Will Comprehensive Immigration Reform Pass in the Senate?

By Tom K. Wong, tomkwong@ucsd.edu, @twong002

Conclusion: 60 filibuster-proof votes for CIR in the Senate are within reach.

 

Will comprehensive immigration reform (CIR) pass in 2013? In a recent whirlwind of events since President Obama’s reelection, the Senate “gang of 8” introduced its draft principles on immigration reform, the President then addressed the nation describing his own vision of reform just one day after the gang of 8 unveiled their blueprint, the White House promptly released more details of the president’s proposal immediately after his speech, the House Judiciary Committee recently held its first hearing on the issue, and then the Senate Judiciary Committee did the same. The momentum that has been building towards CIR, which started well before last November’s election, has shown no signs of slowing down. However, momentum itself is no guarantee that reform will happen.

As a matter of politics, the key question is whether there are enough votes in Congress? More specifically, are there 60 filibuster-proof votes in the Senate and 218 votes in the House? While there are no crystal balls to tell us how legislators will ultimately vote, the recent history of immigration politics in the U.S. provides sufficient information to make informed predictions, not only about how current members of Congress are likely to vote on CIR, but also about what the final bill is likely to include (or not include).

Three key factors are worth noting.

  • First, while the current push for immigration reform represents its own distinct stage, many of the key actors are still the same. Nearly 200 representatives who are currently in the House were also in office during the contentious debate over H.R. 4437 in 2005. In the Senate, nearly 50 sitting Senators were also around during the 2006 immigration reform effort. This provides a rich pool of voting history to analyze and learn from.
  • Second, while we still do not know the details of the Senate or House bills, a) the degree of overlap between the gang of 8’s principles, the President’s proposal, and the 2006/2007 Senate CIR bills, coupled with b) the many lingering stand-alone immigration-related bills that have been introduced in Congress over the past few years, suggests that while the devil remains in the details, many of the details are already known. This is not to suggest that 2013’s bill will perfectly mirror previous reform efforts. Rather, it is to say that in analyzing the hundreds of amendments to CIR 2006 and 2007 and all of the immigration-related bills that have been introduced in Congress since 2005, we can create a map of the road ahead.
  • Third, the November presidential election showed that immigrant communities – and communities of color more generally – are increasingly flexing their political muscle. This adds confidence to our ability to use the demographic makeup of states and districts to inform how we think legislators are likely to vote on CIR.

Senate Prediction

60 filibuster-proof votes for CIR in the Senate are within reach. In fact, the data suggest that there are currently 52 solid yes votes with 19 additional Senators that lean towards voting yes. The 52 solid yes votes are spread across 31 states. They include the Senators who we would expect to support CIR (e.g., Democrat Senators in states like California and New York). This group also includes 8 Republicans. The 19 Senators that lean yes are spread across 15 states and include 10 Democrats and 9 Republicans. A full list of Senators broken down into their predicted categories of support and opposition can be found here.

71 yes votes is an ambitious prediction, and represents a high range. It is only reasonable if we a) expect Democrats who have supported CIR in the past to vote for CIR in 2013, b) expect Democrats without a robust voting record to analyze, but who represent diverse states, to also vote for CIR, c) expect Republicans who supported CIR in the past to vote for CIR in 2013, and d) expect Republicans who may not have supported CIR in the past, but who represent increasingly diverse states, to vote for CIR. A more moderate prediction would remove the last assumption, leaving us with 67 votes. If there is a strong sense that CIR will, indeed, pass, there is likely to be a bandwagon effect during the final passage vote that moves us closer to the high-range estimate. However, if the mood around CIR changes, wherein there is less confidence that a bill will pass, the actual vote total is likely to be less than the moderate prediction.

The underlying assumption of the statistical models used to estimate likely support and opposition is that Senators who represent diverse states, particularly those with large foreign-born and Hispanic/Latino and Asian populations, are more likely to support a CIR bill that includes a path to citizenship than Senators in mostly white states – while considering other factors. One way to interpret these results is thus: to the extent that Senators are responsive to the changing demographics of their states, we can expect them to vote as the model predicts. However, it is not yet clear whether changing demographics – even given the November 2012 elections – will parlay into more legislative support for CIR.

Notes

Step 1 of this analysis models the cloture and final passage votes of the Comprehensive Immigration Reform Act of 2006. Here, I estimate two models. Model 1 focuses on the size of the foreign-born population while also accounting for other important factors (e.g., the party affiliation of Senators, southern state, etc.). Model 2 focuses on the size of different racial and ethnic groups. I construct categories of likelihood, wherein each predicted probability is characterized as a “solid yes” vote, “leans yes,” “leans no,” and “solid no” vote. These categories take into account the confidence intervals around each predicted probability. Figure 2 provides a graphical illustration.

Figure 2. Predicted probabilities and 95% confidence interval for all 100 Senators

Figure 2

Step 2 critically evaluates the predicted probabilities obtained in step 1 by analyzing the actual voting record on immigration-related bills for each Senator (where a voting record exists). For example, if a Senator voted for CIR in 2006 and for the DREAM Act in 2010, I expect the Senator to be categorized as “solid yes.” Conversely, if a Senator voted against CIR in 2006 and against the DREAM Act in 2010, I expect the Senator to be categorized as “solid no.” In all, 38 Senators fell along these lines and all were correctly classified except for 4, which represents an approximately 90% match rate.

Step 3 expands the analysis in step 1 to the full history of voting on immigration-related bills in the Senate from CIR 2006 to present as an additional check. The determinants of past votes on immigration policy are analyzed using a logistic regression model: Pr (Voteit = 1|Xit) = P (β0 + β1Xit + εit) where Xit represents a vector of explanatory variables, including key demographics characteristics, and εit represents the error term. In the Senate, there are nearly 6,000 observations to analyze. Predicted probabilities are then obtained from m = 1,000 simulations for each current member of the Senate using the estimated models.

Step 4 synthesizes all of the data – actual voting records and predicted probabilities – and assigns each Senator to one of the four categories of likelihood described above.