The Sound of Many Funds Rebalancing

Abstract

This paper proposes that portfolio-rebalancing cascades generate noise in financial markets. In particular, we study the rebalancing decisions of exchange-traded funds (ETFs). Stocks are often held by many different ETFs that track a wide variety of benchmarks. We show theoretically that, when many ETFs are involved, small price changes can trigger large rebalancing cascades that wind up affecting unrelated stocks on the opposite side of the market. We then show empirically that, when many ETFs hold a stock, aggregate ETF demand for that stock passes statistical tests for white noise and that aggregate ETF volume predicts total trading volume, liquidity, and insider-trading activity. These results persist when we instrument for aggregate ETF volume with the number of unrelated ETFs in the market — that is, with the number of ETFs that hold neither the target stock nor any stocks held by ETFs that hold the target stock. The number of unrelated ETFs is correlated with the potential for rebalancing cascades but not with characteristics of the target stock, so our results are unlikely to be driven by unobserved stock-level differences.