“Here is a telling statistic. One hundred fifty-three of California's congressional and legislative seats were up in the last election, and not one, I repeat, not one, changed parties. What kind of democracy is that?”

With those words in his 2005 State of the State address, Gov. Arnold Schwarzenegger introduced a proposal to reform the electoral process in the Golden State. The problem with California's congressional and legislative elections, according to Schwarzenegger, is that the Democratic majority in the Legislature has used its control over redistricting to maximize the number of Democratic seats by packing Republican voters into as few districts as possible. It's called gerrymandering, and it has been going on since the days of Elbridge Gerry, the 19th-century Massachusetts politician whose political mapmaking skills gave birth to the term. The difference today is that 21st-century politicians have far more sophisticated mapmaking tools than did Gerry. Computer programs allow legislators to use data on voting patterns and demographic characteristics that are broken down to the city-block level (Monmonier 2001). The result, according to Schwarzenegger and other critics of current redistricting practices, is that most seats are safe for one party or the other and it is almost impossible for the minority party to threaten the majority party's control. The solution, the critics say, is to transfer responsibility for drawing state legislative and congressional districts from the Legislature to independent, nonpartisan commissions such as the panel of retired judges proposed by Schwarzenegger. These nonpartisan commissions would presumably create more competitive districts.

When Schwarzenegger talks, people listen and since his speech, the Governator's proposal has generated favorable reactions from many pundits and editorial writers inside and outside of California. The Los Angeles Times (2005) stated that, “Schwarzenegger could hardly pick a worthier reform than that of taking politicians out of the business of drawing electoral district maps…. Political parties love carving out safe districts for themselves, and this year's gerrymandering is eroding the vitality of California's democracy.” Columnist Cragg Hines of the Houston Chronicle asked, “Can it really be any worse than how most do it now?” and concluded that Schwarzenegger's proposal “certainly seems worth a look-see.” The Christian Science Monitor (2005) opined that, “A panel of judges or a balanced commission seems preferable to lawmakers using sophisticated computer programs to defend their own turf year after year.” Newsday (2005) endorsed a similar plan for drawing county legislative districts on Long Island, arguing that, “Gerrymandering brings perpetual incumbency and bad government.” And the Washington Post (2005), in an editorial titled “A Model of Reform,” summed up its reaction this way: “Go, Arnold!”

It is understandable that political commentators across the United States have been favorably impressed by Governor Schwarzenegger's proposed reform. Lack of competition in elections is a national problem. In the 2004 U.S. House elections, only five challengers in the entire nation succeeded in defeating an incumbent. Of the 435 seats in the House, only 22 were decided by a margin of less than 10 percentage points. But while making elections more competitive is a worthy goal, our research on House elections that have taken place since the 1970s shows that redistricting has not made these elections less competitive and that turning control of redistricting over to nonpartisan commissions would not necessarily increase competition.

We classified House districts as safe or competitive based on the major party vote in the most recent presidential election because the presidential vote closely reflects voters' party loyalties and is not influenced by whether an incumbent is running in a district or how much money the local candidates spend. Districts that voted at least 10 percentage points more Democratic than the nation were classified as safe Democratic; districts that voted at least 10 percentage points more Republican than the nation were classified as safe Republican; districts that were within 5 percentage points of the nation were classified as competitive. In the 2000 House elections, for example, a district that voted 60% or more for George W. Bush was classified as safe Republican, a district that voted 60% or more for Al Gore was classified as safe Democratic, and a district that voted between 45% and 55% for Gore vs. Bush was classified as competitive.

Our normalized presidential vote measure was strongly related to the outcomes of House elections. In the 2000 House elections, 96% of safe Democratic districts were won by Democrats and 91% of safe Republican districts were won by Republicans. Of the districts classified as competitive, 53% were won by Republicans and 47% by Democrats.

The 2000–2002 redistricting cycle is often cited by critics of partisan redistricting as the best illustration of the dangers of gerrymandering because of the extensive use of sophisticated mapmaking technology in drawing district lines. However, between the 2000 and 2002 elections, the number of safe U.S. House districts only increased from 201 to 203 and the number of competitive districts only decreased from 123 to 116.

Figure 1 displays the number of safe and competitive House districts before and after redistricting in the past three redistricting cycles—those that occurred between 1980 and 1982, 1990 and 1992, and 2000 and 2002. Over these three redistricting cycles, the number of safe districts increased by an average of only eight while the number of competitive districts decreased by an average of only two. The biggest change occurred between 1990 and 1992 when the number of safe districts increased by 18. But that change was caused mainly by the creation of new majority minority districts, not by partisan gerrymandering. All of the remaining changes were miniscule.

There is little evidence that redistricting generally makes elections less competitive. In fact, two of the leading scholars in this field, Andrew Gelman and Gary King, have argued (1994) that redistricting, even when done for partisan purposes, has beneficial consequences for democracy. That is because party leaders drawing district lines face a fundamental tension between incumbent protection and maximizing their party's electoral potential. More often than not, the only way to shift marginal districts toward the majority party is to cut the safety margins of incumbents by moving reliable partisans out of their districts. For this reason, partisan redistricting often has the effect of reducing the safety of incumbents.

The assumption that shifting control of redistricting from partisan state legislatures to nonpartisan commissions would dramatically increase the number of competitive districts is not supported by the record of such commissions. In the 2000–2002 round of redistricting, eight states with a total of 75 House districts used nonpartisan commissions to redraw their districts or had their districts redrawn by the courts. In the 2002 elections, 9% of House contests in those states were decided by a margin of less than 10 percentage points compared with 8% in all other states. Of 65 incumbents who ran for reelection in states whose districts were redrawn by the courts or nonpartisan commissions, not one was defeated.

If redistricting isn't responsible for the low level of competition in House elections, what is? We found that two major trends have contributed to a decline in competition in recent years. First, House districts have become less competitive, but not because of redistricting. Most of the change has occurred between redistricting cycles. Between 1992 and 2000, for example, the number of safe districts increased from 156 to 201 while the number of competitive districts decreased from 157 to 123.

Why have House districts become less competitive? For the same reasons that states and counties have become less competitive—Americans are increasingly living in communities and neighborhoods whose residents share their values and they are increasingly voting for candidates who reflect those values. Gimpel and Schuknecht (2003) have demonstrated that a range of geographically linked factors such as immigration, migration, education, income, and religion are contributing to growing geographic divergence in party loyalties among states. Stonecash, Brewer, and Mariani (2003) and Oppenheimer (2005) have proposed similar explanations for growing partisan divergence among congressional districts. Evidence of this trend can also be seen at the county level where the number of counties dominated by one party and the proportion of voters living in such counties have increased dramatically over the past several decades (Bishop 2004; Abramowitz, Alexander, and Gunning 2005).

There is also strong evidence that growing ideological polarization at the elite level has made it easier for voters to choose a party identification on the basis of their ideological preferences (Abramowitz and Saunders 1998; Jacobson 2000). Therefore what many observers describe as “polarization” might more accurately be described as “sorting,” as voters bring their policy and partisan preferences into alignment (Levendusky 2004). As a result of this sorting, Southern and border states that once regularly elected conservative Democrats have been trending Republican (Black and Black 2004) while urban and suburban areas in the North that once regularly elected moderate and liberal Republicans have been trending Democratic (Paulson 2004). The result is that red states, counties, and districts are getting redder while blue states, counties, and districts are getting bluer.

The 1976 and 2004 presidential elections were equally close but there were far more swing states in 1976 than in 2004. In 1976, 20 states with 299 electoral votes were decided by a margin of less than five percentage points. In 2004, only 10 states with 106 electoral votes were decided by a margin of less than five percentage points.

The state of California illustrates these trends. California was once a swing state in national politics. In 1960, 1968, and 1976—three close national elections—Republican presidential candidates carried California by margins of one, three, and two percentage points, respectively. But California is no longer a swing state. In both 2000 and 2004, despite the closeness of the national vote, Democratic presidential candidates carried California by margins of 11 and 10 points, respectively.

As the state of California has become less competitive, so have its counties. Many of the state's urban areas, including the San Francisco Bay Area and Los Angeles County, have become much more Democratic. At the same time, the state's rural areas and small towns, like their counterparts in the rest of the nation, have been trending Republican.

Table 1 compares the competitiveness of California counties in the 1976 and 2004 presidential elections. These data show that there were far more competitive counties in 1976 than in 2004 and that a much larger percentage of the state's voters lived in competitive counties in 1976 than in 2004. In the 1976 presidential election, 46 of California's 58 counties were decided by a margin of less than 10 percentage points and those counties included 72% of the state's voters. But in 2004, only 13 counties were decided by a margin of less than 10 percentage points and those counties included only 21% of the state's voters.

On the other hand, there were far more landslide counties in 2004 than in 1976 and a much larger percentage of California voters lived in landslide counties in 2004 than in 1976. In the 1976 presidential election only two counties in California were decided by a margin of more than 20 percentage points and those counties included only 8% of the state's voters. But in 2004, 36 counties were decided by a margin of more than 20 percentage points and those counties included 64% of the state's voters.

Given the one-sided partisan make-up of so much of the state, it would be difficult for even a panel of retired judges to draw a large number of competitive state legislative and congressional districts in California. And if you think some of the current districts are misshapen monstrosities, try to imagine what a competitive district in the San Francisco Bay Area would look like.

The other major trend contributing to the low level of competition in recent congressional elections is the growing financial advantage enjoyed by incumbents. It now costs well over a million dollars to wage a competitive campaign for a U.S. House seat. Most incumbents can raise that kind of money easily but very few challengers can, even when they're running against an incumbent who appears to be vulnerable.

We measured the potential vulnerability of House incumbents based on the presidential vote in their districts—the most vulnerable incumbents were those in districts that were more supportive of the opposing party's presidential candidate than the nation—for example, Democratic incumbents in districts that voted for Bush in 2000 or Republican incumbents in districts that voted for Gore in 2000.

In the three elections between 1998 and 2002, there were 193 House contests involving such potentially vulnerable incumbents. However, only a small minority of the challengers in these contests were able to mount competitive campaigns: only 33 challengers spent more than a million dollars while 132 spent less than $500,000. Of the 33 challengers who spent more than a million dollars, 25 received at least 45% of the vote and nine were elected. Of the 132 challengers who spent less than $500,000, only nine received at least 45% of the vote and not one was elected.