Hi,
I'm trying to use PLS-SEM (particularly, smartPLS) to model and test measurement & structural relationships involving my endogenous LV (credit performance) & three exogenous LV variables (capacity to pay, character & experience, and business conditions) and their respective MVs that are actually Likert rating scores of individual cases. I understand that the model will yield weights for the internal model variables and path coeffiicients for the external model latent variables. Assuming that the smartPLS calculated weights & coefficients are seen to have sufficient validity and statistical significance, would it make sense to use such weights & coefficients as scoring weights in a credit risk scoring system for prospective bank borrowers? My interest is actually to come with a validated model that can be used for predicting the relative credit worthiness of a loan applicant. Would this be a valid application of PLS-SEM? I'd like to know if the technique has ever been used for this particular purpose. Any reply from anybody would be most appreciated.