DP12737 Liquidity Regimes and Optimal Dynamic Asset Allocation
We solve for the optimal dynamic asset allocation when expected returns, volatilities, and trading costs follow a
regime switching model. The optimal policy is to trade partially towards an aim portfolio with a given trading speed.
In a given state, the aim portfolio is a weighted average of mean-variance portfolios in every state, where the weight
is a function of the probability of transitioning to that state, and the state's persistence, risk and trading costs.
The trading speed is higher in states that are more persistent, where return volatility is higher and trading costs
are lower. It can be optimal to deviate substantially from the mean-variance efficient portfolio (or from the risk-parity allocation) and to underweight
high Sharpe ratio (high volatility) assets, as well as to trade more aggressively the less liquid assets in anticipation of an increase
in their volatility and trading costs. We illustrate our approach in an empirical exercise in which we exploit
time-variation in the expected return, volatility, and cost of trading of the value-weighted market portfolio of US
common stocks. We estimate a regime switching model applied to a dataset of institutional trades, and find that
realized trading costs are significantly higher when market volatility is high. The optimal dynamic strategy
significantly outperforms a myopic trading strategy in an out-of-sample experiment. The highest gains arise from
timing the changes in volatility and trading costs rather than expected returns.