Algorithmic trading and mini flash crashes: Evidence from Austria

We use stock-day level data on the market share of algorithmic trading to analyze whether algorithmic trading affects the frequency of mini flash crashes in the Austrian stock market. We use an instrumental variables approach and the Petrin and Train (2010) control function approach to address endogeneity concerns. We find no evidence that algorithmic trading significantly affects the probability of the occurrence of mini flash crashes. • We study 144 mini flash crashes on the Austrian stock market between 2011 and 2015. • We use panel logit models to relate mini flash crashes to algorithmic trading. • We address endogeneity by using a control function approach. • We find no evidence that algorithmic trading significantly affects mini flash crashes.