| The Q theory of investment is the cornerstone of the investment-based asset pricing.One prediction of the theory is that lower cost of capital stimulates more investment,so current investment negatively predicts future stock returns.However,Cooper,Gulen,and Ion critique the wide usage of asset growth to measure investment.They argue that the Q theory is more suitable for physical capital,whereas empirically investment factors based on long-term and current asset growths have much weaker performances than the investment factor based on total asset growth.We extend the standard Q theory of investment into a two-capital setup in which firms use both physical capital(long-term asset)and short-term capital(current asset)as production inputs.We find this simple extension is capable of explaining the stronger return predictive power of total asset growth than current and long-term asset growths.A novel asset imbalance channel creates negatively correlated comovement between current and long-term asset growths that are unrelated to the discount rate effect.Part of this comovement is cancelled out in the total asset growth,giving rise to its stronger return predictive power.Empirically,once controlling for this comovement,the return predictive power of current and long-term asset growths substantially improves.Furthermore,we document compelling evidences for the model’s prediction that the asset growth effects are more prominent among firms with low asset imbalance.Our results support the Qtheory based explanation for the asset growth effect.Besides,we extend our model to China A share market,empirical analysis shows that A share market after the China splitshare structure reform also exists asset growth effect and asset imbalance mechanism,but A share market before the reform shows no obvious asset growth effect which indicates that the split-share structure reform is effective.Besides,we study the asset growth effect in a mean-variance analysis.In the asset growth deciles,Unlike the negative relation between asset growth and average returns,we document a U-shaped relation between asset growth and return volatility and hump-shaped relation between asset growth and Sharpe ratio,with the bottom of the volatility and peak of the Sharpe ratio both around zero asset growths.The pattern is robust to finer portfolio grids and portfolios sorted by the growth rates of asset components,including current and long-term asset,book equity,and debt;it also exists among stocks with different sizes,market beta and idiosyncratic volatilities.It turns out that this two-factor structure is very important to understand the patterns in portfolio volatility and Sharpe ratio.We derived the analytical relation between portfolio volatility and Sharpe ratio in a model using such a two-factor structure and replicated these observed patterns in the simulated asset growth portfolios.In the data,a linear two-factor model with the market factor and asset growth factor can explain above 80% of asset growth decile returns.In reality,for an investor with a mean-variance preference,a portfolio of stocks with nearly zero asset growth is preferred than the passive broad market portfolio. |
