Vliv vybraných faktorů na výnosnost podílových fondů

Abstract

This master thesis deals with the influence of selected variables on average performance of five commodity funds in 2013 - 2017. The main aim of the thesis is to find and evaluate the impact of selected factors on the performance of commodity funds and to find a statistically meaningful least squares linear regression model. The secondary aim is to apprise readers of collective investment, specifically with the issue of alternative investment. The thesis uses publicly available data from which a linear regression model for econometric analysis is estimated. The explanatory variables in the model are oil prices, unemployment rate, S&P commodity index and gold futures contracts. Least squares linear regression, graphical analysis and statistical tests are applied for methodological procedures. The analysis of the time series revealed non-stationarity of the explanatory variables, which was softened by the rate of economic growth. Furthermore, the time delay of most of the explanatory variables to the explained variable was determined by cross-correlation. Therefore, the explained variable was delayed by one period. From the econometric phenomena autocorrelation, heteroscedasticity, multicollinearity, model specification and normality of residuals are analyzed in the model. The model showed weak heteroscedasticity as well as multicolinearity between oil prices and S&P commodity index. The explained variable was also predicted for the next three periods. Thanks to the transformation of the variables, the mutual influence, which initially seemed insignificant, was proven and a linear model was set with an explanatory power of 87.3% with statistically significant regression coefficients.