Investors' biased expectations and returns predictability -a trading rule test
ILOMÄKI, JUKKA (2009)
2009
Kansantaloustiede - Economics
Kauppa- ja hallintotieteiden tiedekunta - Faculty of Economics and Administration
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2009-06-11
Julkaisun pysyvä osoite on
https://urn.fi/urn:nbn:fi:uta-1-20361
https://urn.fi/urn:nbn:fi:uta-1-20361
Tiivistelmä
In this study a statistical technique is proposed which attempts to recognize statistical arbitrage opportunities. A certain kind of nonlinear cointegration approach is proposed, which we call temporary cointegration. In addition, unidirectional Granger causality is proposed to be an essential aspect for the forecastability of the technique. A mechanical trading rule based on the proposed technique produces economic profits in the testing period January 2000 ???? March 2007. In the trading rule test, the hit ratio for the forecasts becomes 80.8 % and it is statistically significant with p-value < 0.0001.
Because testing of the trading rule yields economically and statistically significant results, further research is needed to find possible causes for these economic profits. We examine this with two techniques. We determine (ex-post) the explanatory power of common risk factors that may produce the trading rule returns by multi risk factor model. In another approach, we forecast (ex-ante) with another procedure (macro-factor model and VAR without ECM) the same observations as in the trading rule and compare the results. After accounting for exchange rate effect, macroeconomic risk factors and lead-lag effect without ECM, the trading rule profits are still economically and statistically significant.
We suggest that when investors who invest to minor investing country assets receive ambiguous information concerning their assets, they may become confused in growing uncertainty. This may lead to a phenomenon where some investors take role models from large financial centers. In other words, there is a herding behavior with a lead-lag effect. The herding may start because investors are risk averters which act as a trigger for this particular herding behavior. If this herding behavior continues long and strong enough, this phenomenon may lead to a temporary relative dependency between particular assets. This may happen because investors suffer from an illusion of validity. There is a possibility that investors are overconfident about the success of their current investing behavior. Therefore, their actions affect the recency bias. We argue that these biases may lead to temporary cointegration with a lead-lag effect between particular asset prices. We propose that unidirectional temporary cointegration is consistent with the existence of herding with a lead-lag effect.
We demonstrate that in a two-asset economy where the investors have standard rational preferences but they may have biased beliefs, specific behavioral biases are sufficient conditions for existence of irrational herding with a lead-lag.
Keywords: temporary cointegration, unidirectional Granger causality, behavioralbiases, herding.
Because testing of the trading rule yields economically and statistically significant results, further research is needed to find possible causes for these economic profits. We examine this with two techniques. We determine (ex-post) the explanatory power of common risk factors that may produce the trading rule returns by multi risk factor model. In another approach, we forecast (ex-ante) with another procedure (macro-factor model and VAR without ECM) the same observations as in the trading rule and compare the results. After accounting for exchange rate effect, macroeconomic risk factors and lead-lag effect without ECM, the trading rule profits are still economically and statistically significant.
We suggest that when investors who invest to minor investing country assets receive ambiguous information concerning their assets, they may become confused in growing uncertainty. This may lead to a phenomenon where some investors take role models from large financial centers. In other words, there is a herding behavior with a lead-lag effect. The herding may start because investors are risk averters which act as a trigger for this particular herding behavior. If this herding behavior continues long and strong enough, this phenomenon may lead to a temporary relative dependency between particular assets. This may happen because investors suffer from an illusion of validity. There is a possibility that investors are overconfident about the success of their current investing behavior. Therefore, their actions affect the recency bias. We argue that these biases may lead to temporary cointegration with a lead-lag effect between particular asset prices. We propose that unidirectional temporary cointegration is consistent with the existence of herding with a lead-lag effect.
We demonstrate that in a two-asset economy where the investors have standard rational preferences but they may have biased beliefs, specific behavioral biases are sufficient conditions for existence of irrational herding with a lead-lag.
Keywords: temporary cointegration, unidirectional Granger causality, behavioralbiases, herding.