Does The Consensus Prevail? Experimental Evidence

We test hypotheses on herd behavior in a simple investment decision through an experimental setting. Subjects were given some fundamental information about a specific firm and asked for a recommendation, buy or sell. This personal judgment was then confronted to the opposed consensus of analysts, in order to determine if subjects revise their recommendations. In this binary choice setting, we show that herding takes place, and is inversely correlated to perceived individual ability. Moreover, when reputation is at stake, conformist subjects are even more prone to follow the consensus in their decision making. Herding behavior is, among other behavioral biases, a frequently cited phenomenon by both money managers and academics to explain booms and crashes in financial markets (Denevow and Welch [1996]). Sometimes, agents on the market are presumed to act according to the behavior of others. Although such behavior is often analyzed in the literature (e.g. Scharfstein and Stein [1990], Bikhchandani, Hirshleifer and Welch [1992], Graham [1999], or Chamley [2004]), empiric evidence is still quite scarce (Welch [2000]). Since Lakonishok, Shleifer, and Vishny [1992], many works (e.g. Wermers [1999], Wylie [2005]) have tried to show clusters among individuals, who act together on the market, compared to a "normal" behavior. However, most of these studies are not convincing since they come up against the detection of actual herding. The fact that individuals acted in the same manner is not always the consequence of herding, and can merely be a common reaction to a common