Below are thoughts from an author of a paper Tyler cited on algorithmic trading (and HFT). There is a project by the UK government on related topics. Some related working papers on HFT are here, here, here, here, and here.
1. Technology has made financial markets work better; improvements in liquidity are large, important, and should result in lower costs of capital for firms; these do not mean that every application of technology is good.
2. There is evidence that investors prefer continuous to periodic trading, but batch auctions as frequent as every few seconds have not been studied.
3. Until technology allows buyers and sellers to better find each other simultaneously, markets need a group of intermediaries; the lowest-cost intermediaries are those closest to the market.
4. Historically, intermediaries were floor traders, now are HFT; floor traders profit from those further from the trading mechanism as do HFT now.
5. What is the best industrial organization for the intermediation sector? i) free entry (HFT) or ii) regulated oligopoly (NYSE specialists, Nasdaq market makers, etc.)?
6. Floor trading had the advantage that within-market relative latency was not so important and the amount of market data produced was small; costs were floor traders’ large advantages and possible collusion.
7. There is yet to be robust empirical evidence linking HFT to declines in market quality or efficiency; Haldane has interesting ideas, but as comments point out, it is difficult to blame HFT more than the economic and euro crises for recent fat tails in asset returns; systemic uncertainty increases fat tails and cross-asset correlations.
Overall, technology applied to intermediation appears to bring benefits with the standard rent seeking costs of intermediaries making money, possible instability (although 1987 showed human markets have their own failings), and technology costs.
Can markets find solutions?
i) If HFT becomes competitive (zero rents), will HFT then resell their technology as brokers? Could this lead to efficiency without negative externalities?
ii) Do dark pools and batch auctions limit part of the “arms race” of technology investment? Significant volumes are already traded in these ways, e.g., the opening and closing auctions. There are many ways investors can avoid HFT. If they do not, is it revealed preference?
If regulations are needed they should target behavior, not certain trading firms, otherwise HFT features will simply be incorporated into other strategies, e.g., a HFT strategy is merged with a mid-frequency strategy.