Introduction and Research Motivation
As the global community intensifies efforts to combat climate change and accelerate the energy transition, the new energy vehicle (NEV) industry has emerged as a central driver of low-carbon transformation in the transportation sector. China, leveraging its policy advantages and market scale, has become a global leader in NEV production and sales. In 2025, China’s NEV production and sales reached 16.626 million and 16.49 million units, respectively, ranking first globally for eleven consecutive years. However, the rapid expansion of this industry has inevitably led to a large-scale retirement of power batteries. According to industry forecasts, by 2030, China will accumulate more than 3 million tons of retired power batteries. This looming “retirement wave” pushes the recycling of EV battery packs from an emerging issue to a critical bottleneck for sustainable industrial development.
The proper treatment of retired EV battery packs is not merely an industrial supporting issue; it is intimately linked to environmental protection and the sustainable utilization of scarce resources. EV battery packs contain complex chemical systems, including electrolytes and cathode materials with heavy metals such as cobalt, nickel, manganese, and lithium. If retired batteries flow into informal recycling channels and are subjected to crude dismantling or random disposal, hazardous substances can seep into soil and groundwater, causing persistent and irreversible ecological damage. From a resource perspective, lithium, cobalt, and nickel are strategic metals contested globally in the energy transition. Efficient recycling of EV battery packs can significantly reduce dependence on primary mineral imports and enhance resource security. The recycling industry for power batteries also holds enormous economic potential, with the market scale expected to reach hundreds of billions of yuan by 2030.
The Chinese government has formulated a policy framework for the recycling of EV battery packs. In 2018, the “Interim Measures for the Management of Recycling of Traction Batteries for New Energy Vehicles” established the extended producer responsibility (EPR) system, clarifying that automobile manufacturers bear the primary responsibility for battery recycling. In 2021, the “Measures for the Management of Cascade Utilization of Traction Batteries” further encouraged the model of echelon utilization to maximize the full-lifecycle value of batteries. Yet, despite these policy efforts, the current recycling system suffers from low efficiency, structural imbalances, and trust deficits. The coexistence of formal and informal recyclers leads to a “race to the bottom” in which informal, low-cost operators offer higher collection prices, diverting retired EV battery packs away from compliant channels. Information asymmetry across the full lifecycle impedes credible assessment of battery health and residual value. The lack of coordination among supply chain stakeholders elevates transaction costs and undermines collaborative recycling. These intertwined challenges require a dual-driven pathway: sound policy mechanisms and technological empowerment through blockchain.
This dissertation aims to address these issues by investigating the optimization of EV battery pack recycling decisions under the dual influence of government reward-punishment mechanisms and blockchain technology. First, I construct a tripartite evolutionary game model involving the government, NEV manufacturers, and consumers to compare static versus dynamic reward-punishment mechanisms. Second, I develop a three-tier closed-loop supply chain model including battery manufacturers, NEV manufacturers, and consumers, and systematically compare four recycling modes (battery-manufacturer-led, NEV-manufacturer-led, hybrid, and alliance) under both scenarios with and without blockchain empowerment. Through analytical derivations and numerical simulations, I aim to provide theoretical guidance for government policy design and enterprise strategic choices.
Theoretical Foundations and Literature Review
The theoretical foundation of this study integrates closed-loop supply chain (CLSC) theory, game theory, government reward-punishment policy, and blockchain technology. CLSC theory extends the traditional linear supply chain by incorporating reverse flows, enabling resource circulation and value regeneration. Game theory, particularly evolutionary game theory and Stackelberg games, provides tools to analyze multi-agent strategic interactions. Government reward-punishment mechanisms are rooted in environmental economics and public policy, aiming to correct market failures through economic incentives and disincentives. Blockchain technology, characterized by decentralization, immutability, and traceability, offers a trust-building infrastructure for transparent and efficient recycling systems.
The existing literature has explored government policies driving power battery recycling from multiple perspectives. Studies have examined various subsidy schemes, carbon trading policies, deposit-refund systems, and reward-punishment mechanisms. For instance, Lou et al. compared volume-based versus capacity-based subsidies and found that volume-based subsidies better improve recycling rates. Evolutionary game models have been applied to analyze the dynamic interactions among governments, manufacturers, recyclers, and consumers. Research by Wang et al. revealed that a single subsidy or regulatory measure is insufficient to achieve the ideal steady state, necessitating combined or phased strategies. In the field of recycling mode optimization, scholars have compared different collection models such as manufacturer-led, retailer-led, third-party-led, and hybrid models. The consensus is that alliance or coalition models often maximize total supply chain profit, while hybrid models improve recycling rates through competition. Blockchain technology in power battery recycling has recently gained attention. Studies have shown that blockchain can reduce information asymmetry, enhance traceability, and increase consumer trust, thereby improving the efficiency and profitability of recycling systems. However, most studies focus on a single recycling scenario or treat blockchain as an exogenous tool, failing to provide a unified comparison across different recycling modes and to quantify the moderating role of consumer trust in blockchain.
Therefore, this dissertation contributes to the literature in two ways: (1) by designing a dynamic reward-punishment mechanism based on the government’s supervision probability and the manufacturer’s recycling probability, and comparing it with the static mechanism within an evolutionary game framework; and (2) by embedding blockchain technology as an endogenous decision variable in a closed-loop supply chain model and systematically comparing four recycling modes under both with- and without-blockchain scenarios, while quantifying the effect of consumer trust in blockchain.
Evolutionary Game Analysis of Power Battery Recycling under Static and Dynamic Reward-Punishment Mechanisms
Problem Description and Assumptions
I consider three bounded rational stakeholders: the government, the NEV manufacturer, and the consumer. The NEV manufacturer chooses either to actively recycle EV battery packs (strategy A) or to passively recycle (strategy P). The consumer chooses either to return the retired battery through formal channels (strategy F) or through informal channels (strategy I). The government chooses either to actively supervise (strategy S) or to passively supervise (strategy N). The parameters used in the evolutionary game model are summarized in Table 1.
| Symbol | Description |
|---|---|
| $C_m$ | Manufacturer’s active recycling cost |
| $I_m$ | Manufacturer’s implicit benefit from active recycling (e.g., brand image, social responsibility) |
| $I_{mc}$ | Maximum benefit from full utilization of retired EV battery pack |
| $\alpha_1, \alpha_2$ | Recycling technology levels under active and passive strategies, with $\alpha_1 > \alpha_2$ |
| $I_{1c}, I_{2c}$ | Consumer’s benefit from formal and informal channel transactions |
| $C_{1c}, C_{2c}$ | Consumer’s cost from formal and informal channel transactions |
| $\mu$ | Consumer’s environmental preference coefficient |
| $L$ | Maximum environmental benefit for the consumer from formal recycling |
| $C_g$ | Government’s active supervision cost |
| $R_m$ | Government’s reward to the manufacturer for active recycling |
| $F_m$ | Government’s penalty to the manufacturer for passive recycling |
| $\beta$ | Reduction ratio of consumer’s informal channel benefit due to government crackdown |
| $\theta$ | Share of the government reward passed by the manufacturer to consumers |
| $R_e$ | Environmental benefit to the government when the manufacturer actively recycles and the consumer uses formal channels |
| $C_{1e}, C_{2e}$ | Environmental governance costs when recycling is suboptimal |
The payoff matrix is shown in Table 2, where $x$ is the probability that the manufacturer chooses active recycling, $y$ is the probability that the consumer chooses formal channels, and $z$ is the probability that the government chooses active supervision.
| Manufacturer | Consumer | Government active ($z$) | Government passive ($1-z$) |
|---|---|---|---|
| Active ($x$) | Formal ($y$) |
$$R_m + I_m + \alpha_1 I_{mc} – C_m – \theta R_m$$ $$I_{1c} + \mu L + \theta R_m – C_{1c}$$ $$R_e – C_g – R_m$$ |
$$I_m + \alpha_1 I_{mc} – C_m$$ $$I_{1c} + \mu L – C_{1c}$$ $$R_e$$ |
| Informal ($1-y$) |
$$R_m + I_m – C_m – C_t – C_r – C_i$$ $$I_{2c} – C_{2c} – \beta I_{2c}$$ $$-C_g – C_{2e} + R_m$$ |
$$I_m – C_m$$ $$I_{2c} – C_{2c}$$ $$-C_{2e}$$ |
|
| Passive ($1-x$) | Formal ($y$) |
$$-F_m + \alpha_2 I_{mc}$$ $$I_{1c} – C_{1c}$$ $$-F_m – C_g – C_{1e}$$ |
$$\alpha_2 I_{mc}$$ $$I_{1c} – C_{1c}$$ $$-C_{1e}$$ |
| Informal ($1-y$) |
$$-F_m$$ $$I_{2c} – C_{2c} – \beta I_{2c}$$ $$-F_m – C_g – C_{2e}$$ |
0 $$I_{2c} – C_{2c}$$ $$-C_{2e}$$ |
Based on the payoff matrix, the replicator dynamic equations for the three populations are derived as follows:
$$F(x) = x(1-x) \big[ z(I_m + R_m + \theta R_m + \alpha_1 I_{mc} – \alpha_2 I_{mc} – C_m + F_m) – (I_m – C_m) \big]$$
$$F(y) = y(1-y) \big[ x(\mu L + \theta R_m z + \beta I_{2c}) + z(\beta I_{2c}) – (I_{2c} – C_{2c}) + (I_{1c} – C_{1c}) \big]$$
$$F(z) = z(1-z) \big[ x F_m + R_m – C_g – (F_m + R_m)x \big]$$
The equilibrium points of the system are obtained by setting $F(x)=0$, $F(y)=0$, and $F(z)=0$. There are eight pure-strategy equilibrium points: $E_1(0,0,0)$, $E_2(0,0,1)$, $E_3(0,1,0)$, $E_4(1,0,0)$, $E_5(0,1,1)$, $E_6(1,0,1)$, $E_7(1,1,0)$, and $E_8(1,1,1)$. The stability of these equilibria is evaluated using the Jacobian matrix and Lyapunov’s first method. The Jacobian matrix is given by:
$$
J = \begin{bmatrix}
\frac{\partial F(x)}{\partial x} & \frac{\partial F(x)}{\partial y} & \frac{\partial F(x)}{\partial z} \\
\frac{\partial F(y)}{\partial x} & \frac{\partial F(y)}{\partial y} & \frac{\partial F(y)}{\partial z} \\
\frac{\partial F(z)}{\partial x} & \frac{\partial F(z)}{\partial y} & \frac{\partial F(z)}{\partial z}
\end{bmatrix}
$$
At each pure equilibrium, if all eigenvalues of the Jacobian matrix have negative real parts, the equilibrium is an evolutionarily stable strategy (ESS). Table 3 summarizes the eigenvalues for the eight pure equilibria.
| Equilibrium | Eigenvalues |
|---|---|
| $E_1(0,0,0)$ | $-F_m + C_g$, $-I_{2c} + C_{2c} + I_{1c} – C_{1c}$, $-C_m – C_r – C_i – C_t + I_m$ |
| $E_2(0,0,1)$ | $-C_g – R_m$, $-I_m + C_m + C_r + C_i + C_t$, $-\mu L – I_{2c} + C_{2c} + I_{1c} – C_{1c}$ |
| $E_3(0,1,0)$ | $-F_m + C_g$, $I_{2c} – C_{2c} – I_{1c} + C_{1c}$, $-C_m + I_m – (\alpha_1 – \alpha_2)I_{mc}$ |
| $E_4(1,0,0)$ | $-C_g – F_m$, $-I_{2c} + C_{2c} + (1-\beta)I_{1c} – C_{1c}$, $-C_m + I_m + F_m + R_m$ |
| $E_5(0,1,1)$ | $-C_g – R_m$, $\mu L – I_{2c} + C_{2c} + I_{1c} – C_{1c}$, $-C_m + I_m + C_r + C_i + C_t + (\alpha_1 – \alpha_2)I_{mc}$ |
| $E_6(1,0,1)$ | $-C_g + R_m$, $-I_m – C_m – C_r – C_i – C_t + F_m + R_m$, $-(1-\beta)I_{2c} + C_{2c} + I_{1c} – C_{1c} + \mu L$ |
| $E_7(1,1,0)$ | $-C_g + F_m$, $(1-\beta)I_{2c} – C_{2c} – I_{1c} + C_{1c}$, $-C_m + I_m + F_m + R_m + (\alpha_1 – \alpha_2)I_{mc} + \theta R_m$ |
| $E_8(1,1,1)$ | $-C_g + R_m$, $-(1-\beta)I_{2c} + C_{2c} + I_{1c} – C_{1c} – \mu L$, $-C_m + I_m + F_m + R_m – \theta R_m – (\alpha_1 – \alpha_2)I_{mc}$ |
Subject to the assumptions that the manufacturer’s active recycling cost exceeds its implicit benefit ($I_m < C_m$) and that the consumer’s informal channel net benefit exceeds that from formal channels, the stable equilibria are determined by three cases:
Case 1: If $C_g > F_m$, the government’s supervision cost exceeds the penalty revenue from the manufacturer; the system converges to $E_1(0,0,0)$: passive recycling, informal channel, passive supervision.
Case 2: If $C_g < F_m$ and $(1-\beta)I_{2c} – C_{2c} > I_{1c} – C_{1c}$, then the system converges to $E_4(1,0,0)$: passive recycling, informal channel, active supervision.
Case 3: If $\mu L + I_{1c} – C_{1c} > I_{2c} – C_{2c}$ and $C_m – I_m < (\alpha_1 – \alpha_2)I_{mc}$, the system converges to the ideal equilibrium $E_5(0,1,1)$: active recycling, formal channel, passive supervision.
Dynamic Reward-Punishment Mechanism
In reality, the government’s penalty for passive recycling is not constant. When the supervision probability is high, violations are more likely to be detected, so a lower penalty suffices. Conversely, when supervision is lax, a higher penalty is needed to deter violations. Similarly, the reward for active recycling should increase with the manufacturer’s recycling probability to reinforce desired behavior. I model the dynamic penalty and reward as follows:
$$F_m^d = \frac{v}{z}, \quad R_m^d = x R_m$$
where $v$ is the dynamic reward-punishment coefficient. Substituting these into the replicator dynamic equations yields the eigenvalues for the six pure equilibria under the dynamic mechanism, as summarized in Table 4.
| Equilibrium | Eigenvalues |
|---|---|
| $E_1^d(0,0,0)$ | $-C_g + R_m$, $-I_m + C_m + C_r + C_i + C_t$, $-\mu L – I_{2c} + C_{2c} + I_{1c} – C_{1c}$ |
| $E_2^d(1,0,0)$ | $-C_g – v$, $I_{2c} – C_{2c} – (1-\beta)I_{1c} + C_{1c}$, $-C_m + I_m + v$ |
| $E_3^d(0,1,1)$ | $-C_g – R_m$, $\mu L – I_{2c} + C_{2c} + I_{1c} – C_{1c}$, $-C_m + I_m + C_r + C_i + C_t + (\alpha_1 – \alpha_2)I_{mc}$ |
| $E_4^d(1,0,1)$ | $-C_g + R_m$, $-I_m – C_m – C_r – C_i – C_t + v + R_m$, $-(1-\beta)I_{2c} + C_{2c} + I_{1c} – C_{1c} + \mu L$ |
| $E_5^d(1,1,0)$ | $-C_g + v$, $(1-\beta)I_{2c} – C_{2c} – I_{1c} + C_{1c}$, $-C_m + I_m + (\alpha_1 – \alpha_2)I_{mc} + v$ |
| $E_6^d(1,1,1)$ | $-C_g + R_m$, $-(1-\beta)I_{2c} + C_{2c} + I_{1c} – C_{1c} – \mu L$, $-C_m + I_m + v + R_m(1-\theta) – (\alpha_1 – \alpha_2)I_{mc}$ |
The dynamic mechanism offers a more adaptive policy framework. Through numerical simulations, I find that the dynamic reward-punishment mechanism significantly accelerates the system’s convergence to the ideal equilibrium compared to the static mechanism. The simulation results (with parameters as given in the original study) show that the convergence speed under the dynamic mechanism is approximately 60% faster than under the static mechanism. In addition, the evolution path under the dynamic mechanism is smoother, without oscillations, whereas the static mechanism exhibits noticeable fluctuations before stabilizing. This demonstrates the superior efficiency and robustness of the dynamic policy design.
Sensitivity Analysis in the Evolutionary Game
I conduct numerical simulations to examine the influence of key parameters. Table 5 summarizes the effects of the government’s penalty $F_m$, the manufacturer’s recycling technology level $\alpha_1$, the consumer’s environmental preference $\mu$, and the profit reduction ratio $\beta$ on the system evolution.
| Parameter | Range | Effect on system evolution |
|---|---|---|
| $F_m$ | 1.5 to 3.5 | When $F_m$ is below the threshold (between 2 and 3), the system converges to $(0,0,0)$. When above the threshold, it converges to $(0,1,1)$. Higher penalty accelerates the transition. |
| $R_m$ | 2.0 to 4.0 | Reward influences the evolution speed but not the final equilibrium. Excessive reward may slow convergence due to diminishing marginal effects. |
| $\alpha_1$ | 0.5 to 0.8 | Above the threshold (0.5-0.6), the system reaches the ideal state. Below the threshold, the system may exhibit cycles or no stable point. |
| $\mu$ | 0.1 to 0.7 | Low $\mu$ leads to $(0,0,1)$; higher $\mu$ leads to $(0,1,1)$. Higher environmental preference accelerates convergence. |
| $\beta$ | 0.1 to 0.4 | Higher $\beta$ speeds up evolution to active strategies but does not alter the equilibrium outcome. |
These results highlight the threshold effect of penalties, the enabling role of recycling technology, and the importance of consumer environmental awareness. The dynamic reward-punishment mechanism internalizes the government’s supervision intensity and the manufacturer’s performance, leading to a more efficient and stable governance regime.
Closed-Loop Supply Chain Model without Blockchain Empowerment
Problem Description and Model Setup
I now shift from the behavioral evolutionary game to an operational perspective, focusing on the decision optimization of EV battery pack recycling in a closed-loop supply chain. The supply chain consists of a power battery manufacturer (B), an NEV manufacturer (V), and consumers. In the forward chain, the battery manufacturer produces EV battery packs and sells them to the NEV manufacturer at wholesale price $w$. The NEV manufacturer assembles vehicles and sells them to consumers at retail price $p$. In the reverse chain, retired EV battery packs are collected through one of four recycling modes: (1) battery-manufacturer-led recycling (denoted as $N_B$), (2) NEV-manufacturer-led recycling ($N_V$), (3) hybrid recycling by both manufacturers ($N_{BV}$), and (4) alliance recycling between the two manufacturers ($N_{(B+V)}$).
The key assumptions are as follows. First, all supply chain members are risk-neutral and maximize their own profits. Second, the market demand for NEVs is $D = \phi + g Q_r – \theta p$, where $\phi$ is the base demand, $g$ is the demand gain coefficient from recycling, $Q_r$ is the total collected quantity of retired EV battery packs, and $\theta$ is the price sensitivity coefficient. Third, the total recycling quantity is $Q_r = a + b r_j – \delta r_k$, where $a$ is the voluntary return quantity, $b$ is the consumer sensitivity to the recycling price, $\delta$ is the channel competition intensity, and $r_j$ and $r_k$ are the recycling prices offered by different channels. Fourth, the unit net profit from recycling is $r = c_o – c + r_b$, where $c_o$ is the cost of using virgin materials, $c$ is the cost using recycled materials, and $r_b$ is the echelon utilization benefit. The government also provides a per-unit subsidy $g_r$ for recycled EV battery packs, and there is a carbon emission saving benefit $\zeta E_r$ per unit recycled. Table 6 lists the parameters.
| Symbol | Description |
|---|---|
| $\phi$ | Base market demand for NEVs |
| $\theta$ | Consumer price sensitivity for NEVs |
| $g$ | Demand gain coefficient from recycling |
| $a$ | Voluntary return quantity of retired EV battery packs |
| $b$ | Consumer sensitivity to recycling price |
| $\delta$ | Channel competition intensity |
| $r$ | Unit net benefit from recycling |
| $r_c$ | Unit fixed recycling cost |
| $g_r$ | Government subsidy per unit recycled |
| $\zeta$ | Carbon trading price |
| $E_r$ | Carbon emission saving per unit recycled |
| $c_o$ | Unit cost using virgin materials |
| $c$ | Unit cost using recycled materials |
| $r_b$ | Echelon utilization benefit per unit |
Equilibrium Solutions for the Four Recycling Modes without Blockchain
I solve the Stackelberg game using backward induction. The battery manufacturer acts as the leader in modes $N_B$, $N_V$, and $N_{BV}$, while in the alliance mode $N_{(B+V)}$, the two manufacturers jointly maximize the alliance profit. The optimal decisions are summarized in Table 7.
| Mode | Wholesale price $w^*$ | Retail price $p^*$ | Recycling price(s) |
|---|---|---|---|
| $N_B$ | $\frac{\phi + g Q_r + c\theta}{2\theta}$ | $\frac{3(\phi + g Q_r) + c\theta}{4\theta}$ | $r_B^* = \frac{a + b(r+g_r+\zeta E_r) – b r_c}{2b}$ |
| $N_V$ | $\frac{\phi + g Q_r + c\theta}{2\theta}$ | $\frac{3(\phi + g Q_r) + c\theta}{4\theta}$ | $r_V^* = \frac{a + b(r+g_r+\zeta E_r) – 3 b r_c}{4b}$ |
| $N_{BV}$ | $\frac{\phi + g Q_r + c\theta}{2\theta}$ | $\frac{3(\phi + g Q_r) + c\theta}{4\theta}$ | $r_B^*, r_V^*$ determined simultaneously |
| $N_{(B+V)}$ | — | $\frac{\phi + g Q_r + c\theta}{2\theta}$ | $r_{BV}^* = \frac{a + b(r+g_r+\zeta E_r) – b r_c}{2b}$ |
Detailed expressions for $N_{BV}$ are more complex due to competition; however, the qualitative insights are clear. First, the wholesale price is independent of the recycling mode because the battery cost and demand parameters do not change across modes. Second, the retail price in the alliance mode is lower than in other modes, because the alliance internalizes the double marginalization effect. Therefore, the alliance mode generates a larger sales volume. Third, the hybrid mode yields the highest recycling rate because two independent channels compete for retired EV battery packs, while the alliance mode achieves the highest total supply chain profit due to resource integration and cost sharing.
Numerical Results without Blockchain
Using the parameter values from the original study (Table 8), I compute the equilibrium outcomes for the four modes. The results are reported in Table 9.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| $\phi$ | 300,000 | $r$ | 29,349 |
| $\theta$ | 1.6 | $g$ | 0.1 |
| $a$ | 0.07 $\times$ demand | $b$ | 1.1 |
| $\zeta$ | 1,000 | $E_r$ | 60 |
| $\delta_B$ | 0.45 | $\delta_V$ | 0.5 |
| $r_{c(B)}$ | 400 | $r_{c(V)}$ | 300 |
| $r_{c(B+V)}$ | 250 | $g_r$ | 1,000 |
| Mode | Supply chain profit (CNY) | Sales volume | Recycling volume | Recycling rate (%) |
|---|---|---|---|---|
| $N_B$ | 4,933,474,747 | 49,700 | 18,228 | 36.68 |
| $N_V$ | 4,859,473,348 | 49,700 | 9,150 | 18.41 |
| $N_{BV}$ | 5,000,859,104 | 49,700 | 19,225 | 38.68 |
| $N_{(B+V)}$ | 6,540,679,996 | 99,400 | 20,050 | 20.17 |
The numerical results confirm that the hybrid mode ($N_{BV}$) has the highest recycling rate, while the alliance mode ($N_{(B+V)}$) yields the largest total supply chain profit and sales volume. The alliance mode, however, has a lower recycling rate in this parameterization because the denominator (sales volume) doubles, making the rate lower despite higher absolute collection volumes. This suggests that the alliance mode is more effective at boosting market demand and overall profitability, whereas the hybrid mode is more effective at channeling retired EV battery packs into the formal recycling system.
Blockchain-Empowered Recycling Decisions
Model Extensions
When blockchain technology is introduced, I add two key variables: the blockchain technology investment level $\lambda$ and the consumer trust coefficient $k$ in blockchain. The demand function becomes $D = \phi + g Q_r – \theta p + k \lambda$, indicating that higher blockchain investment and consumer trust enhance demand. The recycling quantity becomes $Q_r = a + b r_j – \delta r_k + k \lambda$, reflecting that blockchain-enabled traceability increases consumers’ willingness to return retired EV battery packs. The cost of blockchain investment is assumed to be $\frac{1}{2} A \lambda^2$, where $A$ is the cost coefficient. The battery manufacturer bears a proportion $t$ of this cost, and the NEV manufacturer bears $1-t$. The four recycling modes under blockchain are denoted as $Y_B$, $Y_V$, $Y_{BV}$, and $Y_{(B+V)}$. The profit functions for the battery manufacturer (B) and NEV manufacturer (V) in the $Y_B$ mode are:
$$\pi_B^{Y_B} = (w – c + g_r) D + (r + \zeta E_r – r_B) Q_r – \frac{t A \lambda^2}{2}$$
$$\pi_V^{Y_B} = (p – w) D – \frac{(1-t) A \lambda^2}{2}$$
Similar functions are constructed for the other modes. Solving the Stackelberg games yields closed-form expressions for the optimal wholesale price, retail price, recycling prices, and blockchain investment level. For brevity, I summarize the qualitative results in Table 10.
| Variable | Without blockchain | With blockchain |
|---|---|---|
| Wholesale price $w$ | Equal across modes | Higher than without blockchain |
| Retail price $p$ | Lowest in alliance mode | Higher than without blockchain; alliance still lowest |
| Recycling price $r$ | Baseline | Lower than baseline due to trust-driven participation |
| Recycling volume | Baseline | Higher |
| Supply chain profit | Baseline | Higher |
Proposition 3 states that the optimal recycling price under blockchain is lower than that without blockchain. This is because blockchain provides transparent information on the battery’s condition and residual value, reducing consumers’ need for a high price to compensate for information asymmetry. Consequently, consumers are willing to return retired EV battery packs even at lower prices, trusting the fairness of the valuation.
Proposition 4 states that the level of blockchain investment increases with consumer trust $k$. Formal verification shows that $\partial \lambda^* / \partial k > 0$ for all modes, confirming that higher consumer trust in blockchain capabilities incentivizes firms to invest more in the technology across the supply chain.
Numerical Results under Blockchain Empowerment
Using the same baseline parameters plus $A=500$, $k=4$, and $t=0.7$, I compute the performance metrics for the four modes with blockchain. The results are reported in Table 11, alongside the no-blockchain cases for comparison.
| Mode | Supply chain profit (CNY) | Sales volume | Recycling volume | Recycling rate (%) |
|---|---|---|---|---|
| $N_B$ (no blockchain) | 4,933,474,747 | 49,700 | 18,228 | 36.68 |
| $Y_B$ | 4,998,974,780 | 50,260 | 19,367 | 38.53 |
| $N_V$ | 4,859,473,348 | 49,700 | 9,150 | 18.41 |
| $Y_V$ | 4,916,297,388 | 50,158 | 9,608 | 19.15 |
| $N_{BV}$ | 5,000,859,104 | 49,700 | 19,225 | 38.68 |
| $Y_{BV}$ | 5,108,024,171 | 50,456 | 21,906 | 43.42 |
| $N_{(B+V)}$ | 6,540,679,996 | 99,400 | 20,050 | 20.17 |
| $Y_{(B+V)}$ | 6,684,186,668 | 100,068 | 23,037 | 23.02 |
The results demonstrate that blockchain adoption improves supply chain profit, sales volume, recycling volume, and recycling rate across all four modes. The hybrid mode still achieves the highest recycling rate, and the alliance mode remains the most profitable. The profit increase in the alliance mode is the largest in absolute terms, but the percentage increase is relatively modest compared to the hybrid mode, whose recycling rate jumps significantly. This differential effect can be explained by the distinct roles of blockchain in different governance structures.
In single-channel modes, blockchain enhances consumer trust but cannot compensate for the limited channel coverage. In the hybrid mode, blockchain alleviates information asymmetry between two competing channels, curbing unethical underpricing based on hidden information and boosting recycling efficiency. In the alliance mode, blockchain primarily reduces internal coordination and verification costs, improving marginal efficiency on top of an already high profit base.
Sensitivity Analysis in the Blockchain-Empowered Supply Chain
Impact of Blockchain Cost-Sharing Ratio on Firm Profits
Figure 1 (not shown for brevity) illustrates that regardless of the cost-sharing proportion $t$ between the battery manufacturer and the NEV manufacturer, both firms’ profits under blockchain adoption are higher than those without blockchain. This indicates a Pareto improvement from blockchain investment, which creates a natural incentive for both parties to collaborate on technology deployment. The cost-sharing ratio only affects the distribution of profits, not the overall profitability improvement.
Impact of Consumer Retail Price Sensitivity
As the retail price sensitivity $\theta$ increases, the total supply chain profit decreases monotonically across all modes. Higher sensitivity means consumers react more strongly to price changes, reducing demand and thereby profits. Firms must carefully calibrate pricing strategies by accounting for the real level of consumer price sensitivity, which can be obtained through market research and data analytics.
Impact of Consumer Recycling Price Sensitivity
The recycling rate is positively correlated with the consumer recycling price sensitivity $b$. When consumers are highly sensitive to recycling prices, a small increase in the recycling price leads to a large increase in recycling volume, thereby improving the recycling rate. This insight suggests that in markets where consumer price sensitivity is high, manufacturers can leverage targeted price promotions to stimulate the return of retired EV battery packs.
Impact of Consumer Blockchain Trust Level
The most critical moderating variable is the consumer trust level $k$ in blockchain. Table 12 reports the sensitivity results for $k = 2, 4, 6$ across the four modes.
| $k$ | $\lambda_B^*$ | $\lambda_V^*$ | $\lambda_{BV}^*$ | $\lambda_{(B+V)}^*$ |
|---|---|---|---|---|
| 2 | 255.9 | 226.0 | 357.0 | 165.4 |
| 4 | 522.6 | 458.1 | 756.2 | 333.9 |
| 6 | 812.7 | 703.0 | 1253.0 | 508.9 |
| $k$ | $p_B^*$ | $p_V^*$ | $p_{BV}^*$ | $p_{(B+V)}^*$ |
| 2 | 175,444 | 175,399 | 175,522 | 144,228 |
| 4 | 176,237 | 176,046 | 176,605 | 144,542 |
| 6 | 177,640 | 177,164 | 178,711 | 145,079 |
| $k$ | Recycling rate $R_B$ (%) | $R_V$ (%) | $R_{BV}$ (%) | $R_{(B+V)}$ (%) |
| 2 | 37.13 | 18.58 | 39.94 | 22.62 |
| 4 | 38.53 | 19.15 | 43.42 | 23.02 |
| 6 | 40.95 | 20.13 | 50.19 | 23.69 |
| $k$ | Supply chain profit $Y_B$ | $Y_V$ | $Y_{BV}$ | $Y_{(B+V)}$ |
| 2 | 4,947,733,297 | 4,873,413,141 | 5,013,479,192 | 6,639,589,771 |
| 4 | 4,988,277,985 | 4,916,297,388 | 5,108,024,171 | 6,684,186,668 |
| 6 | 5,083,716,123 | 4,989,947,750 | 5,263,076,544 | 6,755,490,573 |
Table 12 confirms that a higher consumer trust level in blockchain not only increases the optimal blockchain investment level but also raises retail prices, recycling rates, and supply chain profits. The effect is most pronounced in the hybrid recycling mode, where the recycling rate rises from 39.94% at $k=2$ to 50.19% at $k=6$. This highlights that consumer trust is a powerful enabler of blockchain’s value creation. When consumers believe in the reliability of blockchain-based traceability, they become more willing to pay a premium for vehicles with blockchain-verified batteries and more willing to return retired EV battery packs even at lower prices. Therefore, firms should not treat blockchain as a purely technical solution; they must also invest in consumer education and trust-building communication to fully realize its benefits.
Managerial Insights and Policy Implications
Implications for Government Policy
The evolutionary game analysis reveals that a dynamic reward-punishment mechanism outperforms a static one in steering the recycling system toward an ideal equilibrium. Policymakers should design intelligent, data-driven regulatory frameworks that adjust penalties based on the actual supervision intensity and rewards based on the manufacturer’s demonstrated recycling performance. Setting a meaningful penalty threshold is essential to deter passive recycling behavior; below that threshold, penalties are ineffective. At the same time, excessive rewards may slow down convergence, so reward levels should be carefully calibrated to avoid diminishing marginal returns.
Moreover, the government should promote the development of recycling technology and digital trust infrastructure simultaneously. Subsidizing R&D in battery recycling technologies and setting technology standards can lower the economic threshold for manufacturers to adopt active recycling strategies. Supporting the creation of a unified blockchain-based traceability platform for EV battery packs would reduce information asymmetries across the entire life cycle and facilitate the adoption of blockchain by supply chain actors.
Implications for Corporate Strategy
Enterprises should choose their recycling mode based on their strategic priorities. If the objective is to maximize supply chain profit, the alliance recycling mode is the most effective, as it eliminates channel conflicts, reduces duplicative investments, and leverages complementary resources. If the objective is to maximize the recycling rate, the hybrid mode offers advantages due to competition between two independent channels. Blockchain technology should be treated as a strategic asset that improves trust, transparency, and operational efficiency. Firms should establish fair cost-sharing and benefit-sharing mechanisms to facilitate joint investment in blockchain infrastructure. In addition, firms need to actively communicate the benefits of blockchain to consumers, building trust that enhances the economic value of the technology.
Conclusion and Future Research
This dissertation has investigated the optimization of EV battery pack recycling decisions under the dual drivers of government reward-punishment mechanisms and blockchain empowerment. By combining evolutionary game theory and Stackelberg game models, I generated the following key findings:
First, government penalties exhibit a threshold effect: only when the penalty exceeds a critical value does it effectively deter passive recycling. Rewards influence evolution speed but not the final equilibrium. The manufacturer’s recycling technology level and the consumer’s environmental preference are crucial internal drivers of a well-functioning recycling system.
Second, a dynamic reward-punishment mechanism, where penalties decrease with supervision probability and rewards increase with recycling probability, leads to faster convergence and greater stability than a static mechanism. This dynamic design provides a more efficient and adaptive policy tool for regulating the recycling market.
Third, in the closed-loop supply chain without blockchain, the alliance mode maximizes supply chain profit, while the hybrid mode maximizes the recycling rate. Blockchain adoption improves both profit and recycling rate across all modes, but the magnitude of improvement varies by mode: the hybrid mode benefits most in terms of recycling rate, while the alliance mode benefits most in profit. Blockchain reduces the optimal recycling price by building trust, yet increases retail prices by boosting consumers’ willingness to pay for transparent and traceable EV battery packs.
Fourth, the consumer’s trust in blockchain is the pivotal moderating factor. Higher trust leads to higher blockchain investment, higher retail prices, higher recycling rates, and higher profits. Without sufficient consumer trust, the strategic value of blockchain remains underutilized.
The findings offer actionable insights for both policymakers and industry practitioners. For governments, the recommendation is to adopt dynamic, adaptive regulatory mechanisms and simultaneously invest in digital infrastructure and consumer awareness programs. For enterprises, the recommendation is to align recycling mode selection with strategic objectives, to cooperate on blockchain investment with fair cost-sharing arrangements, and to build consumer trust through transparent communication and credible traceability.
This research has limitations that open avenues for future work. The models do not explicitly incorporate carbon tax or carbon trading policies, which are increasingly relevant in the era of carbon neutrality. Adding those policy instruments as endogenous variables would allow the analysis of interactions between environmental regulation, economic incentives, and blockchain technology. Moreover, the models assume rational or boundedly rational behavior with a stylized set of parameters. Future research could incorporate behavioral biases, asymmetric information, and multiparty competition with third-party recyclers and echelon utilization firms. Empirical validation through industry data would also strengthen the practical applicability of the conclusions. By addressing these gaps, scholars can continue to refine the dual-driven framework of policy regulation and technological empowerment for a sustainable EV battery pack recycling ecosystem.