# When should we use “internal consistency” test?

From a paper of Gao, Whited, and Zhang (2021), I saw a paparagraph

We deviate from this traditional specification in three ways: we examine only money held by corporations, we include real GDP growth in the regression to control for economic fluctuations over the business cycle, and we use real instead of nominal interest rates. This last choice is for internal consistency with the theoretical framework that follows

I understand the internal consistency from this description

For example, if a respondent expressed agreement with the statements "I like to ride bicycles" and "I've enjoyed riding bicycles in the past", and disagreement with the statement "I hate bicycles", this would be indicative of good internal consistency of the test.

But I am wondering when we should use "internal consistency" and why the authors need to perform the robustness test for it afterwards?

• Firstly, this is a very interesting paper, like all Toni Whited papers!:) The authors adopt a well-known regression in Equation (1) but adjust it slightly (eg use real instead of nominal rates) as this better aligns with their theory. Thus, internal consistency here refers to the empirics matching the theory, within the same paper. As the regression is well-known, some people may be wary of changing it (perhaps they changed it for the results are otherwise not robust?) To alleviate such concerns, the authors offer robustness checks (external consistency with the literature'' if you want to). Aug 24, 2021 at 13:00
• They have real model. Hence inflation/nominal variables such as nominal interest rates are essentially undefined and cannot be used in the theory part. This is likely why using real rates is "internally consistent" with the theory.
– fes
Aug 24, 2021 at 13:34
• Your second quotation is about "internal consistency" in questionnaires and surveys, it is not the meaning relevant here. Rather it is about "consistency with the theoretical framework". Aug 24, 2021 at 16:02