On the Corporate Governance (CG) of Model Validation (Mv): A Micro-CG Illustration of ROEg, Put-Call Parity (Pcp), and Yule-Simpson Paradox (Ysp) As Archetypal Models of Financial Risk Evaluation and Valuation Subject to Mv

This paper builds on Wurts (2018a,b) in a variety of ways. It (1) introduces a Return-on-Equity variable g (ROEg) equation as an additional archetypal model, as an opinion-based analytical model, (2) introduces two additional bright-line tests (BLT) (to address Yule-Simpson Paradox (YSP) and ROEg as archetypes for empirical and opinion-based models, respectively, to complement the BLT for the Put-Call Parity (PCP) as an archetypal analytical model), (3) adds additional attention-direction tools (e.g., the Five Demonstrations (5D) and the Defining and Simplifying Story and Model (DSSM) frame, both to help resolve the broad validation issue) and a host of named analogies that can be applied more broadly to help identify biased persuasion within the rhetoric of models, (4) elaborates more on (4a) corporate governance (CG) issues regarding model validation (MV), (4b) requirements for theories of model validation (TMV), (4c) the archetypal CGMV story, and (4d) the NEP-WED frame. Specifically, (5) a DSSM frame is introduced to provide language to help understand how a model can fail in perhaps an abstract sense; namely, because it cannot answer the question of interest (QOI) provided by the Defining Story of the issue of concern (IOC), with perhaps (6) a cascade of deterioration with respect to the power of an argument as (6a) the Simplifying Story the model is intended to address has lost too much information (i.e., become too “degree negative”), (6b) leading to a Simplifying Model that cannot really answer the QOI (i.e., the Minimal Model Story is insufficient to even answer the IOC presented by the Simplifying Story), (6c) sometimes leading questionable modelers to add information that did not exist (i.e., hence, “degree positive”) in either the Defining or Simplifying Story, creating an accompanying Maximal Model Story laden with unsupportable “degree positive” assumptions that influence conclusions beyond sensible inference. Together, (7) the 5D and DSSM tools provide frames to both (7a) proactively identify inference concerns to avoid having a model blow-up a company and (7b) reactively identify, as forensic analysis, specifically how a model blew-up a company. Together, they address two directions of a common question: what question does the model actually answer, regardless of what advocates claim it can answer and is answering? Key conclusions are justified: (1) the ROEg model helps illustrate how accounting reporting measurements may be inadequate for measuring economic risks of corporate interest, (2) the PCP model helps identify that arbitrage forces may not be strong enough to justify the use of arbitrage-free pricing models in a corporate finance context, and (3) the YSP model illustrates that regression equation coefficient estimates are often inadequate as isolated performance measures of factor sensitivities. And yet, such conclusions can be generalized toward more-complex models.