Dion Bongaerts, Sarah D De Luca, Mark Van Achter
Review of Finance, Volume 28, Issue 6, November 2024, Pages 1953–1989, https://doi.org/10.1093/rof/rfae029
Recently, financial market runs, like the one on SVB bank, have underscored the importance of containing coordination failures in financial markets. While bank runs have been widely studied, similar runs can also ensue in securities markets, like the public equity market, resulting in downward price spirals. The standard regulatory tool to contain such market runs is a circuit breaker: a procedure that halts markets following price moves that exceed a certain threshold value (e.g., returns exceeding -10\%). Yet, circuit breakers are ill-understood and understudied, and it is often unclear how to optimally calibrate the threshold for them to trigger. Anecdotal evidence from China and the US even suggests that improperly calibrated circuit breakers can give rise to rather than mitigate market runs.
We put forward a model to analyze the optimal configuration of circuit breakers. Our model runs over two periods, abstracts from asymmetric information, and features investors that can (but need not) be hit by a common liquidity shock in either period. In the model, a mechanical, competitive and risk-averse market maker with an inventory concern stands ready to absorb any order flow and sequentially passes on any marginal change in disutility to investors through transaction prices (price impact). Investors can submit orders in either period and experience uncertainty in terms of execution order. All investors optimally want to trade in case of a contemporaneous liquidity shock. Absent a shock in the early period, some investors still strategically submit orders in the early period to front-run their peers in case a liquidity shock hits in the late period. This preemptive trading, which we call upfront trading, is inefficient if no shock arrives in the late period and welfare-neutral otherwise. Hence, a circuit breaker that shuts down trading in the early period in the absence of a liquidity shock would be optimal.
Yet, a social planner may not be able to distinguish upfront trading from trading in response to a contemporaneous liquidity shock. In the latter case, curtailing the market in the early period impairs gains from trade and is hence is inefficient. We derive in the context of our model an optimal circuit breaker which is conditional on the probability of a current vs a future shock. If the probability for a contemporaneous liquidity shock is high, the circuit breaker is optimally loose to allow gains from trade to be realized. If the probability of a future liquidity shock is moderately high, there is a lot of inefficient upfront trading and the circuit breaker is optimally tight to contain such upfront trading. Consistent with the failed introduction of the Chinese circuit breaker in 2015, if the probability of a future shock is small but non-zero, a too tight circuit breaker limits losses of upfront trading and thereby creates excessive incentives for doing so. As the probabilities of current and future shocks are crucial for our optimized circuit breaker, we illustrate how to calibrate these probabilities empirically.