engleski

ANALYSIS OF THE RISK-RETURN TRADE-OFF IN THE CROATIAN STOCK MARKET

Positive relation between risk and return on investments is one of the fundamental postulates in financial theory as the risk- return trade-off (the crucial component in the investment assessment process) is based on this fundamental relation. In the last decade research has shown that this relation can be used to improve return estimation by using risk measures as return proxies, if it can be empirically proven that it holds true in a market. In order to test whether exposure to volatility as a measure of risk (but also to other risk factors) is rewarded forming long- short portfolios containing different sample data percentiles is often used to isolate and vary the tested factor exposure. Since earlier research for the Croatian stock market did not provide significant evidence of positive relation between risk and return in this research additional risk measures are tested. Furthermore, the long-short portfolios are decomposed and their components’ behaviour is analysed and compared in order to provide a comprehensive market assessment. The research is based on weekly returns in excess of the T- bill rate for stocks that were included in the CROBEX index in the period from January 2005 till November 2017. Standard deviation, semi- deviation, skewness and kurtosis are tested as risk measures and are calculated for multiple samples based on a 52-week rolling window of returns. For each risk measure stocks were sorted into equally-weighted quartile portfolios. Each quartile portfolio was held for 1 week out-of-the sample producing a time series of 672 out-of-sample estimates of weekly returns. Estimates were calculated separately as arithmetic average return and median of the returns per each quartile and per each risk measure. The obtained data was further used to calculate the cumulative returns of individual quartile and long-short quartile portfolios. Research findings show that unlike for the standard deviation, semi-deviation and kurtosis, skewness as a risk measure seems to be able to capture the risk premia associated with going long in the riskiest quartile portfolio and short in the less risky ones. However, this only seems to hold if median is used as the portfolio return estimator and in the post-expansion period in the stock market. Moreover, the sources of these premia (analysed by decomposing the long-short portfolio and exposing the cumulative returns of the quartile portfolios) reveal that the fall in returns of the less risky portfolios seems to impact the outcome more than the rising returns of the riskiest portfolio. Nevertheless, in the post- expansion period quartile portfolios’ premia follow the risk-return trade-off principle in this case. In the case of other risk measures long-short portfolios cumulative returns do not seem to capture any risk premium with kurtosis arguably exhibiting the worst performance of all. However, research findings also point out that the low-volatility i.e. low-variability anomaly seems to be present in the market in the post-crisis period in the case of standard deviation and semi-deviation respectively. The phenomenon explains the poor performance of the long-short portfolios designed to capture the standard risk-return trade-off and seems to be more pronounced when using the median as the return estimator. On the other hand, when using the arithmetic average the cumulative returns of the both low-volatility and low semi- deviation stocks exhibit stronger growth potential suggesting that the arithmetic average serves as a better estimator of the stock returns which is explained by positively skewed distribution of returns. Research findings also show that during the expansion period in the market none of the cumulative returns analysed based on the tested risk measures seems to follow the risk-return trade- off principle properly, including skewness. If data sample is further reduced by filtering out the less liquid stocks main research findings remain unaffected suggesting that outliers and unreliable data do not distort the presented results. Presented findings reveal that a relation between risk and return on investments exists in the market only in the post-crisis period (positive and anomalous) depending on the risk measure. Questions are raised regarding causes of identified phenomena and some explanations are offered but conclusions warrant further research especially regarding the rewarded risk factors in the Croatian stock market.

Risk-return trade-off, higher order moments, low volatility anomaly

nije evidentirano