Bass Re-visited: Quantifying Multi-Finality

Abstract

As forecasting the diffusion of new technologies results in unreliable predictions when only few trend data are available, scholars and practitioners alike tend to favor foresight efforts of a more qualitative nature to gauge foreseeable futures. We argue and demonstrate that it is feasible to reconcile quantitative forecasting models with (qualitative) scenario development efforts by advancing a set of heuristics that build on known diffusion curves. Rather than focusing on the ultimate set of parameter estimates, we explore an exhaustive range of combinations signaling potential outcomes. We demonstrate this approach for Battery Electric Vehicles in Europe and the US. A multi-dimensional search space is developed consisting of a wide range of values for each parameter implied. Resulting outcomes are assessed in terms of how well they explain current observations (2011-2017) by means of a loss function. The loss function allows us to evaluate the presence of different end states (multi-finality), their frequency of occurrence, and the implied time frames. The insights obtained inform foresight exercises in a number of distinctive ways. First, by incorporating multi-finality in the design of the heuristics, it becomes feasible to qualify multiple pathways ex ante and simultaneously. This allows us to assess the likelihood of - and the time horizon implied by - different scenarios. Second, allowing for multiple pathways and end states directs our attention to the antecedents required for different scenarios to unfold, which can inspire back-casting efforts of a more qualitative nature. Finally, applying the grid logic on more fine-grained levels of analysis (e.g. countries, states, or regions) helps us assess the (differential) impact of policies oriented towards stimulating diffusion. As such, quantitative heuristics become a valuable complement to scenario development exercises, rather than an inferior or even neglected substitute.