COMPARISON OF COPULA FAMILY (GAUSSIAN, ARCHIMEDEAN, AND REGRESSION) IN A CASE STUDY OF COMPOSITE STOCK PRICE INDEX ON INDONESIA STOCK EXCHANGE

Abstract

Stocks are one of the most popular financial market instruments. On the other hand, stocks are an investment instrument that is widely chosen by investors because stocks are able to provide attractive profit levels. Investment is an effort to postpone consumption in the present. Comparing copula families is crucial for selecting the model that best fits the observed data dependency structure. This helps produce more accurate analysis and more meaningful interpretations. This study analyzes different types of copula relationships using the Tau Kendall method, applying it to the movement of the Composite Stock Price Index (IHSG) on the Indonesia Stock Exchange (IDX). The data used are secondary monthly data of IHSG as a response variable, while the explanatory variables are inflation (%), exchange rate (Rp/USD), and interest rate (%) in 2010-2014. The results show the pattern of the relationship between IHSG and its macroeconomic factors on the IDX using copula parameter estimation with the Tau Kendall approach, with the largest log-likelihood fitting results showing a relationship pattern following the Gumbel copula, namely IHSG with inflation, interest rates with IHSG following the Clayton copula, and exchange rates following the Frank copula. Meanwhile, using the regression copula has better interpretation results compared to the Gaussian and Archimedean copula, with an MAPE value of 0.122 with a correlation of 70.63%.