Analysis of manufacturing and retailer blockchain decision based on resource recyclability

3.1 Blockchain adoption decision model under manufacturer recycling

In response to the problems of asymmetric information, distrust, and low efficiency of the actual recycling system in traditional resource recycling models, a manufacturer-led CLSC model is studied and constructed, taking into account the distinction between manufacturer recycling, retailer recycling, and blockchain decision adoption. Among them, two manufacturer-led Stackelberg game models are constructed in the CLSC model, namely the manufacturer recycling model that does not adopt (NA) and adopts (YA) blockchain. The structural diagram of the NA model is shown in Figure 1.

Figure 1

Schematic diagram of NA recycling model structure.

From Figure 1, the manufacturer uses a dual-channel strategy for sales: online direct sales and offline retail distribution. In the forward supply chain, the manufacturer sells products directly to consumers at direct sales prices by opening online channels and supplies products at wholesale prices to retailers, who then sell the products to consumers at retail prices. In the reverse supply chain, manufacturers pay recycling prices to recover waste products from consumers, test and screen these recovered waste products, and then use the qualified waste to remanufacture new products. Introducing blockchain on the basis of the NA model constitutes the YA model, with no significant difference in overall structure compared to Figure 1. However, it utilizes blockchain technology to improve service quality in the CLSC and enterprises’ enthusiasm for recycling, thereby reducing product remanufacturing costs and recycling detection costs and changing the interest relationship between manufacturers and retailers [13,14,15]. On this basis, corresponding assumptions are made for the model, as shown in Figure 2.

Figure 2

Schematic diagram of specific content of research model assumptions.

From Figure 2, the model assumes a total of seven assumptions. The model assumptions are totaled into seven. The first is the volume of direct sales by manufacturers and retail sales by retailers. The second is the volume of recycling. The third is that all recycled used products will be used for product remanufacturing and there is no difference in consumer preferences between new and remanufactured products. The fourth is that all members of the supply chain make decisions based on the principle of maximizing their own benefits. The fifth is that the cost of the blockchain is borne by the manufacturer. The sixth is that the recycling process is profitable. The seventh is that N and Y are used as superscripts to differentiate between service levels before and after the adoption of blockchain technology, the amount of voluntary returns from consumers, the cost of remanufacturing, and the cost of inspection by the manufacturer. Due to space limitations, only three points will be elaborated on in detail. The first step is to set the direct sales volume of manufacturers and retail sales volume of retailers, and the calculation expression of the two is shown in the following equation:

In Eq. (1), E e means the manufacturer’s direct sales volume; λ denotes the market share of direct sales channels; γ indicates the total market demand; δ represents the sensitivity coefficient of sales volume to price; x e refers to the direct selling price of the product sold by the manufacturer; μ expresses the cross price coefficient between two related channels; x s stands for the actual retail price of the product sold by the retailer; m means the sensitivity coefficient of sales volume to the actual service level of the CLSC; g denotes the actual service level of the CLSC; E s stands for retail sales of retailers. Next is the setting of the recycling amount, which is expressed in the following equation:

In Eq. (2), y indicates the recovery amount; d denotes the total amount voluntarily returned by consumers; χ refers to the sensitivity coefficient between the amount of recycling and the actual recycling price; r represents the actual recycling price of recycled waste products; n stands for the actual sensitivity coefficient of recycling volume to service level. Finally, N and Y are used as superscripts to distinguish between the service level before and after adopting blockchain technology, the total amount voluntarily returned by consumers, the actual remanufacturing cost of the product, and the actual inspection cost of the manufacturer. The corresponding parameters of the YA model have changed after the introduction of blockchain, and the study uses the reverse induction method to solve [16,17,18]. The profit function of manufacturers and retailers is expressed as shown in the following equation:

In Eq. (3), Π A YA stands for the manufacturer’s profit function; Π R YA means the retailer function; κ indicates wholesale price; b n refers to the production cost of the manufacturer when manufacturing a new product; b a stands for the unit cost of blockchain; b ni represents the manufacturer’s inspection cost. At this point, the reverse induction method is used to solve the first and second derivatives of x s in the retailer function in Eq. (3). Due to the existence of the optimal retail price, the retailer’s profit is maximized. Therefore, if ∂ Π R YA / ∂ x s = 0 , then x s = ( 1 − λ ) γ + δ x e + μ κ + m g Y / 2 δ . The corresponding expression obtained by substituting this equation into Eq. (3) is shown in the following equation:

The expression of the seafront array obtained by solving the first and second-order partial derivatives of κ , x e , and r for Eq. (4) is shown in the following equation:

In Eq. (5), G means the Hesse matrix, which is a negative definite matrix. At this point, let ∂ Π A YA / ∂ κ , ∂ Π A YA / ∂ x e , and ∂ Π A YA / ∂ r all be 0, and the optimal wholesale, direct sales, and recycling prices obtained from this are expressed in the following equation:

In Eq. (6), κ YA ∗ , x e YA ∗ , and r YA ∗ represent the optimal wholesale, direct sales, and recycling prices. The expression of the equilibrium solution obtained by substituting Eq. (6) into each variable is shown in the following equation:

Substituting Eqs. (6) and (7) into the retailer’s profit function yields a maximum profit of Π R YA ∗ = [ ( 1 − λ ) γ − ( μ − δ ) ( b n − b a ) + m g Y ] 2 / 16 δ , while substituting them into the manufacturer’s profit function yields the manufacturer’s maximum profit as expressed in the following equation:

After comparing and analyzing the profits of manufacturers YA and NA blockchain-related indicators under the recycling mode, corresponding decision-making analysis of recycling entities can be obtained, namely three propositions, as shown in Figure 3.

Figure 3

Decision analysis of recycling entities under the manufacturer’s recycling model.

From Figure 3, by comparing and analyzing the profitability of metrics related to YA and NA of blockchain in the manufacturer’s recycling model, the study arrives at three key propositions. First, Proposition 1 shows that the YA of blockchain technology in the manufacturer recycling model resulted in an increase in wholesale prices, direct sales prices, and retail prices. This is due to the fact that the introduction of blockchain facilitates customer-initiated returns, improves service levels, and reduces remanufacturing costs and manufacturer inspection costs, which pushes up the overall price level. Second, Proposition 2 states that there exists a threshold for the sensitivity coefficient of the recycling volume to the service level, and when this threshold is exceeded, the price after YA of blockchain will be significantly higher than the price NA blockchain, which suggests that blockchain is able to significantly improve the supply chain performance in the presence of high recycling volume. Finally, Proposition 3 emphasizes that changes in the sensitivity coefficients consistently result in greater gains for manufacturers that YA blockchain than those that NA. In particular, the profit advantage from blockchain adoption becomes more pronounced after the sensitivity coefficient of recycling volume to service level exceeds a certain threshold, ensuring that manufacturers continue to earn higher profits after adopting blockchain.

3.2 Blockchain adoption decision model under retailer recycling

Retailers also play an important role in this process. Therefore, in addition to building a manufacturer-led model, a blockchain decision-making model with retailer recycling is studied and constructed, which also includes two models of NA and YA blockchain. The specific content of the NA model architecture is shown in Figure 4.

Figure 4

Schematic diagram of the specific content of NA model architecture.

From Figure 4, the NA model architecture describes the role of the retailer as a central intermediary in the forward and reverse supply chains. In the forward supply chain, retailers purchase products from manufacturers at wholesale prices and sell them to consumers at retail prices. In the reverse supply chain, the retailer recovers used products from consumers at recycled prices and then sells these recycled products to manufacturers at transfer prices for remanufacturing or disposal. Blockchain technology has not been introduced in this model, so information flows between parties in the supply chain are slower, with less transparency and trust, and the recycling process relies on traditional trust relationships between retailers and consumers and manufacturers. The information asymmetry in this architecture may lead to inefficiencies, lower recycling rates, and some uncertainty in the quality control of recycled products, constraining the overall operational efficiency of the supply chain. The study also proposes corresponding model assumptions, as shown in Figure 5.

Figure 5

Model assumptions under retailer dominance.

From Figure 5, the model assumes that the first step is to set the manufacturer’s direct sales and retail sales, and the second step is to set the recycling amount. All the recycled old products will be returned to the manufacturer for product reconstruction, which will incur corresponding costs. Then, members within the supply chain will make corresponding decisions based on maximizing their own interests, using NA and YA superscripts to distinguish the profit functions of different models, and subscripts to distinguish the variable equilibrium solutions of different models. N and Y superscripts are utilized to distinguish variables such as service level before and after adopting blockchain. In the YA model, the manufacturer’s profit function is solved in the same way as Eq. (3), while the retailer’s profit function is expressed in the following equation:

In Eq. (9), a denotes the transfer price. On this basis, the inverse induction method is used to solve the corresponding problem. The Hesse matrix obtained by solving the first and second-order partial derivatives x s and r in Eq. (9) is shown in the following equation:

The Hesse matrix in Eq. (10) represents that the actual profit function of the retailer is a joint concave function of retail and recycling prices, where there exists an optimal retail and recycling price to maximize the retailer’s profit. Therefore, the corresponding expression obtained by setting both ∂ Π R Y R / ∂ X s and ∂ Π R Y R / ∂ r to 0 is shown in the following equation:

The corresponding expression obtained by substituting Eq. (11) into the manufacturer’s profit function in Eq. (3) is shown in the following equation:

At this point, it calculates the first- and second-order partial derivatives of the relevant variables in Eq. (12), and the resulting Hesse matrix is expressed as shown in the following equation:

The optimal expression of wholesale, direct sales, and transfer prices is obtained by setting ∂ Π A YA / ∂ κ , ∂ Π A YA / ∂ x e , and ∂ Π A YA / ∂ r to 0. The optimal wholesale and direct selling prices are the same as Eq. (6), and the transfer price expression is shown in the following equation:

In Eq. (14), a YR ∗ indicates the optimal transfer price. After substituting it into each variable, the relevant expression is the same as Eq. (7), but the expression with some differences is shown in the following equation:

At this point, the retailer’s maximum profit is Π L YL ⁎ = [ ( 1 − λ ) γ + ( μ − δ ) ( b n − b a ) + m g Y ] 2 / 16 δ + [ d Y + n g Y + χ ( b n + b a − b r Y − c n i Y − b r i Y ) ] 2 / 16 δ while the manufacturer’s maximum profit is the same as Eq. (8), and only the subsequent denominator 4 is revised to 8. Based on this, the decision-making analysis content of the recycling subject under the retail recycling model is shown in Figure 6.

Figure 6

Content of decision analysis for recycling entities under retailer recycling mode.

From Figure 6, the retail recycling model also includes three propositions. Firstly, YA blockchain will increase wholesale, direct, and retail prices. At the same time, when the actual sensitivity coefficient of sales volume to service level exceeds the threshold, both direct and retail sales will increase. Secondly, in the retail recycling mode, if there is a recycling volume below a certain threshold, its sensitivity coefficient to service level will increase both the recycling price and transfer price after YA blockchain. Finally, in the retailer recycling model, the profits of manufacturers and retailers depend on different parameters, and the benefits after YA blockchain are greater than those after NA blockchain. At the same time, there can be a sensitivity coefficient of the service level to the amount of recycling that is greater than a certain threshold, making the benefits of YA blockchain higher than those of NA.

Similarly, by combining the manufacturer recycling model with the retailer recycling model, the recycling model adopted by blockchain includes five propositions. Firstly, under the two recycling modes, wholesale, direct sales, retail, recycling, transfer prices, and recycling volume will increase with the actual adoption cost of blockchain, while direct and retail sales will decrease. Second, the wholesale, direct and retail prices, direct and retail sales, and recycling volume of the two recycling models after YA blockchain will increase with the improvement of service level, while recycling and transfer prices will decrease. Then, after YA blockchain, the manufacturer’s profit increases in both modes as the sensitivity coefficient of recycling volume to service level increases. Finally, the wholesale, direct and retail prices, and direct and retail sales are the same for the two models after YA blockchain technology. The profit of manufacturers in manufacturer mode is higher than that of retailers in retailer mode.

4.1 Comparative case analysis

To verify the correctness of the proposition proposed for manufacturing and retail blockchain decision-making, numerical examples were used to validate it. First, the various propositions in the manufacturer and retailer recycling model before and after YA blockchain were verified through numerical examples. Second, the impact of unit cost of blockchain and actual service level of CLSC on various variables in the manufacturer and retailer recycling model after YA blockchain was verified through numerical examples. The results of parameter assignment and parameter assignment before and after YA blockchain are shown in Figure 7.

Figure 7

Parameter assignment and the result of parameter assignment before and after adopting blockchain. (a) General parameter assignment results. (b) Parameter assignment results before and after adopting blockchain.

From Figure 7, in the conventional parameter assignments, γ , λ , δ , μ , χ , and b n were 1,000, 0.4, 0.8, 0.5, 30, and 5, respectively. After YA blockchain, it was 0.6, 3, 0.5, 0.2, 0.3, 10, and 0, respectively. Based on the assigned values, the relevant variables of the positive supply chain were compared accordingly. Among them, the relevant variables in the blockchain model NA and YA by manufacturers and retailers in the recycling mode varied with the sensitivity coefficient (m) of sales volume to service level or the sensitivity coefficient (n) of recycling volume to service level, as shown in Figure 8.

Figure 8

Result of sensitivity coefficient change of related variables with sales volume to service level. (a) Wholesale price comparison and retail direct sales price comparison. (b) Comparison of direct sales and retail sales. (c) Comparison of recycling volume. (d) Comparison of recycling prices.

In Figure 8, in setting the sensitivity coefficient to increase from 0.2 to 0.8, the data showed that as the sensitivity coefficient increased, the service level increased significantly. When the sensitivity coefficient was 0.2, the manufacturer’s service level score was 60 and the sales volume was 120, while when the sensitivity coefficient was increased to 0.8, the service level score improved to 85 and the sales volume grew to 280. This result showed that as consumer demand for service quality increased, parties in the supply chain were forced to increase their service levels to maintain competitiveness and sales volume. The increase in the sensitivity coefficient reflected the positive impact of competitive market pressures on supply chain performance, especially in market environments with high service demand. This result not only verified Propositions 1, 2, 4, and 5, but also demonstrated the facilitating effect of increasing service levels on recycling volumes. However, the sensitivity coefficients were easily affected by factors such as consumer preference, market demand fluctuation, product life cycle, service level, and technological innovation, which showed significant differences in different industries and environments. For example, factors such as market maturity, policy regulation, and consumers’ awareness of environmental protection affect the sensitivity of sales to varying degrees, and the differences between these factors in different environments make the role of sensitivity coefficients variable. The role of these factors in different environments makes the sensitivity coefficient have variability. Next, it analyzed the profits of manufacturers and retailers, and the results are shown in Figure 9.

Figure 9

Manufacturer and retailer recycling models manufacturer profits. (a) The sensitivity coefficient is 0.8, and the blockchain adopts the results of Unit cost of 3 and 3.5. (b) The blockchain adopts the result that the Unit cost is 3, and the sensitivity coefficient m is set to 1 and 0.8.

The manufacturer’s recycling model set a sensitivity coefficient m of 0.8, and the blockchain adopted unit costs of 3 and 3.5. In the retail recycling mode, the unit cost of blockchain adoption was set to 3, and the sensitivity coefficients m were set to 1 and 0.8. From Figure 9, there was no intersection between the profits of YA and NA blockchain when the sensitivity coefficient n was 0.8, and the unit cost of blockchain YA was 3 in the comparison of manufacturers’ profits under the manufacturer’s recycling model. When the sensitivity coefficient m was 0.8 and the unit cost of blockchain YA was 3.5, and the sensitivity coefficient n was less than about 2.7, the actual profit of adopting blockchain was lower than that of NA blockchain. After exceeding around 2.7, the former gradually exceeded the latter, and the profit at the intersection was 36,540 yuan. The same results were observed in the comparison of profits between retailers and manufacturers in the recycling model. When the sensitivity coefficient m was 0.8, the unit cost of blockchain YA was 3, and the sensitivity coefficient n was 1.25. It was the turning point. At this point, the profit of YA blockchain was higher than that of not adopting it. It can be seen that as the blockchain cost increased, the manufacturer’s profit showed a trend of decreasing first and then increasing. The study adopted the inverse induction method, and by calculating the profit function under different blockchain costs, it was clarified that when the blockchain cost was low, the manufacturer’s profit was hit more, a phenomenon that can be attributed to the inevitability of the initial investment. However, when the blockchain cost exceeded a certain threshold, the improved recycling efficiency of blockchain technology made the production cost decrease and the profit starts to rebound. The comparison results of retailer profits under the two modes are shown in Figure 10.

Figure 10

Comparative results of retailer profits under two model models. (a) The result of setting the sensitivity coefficient m to 0.8 and 1 in the manufacturer's recycling model. (b) The results of setting sensitivity coefficients m to 0.8 and 0.01 in the retailer recycling model.

The unit cost of blockchain adoption was set to 3, the sensitivity coefficient m in the manufacturer’s recycling model was set to 0.8 and 1, and the sensitivity coefficient m in the retailer’s recycling model was set to 0.8 and 0.01. From Figure 10, in the manufacturer’s recycling model, when m was 0.8, the profit of retailers who adopt or do not adopt blockchain would not change with the change of sensitivity coefficient n. When m was 1, the trend of change was the same. When m was 0.8 in the retailer recycling model, the change in n value was directly proportional to the retailer’s profit. Retailers who adopt blockchain always had higher profits than those who did not, with the former having a minimum of 28,218. When m was 0.01, when the value of n was higher than about 6, the profit adopted was higher than that not adopted, and the profit at the intersection was 28,010. Combining Figures 9 and 10, Propositions 3 and 6 demonstrated the effectiveness of blockchain decision-making in the constructed model. It also illustrated that the impact of blockchain costs on retailer profits showed a similar trend, but unlike the manufacturer model, the retailer model had flatter profit fluctuations. This is due to that retailers mainly bear the recycling transfer costs rather than the direct remanufacturing costs. Therefore, by comparing these two recycling models, the study revealed the different mechanisms by which blockchain costs affected their respective profits and validated the differences in revenue distribution between the two.

4.2 Impact case analysis

In addition, the impact of example validation mainly focused on the impact of the actual unit cost and service level of blockchain on various variables in the manufacturer and retailer recycling model after adopting blockchain. The variable assignment was the same as Figure 7, with a sensitivity coefficient m of 0.8 and a sensitivity coefficient n of 10. The variation of each variable with the unit cost of blockchain is shown in Figure 11.

Figure 11

The change of each variable with the unit cost of the blockchain. (a) Results of different prices and sales changing with blockchain. (b) The change of recovery price and recovery volume with the unit cost of the blockchain. (c) Change of transfer price with unit cost of blockchain. (d) Changes of manufacturer's profit and retailer's profit with unit cost of blockchain.

Based on Figure 11, the prices of wholesale, retail, and direct sales in both models increased with the increase in the unit cost of blockchain. The direct and retail sales of the two models decreased with the increase in the unit cost of blockchain. In addition, the recycling prices of the two models increased with the increase of blockchain unit cost, but the recycling price growth rate of the manufacturer model was significantly higher than that of the retailer model. At the same time, the recycling volume of the two models showed the same results, while the transfer price of the retailer model increased with the increase of the unit cost of the blockchain, and the growth rate of the manufacturer model in the recycling price was the same. In terms of manufacturer profits, there was a minimum value for both models when the unit cost of blockchain was greater than 0. The minimum value for manufacturers appeared around 25, and the profit was around 362000 yuan. The profit trend under both modes was decreasing first and then increasing. In terms of retailer profits, the retailer model grew as the unit cost of blockchain increased. By simulating different unit cost scenarios, the study clarified that in low-cost blockchain applications, the price and sales volume changes in the supply chain were more moderate, but when the cost was higher, the price rises rapidly and negatively affected the sales volume. This suggested that cost control by blockchain technology was crucial in supply chains, and that reasonable cost sharing and optimization could effectively alleviate the sales pressure caused by price increases. In addition, the impact of service level on related variables is shown in Figure 12.

Figure 12

The impact of service level on related variables. (a) Results of varying prices and sales with service levels. (b) Changes in recycling prices and volumes with service levels. (c) Transfer price changes with service level.

From Figure 12, the improvement of service level has led to an increasing trend in wholesale, retail, and direct sales prices, with wholesale prices maintained between 880 and 900 and retail prices maintained between 1,050 and 1,100. The retail price has always been higher than the other two prices. In addition, under both models, the recycling and transfer prices decreased with the increase of actual service level, while the recycling volume increased with the increase of actual service level. By synthesizing Figures 11 and 12, the different change graphs intuitively displayed the trend of variable changes in different models and also demonstrated the effectiveness of the proposed blockchain decision-making, namely the correctness of propositions 7–11. To summarize, the sensitivity coefficients of related variables such as wholesale price, retail price, and recycling volume with the change of sales volume have a more significant impact on the service level. First of all, as the sensitivity coefficient increases, the service level shows a trend of significant increase, which is because higher sensitivity coefficients imply that consumers pay more attention to the quality of products and services, which in turn forces all parties in the supply chain to improve the level of service to meet consumer demand. It is found that when the sensitivity coefficient exceeds a certain threshold, manufacturers and retailers have to optimize pricing strategies and improve service quality to maintain the growth of sales volume, thus ensuring the smooth operation of the supply chain. In addition, the impact of sensitivity coefficient on recycling volume is also more significant, and higher sensitivity coefficients encourage supply chain parties to pay more attention to improving service level, thus increasing recycling rate and remanufacturing efficiency. Therefore, the sensitivity coefficient not only affects the sales volume, but also has a profound impact on the overall service quality and recycling efficiency of the supply chain.