Bad advice, herding and bubbles

Abstract

Prior to the crash of the housing market, many experts told the public that the upward trend in housing prices was not a bubble, it could be explained by fundamentals. This paper shows that an increase in the propensity of individuals to herd toward trend-chasing behavior caused, for example, by bad advice from experts, increases the likelihood that a destructive bubble will occur.

Keywords:

• herding
• bubbles
• economists

Acknowledgements

I would like to thank the participants at the Workshop on Methodology, Systemic Risk, and the Economics Profession held at Duke University in December 2011, and the participants at the Duke INET Workshop at the 2012 American Economic Association Meetings in Chicago for their helpful comments.

Notes

 1. See Wieland and Wolters (2011) for an overview of the forecasting performance of macroeconomic models before, during and after the crisis.

 2. Allowing agents to switch beliefs based upon a weighted average of many past time periods, as in some of the cases Brock and Hommes (1998) consider, would not fundamentally alter the results.

 3. The modification, as explained below, is in the equations governing the evolution of the share of agents holding each type of belief.

 4. Some belief types, such as rational beliefs with a constant bias, do not produce bubbles at all.

 5. The development of the model through Equation (1) is a condensed version of the model in Brock and Hommes (1998), which is itself based upon the standard asset pricing model.

 6. The excess return is defined as capital and dividend returns in excess of what could have been earned if wealth had been held as a risk-free asset instead, i.e. any excess over the opportunity cost of the investment.

 7. If the shock results in a value for the share that is less than zero or greater than one, the share is reset to the boundary.

 8. This assumption makes bubbles asymmetric, i.e. they inflate slower than they pop. It would be better if the asymmetry arose endogenously, for example, from asymmetry in the loss function due to loss aversion, but that just puts the assumed asymmetry at a different level. The end result is the same.

 9. Fundamentalists do not know the fractions of each belief type, n _h,t .

10. It would be possible to let the value of c be estimated from actual data on past prices, and then condition expectations for the trend-chasers on this estimate.

11. The model runs for total 2000 time periods. The first 1000 time periods prime the model and iron out the effects of starting values, and the second 1000 time periods provide the data for the simulations.

12. Bubbles are defined as prices exceeding fundamental values by more than 50%. Thus, if the fundamental value is 100 and the price rises above 150, it is counted as a bubble. If the price rises to less than 150, it is not.

13. The expectations for fundamentalists are of the form , where δ₀ and δ₁ are functions of the parameters of the model.

14. The random shock in the model can push the model into an unstable region, and as the slope increases, this becomes more and more likely.

15. An increase in the other two parameters of the model, the variance of the shock to dividends, , and the variance of the shock to the share following each strategy, , increases the frequency of bubbles in approximately a straight-line fashion.