The uncertainty of firm's future payoffs is the dominant factor in explaining firm-level stock return volatility at the firm level. However, financial statement numbers (e.g. dividends, accounting earnings), only provide a limited measure of expected payoffs, as they do not reflect firms' fundamentals on a timely basis. We demonstrate theoretically and empirically that information about firms' fundamentals contained in analysts' forecasts, (which we label as "non-accounting information"), is

more »

... ected to influence future stock return volatility. When combined with Ohlson's (1995) linear information dynamics, the accounting version of the Campbell-Shiller model (Campbell and Shiller (1988) and Vuolteenaho (2002) ) implies that if current non-accounting information is more uncertain, then future stock returns are expected to be more volatile. Our empirical evidence supports the theoretical predictions, and the results are valid for measures of both systematic and idiosyncratic volatility. Additional analysis yields some evidence that both favourable and unfavourable news from non-accounting information increases future stock return volatility. Overall, our results indicate that information in analysts' forecasts beyond what is contained in the current financial statements (i.e. dividends, accounting earnings, etc) is associated with fundamentals, and its uncertainty drives the cross-sectional differences in stock return volatility. 2 5 Using data from the merged Compustat, CRSP and I/B/E/S database for the period from 1981 to 2003, we show that the volatility of non-accounting information is positively associated with the variance of future stock returns, even after accounting for other known determinants of stock return volatility. We further examine whether the uncertainty of non-accounting information (mainly) drives the cross-sectional differences in systematic or idiosyncratic risk. As there is no exact theory on these links, it is treated as a purely empirical issue. Following Malkiel and Xu (2003), the CAPM and the Fama-French three factor model (Fama and French 1993, 1996) are used as benchmarks to decompose total volatility its systematic and idiosyncratic components. We find that the significant relation between the uncertainty of non-accounting information and stock return volatility occurs for both the systematic and idiosyncratic components of volatility. Our results are robust to several different measures of stock return volatility, different sample periods, different stock exchanges in the U.S. (Schwert 2002) and different industry categories (Roll 1992). Using different econometric frameworks including pooled time-series and cross-sectional regression, Fama-MacBeth regression and fixed effect regression, we still find a significant relation between the uncertainty of non-accounting information and stock return volatility. The results continue to hold even after accounting for new listing bias in the 1980s (Fama and French 2004; Wei and Zhang 2006), technology bubbles (Chan, Lakonishok and Sougiannis 2001; Schwert 2002), deteriorating corporate earnings (Collins, Pincus and Xie 1999; Givoly and Hayn 2000) and specific properties of analysts' forecasts such as analyst coverage, forecast dispersion and forecast bias (Ajinkya and Gift 1985; Daley, Senkow and Vigeland 1988; Ackert and Athanassakos 1997; Alford and Berger 1999). We finally examine the effect of non-accounting information new (the level of non-accounting information) on stock return volatility. Non-accounting information news can be thought of as an aggregate indicator of all value-relevant events that have yet to have an impact on the financial statements. Previous studies provide consistent evidence that stock return volatility increases after the release of firm-specific news