emigration by making it harder for individuals to finance the costs of migrating — in support of their hypothesis that short-term changes in oil income are associated with higher migration rates. However, the migration hump hypothesis is a theory about the long-term effects of changes in per capita income, not about short-term income fluctuations. Recent models of the migration hump hypothesis distinguish between these long-term effects, which operate through accumulation of human and physical capital and shorter-term effects.¹⁰ For example, Clemens and Mendola¹¹ show that the propensity to emigrate is declining in income levels for given levels of human capital so that in the near term an adverse economic shock can increase out-migration, while in the longer term such a shock, by making education less accessible, can lower the propensity to emigrate. Tests of this hypothesis on monthly data over a four-year period, as presented by Bahar and Hausmann, can only evaluate claims about short-term effects of economic shocks, not about the hypothesis of the long-term effects of development on migration that the migration hump hypothesis is meant to explain.

This distinction is perhaps best illustrated by a recent empirical debate on the migration-development nexus. In a recent paper, Benček and Schneiderheinze¹² argue that, in contrast to the predictions of the migration hump hypothesis, there is a robust negative relationship between emigration and per capita income in cross-country panel data after controlling for country fixed effects. Clemens¹³ has criticized these findings precisely on the grounds that they do not speak to the long-term relationship between income and development. At the same time, Clemens has noted that the findings accurately describe the effects of short-term shocks, citing Venezuela as an example where we should expect there to be a negative correlation between emigration and income. According to Clemens:¹⁴

The [Benček and Schneiderheinze] paper teaches us about the correlation between economic shocks (an oil boom, for example, or a currency crisis) and emigration. If income suddenly drops much lower than the normal trend—as recently happened in Venezuela—we can expect emigration to rise more rapidly.¹⁵ (emphasis added)

One more reason why the findings of Bahar and Hausmann are puzzling is because their data choices appear poorly suited to evaluate their hypothesis. Bahar and Hausmann argue that increases in oil revenues in Venezuela lead to higher out-migration from the country. However, to capture out-migration from Venezuela, they employ data on border encounters of Venezuelans in the United States. A substantial share of the Venezuelans who have crossed into the US over the past four years did not come directly from Venezuela; many had previously settled in third countries — such as Colombia, Peru, or Mexico — and later decided to migrate again due to worsening conditions for migrants in those countries, coupled with a faster post-COVID economic recovery in the US than in much of Latin America.¹⁶ It is therefore unclear why Venezuelans residing in other nations of Latin America would decide to leave those countries in response to sanctions-induced increases in oil revenues accruing to the Venezuelan state. Furthermore, it is curious, given their use of US-centered data, that Bahar and Hausmann fail to engage with the simpler hypothesis that the increased rate of encounters is tied to improved economic conditions in the US. This is especially interesting given that they cite Bahar¹⁷ for methodology yet overlook Bahar’s own conclusion that crossings to the Southwest US border “naturally decrease as the labor market cools.”¹⁸

In this research note, we argue that there is no contradiction between the patterns observed in Venezuelan migration — as captured by apprehensions at the US southern border — and prior theoretical and empirical findings that lead us to expect sanctions-induced export declines to increase migration. We show that critical results reported by Bahar and Hausmann are driven by an incorrect empirical specification, apparently caused by an inadvertent coding error. While Bahar and Hausmann claim to analyze year-on-year variations of migration and oil revenues, they in fact analyze the 12th difference of monthly immigration rates — a calculation that, to the best of our knowledge, has no meaningful economic interpretation in this context. Once Bahar and Hausmann’s coding errors are corrected, their central results disappear. We furthermore show that any remaining patterns in their data are the product of spurious correlations, driven by the strong relationship between global oil prices and US economic conditions.

Replication and Analysis of Bahar-Hausmann Results

Bahar and Hausmann¹⁹ estimate four regressions to analyze the relationship between apprehensions of Venezuelan immigrants at the US southern border and Venezuelan oil revenues. These are given by the combination of two empirical specifications — a levels and a differences estimation — and two measures of the key explanatory variable — oil revenues and oil prices. Their levels specification is given by

where crossings refers to apprehensions of Venezuelan nationals at the US southern border, oil refers to the measure of oil revenues, and m and y are, respectively, month and year fixed effects, while the subindices m and y refer to the month and year of the data. Bahar and Hausmann also present an autoregressive distributed lag specification given by

where Δ denotes the year-over-year difference of a variable,

that is, the difference between its level at time t and its level in the same month of the previous year, t−12. Such differences — also commonly referred to as 12-month differences or seasonally differenced data — are, of course, commonly used in macroeconomics as a form of seasonal adjustment. The specification captures the hypothesis that migration patterns should depend on lagged migration, current and lagged oil revenues, and an error-correction term (ECT) which represents the extent to which current migration levels deviate from their long-term relationship with oil revenues.

For each model, Bahar and Hausmann separately employ two measures of oil revenues. First, they specify the value of Venezuelan oil production, given by the product of multiplying Brent oil prices times Venezuelan production levels. Alternatively, they use the level of Brent oil prices, which they argue suffers from less endogeneity than oil income.²⁰

We begin by considering the specification in seasonally differenced data given by Equation (Eq.) (2). Despite their claim to estimate Eq. (2) using year-over-year log differences, Bahar and Hausmann do not actually estimate this specification. Instead, due to what appears to be an inadvertent coding error, they use Δ¹² instead of Δ in their regressions, where Δ¹² is the 12-th difference operator, i.e.:

Δ¹² is the discrete time analogue of the 12-th derivative of a function. If the first-difference Δ¹ captures the change between and and the second-difference captures the change between the first differences, i.e.,

then the 12th-difference Δ¹² is the rate of change in the 11th-difference Δ¹¹, which in turn is the rate of change in the 10th-difference, and so on.

To the best of our knowledge, there is no economic interpretation of the 12th-difference of time series with meaningful application to the study of migration or its relationship with sanctions-induced changes in economic conditions in countries of origin. In fact, we are not aware of any other prior application of 12th difference measures in applied economics. Nor do Bahar and Hausmann claim that such an interpretation exists. Rather, they claim to be using the seasonally differenced year-over-year variations Δ given by Eq. (3) while instead applying the 12th-differences Δ¹² given by Eq. (4). This mistake appears to have been caused by using the Stata time series prefix D12, which applies to differences of differences, rather than the appropriate prefix S12, which corresponds to seasonal differences.²¹ Yet as the Stata User’s Guide explains, the seasonal and the difference operator do not coincide whenever the difference is taken over more than one period.²²

That the Bahar and Hausmann series do not correspond to seasonal differences is apparent from a visual inspection of the figures they published. In particular, the right two panels of their Figure 1, replicated  in the left of our Figure 2 below in a format similar to that used in their paper, make this clear.²³ The range of variation spans from minus 20 million to plus 40 million for encounters, and from minus 20,000 to plus 20,000 for oil prices. In each case the range of differences is many orders of magnitude larger than the range of levels — an impossible outcome. These clearly cannot be seasonal differences — that is, they cannot represent the simple change between the value of the variable in a given month and its value 12 months earlier.

Figure 2: Original and Corrected Bahar-Hausmann Figures

Figure 2 also shows on the right-hand side the corrected graphs estimated using appropriate seasonal (12-month) single differences. These graphs display reasonable ranges of variation for year-over-year changes. Importantly, they reveal no obvious correlation between the series. For example, oil prices rise substantially between mid-2022 and mid-2023, yet migration levels remain largely unchanged.

Table 1 below shows the effect of reestimating Bahar and Hausmann’s differenced specification, given by Eq. (2) above but using the correct seasonal differences.²⁴ Columns 1 and 2 report results from the original Bahar-Hausmann 12th-difference specification, showing significantly positive effects of changes in oil revenues on changes in Venezuelan emigration to the US. However, once we correct these specifications — as shown in Columns 3 and 4 — we find no statistically significant effects.

Sources: Authors’ elaboration based on Bahar and Hausmann (2025a). Standard errors in parentheses. Asterisks denote significance levels: *−10%, **−5%, ***−1%. All regressions are estimates of Eq. (2).

It is worth noting that the point estimates obtained in our corrected specification, while still positive, are much smaller both in absolute values and in relation to their standard errors. The coefficients corresponding to the oil incomes and price specifications are between one-third and one-seventh the size of the corresponding standard errors, with associated p-values of 0.73 and 0.89, respectively. These specifications are therefore consistent with a wide range of possible relationships between oil revenues and migration: the effects could be large and positive, large and negative, or close to zero. In other words, these results suggest that the data that Bahar and Hausmann proposed using, specifically, the 35 observations of year-over-year changes in border apprehensions of Venezuelans and oil revenues or prices, are not sufficiently informative to draw any conclusions about the validity of competing hypotheses regarding the relationship between sanctions-induced changes in export revenues in countries of origin and emigration to the United States.

We now return to the levels specification estimated by Bahar and Hausmann (Eq. [1]). This specification is not affected by the 12th differencing operation. The results, shown in the first two columns of Table 2 below, indicate a borderline significant relationship between oil revenues or prices and apprehensions of Venezuelan nationals at the US southern border, after controlling for month and year fixed effects.²⁵

The problem with this specification, as Bahar and Hausmann themselves acknowledge, is that a simple time series regression of two variables — particularly over a short time span — is prone to generating spurious correlations. Even if the variables are unrelated, nonstationarity can lead them to exhibit correlated stochastic trends, leading to misleading significant relationships. The fact that time series regressions are prone to such spurious correlations is well-known in econometrics, and it is precisely why levels regressions of trending variables are rarely used to test causal hypotheses.²⁶ Attenuating this unwanted effect is exactly the rationale behind the above difference model of Eq. (2). The fact that this model, when properly estimated, yields statistically insignificant coefficients — and whose point estimates are orders of magnitude smaller than in the levels specification — adds additional evidence in favor of the hypothesis that the results of the levels model are spurious.

Nevertheless, one could argue that — particularly given the lack of informativeness of the differenced specification — the presence of a correlation in the estimation of Eq. (1) is at least noteworthy, as it suggests that migration is increasing at the same time as oil prices or oil revenues are rising. While any causal interpretation should be approached with caution and strong policy conclusions should be avoided, the result may still appear counterintuitive, given the expectation that declines in oil revenues would lead to increased migration.

However, it is worth bearing in mind that hypotheses about emigration must properly consider not only changes in conditions in countries of origin but also, importantly, changes in conditions in countries of destination. In the language of migration studies, they must account for what are commonly referred to as “push” and “pull” factors.²⁷ When assessing a correlation such as the one discussed here, it is essential to examine whether the most reasonable interpretation traces it to conditions in the country of origin or of destination.

This consideration is particularly relevant given the nature of the data used by Bahar and Hausmann, which refers to apprehensions of Venezuelan nationals attempting irregular entry into the United States via the Mexico border. In this context, it is reasonable to expect that conditions in the United States play an important role in shaping migration trends. Moreover, as we have already discussed, many of the Venezuelan nationals attempting to enter the US were not residing in Venezuela immediately prior to their decision to migrate. This suggests that conditions in Venezuela may not be the most relevant determinant of their migration behavior. Consequently, any correlation between Venezuelan oil revenues and migration in this series could potentially reflect broader variations in global economic conditions that influence both oil revenues and the attractiveness of migrating to the United States.

Figure 3 below illustrates one potential reason for concern: It shows a striking negative relationship between Brent oil prices — the measure that Bahar and Hausmann claim is least contaminated by endogeneity for the purposes of their analysis — and the US unemployment rate, a standard indicator of labor market conditions. A low unemployment rate reflects a tighter labor market, which should make it more attractive for migrants to attempt to enter the United States.

Notably, US unemployment was high during the early phase of the COVID pandemic, which also marks the beginning of the period analyzed by Bahar and Hausmann. During this same period, oil prices were low due to the global COVID-induced recession, and they began to recover at the same time that the US labor market improved. This pattern offers the alternative hypothesis that what is driving increased migration by Venezuelan nationals is not higher oil revenues in the hands of the Maduro regime but rather the broader recovery of the US economy between 2020 and 2024. Put differently, while global oil prices may be reasonably exogenous to economic conditions in Venezuela, they are clearly endogenous to economic conditions in the United States, a country which accounts for one-fifth of global oil consumption — and whose economic recovery at the time closely tracked that of other large economies that also are key sources of global oil demand.²⁸

Figure 3: Oil Prices and US Unemployment Rate, 2020–2024

Sources: Authors’ elaboration based on US Energy Information Administration (2025) and FRED (2025).

Table 2 below evaluates this hypothesis more systematically. Columns 1 and 2 reproduce the results from Bahar and Hausmann, which show a positive, borderline significant relationship between emigration of Venezuelan nationals to the United States and oil revenues or prices. However, these regressions do not include alternative explanatory variables, such as indicators of US labor market conditions.

In Specification 3, we run the same specifications but use the US unemployment rate instead of Venezuelan oil revenues and find a strong, statistically significant negative relationship between unemployment and Venezuelan migration to the US. Importantly, the R2 of the regression increases once the US unemployment rate is included, suggesting that its explanatory power is greater than that of Venezuelan oil revenues. In other words, using Bahar and Hausmann’s own specification, we could argue that Venezuelan migration is not driven by oil prices or revenues in Venezuela but rather by a simpler and more intuitive hypothesis: More people want to migrate to the United States when there are more jobs available there.

Specifications 4 and 5 test these competing hypotheses against each other by including both Bahar and Hausmann’s measures of oil revenues and prices alongside the US unemployment rate. As expected — given the strong negative correlation illustrated in Figure 3 — there is substantial collinearity between the variables. This inflates standard errors and weakens the precision of both estimates. Nevertheless, the coefficient on US unemployment remains closer to statistical significance, with a t-statistic close to two, whereas the coefficients on oil revenue variables show point estimates of similar magnitude to their standard errors and thus far from conventional significance thresholds.

Sources: Authors’ elaboration. Standard errors in parentheses. Asterisks denote significance levels: *−10%, **−5%, ***−1%. All regressions are estimates of Eq. (1) and thus include month and year fixed effects.

In sum, the limited data on border apprehensions do not allow us to definitively distinguish between the competing hypotheses that migration is driven by changes in Venezuelan oil revenues or by US labor market conditions. However, given the extensive literature showing that labor market conditions at destination significantly influence migration flows,²⁹ the most reasonable interpretation of this pattern is that the observed correlation between Venezuelan migration and oil revenues is a spurious one, arising from the omission of a key explanatory variable: economic conditions in the destination country.³⁰

Conclusion

We have shown that the hypothesis put forward by Bahar and Hausmann — that higher oil revenues for Venezuela lead to increased emigration — finds no conclusive support in the data that they use. The inferences they draw are based on an incorrect empirical specification, apparently stemming from a coding error. Correcting the error results in small and insignificant point estimates, and the size of the associated confidence intervals suggests that this limited dataset lacks enough information to allow us to meaningfully evaluate hypotheses about determinants of Venezuelan migration. Moreover, the results that are not affected by this methodological mistake are still distorted by the failure to account for US economic conditions as a potential driver of immigration.

In short, the correlations presented by Bahar and Hausmann do not demonstrate that increases in Venezuelan oil revenues cause higher emigration, as they suggest. Rather, they appear to reflect the fact that improved US labor market conditions — something that is itself correlated with global demand and oil prices — make migration to the United States more attractive. Taking this into account, the alleged positive link between Venezuelan oil revenues and migration that is the central piece of evidence for their hypothesis disappears.

The stakes involved in correctly understanding the relationship between sanctions and migration are significant. If the decision to impose maximum pressure sanctions on Venezuela once again plunges the country into a deep recession, it could trigger a severe deterioration in living standards — this in a country that has already endured one of the worst humanitarian crises documented in modern economic history. It is therefore imperative that US foreign policy in this area be grounded in solid evidence and rigorous research. Regrettably, the recent work by Bahar and Hausmann does not strengthen this evidence base.

References

Attia, Hana, Julia Grauvogel, and Christian von Soest. 2020. “The Termination of International Sanctions: Explaining Target Compliance and Sender Capitulation.” European Economic Review 129: 103565. https://doi.org/10.1016/j.euroecorev.2020.103565

Bahar, Dany. 2024a. “Not a Border Crisis, but a Labor Market Crisis: The Often Overlooked ‘Pull’ Factor of U.S. Border Crossings.” Journal of Policy Analysis and Management: 1–7. https://doi.org/10.1002/pam.22665

Bahar, Dany. 2024b. “Replication Data for: Not a Border Crisis, but a Labor Market Crisis: The Often Overlooked ‘Pull’ Factor of US Border Crossings.” Version 2, Harvard Dataverse. December 4. https://doi.org/10.7910/DVN/HEJWXW.

Bahar, Dany, and Ricardo Hausmann. 2025a. “Sanctions on Venezuela Are Not Driving Migration to the US Southwest Border: An Empirical Assessment.” CGD Note 382. Washington, DC: Center for Global Development. https://www.cgdev.org/sites/default/files/sanctions-venezuela-not-driving-migration-southwest-border.pdf

Bahar, Dany, and Ricardo Hausmann. 2025b. “Stopping Venezuela’s Exodus Hinges on Restoring Hope.” Americas Quarterly, March 24. https://americasquarterly.org/article/stopping-venezuelas-exodus-hinges-on-restoring-hope/

Benček, David, and Claas Schneiderheinze. 2020. “Higher Economic Growth in Poor Countries, Lower Migration Flows to the OECD: Revisiting the Migration Hump with Panel Data.” Kiel Working Papers 2145. Berlin: Kiel Institute for the World Economy. https://www.ifw-kiel.de/publications/higher-economic-growth-in-poor-countries-lower-migration-flows-to-the-oecd-revisiting-the-migration-hump-with-panel-data-21624/

Benček, David, and Claas Schneiderheinze. 2024. “Higher Economic Growth in Poor Countries, Lower Migration Flows to the OECD – Revisiting the Migration Hump with Panel Data.” World Development 182, 10665. https://doi.org/10.1016/j.worlddev.2024.106655

Caballero Reina, Juan, Jesus Crespo Cuaresma, Katharina Fenz, Jakob Zellmann, Teodor Yankov, and Amr Taha. 2024. “Gravity Models for Global Migration Flows: A Predictive Evaluation.” Population Research and Policy Review 43, no. 29. https://doi.org/10.1007/s11113-024-09867-6

CBP. See US Customs and Border Protection

Chiswick, Barry, and Timothy J. Hatton. 2003. “International Migration and the Integration of Labor Markets.” In Globalization in Historical Perspective, edited by Michael D. Bordo, Alan M. Taylor, and Jeffrey G. Williamson, 65–117. Chicago, IL: University of Chicago Press.

Clemens, Michael. 2020. “The Emigration Life Cycle: How Development Shapes Emigration from Poor Countries.” IZA Discussion Papers, No. 13614. Bonn: Institute of Labor Economics (IZA). https://www.econstor.eu/bitstream/10419/224056/1/dp13614.pdf

Clemens, Michael, and Mariapia Mendola. 2020. “Migration from Developing Countries: Selection, Income Elasticity, and Simpson’s Paradox.” CGD Working Paper 539. Washington, DC: Center for Global Development.

Czaika, Mathias, and Constantin Reinprecht. 2022. “Migration Drivers: Why Do People Migrate?” In Introduction to Migration Studies, edited by P. Scholten. Springer, Cham. https://doi.org/10.1007/978-3-030-92377-8_3

Federal Reserve Economic Data (FRED). 2025. “Unemployment Rate, Percent, Monthly, Seasonally Adjusted.” Federal Reserve Economic Data, Federal Reserve Bank of St. Louis. Accessed March 27, 2025. https://fred.stlouisfed.org/series/UNRATE

Felbermayr, Gabriel, Clifton Morgan, Constantinos Syropoulos, and Yoto V. Yotov. 2021. “Understanding Economic Sanctions: Interdisciplinary Perspectives on Theory and Evidence.” European Economic Review 135: 103720. https://doi.org/10.1016/j.euroecorev.2021.103720

FRED. See Federal Reserve Economic Data

Goodman, Joshua. 2021. “Driven by Pandemic, Venezuelans Uproot Again to Come to US.” Associated Press, October 10. https://apnews.com/article/venezuela-immigration-business-coronavirus-pandemic-health-72e16118a21cf9ae3d0d4c9204f24643

Granger, Clive, and Paul Newbold. 1974. “Spurious Regressions in Econometrics.” Journal of Econometrics 2, no. 2: 111–120.https://doi.org/10.1016/0304-4076(74)90034-7

Grogger, Jeffrey, and Gordon Hanson. 2011. “Income Maximization and the Selection and Sorting of International Migrants”. Journal of Development Economics 95: 42–57. https://doi.org/10.1016/j.jdeveco.2010.06.003

Gutmann, Jerg, Matthias Neuenkirch, and Florian Neumeier. 2023. “The Economic Effects of International Sanctions: An Event Study.” Journal of Comparative Economics 51, no. 4: 1214–1231. https://doi.org/10.1016/j.jce.2023.05.005

Gutmann, Jerg, Pascal Langer, and Matthias Neuenkirch. 2024. “International Sanctions and Emigration.” Journal of Economic Behavior & Organization 226: 106709. https://doi.org/10.1016/j.jebo.2024.106709

Hatton, Timothy, and Jeffrey Williamson. 2008. Global Migration and the World Economy. Cambridge, MA: The MIT Press.

Hufbauer, Gary, Jeffrey J. Schott, Kimberly Ann Elliott, and Barbara Oegg. 2007. Economic Sanctions Reconsidered, 3rd ed. Washington, DC: Peterson Institute for International Economics.

Idrobo, Nicolás. 2024. “Why Economic Sanctions Backfire: The Role of Emigration in the Venezuelan Case.” September 29. https://idrobo.github.io/files/NicolasIdrobo_JMP.pdf

Münz, Rainer. 2008. “Migration, Labor Markets, and Integration of Migrants: An Overview for Europe.” SP Discussion Paper No. 0807. World Bank, April. https://documents1.worldbank.org/curated/en/495481468252864611/pdf/433370NWP08070Box327357B01PUBLIC1.pdf

Neuenkirch, Matthias, and Florian Neumeier. 2015. “The Impact of UN and US Economic Sanctions on GDP Growth.” European Journal of Political Economy 40, part A: 110–125. https://doi.org/10.1016/j.ejpoleco.2015.09.001

OPEC. Various Years. Monthly Oil Market Report. Vienna: Organization of the Petroleum Exporting Countries. Accessed March 28, 2025. https://www.opec.org/monthly-oil-market-report.html

R4V Platform. 2024. “Refugees and Migrants from Venezuela.” Panama City: Regional Interagency Coordination Platform for Refugees and Migrants of Venezuela. Accessed March 27, 2025. https://www.r4v.info/en/refugeeandmigrants.

Rodríguez, Francisco. 2023. “The Human Consequences of Economic Sanctions.” Journal of Economic Studies 51, no. 4: 942–963. https://doi.org/10.1108/JES-06-2023-0299

Rodríguez, Francisco. 2024. “Sanctions and Venezuelan Migration.” SSRN, December 26. http://dx.doi.org/10.2139/ssrn.5072710

Simon, Herbert A. 1954. “Spurious Correlation: A Causal Interpretation.” Journal of the American Statistical Association 49, no. 267: 467–479. https://doi.org/10.2307/2281124

Splinter, Melody, and Jeroen Klomp. 2021. “Do Sanctions Cause Economic Growth Collapses?” In Netherlands Annual Review of Military Studies 2021, edited by R. Beeres, R. Bertrand, J. Klomp, J. Timmermans, and J. Voetelink. The Hague: Asser Press. http://dx.doi.org/10.1007/978-94-6265-471-6_7

StataCorp. 2023. Stata 18 User’s Guide. College Station, TX: Stata Press.

Todaro, Michael. 1969. “A Model of Labor Migration and Urban Unemployment in Less Developed Countries.” American Economic Review 59: 138–148. https://www.jstor.org/stable/1811100

Turkewitz, Julie, and Isayen Herrera. 2023. “Why Are So Many Venezuelans Going to the United States?” The New York Times, September 24.

UN High Commissioner for Refugees (UNHCR). 2022. “Venezuela Situation: Global Report 2022.” UNHCR Global Focus. https://reporting.unhcr.org/venezuela-situation-global-report-2022

UNHCR. See UN High Commissioner for Refugees

US Customs and Border Protection (CBP). 2025. FY22 – FY25 (FYTD) Nationwide Encounters by Area of Responsibility – February. Accessed March 27, 2025. https://www.cbp.gov/sites/default/files/2025-03/nationwide-encounters-fy22-fy25-feb-aor.csv

US Energy Information Administration. 2023. “Oil and Petroleum Products Explained.” Washington, DC: US EIA. Accessed March 27, 2025. https://www.eia.gov/energyexplained/oil-and-petroleum-products/use-of-oil.php

US Energy Information Administration. 2025. “Europe Brent Spot Price FOB (Dollars per Barrel).” Washington, DC: US EIA. Accessed March 27, 2025. https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=RBRTE&f=M

Van Hear, Nicholas, Oliver Bakewell, and Katy Long. 2017. “Push-Pull Plus: Reconsidering the Drivers of Migration.” Journal of Ethnic and Migration Studies 44, no. 6: 927–944.https://doi.org/10.1080/1369183X.2017.1384135

Vermeulen, Maite. 2020. “How I Ended up in a Scientific Spat About Migration Figures and What I Learned From It.” The Correspondent, October 14. https://thecorrespondent.com/747/how-i-ended-up-in-a-scientific-spat-about-migration-figures-and-what-i-learned-from-it/98789433039-1dadd2ed

Westerlund, Olle. 2003. “Employment Opportunities, Wages and International Migration in Sweden 1970–1989.” Journal of Regional Science 37, no. 1: 55–73. https://doi.org/10.1111/0022-4146.00043

Acknowledgements

We thank Dan Beeton, María Eugenia Boza, Debarati Ghosh, Nicolás Idrobo, Dorothy Kronick, Alex Main and Mark Weisbrot for comments and suggestions and Luisa Elena García for excellent research assistance. Data and code for our results is available at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/7CXX8U. All errors are our responsibility.