Texas Democrats are giddy over the possibility that demographic changes might turn the Lone Star State “blue,” but the numbers suggest Texas will lean “red” for a long time.
In the last issue of The New Republic, a big graphic estimated that the combination of demographic changes, immigration reform, and higher Hispanic turnout wouldn’t turn Texas blue until the early-2030s. To win sooner, Democrats need big gains among the state’s white voters. You can see those estimates here.
This post documents how I arrived at that conclusion.
First, I projected the composition of the state’s voting eligible population. That starts with population projections from the Texas State Data Center, which provides the state with population projections based on data from the last Census. The projections assume that migration continues at the frenetic pace of the last decade.
Over the next few decades, the TSDC forecasts that the white share of the Texas population will decline steadily. Non-Hispanic whites will be just 29 percent of the population by 2036. Hispanics will be a majority. With numbers like that, it's easy to see why Democrats are so optimistic.
But these projections are for the entire Texas population—not the voting eligible population of U.S. citizens over 18 years old. Unfortunately for Texas Democrats, a disproportionate share of Hispanics are under age 18 or non-citizens. As a result, the voting eligible population—U.S. citizens over age 18—is much whiter than the state’s population. According to the American Community Survey, only 26 percent of eligible voters were Hispanic in 2010.
Projecting the evolution of the voting eligible population starts by taking the TSDC’s population projections and removing people under age 18. Then, we need to subtract non-citizens. That’s more difficult. The biggest factor is young Hispanics: Most were born in the United States, even if their parents are legal permanent residents, or even undocumented. As a result, 92 percent of young Hispanics are citizens compared to 67 percent of Hispanics over age 18, according to the American Community Survey.
As young Hispanics reach voting age and older Hispanics depart, the share of voting age Hispanics with U.S. citizenship should increase steadily. This projection estimates it will rise to about 82 percent by 2036. We can be modestly confident in this estimate, since almost all of 2036’s voting age Hispanics are already alive. However, it doesn’t account for two counter-veiling factors: the number of Hispanic legal permanent residents who become U.S. citizens, and the number of new Hispanic immigrants without citizenship. To date, it seems that these issues are canceling out, since the Hispanic citizenship rate has been relatively constant, but a better estimate would account for these two factors with more precision. Nonetheless, we can say with great confidence that the Hispanic share of eligible voters will remain beneath their share of the population for the foreseeable future.
Will these demographic changes be enough to put Democrats over the top? If so, when? That depends on turnout and how well Democrats can perform with given demographic groups. Of course, it’s not possible to predict the future with any precision. Fortunately, that’s not the point—although a few commentators have accused me of holding that objective. The point is to test whether demographic changes could put Democrats over the top if they keep dividing the electorate as they did in 2012.
Unfortunately, we don’t know the full story of how Texas voted in 2012, since there wasn’t an exit poll. What do we have? The actual results, the 2008 exit polls, and the November 2012 Supplement to the Census Current Population Survey (CPS), which provides solid turnout estimates. We can use these data to make some rough guesstimates of how Obama performed last November.
The CPS is the only available turnout estimate for 2012, so we’ll take it as a given. That’s fine by me, since the CPS is probably better at gauging turnout than the exit polls. But the CPS shows that the Texas electorate was extremely diverse in 2012: just 59 percent of the electorate was white. That’s pretty astonishing, since Romney won the state by 16 points. That means that Romney must have done extraordinarily well among white voters.
How well did Romney do among whites? Nationally, Obama did even better among non-white voters than he did in 2008, but lost considerable ground among whites. If that was true in Texas, then we could plug the 2008 exit poll results into the CPS turnout estimate, and deduce a relatively high estimate of Obama’s share of the white vote.
According to the 2008 exit polls in Texas, Obama received 98 percent of the black vote and 63 percent of Hispanics. We’ll plug-in 58 percent for “other,” the same amount Obama received nationally in 2012. If we assume Obama repeated those performances in 2012, then Obama only won 18.9 percent of white voters in Texas.
For the purposes of the projection, I plugged in that Obama did worse among blacks (93 percent) but better among Hispanics (65 percent) than 2008, since that seems reflected in the national exit polls and the county-level data. That hastens Democratic gains in Texas, since Hispanics are the growing segment of the population, but doesn’t move the estimate for white voters.
Did Obama really do so bad among white voters in Texas? Even if he didn’t, it was probably pretty close. Let’s start with 2008. Even in those extremely favorable circumstances, the exit polls found that Obama only won 26 percent of the Texas white vote. From that perspective, 18 or 19 percent isn’t preposterous, although it remains a pretty significant drop.
One way to confirm our assumptions is to look at the actual results. If Obama suffered huge losses among white voters, it should be evident in white counties and legislative districts. The county-level results confirm that Obama suffered serious losses among white voters. The first map shows the change in Obama’s support between 2008 and 2012. The counties in red are the counties where Obama did better than he did in 2008. The counties in blue are where the president did worse.
The second map shows Texas by race, where green is plurality or majority non-Hispanic white and yellow is Hispanic.
As you can see, the president suffered big losses in the whitest areas—the president did at least a net-10 points worse than 2008 in the darker blue, darker green counties strung across the middle of the state. On the other hand, Obama held strong or even made gains in heavily Hispanic areas, like along the Rio Grande.
Similarly, we can consider the whitest counties in Texas, where whites represent more than 80 percent of the population. In 2008, Obama won 22.8 percent of the vote in these counties; in 2012, he won 17.8 percent. That 5 point decline is broadly consistent with the earlier, 7 point estimate based on the exit polls and CPS.
I have to emphasize that this isn’t an authoritative estimate. It’s a guestimate, necessary to make up for the absence of exit poll data. Mixing and matching the CPS and the exit polls is not necessarily advisable, as I’ve written about before. In this instance, however, I think it’s the best we can do, even if it’s imprecise. For what it’s worth, I suspect the guesstimate slightly underestimates Obama’s support among white voters, since the CPS probably overestimated black turnout. If you put a gun to my head, I’d probably say Obama received 20 percent of the white vote.
But whether Obama got 16 or 18 or 20 or 22 percent of the white vote doesn’t significantly influence the trajectory of the projection. Paradoxically, a stronger Obama performance among white voters actually slows the pace that demographic change helps Democrats. Why? If Obama received a larger share of the white voters, then he must have done worse among non-white voters, either due to lower turnout or support. If true, then Texas Democrats have even more turnout work to do, or the new, non-white, Democratic-leaning voters might be a little less Democratic than this projection assumes.
The final piece of the puzzle is turnout. If the CPS is right, and it is for the purposes of these projections, then Obama actually had a pretty good turnout in 2012. The CPS shows that 63.1 percent of blacks turned out, compared to 60.9 percent of non-Hispanic whites—the latter of which is far lower than it was in 2004. Since that’s probably not sustainable, the “status quo” projection assumes an average of 2004 and 2012 turnout. For that reason, the “status quo” projection doesn’t show the Democrats making many improvements between 2012 and 2016. In Texas last November, the Democratic turnout advantage compared to the 2004-2012 average was roughly equal to four years of demographic change. Not too bad.
On the other hand, the “improved Hispanic turnout” projection is a true best case scenario. It takes Nevada’s 2012 turnout across the board, which basically means historic black turnout, low black turnout, and a 26 percent increase in Hispanic turnout compared to Texas in 2012. Realistically, Democrats can’t pull that off, since the state’s Hispanic population is less urban. But I can’t imagine too many Democrats complaining that I didn’t give them a fair shot: this is a scenario where Democrats turnout one million new Hispanic voters by 2016, preserve Obama’s historic black turnout, and avoid resurgent Republican turnout among whites.
Those huge efforts would make a difference, combining with demographic change to reduce the white share of the electorate from 59 percent in 2012 to 53.6 percent in 2016. That cuts the GOP’s margin down to 12.9 points—about as well as Obama did in 2008. But without more support from Texas whites, Democrats aren’t getting over the top for a long time.
It shouldn’t be too surprising that Democrats will need gains among white voters to win Texas. The Democratic rebound in Virginia and North Carolina was greatly assisted by demographic change and higher black turnout, but it also required an influx of northern moderates and liberals to northern Virginia or the Research Triangle. According to the exits, Obama won 38 percent of the white vote in Virginia last November, compared to Kerry’s 32 percent. In North Carolina, the exits say Obama won 31 percent of the white vote, up from Kerry’s 27 percent. Without those gains among whites, Obama narrowly loses Virginia and North Carolina isn’t exactly competitive.
So far, there aren’t signs of a pro-Democratic trend among white voters in Texas. In fact, Obama’s performance among Texas whites last November was historically bad: Probably the worst in the history of the party. It’s not hard to see why Democrats do so poorly among Texas whites. Half are evangelical Christians and the rest are pretty conservative, too. The cultural issues aiding Democrats in Denver and Fairfax aren’t helping in Dallas. The state is far more dependent on carbon and natural resources than the post-industrial, creative-class metropolitan areas on the coasts.
While one might expect migration to help Democrats in the suburbs, it hasn’t happened yet. It’s pretty clear that the whites moving to Dallas and Houston aren’t the secular Democratic-leaners from the Northeast piling into Virginia and North Carolina, but deeply conservative and Evangelical Republicans from the Midwest and South. In Montgomery County, Houston’s whitest, most affluent suburb, Obama won just 19 percent of the vote—the worst Democratic performance in the history of the county. White precincts in Harris County, home to Houston, basically voted unanimously for Romney. Democrats have made gains around Dallas, but it’s hard to distinguish those gains from the influx of non-white voters. The congressional district results suggest that Republicans are still doing extremely well. Young voters won’t bail out Democrats either—Obama only won 30 percent of young Texans in 2008. He probably did significantly worse in 2012, although we don’t know for sure. And sorry folks, but Austin is really just a small blue island in a red sea.
The best argument for a Democratic rebound among white voters is the possibility that another Democrat, particularly a white one, might have more appeal among Texas whites. But a white Democrat wouldn’t do as well among blacks, and another Republican could easily do better among Hispanics. In a state like West Virginia, that’s a good trade for a white Democrat. But in a state with a huge minority population like Texas, it’s not clear that a white Democrat would do much better. And in the 2012 Senate race, the Democratic nominee lost by the same 16 point margin as the president. Yes, Paul Sadler did slightly better among whites, but it was canceled out by doing slightly worse among non-white voters. And in a state where half of whites are evangelicals, there’s only so much room for Democrats to improve—at least if they keep nominating progressives.
The point isn’t that Democrats can’t do better among Texas whites. Maybe the next wave of young Texans will get more Democratic. Maybe the next wave of migrants will be more Democratic. Maybe Democrats will nominate a relatively conservative southerner. It's all possible. The point is that they must if they intend to win any time soon. The growing Hispanic share of the population won't be enough. And so far, there aren’t any signs of Democrats making big inroads among Texas whites. It might come some day, but it hasn't; and there are plenty of reasons to question whether it will.