I further investigate changes in the determinants of earnings persistence during earnings
strings, and the results suggest the following: First, the positive association between past
earnings growth and current earnings growth decreases when firms approach breaks in their
earnings strings. Second, the coefficient estimates of persistence determinants do not exhibit
temporal shifting patterns in any systematic manner. Third, the partial adjustment model’s
explanatory power is at its lowest level before the break, which suggests that economic factors
have lower explanatory power to earnings growth in this period.