Asset Fire Sales or Assets on Fire?

38 Pages Posted: 4 Dec 2019

See all articles by Simon Schmickler

Simon Schmickler

Princeton University, Department of Economics

Date Written: November 17, 2019


Using high frequency data and machine learning methods, I propose a new method to isolate a plausibly exogenous component of mutual fund flows and use it as an instrument to revisit classic empirical questions in Finance because previous methods are vulnerable to a reverse causality critique.

The idea that mutual fund flows induce fire sales which drive asset prices away from fundamentals has been fruitful. Based on this idea the Asset Pricing literature finds markets are inefficient and fragile and the Corporate Finance literature shows market misvaluation distorts real outcomes. First, I argue these findings are partially driven by reverse causality. Assets on fire reduce mutual fund returns which trigger outflows. Empirically, this becomes apparent when increasing the frequency of the standard event study from quarterly to daily; returns precede flows. Second, I suggest a solution. In contrast to quarterly flows, an instrument constructed from daily surprise flows is exogenous to fundamentals. Also, this instrument can be strengthened by training machine learning models to predict how mutual funds trade in response to flows. Third, I use surprise flows to reevaluate important findings in the literature. Overall, while I confirm most findings qualitatively, the new estimates imply that equity markets are more efficient, less fragile and less distortive than suggested.

Keywords: Mutual Fund Flows, Fire Sales, Market Feedback Effects, Return Predictability, Trading Predictability, Machine Learning, Neural Nets

JEL Classification: G12, G23, G32

Suggested Citation

Schmickler, Simon, Asset Fire Sales or Assets on Fire? (November 17, 2019). Available at SSRN: or

Simon Schmickler (Contact Author)

Princeton University, Department of Economics ( email )

Princeton, NJ
United States

Here is the Coronavirus
related research on SSRN

Paper statistics

Abstract Views
PlumX Metrics