The risk of negative campaigning from a decision theoretical perspective

Running a negative campaign is often seen as risky for a politician because it may backfire. This has led to a substantial body of empirical work which examines whether negative campaigning helps or hurts the candidates who engage in this tactic on average. However, the focus on the average may follow the colloquial definition of risk, as something bad that may happen, but not the decision-theoretical definition. According to this definition, risk is not the average outcome, but the variance of the (expected) outcome. I will argue that a decision-theoretical assessment of negative campaigning provides an interesting new perspective. According to decision theory, risk is not necessarily a bad thing. Its (un)desirability depends on the preferences of the decision maker and, more importantly in this case, on the situation. For example, a candidate who can expect to lose when nothing changes may benefit from ‘shaking things up’. Either she will lose by a landslide (which seems not much worse than losing by a thinner margin) or she may eke-out a win. In this paper I will analyse this problem theoretically, considering both optimal and boundedly rational decision making by candidates. This decision-theoretical perspective also raises new empirical questions. Firstly, is negative campaigning really riskier in a decision theoretical sense? Or put differently, are election results more variable when one or more candidates engage in negative campaigning? I will assess this question by reanalysing data on the effects of negative campaigning. Secondly, do candidates engage in negative campaigning when this would be to their benefit, or do they perhaps appear to suffer from known decision-making biases? This question is harder to answer using existing data. I will discuss what can be said about this based on the existing literature and what data would be required to answer these questions more in-depth.