Earnings persistence coefficient
factor.formula
Earnings persistence model (time series regression):
in:
- :
The annual earnings measure for company j in year t, usually using a standardized value (e.g., divided by total assets or total equity) of a measure such as earnings per share (EPS) or net income. Using a standardized value eliminates the effect of differences in company size.
- :
The regression intercept term for company j represents the expected profit in year t when the profit in year t-1 is 0.
- :
The earnings persistence coefficient of company j indicates the degree of influence of earnings in year t-1 on earnings in year t, i.e. the autocorrelation of earnings. This coefficient is a key measure of earnings persistence.
- :
The regression residual term for company j in year t represents the earnings volatility that the model cannot explain and is assumed to be normally distributed with mean 0.
factor.explanation
The value range of the earnings persistence coefficient ($\phi_{1,j}$) is usually between -1 and 1. The closer the value of $\phi_{1,j}$ is to 1, the stronger the earnings persistence is, that is, the current earnings have a strong predictive ability for future earnings, the earnings quality is high, and the company's profitability is relatively stable. The closer the value of $\phi_{1,j}$ is to 0, the weaker the earnings persistence is, the current earnings have limited predictive ability for future earnings, the earnings may be affected by one-time or temporary factors, and the profitability volatility is large. $\phi_{1,j}$ may also be negative, indicating that the current earnings are negatively correlated with the previous earnings, which is relatively rare, but may mean that the earnings are affected by special events or accounting operations. In practice, it is necessary to combine industry characteristics and company fundamentals analysis to more accurately assess the persistence of earnings.