A Choice-Based Diffusion Model for Multi-Generation and Multi-Country Data

Abstract This study proposes a model that enables us to investigate the multi-generation and the multi-country diffusion process simultaneously. Many former studies focus on only one of the dimensions since it is difficult to integrate both dimensions at the same time. Our proposed framework can explain both diffusion processes by capturing the common trend of multi-generation diffusion process and the country-specific heterogeneity. We develop the choice-based diffusion model by decomposing the choice probability of adoption into two components; the first component explains the individual country heterogeneity depending on the country-based variables while the second component captures the common trend of multi-generation diffusion process with the generation-based variables. We apply the model to 3G and 4G connections across 25 countries. Empirical result shows that it is not easy to use individual country level model for most countries due to the lack of data points. Our pooled model outperforms several individual country models according to the fitting and forecasting measures. We find that each country's market competitiveness and the market price affect the rate of diffusion and show that random effects of 3G and 4G are positively correlated. This framework provides the fine prediction capability even with few data points and valuable information for formulating policies on a new generation.