Betting and Financial Markets Are Cointegrated on Election Night

Abstract

We present a model linking prices of political binary options to financial assets that applies in the very particular circumstance of the overnight session following an election. Contrary to most of the existing literature, the model is derived from economic first principles and applies in a general setting. We find that under suitable assumptions, election and financial markets will be cointegrated. Deviations from risk neutrality lead to the presence of a non-linear term relating to risk in the cointegrating relationship. The model is tested on three recent political events: The 2014 Scottish independence referendum, the 2016 Brexit referendum and the 2016 US presidential election. Strong support is found for two events (the Brexit referendum and the 2016 Trump win). We find that weak market efficiency broadly holds although there are violations of the order of minutes to tens of minutes. This is apparently caused by betting markets leading financial markets, a phenomena that is observed for all three events. This finding is consistent with the conclusion of the existing literature that prediction markets have superior forecasting ability to other methods. A realistic ex-ante trading strategy is presented for Brexit that profits from these inefficiencies. However, the success is not repeated for the 2016 presidential election. This is due to an apparent deviation from risk neutrality that is not observed on the night of Brexit.

Keywords: Elections, Election market, Political risk, High frequency data, Pricing of risk

JEL Classification: C51, D72, G12, G14, G15