Financial system stress: From empirical validity to theoretical foundations

A review of financial system's stress measure (Gramlich et al., 2010; Oet et al., 2011) reveals not only the absence of theory of financial stress, but also the absence of search for theory. Previous research has proceeded by accreting intuitive choices of variables to serve as components of stress in financial system or specific markets. In addition, the composition of the financial system itself has not been thoroughly tested by previous studies. Thus, parsing out of the financial system into latent factors that may serve as internally reliable and valid constructs has been left largely to intuition and chance. A valid understanding of the latent factors that form the financial system stress is vital. Yet, the problem is vicious as the dynamics of agent interaction are difficult to trace: the agents exhibit adaptive behavior, and the financial system evolves in response to change. This study conducts an integrative systematic review and empirical analysis of financial stress in an evolving financial system to construct a parsimonious set of causal factors that drive the financial system dynamics. The empirical analysis parses out principal axis factors utilizing longitudinal exploratory factor analysis. The resulting factors' correlational and Granger causality structures are applied via confirmatory factor analysis to test the hypotheses of conditional process analysis on the latent stress factors. Our analysis of empirical validity of current stress measures highlights a number of serious problems with the current a priori stress construction. The empirical validity leads us to posit a new theoretical foundation for a deeper understanding of financial stress-one that can adapt to a change in financial system structure and be useful across financial system of diverse architectures (market-based vs. intermediate) and development (emerging or developed).