A new empirical paper on the dot-com bubble of the late nineties presents some serious challenges to the efficient market hypothesis. The authors report that 'institutions trade in the direction of clear mispricing in a small sample of equity carve-outs', the mispricing being a parent firm's share in a dot-com venture that exceeds the value of the parent firm (yes, that means that part of company A is worth more than company A altogether!). This should seriously worry anyone who believes that experts' use of arbitrage is a rationalising force in financial markets.
Furthermore, the paper takes some of the bubble blame away from individual investors by identifying big institutions (such as hedge funds) as the main drivers behind the bubble. This finding again (remember this guy?) contradicts the idea that the dot-com bubble was fuelled by personal investors, which seems like a textbook case of mistaking correlation for causality.