What this research found
Which lever moves electric two-wheeler sales in India further, purchase subsidies or charging infrastructure? A panel of 14 states covering 2019 to 2024 — generated synthetically for the exercise — was fitted with fixed-effects and random-effects models, and charging station density emerged with an elasticity 1.76 times that of subsidy amount, 0.4919 against 0.2793. A separate Monte Carlo simulation of ownership costs put five-year savings for an electric two-wheeler at 48.7% in Tier 1 cities and 37.0% in Tier 2.
- In the preferred random-effects specification, charging station density carries an elasticity of 0.4919 against 0.2793 for subsidy amount, a ratio of 1.76. Density is also the more statistically reliable term, at p = 0.0021 against p = 0.058 for subsidies.
- The ordering survives in the fixed-effects specification, which accounts for 96.5% of variance: each additional unit of charging density is associated with roughly 7,440 more units of sales (p < 0.001), against a subsidy coefficient of 0.5616 (p = 0.009).
- Petrol price is the unstable regressor. It is strongly positive in the pooled model (528.71, p < 0.001) but turns negative and insignificant once state fixed effects absorb cross-state differences (-52.38, p = 0.692).
- The ownership-cost simulation, run over 10,000 iterations and a five-year horizon, puts electric two-wheeler total cost at ₹175,037 against ₹341,218 for petrol in Tier 1 cities, a saving of ₹166,182 or 48.7%. In Tier 2 cities the comparison is ₹173,092 against ₹274,605, a 37.0% saving, and modelled noise falls by about 31 dB in both tiers.
- Congestion widens the gap. Annual idling cost savings run to ₹17,788 in Tier 1 cities against ₹7,930 in Tier 2, and per-kilometre running cost is ₹4.23 for electric against ₹8.08 for petrol in Tier 1.
How it was done
A panel of 14 Indian states spanning 2019 to 2024, 84 observations in all, was generated synthetically with electric two-wheeler sales as the dependent variable and subsidy amount, charging station density, GDP per capita, petrol price and urbanisation rate as regressors. Variance inflation factors all fell below 10, and both entity fixed-effects and pooled random-effects models were estimated, with a Hausman test pointing to random effects. Elasticities were then computed to put the subsidy and infrastructure channels on a common scale. Separately, a 10,000-iteration Monte Carlo simulation compared five-year ownership costs, per-kilometre running costs, idling waste and noise for electric against petrol two-wheelers in Tier 1 and Tier 2 cities. The work was written up as a 28-page report with five figures and more than 55 references.
Data sources
- Synthetic state panel — 14 Indian states, 2019–2024, 84 observations covering EV two-wheeler sales, subsidy amount, charging station density, GDP per capita, petrol price and urbanisation rate
- ICCT 2024 analysis of India's FAME II incentive scheme
- IISD 2024 SAVi sustainable asset valuation
- BloombergNEF 2025 battery cost data
- WHO 2018 environmental noise guidelines
- More than 55 references spanning incentive policy, charging infrastructure, ownership economics and panel-data methods
Limitations
The panel is synthetic rather than observed, so the coefficients describe the structure of the generated data and cannot be read as measurements of how Indian states actually responded to policy. The subsidy coefficient also falls short of significance at the 0.05 level in the preferred specification, and petrol price switches sign between specifications.
Figures from this analysis
How this research was produced
K-Dense Web planned and ran this transportation investigation end to end — gathering the sources, carrying out the analysis, producing the figures, and drafting the report. The full session transcript, including every intermediate step, is available to view.
