This paper provides market microstructure evidence, including price discovery, magnet effect and volatility spillover effect, on the failed market-wide circuit breakers in China’s stock exchanges on January 4 and 7, 2016. The results for event samples are compared with non-event samples which control for normal trading behaviour around information release but absent regarding trading halts. We find trading halts provide no “time-out” cushion given that strong positive return autocorrelation on the days circuit breakers were triggered, and no improvement in the informativeness of stock prices given the smaller price contribution during halt relative to non-halt periods. Traders face time constraints from impending trading halts which force them to quickly close their positions as information asymmetry peaked, thus causing price to accelerate towards its limit (a magnet effect) and triggering a subsequent halt. Volatility of within-stock spills over into post-halt and overnight non-trading periods while cross-sectional volatility spillover between A- and B-shares increases in post-halt periods. Collectively the empirical evidence justifies the abolition of market-wide circuit breakers by government policy-makers and/or regulators. 

Keywords: Market microstructure; Circuit breaker; Price discovery; Magnet effect; Volatility spillover; China
JEL Classification: G10; G14; G18​