Stop-Loss Orders and Price Cascades in Currency Markets
Carol L. Osler
Brandeis University - International Business School
FRB of New York Staff Report No. 150
In this paper, I provide evidence that currency stop-loss orders contribute to rapid, self-reinforcing price movements, or price cascades. Stop-loss orders, which instruct a dealer to buy (sell) a certain amount of currency at the market rate once the rate has risen (fallen) to a prespecified level, generate positive-feedback trading. Theoretical research on the 1987 stock market crash suggests that such trading can cause price discontinuities, which would manifest themselves as price cascades.
My analysis of high-frequency exchange rates offers three main results that provide empirical support for the hypothesis that stop-loss orders contribute to price cascades: (1) Exchange rate trends are unusually rapid when rates reach exchange rate levels at which stop-loss order have been documented to cluster. (2) The response to stop-loss orders is larger than the response to take-profit orders, which generate negative-feedback trading and are therefore unlikely to contribute to price cascades. (3) The response to stop-loss orders lasts longer than the response to take-profit orders. Most results are statistically significant for hours, although not for days. Together, these results indicate that stop-loss orders propagate trends and are sometimes triggered in waves, contributing to price cascades. Stop-loss propagated price cascades may help explain the well-known fat tails of the distribution of exchange rate returns, or equivalently the high frequency of large exchange rate moves. The paper also provides evidence that exchange rates respond to noninformative order flow.
Number of Pages in PDF File: 44
Keywords: stop-loss, exchange rates, currency markets, high-frequency, portfolio insurance, order flow
JEL Classification: F1, G3
Date posted: July 31, 2006
© 2015 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo1 in 1.188 seconds