Why Don’t We All End Up at Super-Super-Duper Tuesday?

Talking Points Memo has a piece suggesting that the decision of Florida to move up its primary date is having knock-on consequences – the South Carolina GOP is threatening to follow suit. What’s interesting to me is why we don’t see more than this. In the very simplest game theoretic model of this, assuming that all states benefit from being earlier in the primary sequence, and are hurt by being later in the sequence, you would expect that all states would converge on holding their primaries on the same day – the very first day possible. You would also expect that threats like the one in New Hampshire law (that New Hampshire will always hold primaries one week before any similar contest in another state) to be non-credible, since if everyone ends up converging on the earliest date possible, New Hampshire won’t have that option. But – even if we see persistent pressures on states to move earlier in the calendar, we don’t see the kind of free-for-all that this theory would predict. Why not? Are the strategic motives of states more complicated than in this simple model (presumably they are: but in what way?). Do some states simply not have any reason to care where they are in the process (while other states, like Iowa and New Hampshire do?). Not being a scholar of US politics, I don’t know if there is a literature on this, and if so, whether it says anything useful, but would be very interested to know more about what is preventing the expected race to the bottom from taking place.

15 Responses to Why Don’t We All End Up at Super-Super-Duper Tuesday?

  1. Rob September 29, 2011 at 12:37 pm #

    I think one reason why we don’t see the mad dash to be first, or nearly first, has to do with actually giving the candidates a chance to campaign in the state. If everyone were scheduled early on candidates would simply focus on the biggest prizes. The only reason that doesn’t happen in a general election is because many states are predictably blue or red.

    Second, there is no real advantage to going first or second or third. The early states get more contact hours with all the candidates, but the later states get more exposure with the candidates that are truly viable. The only real disadvantage is being in the bottom tier of states when a nomination contest is not competitive, like the GOP race after Super Tuesday in 2008. The Democratic field, on the other hand, was a two candidate race down to the wire, or nearly so. So a part of the calculation for a state to schedule a primary date may be to maximize influence on the remainder of the field rather than dividing it’s delegates too thinly by going too early.

  2. dick winters September 29, 2011 at 1:11 pm #

    One constraint on the “race to the first” or “…to the convergent date” is that NH’s threat is a credible one. The state and locals are so “good” at what they do — election administration — that they can afford to wait everyone else out, and then re-jigger the schedule so that they are a week ahead.

  3. Keith Smith September 29, 2011 at 1:37 pm #

    I think part of the story has to do with state-level politics. A balance has to be struck between the cost of administering elections and the desire to be influential in the process. In California, for example, where we hate political parties and try to constrain them as much as possible, the state pays for the primaries. During the 2008 cycle, we had two sets of primaries–the presidential primaries in February and the primaries for everyone else in June. Doing so costs a lot of money, though. The state moved the presidential date up in 2008 in order to get a better return from campaign spending (legislative leaders said they were tired of being the ATM for the rest of the country and wanted the money spent in California). This time round, they’ve moved the primaries back to June because they decided it wasn’t worth the cost and effort to have a separate set of events. In the background, though, are state-level partisan considerations. In 2008, there was a race that the Democratically controlled legislature cared about influencing. This time, not so much.

  4. Michael Hagen September 29, 2011 at 1:41 pm #

    Some of the calculations surely have to do with the presidential nominating process itself. But those who ultimately decide when their state’s primary will be held—state legislators—have a host of more parochial considerations in mind as well. In Pennsylvania in 2004, then-Governor Rendell created a task force to study election reform, and moving up the date of the primary—normally in mid-April—was on the agenda. The major impediment explicitly acknowledged by the task force (http://www.portal.state.pa.us/portal/server.pt/community/election_reform/12652) was that moving the state’s primary to early March would have the effect of requiring candidates for state offices to circulate their nomination petitions during the holiday season, and no one liked the sound of that. The task force considered getting around that problem by leaving the primaries for state offices in April and conducting a separate presidential primary in early March, but that would have meant paying the administrative costs of two elections rather than just one, and the benefits of that option were judged to be outweighed by the costs. At least one more big consideration went unacknowledged in the task force report: moving the Pennsylvania primary to a spot on the calendar that would lead presidential campaigns and many more voters to sit up and take notice would yield primary electorates (Pennsylvania primaries are closed) far different in size and composition from the one that candidates for state offices can normally expect, and that prospect didn’t appeal much to incumbent legislators. Although the task force wound up recommending that the primary be moved up, and the legislature held hearings on the proposal, the 2008 Pennsylvania primary was on April 22. Of course, thanks to structure and circumstances, we got to see a pretty good show nonetheless, on the Democratic side at least, so political junkies here had only the usual reasons to be bitter.

  5. Scott Monje September 29, 2011 at 2:14 pm #

    Is there an “earliest date possible”?

  6. Adam S. September 29, 2011 at 2:15 pm #

    To add to Michael Hagen’s point, a similar scenario played out in Texas in 2008. The legislature favored moving our March 4 primary to Super Tuesday, but the bill was blocked at the last minute by state senators who had been lobbied by local election officials. They objected to moving the filing deadline to December because, in addition to it being the holiday season, an earlier deadline would have forced potential candidates who held other statewide offices to resign earlier.

    Then along came the 2008 Democratic primary, where Texas actually mattered. Accordingly, there was no effort to move our primary up this time.

  7. Paul G. September 29, 2011 at 4:38 pm #

    I agree with what Keith wrote above (our primary is in May, and it’s mainly a local affair: we have a number of non-partisan races that are essentially decided in that election, and often have referenda or initiatives on the ballot).

    Second, we’d have to consider whether there is really much gained by being early. Some additional advertising dollars, some candidate visits, a week or more in the national spotlight.

    But since we’re playing game theory, is there any marginal benefit to the state legislators, state policy makers, or the political establishment of being early? It may be that a state has been quite happy with the presidential field in the past. It may be that a state has an influential delegation and doesn’t need (or can’t afford–high profile primaries surely have a cost as well as a benefit) the “bump” provided by an early primary.

    We all talk about ethanol and Iowa, but I haven’t seen much literature (the ethanol stuff is mostly anecdotal and could be a result of influential members of Congress and the makeup of committees) that shows that being early really makes much of a difference.

  8. Acilius September 29, 2011 at 5:09 pm #

    New Hampshire’s law gives the secretary of state the authority to set the date of the presidential primary, while other states require the legislature to pass a statute setting the date. That gives New Hampshire a shorter reaction time than other states, which is one of the reasons why they’re so good at staying first.

    Also, New Hampshire does tend to punish candidates who go along with efforts of other states to pre-empt that state’s traditional first-in-the-nation status. So in 1996, Delaware voted early. Malcolm Stevenson Forbes, Junior, was the only Republican to campaign there. Forbes’ win in Delaware was ignored, while his bad performance in New Hampshire did hurt his prospects. The message to subsequent candidates was that they were not especially likely to benefit from participation in pre-NH contests, while they were certain to pay a price for doing so.

  9. Tom Hammond September 30, 2011 at 12:16 am #

    I’d like to push the point by “Rob” a little further. I’ve long thought about this strategic choice by a state (or a state party, or a faction of a state party that favors a particular candidate) as a choice between ‘setting the agenda for later states’ versus ‘choosing the final nominee from the agenda set earlier.’ It is not obvious to me which choice — setting the agenda versus choosing from that agenda — would yield the highest utility for a state (or party, or faction), at least a priori. Most media attention seems to focus on those states that want to go early, but the fact that by no means all states are choosing to pursue this particular “go early” strategy suggests that they see some advantages in going late (e.g., they get to influence, or even determine, who the final nominee will be, from the ‘short list’ established by the early primaries).

    I have long thought it would be instructive to set this problem up as a multi-state non-cooperative game, and then ask whether this game has any kind of Nash equilibrium as to when the states choose to hold their primaries: i.e., “as long as no other state changes when its primary is, I won’t change when mine is.” I don’t have any particular intuitions as to the answer to this question. The fact that states keep jockeying with each as to when their primaries are held might seem to suggest that there is not a Nash equilibrium; that is, for any strategy (choice of date) by the other states, the last state can make a choice of date that leaves it better off and one or more other states worse off, and this is true for each state. If no Nash equilibrium exists, this alone might account for why the states keep jockeying with each other as to when to hold their primaries.

    Of course, this strategizing by a state as to when to hold its primary is complicated by additional factors, such as how many candidates are running from a given “wing” or “faction” of the party (“factional splitting”), and what rules each state uses to allocate its delegates among candidates (e.g., statewide winner-take-all, statewide proportional, or districted winner-take-all). Since these also vary from year to year, it makes multi-year comparisons and data analysis more complex.

    I tried to wrestle with a few of these questions a long time ago in a 1980 article in the Western Political Quarterly titled “Another Look at the Role of ‘The Rules’ in the 1972 Democratic Primaries.” To the best of my knowledge, not much has been done along these particular lines since, other than in a handful of subsequent studies by Paul-Henri Gurian at the University of Georgia in Athens. Most political science attention seems instead to have been focused on the role of media, the role of early money, and the role of “momentum.” To my mind, the impacts of sequence (the topic of this debate), of factional splitting, of the delegate allocation rules, and the interactions among these three variables, have been unwisely neglected in studies of nomination contests.

  10. DavidN September 30, 2011 at 4:38 am #

    Game theoretically the New Hampshire law can be construed as deterrence. Analogous to companies advertising to match or undercut price of anyone who is cheaper than them.

  11. Henry Farrell September 30, 2011 at 10:04 am #

    Thank you all – and in particular Tom Hammond for these insights. Tom – that sounds like a great article waiting to be written (by sheer coincidence, I was telling my Ph.D. students yesterday in class that they really, _really_ needed to read the “Why Politics Is More Fundamental Than Economics” piece as a really nice example of how to use simple mathematics to convey important and non-obvious arguments about the relationship between institutions and efficiency, or lack of same).

  12. will September 30, 2011 at 12:24 pm #

    I get that states don’t all want to be first. But what’s the deal with really late primaries, such as South Dakota in June? It’s only on rare occasions that they would be meaningful. In 2000, 2004, and 2008 (on the Republican side) all but one of the candidates had dropped out before the primary calendar was done. This means many would-be primary voters are totally disenfranchised, just as much as if their state had been stripped of delegates.

  13. Tom Hammond September 30, 2011 at 3:48 pm #

    A trio of further thoughts, the first two stimulated in part by Will’s observation.

    First, let’s consider an extreme case of primary-sequencing dynamics. If all states try to front-load their dates, with all of them having primaries on the same day, the consequence is that the net impact of any one of them will be diluted. (This would, in effect, create something like to a “national” primary by default.) And with the diversity among the states ideologically, ethnically, religiously, etc., and assuming some comparable diversity among the candidates, this would suggest that no candidate would be likely to come away from this first set of primaries with anything like a commanding lead, much less a majority of the delegates. (Indeed, with a single early “national” primary, no candidate would be likely to arrive at the convention with a majority of delegates, which then means some heavy-duty brokering and bargaining would be needed to select a nominee.) With the influence of each of these first states thereby diluted, and with no candidate likely to come out of the first set of primaries with a commanding lead, at least some of the states might then rationally choose to move their dates a little later, hoping to capitalize on the likelihood that several candidates will still be viable and on the likelihood that the later states would therefore be more likely to be the ultimate kingmaker(s).

    Of course, one could play with the opposite extreme example, with the same logic following. For example, assume that all states will want to be final kingmakers, and thus will schedule their primaries to be on the last possible date. But again, the impact of any individual state will be diluted as a result, and so at least some of these states might try to increase their influence by moving their primary dates earlier, so as to gain some agenda-setting influence, since no one state will be able to have as much kingmaking influence as it might prefer.

    Thus, while there may not be any particular overall Nash equilibrium, it does seem that at least some aggregations of dates — everybody on the first day, everybody on the last day — would certainly NOT be in any kind of equilibrium, and at the least should not be expected empirically.

    Second, if it seems reasonable to think (at least it seems reasonable to me!) that states will realize, before primary season actually starts, that there is a lot of uncertainty as to how the results of the primaries will ultimately turn out (due at the very least to uncertainty as to how many candidates from which faction will actually run), then it may not be irrational for a state to hold its primary at a rather late date (as with Will’s case of South Dakota). In effect, the state would be making a guess that two (or more) significant candidates will still be viable that late in the race. The state will be wrong some proportion of the time; this is always true in a world with uncertainty. But sometimes the state won’t be wrong, and in that case it may be very valuable for the state to be the kingmaker. This may be true often enough empirically to make this strategy a rational one: think of the Obama-Clinton race which went down to the wire in 2008, and the Humphrey-McGovern race in 1972 which was effectively decided by California on the last primary date, if I recall correctly.

    Third, the status of small states like South Dakota may be a little tricky to fully understand. Of course, one problem for a small state is that, if they are not the very first, or among the very first (e.g., like Iowa or New Hampshire), and if their primary date is the same as a much larger state, then such a state may simply not attract much attention, and thus would have relatively little impact on the outcome. However, I would certainly predict that small states would at least avoid setting their dates on the same day as a much bigger state, if that is possible.

    I do think, though, that we will find examples in which small states have scheduled their dates at the same time as a big state, which contradicts my logic here. (I vaguely remember that in 1972 and 1976 — don’t trust my memory, though! — that South Dakota had its primary on the same date as California.) It may simply be that, in general, small states are always going to be in a pickle (a well-defined term in game theory…), precisely because they are small. And if their small size means that they are not going to have much influence no matter what they do, then non-strategic factors may be expected to influence their decisions. (E.g., in South Dakota, are they concerned about simply getting the fields ploughed and the seeds in the ground in the spring before electioneering needs attention, and since the arrival date of warm spring weather suitable for planting is always unpredictable, the resulting choice is to hold the primary as late as possible?)

    Third, as much as I appreciate Henry Farrell’s gracious reference to “Why Politics Is More Important Than Economics,” I should acknowledge that my coauthor Gary Miller had the original insights which stimulated that paper. Gary should get most of the credit!

  14. Scott Monje October 1, 2011 at 10:44 am #

    Would small states benefit from grouping several together on a Super Small Tuesday?

  15. Tom Hammond October 3, 2011 at 8:20 am #

    To respond to Scott Monje’s question, plus a couple of other thoughts:

    1. Would small states benefit from grouping several together on a Super Small Tuesday?

    Response: I have somewhat ambivalent views here.

    On one hand, if a small state and a big state are holding their primaries on the same day, then the media attention will tend to go to the big state, plus the big state will, by definition, have more delegates at stake than will the small state.

    On the other hand, just because a state is small does not mean that it shares any kind of interests with other small states. I’m inclined to think that a key issue is not so much big-vs-small states, but state parties (or factions of state parties) with particular ideological or philosophical views (e.g., “liberal” vs. “moderate” vs. “conservative” parties or factions having primaries on the same date as other state parties (factions, etc.) with other ideological or philosophical views.

    In either case, the question for any one state (party, faction) is how it can maximize the impact of its views on the final outcome (i.e., who the nominee is). So it is not clear to me whether a state would most benefit from (a) having its primary on the same date as a big state with similar views to its own, or (b) having its primary on the same date as a big state with different views from its own, or (c) having its primary on the same date as several other small states with similar views to its own, or (d) having its primary on the same date as several other small states with different views from its own, or (e) having its primary on the same date as several other small states with a diversity of views.

    2. It may well be the case that, for any one state (party, faction), what matters is not how much impact that that state (party, faction) had on who becomes the final nominee, but on whether that state (party, faction) likes or dislikes the final nominee. If I am a small state, for example, I may not care if my state has much, if any, impact on who the final nominee is, as long someone I like a lot gets the nomination. Indeed, if I could always be 100% certain that a someone I like a lot gets the nomination, I might not care if my state had no contest at all.

    3. For a general model of the nomination process, I’m beginning to realize that a couple of key variables would be (a) the number of states holding primaries, versus (b) the number of “available dates” for holding primaries. If (a) is greater than (b), then it is more likely that any one date will see two or more state contests on that date; if (b) is greater than or equal to (a), then it is at least possible that each state could have its own primary on a unique date, with no other state holding a primary that day. If the latter relationship holds, then the question considered in #1 above could in principle be avoided. If the former relationship holds, then the question in #1 above cannot be avoided.

    4. My fourth point is that the date in the nominating season will influence what a candidate cares most about at that time; that is, the candidate’s strategy will be time-dependent. Early in the nominating season, the candidate’s goal is at least to survive, and hopefully to prosper, by winning or placing “better than expected” in the early contests. At this stage, the candidate will not really care very much how many delegates he wins. But for those candidates who survive until the later stages (which means they have either won a number of early primaries, or at least have consistently done “better than expected” in these early contests), the candidate’s concerns will increasingly be to maximize how many delegates are won. This is why, as much initial discussion suggests (and my 1980 Western Political Quarterly paper argued), the delegate allocation rules can matter a lot. Especially late in the season, the delegate allocation rules can make or break a candidate’s chances of winning the election (though of course by then, how many delegates the candidate has won will have been influenced by the delegate allocation rules in the earlier primaries or caucuses).

    5. Whether a candidate puts a lot of effort into any one state will presumably be influenced by (a) how many other candidates from the various factions are running in that state, and (b) what the delegate allocation rules are in that state. (Item (b) will likely be more important in later primaries rather than earlier ones.)

    For (a), in a state with a primary electorate of 50% “liberal” and 50% “conservative,” and if there are 5 liberals running and just one conservative, and if the state rule is Winner-Take-All, it will be very unlikely that a liberal candidate will be able to win any delegates. Thus, liberal candidates might think about not running in that primary. But, of course, if several of these liberal candidates were thinking of not running, then if just one liberal runs, he might have a respectable chance to winning all the state’s delegates. Which of course suggests that these other liberals might NOT want to avoid the primary after all. It’s not clear to me what the liberal candidates’ strategies should be here.

    In contrast, if a state’s delegate allocation rules are proportional in some sense (the candidate will get x% of the delegates if he wins x% of the vote), then there is less strategizing that a candidate will need to engage in: more candidates will likely enter because for each candidate the chances of winning at least some delegates will be high.