An empirical study of the relation between stock price and EPS in panel data: Korea case

Abstract

This study investigates the relationship between stock price and earnings-per-share using Korean stock market data. The nonstationarity of the data has been managed. In particular, recently developed panel cointegration techniques are used, which are known to be more powerful than individual cointegration methods. The tests applied to the panel data as a whole showed the cointegration relationship between the stock prices and EPS (earnings-per-share), while tests for the individual stock prices could not detect the cointegration. Therefore, there is now some evidence in support of weak mean-reversion for the PER (Price-Earnings Ratio). However, only mixed evidence was obtained for the hypothesis that the coefficient of EPS to the stock prices is equal to one. In short, stock prices do seem to move with firm fundamentals, in the long-run and on average, but not necessarily at the same rate.

Acknowledgements

The authors thank Prof. Joonhwa Rho of Chungnam National University for providing the data.

Notes

¹ The working paper was originally written in 1992.

² The terms ‘mean-group’, or ‘group-mean’ are mixedly used with ‘between-group’ in some papers of panel regression. But the current study is going to use ‘between-group’ only.

³ Coakley et al . (2001) showed the small sample properties for these three estimators (that is, POLS, FE, BG) using Monte Carlo simulations.

⁴ The estimators only have nuisance parameter free normal distribution when the regressors are exogeneous, even under cointegration. The authors of the present study appreciate the referee for important comments.

⁵ Firm-level detailed data can be retrieved from http://klcaline.klca.or.kr for the member of the association.