Christmas cards, Easter bunnies, and Granger-causality

This paper takes a closer look at the conceptual grounds of the notion of causality in Granger's sense. We start with the often jokingly made remark that 'Christmas card sales Granger-cause Christmas'. Then, we extend the example to the more challenging case of chocolate Easter bunny sales and Easter. We show that any references to Granger-causality in these cases are due to the misinterpretation of the concept. Moving further on methodological grounds, we argue that the concept of Granger-causality calls for a multivariate framework of analysis. This is because taking all available relevant information into account is indeed required in Granger's definition of causality. This is also in line with rational behaviour and learning under imperfect and incomplete information. The implications of employing a multivariate framework of analysis is discussed in terms of the additional insights it brings; namely, direct, indirect, and spurious cases of Granger-causality. Finally, we examine the semantics of the definition of causality in Granger's sense.