Responses in divergence of opinion to earnings announcements: Evidence from American depository receipts

The basic asset pricing model is developed under the presumption of homogeneous beliefs; However, Miller (1977) puts forward heterogeneous beliefs along with other early studies. His appealing theoretical verbal model suggests that rather than reflecting the average expectation of all market participants, stock prices reflect the valuation set by the most optimistic investors due to the differences of opinion and short-sales constraints. He predicts the convergence of divergent opinions over time "primarily because the passage of time often resolves certain uncertainties about the future of a company" (Miller 1977, p. 1155). Among the prior empirical studies which provide mixed evidence of Miller hypothesis, the most recent study by Berkman et al. (2009) endeavors to document the negative relationship between excess returns and differences of opinion in a three-day window around earnings announcements when there are binding short-sales constraints.;Yet to date there has been little empirical research investigating how divergent opinions affect asset prices of foreign stocks. This dissertation takes a step in this direction. Using the American Depositary Receipts sample, we adopt an event study methodology and use multivariate regressions to examine the host-market response in divergent opinions to the earnings announcements. Overlooked by prior literature, asymmetric reactions to good and bad earnings surprises are allowed in the study. Country-level factors from home market are introduced with firms-characteristic factors to capture the cross-sectional excess returns in the presence of belief dispersion and host-market short-sales constraints.;Our sample contains 553 ADRs with 13378 firm-quarter observations from 52 countries. Results from regression analysis show that consistent with Miller hypothesis, quarterly earnings announcements indeed help reduce opinion divergence in ADRs by documenting the negative relation between differences of opinions and excess quarterly earnings announcement returns. Our findings are robust when controlling the financial leverage, illiquidity, analyst forecasts, post-earnings announcement drift and price momentum. Moreover, we find investors do process information asymmetrically based on good and bad earnings shocks when use TURN as the DIVOPN proxy. We observe the price divergence when good earnings are released. Non-news group and bad news group experience the same price convergence. However, we do not find the positive relation between host-market short-sales constraints and excess earnings announcement period returns, even conditioning on home-market short-sales restriction. Last, we include the enforcement of insider trading law, legal origin, investor protection, rating on accounting standard and short-selling feasibility from home market into our baseline model. These home-market country-level factors do not account for our findings.