Tax coal or subsidize clean capacity? A differential game of incumbent utilities’ energy transition

Electricity decarbonization requires both replacement investment and decisions about the remaining life of coal assets. This paper develops a Stackelberg differential-game benchmark with an incumbent utility that builds clean capacity and retires coal, and an independent clean developer. Binding organizational constraints make construction and closure substitutes; binding finance can instead make them complements when retirement releases net funds. Annual numerical experiments compare five policy classes using a common post-window continuation rule, incumbent-value limits and explicit capacity-gap ceilings. In the normalized China-scale baseline, the highest-welfare unconditional candidate found within the tested policy class coincides with clean support alone. Retirement-linked support combined with an early coal charge raises welfare by 1.07% relative to that benchmark and induces 0.0317 TW of accelerated retirement without an aggregate capacity gap. Freezing the same realized support and charge paths removes almost all accelerated retirement, identifying a behavioral contribution from the link. The advantage is conditional: it disappears under zero continuation values, and declining clean-capacity credits can reverse fixed-policy rankings. Wider support bounds also change the preferred instrument mix. The results identify how shared adjustment resources and dependable replacement capacity condition policy choice; the numerical rates are unestimated policy experiments for an existing portfolio, rather than national forecasts.