Hello everyone,

This question is from a paper that I recently read; Laeven.L, and Levine.R, (2009), "Bank governance, regulation and risk taking", Journal of Financial Economics 93, pp.259-275. In a nut shell, one of the major hypothesis tested by the authors is the impact of bank regulations (original variable of interest) on bank risk (original dependent variable). The associated model (I refer this as the original model), simply is as follows;



where Y represents risk of bank b in country c and X represents country level measures of *bank regulations.

In a subsequent section, the authors mention that they allow for the joint determination of bank risk and bank valuations(since bank regulations (original variable of interest) might influence bank risk (dependent variable) by affecting bank valuations;i.e. they control for the endogenous determination of bank risk and bank valuation and test whether an association exists between risk and bank regulations independent of bank valuation.

To test this, authors implement 2SLS regression. In the first stage, they model Bank valuation as a function of all the variables included in the original model (please see above equation) and instrumental variables for bank valuation. In the second stage, bank risk (original dependent variable) is modeled with all the variables included in original model and in addition, with bank valuation.

My question is, according to my understanding, in the first stage of 2SLS regression, the dependent variable should be the endogenous variable included in the original model (I assumed that bank regulations should be the endogenous variable). But could someone please explain how come they have included bank valuation as the endogenous dependent variable in first stage where bank valuation was not even included in the original model?

Any help is much appreciated. Thank You.