Hal Varian's column cites research on the irrationality of small investors during the dotcom bubble.
First, there were significant differences of opinion about the value of Internet stocks, with retail investors tending to be much more optimistic than insiders or institutions.
Second, there were significant restrictions on short-selling those stocks, a way of betting that they would decline. This prevented the pessimistic expectations from being incorporated in market prices.
While the paper to which Varian refers, by Eli Ofek and Matthew Richardson, is not on line, a related paper says that
While the fact that the price gets bid up too much in the first place is clearly anomalous, we provide arguments and evidence that the lock-up effect is not arbitrageable. In fact, trading costs, the difficulty of shorting newly-public stocks, and short-term capital gains faced by the original shareholders, all help explain this fact.
I am skeptical of this research. First of all, the estimate of the effect of the lock-up on share prices is only a few percentage points. It does not explain how Internet stocks came to be overvalued by 1000 percent. The challenges on short sales could have been overcome using derivatives. The options exchanges had created indexes of Internet stocks as early as 1996. Buying puts on those indexes, or buying puts on the Nasdaq index as a whole, was a way to bet on the bursting of the Internet bubble that did not suffer from the institutional impediments to short selling.
The other component of the analysis is that small investors allegedly are less rational than professional investors. However, there may be institutional explanations for the observation that the proportion of individual vs. institutional ownership is higher for new stocks than for seasoned stocks.
For Discussion. The crash of 1929 was blamed in part on short selling. This study says that the crash of 2000 should be blamed in part on restrictions on short selling. Which of these analyses, if any, is correct?