The Political Economy of Veto Overrides. Evidence from Justice Against Sponsors of Terrorism Act (JASTA)

1 Introduction

Presidential vetoes have significantly reduced over the years with a 69 % decline since President Clinton’s administration. Veto overrides have always been low in number with the highest number of veto overrides being twelve in President Ford’s administration (McKay 1989). The past two administrations have had only two veto overrides in total and there has been none in the current administration. President Obama issued a total of twelve vetoes and the Justice Against Sponsors of Terrorism Act (JASTA) was overridden by Congress and passed to law (Daugirdas and Mortenson 2017). The Senate voted 97-1 and the House voted 348-77 to override the veto. Due to the attainment of two-thirds majority rule, JASTA was passed into law in 2016.

JASTA modified the Foreign Sovereign Immunities Act (FSIA) of 1976 by allowing U.S. citizens to sue foreign governments for terrorist attacks on US soil (Watkins 2018). The bill was largely supported by families of 9/11 victims who sought the ability to take legal action against foreign entities suspected of financing terrorism. President Obama vetoed JASTA on September 23, 2016, arguing that it could weaken sovereign immunity, strain US relationships with foreign allies, and expose US officials to lawsuits abroad (Obama 2016). Despite these concerns, Congress overwhelmingly voted to override the veto, demonstrating a rare moment of bipartisan consensus.

Congress’s decision to override the veto was significant, as veto overrides have become increasingly rare. Rising partisan divisions and the increasing concentration of power in the executive branch have made bipartisan cooperation less common, making overrides more difficult. The fact that veto overrides are becoming less frequent underscores the importance of studying cases like JASTA, where Congress exercised its authority despite executive opposition. JASTA’s passage showed that constituent demands could sometimes outweigh executive authority, especially in cases with strong public support. Lawmakers were under pressure from 9/11 victims’ families, making it difficult to justify upholding the veto. The overwhelming support for JASTA suggests that when public opinion is strong, Congress is more likely to challenge the president’s decision.

This paper seeks to unravel the dynamics behind legislator decision-making when it comes to veto overrides. Indeed, there are many factors that could potentially influence the decision behind a veto override. A combination of party affiliation, constituent concerns, and broader geopolitical considerations shape legislative decisions on matters of national interest (Trubowitz 1998). The 2016 Justice Against Sponsors of Terrorism Act (JASTA), in which Congress overrode President Obama’s veto, provides a valuable case to examine these dynamics.

This event raises important questions: Does party affiliation significantly impact how legislators vote on national security issues, and did representatives from states most affected by the 9/11 attacks, such as New York, vote in line with their constituents’ interests? The JASTA vote demonstrated not only the role of partisan politics, but also the influence of regional concerns, where lawmakers from states that suffered the most from 9/11 may have prioritized victim reparations over party lines.

JASTA also raises questions about the balance of power in foreign policy. The executive branch usually handles diplomatic relations, but Congress demonstrated that it can act independently when domestic concerns are strong. The law also brought attention to the issue of sovereign immunity and whether other nations might pass similar laws targeting the US. Some critics argue that JASTA could lead to legal actions against US military personnel and diplomats in foreign courts. However, supporters believe that the law is necessary to provide justice for victims of terrorism and to hold foreign governments accountable.

JASTA’s passage highlights the way national security, public opinion, and legislative power interact. The rare veto override shows that when public pressure is strong and bipartisan support exists, Congress is willing to push back against the executive branch. This analysis sheds light on how political identity and local interests shape legislative behavior in high-stakes decisions.

This case provides insight into how domestic political concerns influence legislative decisions, especially when they involve foreign policy and national security.

2 Empirical Approach and Data

3 Empirical Results

Table 2 shows the results of the probit regression and Table 3 shows the corresponding average marginal effects (AME) that provides information on the various factors that influence House representatives to vote ‘Yea’ on the bill. Table 2 with the probit model shows the general direction and statistical significance of the variables in the model while the AME results quantify the probability of a ‘Yea’ vote.

The variable Party affiliation is significant across all models, which means that being a Democrat significantly reduces the chances of a legislator voting ‘Yea’ to the veto override. This is the main relation in the model and is captured by Figure 1. Figure 1 plots the party affiliation against the probability of voting Yea. As in the table, being a Democrat reduces the chances of voting Yea in the veto override. The AME results indicate that Democrats are about 21–23 percentage points less likely to vote in support of the veto override. This result is expected since President Obama was a Democrat, therefore, legislators who were Democrats were less likely to vote in support of overriding the President’s veto decision.

Figure 1:

Probability of voting ‘Yea’ by party affiliation.

Legislators from states that were adversely affected by the 9/11 crash were significantly more likely to vote in support of the veto override by 9–12 percentage points. This suggests that house representatives were most likely voting in consideration of their respective constituent’s interest. The results for the variable for the age variable indicate that the older the legislator, the less likely they are to vote in support of the veto override but the effects weaken with the addition of controls. The squared term for age does not significantly affect voting behavior as shown in model 3.

Whether a legislator is up for re-election is also significant in the probit model and in the first AME model indicating that legislators facing re-election are more likely to vote in support of the veto override. This suggests that electoral incentives play a crucial role in influencing voting behavior.

Other legislators’ characteristics like race, gender and military background are found to be insignificant, suggesting that these characteristics do not systematically influence the legislator’s decision to vote against or to support the veto override.

For median voter characteristics, the percentage of constituents who have a bachelor’s degree is consistently associated with a lower probability of voting in support of the veto override. The AME results indicate that an increase in the proportion of constituents who have a bachelor’s degree decreases the probability of voting ‘Yea’ by 1 percentage point. This highlights the importance of constituent education attainment in legislator voting decisions.

The variable for median income is included, then we also add a squared term to test the non-linearity of median income. In model 3 of Table 2, both median income and its squared term do not have a statistically significant relationship to voting in support of the veto override. Both linear and quadratic terms are insignificant in Model 3, suggesting that there is no significant curvilinear relationship between median income and the probability of voting ‘Yea’. The percentage of immigrants, the percentage of veterans and the percentage of males are not statistically significant, indicating that these variables do not systematically affect the voting behavior of legislators.

Table 4 shows the probit results from the full model with the addition of interaction terms. Interaction terms are added to test if the effect of party affiliation is conditioned by legislator and district level characteristics. In model 2 and 3, the variable for party affiliation maintains significance and direction. However, in model 1 and 4 of Table 4, party affiliation loses significance, therefore, suggesting that the inclusion of interaction terms alters its effect. The variable adverse state is consistent with previous models, indicating a positive and statistically significant relationship. Race, gender, higher education and military are still insignificant in the models.

Age is consistently negative in the model showing that the probability of voting in support of the veto override increases with age, however, the result is insignificant. Similarly, being up for re-election is positively associated with voting ‘Yea’ due to potential electoral incentives. This holds throughout the models but loses significance when the variable for party is interacted with higher education. The coefficient on median income and its squared term are not significant, suggesting a lack of effect of median income on the voting decision. The percentage of constituents with a bachelor’s degree also remains significant in the interaction models. The interaction terms are not significant across the whole model, meaning that party affiliation does not substantially change based on military, veteran or education characteristics.

4 Conclusions

The findings show that party affiliation is the strongest predictor of House representatives’ decisions on the veto override. Democrats were significantly less likely to support the override, which is consistent with partisan alignment, since the veto came from a Democratic president. This reinforces the idea that party loyalty remains a dominant force in legislative behavior.

Beyond partisanship, electoral incentives and constituent interests also played a role. Representatives from states most affected by the 9/11 crash were more likely to support the override, suggesting responsiveness to district-level economic and social concerns. Legislators who are running for reelection were also more likely to vote in favor, highlighting the role of electoral pressures (Mayhew 2004). These results align with theories of political representation, particularly the delegate model, which suggests that legislators act according to their constituents’ preferences, while the trustee model assumes that they rely on their own judgment (Fenichel Pitkin 1967). The significant effects of district-level characteristics, such as economic hardship and education, suggest that legislators respond to constituent demands, supporting the delegate model in this context.

The results also provide evidence for political ambition theory (Schlesinger 1966), which argues that legislators make strategic choices based on their career goals. The positive relationship between reelection status and voting behavior suggests that legislators prioritize electoral considerations when making policy decisions. Furthermore, the association between constituent education levels and voting outcomes aligns with economic voting theory (Johnston 2009), which suggests that voter characteristics influence legislative decisions.

Overall, the findings demonstrate that while party affiliation is the most significant determinant of voting behavior, electoral incentives and constituency characteristics also shape legislative decisions. These results contribute to a broader understanding of congressional voting patterns and suggest that party loyalty, strategic electoral considerations, and district-level factors interact in complex ways. Future studies could examine how these dynamics evolve across different policy areas, particularly as partisan polarization continues to influence congressional decision making.