A PSA on California’s Top-Two “Primary”
It isn't a primary and it is an illustration of how electoral rules affect behavior.
Yesterday, California held what has been widely called “primary elections.” However, this is not what they were. A primary election is, by definition, an election held within a given a party to determine the sole nominee of that party for the purposes of then competing for office in the general election. Primaries can be closed (requiring advanced registration by the voter) or open (the voter chooses their party affiliation on the day of the primary election without actual registration),* but in each case a party’s self-identified voters are picking a nominee who then will be on the general election ballot.
California (along with Washington) fully uses a system that is properly described as a two-round system where it is impossible to win the first round, but rather that the top two candidates from the first round advance to the second. The margin of victory is irrelevant, unlike a standard two-round system that will declare a final winner in the first round of voting if one candidate wins 50%+1 or more of the vote.**
So, the elections in California were not about party nominations, they were open competitions wherein candidates self-nominated and wherein each candidate is trying to win a slot to move on to the next round. These contests are simultaneously inter-party and intra-party. While designed to try and moderate the ideological leanings of the eventual contestants, they do not achieve that goal, if anything because a given district is unlikely to have a normal distribution of ideological perspectives, and this would be even more true of the active voting population. Single-seat districts tend to have concentrations of particular groups of partisans to begin with (due to geographical self-sorting) and gerrymandering exacerbates this natural tendency. Most districts (especially if we are talking about the US House) are not competitive.
For example, Nancy Pelosi’s district is, as one might expect for the Bay Area, heavily Democratic. Yesterday she won 68.5% of the vote, with her nearest competitor, Republican Lisa Remmer, winning 10.0% of the vote. Assuming the certified results do not change, this means that despite a 58.5% margin of victory over Remmer, she and Pelosi will face off in November. Note that the total Democratic vote-share in the district was 86.5%. The likely outcome of the November race will be Pelosi winning ~90% of the vote. Such an outcome understandably raises the question of why even have the second round.***
A counter-illustration to the way that California works is Alabama, which uses a two-round open primary system: partisans compete in self-contained elections and must win 50%+1 of the vote to win the nomination outright. So, for example, the incumbent Republican governor, Kay Ivey, won 56.1% of the vote last night, and thereby won the GOP nomination outright (as did Democrat Walt Maddox with a similar vote total). However, in House District 2 (a race I plan to write more about), the incumbent, Martha Roby is facing a run-off against challenger, and former US Representative, Bobby Bright because no one won the requisite number of votes to secure the nomination.
A great illustration of the fact that the top two system is not a primary is the US Senate race in California. The incumbent, Diane Feinstein, won 43.8% of the vote and the second-place finisher Kevin de León, a Democrat also, won 11.3%. As such, the race in November (pending final certification) will be Democrat v. Democrat. That is no primary. It should be noted that there were 32 candidates who received votes in that contest, ten of whom were Democrats, 11 were Republicans, 1 was a Libertarian, 1 was from the Peace and Freedom Party, and nine were Independents.
This type of system typically incentivizes entry into the competition because contestants know that they are not competing for a plurality win but, rather, second place (and that the margins do not matter). Since second place can be won by fairly small vote shares (see the previous two examples) there is an incentive towards fragmentation. Such fragmentation can lead to one of the major parties being shut out of the general election because there is no good mechanism for partisan coordination. For example, if there had been fewer Republican candidates in the Senate race it is likely those votes would have coalesced enough around the most popular Republican so that that candidate would have advanced to the second round for a two-party contest.
In fact, the Democratic Party was concerned it might not make the November ballot in a number of House districts it thinks will be competitive (via the LAT: Democrats look like they’ll make it through California’s House primary. Here’s what’s next).
Tuesday’s primary could have been disastrous for Democrats in California. With the largest portion of their national House battleground in play here, there had been concern for months that the state’s primary, which advances the top two vote-getters regardless of party, would leave voters with two Republicans to choose from in several key races.
[…]
By early Wednesday, it appeared it had all paid off, with Democrats poised to slide into second place for all 14 GOP-held districts in California. Democrats have targeted 10 of those and their candidacies will matter most in the seven seats where voters sent a Republican House member back to Congress but chose Hillary Clinton for president in 2016.
And while Democrats were worried about House contests, the CA GOP was worried about the governor’s race. Specifically, California Republicans are lucky that they appear to have managed the second place slot in the governor’s race, as there was a chance neither of the major statewide races (US Senate and governor), would have had a GOP nominee, which likely would have repressed GOP turnout which, in turn, would have had negative effects for the party on down-ballot contests (especially the US House).
The top two system does not lead to more moderate nominees and, worse, it further weakens already weak parties–and more could be said on these counts from both an empirical and normative POV. One thing is for sure: they are not primaries. Further, California’s system is a illustration of how electoral rules matter in terms of party, candidate, and voter behavior.
___________
*There are some variations to how this works in given states, but the basic categories capture the essentials.
**There are some other variations one might find (see, e.g., presidential elections in Costa Rica), but again, these are the basics.
***Note that the only reason to have a second round is that SCOTUS ruled some years ago that the actual election of members of Congress has to take place on, or after, election day in November. This was in response to a similar system in Louisiana which was declaring victors after the “primary” round if a candidate won more than 50% of the vote.
The Republicans are still free to coordinate however they like prior to the primary date. They could have a nominating convention. They could have their own party primary. They only catch is they have to do so without the taxpayers paying the bill for it. Which makes sense to me. As an Independent in Pennsylvania, I’m denied any voice in the state primaries because they’re supposedly private party functions, yet I’m still expected to pay for them.
Futhermore, so what if there’s no Republican on the ballot in November? If they couldn’t get even 10 percent of the votes, why do they deserve a slot over a more popular Democrat candidate just because they’re a Republican?
In what way are Feinstein and de Leon not moderate nominees? Particularly in contrast to Bradley?
@Stormy Dragon: The question is not what a particular race produces, but what the system purports to produce, and what it does produce. This system was touted as a means of producing, on a broad level, moderate candidates. It doesn’t work that way.
And look, you can argue that the right outcome for a statewide race in CA is a moderate Dem and a more left-leaning Dem, which is what that race produced. That is a different argument than the argument behind the design of the top two system. Also: there is a difference between comparing statewide races with district races–especially given the size of CA’s population.
@Steven L. Taylor:
You made a specific claim with absolutely no evidence provided. Can you provide any concrete examples of cases where California’s primary system led to extremist candidates being selected over moderate ones?
@Stormy Dragon: You make a fair point insofar as I have not provided specific evidence in this post, and I will see what I can do in perhaps an additional post at some point. At the moment all I can offer, without an investment of time, is that there is a literature, based on a relatively small number of election cycles (this has only been in place since 2012) to suggest that this is not happening as designed. Further, the basic nature of single-seat districts is not (as I do note in the post) conducive to that outcomes. The top two model seems to assume a Downsian distribution of ideology in a district (a bell curve with moderates in the middle). This is not the case. As such, the proposal is heavily flawed if the goal is as stated.
This is an area of expertise of mine, which you can take as will.
Fundamentally what the system does, for good or ill, is encourage a large number of self-nominated candidates all competing for a slot in the general election. It undercuts the value of party label as a signalling device (e.g., what does “Republican” mean if there are 12 of them in a given contest?). The ability of voters to even know who is moderate and who is extreme is limited.
It really takes away the significance of party labels in general.
There are comparative noted to be had for Japan under SNTV and Colombia until 2003–it is a system that encourages personalistic politics because it all but removes the filtering function of party labels.
And with extremely fragmented races, you really do not necessarily get an especially representative outcome in the general–probably the incumbent versus a nobody.
Also: if I can win the second slot with 10% of the vote–that is not a mechanism that limits potential extremism. Those kind of numbers are probably more likely to open the door to more extreme candidates than is a normal primary.
I cannot win an actual primary with 10% of the vote, for example.
@Steven L. Taylor:
Do you think an IRV system would produce the moderating effect? For those passionate about electoral reform (I count myself in those ranks, but maybe not the *most* passionate) it seems 2-round and IRV systems are favored. I had my misgivings about California’s system before it was implemented–namely that the voting population during the ‘primary’ is naturally skewed to the political wings, removing a democratic force for moderation. IRV, however, is intriguing.
@Steven L. Taylor:
I’ll defer to your expertise on the matter then, but thanks for providing some specifics in your follow ups.
@Neil Hudelson: I would prefer IRV to top two, yes. I do not think that it would lead to a moderating effect because of the preexisting ideological distribution of voters in most districts.
I do have general concerns about primaries in general, because as you note, they tend to attract only a certain slice of voters. I have moved over time from the position that primaries are a democratic good to thinking that more party controlled nominations would actually provide better (in the sense of being representative) choices for voters. But that is another (albeit ongoing) discussion.
Of course, if we could shift to some sort of PR a lot of these problems would solve themselves.
@Stormy Dragon: Thanks–and I noted that insofar as I am not pulling this out of thin air 😉
@Stormy Dragon:
But they didn’t/don’t. There must be some reason why. My guess: because there is *nothing to stop the Republican also-rans from running as “independent”, perhaps with publicizing how “conservative” or “pro-Trump” or whatever they are. So you still have a bunch of wolves in the race, even though all but one are wearing sheep costumes.
The alternative is to do what the DCC did – just advertise heavily which candidates the Democratic establishment thinks are most viable.
*this assumes mechanisms for party discipline are ineffective, as they most likely are.
@Stormy Dragon:
@Charon‘s comment reminded me: you are correct in the abstract, but the reality is that the nature of the rules make coordination very, very difficult (to the point of near impossibility). The RNC can try harder than they did, to be sure, but again: we are essentially talking about a process of self-nomination, which makes coordination a huge problem.
@Charon:
And even that is a highly limited form of coordination.
And the fact that they did so underscores how primaries, and especially Top Two, undercuts party influence over nominations.
@Steven L. Taylor:
It also consumes a lot of money the party would prefer to spend on something else.
Whenever I read about different systems and the contrast with the norm, I am reminded of what I modestly call Kathy’s First Universal Law: There are downsides to all choices.
One of the things I’ve learned over the years in business and management is that people have an overwhelming desire to fix immediate problems by changing the process. These changes almost always have unintended side effects, often bad, so my reaction to such proposed changes has become inherently conservative, i.e. work with the existing system and tweak it if necessary.
It’s not an exact corollary though. What I actually tell people is “Never fix a managerial problem by changing process and never fix a process problem by changing management.” The difficulty with elections is that, absent a strong party, there is little effective management and so we are left with only the “process” stick left to whack thing with.
@Kathy:
MarkedMan’s First Universal Law is similar: At a deep enough level, everything is complicated.
@Kathy:
Well, sure. There is no perfect choice. But that does not mean that some choices aren’t superior to others.
@MarkedMan:
Except that we know that the process used affects whether or not the system produces strong parties. They are not wholly separate variables.
@MarkedMan:
You know they’re all derivations of Chaos theory and Fractal math, right? 😉
@Steven L. Taylor:
Absolutely. It’s just that in contrasting any two systems, the downsides of one are often left out. I’m not saying this applies in your post.
This is related to the absolute solution. The idea that things ought to be just one way, and if/when that way doesn’t work out, then the opposite must be the solution. That’s partly how the opioid epidemic came to be.
The corollary to Kathy’s First Law is: you should adopt the desirable solution with the downsides you can tolerate, ignore, make up for, afford to pay for, etc.; keeping in mind the ways you cope may have second-order downsides as well.
I really dislike the top two primary system, but I don’t see it changing any time soon. It might be nice if there was at least a write-in option on the November ballot, although that has drawbacks as well.
At the same time, if the second place finisher had been Villaraigosa, perhaps the fall election might have been a nasty Democratic internecine fight that may have hurt the Dems…
@Steven L. Taylor: Yes. I should have been more clear. It was an observation in what might have led Californians to make such a bad choice
This, to me, is very revealing. This top two primary system was touted as awesome … right up until the point where it was going to produce Republican-only general elections. I can’t help but think the entire point of this is to shut the GOP out of elections. If Alabama had a similar system and the recent senate election had been between two Republicans (one of whom was Roy Moore), I have no doubt that it would have been as the partisan hackery it is.
@MarkedMan:
You know, I think I bring up air accident investigations too often. The format and method focus on identifying causes and contributing factors, and then fixing them by whatever means are necessary, and it’s not punitive at all.
On first glance, then, one would think: why not take that approach with every other problem? After all, air accidents trend continuously downward, right? Commercial aviation gets safer every year, right? and one can point to many implemented solutions that either prevented accidents later on, or at least saved many lives (like using Cockpit Resource Management in the case of the United DC-10 that crashed in Iowa after losing an engine and all flight controls, or the Qantas A380 with the exploding engine that perforated the wing).
But there are downsides to all choices.
Air accident investigations and the solution implemented are focused on one single factor: safety. That’s not necessarily the focus of every other area, and in some cases it’s not even a factor. You may lose a contract or a sale through mistakes and shoddy practices, but no one will die or be injured by it.
IMO, the idea of investigations being non-punitive does apply, or can apply, far more broadly. The notion is that people will come forward and report their own mistakes and shortcomings if they know they con’t suffer adverse consequences for doing so.
Past that, the focus should be on identifying not just the error, or errors, but what led to it, and then how to prevent it later on. And also identifying deficiencies, obstacles, bottlenecks, etc. and how to get past them. With the form of the solution being less important than its result.
Then the problem is watching out for the downsides. For instance, you may make sure to finish all projects on time if everyone works longer hours, but that will cause burn out in short order.
At my job we work with inflexible deadlines. Often we hand things off just hours before the deadline, and that’s definitely not good, especially for the parts that are made or checked near the end. Everyone’s tired, fed up, it’s late, and we all just want to finish in any way and go home.
The reason I think these top two primaries will increase partisanship is this: we’re familiar with how the primary-general system works. Politicians tend to cater to the base in the primary and then tack back to the center for the general. Remember Romney’s etch-a-sketch thing. But when the race is between two people of the same party, the balance point is not between the right and left wing; it’s between, say the moderate right wing and the extreme right wing. It’s the point Bartlett made about how divided government tends to be moderate while united government tends to be extreme: you make deals with the outlying wing of your own party instead of the other party.
@Hal_10000:
I am not sure much of anyone thinks that this system is “awesome.”
Also, note that the Republicans were worried about being shut out of the governor’s race and were shut out of the senate race.
While I am sure someone, somewhere, may have made the claims you are making, the stated goals of the top two system were not shutting out Republicans (or anyone else).
@Hal_10000:
You are making a false assumption about ideological distribution, especially in congressional districts. What is “the center” in Pelosi’s district, for example?
This is what I mean about a fundamental error in the top two design: it assumes that districts have centers that are in the center of the spectrum. This is often not the case due to geographic sorting and gerrymandering.
@Steven L. Taylor:
You will get ‘moderate’ candidates if there are enough moderate voters to make that happen.
Where I live, in Sonoma County, in the coastal country suburbs 65 miles north of San Francisco, I am what passes for a ‘moderate’ – I voted for Dianne Feinstein (definitely moderate), I favor Free Trade, I supported a reduction in the Corporate Tax Rate, and for the most part I have generic liberal views on environmental and civil rights issues. That said, by almost any standard elsewhere in the country, particularly in Red States, I would be seen as a leftist, a progressive. Many of the people in my neighborhood were supporters of Bernie Sanders and/or Jill Stein, and countywide, Hillary Clinton received 71% of the vote, Trump 22%.
In Marin County, about 40 miles south, there’s an affluent and well-educated base that voted 79% Hillary and 16% Trump. Trump got less than 10% in San Francisco, and most people don’t realize that the Chinese-American community is the middle class base of the city. In Silicon Valley Trump got 21%. Are Trump’s tax cuts going flip Silicon Valley voters? I don’t see it.
Trump is not moderate, so he and his people are are going to have a hard time presenting themselves as a ‘moderate alternative’ to Democratic Party candidates. To Republicans, they are the non-communist non-Venezuelan alternative to HugoChavezesque Democrats.
2018 is all about turnout. Those ‘moderate voters’? Where are they? Seems to me that people SAY they’re moderate but that does not translate to moderate electoral results.
@al Ameda: “Moderate” is a relative term, to be sure. But in a district that was 71-22 for HRC, the skew is clearly on the American “left” (which is a moderate in a broadly comparative sense-I get that).
The bottom line for this post is that your district does not fit the alleged goals of top two (but, then again, most districts don’t and even those which come close don’t behave as expected).
@Stormy Dragon: Although California has used the top-two in regular elections only starting in 2011, it used a blanket primary in all special elections for state office and congress between 1967 and 2010. Both John Birch Society members who were elected to Congress from California were first elected in special elections in which all candidates appeared on the same ballot, in the first round. Also David Duke was elected to the Louisiana legislature in a top-two election.