Sustainable E-learning? A flash in the pan or a lasting change for workers

• The pandemic spiked workers’ e-learning; demand reverted post-COVID reopening. • DID panel (2019-2023) with TAM reveals crisis-driven, time-varying motives. • Service and high-skill staff adopted the fastest but abandoned e-learning quickest. • Manufacturing and low-skill groups showed more minor yet more persistent use. • Results guide policies to convert emergency e-learning into lasting workforce skills. This quantitative, observational, quasi-experimental inquiry analyzes the intertemporal e-learning behaviors of full-time workers in Taiwan using nationally representative, repeated cross-sectional survey data. Methodologically, the analysis links three waves of Taiwan’s Digital Development Survey (2019 = 5,706; 2022 = 6,860; 2023 = 7,116) to a two-period, two-group Difference-in-Differences (DID) estimator and embeds the survey’s perceived-usefulness and ease-of-use items in a temporal-contingency extension of the Technology Acceptance Model (TAM). By integrating the TAM with the DID method, the study reveals a substantial surge in e-learning demand during the peak of the COVID-19 pandemic. However, this demand quickly reverted to pre-pandemic levels after the pandemic, indicating that e-learning primarily functioned as an emergency response rather than instigating a long-term shift in workers' learning behaviors. This pattern holds consistently across workers from various industries and skill levels. The study provides empirical validation for the TAM framework in exogenous shocks, emphasizing the significant roles of perceived usefulness and ease of use in shaping workers' e-learning behaviors. Additionally, it contributes to the literature by offering insights into the heterogeneous demand for e-learning across different skill levels and industries. The findings highlight the short-term volatility of e-learning demand in response to external shocks, extending the theoretical understanding of e-learning behaviors during unforeseen events. These results have important implications for policymakers, suggesting that strategies to promote the long-term normalization of e-learning should address its inherent volatility and focus on fostering sustained engagement beyond crisis-driven adoption.