Bubbles, Blind-Spots and Brexit

Abstract

Abstract:In this paper we develop a well-established financial model to investigate whether bubbles were present in opinion polls and betting markets prior to the UK's vote on EU membership on 23 June 2016. The importance of our contribution is threefold. Firstly, our continuous-time model allows for irregularly spaced time series-a common feature of polling data. Secondly, we build on qualitative comparisons that are often made between market cycles and voting patterns. Thirdly, our approach is theoretically elegant. Thus, where bubbles are found we suggest a suitable adjustment. We find evidence of bubbles in polling data. This suggests they systematically over-estimate the proportion voting for remain. In contrast, bookmakers' odds appear to show none of this bubble-like over-confidence. However, implied probabilities from bookmakers' odds appear remarkably unresponsive to polling data that nonetheless indicates a close-fought vote.

The bilateral trade effects of announcement shocks: Brexit as a natural field experiment

“…15 Saia (2017)'s study also produces quite a narrow confidence interval. 16 These are derived from data cited in Fry and Brint (2017), although some approximation had to be used in converting daily to monthly data. 17 Using a donor pool selected by revealed comparative advantage (RCA), the estimated percentage difference between the actual and the counterfactual for the extra-EU export flows is about À15% (standard errors 1.427).…”

An options-pricing approach to election prediction

“…As far as finance is concerned the link between vote share and subjective probabilities obtained via a betting argument is important philosophically (Taleb, 2017;2018). Further, recent concerns over Socially Desirable Response Bias in opinion polls (Brownback and Novotny, 2018;Coppock, 2017;Kimball, 2019) may be analogous to bubbles and mis-pricing in financial markets (Fry and Brint, 2017).…”

“…Allied to the above the original motivation behind this work was to try and model bias in opinion polls using a jump-process model of the form of Johansen et al (2000) or Fry (2012). For an early attempt at work along these lines see Fry and Brint (2017). However, as helpfully pointed out by a reviewer, in this case the time of the jump, the known election date, is deterministic meaning that arbitrage-free pricing approaches do not apply.…”

“…The interplay between finance and politics is interesting (Prechter, 1999;Wang et al, 2009). Of particular interest is the analogy between opinion polling and behavioural finance (Fry and Brint, 2017;Wu et al, 2017). This is given added prominence by recent concerns raised over behavioural effects in polling and apparent forecasting inaccuracies (Hopkins, 2009;Coppock, 2017;Brownback and Novotny, 2018;Kimball, 2019).…”

Prediction markets: Theory, evidence and applications

“…A separate focus of attention has been the design of the market mechanisms used in prediction markets (Chen, 2011;Klingert and Meyer, 2018;Spann and Skiera, 2003). There has also been a developing interest in recent years in the idea of combining forecasts from a range of different forecasting methodologies (Graefe et al, 2013;Pathak et al, 2015;Rothschild, 2015), and some have questioned aspects of the performance of markets relative to other forecasting methodologies (Atanasov et al, 2017;Erikson and Wlezien, 2008;Fry and Brint, 2017). Others have addressed how these markets might be influenced by attempts at market manipulation (Buckley and O'Brien, 2017;Hanson and Oprea, 2009).…”