Platformization without platform data: A latent variable approach

Digital multi-sided platforms intermediate a growing share of household expenditure, yet direct cross-country measurement of platformization remains infeasible owing to the absence of publicly available data. This paper treats platformization — defined as the ratio of household consumption expenditure intermediated by platforms to total consumer spending — as a latent variable and estimates it using a Multiple Indicators Multiple Causes (MIMIC) model for a panel of 86 countries over 2000–2023, drawing on the World Bank’s World Development Indicators. The selection of causal and reflective variables is grounded in the Aghion–Howitt endogenous growth model, operationalizing the creative destruction mechanism in the context of platform economics. The methodological contri­bution consists in applying a Mundlak decomposition within the MIMIC specification, separating short-run (within-country) and long-run (between-country) determinants of platformization while preserving the random-effects structure required for latent variable identification. Bootstrap analysis and leave-one-country-out procedures identify a robust core of determinants: financial depth, broadband access, regulatory quality­, urbanization, and R&D expenditure (the latter exhibiting a negative between-effect interpreted as a crowding-out effect: countries with lower average R&D intensity exhibit higher platformi­zation because they are recipients of platform technologies originating in a small number of R&D-intensive exporting economies). Country-level estimates reveal conditional β -convergence alongside persistent absolute gaps among income groups, consistent with technology diffusion under institutional heterogeneity. The resulting estimates are benchmarked against independent point estimates from the literature and can serve as a basis for cross-country comparison of platformization levels and assessment of long-run regulatory effects.