Structural Equation Modeling in Archival Capital Markets Research: An Empirical Application to Disclosure and Cost of Capital

Structural equation modeling (SEM), an empirical methodology underutilized in archival research, enables researchers to examine paths linking constructs. SEM consists of two components: a measurement model that generates common factors from observed variables and a path model that links the factors. We discuss SEM's components, estimation, advantages, best practices, and limitations. We illustrate SEM with an application to disclosure research. Unlike some prior research, we find voluntary disclosure quality is negatively associated with cost of capital, both directly and indirectly through information asymmetry, even after controlling for earnings quality's direct and indirect associations with cost of capital. We believe SEM offers fruitful avenues for future research because it allows flexibility in modeling relations guided by theory, enables tests of underlying theoretical mechanisms, provides tools to address measurement error and missing data, and estimates simultaneous equations. SEM may be useful in settings that currently use path analysis or principal component analysis.