Did large institutional investors flock into the technology herd? An empirical investigation using a vector Markov-switching model

This article investigates whether large non-bank institutional investors herded during the dot-com bubble of the 1990s. We use the vector Markov-switching model of Hamilton and Lin (1996) to analyse the technology stockholdings of 115 large institutional investors from 1980 to 2012. By imposing different restrictions on the elements of the transition probability matrix, we are able to test for various lead/lag scenarios that might have existed between the technology stockholding of each investor and that of the residual market. We find that only 17.4% of the investors in our sample herded during the dot-com bubble. Thus, during the dot-com bubble, herding among large institutional investors was not an especially widespread phenomenon. Among those investors that herded, 80% herded during the run-up, 10% during the collapse and 10% during both phases of the dot-com bubble. About 23% of all investors in our sample exited from the technology sector before the bubble collapsed. These results seem to support Abreu and Brunnermeier's (2003) theory of bubbles and crashes.