International markets becoming more correlated |
Back |
Karena Knaggs wrote a prize winning essay analysing evidence for increasing correlation among international stock markets for Trinity College’s Foresight Business Journal. The following are extracts. |
The key premise underlying diversification is that all assets do not move in lock step with one another. The basic idea of correlation is that, if two markets move precisely in tandem they will have a correlation of one, if they move precisely in opposite directions their correlation is -l, and a correlation of 0 means that their movements are unrelated.
Empirical findings
When we try to examine correlation between markets we are faced with any number of possible reasons why these markets are not correlated. One significant report on correlation of international markets was by Roll (1992). He investigated the pattern of correlations across countries and investigates whether that pattern could be attributed to:
• index construction; behaviour attributed to a technical aspect of an index’s construction, which could include number of stocks and concentration of markets,
• industrial compositions; the industrial structure of a country’s stock markets, i.e. what types of companies are featured on that exchange, utilities, telecommunications, retail companies etc. This is important with respect to the business cycle and how certain types of industries react in certain economic conditions,
• foreign exchange markets; problems of exchange rate behaviour.
Roll’s study found that correlation coefficients were generally quite modest in size except between countries whose economies are very closely linked. One reason given for this was time zone differences.
Time leads and lags
Roll found certain countries displayed more significant lead or lag characteristics with respect to what was affecting the market. For the lead coefficients the two most significant were Canada and the US and Roll attributes this to timezone location.
To test for the possibility that trading daytime differences were reducing correlations, multiple regressions were calculated. The results show little differences for countries in the same time zone, such as European countries.
Another factor Roll uses to explain low correlations may be global industry shocks which affect local index returns via effects on domestic industries. This may cause an exchange market intervention by monetary authorities and this intervention could possibly distort world relative prices of financial services and distort relative prices in all industries, obviously changing the exchange rate.
A study by Copeland in 1998 further examines the issues of leads and lags in the flow of information among security exchanges around the world. It found that the US has a statically significant one day lead over markets in Europe and Asia; that no significant lead extends beyond one day; that changes in foreign exchange rates contribute to the links among markets; and that the industries designated global were more sensitive to leads than local industries.
The 1987 crash
In a 1998 study by Meric, the long-run co-movements of international stockmarkets before and after the 1987 crash are compared. The study, of the world’s 10 largest stock markets (USA, Japan, France, Germany, Switzerland, UK, Austria, Hong Kong, Japan and Singapore) used 76 monthly observations for the post crash period compared to 76 months preceding the crash. It found that correlation between national stock markets increased substantially and therefore the benefits of international diversification decreased considerably after the crash.
Moving closer
The study showed the average correlation coefficient of each stock market with the other 9, for the 76 pre- and post- crash periods shows that the average correlation co-efficients of all ten stock markets increased substantially from the pre-crash to the post-crash period. The average correlation co-efficient for all 10 increased from .316 in the pre-crash period to .44 in post-crash period, an increase of 39.2 p.c..
Increase in European correlation
The co-movements of European markets have increased substantially and are likely to continue to grow. Two recent developments are helping this to happen. First, the diversification of sales and production activities across Europe and, second, the advent of EMU.
Even before EMU, European companies had diversified sales and production through Europe. This reduces the link between a positive business climate and a good stock market performance in a single country.
Variability of exchange rates and interest rate differentials explains a significant part of the divergence of individual national stock markets. So the implementation of EMU should align the movements of national European markets ever more. The highest and lowest correlations reveal that the Dutch stock exchange is the most integrated in Europe whereas the peripheral markets of Italy and Spain are the least integrated. This should be expected because the Dutch economy is greatly intertwined with other European economies and large pan-European companies dominate its stock exchange. Low correlation countries are where European integration only recently took off and companies focus on local markets.
In comparison to the Meric study however there is considerable disagreement with respect to the 1987 crash. Freimann argues that after the effects of the crash had died down, globally measured correlations receded to levels experienced before the crash but that changes in the European stock market structure were permanent.
The future
It was probably a combination of these two factors which led to increased correlation within Europe. Although the increase in European correlation is significant, it is far from complete. In the future, the implementation of EMU should give stock market integration in Europe a further boost and eventually correlations between core and peripheral markets should become virtually indistinguishable.
What will this mean for investment managers? Their ability to reduce overall portfolio risk through European diversification will be significantly reduced. |
Karena Knaggs is a recent business graduate from Trinity College.
|
Article appeared in the June 2000 issue.
|
|
|