Managing fundamentals versus preferences: Re-balancing portfolios and stock returns

What can granular data on investor holdings tell us about stock price variation? I model the growth rate of a portfolio manager's holdings based on evolving asset fundamentals by including demand for asset-specific characteristics in a portfolio optimisation function. Alongside changes in asset characteristics, the manager re-allocates wealth according to evolving preferences. This introduces memory into the portfolio management problem, as past investments inform the choice for new allocations. Using the model, I decompose the growth rate of mutual fund holdings by the effect of i) changing stock characteristics, ii) new preferences, and iii) mean reversion in latent demand. I nest these estimated components, by aggregating holding growth rates by the fund's total net assets, into an expression for stock price growth. I find that changing preferences explain at least as much variation in stock prices as changes in fundamentals. This demonstrates the importance of studying heterogeneity in investor preferences, and their evolution, in furthering our understanding of stock market phenomena.