Mind the Gap: Identifying Latent Objective and Subjective Multi-Dimensional Indices of Well-Being

Abstract

Within the OECD Better Life Initiative, the Better Life Index (BLI) represents a major attempt to measure well-being and societal progress beyond GDP, following up the recommendations outlined in the Stiglitz-Sen-Fitoussi Commission report. Using a Structural Equation Modeling (SEM) approach, we estimate BLI as a latent construct starting from eleven underlying dimensions of well-being. This method, based on variance-covariance matrices, allow us to study the interrelations and causal-relationships across well-being determinants and across the underlying drivers of well-being. In our analysis we utilize two different comparable OECD datasets for the year 2012, one based on average country-level data reflecting well-being outcomes, the other one on microdata reflecting people's stated preferences on well-being indicators. In order to deal with the idiosyncratic structures of the datasets, we apply two Structural Equation Modeling techniques - bootstrapped SEM and Generalised SEM MIMIC - to estimate the relative weights and rankings of BLI dimensions. We then build an 'objective' BLI measure predicted from the national-level data, whereas a 'subjective' BLI is obtained using the new OECD microdata. Finally, we conclude our analysis comparing the objective and subjective BLI dimensions weights and country rankings and discussing the main policy implications.

Keywords: Structural Equation Modeling, Latent Multidimensional Index, Beyond GDP, SubjectiveWell-Being Index, ObjectiveWell-Being Index, Stated Preference, Generalised SEM MIMIC, GSEM, Bootstrapped SEM, Small Sample Size, Weights

JEL Classification: C43, C83, D12, I31