LAUSR

Abstract: Conventions, or “that the existing state of affairs will continue indefinitely, except in so far as we have specific reasons to expect a change” (Keynes 1936), play a central role in over-the-counter markets. For instance, by allowing expectations about the future to become more harmonized and orderly, they act as stabilizers for the provision of liquidity. Conventions might, of course, change at any time. Nonetheless, by being attached to the daily trading routine and/or integrated within the institutional structure, the confidence in their relevance and validity can be long-lasting. In the foreign exchange market, in particular, where prices are quoted to end-users on demand, market-making banks rely on a convention to quote prices to each other to maintain liquidity. However, the rise of algorithmic and high-frequency trading poses a practical as well as a theoretical challenge to such conventions. By reacting ultra-fast to new information, including to new limit orders submitted by others, markets largely populated with algorithmic traders have become susceptible to a withdrawal of liquidity at an unprecedented speed and scale. Using a high-frequency dataset provided by Electronic Broking Services (EBS), we investigate the process of liquidity withdrawal from the foreign exchange spot market. By doing so, we consider the crowding out of conventions associated with liquidity provision, traditionally upheld through mutual understanding among financial institutions – in other words, reciprocity and trust among humans.