U.S. commodity futures markets have been transformed by the arrival of electronic trading and the subsequent domination of algorithmically driven high frequency traders (HFTs). In the context of this structural change, we investigate the reliability and the consistency with which electronic financial markets with HFTs provide transactional liquidity services. Our analysis is based on proprietary account-level intraday data from U.S. oil futures markets. Using data from 2006 (before electronification) and 2011, we show that, unlike the erstwhile locals in futures pits, HFTs significantly reduce their participation as well as their liquidity provision in periods of high and persistent (i) volatility, (ii) order imbalances, and (iii) bid ask spreads. The changes in liquidity provision that we observe in an 8-month period around the 2008 Lehman bankruptcy and amid the 2011 Libyan political crisis support our main conclusions. At the same time, however, we also establish that trader anonymity in electronic markets does not necessarily make such markets less conducive to facile adjustment of severe information asymmetries. Indeed we show that, in contrast to HFTs, some other types of machine traders – who trade less frequently, but as fast and as intensely when they do trade – do not pull back under stress: to the contrary, they increase both their participation and their liquidity provision. Overall, our results provide evidence that the nature of the algorithm matters, and that focusing solely on HFTs is likely to distort perceptions of liquidity provision’s potential fragility in algorithmically driven electronic markets under stress.
Back to Workshop III: Commodity Markets and their Financialization