Editorial note: We have removed the sub-headers from this paper that detailed the mathematical models used to analyze the claims made by Bahar and Hausmann. It is worth noting that Hausmann served as Planning Minister during the neoliberal second term of President Carlos Andrés Pérez and was later designated by Juan Guaidó as representative to the Inter-American Development Bank.
By Francisco R. Rodríguez, Giancarlo Bravo – Dec 18, 2025
Introduction
In a working paper and accompanying op-ed published in March of this year, Dany Bahar and Ricardo Hausmann claimed to find evidence of a positive relationship between Venezuelan oil income and migration flows to the United States. Rodríguez, Rosnick, and Bravo showed that these results were driven by a coding error: The authors had incorrectly used a twelfth-difference operator instead of year-on-year seasonal differencing. After correcting this error, the statistical significance of their results disappeared. Further, the observed correlation between Venezuelan migration flows and oil revenues was not robust to the inclusion of controls for labor market tightness in the United States.
Bahar and Hausmann 3 acknowledged that their results lost statistical significance after correcting the coding error yet argued that their positive point estimates still provided evidence against the hypothesis that sanctions — which reduce Venezuelan oil revenues — were a driver of Venezuelan emigration flows to the United States.
In this paper, we address a separate problem with the Bahar-Hausmann approach: their incorrect application of cointegration methods.
Bahar and Hausmann’s results come from two empirical specifications that are premised on the existence of a cointegrating relationship between migration and oil revenues. Citing results from an Engle-Granger cointegration test, they claim to find evidence of such a relationship. Yet those results come from misapplying a misspecified, nonstandard version of the test and disappear once the error is corrected.
The cointegration results reported by Bahar and Hausmann derive from applying an Engle-Granger test to first differences of their variables of interest. However, the Engle-Granger test is designed to be applied to the levels of variables that are integrated of order one, not their first differences. By first differencing the series before applying the Engle-Granger test, Bahar and Hausmann are incorrectly testing for cointegration between two stationary series. Because the test evaluates the stationarity of the residuals from a regression involving potentially cointegrated series, applying it to stationary first differences will virtually guarantee a spurious
rejection of the null of no cointegration.
As we show below, once we correct this misspecification, the evidence of cointegration in the Bahar and Hausmann data disappears. This implies that their empirical framework is inadequate and uninformative with respect to the existence of a long-run relationship between oil revenues and migration flows of Venezuelan nationals to the United States.
Conclusions
This research paper has argued that the claim made by Bahar and Hausmann— that sanctions on Venezuela’s oil industry are not a driver of migration flows to the United States—finds no support in their own data. We have shown that Bahar and Hausmann’s results were based on their use of a nonstandard and misspecified test to evaluate the existence of a long-run cointegrating relationship between Venezuelan migration to the United States and Venezuelan oil revenues. Once their mistake is corrected, evidence of cointegration disappears, rendering their empirical approach invalid.
We have established that when the Bahar-Hausmann test is applied to independent weakly stationary series, the probability that it will find evidence of cointegration 100 percent of the time, even when such a relationship does not exist, approaches one as the sample size increases. We also provide Monte Carlo simulations on time series of the length used by Bahar and Hausmann that yield a false positive rate of 100 percent for two independent non-cointegrated unit root processes. These results imply that the Bahar-Hausmann approach is guaranteed to find evidence of cointegration even when no such relationship exists. We also showed that when one runs
cointegration tests on the levels of logarithms—the only formulation consistent with the empirical specifications adopted in their time series regressions—there is no evidence of cointegration in these tests. This renders their empirical approach, including both the level regressions and ARDL specifications, as inherently misspecified and therefore uninformative about the true nature of the relationship.
These results are not surprising. As discussed in greater detail in Rodríguez, Rosnick, and Bravo, there are reasons to be skeptical about the use of data on US border encounters of Venezuelan nationals to assess hypotheses about the effects of sanctions on migration flows. A large share of Venezuelans who attempt to enter the United States today have spent several years living outside of Venezuela, often in countries like Colombia, Peru, or Mexico. It is unclear why their decisions to migrate to the United States would be affected by fluctuations in the resources under the control of Nicolás Maduro’s government. As we also show there, any residual
correlation between migration and oil prices is explained by the fact that oil prices were strongly correlated with economic conditions in the United States in a period that includes the COVID pandemic. Strikingly, Bahar argues that US labor market conditions are a key determinant of border crossings, yet Bahar and Hausmann make no effort to control for this effect.
Moreover, directly testing for the impact of oil revenues on migration flows fails to capture the broader economic dynamics likely to influence migration decisions. Since 2012, Venezuela has suffered the largest documented economic collapse to occur outside of war, and the dramatic increase in emigration during this period suggests that deteriorating economic conditions have been a central factor. In this context, it appears more reasonable to first assess the impact of economic sanctions on Venezuelan economic performance and to separately assess how deteriorating living standards have affected migration decisions.
Bahar and Hausmann have proposed an interesting hypothesis: that expectations of regime change may themselves be a major driver of migration. This idea is not necessarily inconsistent with the view that deteriorating incomes—partly due to sanctions—have contributed to rising emigration. The possibility that migration decisions are shaped by potential migrants’ expectations about future economic and political conditions appears inherently plausible.
More empirical work is certainly needed to better understand the extent to which economic and noneconomic factors have contributed to Venezuelans’ emigration decisions and to clarify how past and current conditions—as well as expectations of change—shape these choices.
While we believe that Bahar and Hausmann’s empirical work has failed to yield informative answers to these questions, we consider the questions they have raised to be highly relevant and deserving of further empirical investigation.
Francisco Rodríguez is a senior research fellow at the Center for Economic Policy and Research and a faculty affiliate at the University of Denver’s Josef Korbel School of International Studies.
Giancarlo Bravo is the head of research of Oil for Venezuela.
(CEPR)
From Orinoco Tribune – News and opinion pieces about Venezuela and beyond via This RSS Feed.