Over the past thirty years, the cumulative effects of direct democracy have served to decrease both levels of taxation and spending in the American states. Yet conservative budgetary policy measures passed during this time period were likely to occur in liberal states. Using data on over thirty separate ballot issues, the author offers a solution to this puzzle, demonstrating that distrusting government serves as a consistently robust predictor of conservative policy choice. The implication of this finding is that citizens often choose conservative policy outcomes because they believe that government is functionally incapable of giving them what they want.
direct democracy, political trust, Proposition 13, fiscal policy
This article asks a simple, relevant, and yet unanswered question: why has direct democracy produced conservative policy in the United States? The work of Matsusaka (1995; 2004) and others (e.g., Camobreco 1998; Besley and Case 2003) demonstrates that states that have adopted ballot initiatives tend to spend and tax less, on average, than states without direct democracy. While some have questioned whether this shift can be interpreted as policy responsiveness (Lascher, Hagen, and Rochlin 1996; Camobreco 1998), a growing body of literature argues that public policy produced by direct democracy is responsive to reasonable preferences by members of the electorate (Matsusaka 2004; Gerber 1996, 1999; Lupia and Matsusaka 2004). In other words, direct democracy produces more majoritarian democratic policy because it is more democratic/majoritarian.
With this argument as a backdrop, I engage another literature-that on the policy implications of political trust (Hetherington 2005)-which argues that public preferences about social programs have become more conservative not because of a conservative shift in ideology but because of declining levels of confidence in government to faithfully and efficiently execute these programs. A key component to this argument is that conservatism and distrust are not the same thing, but both play important roles in shaping policy preferences. In this article, I test whether political trust influences the way in which individuals choose in ballot initiative elections. While instrumental and political determinants remain important in these elections, confirming many of the findings from the extant literature, I demonstrate that political distrust plays a prominent role in predicting conservative policy choices in ballot measure elections, particularly those that deal directly with tax-andspend policy or have significant fiscal implications. In concluding, I suggest that distrust as a motivator of conservative voting behavior clarifies the mechanism of preference formation in ballot measure voting and poses a solution to the puzzle that liberal states seem to adopt conservative policy via the ballot initiative process.
2. Voter Choice in Direct Democratic Elections
Currently, the literature has identified several ways to predict systematic choice in ballot initiative elections. The choice problem, as defined by Magleby (1984), is that American politics research has demonstrated repeatedly that voters know very little about politics (Campbell et al. 1960; Converse 1964; Delli Carpini and Keeter 1996). Given that the average ballot initiative requires a graduate- level education to comprehend (Magleby 1984), how can voters possibly behave rationally under such conditions? The problem is exacerbated by the fact that initiatives are policy issues and do not contain explicit partisan cues on the ballot. Literature over the last twenty years, however, has demonstrated that (1) low-information voters behave as if they had higher levels of knowledge (Lupia 1994; Bowler and Donovan 1998), (2) voters are in some instances able to vote instrumentally or express selfinterest (Sears and Citrin 1982; Bowler and Donovan 1998), (3) when they are unsure of the meaning of initiatives, voters default to "no," thus preserving the status quo (Key and Crouch 1939; Bowler and Donovan 1998), and (4) despite the lack of explicit textual partisan cues on the ballot, explicit partisan cues are readily available in the information environment, and partisan voting behavior is the rule, not the exception, in ballot measure elections (Branton 2003; Kahn and Matsusaka 1997; Smith and Tolbert 2001). …