Information Processing in an Asset Market Experiment with Algorithmic Arbitrage Trading

In a novel laboratory setting in which scheduled information changes the expected asset value, we test the efficient market hypothesis. The scheduled information contains either good or bad news. The data show that average prices are not different from the fundamental value. However, asset prices underreact to news; we observe asset prices below fundamentals after good news and above after bad news. There seems to be a behavioral asymmetry; absolute price deviations are larger after bad news than after good news. In the course of a period, the asset price drifts towards the fundamental value. Following the literature that hypothesizes arbitrage trading would push asset prices towards fundamentals, we have an arbitrage robot trader interacting with human subjects in the experimental market to find no support for this hypothesis.