When the Personal Vote Is Not Enough: The Failure of Charter Reform in Columbia, South Carolina

251-26

Perhaps the most important institutional decision local political actors make is how to structure their governments.While virtually all mayors and city councilors inherit a form of government, that form can be changed through a variety of mechanisms-the most common of which is to put the choice up for popular vote through a referendum.Thus, the decision to reform relies on the political will of elected officials to place the measure on the ballot and campaign for it as well as the choice of the people to turn out and select the reform option.
While there are a variety of different institutional design options that elected officials can tackle, perhaps the most fundamental is the choice between council-manager and mayor-council systems.Although most cities adopt a structure that takes on characteristics of both types (Fredrickson & Johnson, 2001;Frederickson, Johnson, & Wood, 2004a, 2004b), the choice of form of government remains a fundamental decision that can affect service delivery and the quality of governance more generally.According to Svara and Watson (2010a): Form is the constitutional and legal basis for assigning authority and functions to officials in government and creates its overall framework.Form shapes the nature of official roles and channels interactions into likely patterns of relationships, that is, who talks to whom, who gives instructions to whom, and how are those instructions interpreted and acted on by the recipient.(p.4) Because of the highly politicized and salient nature of such a decision, it should be no surprise that cities often choose to change their government structure.Indeed, during the early twentieth century, many local governments abandoned the traditional mayor-council system for the reform-minded council-manager government.In the latter half of the century and into the early 2000s, many cities returned to the mayor-council government, while some hold-outs from the old system adopted the council-manager form (Svara & Watson, 2010a).
As we discuss in more detail in the pages that follow, a number of scholars have undertaken macro-level examinations of why cities choose various forms of government (e.g., Choi, Feiock, & Bae, 2013), but few studies have looked at the dynamics of choice within individual cities.In this paper, we use the case study of Columbia, South Carolina-a relatively diverse, mediumsized city-to gain a better understanding of why sometimes even popular mayors cannot convince a majority of the voting public to abandon the manager-council form of government and cede more power to the mayor.

Local Government Structure
Although there are a number of smaller variations, the five most basic forms of city structure are the mayor-council, council-manager, commission, town meeting, and representative town meeting.Even listing these five potential structures overstates the choice most governments face, however, as over 90% of municipal governments with populations over 2,500 employ one of two forms of government: the mayor-council or the council-manager form (Svara & Watson, 2010a).Table 1 lists the basic characteristics of these two forms of government.
The mayor-council system is familiar to most Americans.It is similar to the presidential system employed at the federal level.In brief, the mayor acts much like a president.S/he is separately elected and presides over the executive functions of the state.The council, under this system, resembles Congress at the national level i.e., acting as the legislative branch of government.Much like the federal government, the mayor-council system "incorporates separation of powers and formal checks and balances between the executive and the legislature" (Svara & Watson, 2010a, p. 1).
Conversely, in the council-manager system, the power to make policy rests with the council, usually headed by the council president or council chair.Although the mayor is on this council, a professional city manager is hired to make most of the day-to-day decisions such as personnel, budget, and implementation of policies (Ulbrich, London, & Lucka, 2004).The councilmanager form is often compared with a parliamentary system "based on unified authority assigned to the legislative body that controls the executive branch" (Svara & Watson, 2010a, p. 1).
A number of studies have cataloged both the causes and the effects of city structure.For example, city size has been listed as an important determinant of whether a city has a councilmanager or mayor-council form of government.The largest cities in the US are much more  Fredrickson, Johnson, & Wood (2006).
likely to have a mayor-council system.Even when large cities do adopt the council-manager form of government, they are more likely to abandon it in favor of the more traditional mayorcouncil form (Protasel, 1988).Summing up this trend, Svara and Watson (2010a, p. 11) note "The mayor-council form seems to match the large city with extensive conflict, and the councilmanager form matches the small, harmonious community." Political and economic factors may also explain reform adoption.For example, cities located in counties that expressed greater support for Republican candidates for president, and cities with larger minority populations were more likely to adopt and maintain the council-manager form of government prior to 1965.There is also evidence suggesting that economic hardship may spur cities toward charter reform (Choi, Feiock, & Bae 2013).
Studies on the effects of city structure date back, at least, to Lineberry and Edmund (1967) who concluded that council-manager governments spend less money and collect fewer taxes than unreformed cities (although recent studies have questioned the durability of this finding [e.g., Craw, 2008]).Summarizing these articles, Carr (2015) notes that there are roughly 10 domains where council-manager governments may outperform mayor-council governments.Although findings lauding the council-manager form are far from unanimous, there is at least some evidence that senior officials in council-manager governments spend more time developing policy solutions, managing the municipality, and finding and applying innovative policies than do senior officials in other forms of government.
Figure 1 presents the percent of local governments that employ mayor-council or councilmanager governments.As shown, there was a decline in the percentage of governments with the mayor-council form from 1984 to 2012.In contrast, the percentage of governments employing the council-manager form has increased.Though we do not have national data from every year, the trends displayed in Figure 1 indicate that the council-manager form of government became the most popular form of government sometime in the late 1990s.
Despite this general trajectory, there is still considerable variation between and within a state.Many large cities, for example, have eschewed the trend toward reform structures (Svara & Watson, 2010b).State culture clearly has an important influence as well.The council-manager government has not taken hold in Wisconsin (Simmons, 2001), whereas in the state of North Carolina every city with over 25,000 people employs the council-manager form of government (Upshaw, 2014).Clearly the choice of local government form is an important one, and one that can change over time.Below we examine what we know about how often and why cities change their local governments.

Efforts to Change Local Government Form
The decision to change a local government form is more common than many might thinkparticularly among larger cities.According to Svara and Watson (2010a), just 3% of cities between 5,000 and 100,000 held referenda about changing government forms between 1990 and 2010.However, they found that 6% of cities between 100,000 and 249,000 and 17% of cities with populations over 250,000 held referenda about charter reform over the same time period.All told, almost one in every five larger cities considered a change in city structure in just two decades.While most of these changes were from mayor-council to council-manager, there was a movement back to mayor-council government in many American cities during the latter half of the twentieth century and into the early 2000s.Of the 191 largest cities in the country, 20 changed from mayor-council to council-manager from 1965-2005, whereas 26 abandoned the reform-style council-manager government and opted instead for the more traditional mayorcouncil government (Choi et al., 2013, p. 732).In summary, charter reform is fairly common, and scholars have successfully identified the types of cities that are expected to adopt and eschew charter reform.
Of the cities that have changed their form of government, we know the most about the 15 profiled in Svara and Watson's (2010b) edited book, More Than Mayor or Manager.These indepth case studies provide important lessons about when cities choose to change or retain their form of government.Nine of the 15 cities profiled in the book had a council-manager government and were attempting to change to a mayor-council form.In the cases where change was successful (St.Petersburg, Florida; Spokane, Washington; Richmond, Virginia; San Diego, California; and Oakland, California), most had experienced economic distress and/or racial strife.In three of these five cities, the manager had undergone increased scrutiny of personnel practices.None of the four cities that were unsuccessful in reform experienced such problems.
Further, in a few cases (most notably, Oakland), there was a highly popular and successful political entrepreneur advocating for change (the mayor himself).
The successful reform movement in Oakland offers a number of important lessons for understanding change in other cities.As noted above, the mayor (Jerry Brown) was extremely popular and campaigned extensively to increase mayoral power.His argument was that without charter reform he was unable to enact the policies that he had campaigned on (Mullin, 2010).After charter reform was approved by 70% of the voters, Brown's popularity soon became the prevailing explanation for victory.As the San Francisco Chronicle proclaimed, "With his strongmayor initiative (Jerry) Brown succeeded where three mayors had tailed before him.His popularity cut through the policy-wonk haze that made previous strong-mayor proposals intriguing to Oakland voters but slightly frightening" (DelVecchio & Holtz, 1998;cited in Mullin, 2010, p. 125).The notion that the mayor has coattails that can cause citizens to cede more power to him through charter reform is known as the "personal vote hypothesis" (Mullin, 2010).
A second explanation for the success of charter reform in Oakland centered on the support from other elites.As is common in low-information referenda campaigns (Paul & Brown, 2006), elite endorsements in Oakland may have helped give voters cues that the change to a strong-mayor system was a good policy change.Some scholars (e.g., Cropf & Swanstrom, 2005;Mullin, 2010) have included precinct (or ward) level analyses of the vote on charter reform.This approach has allowed them to evaluate the often-proffered explanations about what types of people supported the reform effort.While precinct-level analyses cannot determine whether a person with a set of characteristics was likely to support reform, they can allow us to make inferences about relatively small geographic areas that tend to be home to certain types of people.Precinct-level analysis can also be helpful for practitioners who may want to advocate for either side, providing clues about where and how to target their campaigns.In particular, Mullin's precinct level analysis in Oakland led her to conclude that Mayor Brown's coattails were not as strong as many had presumed.Support for the mayor-council government was only weakly associated with Brown's vote share in the previous election.While existing case studies and analyses have taught us a great deal about why cities adopt or eschew charter reform, battles over charter reform continue to wage across the country.One particularly intriguing case took place in Columbia, South Carolina in 2013.

The Case of Columbia
On the heels of a resounding re-election victory in November 2013, Columbia, South Carolina Mayor Steve Benjamin turned his attention not to crime, downtown redevelopment, or other common local policy issues but rather to the most basic question of politics: Who has the power?Columbia, like the majority of US cities operated under the council-manager form of government (with an elected mayor).Citing the difficulties of running a city on a part-time basis, Benjamin proposed that the city should opt for the mayor-council system.According to Benjamin, "It's a full-time job, and anybody who commits should be willing to do it as a full-time job" (Cope & Hinshaw, 2013).
Initially, Benjamin left it to the City Council to make this change without involving Columbia voters directly.The council, however, rejected this proposal.As a result, on August 2013, Benjamin released a statement that he intended to hold a single-issue referendum changing the city's government from a mayor-council system to a strong-mayor system (Bell, 2013).
Advocating for charter reform was Benjamin's first substantive act after re-election, but he only had two months to mount a campaign in support of this change.Benjamin had a number of reasons to be optimistic.By any measure of electoral success, Benjamin was a popular mayor.He won his initial term in a runoff election by a 56-44 margin over sitting Councilman Kirman Finlay.He expanded this margin in 2013 when he beat businessman Moe Baddourah by an astounding thirty percentage points.In brief, the personal vote was on his side.
In addition to his electoral success, the council had recognized that the mayor's job had become more complex over the years and increased his salary from $17,500 when he won his first term to $75,000-an increase that took hold when he took the oath for his second term (Wallace, 2013).A further increase to $160,000, as specified in the charter change proposal, might therefore not look like as drastic of a jump.Faced with the decision of how to lobby for this change, as allies he reached out not just to Columbia-based politicians but also to prominent politicians who served at the state level (such as Governor Nikki Haley) or who were recognized as national leaders in local government administration (such as longtime Charleston Mayor Joe Riley).
Benjamin lined up high-profile support for changing the government structure in Columbia.Supporters included Governor Nikki Haley, former Governor Jim Hodges, former state Attorney General (and current Lieutenant Governor) Henry McMaster, and longtime Charleston Mayor Joe Riley (Leblanc & Monk, 2013).The face of the campaign against the referendum was former Richland County Council member Kit Smith.Other people who came out against the referendum were City Council members Moe Baddourah (Benjamin's opponent in the 2013 mayoral contest) and Tameika Isaac, and former president of the state bar association Leevy Johnson.
In order for this issue to be put up for a vote, Benjamin needed 15% of the voters to sign a petition.He had to go this route once the reform proposal was voted down by the City Council by a 4-3 margin.Another important factor in this initiative was that the Benjamin team had only 120 days after the petition was turned in to put the issue up for a vote.This was a narrow window to get all of the work done before it appeared on the ballot.
The opposition used a number of strategies to rally support.Johnson, a respected African-American leader campaigned against the change, telling voters "if it's not broken, don't break it" (LeBlanc & Monk, 2013).Opposition forces also enlisted Howard Duvall, former director of the Municipal Association of South Carolina, who told The State newspaper that a key strategy was to separate the decision about the form of government from Benjamin's popularity (LeBlanc & Monk, 2013).Duvall said that "We were never anti-Benjamin…We were just pro-professional manager" (LeBlanc & Monk, 2013).
The referendum took place in December of 2013.On the day of the vote, election returns indicated that about 11,808 people showed up to vote on the referendum; of this, 57% of the people voted "no" to the referendum and wanted to keep the current council-manager system.As an indicator of the low turnout, 11,757 people had signed a petition stating they wanted to vote on the current government system, about the same number who showed up to vote on the referendum (Leblanc, 2013).Only 43% of the total voters voted "yes" to change to a strongmayor government system.This 14% loss for Benjamin's proposal surprised many observers and was considered a substantial defeat for the popular mayor.
The case of Columbia was purposively selected; and, based on Patton's (1990) sampling strategy it represents an extreme case.The notable failure of Mayor Benjamin is worthy of investigation and has the potential to be instructive for scholars interested in the influence of the personal vote.
There are a number of other reasons the reform measure in Columbia is worthy of investigation.First, the Columbia decision took place after extant works on city-level decisions (Choi et al., 2013;Svara & Watson, 2010b).Further, there has been renewed scholarly interest in reform decisions, but most of this work has focused on the largest cities in the country (Choi et al., 2013;Svara & Watson, 2010b), leaving cities like Columbia (population 134,385) unexamined.Considering that city size was a clear influence on the choice of government form, this is an important oversight.This is also a politically consequential oversight because medium or midsized cities were responsible for more growth in the 1990s than large cities (Vey & Forman, 2000) and patterns in the early 2000s reflect similar trends (Cohen, 2015).
Apart from its status as a midsized city, Columbia also marks a particularly interesting case study because Mayor Benjamin was a popular and vocal (perhaps the most vocal) advocate of the change, offering us a clear test of the impact of the personal vote hypothesis.The fact that the vote took place in a special election, rather than alongside a regular cycle election, means that we can infer that turnout was not related to other races but is rather a direct reflection of interest in this referendum.

Data and Methods
To take a closer look at the 2013 referendum, we rely on precinct-level election returns from the 2012 presidential election, the 2013 mayor's race, and the 2013 referendum (South Carolina Board of Elections, 2016).We present several figures showing bivariate relationships, and we also compute an ordinary least-squares (OLS) regression model predicting precinct level support for reform (model diagnostics indicate that OLS is appropriate in this case).The dependent variable in our regression model is the percent of voters supporting a change in the form of government in each precinct.
Although there are a number of potentially important explanatory variables, we follow the advice of methodologist Christopher Achen (2005), opting for a more parsimonious model with just three independent variables in our regression.To test the personal vote hypothesis, we created an independent variable measuring precinct level support for Benjamin during his reelection bid.We evaluate the elite influence hypothesis with a variable indicating whether the precinct was represented by a city councilmember who supported charter reform.We concede that this measure does not capture the influence of statewide elected officials, or mayors from other municipalities, but it does provide a measure of support for charter reform by the city council leader representing the individuals and, in particular, voting precincts.Further, because city councilors are likely the political actors that citizens have the most contact with, it stands to reason that many citizens would use the opinions of their councilmembers as cues for how to vote on local charter reform.Finally, we investigate the influence of voter mobilization with an independent variable measuring precinct-level voter turnout. 1 Descriptive statistics are presented in the appendix.

Results
To begin, we investigated three bivariate relationships.Figure 2 shows voter turnout in each precinct and support for reform.As can be seen, there is a negative relationship between turnout and support for change in the form of government (r=-0.70).Precincts with the highest turnout, with over 30% of voters casting ballots, generally had the lowest levels of support for change.Next, we wanted to see if precincts represented by council members who supported change in the form of government experienced higher percentages of support for reform among the voting public.Here, we find that nearly 60% of residents who lived in precincts where the council member supported changing the form of government voted for reform.Conversely, only about 45% of voters voted "yes" in precincts where the city council candidate did not support change in the form of government.
Lastly, we estimated an OLS regression model to learn more about the types of precincts that supported change in the form of government.Again, the dependent variable for the model was percent voting "yes."The independent variables were percent of voter turnout in the special election, percent in support of Benjamin in 2013, and a dummy variable for whether the council member representing the precinct supported change in the form of government.The results of our regression model appear in Table 2.
As shown in Table 2, precincts with higher voter turnout were less likely to support change in the form of government.As we expected, precincts that supported Benjamin at higher rates were more likely to support change.This finding indicates support for the personal vote hypothesis.The variable measuring whether the precinct was represented by a council member who supported change in the form of government did not achieve statistical significance.As a result, we did not find support for the elite influence hypothesis.In addition to determining the statistical significance, we computed predicted quantities for the statistically significant coefficients i.e., the turnout variable and the support of Benjamin variable.The predicted percentage "yes" vote in precincts with the highest turnout was just 18%, while the predicted "yes" vote in precincts with the lowest turnout was 31%.Regarding the effects of support for Benjamin, the predicted quantities indicate that precincts with the highest levels of support for Benjamin voted "yes" at 42%, while precincts with the lowest support for Benjamin voted "yes" at 49%.
It is too simple to conclude that Mayor Benjamin's coattails weren't strong enough.Indeed, charter reform did fare better in precincts where Benjamin performed better.Interestingly, though, it appears that while Benjamin might have been successful in getting his strongest supporters to give the mayor more power, he was unsuccessful in turning out the vote among those who were not so clearly in his camp.Indeed, precincts with higher levels of turnout showed lower levels of support for the charter reform effort.Simply put: The anti-reform advocates successfully targeted precincts where they were likely to win.This effort was unrelated to elite influence in the form of council support.

Discussion
This study investigates an unsuccessful attempt at changing the form of government.A popular mayor was unable to convince voters to change the form of government from council-manager to strong-mayor.This defeat occurred just two months after winning a convincing re-election.
We highlight a number of factors that contributed to the referendum's defeat.With just two months to build support, there was not much time for Benjamin and his supporters to make the case for the mayor-council form of government.In an August 11, 2015 phone conversation with Mayor Benjamin (Benjamin, 2015), he spoke about many of the lessons he had learned through this experience.First, he wished he had telegraphed his intentions sooner.He believes that this would have allowed him to have informed the public earlier on how important the issue was, allowing more people to show up and vote on the referendum.It was also interesting to note that Mayor Benjamin believes, in part, that his team had been "victim of their own success."Because they had gotten so much accomplished under the current form of government, where the mayor did not have much power, they were still able to accomplish many goals that they set for themselves.
In addition, the lack of support from the Columbia City Council also made passage of the referendum more difficult.This 4-3 vote had important consequences.First, it meant that reform supporters had to collect signatures to get the issue on the ballot.Second, and perhaps more importantly, the referendum vote did not take place at the same time of Mayor Benjamin's re-election.We suspect the results would have been much closer, perhaps even passing, if the referendum vote occurred in November instead of December.
Our investigation of precinct-level data showed some important trends as well.First, there was a negative correlation between voter turnout and support for change in the form of government.
The precincts where the turnout was the highest were also the precincts where support for change was the weakest.This provides some indication that the "no" vote was more effective in mobilizing voters than Benjamin's "yes" campaign.We also found support for the personal vote hypothesis.In our regression model, we discovered a positive correlation between support for Benjamin in 2013 and support for change in the form of government.It certainly makes sense that those who supported the mayor wanted to give the mayor more power.We did not find support for the elite influence hypothesis, as we measured it, suggesting that citizens do not adopt the opinions of their city councilors on charter reform referenda.These results should provide fodder for future studies examining charter reform in America's cities.

Conclusion
So what does the future hold for Columbia, South Carolina?In an interview, Mayor Benjamin left open the possibility of revisiting this issue in the future, noting that, "The big battles are worth fighting."The potential for future change could take two forms.First, according to South Carolina law, the referendum can be introduced again in four years.More likely, however, is the possibility that Columbia may not make a wholesale change but instead may slowly adopt characteristics of the mayor-council government, thus increasing power incrementally rather than in a punctuated fashion.This type of change may be more politically palatable and would be consistent with how change has taken place in many other cities (Fredrickson & Johnson, 2001).
In the end, our work suggests that efforts to change the form of government are both an active movement in many local governments and a reminder that these efforts are not always successful.Indeed, if there was any case where we might expect the people to side with a strongmayor form of government, this would be it.Benjamin had a landslide victory just a few months earlier and was able to garner support from many of the most prominent figures in the state.Although Columbia is not a large city by most definitions, it is the largest city in South Carolina; and, thus, it may have been viewed by many as a prime candidate for a strong-mayor system.The fact that the voters rejected this reform is certainly viewed as a negative by the supporters of the reform initiative, but, through a different lens it is a positive view of the ability of voters to express their opinions and separate policy from people.

Notes
1. We should also note that we collected basic demographic information at the precinct level (South Carolina Voter Registration Demographics, 2016).We were particularly interested in the effects of a precinct's racial makeup on support for reform.However, the nonwhite percent was highly correlated with support for Benjamin (r=0.87),so it is difficult, if not impossible, to tease out the relative effects of these two potential factors.

Figure 1 .
Figure 1.The Two Major Types of Local Government Form

Figure 2 .
Figure 2. Voter Turnout and Support for Change in the Form of Government

Figure 3 .
Figure 3. Support for Benjamin in 2013 and Support for Change in Form of Government

Table 1 .
Differences in Classic Structure of Mayor-Council and Council-Manager Forms of Government

Table 2 .
Predicting Support for Change in the Form of Government Entries in the left column are OLS regression coefficients.Numbers in parentheses are standard errors.Entries in the right column are minimum and maximum predicted quantity ranges.