-
Disagreement and Stock Prices in the JASDAQ ―An Empirical Investigation Using Market Survey Data
Abstract
This article empirically examines “disagreement” models using JASDAQ market data by exploiting institutional investors’ forecasts of future stock prices. We use the standard deviations of the one-month ahead forecasts of stock prices in the QSS Equity Survey as the measure of disagreement in the market. The results indicate that an increase in disagreement is associated with an increase in contemporaneous stock returns and lower average expected returns. In terms of the latter, while the survey data provides an average assessment of marketwide expectations, when disagreement is high the current market price tends to reflect the opinions of more optimistic market participants. These results contrast with comparable findings using TOPIX data (representing larger firms on the Tokyo Stock Exchange’s first section) that contradict the predictions of disagreement models. One reason posited is that firms on the JASDAQ market are much smaller and the number of market participants more limited. Accordingly, institutional “limits of arbitrage”, such as short-sale and liquidity constraints, are more binding and their influence on stock prices is thereby greater.
Introduction
This paper uses Japanese data to test the empirical implications of recent developments in behavioral finance, which we refer to as “disagreement” models. These models, starting with seminal contributions by Miller (1977), and Harrison and Kreps (1978), and surveyed in Hong and Stein (2007), display two key elements. First, they assume some disagreement or difference in opinion among investors over the valuation of assets. Second, some institutional factors, typically short-sales constraints, prevent the arbitrage mechanism from working completely. As a result, the market price tends to reflect the expectations of optimistic investors. Accordingly, the informational role of asset prices is partially confined and the market price can be persistently higher than the fundamentals.