How Market Intervention Can Prevent Bubbles and Crashes

Abstract

Using an agent-based model (ABM) with fundamentalists and chartists, prone to develop bubbles and crashes, we demonstrate the usefulness of direct market intervention by a policy maker, documenting strong performance in preventing bubbles and drawdowns and augmenting significantly the welfare of all investors. In our ABM, the policy maker diagnoses burgeoning bubbles by forming an expectation of the future return of the risky asset in the form of an exponential moving average of the excess return over the long-term return. The policy maker invests in the risky asset when he detects a small deviation of the return from the long-term growth rate in order to construct an inventory that he draws upon later to fight future market exuberance. Then, when this deviation between the current growth rate and the long-term growth rate exceeds the policy maker's tolerance level, he starts to sell the risky asset that he has accumulated earlier, in a countercyclical fight against future price increase. We find that the policy maker succeeds in preventing bubbles and crashes in our ABM. In simulations without bubbles, the policy maker behaves similarly to the fundamentalists and his impact is negligible, following the principle of "Primum non nocere". In simulations where bubbles form spontaneously as a result of the noise traders's strategies, the policy maker's intervention reduces the average drawdown by a factor of two when his market impact becomes significant. We find that the policy maker intervention improves all analysed metrics of market returns, including volatility, skewness, kurtosis and VaR, making the market less turbulent and more stable. The combination of fewer bubbles and crashes, lower market risks and the stability of the long-term growth rate make the policy maker intervention to improve the welfare of all investors as measured by their risk-adjusted return, increasing the Sharpe ratios from approximately 0.3 to 0.5 for noise traders, from 0.6 to 0.8 for fundamentalists as the market impact of the policy maker increases to the level of the fundamentalists. We also test the sensitivity of these results to variations of the key parameters of the strategy of the policy maker and find very robust outcomes. In particular, the conclusions are unchanged even under very large miscalibrated long-term expected returns of the risky asset.

Keywords: financial bubbles, agent-based model, arbitrageurs, prediction, noise traders, fundamentalists, market intervention

JEL Classification: C53, C63, E58, E60, G01, G17, G18