A prominent feature of subnational governments in the US is that they face legal constraints that prevent them from raising taxes or spending levels by any amount they may want. Indiana, for instance, restricts the local budget levy to a fraction of statewide nonfarm personal income growth. There are ways to circumvent these limits, however, and a popular way is to allow for voter overrides. You will recognize this as something like “the city can increase its budget up to 5% without voter approval this year, but anything more than that will require a special voter referendum.”

A recent JPubE paper by Stephen Coate and Ross Milton investigates the optimal determination of the limits in the presence of voter overrides.

This paper studies optimal fiscal limits in the context of a simple political economy model. A politician chooses the level of taxation for a representative citizen but is biased in favor of higher taxes. A constitutional designer sets a tax limit before the citizen’s preferred level of taxation is fully known. The politician is allowed to override the limit with the citizen’s approval. The paper solves for the optimal limit and explains how it is impacted by the possibility of overrides. The paper also shows that the citizen’s welfare can be enhanced if the designer imposes a limit on the politician’s override proposals.

The authors take inspiration from the literature on the delegation problem, where an agent must choose a policy that impacts both principal and agent’s payoffs. The payoffs depend on a state of nature, which prior to the policy choice will only be observed by the agent. Here, the policy is the level of spending or taxation that impacts welfare of a representative citizen (the principal) and the politician (the agent) is assumed to be biased in favor of a larger spending. Coate and Milton extend this to determining the optimal fiscal limit, which sets up their main contribution: comparing to a extension where voters can override the limit. This is done by allowing the politician to propose a policy in excess of the limit which is subjected to the vote. If voters pass the proposal, this is adopted; if not they offer a second proposal which respects the limit. The primary tool here for the politician is agenda-setting power over the citizen. Here is their summary of the main results:

There are three main results. First, with overrides, the optimal limit is at least as stringent as without. […] Second, with overrides, the typical monotonic relationship in delegation models between the agent’s bias and the tightness of the limit does not necessarily arise. […] Third, the institutional arrangement consisting of a limit and an override provision can always be strictly dominated by an arrangement that also specifies an override limit. Limiting the proposal the politician can make at the override stage prevents him from fully exploiting his agenda-setting power. The availability of an override limit changes the calculus underlying the regular limit. Since a limit increases the benefit of overrides, the regular limit is often tighter, making overrides more likely.