Democracy for Realists: Why Elections Do Not Produce Responsive Government. Christopher H. Achen & Larry M. Bartels. Princeton University Press. 2016.
The authors challenge the cherished American notion that general citizens obtain the government policies they want by democratic elections. They argue that election outcomes are essentially random and do not validate voters’ policy preferences. They begin by examining political science theories of how elections transmit the preferences of ordinary people to be enacted by government. They divide these into the three older theories of populist voting, leadership selection voting, and retrospective voting and the newer theory of group identity voting.
In the populist model (folk theory), voters know their policy preferences and have them implemented either by direct democracy or by representative democracy. In direct democracy, voters rule by choosing policies themselves via initiative and referendum procedures. In representative democracy, voters elect candidates whose policy preferences are most similar to theirs to represent them in assemblies that enact their preferences.
The leadership selection model dispenses with the notion that the voters themselves decide issues by electing candidates to carry out their will. Instead, democracy means only that voters have the opportunity of accepting or refusing the individuals who are to rule them. Thus voters don’t need to know policy so long as the leaders they elect make the best political decisions for them in order to compete for their votes.
The retrospective model regards voters as merely appraisers of past events, performance, and actions. Thus election outcomes hinge not on ideas, but on public approval or disapproval of actual past performance of incumbent political leaders. The authors compare this form of voting to driving by looking in the rear view mirror. They state that it works about as well in government as it would on the highway.
The authors present voluminous information to show that none of these models satisfactorily explains election results. For the populist and leadership selection models, they show that voters have insufficient understanding of their own perceived political preferences and those of their parties and candidates to vote on this basis. In addition, voters are commonly mistaken in highly partisan directions about easily measureable facts, such as crime rates and changes in deficits. Consequently, numerous studies show very little correlation between voters’ preferences and those of the parties and candidates they elect and the actual political outcomes that result.
For the retrospective model, the authors show that voters have insufficient understanding of whether times have been good or bad and whether government is responsible for perceived changes to vote on that basis. For instance, votes for incumbents have been shown to fall significantly after acts of nature, such as shark attacks and hundreds of droughts and floods, for which government clearly is not responsible. This is compounded by the incentive for political and ideological entrepreneurs to construct self-serving explanations and solutions for people’s hardships, which are then amplified by the mass media.
For retrospective voting, only the state of the economy is correlated with election results, but even that correlation is limited. When a president’s term is divided into sixteen quarters, only the two before the election matter. A president will be punished for an economy that does well for four years overall, but tumbles for the last two quarters, and he or she will be rewarded for the reverse. Politicians know this, so income growth usually increases during the last years of four year presidential terms, likely as the result of political manipulation of the economy. Thus, even for the economy, overall performance matters much less than very short term performance just before elections.
The authors argue that these older models of voting do not explain even the great partisan realignments, such as with the New Deal and with the Civil Rights Acts, as for changing policy preferences. For the New Deal, incumbent Republicans were punished for economic collapse in 1932, and incumbent Democrats were rewarded for recovery in 1936. However, contemporaneous elections in state governments and many foreign countries showed that incumbents of opposing parties were similarly punished and rewarded whether they were liberal or conservative. Hence, these election results were not directed at specific policies, but rather at incumbents. In addition, the recession of 1938 led to Democratic losses in congress and state governments. If the presidential election had been held that year, the great realignment might not have occurred.
This brings us to the group identity theory of voting. The authors conclude that the primary sources of voting behavior are partisan loyalties, social identities, group attachments, and myopic retrospection, not policy preferences, ideologies, or realistic assessment of circumstances. Party is the strongest identity, but others include race, ethnicity, religion, social class, and region. Identities are emotional attachments that transcend thinking and may trump facts and policy reasoning. Voters first choose, or commonly inherit the choice of, a party validating their political and social identities, and only then adapt their policy choices to fit those of their candidates and parties. Hence, in thinking about politics, it makes no sense to start from issue positions.
Consequently, the authors find that election outcomes are essentially random choices among the available parties—musical chairs. When the party balance is close, which it usually is in two party systems according to the “Law of the Pendulum,” outcomes turn on the voting choices of “pure independents” who do not even lean toward one party or the other. These “swing voters,” who are the least informed and the least engaged, are often swept along by the familiarity of an incumbent, the charisma of a fresh challenger, or a sense that it is “time for a change.” Hence, elections do not produce policy mandates, even when they are landslides.
Even well-informed and highly educated citizens are not exempt from these findings. They are likely to have more elaborate and internally consistent worldviews that just reflect better rehearsed rationalizations. Indeed, they are often more subject to partisan and confirmatory bias than less attentive voters. The authors emphasize that “this is not a book about the political misjudgements of people with modest educations. It is a book about the conceptual limitations of human beings—including the authors of this book and its readers.”
Given these findings about elections, what is good about democracy? First, elected governments are accepted as legitimate, which facilitates peaceful and orderly transfers of power. Second, in well-functioning democracies, parties that win office are inevitably defeated in later elections, sometimes due to random events, such as droughts, floods, or untimely economic slumps. This inevitable turnover is key to preventing any one group or coalition from becoming too entrenched in power and leading to the abuses of dictatorships or one party states. Third, electoral competition provides incentives for rulers to tolerate loyal opposition. Fourth, in well-functioning democracies, reelection-seeking politicians will strive to avoid being caught violating consensual ethical norms.
Given the limitations of voting, what are the concerns about democracy? After scrupulous efforts to present data in a nonpartisan manner throughout the book, the authors reveal what some would argue is partisan bias in only the last several pages. In their view, “more effective democracy would require a greater degree of economic and social equality.” Power imbalances are very large in favor of the wealthy, the educated, corporations, major media, ethnic majorities, and racial majorities. Organized, powerful, often minority policy demanders routinely get what they want at the expense of less powerful, unorganized majorities. Hence, the authors believe the folk theory of democracy should be abandoned in favor of the group identity theory to better understand the contributions and limitations of citizens, groups, and political parties in the search for political and social progress.
