Using a sample of A-share listed firms from 2007 to 2022 and employing a difference-in-differences (DID) approach, we evaluate the impact of social security premiums reduction on corporate labor share and its mechanisms. The results demonstrate that the social security premiums reduction significantly increases labor share in labor-intensive enterprises. Specifically, compared to firms with lower labor intensity, the policy elevates labor share in high labor-intensity enterprises by approximately 2.45% relative to the sample mean. The effect primarily operates through alleviating corporate financing constraints and enhancing human capital agglomeration effects. The policy’s positive effects are more pronounced for enterprises in regions with heavier social insurance expenditure burdens, non-high-tech enterprises, enterprises facing higher market competition intensity, larger-scale enterprises. Furthermore, the policy significantly raises labor share for ordinary employees while exerting a negative impact on executive compensation. This study expands the research on the income distribution effects of social security premiums reductions from the perspective of factor income distribution, providing micro-level evidence for optimizing social security system and improving primary distribution mechanisms, while providing policy insights for promoting common prosperity.
Securing poverty alleviation gains is key to rural revitalization, with financially strained households at significant risk of falling back into poverty. Considering the randomness of agricultural yield and the limited budget, this paper studies the planting decision-making behaviors of multiple farmers, and compares the impact of different subsidy policies on farmers’ income, consumer surplus, and social welfare. The study finds that the loan subsidy can better stimulate farmers to expand the planting area, bring more agricultural products, and yield higher is higher, but the planting subsidy can allow farmers to obtain higher income. In addition, when the government subsidy budget is small, the loan subsidy can help the government create greater social welfare; When the government subsidy budget is large, production subsidy can bring higher social welfare. This paper evaluates the performance of different subsidy policies when only some farmers in the regional market are subject to capital constraints. When the government budget is limited, the choice of subsidy policies by the government is related to the incidence of the risk of falling back into poverty in that region. Moreover, the paper indicates dual subsidy policies do not yield better performance than single subsidy policies.
Exploring the impact and mechanism of economic growth targets on land transfer structure is of great significance for improving the target assessment system and optimizing the allocation of land resources in the new era. Based on data from 284 prefecture-level cities in China from 2011 to 2020, we conducted empirical tests using the system generalized method of moments (GMM) estimation and the panel mediation effect model. The results indicate that: The higher economic growth targets promoted the adjustment of land transfer structure toward increasing commercial and residential land, and the conclusion remained valid after a series of robustness tests. The increase in the economic growth target led to greater financial pressure and intensified the crowding-out effect of real estate investment, thereby positively driving the transfer of commercial and residential land. The economic growth target had a notably stimulative effect during the phase of the economy transitioning into high-quality development, as well as in coastal cities, large cities, and cities with population influx. It had a stronger impact on the supply of commercial land transfer area, which promoted the rise of the average floor price of commercial and residential land, thus aggravating the distortion of land transfer price. This research not only enriches the literature on economic growth targets and land transfer structure but also provides empirical evidence for improving land supply model and better promoting high-quality economic development.
Preventing and mitigating potential systemic financial risks that may result from climate change and low-carbon transition is a major theoretical and practical concern at present. Climate risks mainly include physical risks, which urgently require a comprehensive and profound low-carbon transition for economies, and transition risks, which emphasize that the transition to a low-carbon economy also brings risks. This paper focuses on the risks of low-carbon transition, regards the low-carbon city pilot as a quasi-natural experiment, and uses the experience data of 232 urban and rural commercial banks in China from 2008 to 2022 to deeply explore the impact and mechanism of low-carbon transition on the systemic risk level of banks, as well as the significance of policy timing and location in the prevention and control of transition risks. The study finds that under the impact of low-carbon transition, the systemic risk level of banks has increased significantly, and banks in cities that are environmentally friendly and adhere to high-quality development are less affected. Mechanism analysis shows that the impact of low-carbon transformation on the systemic risk of banks has a channel that starts from the micro level and gradually transmits, and the low-carbon transformation will impact the production and operation of local industrial enterprises, and transmit the risk to the bank through the relevant loan business, and eventually bring about an increase in the systemic risk level of the bank. Further research shows that the timing of low-carbon transition policies will affect the above-mentioned transition risks, and the increase in bank systemic risks will be lower when low-carbon transformation is promoted in the upward stage of the economy. In addition, the resilience of the city’s economy will also affect the performance of transition risks, with banks with weak economic resilience in cities experiencing a significant increase in systemic risk levels under the impact of low-carbon transition, while banks with stronger economic resilience in cities are almost unaffected by the shock. This paper has important policy implications for China to actively and steadily promote the “dual carbon” goal and prevent systemic financial risks during the low-carbon transition period.
In recent years, the tail risk cross-market resonance and contagion effects have become more and more prominent, which put forward higher requirements for the sound market expectation management mechanism. Based on the time-varying conditional prediction distribution framework of deep learning parameter optimization, this paper realizes the classification of tail risks of financial asset prices in stock, bond and foreign exchange markets, and empirically analyses the impact of investor sentiment in the evolution, resonance and contagion of tail risk based on the interdependent dynamic and contagion effect of tail risk. This paper finds that:1) The evolution trend, interdependent dynamic and contagion effect of the financial asset tail risks have stage characteristics. As the process of financial openness, the interdependence of exchange rate price tail risk and the spillover effect of bond price tail risk have increased. 2) Compared with global investor sentiment, domestic investor sentiment has a more significant effect on the tail risk of financial asset prices, with improved domestic investor sentiment helping to smooth out the tail risk of stock and exchange rate prices, while deteriorating domestic investor sentiment will exacerbate the resonance and contagion between tail risks. 3) Relative to financial condition, the economic condition recovery exerts a stronger positive impact on domestic investor sentiment. Meanwhile, policy uncertainty significantly constrains the sentiment recovery. The findings offer valuable insights for enhancing financial stability through expectation management.
With the increasing emphasis on sustainable development, corporate environmental, social, and governance (ESG) performance has become a focal point for investors, significantly influencing their investment decisions. However, how ESG performance affects investors’ risk perception remains underexplored. This paper utilizes textual data from earnings conference call questions of A-share listed companies from 2009 to 2021, extracting risk-related information through textual analysis to examine the impact of corporate ESG performance on investor risk concerns. The findings reveal a significant negative correlation between corporate ESG performance and investor risk concerns, indicating that strong ESG performance can reduce investors’ concern about corporate risks. This conclusion remains robust under various tests, including instrumental variable methods and difference models. Further analysis shows that the baseline results are more pronounced in state-owned enterprises, companies with high annual report complexity, a high concentration of unsophisticated investors, and a higher proportion of institutional investors. These results suggest that corporate ESG performance plays a critical role through the mechanism of information substitution. Additionally, the paper finds that strong ESG performance can mitigate the exacerbating effect of investor risk concerns on financing constraints. The findings of this paper provide new empirical evidence for understanding the impact of ESG performance on investors’ information processing capabilities and offer important practical insights for companies aiming to enhance ESG performance to reduce risk attention and alleviate financing constraints.
In the private equity secondary market, information insufficiency constrains asset pricing and market liquidity. Private equity secondary funds (S funds), however, as specialized trading entities, are able to complete transactions at higher prices under comparable conditions. To explain this difference, this paper develops an information-economics framework in which due diligence is modeled as an endogenous expansion of investors’information sets, while information precision is treated as a continuous state variable. We show that changes in information precision affect trading outcomes by altering the risk compensation required by risk-averse investors. Embedding the trading process in a Nash bargaining framework with Bayesian updating, we characterize a minimum information-precision threshold below which trade cannot occur. The model implies that improvements in information precision reduce uncertainty premia, increase investors’ willingness to pay, and facilitate trade once the threshold is exceeded. These results provide a theoretical foundation for understanding how S funds improve pricing efficiency and market liquidity through information production rather than liquidity provision alone.
Under the context of increasing volatility in the global financial environment and growing economic vulnerability across multiple countries, precisely identifying and measuring upward and downward risks in each economy has become a critical issue in macroprudential management and policy regulation. This study employs the concept of economic risk volatility vectors to dynamically capture the effects of macroeconomic volatility on different quantiles of GDP growth distributions, utilizing a time-varying connectivity approach and frequency-domain decomposition methods to construct the connectivity network of economic risk volatility vectors among G20countries. The research uncovers differences in short- and long-term connectivity and explores the network’s structural characteristics and drivers. Results indicate that, facing significant external shocks, developing countries experience intense volatility in downward entropy and pronounced expected short-term gaps, whereas developed countries demonstrate lower vulnerability and higher economic resilience. The economic risk volatility vector network exhibits asymmetric spillover characteristics, with downward risk volatility vectors centered around developed economies, whereas upward risk volatility vectors are predominantly driven by developing countries, albeit lacking long-term sustainability. Cross-border capital flows and credit are identified as key driving factors influencing the connectivity of economic risk volatility vectors among G20 countries. In downturn periods, developed economies leverage their mature financial markets and investment channels to rapidly respond globally, reinforcing their leadership in risk spillovers. Conversely, during upturns, developing countries become attractive targets for international capital flows due to higher external returns, thus leading the upward spillover processes globally. However, their leading role in this context is not sustainable in the long run.
Policy adjustments triggered by global climate change and their associated uncertainties are closely related to the precious metals market. This paper examines the spillover effects between climate policy uncertainties and the international precious metals market, as well as their dynamic evolution patterns, from the perspectives of time domain and frequency domain respectively, using the TVP-VAR-DY and TVP-VAR-BK models. The empirical evidence reveals that there are significant spillovers between climate policy uncertainty and the precious metals market, which intensify notably during major climate policy and crisis events, and primarily driven by short-term factors. Meanwhile, the spillover effects of climate policy uncertainty on palladium and gold are stronger than those on platinum and silver, and palladium exhibits high stability in the system. Based on spillover network analysis, it is found that platinum and palladium act as stable risk transmitters and risk receivers in the system, respectively. In the short term, the fluctuation of precious metal markets can affect climate policy uncertainty, while in the medium and long term, climate policy uncertainty is a net risk exporter, indicating that the impact of climate policy uncertainty on the precious metals market has a significant time-lag effect.
With the aim of accurately identifying high-risk defaulting customers and improving the credit risk management level of commercial banks, this paper proposes an innovative imbalanced credit evaluation model that can replace the multilayer perceptron (MLP). One is to replace the traditional evaluation models built on the MLP with Kolmogorov-Arnold networks (KANs) based on the Kolmogorov Arnold representation theorem, in order to avoid the phenomenon of fixed activation functions and inability to adjust according to the characteristics of credit data in MLP, which limits its flexibility in characterizing the highly complex nonlinear relationship between credit evaluation indicators and default status; The second is to use class balanced loss to modify the cross entropy function in KANs, alleviating the weak applicability of KANs with cross entropy as the objective function to imbalanced samples. The empirical results of multiple datasets with different credit characteristics, 11 comparative models, 5 model evaluation criteria, and two different training and testing set partitioning ratios verified the applicability and robustness of the proposed model for predicting defaults on credit data with different imbalanced ratios. This article explores the practical feasibility of using KANs as an alternative to MLP, as well as their applicability and effectiveness in the field of financial risk assessment, which can provide new ideas and methods for credit risk management in commercial banks.
Addressing corporate carbon dependency is crucial for the high-quality development of the manufacturing sector, whether the emerging blockchain technology can provide new impetus for corporates to reduce carbon emissions at the source remains a question. Drawing on endogenous growth theory, this study develops a theoretical model that incorporates blockchain technology and carbon dependency within a corporate pollution and carbon reduction framework. Empirically, this study analyzes panel data from A-share listed manufacturing corporates from 2007-2022, using the debiased machine learning model to assess the effects of blockchain technology on carbon dependency. The results show that: 1) Blockchain technology substantially mitigates carbon dependency, and this result holds up under rigorous robustness tests. 2) Heterogeneity analysis shows a more pronounced effect of blockchain technology among capital-intensive and labor-intensive firms, as well as those with lower external financing dependence and slower growth rates. 3) The study also confirms that blockchain technology plays the functions of management empowerment and innovation empowerment, facilitating carbon dependency reduction by enhancing total factor productivity and promoting green innovation. 4) Additionally, blockchain technology notably alleviates financial distress and boosts firm value and environmental performance. Meanwhile, there is a dynamic effect of blockchain technology on carbon dependency. This study provides a systematic examination of how blockchain technology reduce corporate carbon dependency, offering empirical evidence and policy recommendations for advancing the green and low-carbon transformation of the manufacturing industry.
Knowledge-sharing platforms play a critical role in the dissemination and innovation of knowledge. However, the platforms face a tough challenge in inducing members to share their knowledge. To improve the sharing motivation of members, the platforms need to establish effective incentive mechanisms. Consequently, based on the low-carbon knowledge-sharing platform (hereafter called carbon platform), this paper develops a differential game model that incorporates both the platform and members with varying abilities, analyzing their decisions and utilities under two incentive models: Activity-based and value-based, considering various characteristics(complementary collaboration, mutual learning, and competition among members). The findings of this study suggest that either incentive model could be optimal depending on the proportion of high-ability members within the platform. When the proportion of high-ability members is low (or high), the platform should adopt the activity-based (or value-based) incentive model. Secondly, we find an extreme case resulted in the impact of knowledge conversion rates on member-sharing behavior. When the conversion rate exceeds a certain threshold, members could gain sufficient utility by learning from others, leading them to reduce efforts to share knowledge. The research also reveals that as the benchmark competition intensity increases, member effort decreases, but their utility does not necessarily decline. Finally, we extend the main model to discuss the conditions under which members engage in free-riding and the case where members obtain utility from low-carbon knowledge output. The findings of the extended model demonstrate the robustness of our results as demonstrated in the main model. This research provides theoretical support for the selection of incentive mechanisms and operational management strategies for knowledge-sharing platforms.
The effective transfer of construction equipment knowledge across projects can improve the management performance of construction projects and is an important research direction in construction project knowledge management. In order to reduce the noise of construction equipment knowledge transfer across projects, fully extract available construction equipment knowledge, and enhance the adaptability of cross project knowledge on target projects by utilizing multiple types of knowledge, this paper constructs a TBMKF framework for construction equipment knowledge transfer across projects based on temporal blocking and multi knowledge fusion. Firstly, using the maximum entropy temporal blocking method TBME, the domain distribution noise of the source project is eliminated, and the consistent distribution within the domain is obtained multiple subdomains with the greatest inter domain differences; Secondly, a knowledge extraction model Res_CL was constructed, extract the spatiotemporal information contained in high-dimensional temporal data in the subdomain. Finally, a multi knowledge feature fusion method MKF was designed, which utilizes domain specific knowledge of subdomains and distribution differences between subdomains and target domains, resulting in a multi knowledge fusion model that is more suitable for the target project. To verify the effectiveness of the proposed method, this paper conducted experiments on six multi time step knowledge transfer tasks using real data from eight shield tunneling construction projects from different geological environments and equipment. Compared to the baseline model, the average improvement in MAE values of the TBMKF framework is 29.27%, 18.95%, 35.37%, 25.46%, 51.73%, and26.17%, respectively. It can effectively assist in the prediction task of the target project, improve the knowledge management level and performance of the construction project organization.
This paper examines backward shareholding, where a downstream firm holds shares in an upstream firm, and incorporates the partial control of the upstream firm by the downstream firm, represented by the control coefficient. We construct a supply chain game model under both single-channel and dual-channel modes, and investigate pricing decisions and channel encroachment strategies within the context of a ‘backward shareholding+ joint control’ relationship. The results show that in a single channel, both wholesale and retail prices decrease with the control coefficient of the downstream firm; whereas in dual channels that involve direct selling, retail prices are unaffected by the control coefficient and shareholding ratio. Meanwhile, when the downstream firm has greater control coefficient and the acceptance of direct channel is low, the upstream firm is more inclined to open direct channel as the downstream firm’s shareholding ratio increases. On the other hand, when the acceptance of the direct channel is high, the upstream firm is willing to open direct channel, but this is not always in favor of its original shareholder. As the control coefficient of the downstream firm increases, opening the direct channel may harm the original shareholder of the upstream firm but benefit the downstream firm. Moreover, when the channel encroachment issue is taken into account during the equity transfers, there are specific conditions for the upstream and downstream firms to achieve backward shareholding cooperation. When the acceptance of the direct channel is relatively high, the downstream firm’s control coefficient is low yet its shareholding ratio is high, there won’t be an equity transfer price that both parties can agree upon, and the equity transaction may fail.
With the boom in social media, influencer marketing (IM) leveraging the key opinion leader (KOL) has become a prevalent way for firms to intervene in consumers’ quality awareness. Firms often increase product variants (products with identical core functions but differing in attributes such as color or specification) to match KOL preferences and thereby improve the probability of positive reviews. However, this may reduce the credibility of reviews, weaken consumer trust, and ultimately undermine firm profitability. Based on this, this paper constructs a two-period model to analyze optimal strategies and the effectiveness of IM under two pricing modes— commitment pricing and responsive pricing — and further explores the firm’s optimal pricing-mode selection. The results show that: 1) IM is not always beneficial and may even reduce profits. Under responsive pricing, IM increases firm profitability only when the profit discount loss is relatively low, whereas under committed pricing, IM diminishes profits when both consumer patience and the firm’s profit discount loss are high. 2) When IM is not implemented, responsive pricing is the optimal choice. When IM is adopted, responsive pricing performs better when the firm’s preferred sales period and the consumers’ preferred purchase period are misaligned, whereas committed pricing performs better when they coincide.
Many software developers sell software and services to users both through downstream service providers (consulting companies) and directly to users, which means that there is channel encroachment by software developers. This paper constructs the “Authorization” and “Authorization
This paper investigates the event-wise cloud manufacturing task scheduling under the uncertainty of task service time, where an event-wise hybrid ambiguity set, encompassing Wasserstein metric, expected dispersion, mean and support set, is used to characterize this uncertainty. Considering the distributionally robust chance constraints on the maximum manufacturing capacity of manufacturing cloud services, a distributionally robust chance-constrained model for task scheduling is proposed to minimize the normalized weighted sum of service cost, energy consumption, and worst-case expected task completion time, which is equivalently transformed to mixed-integer linear programming model by the strong dual theory. To solve the resulting model, we develop a Benders dual branch-and-cut algorithm, incorporating in-out and aggregated sample Benders dual cut generation strategies. Finally, a real-world case of a group enterprise producing large cement equipment in Tianjin, China, are used to verify the performance and our model and algorithm, and draw management insights.
The path optimization problem of the truck-drone delivery faces multiple economic, social and environmental constraints. This paper comprehensively considers the “triple bottom line” (TBL) principle of the unity of economy, society and environment, and studies the multi-objective path optimization of truck-drone delivery in the instant delivery scenario. The economic, social and environmental indicators are respectively characterized by comprehensive distribution cost, the customer’s time perception satisfaction and the comprehensive carbon emissions of truck-drone under variable load. In the quantification of social goals, a non-linear perception satisfaction model is constructed based on the coexistence of soft and hard time windows as constraints, and considering the psychological characteristics of customers’ “bounded rationality” based on prospect theory. In terms of quantifying environmental goals, considering the synergistic state of trucks and drones and the delivery of goods listed in the orders, the comprehensive carbon emissions of trucks and drones under variable loads are quantified. For truck-drone delivery, the coding and decoding methods are designed to represent the feasible paths, the convergence algorithm is accelerated by crossing and conflict detection, and the local search strategy is combined to enhance the diversity of the population, so that an extended non-dominant ranking genetic algorithm adapted to the model is proposed to solve the model. Finally, the effectiveness of the model and algorithm is verified by performance test, algorithm comparison and model comparison. The results show that: 1) Considering customer psychological perception and truck variable load can help accurately quantify customer time perception satisfaction and comprehensive carbon emissions. 2) Under the condition that other conditions remain unchanged, the use of truck-drone delivery can complement the advantages of two delivery modes in terms of load, range, energy consumption, and speed, taking into account scenarios with multiple customer scales. Compared with the route scheme of using only truck distribution, the use of truck-drone delivery can reduce the comprehensive delivery cost by 9.87%, improve the customer’s time perception satisfaction by 10.31%, and reduce the comprehensive carbon emission by 16.31% on average, and with the increase of customer scale, the optimization effect of the cooperative delivery scheme is gradually enhanced. 3) Coordinating the conflict between the three goals of comprehensive delivery cost, customer’s time perception satisfaction and comprehensive carbon emission can provide a decision-making basis for logistics service providers to make a satisfactory path plan under different target preferences.
Based on two failure modes: Internal degradation and external shocks, this paper develops reliability and maintenance strategy optimization model for a consecutive-
Mobile wireless sensor networks (MWSNs) are widely used in fields such as military and industry. Accurate assessment and proper optimization of reliability are highly important for ensuring a safe and stable operation. Previous studies fail to jointly consider the effects of node mobility and wireless communication randomness, and are focused on single aspect of practical requirements in terms of connectivity and timeliness. Thus, they are unsuitable for real MWSNs. In this study, the signal-to-interference-plus-noise ratio model is built to depict wireless communication randomness, the bidirectional communication range is defined according to wireless communication mechanism, and its distribution is derived based on the Laplace transformation. The 2 terminal transmission reliability is defined by incorporating the requirements of connectivity and timeliness under the joint effects of node mobility and wireless communication randomness, and accordingly an approximation algorithm for the novel network reliability index is proposed based on the conditional event analysis and Monte Carlo sampling. A bi-objective optimization problem of network reliability and energy efficiency is constructed, and a method to determine the optimal transmitting signal power of a node is provided by combining the weighted sum method with the fuzzy satisfying method. Finally, a case study is used to demonstrate the effectiveness of the proposed method.
Designing a logistics network that robustly matches the demands of routine operations and emergency response is a critical issue for balancing dual-use requirements, enhancing societal resilience, and optimizing resource allocation. To address this challenge, this study proposes a dual-use logistics network design framework based on a two-stage distributionally robust optimization model. In the first stage, the model optimizes decisions regarding warehouse location selection, emergency supplies pre-positioning, Daily material operation capacity, and road reinforcement construction. In the second stage, a Wasserstein ambiguity set is constructed to characterize the uncertainties in pre-disaster routine demand, post-disaster emergency demand, and road damage. Using this uncertainty representation, the model integrates the optimization of transportation planning for both routine and emergency scenarios, ensuring the efficiency and reliability of the logistics network. The nonlinear problem is reformulated into an equivalent linear model and solved using a column-and-constraint generation (C&CG)algorithm. A case study based on the July 2021 Henan flood was conducted to validate the proposed framework. The results indicate that the dual-use logistics network significantly improves network operational performance and emergency response efficiency through resource integration, while the construction cost of dual-use warehouses has a substantial impact on network layout decisions, underscoring the importance of controlling facility costs for practical implementation. Furthermore, the Wasserstein-based distributionally robust optimization model effectively supports the robust matching of dual-use demands, enhancing the adaptability and fault tolerance of the logistics network in the face of risks. The findings emphasize the necessity of promoting dual-use facility construction tailored to local conditions and ensuring efficient operations through scientific planning. This approach leverages the synergies of the dual-use model to enhance emergency response capabilities and optimize resource allocation.
The urban interdependent systems (UISs), represented by power supply and water supply systems, provide essential services that are vital to societal welfare and functionality. Enhancing the resilience of UISs is beneficial for maintaining the safe operation and promoting high-quality development of cities, which is not only related to pre-risk preparedness decisions and post-risk response decisions, but also involves the fairness concerns derived from UISs’ public attributes. This paper develops a model that aims at maximizing resilience of UISs based on risk preparedness and risk response decisions with fairness consideration. The model is a two-stage robust optimization model with decision-dependent uncertainty, and the model is reformulated as four level mixed-integer linear programming model and solved using an adapted column and constraint generation algorithm. A case study validates the necessity of integrating risk preparedness and risk response decisions and provides comprehensive decision schemes for resilience optimization of UISs under various risk severities and resources, and then the impact of fairness concerns on the resilience of UISs and the fairness of risk preparedness and risk response decisions is analyzed.
Short videos have become a crucial channel for the public to access information on social media. Audiovisual features are key aspects of short videos, and this paper focuses on exploring their impact on the dissemination of health behaviors in health-related short videos. Based on 3483 health behavior video logs from influencers on a Chinese short video platform, this study utilized machine learning techniques to mine both structured and unstructured data (including audio, visual, and comment-related data, totaling 82,972 data sets). It employed SHAP value algorithms and econometric methods from machine learning to analyze relationships between variables and conducted behavioral experiments to verify causal mechanisms. The results reveal that the match between audiovisual features significantly enhances the dissemination of health behaviors. Specifically, when high voice arousal matches with high visual variation or low voice arousal matches with low visual variation, it triggers a higher level of endorsement from the public towards influencers, thereby increasing the public’s intention to adopt health behaviors. This paper empirically demonstrates, based on video mining, the effectiveness of short videos in disseminating health behaviors, providing managerial insights for promoting the social mobilization mechanism of the‘Healthy China Initiative’.
As a novel approach to organizing and managing scientific research projects, the horse-racing system features multi-agent parallel research, parallel funding, phased assessment and elimination, and selective support for the best performers. Focusing on the characteristics of horse racing system, this paper constructs a stage assessment incentive differential game model including output subsidy mode, R&D effort-dependent subsidy mode, and a hybrid subsidy mode of cost-sharing and benefit-sharing. According to the observability of R&D efforts, front-end or back-end subsidies of production activities and other factors, this paper analyzes the dual-use technology projects of military-civilian collaborative innovation, and discusses the applicable conditions and implementation effects of three classical policy subsidy modes. The study finds that under the output subsidy mode and R&D effort-dependent subsidy mode, the implementation of the horse racing mechanism can reduce the equilibrium R&D effort level of military and civilian enterprises, while increasing their profits, without altering the optimal intensity of incentives. Under the optimal intensity of incentives, the higher the input efficiency of civilian enterprises, the stronger the sensitivity of the incentive mechanisms to the performance of the horse-racing mechanisms. There exists an equilibrium state of incentive compatibility between military and civilian enterprises, at which the hybrid subsidy mode of cost-sharing and benefit-sharing can achieve the optimal allocation. Therefore, military enterprises should optimize incentive subsidy modes and intensity to enhance R&D efficiency and fund utilization rates, while avoiding excessive subsidies to civilian enterprises to prevent undermining sustained R&D capabilities.
The traditional grey prediction model ignores the network structure information in the sample data. Therefore, this paper proposes a grey prediction model based on directed visible graph network. The model converts the time series into corresponding networks and uses the Leicht-Holme-Newman-2 to calculate the similarity of each node. The similarity is converted into weights and a network index is proposed to process the original sequence to further obtain the network structural information in the original data, with the aim of considering the internal structural changes in the time series from a network perspective. Finally, a fractional order grey prediction model will be established based on the sample data processed by the directed visible graph network with grey wolf optimization algorithm to improve the performance. Therefore, the model was applied to predict relevant indicators in the ESG field. The research results showed that the novel model had better prediction performance than other classical models. At the same time, the method proposed in this paper to mine data from a network perspective can also be applied to many other prediction models.
