Counterfactual Approach for Sequential Mediators in Mediation Analysis: A Suitable Methodological Approach for the Banking and Finance Literature?

Mediation analysis can be defined as a family of statistical methods whose aim is to extract information about the causal/associational mechanisms through which the aforementioned explanatory variable X affects the outcome variable Y. More specifically, this chapter focuses on sequential mediators. The simplest sequential mediator case can be described as follows: a primary explanatory variable X affects an output variable Y both directly (direct effect) and indirectly (indirect effect). The indirect effect is due to two (or more than two) mediators M1 and M2, where M1 affects M2. In this chapter, I show that such statistical methodology is suitable and effective for various empirical settings in the banking and finance literature to identify causal relationships between financial statements, corporate governance, and a wide range of other variables. This methodology has already been widely employed in the literature to analyze the effects of high-frequency trading, the determinants of the financial performance of financial intermediaries, and customer propensity to use electronic banking services. However, several other topics may be investigated by means of this methodology. For instance, the growing literature on the determinants of bank risk-taking would enormously benefit from the use of mediation and sequential mediation analysis, to offer a more comprehensive view of the relationship between bank risk-taking and a wide range of potential explanatory, moderating and mediating variables such as profitability, capitalization, business model, funding structure, corporate governance mechanisms, and institutional factors.