Test-driving California’s Election Reforms
SummaryIn the June 2012 primary, California tested two important electoral changes: new legislative and congressional districts drawn by an independent citizens commission and a “top two” primary system. The results suggest the reforms produced some changes—in particular, more open seats and more competition. However, there was also a great deal of continuity with recent elections: most candidates endorsed by a major party and all incumbents are advancing to the fall election and partisan outcomes were broadly in line with what might have been expected under the old primary system. Over time, the reforms may produce more radical change, but the first step on the road of reform has been a small one.
California tested two electoral innovations in the recent June primary. First, the state used congressional and state legislative districts drawn by an independent commission rather than the state legislature. Second, it employed an open “top two” primary that allowed voters to cast a ballot for any candidate for each office, regardless of party, with the two candidates receiving the most votes, again regardless of party, advancing to the fall election. This replaced a “semi-closed” system that required registered Democrats and Republicans to vote for candidates of their own party, and only sometimes gave independents the option to vote in partisan races. Only the presidential nomination continues to use this old system, while all other state and federal contests now employ the top two.1
Supporters of reform wanted to enliven California democracy by offering more choices and making officeholders more accountable to voters through competitive elections. Many also hoped these changes would help reduce political gridlock by increasing the number of representatives who either appeal to the center of the political spectrum or take a more problem-solving approach to governing.
It is still too early to evaluate most of these hoped-for effects. Instead, this report offers some preliminary analysis to gauge how successful the reforms have been at changing the electoral process. How competitive were the primary elections, and how engaged were voters? How much money did candidates raise? Did the reforms change the election outcomes from what we might have expected under the previous primary system?2
So far, the reforms appear to have opened up the process to more candidates by upsetting the status quo. The redistricting helped move many incumbents out of their comfort zone.3 In races for Congress and the state legislature, the average incumbent ran for a seat where 45 percent of the constituents lived in territory that was not part of the incumbent’s old district. In fact, 41 percent of incumbents were running to represent a district with more constitutents from new territory than old. Many either retired or ran for a different office instead, leading to high numbers of open seats: 35 (out of 80) in the state Assembly, 9 (out of 20 available this year) in the state Senate, and 9 (out of 53) in the congressional delegation.4
The redistricting also increased the number of seats likely to be competitive between the two major parties in the fall election, though the tendency for Republicans and Democrats to live in different parts of California prevented radical change on this score. A rough rule of thumb is that a seat is competitive if the relative share of party registrants never exceeds a 5-point advantage for Republicans or a 10-point advantage for Democrats.5 By this definition, there are 10 competitive Assembly seats, 6 state Senate seats, and 10 House seats in this election cycle, compared to 9, 0, and 4 in 2010. Most of the competitive seats fall in one of three places: the Central Valley, the central coast between Los Angeles and the Bay Area, and the Inland Empire east of Los Angeles. As an example, the competitiveness of Assembly seats can be seen in Figure 1. To better reflect the party balance in the state, Figure 2 distorts this traditional map to reflect the registered voter population.6
Figure 1. Assembly districts with competitive party registration are concentrated in pockets of the state
Figure 2. Democratic strength in the state becomes clearer with population-based map
The top-two primary altered the strategic context for elections in certain important ways. Under the old system, only one candidate from each party could advance to the fall election. In uncompetitive seats, that meant the dominant party’s strongest candidate could eliminate all serious challengers in the primary. But strong challengers from within the party can now survive to the fall election, where they get a second chance and can try to broaden their appeal. As a result, incumbents in uncompetitive seats were much more likely to face an intra-party challenge: 42 percent faced such a challenge this year, compared to an average of 19 percent from 2002 through 2010.7
Supporters of reform wanted to enliven California democracy by offering more choices and making officeholders more accountable to voters through competitive elections.
These extra challengers helped produce closer outcomes for incumbents. Winning incumbents led their top opponents, whether same-party or cross-party, by a much smaller gap than in previous years (an average of 28 points this election, down from an average of 39 over the last decade). Meanwhile, the margin for winning non-incumbents was about the same (16 points before vs. 14 points now). The odds of claiming an outright majority in the primary—making victory much more likely in the fall—were also lower: 59 percent of first-place candidates this year, compared to 75 percent in elections from 2002 to 2010.
Though outcomes were closer, they were not always close or unexpected. Every incumbent who ran is advancing to the fall, along with virtually every non-incumbent candidate endorsed by a major party (101 out of 113). Moreover, 88 of 102 incumbents led their closest opponent by more than 10 points, as did 98 of 113 endorsed candidates who are not incumbents. These numbers suggest that politicians who are part of the political establishment did well in this first trial under the new system.
One of the most intriguing innovations of the top-two primary is the prospect of fall contests between two candidates of the same party. There will be 28 such races: 18 for the state Assembly, 2 for the state Senate, and 8 for the U.S. House. All but one of these races is for a seat that likely would not have been competitive otherwise in the fall. The one exception is Congressional District 31, near San Bernardino, where Democratic registration outnumbers Republican registration by 5 points but two Republican candidates will face each other in the fall.8
The top-two primary also produced another significant change: minor party candidates will be almost completely absent from the fall ballot. Only three races will feature such candidates, and in each race the minor party candidate was a write-in against an incumbent who was otherwise uncontested.
In addition, five races will feature an independent, or “no party preference” (NPP) candidate. None of these NPP candidates has been registered with a minor party in California in the last 10 years, and four of them were registered with a major party as late as 2011.
Both supporters and opponents of reform have expected more money to flow into elections this cycle. The new top-two system not only offered candidates the chance to reach out to a wider range of voters in the primary (arguably costing more money), but also produced same-party contests in the fall general election for seats that would have had no serious fall contest under the old system. The redistricting might also force more fundraising by producing more open seats and more competitive races in the general election.
For every chamber (U.S. House, state Senate, and state Assembly), we counted all the money raised by candidates or spent by independent expenditure committees (either for or against a candidate) and divided by the total number of candidates.9 By this measure, signs of increased campaign money have been limited so far to congressional races alone. The average for House candidates was $333,509 in the 2012 primary, compared to $234,287 in 2010 and $241,626 in 2008. Money flowed more freely for both incumbents (up 43% on average from 2010) and non-incumbents (up 118% on average).10 By contrast, the average for Assembly candidates was $202,600, compared to $203,524 in 2010 and $215,441 in 2008. The average for State Senate candidates was somewhat more this year—$325,726 compared to $254,387 in 2010—but their numbers are down overall from several years ago.11
Figure 3. Only congressional races saw outsize money this primary election
There are reasons to think that candidates will need to raise more money this fall than in recent general elections. The redistricting has increased the proportion of competitive races between Democrats and Republicans, and the new same-party races will likely force a tight contest in many places that never would have had one under the old system. Nonetheless, it is important not to exaggerate this possibility. The top vote getter has already received an outright majority of votes in 83 of 125 cross-party races. In 32 of the remaining 42 cross-party races, one of the top two vote getters has received more than twice as much money as the other.12 While such a large financial advantage does not guarantee victory, it makes it more likely.
There are more signs of competition in the same-party contests, but still a number of easy races. The first-place candidate received an outright majority in 8 of the 28 same-party contests. Of the remaining 20 contests, 8 feature a first-place candidate who raised more than twice as much as the second-place candidate. That leaves 12 same-party races—2 for Congress and 10 for the Assembly—where the contest in the fall shows more signs of a tough fight.
There are reasons to think that candidates will need to raise more money this fall than in recent general elections.
In fact, despite all the changes, money still flowed disproportionately to establishment candidates. The average contested incumbent accounted for 89 percent of the money going to candidates of the same party, well in line with previous years. Likewise, non-incumbent candidates who were endorsed by a major party did quite well relative to the competition they faced from within their own party: they accounted for 77 percent of same-party money, compared to just 20 percent for those who did not receive an endorsement.
Did the reforms increase crossover voting in a way that changed the basic dynamics of elections? The top-two system allows all voters, not just independents, to cross party lines, and it allows them to do so race by race. If voters took advantage of this new freedom, the correspondence between party registration and party vote should be far less exact than it was under the old system, since voters may now switch their preferred party from one race to the next. The weaker the connection between registration and voting, the easier it is to conclude that the top two has changed voting behavior in the primary.13
If we assume the relationship between party registration and party voting from the previous system still held for the top two, we can account for 89 percent of the range of voting results across districts, missing the actual result by about 6 percentage points on average. This does not mean that crossover voting did not occur at all, or that it did not decide the outcome of certain races. However, despite new choices for voters, the outcome of the 2012 primary resembled what we might expect from a primary election where virtually no crossover voting occurred.14
On balance, the first test of these important reforms suggests that limited change has taken place so far. The reforms have encouraged higher turnover in the state’s political delegation and more fresh faces on the ballot, and that has contributed to a higher number of competitive outcomes.
But incumbents and other candidates connected to the major parties—who were key targets of the reforms—have been quite successful so far, and the vote shares were broadly in line with what we might have expected under the old primary system. The reforms have also made it significantly more difficult for minor parties to get on the fall ballot.
In short, the reforms have yet to produce a significant shift in the electoral status quo, and in some ways have reinforced it. Even the flow of campaign money, at least for the primaries, has not increased dramatically. And it remains to be seen whether the small shifts that have occurred will alter the way that representatives actually govern when serving in office. But change in any political system is slow and incremental. We will likely have many years to watch these reforms unfold, and the June 2012 primary election may prove to be the first small step toward a larger transformation.
The authors would like to thank Jack Citrin and Sherry Bebitch-Jeffe for their insightful comments on earlier drafts. Inside PPIC, this report has benefited greatly from the suggestions and effort of Mark Baldassare, Abby Cook, Bob Gleeson, Dave Lesher, Kate Reber, Mary Severance, Linda Strean, and Lynette Ubois.
- These are only the most recent in a long line of efforts to reform elections in California. Efforts to implement a commission for redistricting date as early as the 1920s, and were especially active in the 1980s and 2000s (Vlad Kogan and Eric McGhee, “Redistricting California: An Evaluation of the Citizens Commission Final Plans,” California Journal of Politics and Policy 4, no. 1 ). The state also employed an open “blanket” primary in 1998 and 2000 that strongly resembled the top-two primary in a number of ways. It was struck down by the courts in 2000, and the state used the semi-closed system from 2002 until this year. Most analyses suggest the blanket primary elected a slightly larger number of moderate representatives, especially to the state Assembly. See R. M. Alvarez and B. Sinclair, “Electoral Institutions and Legislative Behavior: The Effects of Primary Processes,” American Politics Research ( forthcoming); Bullock and Clinton, “More a Molehill Than a Mountain: The Effects of the Blanket Primary on Elected Officials’ Behavior from California,” Journal of Politics 73, no. 3 (2011): 915–30; McGhee, “Technical Appendix,” Open Primaries (San Francisco: Public Policy Institute of California, 2011), www.ppic.org/content/pubs/other/210EMAI_appendix.pdf.
- In order to compare the 2012 primary election results with previous primary elections, we ranked the vote totals of all candidates for a single seat in previous elections as if they competed together in a single race regardless of party.
- We define a candidate as the incumbent if they were indicated as such by the Secretary of State. The only exception is the two congressional races where two sitting members of Congress face each other; in those cases, the Secretary of State labeled only one candidate the incumbent, but we considered both candidates incumbents for the sake of our analysis.
- This compares with an average of 30 open seats for the Assembly, 10 for the Senate, and 3 for the House over the last decade.
- Specifically, we calculate the share of all registrants (including independents and minor-party adherents) who are Democrats and the share who are Republicans and take the difference between these two numbers. If this difference falls between 5 points Republican and 10 points Democratic, we consider the seat competitive. The asymmetry of this range reflects the comparatively uneven turnout among Democrats and their greater likelihood to cross party lines. While this definition is not perfect, it does capture the general range within which competitive seats tend to fall.
- This map was created in ArcMAP with an add-on utility that creates Density Equalizing Cartograms using the methodology developed by Mark Newman and Michael Gastner at the University of Michigan (Newman and Gastner, “Diffusion-based Method for Producing Density-Equalizing Maps,” PNAS 101, no. 20 : 7499–7504). Density Equalizing Cartograms change the shape of map polygons so that their size is based upon another attribute such as population. The size and shape of the polygons are changed, sometimes dramatically, but their original neighbors remain neighbors, and no new neighbors or new gaps are added.
- One might argue that the redistricting produced the increase in same-party competition for incumbents. However, the change was much more modest for incumbents running in competitive seats (21% this year compared to 15% over the last decade) and incumbents whose districts changed a great deal were not notably more likely to face competition than those whose seats had changed very little. Thus, the new primary system is a more likely cause.
- Idiosyncratic conditions may have played into this result. Two Republicans competed with four Democrats, and the top Democratic finisher was only 2.2 percent behind the second-place Republican.
- This excludes “soft” money spent by parties on campaign activities with a close nexus to traditional party building. This can include slate mailers to party members, get-out-the-vote drives, and even consultant salaries. There are ways to estimate the total amount of soft money in an election, but identifying the portion of national party money spent on California congressional races would be difficult. By the same token, committees that run issue advertisements which do not expressly advocate for or against the election of a candidate are difficult to connect to specific races, so we have omitted their totals as well.
- It does not appear that the increase in congressional races is driven by outliers. The patterns over time are very similar if the top 10 percent of candidates in terms of money are removed from each chamber in each year.
- This money includes all direct donations to candidates as well as any independent expenditures either for or against. We were only able to obtain electronic filing reports from the Secretary of State for the 2012 election by the time of this report; the numbers from earlier years come from the National Institute on Money in State Politics, which includes all candidates. Electronic filing is required when a candidate raises or spends more than $25,000, or when an independent expenditure committee spends more than $50,000, so electronic filers include all the most active and competitive candidates and organizations. For the sake of the calculations here, we assumed that all non-electronic filers in 2012 raised and spent no money. When we limited the earlier numbers to electronic filers alone, it did not change the substantive conclusions.
- Since we are now using money as a measure of competitive advantage, rather than a general measure of outside influence, we exclude independent expenditures against candidates from this calculation. The numbers are very similar if this money is included.
- This analysis is best at telling us whether any crossover voting that occurred mattered to election results; it is far less effective at telling us exactly how much crossover voting occurred. For example, in a race where every new Democratic crossover voter was matched with a Republican crossover voter, the overall outcome would be identical to a race where there had been no crossover voting at all. Similarly, in uncompetitive districts crossover voting by members of the minority party could be quite high without having much impact on the result, simply because there are not many minority party members in the first place.
- To obtain these estimates, we first calculated a two-party vote share for each district: the proportion of all Democratic and Republican votes that went to Democratic candidates. We then regressed the primary party vote on the Democratic and Republican shares of registered voters for 2002 through 2008, limiting the analysis to districts with at least one candidate from each major party on the ballot in the primary. Next, we generated out-of-sample predictions from this model for 2012, and regressed the actual outcome in each year on these predicted values. The numbers reported in the text are the adjusted R2 and root mean squared error for this regression. For comparison, we also predicted the outcome for 2010 with the same model. Not surprisingly, the fit is better, accounting for 97 percent of the variance and missing the actual result by about three percentage points. The coefficients and model fit for the regressions are available from the authors upon request.