Anchoring and adjustment biases and local government referenda language

Objectives: Local governments rely on referenda for many aspects of their finances. To take one example, local governments often use referenda as a way to gain permission from citizens to borrow for funding capital expenditures. Essentially, a referendum question amounts to an opportunity for the voter to perform a benefit-cost analysis. In relying on citizens and their votes to decide whether the jurisdiction should have the authority to issue debt, we are implicitly assuming that individuals will appropriately assess both the utility from the proposed uses of the funds and the cost of borrowing from their perspective. However, there are more than a few reasons to believe that this may not be the case.

Many researchers have pointed out heuristics that individuals use in assessing questions of the type reflected in referenda questions and the biases caused by using those heuristics. In this study we concentrate on the anchoring and adjustment heuristic and its effect on perceived cost to the voter of the project. We will assess voters’ ability to translate information from the referendum language into the cost that their household will bear from the proposed program.

Data/Method: We employ a survey methodology to assess how much a voter perceives that their household will have to pay in order to support a program of the size detailed in the referendum. Individuals will receive a mock referendum question. The language of the question will explain the cost of the project, varying the format of the cost across forms to include (1) the total cost of the project; (2) the amount of the cost per capita; and (3) the cost broken down into the amount that the owner of a median value house will have to pay in increased taxes per year. The respondent will then be asked to assess how much their household would have to pay per month in order to support the project if it is accepted. Given the predictions of the theoretical model, we expect that the expected costs will vary systematically according the form of the cost provided.  

Results: The results of our research will be relevant both from an academic research perspective but also to practicing public finance and election officials. The lessons that can be gleaned from our results should help design referenda language that is useful in determining citizens’ willingness to pay for public programs and services.