Value relevance of loan loss provision components and the choice of model specification

This paper investigates the level of paradigm development for estimation procedures used in identifying value relevance of loan loss provision components. The multiplicity of loan loss provision models poses a challenge to the researcher to select a specification while understanding the capital market impact of the discretionary and nondiscretionary components. We examine the differences that can emerge in understanding the value relevance of the components of the loan loss provision by use of different loan loss provision estimation models, market valuation models, scale-related deflators, and regression techniques. We develop estimates of the discretionary and non-discretionary components of the loan loss provision using nine different specifications and two scale-related deflators. We test the value relevance of these estimates on the market capitalization using two different valuation models and different regression approaches. Using a sample of banks in India, our empirical results show that the value relevance of the components of loan loss provision is conditioned on the choice of loan loss provision specification, the valuation model, and the regression approach.