Dynamic Electric-Vehicle Adoption in India: A Genuinely Coupled Bass–Agent-Based Simulation Framework with Dynamic PESTL Integration

Electric-vehicle (EV) diffusion in large, heterogeneous, and price-sensitive markets such as India cannot be adequately represented by either aggregate diffusion curves or agent-based models (ABMs) in isolation: the former cannot express heterogeneous buyers or shifting policy, and the latter are rarely anchored to a calibrated national diffusion trajectory. This paper develops a dynamic mathematical model of EV adoption among Indian passenger-car buyers that couples a Bass diffusion model at the macro level with an ABM of heterogeneous consumers at the micro level, with both layers driven by a time-varying PESTL (political, economic, social, technological, legal/environmental) representation of the market environment. Unlike earlier hybrid proposals in which the two layers are described together but solved independently, this framework implements a genuine two-way coupling: in every simulated year, 5000 agents positioned on a small-world social network make discrete vehicle-choice decisions based on deterministic utility; the resulting peer-adoption share and the normalised PESTL drivers jointly determine the time-varying innovation coefficient p(t) and imitation coefficient q(t); a bounded Bass equation with a time-varying saturation ceiling then advances the national EV share of new-car sales; and that macro share re-enters the agents’ social-influence term in the following period, closing the loop. The model is anchored on current Indian parameters—a passenger-car EV penetration of approximately 4% in 2025, volumetric battery-pack cost near US$108/kWh, and around 29,000 public charging points—and run under baseline, aggressive-support, and stalled-support scenarios to 2050, together with a 3 × 3 scenario matrix cross-combining three PESTL environments with three agent-population profiles. For the illustrative calibration, the baseline scenario yields approximately 29.5% EV sales share by 2030—narrowly below the Government of India’s 30% target for passenger cars—and approximately 30.4% under the aggressive-support scenario. A formal two-way ANOVA-based variance decomposition, with 10 independent random-seed replications per scenario cell (90 observations in total), indicates that heterogeneous agent characteristics account for 81.3% of the variance in 2030 outcomes, substantially exceeding the 16.8% attributable to the PESTL policy environment, with the remaining 1.9% reflecting interaction and stochastic noise. One-at-a-time sensitivity analysis identifies the peer-influence weight as the parameter with the largest overall effect on 2030 penetration, with the purchase-subsidy multiplier remaining the most effective lever among instruments directly controllable by policy. The emergent Bass coefficients fall within established meta-analytical ranges, providing internal plausibility validation. The framework is offered as a transparent decision-support and scenario-analysis tool for policymakers and industry planners, not as a point forecast of future EV sales.