Formal Identification of Sentiment Effects in Asset Markets

It is generally accepted that excessive exuberance or gloom in investor sentiment contributes to booms and crashes in asset prices but, because of its complex interaction with other aspects of the valuation process, these effects are not easy to identify with statistical confidence and this limits the scope for crafting an adequate and early policy response. To fill this gap, we develop an ex ante valuation approach that assigns different measures of sentiment to separate roles in the valuation equation. One measure of sentiment is assigned to capture risk aversion effects, while a broader-based investor sentiment index is assigned to capture changes in the perceived prospects for long-term earnings growth. The ratio of equity valuation when sentiment variables are included in the valuation exercise to the valuation when they are excluded is an obvious indicator of sentiment effects but this is difficult to assess statistically. We show that the ratio of average squared ‘implied' long-run earnings growth ‘with' and ‘without' sentiment produces a sentiment indicator that can be assessed with statistical significance. Out-of-sample testing using the Dow 30 index shows that sentiment effects can often be confidently identified as widespread, significant and large. We find that the growth ratio is more prescient as an early warning indicator of mis-valuations. Being able to draw attention to such statistically verifiable arbitrage opportunities in a timely fashion offers macro-prudential policy makers a more targeted policy response than making alterations to poorly focused policy instruments, such as interest rates