The Democratic Veepstakes, 2008 Edition

There’s been a lot of talk lately (click here or here for examples) about a “dream” Democratic ticket of Clinton and Obama. As it happens, the quadrennial “veepstakes” is a topic on which my GW colleague Paul Wahlbeck and I published an article (here, gated) a decade ago in the American Political Science Review. So, pretending to be an expert, let me offer some very early thoughts on the matter.

First, let’s look at the prevailing wisdom about who’s likely to end up as the Democrats’ vice presidential nominee. The answer obviously depends on who ends up as their presidential nominee, but let’s leave that small detail aside for a moment. To tap into the current conventional wisdom, I visited the intrade website earlier today to see what their prediction market indicates about who are seen as the front-runners for the Democratic vice presidential nomination. Here are the current futures contract market prices for the 16 individuals that intrade lists, and for the “field.”

12.3 Barack Obama
10.2 Evan Bayh
11.7 Al Gore
08.7 Wesley Clark
07.5 Bill Richardson
06.0 John Edwards
04.2 Joe Biden
00.5 Tom Vilsack
04.0 Jim Webb
00.8 Mark Warner
00.8 Tom Daschle
01.4 Ted Strickland
03.8 Hillary Clinton
00.3 Bob Kerrey
00.5 Sam Numm
00.1 Christopher Dodd
27.7 Field (any other candidate)

In short, future traders see the race as wide open at this point, with Obama, Bayh, and Gore commanding the only double-digit bids and with the favorite investment at this point being none of the above, i.e., the field.

Editorial comment: The idea that Al Gore might end up as the running mate of either Hillary Clinton or Barack Obama seems pretty far-fetched, and so, for that matter, does the idea that Hillary Clinton might end up as Barack Obama’s running mate.

For a different take on these scenarios, I ran a pared-down version of the statistical model that Paul and I used to analyze presidential nominees’ running mate choices, 1940-1996. For present purposes, the only predictors in the model were those that had displayed statistically significant effects in our 1940-1996 analysis: whether the presidential nominee and a potential running mate were at least 10 years apart in age (that is, the “age balance” of the potential ticket), whether they had previously competed against one another for the presidential nomination, whether they had competed against one another in this year’s nomination contest, and their home state’s percentage of the 538 Electoral College votes. I then used those results to simulate the probabilities that Hillary Clinton would choose one of the 15 other individuals from the intrade market as her running mate and that Obama would do the same. Here are the results, first for Clinton’s first choice of a running mate and then for Obama’s.

.190 .xxx Barack Obama
.xxx .216 Hillary Clinton
.067 .034 Evan Bayh
.067 .074 Al Gore
.046 .050 Wesley Clark
.025 .027 Bill Richardson
.032 .016 John Edwards
.022 .024 Joe Biden
.050 .055 Tom Vilsack
.079 .088 Jim Webb
.079 .041 Mark Warner
.037 .041 Tom Daschle
.139 .152 Ted Strickland
.042 .046 Bob Kerrey
.095 .104 Sam Nunn
.029 .032 Christopher Dodd

Thus, according to our model, as in the intrade market, if Obama’s bid for the presidential nomination falls short, then he looks like the favorite for the second spot—but far from a prohibitive favorite. If the tables are turned and Obama heads the ticket, then our model shows Hillary Clinton as his favored running mate. I’ve already expressed my opinion that this outcome strikes me as highly unlikely, so do I not believe my own model? Well, it’s not quite as simple as that. Whereas the quesiton in the intrade market is who will end up as the running mate, the question in the Sigelman-Wahlbeck analysis is who will be the presidential nominee’s first choice as a running mate rather than who will end up on the ticket—and the first choice often doesn’t end up on the ticket. In this connection, I can well imagine, for example, presidential nominee Obama asking Hillary Clinton to join him on the ticket, even though I can’t believe that she would accept his invitation.

Some cautionary comments:

  • The Sigelman-Wahlbeck analysis focused exclusively on the pool of “finalists.” Some of the individuals mentioned in the intrade market aren’t going to be finalists. That’s one basis of incomparability between the intrade numbers and the model-based ones.
  • As a consequence, the Sigelman-Wahlbeck forecasts have no “field” option—another basis of incomparability.
  • It’s very early, and these numbers—both intrade’s and ours—should be taken with a ton or so of salt. In that respect, I will note that in 2000 Paul and I used our 1940-1996 results to try to predict who would be the Democratic and Republican vice presidential nominees that year. Among those we considered, neither Joe Lieberman nor Dick Cheney finished in their party’s top dozen. One could take that as an indication of the weakness of our model, or—my preferred interpretation—as proof (as if further evidence were needed) that both presidential nominees that year chose the wrong running mate.

5 Responses to The Democratic Veepstakes, 2008 Edition

  1. Matt Jarvis February 7, 2008 at 4:52 pm #

    Lee, your 1998 paper is quite fun–glad you posted the link to it.

    One question: so, size of state matters for the obvious “favorite son” hypothesis (even though us political scientists know it to be bunk, candidates think it might matter, as you note in the article). However, in the more modern period, where the partisan leanings of states are more fixed than they were for your 1940-1996 period, is it possible that the right variable might be some kind of “how many electoral votes would that person’s state likely deliver?” (You speculate on something like this in the paper, but it is 10 years later, so I’m curious what your thoughts on this are)

    So, under that logic, Lieberman makes sense because he might appeal to Florida (not his home state, but a special case?) which had been predicted to be a swing state with a pile of votes. Cheney should fit into your insider-outsider balance (sort of, he was also “gravitas”–generally not a problem for candidates, but a Bush problem in 2000). However, I don’t really expect anyone from Texas or California to get picked by either side because, even though the prize is big, the chance that the VP changes who wins them is slim.

    What’s odd to me about what I’m saying, though, is that it would seem to lead to Strickland or possibly Webb. Currently, 94.6% of the market would disagree with those picks (and I’m probably in that 94.6%, in the “not willing to wager my money yet” box).

    Also, just on the model, what happens if you don’t focus on the rivalry (this time or last) and focus on the question of whether or not the potential VP had RUN before, including just a few months ago?

  2. Dan Klein February 7, 2008 at 7:04 pm #

    Intriguing hypothesis; however, you seem to have mixed your Virginian Warners up. As your first table indicates, Mark Warner is the fellow favored on Intrade; octogenarian Republican John Warner seems like an unlikely pick.

  3. Scott McClurg February 7, 2008 at 8:42 pm #

    Out of curiosity, how different do you suppose the “finalists” pool you looked at in the paper matches what this intrade pool looks like? For example, I could see Obama considering someone like the female governor of Arizona very seriously even if she isn’t getting any love in the intrades market. That might explain the discrepancy between your predictions after 96 and your lack of success (the finalist pool is not the same as the pool you analyze for predictions).

  4. Oberon February 8, 2008 at 8:31 am #

    Unfortunately your model doesn’t include the possibility of a VP with $5 billion…i.e. Bloomberg

  5. Lee Sigelman February 8, 2008 at 12:02 pm #

    Dan: Let’s see — Mark is the one who was married to Elizabeth Tayl… No, that was John. Actually, I do know the difference. Mark is an alum of the very university where I draw my paycheck. I simply brain-locked and wrote “John” instead of “Mark.” Thanks for catching it; I’ve now made the correction.

    Oberon: I don’t see Michael Bloomberg as anyone’s vice president. I don’t think he perceives himself as a second-position kind of guy.

    Scott: I can see some of those in the intrade market as being “finalists.” But others — nah. Hillary Clinton just isn’t going to ask Tom Daschle, one of Obama’s mentors, to be her running mate. That’s one reason why the accuracy of this type of model depends in part on how early in the season it is. At this point, we don’t really have much of an idea about who the prospective presidential nominees might see as belonging to their running mate pool. Later this will become clearer. Not clear, but clearer.

    And Matt: Good point about weighting the state’s Electoral College vote by the party’s chances of winning the state. If we ever come seriously back to this stream of research, we’ll take that into account.

    My hunch — just that — is that Strickland and Webb and Mark (not John) Warner may well end up as finalists. Each has something to offer.

    It wasn’t clear in my write-up here, but our handling of the “run before” variable was tripartite: Ran this time against the presidential nominee; has never competed against the presidential nominee; or competed against the presidential nominee in the past but not this time. The latter got dropped as the excluded or reference category in the conditional logit regression model. We didn’t consider whether a finalist had ever sought the presidency in a year when this year’s presidential nominee wasn’t involved, but that, too, would be worth considering.