A newly published economic analysis of China’s electric vehicle subsidy system finds that tiered subsidy multipliers tied to battery energy density can push automakers toward higher-technology batteries, but the overall effect on the average energy density of newly launched electric vehicle models is limited. The study, led by Hu Yunyi, Wang Feng, and Yin Haitao, examines how the design of electric vehicle purchase subsidies affects product technology selection at the manufacturer level. It uses a profit-maximization framework and counterfactual simulations to evaluate the tiered subsidy adjustment multiplier applied to China’s electric vehicle market from 2017 to 2021.
The central question is whether a subsidy schedule that pays more for electric vehicle models with better battery performance can encourage producers to choose more advanced technology. The study’s answer is nuanced. Compared with a uniform subsidy multiplier, the tiered design does encourage more new electric vehicle models to use batteries with higher energy density, especially around the 160 Wh/kg threshold. However, because automakers tend to cluster near subsidy thresholds rather than push far beyond them, the policy raises the average energy density of new electric vehicle models by only about 1%. A more technology-sensitive design, such as a quasi-linear subsidy schedule linked to energy density, could produce an average increase of about 10%, according to the counterfactual results.
The findings matter for electric vehicle policy because subsidies are often used not only to expand adoption but also to steer manufacturers toward better products. The study suggests that the architecture of an electric vehicle subsidy matters as much as its total budget. A tiered electric vehicle subsidy may move the market in the intended direction, but its threshold effects can concentrate technology choices at specific battery energy density levels. For electric vehicle regulators seeking both technological progress and fiscal discipline, the results point to the need for careful calibration of subsidy steps, eligibility thresholds, and the relationship between subsidy levels and battery performance.
1. The study examines how tiered electric vehicle subsidies influence battery technology selection
The analysis focuses on China’s pure electric passenger vehicle market and uses battery mass energy density, measured in Wh/kg, as the key technology indicator. Battery energy density reflects how much energy a battery pack can store per kilogram of weight. In electric vehicle design, higher energy density can support longer driving range, lighter battery packs, or both. The study asks whether the tiered subsidy adjustment multiplier, which raises the subsidy amount for electric vehicle models with higher battery energy density, changes the battery technology choices made by automakers.
The policy context is important. China’s electric vehicle purchase subsidy system combined a range-based subsidy standard with a battery energy density adjustment multiplier. The range-based component was intended to encourage electric vehicle models with longer driving range. The energy density multiplier was added to discourage automakers from pursuing range gains only by reducing vehicle weight or motor power while neglecting battery technology. Under the tiered design, electric vehicle models with higher battery energy density could receive a larger subsidy multiplier, creating a direct financial incentive for automakers to select more advanced batteries.
The study examines this mechanism through the lens of firm decisions. Instead of treating electric vehicle technology as a demand-side outcome alone, it models the automaker’s choice of battery technology as part of a profit-maximization problem. The automaker selects among available battery types, considering the subsidy received, the cost of the battery, the effect on vehicle characteristics, and the resulting impact on model profit. This supply-side perspective is central to the analysis because the electric vehicle subsidy schedule is designed to influence what manufacturers produce, not only what consumers buy.
2. Policy background: electric vehicle subsidies and battery energy density thresholds
China began nationwide promotion of electric vehicles with purchase subsidies that varied by driving range. In 2016, for example, pure electric passenger vehicle subsidies depended on range categories, with higher subsidies for longer-range models. Such a design was meant to compensate for the higher cost of electric vehicle models with larger batteries and longer range. However, range can be increased through means other than battery technology, including vehicle lightweighting and reduced motor power. That raised concerns that electric vehicle makers might optimize for subsidy eligibility rather than improve the underlying technology.
To address this, the subsidy system introduced a battery system energy density adjustment multiplier. In 2017, pure electric passenger vehicle models were required to have battery system mass energy density of at least 90 Wh/kg, and models above 120 Wh/kg received a multiplier of 1.1. In 2018, the requirements and multipliers became more granular: models in the 105–120 Wh/kg range received a 0.6 multiplier, 120–140 Wh/kg received 1.0, 140–160 Wh/kg received 1.1, and 160 Wh/kg and above received 1.2. In 2019, the minimum requirement rose to 125 Wh/kg, with a 0.8 multiplier for 125–140 Wh/kg, 0.9 for 140–160 Wh/kg, and 1.0 for 160 Wh/kg and above.
These changes created visible thresholds in the electric vehicle market. The study reports that many newly launched electric vehicle models clustered near the subsidy thresholds, especially around 140 Wh/kg and 160 Wh/kg. This pattern suggests that automakers respond strategically to the design of the electric vehicle subsidy. The threshold structure can encourage battery upgrades, but it can also concentrate technology choices at the minimum level needed to qualify for a higher multiplier. The key policy question is whether this threshold effect is strong enough to improve the market’s average technology level and whether alternative designs could do better.
3. Data and market patterns in China’s electric vehicle catalog
The study uses data from China’s recommended model catalog for pure electric passenger vehicles from 2017 to 2021. The catalog contains key parameters such as battery energy density, driving range, and battery type. Inclusion in the catalog is a prerequisite for receiving subsidies, so the data cover the main electric vehicle models in the domestic market. The researchers match catalog model codes with vehicle names and parameters from an automotive information website, and they cross-check the data for accuracy. The final analytical sample includes 2,571 unique pure electric passenger vehicle models after removing missing values, duplicates, and some non-passenger vehicle categories.
The data show substantial variation in electric vehicle battery technology. The lowest battery energy density in the sample is 91.1 Wh/kg, while the highest is 206 Wh/kg. The average is 143.22 Wh/kg. Average driving range is 348.43 km, with a minimum of 100 km and a maximum of 1,008 km. Average battery pack energy is 46.26 kWh, and the average manufacturer-guided price is RMB 158,500. The average subsidy in the sample is RMB 16,900, with a maximum of RMB 60,000 for models eligible for the highest support. By battery route, about 72% of models use ternary lithium batteries, 18% use lithium iron phosphate batteries, and 10% use other battery types.
| Year | Model count | Energy density (Wh/kg) | Subsidy (RMB 10,000) | Driving range (km) | Battery energy (kWh) | Price (RMB 10,000) |
|---|---|---|---|---|---|---|
| 2017 | 358 | 117.45 | 1.54 | 223.22 | 32.63 | 16.39 |
| 2018 | 471 | 140.55 | 2.11 | 311.05 | 42.32 | 15.06 |
| 2019 | 713 | 148.14 | 1.86 | 358.07 | 47.17 | 14.46 |
| 2020 | 547 | 149.90 | 1.49 | 391.20 | 50.56 | 16.61 |
| 2021 | 482 | 150.10 | 1.37 | 415.13 | 54.00 | 16.85 |
The annual pattern shows clear technological improvement in the electric vehicle market. Average battery energy density rose from 117.45 Wh/kg in 2017 to 150.10 Wh/kg in 2021. Average driving range increased from 223.22 km to 415.13 km over the same period. Average battery pack energy rose from 32.63 kWh to 54.00 kWh. Average model prices remained relatively stable, moving from RMB 163,900 in 2017 to RMB 168,500 in 2021, with a low of RMB 144,600 in 2019. Average subsidies first rose and then declined, from RMB 15,400 in 2017 to RMB 21,100 in 2018, before falling to RMB 13,700 in 2021.
The distribution of electric vehicle battery technology also reveals threshold clustering. A large share of models used batteries at or just above the subsidy thresholds. In 2018, the 140 Wh/kg band accounted for 53.50% of new electric vehicle models, and in 2019 it accounted for 49.51%. By 2021, the 160 Wh/kg band accounted for 22.97% of models, while the 140 Wh/kg band accounted for 22.15%. The share of models with batteries at 160 Wh/kg and above increased over time, reaching a meaningful portion of the electric vehicle market. The study cautions that these descriptive patterns cannot by themselves prove that the tiered subsidy caused the shift, because battery costs, battery availability, consumer preferences, and driving-range choices also changed.
| Energy density band (Wh/kg) | 2017 | 2018 | 2019 | 2020 | 2021 |
|---|---|---|---|---|---|
| Below 105 | 32.51 | 0.21 | 0.42 | 2.90 | 3.05 |
| 105–119.9 | 15.15 | 2.97 | 0.70 | 4.17 | 7.93 |
| 120–129.9 | 28.93 | 17.20 | 5.87 | 7.26 | 6.10 |
| 130–139.9 | 12.40 | 6.79 | 5.17 | 7.26 | 7.72 |
| 140–149.9 | 9.09 | 53.50 | 49.51 | 31.22 | 22.15 |
| 150–159.9 | 1.93 | 11.89 | 5.45 | 3.27 | 7.52 |
| 160–169.9 | 7.22 | 29.37 | 23.96 | 22.97 | |
| 170–179.9 | 0.21 | 2.38 | 11.25 | 13.01 | |
| 180 and above | 1.12 | 8.71 | 9.55 |
4. A structural model links electric vehicle profit maximization to battery choice
The study builds a structural model in which an electric vehicle model’s expected profit depends on model characteristics, marginal cost, and subsidy. The automaker chooses among available battery types for a given electric vehicle model. Each battery type is defined by energy density and technology route. The researchers divide battery energy density into nine bands and battery technology into three routes, yielding 27 battery types. They also group electric vehicle models by driving-range category and catalog batch. The market share of a battery type within a range category is modeled as a function of subsidy, range, battery characteristics, and battery-type-by-year effects.
The empirical model estimates how battery choices respond to subsidies and other factors. The dependent variable is the log difference between the share of models using a given battery type and the share using a baseline battery type within the same range category and batch. The explanatory variables include the subsidy difference between the battery type and the baseline, the log driving-range difference, battery-type-by-year fixed effects, batch fixed effects, and range fixed effects. Standard errors are clustered at the range-category level in the main specification. The fixed effects absorb time-invariant differences across range categories, common shocks across batches, and changes in the relative profitability of battery types over time.
The baseline battery groups are chosen as widely used battery types in each year. For 2017, the baseline is a 120 Wh/kg ternary lithium battery. For 2018 and 2019, the baseline is a 140 Wh/kg ternary lithium battery. For 2020 and 2021, the baseline is a 160 Wh/kg ternary lithium battery. This approach preserves observations and provides a sensible reference point for each year’s electric vehicle market. The model’s estimates are then used to recover the relative contribution of different battery types to electric vehicle model profit.
5. Empirical estimates show subsidies and range affect electric vehicle model profit
The regression results show that subsidies have a positive and statistically significant effect on the relative profitability of electric vehicle models. In the preferred specification, the subsidy coefficient is 0.888 with a clustered standard error of 0.136. The study interprets this as meaning that, holding other factors constant, an additional RMB 10,000 in subsidy raises an electric vehicle model’s profit by about RMB 8,880. The coefficient on log driving range is 1.484 with a clustered standard error of 0.413. The study’s interpretation is that a 1% increase in driving range is associated with about RMB 15,000 higher model profit.
| Variable | Coefficient | Clustered standard error | Significance |
|---|---|---|---|
| Subsidy (RMB 10,000) | 0.888 | 0.136 | 1% |
| Log driving range | 1.484 | 0.413 | 5% |
| Sample size | 540 | ||
| Adjusted R-squared | 0.579 | ||
| Battery type by year fixed effects | Yes | ||
| Batch fixed effects | Yes | ||
| Range fixed effects | Yes | ||
| Clustering level | Range category |
The study runs several robustness checks. It changes the baseline battery type to a 140 Wh/kg ternary lithium battery. It trims observations with market shares below the 5th percentile or above the 95th percentile. It restricts the sample to the two mainstream battery routes, ternary lithium and lithium iron phosphate. It adds range-by-year fixed effects. It clusters standard errors at both the batch and range-category levels. It replaces the log of driving range with the absolute value of driving range. The estimated subsidy effects remain positive and significant across these specifications, supporting the main conclusion that electric vehicle subsidies affect model profitability and therefore battery selection.
The researchers also discuss endogeneity. They argue that the battery-type-by-year fixed effects absorb much of the variation that could confound the subsidy estimate. They note that finding a credible instrument for the battery-type choice is difficult and may not add explanatory power. They also point out that battery prices are often determined in later production and sales stages, so battery suppliers are unlikely to set prices based on the number of electric vehicle models choosing a battery type during the design stage. At the same time, they acknowledge that if battery prices differ across batches within a year, measurement error could lead to an overstatement of the estimated effect of battery type on model profit.
6. Battery technology contributions to electric vehicle model profit shift over time
The study estimates the relative contribution of different battery energy density levels to electric vehicle model profit. In 2017, lower energy density batteries had a stronger positive contribution to model profit, while the highest energy density batteries had a clear disadvantage. For example, the 100 Wh/kg battery had a relative contribution of RMB 1,230, while the 150 Wh/kg battery had a relative contribution of negative RMB 10,710. This pattern suggests that, in the early period, low-density batteries were more cost-effective for electric vehicle makers.
Over time, the advantage of low-density batteries weakened, and higher-density batteries became more favorable. In 2018 and 2019, the baseline battery types contributed more to model profit than most alternatives. By 2020 and 2021, higher energy density batteries began to show positive relative contributions. In 2021, the 130 Wh/kg battery had a relative contribution of RMB 5,100, the 140 Wh/kg battery had RMB 3,190, and the 170 Wh/kg battery had RMB 6,990. The 180 Wh/kg battery had a relative contribution of RMB 2,110. The 150 Wh/kg battery remained negative at negative RMB 1,800. These estimates suggest that the profit contribution of higher energy density electric vehicle batteries increased as battery technology improved, costs changed, and consumer preferences shifted toward higher energy density.
The study does not interpret these estimates as direct measures of battery cost alone. The relative contributions combine the effect of battery characteristics on vehicle pricing and the effect of battery costs on model profitability. The estimates also reflect consumer preferences for battery types and the competitive conditions in the battery market. The researchers use these estimates as inputs for counterfactual simulations of alternative electric vehicle subsidy designs.
7. Counterfactual findings: tiered electric vehicle subsidies versus uniform subsidies
The core counterfactual compares the observed tiered subsidy adjustment multiplier with a uniform subsidy multiplier of 1 for all electric vehicle models, holding the total subsidy budget the same. Under the uniform policy, all battery energy density levels receive the same adjustment multiplier. Under the tiered policy, higher energy density levels receive higher multipliers according to the actual policy schedule. The researchers calculate the probability that an electric vehicle model chooses each battery type under the uniform policy and then infer the number of models that would use each battery type. They compare that simulated distribution with the observed distribution under the tiered policy.
The results show that the tiered electric vehicle subsidy design shifts model choices toward higher battery energy density. Compared with the uniform subsidy, the tiered design reduces the number of models choosing lower energy density batteries. The 100 Wh/kg band falls by 12%, the 110 Wh/kg band falls by 16%, the 130 Wh/kg band falls by 11%, and the 140 Wh/kg band falls by 3%. The 120 Wh/kg band falls by 3%, and the 150 Wh/kg band falls by 3%. At the same time, the number of models choosing higher energy density batteries rises. The 160 Wh/kg band increases by 14%, the 170 Wh/kg band increases by 9%, and the 180 Wh/kg band increases by 9%.
| Energy density band (Wh/kg) | Model count under uniform subsidy | Model count under tiered subsidy | Change (%) |
|---|---|---|---|
| 100 | 58 | 51 | -12 |
| 110 | 66 | 56 | -16 |
| 120 | 266 | 257 | -3 |
| 130 | 195 | 174 | -11 |
| 140 | 909 | 880 | -3 |
| 150 | 155 | 150 | -3 |
| 160 | 418 | 477 | 14 |
| 170 | 133 | 145 | 9 |
| 180 | 97 | 106 | 9 |
The largest positive effect appears at the 160 Wh/kg band, which is the highest tier in the subsidy schedule for part of the sample period. The study explains that automakers have a strong incentive to choose batteries that just reach the highest subsidy multiplier. Choosing batteries above 160 Wh/kg does not increase the subsidy multiplier, so the additional battery cost may outweigh the benefit unless the higher energy density also supports a higher vehicle price or other advantages. As a result, the tiered design encourages electric vehicle models to reach the threshold but does not necessarily push them far beyond it.
The aggregate effect on average battery energy density is correspondingly modest. The study estimates that the tiered subsidy multiplier raises the average energy density of newly launched electric vehicle models by about 1%. This is a positive effect, but it is much smaller than the threshold-level changes might suggest. The researchers note that this estimate may be a lower bound because the model focuses on battery choice within a given driving-range category and does not fully account for the possibility that the tiered subsidy could also encourage automakers to produce electric vehicle models with longer driving range. Longer range could indirectly raise average battery energy density.
8. Technology-route effects: ternary lithium and lithium iron phosphate electric vehicle batteries
The study also examines how the tiered electric vehicle subsidy affects different battery technology routes. Ternary lithium batteries are the most responsive. Under the tiered subsidy, low-energy-density ternary lithium battery choices fall sharply. The 110 Wh/kg ternary lithium band falls by 8 models, or 20%. The 130 Wh/kg band falls by 11 models, or 11%. The 140 Wh/kg band falls by 19 models, or 3%. At the same time, higher-energy-density ternary lithium choices rise. The 160 Wh/kg band rises by 55 models, or 14%. The 170 Wh/kg and 180 Wh/kg bands each rise by 9%.
Lithium iron phosphate batteries are also affected, but the pattern differs. The tiered subsidy reduces the number of electric vehicle models using lithium iron phosphate batteries at or below 130 Wh/kg by more than 12%. For higher energy density lithium iron phosphate batteries, the number of models remains unchanged. The study notes that, in the sample period, there are no lithium iron phosphate battery types at or above 180 Wh/kg. Other battery technology routes are less affected by the tiered subsidy design, although some low-density other-route batteries decline and some high-density other-route batteries increase.
These results matter for the electric vehicle industry because battery chemistry and battery energy density are linked. Ternary lithium batteries can offer higher energy density, but they may be more costly. Lithium iron phosphate batteries may be lower cost and safer, but they often have lower energy density. The tiered subsidy design can therefore influence not only how much energy density the electric vehicle market adopts but also which battery technology routes gain share. The study’s results suggest that the tiered design favored higher-energy-density ternary lithium electric vehicle batteries during the sample period, while low-density lithium iron phosphate batteries lost ground.
9. Alternative electric vehicle subsidy designs could deliver larger energy-density gains
The study simulates two alternative tiered electric vehicle subsidy designs to see whether different schedules could produce stronger technology effects while keeping the total subsidy budget unchanged. The first alternative expands the subsidy gap between tiers. For 2017, batteries at 120 Wh/kg and above receive a multiplier of 1.2. For 2018, batteries at 160 Wh/kg and above receive 1.2, while batteries below 120 Wh/kg receive 0.6. For 2019, batteries at 160 Wh/kg and above receive 1.2, while batteries below 140 Wh/kg receive 0.6. The second alternative uses a quasi-linear schedule in which the subsidy multiplier is linked to the battery energy density range. In each batch, the lowest energy density model receives no subsidy, and the subsidy increases arithmetically with energy density, while the total subsidy amount is held equal to the actual policy.
The counterfactual results show that both alternative designs perform better than the actual tiered schedule in promoting higher battery energy density. The expanded-gap design reduces the number of electric vehicle models choosing batteries below 150 Wh/kg and increases the number choosing 150 Wh/kg and above. The quasi-linear design has an even stronger effect, especially for the highest energy density bands. Under the quasi-linear design, the number of electric vehicle models with 180 Wh/kg batteries rises substantially. The study reports that a subsidy multiplier quasi-linearly related to energy density could raise the average energy density of newly launched electric vehicle models by about 10%, compared with the roughly 1% effect of the actual tiered design.
The policy implication is that the relationship between subsidy and technology should be designed with care. A tiered schedule with discrete thresholds can move electric vehicle technology, but it can also create bunching at the thresholds. A schedule that links subsidy more continuously to battery energy density can avoid some of this bunching and provide stronger incentives for electric vehicle makers to choose technologies above the minimum threshold. However, a quasi-linear schedule may be more complex to administer and may require better information about battery costs and consumer preferences. The study suggests that policymakers should collect technology cost information and run policy simulations before choosing the number of tiers, the size of the gaps between tiers, and the level of support for each technology band.
10. Policy implications for electric vehicle subsidy design
The study’s findings lead to several policy lessons for electric vehicle subsidies and for other industries that use performance-based subsidies. The first lesson is that subsidy standards should match the policy objective. If the main goal is to promote the most advanced electric vehicle technology, the subsidy schedule should provide the largest incentive for the highest technology levels. If the goal is to promote gradual technology upgrading while controlling fiscal costs, a linear or quasi-linear electric vehicle subsidy may be more effective than a schedule with large discrete jumps.
- Electric vehicle subsidy thresholds should be calibrated to the cost differences between battery technologies, so that the subsidy gap is large enough to compensate automakers for choosing higher energy density batteries.
- Electric vehicle subsidy design should account for threshold clustering. If automakers bunch at a specific battery energy density level, the policy may raise the minimum technology standard without generating broad gains in average technology.
- Electric vehicle subsidy schedules should be updated as battery technology advances and consumer preferences change. A schedule that is appropriate in one year may become outdated as battery costs fall and new electric vehicle models enter the market.
- Electric vehicle policy evaluation should use structural models and counterfactual simulations when traditional control groups are unavailable, because electric vehicle subsidies often apply to all manufacturers in the market.
- Electric vehicle subsidy design should consider both direct effects on battery choice and indirect effects on driving range, model design, and consumer adoption, because these channels can influence the overall technology outcome.
The study also emphasizes the importance of policy timing. In fast-moving electric vehicle markets, battery technology and consumer preferences can change quickly. A subsidy schedule that is fixed for too long may become misaligned with the market. Regular evaluation, technology forecasting, and feedback from industry can help policymakers adjust electric vehicle subsidies in a timely way. The researchers argue that this is especially important for emerging industries where technological progress is rapid and the cost of policy error can be high.
11. Research contribution and limits in the electric vehicle policy debate
The study contributes to the electric vehicle policy literature by providing supply-side evidence on how tiered subsidies affect product technology selection. Much of the existing research on electric vehicle subsidies focuses on consumer demand, including purchase intentions, adoption rates, and the effects of subsidy phase-outs. This study shifts attention to manufacturers and shows that electric vehicle subsidy design can influence the battery technology built into new models. It also offers a structural framework that can be used when a policy applies to an entire market and a clean control group is unavailable.
The researchers acknowledge several limitations. The analysis focuses on the second stage of the automaker’s decision, namely battery choice after driving range has been determined. It does not fully model how the subsidy affects the choice of driving range itself. The data do not allow direct measurement of battery costs or the precise price premium associated with battery characteristics, so the empirical model combines cost and price effects into battery-type-by-year fixed effects. If battery prices vary across batches within a year, the estimated profit contribution of battery types may be overstated. The study also notes that its estimate of the average energy density effect is likely a lower bound because it does not capture the indirect effect of the subsidy on longer-range electric vehicle models.
Despite these limits, the study provides a clear message for the electric vehicle policy debate. Tiered electric vehicle subsidies can influence technology choices, but the design details determine whether the policy produces broad technology gains or simply shifts the market toward subsidy thresholds. The difference between a 1% average energy density gain under the actual tiered design and a roughly 10% gain under a quasi-linear design is substantial. For electric vehicle regulators, the results suggest that the choice between discrete tiers and continuous incentives is not a minor administrative detail but a core element of policy effectiveness.
12. Conclusion: electric vehicle subsidy design must move beyond threshold compliance
China’s experience with tiered electric vehicle subsidies shows both the promise and the limits of performance-based industrial policy. The policy succeeded in encouraging more new electric vehicle models to adopt higher battery energy density, and it reduced the number of models using lower-density batteries. The 160 Wh/kg battery band gained more than 14% more models under the tiered design than under a uniform subsidy. These are meaningful shifts in the electric vehicle market and demonstrate that manufacturers respond to financial incentives tied to product technology.
Yet the policy also produced threshold clustering. Many electric vehicle models landed near the subsidy cutoff rather than pushing to the highest available technology. As a result, the average energy density of new electric vehicle models rose by only about 1%. The study’s counterfactual analysis suggests that a quasi-linear electric vehicle subsidy linked more continuously to battery energy density could raise average energy density by about 10%. That gap highlights the importance of subsidy architecture for electric vehicle technology progress.
The broader lesson is that electric vehicle subsidies should be evaluated not only by how much they increase sales but also by how they shape product design. A well-designed electric vehicle subsidy can encourage automakers to adopt better batteries, improve driving range, and reduce costs. A poorly calibrated schedule can create bunching, distort technology choices, and waste fiscal resources. As electric vehicle markets mature and battery technology continues to evolve, policymakers will need to adjust subsidy designs with greater precision. The study’s framework offers a way to simulate alternative electric vehicle subsidy schedules before implementation and to assess whether a policy is likely to deliver real technology gains or only threshold compliance.