Days before the polls closed on November 6, 2012, the Obama team was predicting publicly that the electorate would be less white than in 2008, and they were right. Their argument went like this:
"Since the campaign began, Team Obama has assumed that whites would make up 72% of voters, down from 74% in 2008. (The logic: This percentage has decreased in every U.S. election, and the minority population continues to grow.) On the other hand, Republicans counter that due to greater GOP enthusiasm, the white percentage could very well stay the same as in 2008 -- or even be a tick higher. And according to our NBC/WSJ pollsters, the final outcome here could sway the election. If you assume that both sides maximize their margins (Romney wins whites by 20 percentage points and Obama wins minority voters by 60 percentage points), whites making up 75% of the electorate would give Romney the edge with the popular vote by a fraction of a point. But if the white percentage is 74%, Obama would win the popular vote by a fraction of a point. And if it’s 73%, Obama wins by one point. Bottom line: Watch the white percentage in tomorrow night’s exit polls; it will tell you more about where Virginia or Iowa or Colorado or Wisconsin will go."
Seems awfully prophetic in retrospect.
Now, of course it is far too early to speculate what the 2016 electorate might look like, but let's do it anyway by looking at past trends. As the chart below notes, the demographic breakdown of the 2012 electorate did seem highly predictive in hindsight.
The white vote, as a proportion of the entire electorate, has decreased in every election since 1992, when it topped out at 87%. The same is true for the Latino vote, dipping as low as 2% of the electorate in 1992, until reaching 10% in 2012. African American voters have not seen a decline in their numbers since 1992. Another way to look at these stats is to consider that the white portion of the electorate has fallen from 88% in 1980 to 72% in 2012, while the non-white share of the vote has grown from 12% in 1980 to 28% in 2012.
If one were going to guestimate what the 2016 electorate may look like, it would be important to consider that the share of the white vote has not fallen MORE than 4% from one election to the next since 1992, and has not fallen LESS than 2% per election year since 1992. Based on that trend, it wouldn't be too surprising to see the share of the white vote drop to anywhere from 68-70% in 2016.
Just as reliable as a drop in the share of the white vote has been an increase in the proportion of the Latino vote. Latino's have increased their vote share anywhere from 1-3% in each election year since 1992. Therefore, one could easily expect their vote share to be anywhere from 11-13% in 2016.
The African American vote had remained relatively stable from 1980-2004, ranging from a low of 8% in 1992, to a high of 11% in 2004. But the nomination of Barack Obama brought them out in record numbers, and they comprised 13% of the electorate in both 2008 and 2012. But the steady downward trend in the white vote, and upward trend in the Latino vote, is less discernible with the African American vote. Blacks made up 10% of the electorate in every presidential election from 1980-1988, before suddenly dropping to 8% in 1992. Their numbers recovered in 1996 to 10%, remained there in 2000, ticked slightly upward in 2004, then jumped up in 2008. The question remains (and will until election day 2016): can anyone other than Obama, or at least a non-African American Democrat, keep African American turnout at 13% or higher? Also, could a non-black Democrat win as high a percentage of the black vote as Obama (95% and 93% of the vote in 2008 and 2012 respectively)?
While the trend for the number of voting Asians is on the rise, it has only grown from 1% to 3% of the total electorate since 1992. In other words, the trend has shown that it takes a few presidential cycles for the Asian vote to gain even one percentage point.
Those identifying as "other" represent the only non-white voting subgroup that saw a drop in their share of the electorate from 2008 (from 3% to 2%).
Given the above stats, what's so hard to envision about a 70% white electorate in 2016, with a 30% non-white electorate (13% black, 12% Latino, 3% Asian, 2% other)? Such a make-up represents a 2% drop in the share of the white vote from 2012, and a 2% increase in the share of the Latino vote (African Americans, Asians, and "other" hold steady at their 2012 levels). For an example of how complicated that can make things for the Republicans, here's what the 2012 results would have looked like had the electorate looked more 2016-like (extrapolated from 2012 exit polling data):
Romney would have lost to Obama 51.3% to 46.9% in the hypothetical 2016 electorate explained above (as opposed to the actual 50.6 - 47.8%). Now, of course, this result makes a lot of assumptions that are intuitively unlikely. For example, one could argue the African American vote will make up a smaller portion of the electorate in the absence of an African American candidate. They could also argue that the Republican share of the African American vote may return to pre-Obama levels (which before 2008 had ranged from 8-12% since 1980; it was 4% in 2008, and 6% in 2012). It also assumes that Republicans will perform near their historical low with Latinos (21% in 1996), as opposed to their historical high (44% in 2004), while assuming they will again perform near their historical high with whites (which was 66% in 1984, with the low being 41% in 1992).
Assumptions aside, the hypothetical above shows you how difficult it could be for the GOP to win future national elections without doing 1 of two things: 1) returning to Reagan/Bush '84-era performance with whites (66%), OR 2) returning to Bush/Cheney '04-era performance with Latinos (44%). Let's take a look at the numbers in both scenarios, holding all other race/ethnic groups constant with our hypothetical 2016 electorate.
Possible 2016 result weighted to hypothetical 2016 electorate AND 1984 Dem/Rep vote share amongst "white" voters:
Republicans would pull out a 4 point win (51.8 - 47.8%) if they were somehow able to replicate Reagan/Bush '84's performance amongst whites, assuming our hypothetical electorate actually resembles 2016 turnout.
Possible 2016 result weighted to hypothetical 2016 electorate AND 2004 Dem/Rep vote share amongst "Latino" voters:
Republicans would draw even with Democrats in 2016 if they were somehow able to replicate Bush/Cheney '04's performance with Latinos, assuming our hypothetical electorate actually resembles 2016 turnout.
Obviously, the more attractive option for Republicans would be if they could somehow manage to capture a similar portion of the white vote they received in 1984. But history makes that incredibly unlikely. Ronald Reagan's 66-34% defeat of Walter Mondale with whites in 1984 accompanied a landslide election and was not repeated before (Reagan beat Carter 56-36% among whites in 1980), nor after (HW Bush defeated Dukakis 59-40% among whites in 1988). In fact, Romney received the second highest percentage amongst white voters (59%) in any election since 1980.
Competing for a greater share of the Hispanic vote makes sense for a couple of reasons. For starters, the Hispanic share of the electorate is growing, while the white share is shrinking. Why should the GOP fight to make further inroads in a shrinking subgroup of the electorate they already have locked down? Second, until recently (2008), the GOP has been somewhat competitive with Latino voters, receiving over 1/3 of their vote in 2004, 2000, 1984, and 1980. This would indicate that popular Republicans can do well with Hispanics (see Reagan and GW Bush), while popular Democrats DOMINATE with Hispanics (see Clinton and Obama).
Whether they like it or not, the GOP may discover soon that they can either get to 40% or better with Latinos on a regular basis, or George W. Bush may remain the last Republican President for some time to come.