Changes in cashflows and changes in "discount factors":
http://www.economics.harvard.edu/faculty/stein/papers/aer-p&p-micro-macro.pdf: Samuelson’s dictum has been made concrete in empirical work by John Campbell (1991) and Tuomo Vuolteenaho (2002). Campbell shows that innovations to aggregate stock returns are largely transitory: less than half of the variance in returns is due to changes in expected cashflows, leading to predictable reversals in future returns. If one adopts a behavioral perspective and thinks of return predictability as being indicative of investor sentiment, this is consistent with Samuelson’s idea that the aggregate market is quite inefficient, with a large fraction of its price variation being due to non-fundamental factors.
In contrast, Vuolteenaho shows that innovations to idiosyncratic firm-level returns are largely permanent: most of the variation in returns is driven by cashflow news, with a smaller component due to time-varying expected returns. This suggests that firm-level relative prices are reasonably efficient, with a smaller role for sentiment.