Economic Note
03Sep02
Kahlil Rowter with statistical assistance by Yulia Ansari
The Jakarta Stock Market: Regional Influence Regional stock indices movements are commonly used as explanation for swings in the Jakarta market. Not all the time, mind you, only when the directions are similar. Conversely when the direction differs domestic reasons are then invoked.
How other market indices influence one’s own can be attributed at least to two factors. First, when a few markets are lumped together, a rising index in one member means an automatic price re-evaluation of the others. Second, rationale for changes in one member’s affects other members through contagion.
For example: if Korea’s index declines due to concerns over a slowing US economy, then a similar rationale should also affect Taiwan. From a statistical point of view, the interesting question is does movements in regional indices affect the JCI? If the answer to the first question is yes then we can ask next do the regional indices individually affect the JCI or are there a cascading effect taking place whereby indices from a select group of countries first affect other indices as well as the JCI directly and these affected countries indices then go about influencing the JCI? The data used here is daily indices from Jakarta (Jakarta Composite), Hong Kong (Hang Seng Composite), Tokyo (TOPIX 1st section), Korea (Korea Composite Index), Singapore (Singapore Stock Exchange All Shares), Thailand (Stock Exchange of Thailand), and New York (NYSE). Some would argue that the Dow Jones Industrial is a better representation of how brokers perceive the US market. This might be the case. However the NYSE and the DJIA correlates very well and therefore are interchangeable.
The range is from 04Jan00 through 29Aug02. The period chosen was to enable a complete match of all the indices included. This leaves out 201 observations out of the possible 694. Despite the fact that having a longer observation period would be useful, using only from 2000 onwards leaves out the challenge of explaining major structural changes that occurred during the crisis. Therefore the period covered here can be envisaged as a post-crisis exercise. The first order of business would be to find the correlations.
And sure enough we find that significant correlations are present among the various indices (see table 1). Now let me posit that the JCI is dependent in the sense that JCI gets influenced by the other indices depicted here. Going by this hypothesis we can re-arrange the correlations table into columns for each country in which the entries are sorted. In addition for each country we take the other country’s index which has the strongest correlations. This country’s column is presented next and the same procedure is repeated (see table 2). What we now have is an ordering of indices which directly affect the JCI and those that indirectly do so. To better illustrate the indices and their correlations we have put them in a graph (see graph 1). This shows how the indices influence each other first before ultimately affecting the JCI. Take HK for example.
Its correlation with the JCI is less than 50%, but indirectly it works its influence through The Philippines then through Singapore, Korea, and lastly Thailand. It should of course be noted that the correlations run the entire length of the observation period and that there is no guarantee it represents day-to-day movements. As a qualifier to the correlations matrix we also performed an econometric exercise to ascertain that the causalities run one way. Our Granger test results affirm that all indices with the exception of the NYSE affect the JCI and not the other way around.
Bottom line The Jakarta composite index is directly and indirectly influenced by regional indices (Thailand, Korea, Singapore, The Philippines, Hong Kong, Tokyo), and the NYSE. The largest influence comes from Thailand followed closely by Korea. The NYSE has the least direct influence but its affect is cumulatively influence is working its way through other indices in the region. This result does not show that the TIP countries (Thailand, Indonesia and The Philippines) are grouped as a similar market.