Answers are grounded in this paper and cite their sources.
Firm-level diffusion research routinely reads the coefficients of a usage regression as determinants of adoption. This paper starts from a proposition about what such regressions identify. A usage regression identifies which firms are users, the rank ordering of the diffusion literature; it is silent about when a given firm adopts. Only estimates of the inter-firm diffusion process, with firm heterogeneity handled, can inform us about when firms adopt and why. A linked census panel of 19,262 Korean enterprises, 106,397 firm-years and nine digital technologies, 2017–2024, illustrates the point. In usage regressions alliance participation carries odds ratios between 3.5 and 4.1 and patent holding about 2. In adoption hazards the alliance association at initiation is entirely between firms (3.08) and absent within them (1.05), and under firm fixed effects on the opportunity-set margin it reverses (0.79); patent holding, a knowledge-stock proxy, strengthens within firms (2.57). An industry adoption-share proxy for epidemic effects raises the hazard only within firms, technology-type alliances behave like other alliances, complementarity among the technologies is a between-firm pattern, and the intensive margin shows no within-firm response to either covariate. Which firms use and when firms adopt are different questions, and the census answers them differently.
No takes yet. Share an insight, caveat, or question.
Hosung Kim (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: