In
traditional finance theory, the efficient market hypothesis suggests that
investors can not make any abnormal returns given the same amount of risk. Investor
rationality has always been a topic studied in behavioral finance, as there are
many anomalies in the stok market caused by investor irrationality. The
existence of psychological barriers is related to the belief in the
predictability of stock price and contradictive to both market efficiency
hypothesis and the assumption of rational investors. In this point of view,
studies which find evidence for existence of psychological barriers contribute
to the literature on market anomalies. The effect of pscyhological barriers can
be explained by an anchoring effect. Furthermore, these barriers are often
associated with support and resistance levels and with a bandwagon effect. But
the existence of psychological barriers in financial markets is a relatively
undocumented anomaly. Also there is no prior academic research on psychological
barriers on Borsa İstanbul.
This
study investigates whether psychological barriers exist in a Turkey Stock
Exchange and major Turkish stocks. Therefore two Stock Index and the daily
closing price of ten stocks are examined with respect to the price clustering
around the barrier and also return and volatility. Generally two different
statistical analysis have been used in studies of the uniformity of digits: the
chi-square and dummy regression analysis. A third approach uses GARCH analysis
to assess the differential impact of being above or below a barrier. In this
study both these statistical method and GJR-GARCH analysis is used to test of
barriers existence.
According
to chi-square test and dummy regression results, there is strong evidence for
the existence of systematic behavior at the BIST 30 index at the 10000s barier
level. The frequencies within the barrier band are significantly lower then
frequencies outside the barrier band. But for most of the examined stock there
is strong evidence in the opposite direction: higher frequencies inside the
barrier band and lower frequencies outside. According to GJR-GARCH results,
there is no significant difference in the conditional mean for the periods
immediately before and immediately after an upward and downward crossing a
supposed barrier. But, there is significant difference in the conditional
variance for the periods immediately before and immeditely after an upward and
downward crossing of a supposed barrier.
Sümeyra Gazel / Erciyes
University, Institute of Social Sciences / Ph.D. Thesis, 2013