A predicted P/E can be estimated from linear regression of historical P/Es on its fundamental variables, including expected growth and risk. While such empirical analysis can provide an analyst with useful insight, there are three main limitations:
- The predictive power of the estimated P/E regression for a different time period and/or sample of stocks is uncertain.
- The relationships between P/E and the fundamental variables examined may change over time.
- Multicollinearity is often a problem in these time series regressions, which makes it difficult to interpret individual regression coefficients.