I've written before that the forex market microstructure is one of the more interesting finance research areas at the moment (see What economists can learn from currency traders). Richard Lyons of the Haas School of Business has a new paper: Myths about the micro approach to exchange rates (PDF).
We address four common misconceptions about micro-based research on exchange rates: (1) public news arrivals account for most exchange rate variation; (2) allocative trades do not convey information; (3) order flow is easy to measure; and (4) transactions obviously drive prices. Though few people subscribe to all four, most people subscribe to at least one. Dispelling these misconceptions should stimulate progress in this area.
...The four misconceptions addressed here are prevalent. Most people subscribe to at least one. They are certainly not the only misconceptions associated with microbased exchange rate research. We chose to focus on these because the broader issues they raise are fundamental.
This paper was given at the ESF workshop on High Frequency Econometrics and the analysis of Foreign Exchange Markets, held 26-27 June 2006 at Warwick Business School. Some of the other papers (available online) are also of interest.