Model building and estimation strategies for implementing the Balanced Scorecard in Health sector

Over last two decades, the concept of the Balanced Scorecard has had broad application by the health sector internationally, including Hospitals systems and national healthcare systems or organizations. However, the lack of literature on causal-effect relationships between different types of dimensions and indicators poses difficulty in conceptualising and implementing a quality evaluation system based on Balanced Scorecard. Methodologically, the most natural context for Balanced Scorecard conceptualization and estimation deals with Structural Equation Models with latent variables. Partial Least Squares Path Modelling has found increased applications, thanks to its ability to handle complex models. However, the lack of a global optimization criterion makes it difficult to evaluate this procedure. The aim of this article is to propose a methodological conceptualization of the Balanced Scorecard in a new context, as the Health sector, using a suitable statistical approach to estimate causal relationships among specified latent dimensions, together with a model building strategy, a necessary step when expert knowledge is too weak to build a robust and well suited model. Specifically, within the Structural Equation Models framework a two-step model building strategy is presented; the first step build the measurement models based on a clustering (around latent variables) technique and the second step build the structural model based on partial correlations and a procedure that selects the best model in terms of predictive power, measured by the mean of the R ² for the endogenous latent variables. Finally, an application based on administrative archives of Lombardy region (Italy) illustrates the presented methodology.