Batto: The adverse consequences of trust in government

Batto. 2005. The adverse consequences of trust in government. Presented at the Midwest.

Overview

Many in the literature have argued that trust is essential to democracy. It enables governments to get things done, especially when short-term costs are necessary to reach long-term goals. But Batto derives several hypotheses from the principal-agent assumptions that suggest that trust has a dark side. He does not claim that it is unnecessary, only that it is a two-edged sword. Specifically, those who trust the government more are more likely to support removing checks/balances (and other impediments to governmental action, like an opposition) and generally have authoritarian attitudes. Using survey data from five new democracies in East Asia, Batto finds evidence in favor of these hypotheses.

Model: Principal-Agent Delegation

A principal delegates a task to an agent in order to save her own resources. Yet she must bear certain costs if she wishes to prevent agency slack. She can use to general means to prevent agency slack, structual and managerial. Managerial means are performed during the course of the delegation. These including screening, monitoring, and sanctions. Structural means are the rules within which delegation occurs. This includes checks/balances (having multiple agents), constraining procedures (like an APA, something that requires the agent to do certain things), and defining the scope of delegation.

The principal must do these things because she does not trust the agent. But if she does, why do them? The delegatory relationship will be more beneficial to her if she can reap all the returns of delegation without bearing the costs of oversight and constraining her agent.

Thus, principals who trust their agent will favor removing constraints on the agent's behavior.

Data

East-Asian Barometer data in five new democracies.

Findings

See Table 3. Sure enough, trust is a significant predictor of five attitudes about democracy. In each of the five regressions, respondents who trust the government more are significantly less likely to favor preserving democracy.

Though these findings are statistically significant (i.e. non-zero), however, I'm not sure I'm persuaded that they are substantively important. Trust ranges from 7-28 (an additive index of seven four-point questions). The five dependent variables all range from -2 to 2. The coefficients on "trust" are very tiny. Still, they are suggestive, and they certainly go in the right direction for Batto's argument.