Causality Patterns Between International Financial Markets
Causality Patterns Between International Financial Markets
Test problem
Financial
markets
Europe
America
Asia Finland Causal relationship?
Time lag?
Stability?
Causality
A concept that has puzzled philosophers and scientists for hundreds of years starting from the ancient Greece
A common feature in all definitions throughout the years is the relation between cause and effect
The first definitions too abstrtract for empirical testing
Hume (1740) made a step towards empirical testing of causality emphasizing that causation is a relation between experiences rather than one between facts
Hume’s view is an analogy to classcal statistical inference
Causality
Hume recognized three basic criteria for causality
Spatial/temporal contiguity
Temporal succession
Constant conjunction
Feigl (1953) defined causality as ”predictability according to a law or set of laws”
Suppes (1970) operationalized the predictability in terms of probability:
One event is the cause of another if the appearance of the first event is followed with a high probability by the appearance, and there is no third event that we can use to factor out the probability relationship between the first and second event
Causality in the Granger sense
The practical solution to the problem of statistically measuring causality between observed time series was presented by Granger (1969)
Granger’s definition is a combination of Feigl’s, Hume’s and Suppes’ concepts
Granger defined causality between two variables X and Y in terms of one-period predictability
Causality in the Granger sense
Variable X is said to cause another variable Y with respect to a given universe or information set that includes X(t) = {Xt, Xt-1,…} and Y(t) = Yt, Yt-1,…} if Yt+1 can be better predicted by using the information in X(t) than by not doing so, all other relevant information being used in each cases
Causality in the Granger sense – Statistical definition
A significance test of the difference between residual sum of squares from a linear autoregressive equation estimated in unconstrained and constrained forms:
Time in measuring causality
The concept of causality itself contains an implicit assumption of a certain time sequence between the effects, i.e. cause precedes the effect in time
The time lag varies depending on the data used from a fraction of a second (chemical processes) to decades (medical research)
In testing causal relationships between financial markets daily or weekly observations most common, even though the importance of intraday data is growing
Problems with the nonsynchronous market open times
The Finnish financial markets – some special characteristics
Rapid growth in the 1990’s
30-folded market capitalization in nine years
Total equity turnover in 2000 more than 200 times that of 1991
Thin in international comparison
158 listed series at the end of 2000
Extreme dependence on the IT sector and especially on one share series, Nokia
66% of turnover and 70 % of market capitalization in 2000
Statistically anomalous and problematic in modeling and predictions
Yearly Equity Turnover (Million €) of Helsinki Stock Exchange
Comments