The causality is the principle or origin of something. The concept is used to name the relationship between a cause and its effect, and can be used in the fields of physics, statistics, and philosophy.
The physics holds that any event is caused by an earlier. Therefore, if the current state of something is precisely known, it is possible to predict its future. This position, known as determinism, was nuanced with the advancement of science.
According to the principle of causality, every effect always has a cause. The principle of uniformity adds that, in identical circumstances, a cause always produces the same effect.
For philosophy, causality is the law by virtue of which effects are generated. Philosophers consider that the fact of any event is caused by a cause and indicate three conditions for A to be the cause of an effect B: A must occur before B, whenever A occurs, B must occur and A and B must be close in time and space.
The statistics, in turn, maintains that causality is a necessary relation of co-occurrence of two variables.
The notion of causality is also present in popular wisdom or informal knowledge. Several sayings spread this idea, such as “you will reap your sowing” or “whoever sows winds gathers storms”. These phrases are not linked to scientific or factual facts, but have their value in the belief that people’s behavior inevitably has consequences.
Granger causality test
Clive WJ Granger, an economist born in 1934 in Great Britain and winner of the Nobel Prize in Economics in 2003, was the author of a statistical hypothesis test whose objective was to determine if a time series (also called chronological, is a sequence of data) served to predict another.
In general, statistical regressions (a phenomenon whereby an extreme measurement tends to get closer to the mean after a second observation), reflect mere correlations, but Granger claimed that causality in economics could be shown through some kind of test.
It is worth mentioning that, since true causality is a deeply philosophical question, experts in econometrics (a branch of economics that uses various statistical and mathematical resources to carry out analyzes, interpretations and predictions about economic systems) maintain that the Granger test it can only return predictive causal information.
The test, which allows finding out if a variable can offer useful results to predict the value of another, provided that its character is unidirectional or bidirectional, requires the comparison of the present behavior of a time series X with the past one, to deduce if it is capable to predict the behavior of a time series Y. If the result is positive, it can be said that the result X causes in the Granger sense the result Y, and their behavior is considered unidirectional.
If, on the other hand, everything explained in the previous paragraph takes place, and the fact that the result Y allows to predict the result X is added, then we are in the presence of a bidirectional behavior: both results cause each other.
Granger causality has certain limitations, since it is not true causality. For example, if both X and Y are part of the same process with different time intervals, one of them may not rule out the alternative hypothesis (also called alternative, it offers a different solution to that proposed by the main hypothesis and by the null, that is, the opposite). However, manipulating one of them would not show any change in the other. Simply put, the Granger test was designed to treat pairs of variables, so using three or more can give confusing results.