Where Corn Don't Grow

Much has been written already about whether 2008 is a re-aligning election. I suppose that this is a small fraction of what will be written in the coming months. I laid my thoughts out here, here, here, here and here about whether 2008 would be re-aligning, and my answer is still a cautious “no.” I’ll have more to say later on the subject, but without getting too “Karl Popper” on you, I find most of the arguments for emerging re-alignments to be the worst kind of historicist junk. Just as an example, while it is true that the youth vote voted overwhelmingly Democratic, this forgets that eight years ago it was split. There’s nothing to say that four years from now it won’t be split again, especially if Obama has a rough time of things (indeed, a large part of why that generation is heavily Democratic for now is that twenty-somethings’ political memory consists of a competent Democratic President and an incompetent Republican President).

There are certainly arguments to the contrary (for one thing, the left dominates many of the information avenues to these voters, which will probably take more than one or two cycles to remedy), but we have no way of sorting them out with anything other than mere guesswork. In short, I find these arguments about as convincing as I found arguments in 2004 that we were encountering a permanent Republican majority (“We won 97 of the 100 fastest-growing counties!!!!”). A re-alignment may be brewing, but we really won’t be able to know until we are well into it.

For now, I’d like to take a close look at how Obama built his map. We note at the beginning his solid, but not overwhelming, majority in the electoral college, which presently looks to be about 364 to 174. Again, this is a solid win, but it is not LBJ 1964. This is fairly remarkable, considering that you have to go back to 1952 or possibly even 1920 to find an open race where the climate was similarly inhospitable to the incumbent party. Both of those resulted in double-digit popular vote wins for the out-party running into the 400s in the Electoral College.

Michael Barone links to a map of how counties voted, courtesy of the Washington Post here. There’s also a cool 3-D map you can click on, though it takes an awfully long time to load. But this isn’t particularly illuminating, as all it shows is the binary choice of Obama/McCain. But the difference between a county or state going 50-49 Obama and 50-49 McCain isn’t as interesting to me as the difference between a county or state going 10-90 McCain as opposed to 10-90 Obama. We need maps that show gradations of red, purple, and blue. A map like this is more useful.

But rather than take the absolute results, what I would like to do is look at the Partisan Voting Index (with a doff of the hat to Charlie Cook for the term) for the states. What this does is take the states and compare them against the candidates’ national average. In other words, if a candidate wins nationally with 55 percent of the two-party vote, and wins a state with 55 percent of the two-party vote, the PVI is zero. This basically allows us to look at what states fall on the right and the left side of the political spectrum as it manifests in a particular election. More importantly, by taking the states and reducing them to the same baseline (e.g. R/D+0), it also allows us to compare elections with very different national results, to see how much the underlying map has changed.

With the amount of spending, advertising, registering, and organizing that occurred in this election, targeted as it was at certain states, we should expect to see significant distortions in the maps. This is especially true if a re-alignment is taking place, as re-alignments are almost always accompanied by substantial regions of the dominant parties electoral coalition breaking off. A classic example here are the 1928 and 1932 elections. Given the substantial upheaval in the electoral college, it should be unsurprising that in 1932, only four states were within a 2 points of their PVI in 1928.

Below are the relative maps for 2004 and 2008. Pure purple means that the state’s PVI is between -2 and 2; slightly bluer means a Dem PVI of 3-4; each gradation means another two points until you hit the darkest blue or red, which means a PVI of 10+.

I will let readers draw their own conclusions, but to me the maps look largely identical. Despite a billion dollars spent between the two candidates in carefully targeting states, each candidate ended up, relatively speaking, within 2.5 points of where their counterparts ended up four years earlier, in thirty-one states.

These include several critical swing states where heavy advertising, organizing, and registering of voters should have moved them more than the national average. In the end NM, NV, VA, and CO moved about as little as did states that were ignored by the campaigns such as VT, IL, CA, and ID. What states moved?

Using 2.5 as the threshold for substantial change (e.g. rounds to 2 or less) we find that AL (3.6), KY (4.1), AZ (4.7), MA (5.1), WV (5.5), AK (5.7), TN (5.8), OK (5.9), LA (7.7), and AR (10.6) moved toward Republicans. For the Democrats the big movers were UT (2.8), DE (3.2) MT (3.7), NE (3.6), ND (4.2), IN (5.4) and HI (12.7).

 For the Republicans, we would expect AZ and AK to stick with McCain/Palin, as they are the home states of the nominees. Similarly, we would expect DE and HI to move toward the Democrats, especially since in 2004 Bush performed unusually well in HI for a Republican (Hawaiians have historically supported incumbent Presidents relatively strongly regardless of party affiliation). The remaining states break down pretty simply: Democrats have the Northern Plains states and Indiana, Republicans essentially have the Jacksonian diaspora and MA.

 That Massachusetts actually moved toward the Republicans, relatively speaking, may be a bit of a cautionary tale for Obama. This isn’t a fluke result – it was picked up in the polls throughout the election. There’s a couple of theories I’ve been toying with here, none of which is fully satisfying. The one I’m inclined toward is that many MA voters had heard Deval Patrick run the exact same campaign two years earlier, only to have him disappoint mightily once in office. Perhaps some MA voters brought a “fool me once, shame on me” attitude to the ballot box. Other theories are that this just represents the effects of Kerry leaving the top of the ballot (although this fails to explain why Texas didn’t move similarly toward Obama), or that Democrats have already “maxed out” among Democrats here (although it doesn’t explain VT and RI roughly keeping pace with the national average). Interestingly, neither Kerry nor Obama have matched, even in absolute terms, Gore’s 27-point drubbing of Bush – with another 6% for Nader!

 To explore the Jacksonian diaspora/Northern Great Plains/IN results, we need another couple of maps.

 Below is a map of relative change at the county level from 2004 to 2008. This is similar to the PVI approach, but it measures the change in a county rather than the result. Since the point spread moved about 9 points toward Democrats, we would expect, if only national factors played a role (e.g., housing collapse, bailout, Wall Street collapse, etc.) and everyone reacted similarly, that we would see every county move 9 points toward the Democrats. However, if you lived in a county where your own home’s value tanked by 10%, you will likely have an even more visceral reaction than most people. Your county will likely swing more than the country as a whole. In other words, if there are other factors at work than the national mood, you should see a swing of more or less than the average national mood resulted in

 So what I’ve done is colored counties that swung 8-10 points toward the Democrats in white. For every additional two point swing toward the Democrats (11-12, 13-14, etc), I’ve increased the blue tint of the county, and vice-versa. It is important to bear in mind that light red counties still moved toward Obama in absolute terms. Our goal here is to try and isolate the national effects. States with minimal local response to national conditions should be largely white – WA and to a lesser extent OR are a good examples of this. I didn’t do New England because the data at CNN is compiled at the town level, and my maps are at the county level (some day I will draw town maps of New England, but probably not until I retire and my kid is at college).

This map is noisier than our statewide map, but that is to be somewhat expected, since we have several thousand more datapoints. Before we get into the relevant states, note some salient features. First, it is safe to say that the Republican party’s position on immigration did not help it with Hispanics – note the Rio Grande Valley’s vote swing.

 Also, take a look at California. Note how blue the Central Valley is! This is likely a combination of the Hispanic vote and the fact that these counties – Merced, Stanislaus, Modesto – are the hub of the housing debacle. Riverside/Ontario in the south was similarly affected. We also note the blue “V” in Colorado – this largely represents the I-25 corridor through Denver and its suburbs, Broomfield, Boulder, and Fort Collins. Indeed, it was only the resistance of the remainder of the state to Barack Obama’s message that kept it from swinging more heavily.

 Take note of the fact that it is not the inner suburbs that swing heavily toward Obama. Cook county had a modest swing, DuPage a stronger swing, but the GOP was killed in McHenry and Kendall counties. In Virginia and Maryland, fast-growing Loudoun, and Prince William counties swung heavier than Fairfax. Montgomery, Delaware and Bucks County didn’t swing as much as Lancaster and Berks counties (there’s a typo for Delaware in my spreadsheet that I caught too late; it should be coded white).

 I have a theory here. When I started out as a newly-minted young professional, the lawyers in my firm consistently gave two pieces of financial advice: Buy a home for tax purposes and max out your 401(k).  Hundreds of thousands of young professionals followed that advice, and they settled in the outer suburbs.

 In the short term, that advice has not looked particularly sage. These young, affluent voters have seen their home values collapse and had their 401(k)s wiped out. These are likely their only two investments. It stands to reason that they would revolt against the status quo more than inner suburban dwellers, whose houses weren’t as hard hit and who likely had other investments saved up.

 Finally, note that the black belt across central Mississippi, Georgia, Alabama, and North Carolina is fairly white, indicating no change. This represents a limitation of my approach. Obviously these counties were critical to Obama’s victory. But they were already heavily Democratic. This is an area where turnout, not swinging votes, made a difference. But this map won’t pick up turnout (on a similar note, some cities such as Philly and San Fran are light red, because there the Democratic vote really was “maxed out” to the point that a further 10-point swing was basically impossible if any Bush voters remained). You just have to keep than in mind when interpreting the maps. 

Anyway, with respect to the states that swung toward McCain, the first thing we notice is our old pal Appalachia.

Our correlation here isn’t as strong as it was in the Hillary/Obama primary, nor would we expect it to be. After all, many of these counties, especially in Pennsylvania, were already Republican, and hence were unlikely to swing further toward McCain. Nonetheless, we note heavy swings toward McCain (+10 points in absolute terms) in Southwest West Virginia, Western Virginia, and Eastern Kentucky. Some of these counties have not gone Republican in a long time – I would venture to guess in some instances some have not gone Republican since the UMW organized them in the 30s. (If you want to see a larger version of this map, click here). 

Indeed, this swing is part of what kept Virginia and North Carolina from moving more than the national average – the relative stability, or even pro-McCain, movements in this part of the state counteracted the movement in Northern Virginia, Richmond area, and suburban North Carolina (note also that in Virginia, the outer NoVa counties such as Loudoun and Prince William moved a lot more than inner counties such as Fairfax and Arlington. This is again likely a manifestation of the rather severe housing contraction in those counties.

(By the way, if there is any state in particular you are interested in, e-mail me at Sean. Oxendine@gmail.com, and I will send you a map. No, I won’t send you all of them (mostly because of time constraints in uploading and sending all of them), but if you want one or two, I’ll see what I can do). 

At any rate, we notice, as we noticed in the primary, that there is a “tail” to the south and west where Obama underperforms. These are counties that were settled by pioneers from Appalachia, and some of the Jacksonian heritage lives on. This, I think, explains the movement in AR, OK, NE TX, and some of Northern Louisiana. To be honest, the movement in Southeastern Louisiana surprised me somewhat – I guess Cajuns don’t like Obama much? 

That leaves Obama’s over performance in the Great Plains. For this I think it is useful to zoom in a bit: 

The temptation is to say that Obama over performed here because of rural areas. Yet I find this unsatisfying. Why did Eastern NE move toward Obama, but Western NE did not? Why did KS not move? One variable that I thought might be at work here was a change in party ID as a result of Obama’s voter registration drives and organization there. After all, there is probably more low hanging fruit in NE than in NC, so we might see more returns on the investment than we saw there. But in ND, the site of one of the strongest swings, Democratic registration was only up a point. SD, IN, and NE saw four, four, and five point increases, roughly the national average. Organization and registration is probably part of the story, but not the whole story. 

Another portion of the answer, I think, lies in this graphic:

 If we take the map of “Major Corn Production” found at the link above and lay it over our “change map,” we get something that looks like this: 

Obviously it is not as pretty as our primary Appalachian map, nor should we expect it to be. There are a ton of variables at work here. Nonetheless, at least along the southern and eastern border, "major corn producing counties" seems to give a pretty good indicator whether a county swung substantially toward Obama. This makes sense, because as you get into Missouri, Kansas and Oklahoma, and into Western Nebraska and South Dakota, you get into wheat and ranching country, respectively. I’m still not sure what happened in Northwestern MN or Eastern ND, though I’ll note that the “major corn producing county” label is an imperfect “yes/no” label, and I’d be interested to see how much corn gets produced in those counties (or in the rest of Southern WI and the LP of Michigan) (I note here  that corn is a crowing crop in North Dakota). 

What does this matter? This gets back to an older post from Jay Cost, speculating that McCain had a rural problem. Namely, an ethanol problem. Every time that McCain talked about cutting wasteful government spending, I am willing to bet that every rural Iowa, Indiana, Western Ohio, and Eastern Nebraska corn farmer looked out over his cornrows and thought “he’s talking about me, by gum.” Especially since McCain’s opposition to ethanol is not exactly a secret. Farmers in this region are historically Republican, but this is by no means set in stone; the fact that five of the six representatives sent to Washington are Democrats is a relic of the 1980s, when the states were fairly close to the national average in the midst of a severe farm contraction. 

By the way, if anyone has a better dataset for corn production by county in the US (ie with actual numbers), I’d love to see it. 

At any rate, the bottom line is that as Republicans embark on a “cut pork” binge, it is important to remember that one man’s pork is another man’s paycheck, and this can hurt in critical states. 

That leaves Montana and Utah to explain. One state was heavily organized by Obama, and one was not. It think Montana was never as heavily Republican as it appeared in the Bush elections. After all, it did go for Clinton in 1992, and the Western portion of the state was historically heavily unionized and left-leaning (and while it isn’t a “clean” map there, the Western half of the state does look bluer than the Eastern, but this could be me just seeing what I’m looking for). Without the cultural affinity for rural-accented President Bush, these counties may have reverted to form. As for Utah, I note that Democrats’ portion of the electorate was only up a point or two from 2004, while Republican identification collapsed by eight points (for a mere 50-21 advantage, as opposed to 58-19 in 2004). Bitterness over Mitt Romney not being on the ticket? Anger among moderates over the church’s engaging in the gay marriage proposition in California? I really don’t know, and am open to suggestions. 

Two other quick thoughts. I think that most of the panic on the right is ill-advised. As we’ve seen, the underlying red-blue map is still largely intact, except for a few candidate-specific quirks. It shifted 10 points to the left, but that is to be expected in a year when our incumbent has a 25% approval rating, the economy collapsed, the housing market collapsed, the US was involved in two wars that are going mediocre at best, and the Republican candidate responded to the defining moment of the campaign by freaking out. Given that you probably have to go back to 1952, if not 1920 to find the country in similar straits in an open election, it is amazing that the shift wasn’t greater, and that Republicans were actually in a position to win until the bailout bill arrived (and I don’t think this underperformance is just racism, as ran roughly the same as Congressional Democrats nationwide. 

In other words, I think a lot of this talk of remaking the party is overblown. This isn’t to say that there isn’t work to be done, especially reaching out to younger voters. But we aren’t building from the ground up either. In a terrible two-cycle election we got 46% of the vote, which isn’t a majority, but isn’t terrible either. 

At any rate, the ball is largely in the Democrats’ court. If they deliver, the opportunity is theirs. If they don’t – and Obama is coming in at a pretty rough time to have any hope of delivering – the country will probably forget what it hated about Republicans pretty quickly (see, e.g., 1982 midterm elections). If/when Democrats screw up, the script will be written for us, just like it was written for Democrats. 

On a related note, this is a great opportunity for us in the House/Senate. Indeed, I’m not all that terribly broken up about McCain losing, as a McCain victory would almost certainly guarantee a strong Democratic showing in 2010, which allows them to control redistricting. And it would probably only forestall the inevitable arriving in 2012. 

But Republicans should remember that being out of power is good for Congressional majorities. I’ve outlined below the number of Republicans in Congress since 1932, and shaded the party that controlled the Presidency blue for a Democratic President, red for Republican. 

I think the correlation here is pretty clear. It would be highly unusual for Republicans not to make gains during the Obama Presidency. And this, I think, is where the developing RightRoots can, I think, make its greatest impact. Fundraising and messaging for the President is great, but $750,000,000 didn’t really move the dial all that much for Obama. Where big money can help is in Congressional races, where $100K to a candidate in seed money can make a huge difference. We need help identifying key races – it is embarrassing that we apparently had a 2-point race with a solid candidate in NY-24, and no one knew. Similarly, we came a lot closer to knocking off Chet Edwards in Texas than we did Jack Murtha, but we knew nothing about the candidate. 

The leftroots would have picked up on these guys. We missed them. This shouldn’t happen again.