A Surprising Development: Tests of the Capital Asset Pricing Model and the Efficient Market Hypothesis in Turkey's Securities Markets
27 Pages Posted: 24 Jul 2001
Date Written: June 2001
This paper performs two tests of the efficient market hypothesis in the markets for Turkish securities using GARCH-M methods. The first test is a joint test of the Efficient Market Hypothesis and a Multifactor Capital Asset Pricing Model. We wondered if a Capital Asset Pricing Model with inter-temporal variation of risks would shed any light on the reason for surprisingly low estimates of beta in the Classical Capital Asset Pricing Model for Turkey and other Emerging Markets. We also wondered whether, because of the newness and high volatility of Turkey's financial markets, we could find market inefficiencies. We find that we cannot reject the null hypothesis that the CAPM applies and Turkish securities markets are efficient. The first test analysis also provides an estimate of the cost of capital in Turkey and gives us an estimated value of the beta of Turkey's common stock with the World stock portfolio of about one. The second test develops a forecasting model for returns to a specific portfolio, long the ISE Dollar Index and short the S&P 500 composite. We sought once again evidence of market inefficiency by asking whether returns to this portfolio were greater than or were commensurate with its systematic risks. We find that simply buying and holding the portfolio produced a low ratio of return to risk, suggesting that the portfolio is not on the efficient frontier of the Capital Market line. But we found that an active trading rule generates very high returns. Normally this is evidence of weak form inefficiency. But because portfolio risks also vary over time we still cannot reject the efficient market hypothesis. A surprising development.
Keywords: Efficient Market Hypothesis, Turkey, Multifactor Capital Asset Pricing Model, Cost of Capital, GARCH-M
JEL Classification: F3, F36, G1, G15
Suggested Citation: Suggested Citation