Which Green Technology to Subsidize? Evidence from Electric Vehicles in South Korea

The Substitution Trap in Green Subsidies

Governments worldwide are pouring billions into transitioning transportation from internal combustion engines (ICEVs) to cleaner alternatives. The standard playbook is simple. Identify the cleanest technology—usually Battery Electric Vehicles (BEVs)—and subsidize them aggressively to drive adoption. But does targeting the "cleanest" tech actually maximize the reduction in greenhouse gas (GHG) emissions?

The core of the problem lies in a potential mismatch between absolute cleanliness and market impact. Even though electric cars (BEVs) are cleaner than hybrids (HEVs), giving money to buy hybrids might actually help the planet more. This is because hybrids are easier for people to switch to. This helps get more people out of gas-powered cars. Currently, BEV subsidies only work well if the electricity used to charge them comes from very clean sources.

The dilemma of intermediate technologies

The authors of this paper investigate a specific tension. Can an intermediate technology, which is objectively "dirtier" than the cleanest available option, actually be more effective at reducing total emissions? The question is not just about how much carbon one car saves. It is about how much carbon is prevented by moving a consumer away from a high-emission vehicle.

If a subsidy for a BEV primarily causes a consumer to switch from a slightly more efficient gasoline car to a BEV, the net gain is modest. However, if a subsidy for a Hybrid Electric Vehicle (HEV) causes a massive wave of consumers to abandon traditional internal combustion engines, the aggregate emission reduction could be much higher. The researchers aim to quantify this tradeoff by modeling the South Korean passenger vehicle market. In this market, BEV subsidies are heavy but HEV subsidies have largely vanished.

Cracks in the "Cleanest is Best" orthodoxy

Common wisdom suggests that policy should always track the frontier of technological cleanliness. The assumption is that by incentivizing the lowest-emission vehicles, you move the entire fleet toward the theoretical minimum. However, this ignores the reality of consumer substitution patterns.

Previous approaches often failed to account for the fact that BEVs and HEVs compete against the vast incumbent fleet of ICEVs. If the "cleanest" technology is too expensive or inconvenient, it may only capture users who were already planning to switch. This can cannibalize other low-emission upgrades rather than displacing high-emission ones. Furthermore, many models overlook the lifecycle emissions of the electricity grid itself. As shown in, a vehicle's total footprint includes both the manufacturing "vehicle-cycle" and the operational "fuel-cycle" (emissions from producing and consuming fuel).

If the grid is carbon-intensive, the perceived advantage of a BEV begins to erode.

Modeling the substitution mechanics

To test this, the authors implement a sophisticated Random Coefficients (RC) logit demand system. This is a statistical model used to estimate how consumers choose between different products. It maps how consumers actually behave using product-province-year data from 2012 to 2023. They incorporate "micro-moments"—integrating consumer survey data and millions of vehicle inspection records—to account for mileage heterogeneity (differences in how much people drive). This is critical. High-mileage drivers are more sensitive to fuel costs, making them the most impactful targets for emission reductions.

The investigation proceeds in two stages. First, they build a Bertrand price-setting equilibrium model. This simulates how manufacturers respond to subsidies. It ensures the model accounts for "pass-through"—the degree to which a subsidy actually lowers the price for the consumer versus being kept by the manufacturer. Second, they run a counterfactual. They ask what happens if they take the existing South Korean BEV subsidy budget and reallocate it uniformly to HEVs. They use the GREET model to ensure their lifecycle GHG calculations account for everything from battery production to the carbon intensity of the local power grid.

Why hybrids win the math

The results are striking. The authors find that while BEVs are technically cleaner, reallocating the existing BEV subsidy budget to HEVs would reduce total GHG emissions by an additional 47%.

The reason is found in the diversion ratios (the fraction of sales moved from one product to another). The study demonstrates that HEVs are perceived as much closer substitutes for traditional gasoline and diesel vehicles than BEVs are. When the price of an HEV drops, it pulls a massive volume of sales away from ICEVs. In contrast, BEV subsidies tend to trigger "within-category" substitution. This happens when consumers simply switch from one type of low-emission vehicle to another, leaving the high-emission incumbent untouched.

The effectiveness of the BEV strategy is also tethered to the grid. As shown in, the emission savings from BEV subsidies only overtake HEV subsidies once the electricity generation mix becomes sufficiently decarbonized.

For South Korea, the authors calculate that the grid would need to become approximately 45% cleaner. This would mean reaching levels seen in Portugal. Only then would a BEV-focused policy become more effective than an HEV-focused one.

Shifting the policy target

This research suggests that the "cleanest technology" is a moving target. It depends heavily on the surrounding infrastructure. If a country's electricity grid is still heavily reliant on fossil fuels, subsidizing hybrids may be the more mathematically sound path to rapid decarbonization.

There are two immediate implications for policy engineers. First, the "greenness" of a subsidy is not a static property of the vehicle. It is a function of the national energy mix. Policymakers cannot decouple vehicle incentives from grid decarbonization targets without risking inefficiency. Second, if the goal is pure emission abatement, the "substitution effect" is arguably more important than the "absolute effect." The substitution effect measures how much you displace the dirty incumbent.

However, the authors note a caveat. Their model abstracts from industrial policy (government efforts to support domestic industries). Many nations likely subsidize BEVs to build a domestic high-tech manufacturing base. A logical follow-up would be to model the long-term "learning-by-doing" effects. This would ask if the higher initial volume of BEVs eventually drives down battery costs enough to make the substitution effect irrelevant in the future.

Figures from the paper