Using A-share listed companies as a sample, this paper investigates peer effects in corporate cash dividends policy from the perspectives of propensity and amount of dividend payments. We find robust evidence that firm's cash dividend payment propensity is significantly influenced by the policies of their industry peers. Moreover, peer influence on cash dividend payment is more pronounced among high-growth and low-cash flow firms. Meanwhile, we find little evidence about peer effects in dividend payments amount. These results indicate that under the background of semi-mandatory dividends rules, firms with high growth and low cash flow have to consider their peers' dividend policy and respond properly to obtain refinancing qualifications. However, what firms focus on is not their peers' payout amount but their propensity to pay dividends. Further research shows that information learning, industry rivalry and CEO reputation concerns are main driving factors of peer effects of corporate cash dividends policy. This paper helps us get better understanding of corporate payout policy in special institutional background and enriches corporate peer effects research.

In recent years, the rapid development of shadow banking not only benefits the economic society, but also brings certain risks to the stability of the financial market. It is of great practical significance to deeply study the risk transmission channels of shadow banks and to put forward appropriate regulatory countermeasures for timely resolving financial risks. Through the structural vector autoregression (SVAR) model based on directed acyclic graph (DAG), the dynamic risk transmission among the shadow banks and commercial banks, money supply, securities/bond market, real estate and macro-economy has been investigated in this work, and the risk contagion of shadow banks from three transmission channels:Monetary policy, asset price and real economy has also been studied. The results demonstrate that, whether short-term or long-term, the risk of shadow banks is mainly affected by monetary policy channels, that is, shadow banks will amplify the transmission effect of monetary policy to a certain extent, and when commercial banks are subjected to risk shocks, shadow banks will be more likely to be exposed to risk. In the short term, the risk of the shadow banks will be transmitted to the real economy through the exchange rate, while in the long term, the risk of shadow banking will be conducted to the bond market through the asset price channel.

The executive compensation, as the core problem in the principal-agent relationship of modern enterprises, has always been the focus of the society. Taking private listed companies as samples, this paper first studies the conditional quantile partial effects of company size, earnings per share, taking, current liability and executive shareholding proportion on executive compensation at different quantile levels by using conditional quantile regression, and the results show that their effects are various at different quantile levels. Then, in order to obtain the general marginal effects of these factors, their unconditional quantile partial effects are analyzed by using unconditional quantile regression. We find that company size, earnings per share and taking have diverse positive correlations with executive compensation at different quantile levels, while current liability is negatively correlated. Executive shareholding proportion is positively related to the executive compensation at low quartile levels, but its effects on the middle and high quartile levels are not significant. Finally, the relative results are compared and it is found that the conditional quantile partial effect and unconditional quantile partial effect of every influencing factor are different, and unconditional quantile partial effects have more realistic value.

Among traditional volatility measurements, normal covariance estimators are not able to distinguish the downside risk and upside gains of asset return, while traditional lower partial moment estimators are asymmetric and impossible to sum up. Therefore, this paper introduces a new risk measurement called realized semi-covariance (RSCOV) to conduct volatility forecasting and portfolio optimization. Based on decomposition of realized covariance matrix, we test it on two common diversification investing strategies, equally-weighted risk contribution (ERC) strategy and global minimum variance (GMV) strategy. To perform forecasting, we adopt online weighted ensemble (OWE) algorithm in machine learning domain to boost the out-of-sample performance of HAR-RV. Compared to existing covariance or realized covariance, we find that realized downside semi-covariance matrix, that only contains information about negative volatility, can be used to better balance the risk contribution of assets in portfolio. Then, using high-frequency data of A share market spanning from 2011 to 2018, empirical result shows that our OWE-HAR-RV can outperform HAR-RV in monthly prediction. Lower RSCOV can be applied to ensure risk parity and minimum variance portfolio strategies to achieve better allocated asset weights and lower maximum loss while maintaining certain portfolio return.

The paper investigates the asymptotic properties of quasi-maximum likelihood estimators of the DSAC panel model with fixed effects when the space system is stable and both n and T are large. The paper shows that when using the transformation approach, the quasi-maximum likelihood estimators yield a bias of O(1/T) order in the general case. When (n-1)/T→0, the estimators converge consistently to the true value with the rate of √(n-1)T. When (n-1)/T→∞, the estimators converge to a degenerate distribution at the rate of T. The estimators obtained by the direct approach yield a bias of max(O(1/T),O(1/n)) order in the general case. When n/T→0 and n/T→∞, the estimators are converge to different degenerate distributions at the rate of n and T respectively. The bias corrected estimators have better finite sample properties than the quasi-maximum likelihood estimators. When n/T³→0, the bias corrected estimators obtained by the transformation approach are √(n-1)T consistently converge to the true value. When n/T³ and n³/T both tend to 0, the bias corrected estimators obtained by the direct approach converge to the true value consistently with the rate of √nT. The direct approach can consistently estimate the individual effects and time effects while transformation approach cannot. The finite sample property of the DASC panel model with fixed effects is better than that of DSAR panel model when the error term has the spatial correlation structure. Finally, an empirical research example shows the application value of the DSAC model.

The promotion and use of the green housing is an important way to reduce building energy consumption and achieve emission reduction targets. Based on the perspective of generalized trust, starting from the internal psychological factors and external situational factors, the research model for the influencing factors of urban residents' willingness to pay for green housing is systematically constructed. And the first-order and higher-order moderating effects of generalized trust and the boundary conditions influencing the effect are explored. The results show that, for internal psychological factors, generalized trust has a significant positive moderating effect on the relationship between green residential cognition and willingness to pay, but it is vulnerable to the negative influence of loss aversion psychology. Generalized trust has a significant positive moderating effect on the relationship between residents' environmental concern and willingness to pay, and is susceptible to the positive influence of residents' advertising appeals (source reliability appeals and self-interest appeals). Generalized trust has a significant positive moderating effect on the relationship between residents' moral identity and willingness to pay, and is easily affected by residents' high construal level. For external situational factors, generalized trust has a significant positive regulating effect on the relationship between social atmosphere/group pressure and willingness to pay. And the authoritative certification mark helps to increase residents' initial trust in green housing, and promote the formation of social atmosphere in which residents actively purchase green housing. Suggestions are provided at the level of government, developers and individual residents based on the empirical research conclusions.

This paper studies a supply chain consisting of a manufacturer, a retailer, and consumers with low-carbon environmental awareness, in which the retailer is the leader and need to choose among three possible channel structures:A pure physical channel, a pure online channel, and dual channels. This paper builds a supply chain model considering the differences in both the unit carbon emission difference and channel operating cost between physical channel and online channel, and focuses on studying the retailer's channel strategy choice. We show that the unit carbon emission difference between physical channel and online channel is one of the significant factors driving the retailer's optimal channel strategy choice. In particular, given other factors, the retailer's optimal channel strategy will change from pure physical channel, to dual channels, and to pure online channel, as the unit carbon emissions level of physical channel relative to online channel increases. Meanwhile, the effect of unit carbon emission difference between channels on the retailers' channel selection strategies will be moderated by the difference in channel operating costs between channels. Furthermore, this paper also analyzes the retailer's optimal pricing decisions and find that when the retailer chooses a dual-channel strategy, the retailer should reduce the optimal retail price of the physical channel with the increase of the unit carbon emissions level of physical channel relative to online channel and the optimal online retail price depends on the physical channel's unit carbon emission level relative to the online channel. Further, the retailer can influence the manufacturer' optimal wholesale price by adding new online channel, thereby increasing the retailer' profits.

This paper studies a manufacturer's two-dimensional product differentiation and pricing strategies when selling products in different channels. On the basis of the horizontal and vertical dimensions, we consider three possible product differentiation strategies. We find that the optimal pricing strategies in the cases of horizontal and vertical differentiation are uniform pricing and differentiated pricing, respectively. The comparisons among differentiation strategies suggest that when only single-dimensional differentiation (i.e., one horizontal differentiation or one vertical differentiation) is available, product indifference strategy can be the optimal strategy for the manufacturer under some conditions. Specifically, under the condition that only the horizontal differentiation strategy is available, when product indifference strategy is optimal, the manufacturer optimally chooses to produce high-quality products; however, in the case that only the vertical differentiation strategy is available, he chooses to produce only low-quality products. When both horizontal and vertical differentiation are available, with the increase of production cost, the horizontal differentiation gradually dominates the vertical differentiation; however, as long as the level of horizontal differentiation is high enough, regardless of the cost, the horizontal differentiation is always optimal for the manufacturer. Besides, we extend the situation of adopting horizontal and vertical differentiation strategy simultaneously.

To explore the internal law of secondary supply chain coordination by buy-back contract under the condition of price randomness and retailer risk aversion with asymmetric production cost information. A buy-back contract model under new conditions is constructed and solved to analyze the impact of information asymmetry and risk aversion on each decision variable in the supply chain. The simulation results show that under the condition of price randomness, no matter the information is symmetrical or not, as long as the retailer has risk aversion, every decision variable in the supply chain will have bifurcation mutation. No matter whether the retailer is risk averse or not, the asymmetric information of production cost will bring additional benefits to the supplier, but it will damage the benefits of the retailer and the whole supply chain as well. The more asymmetric the information is, the greater the amplitude of various decision variables in the bifurcation mutation region will achieve. The conclusion is that the bifurcation mutation is a special phenomenon after the price randomness and the participants' risk aversion coupling. Supplier can generate additional revenue by withholding private information, but at the expense of retailer and supply chains; The best way for retailer to deal with the asymmetric information of production cost is to make the information of production cost transparent at the lowest cost. Retailer face various external risks with a stable attitude, which is more conducive to scientific decision-making.

Considering the deficiency of existing propagation dynamics models, as well as the transmission characteristics of 2019 novel coronavirus (2019-nCoV) and prevention-control measurements, a new infectious disease model is established, which included eight compartments, i.e. susceptible, uninsulated, isolated, diagnosed, asymptomatic, cured after diagnosis, cured after asymptomatic and died of disease. In order to simulate the pressure of a large number of isolated patients on the operation of isolation points and the pressure of a large number of confirmed patients on the operation of hospitals, the saturation characteristics of state transfer parameters are considered in the model. In the aspect of parameter configuration, the time-varying characteristics of state transition parameters in different stages of epidemic development are analyzed. A parameter identification model is established and solved by Markov chain Monte Carlo algorithm. In addition, the risk indicators of epidemic transmission are established from multiple dimensions to comprehensively assess the risk of epidemic transmission. Through the case studies of coronavirus disease 2019 (COVID-19) in Wuhan and foreign countries (America and Spain), the results show that the calculated results are in better agreement with the official data and reveal the mechanism of epidemic transmission compared with other classical dynamics models.

With the implementation of the policy of targeted poverty alleviation, the impoverished people under the current standards by 2020 will all be lifted out of poverty. Nevertheless, the solution to the existing problem of impoverished people does not mean that there will be no more impoverished people after that. Basic medical insurance, critical illness insurance program and medical assistance are important guarantees to prevent the emergence of impoverished people. The establishment of a tripartite system of sharing medical expenses, which may still require additional reimbursement, is an important basis for ensuring that there is no more impoverished by diseases and back to poverty due to illness. In this paper, we consider the problem of sharing additional expenses of medical insurance, introduce the theory of supply chain and game theory, and transform the problem into a multi-party cost-sharing problem. Firstly, we have constructed a cost-sharing model, to simplify the calculation of the proportions of parties, assume in which the basic medical insurance covers all the additional expenses, and critical illness insurance program and medical assistance achieve the purpose of sharing the additional reimbursement expenses by assuming the basic reimbursement expenses of the former. Then, the contract-bargaining process is composed of two Nash bargaining models. The problem of conflict and cost sharing is resolved according to the result of tripartite consultation, and the optimal decision of final cost sharing ratio is obtained. Finally, through the mathematical analysis of above-mentioned results, the more additional costs are borne, the more the basic costs are borne by critical illness insurance program and medical assistance, and the validity and correctness of the model are proved by data simulation.

The implementation of social responsibility has significant effect on political risk prevention and control in host countries. However, under the complicated and volatile social background of host countries, transnational corporations may either be criticized for involving too much in fulfilling their social responsibility or lose the opportunity of getting further cooperation because of their “insufficient” performance. Therefore, the decision-making on social responsibility performance has been extensively concerned as a prominent challenge for transnational corporations. From the perspective of implementing social responsibility, this paper portrayed various cooperation strategies between host countries and transnational corporations by studying on incentive contract, restraining contract and non-discrimination contract, and constructed differential game models on political risk prevention and control for the foreign investment reciprocity of transnational corporations. The research shows that the political environment stability in host country has significant impact on mutual profit and path selection on optimal agreement. For transnational enterprises, small reputation decay rate makes incentive contract the optimal choice in short term cooperation, and non-discrimination contract performed better than others in long term cooperation, while complete restraining contract has the worst effect. However, transnational enterprises, which can inefficiently fulfilled their responsibility, may be immune to incomplete restraining contract. When reputation decay rate gets higher, incentive contract is still optimum, while complete restraining contract, which remain to be the worst, may even erode the invested capital of the corporation.

Assume that the natural gas importing country carries out reserve and price control for the sake of economic benefit and supply safety, while the natural gas exporting country makes investment in exploitation technology so as to maximize economic benefit. Under this condition, a dynamic game model of an importing country and an exporting country is presented. The optimal control theory is used to analyze the optimal natural gas consumption and reserve of the importing country, as well as the optimal technology investment and export quantity of the exporting country. The research shows that the investment efficiency and the investment cost coefficient, as well as the reserve preference and selling price, have an important influence on the import and export strategy. The steady-state utility of reserve and the total utility of the importing country are both inverted “U” in relation to the selling price. And when the former reaches the maximum, the selling price is lower than that of the latter. The importer's price control may cause the reserve to deviate from its initial purpose, and even lead to a result completely opposite from the expected. The study also finds that, while the exporting country could use its monopolistic power to avoid the adverse effects generated by the gas-importing country, such as price controls, it may also lose the potential increase in market share due to changes in the importer's reserve preferences.

Following the idea of decomposition-reconstruction-subsequence forecasting-ensemble, a combined forecasting model based on variational mode decomposition (VMD) was proposed. The model was constructed by selecting suitable decomposition model, optimizing reconstruction method, choosing appropriate subsequence forecasting method and ensemble method. And it was used to forecast the China containerized freight index (CCFI) and analyze the volatility characteristics and economic connotations of CCFI. Firstly, The time series CCFI was decomposed into multiple modal components by using VMD. Secondly, The modal components were reconstructed into high frequency, medium frequency, low frequency and trend subsequences, which means short-term market imbalance factors, seasonal factors, major events and market supply and demand respectively. Here, the fuzzy C-clustering algorithm was used to reconstruct the modal components, and its parameter C was optimized by component time-scale analysis. The economic meaning of each subsequence was explored by analyzing its volatility characteristics. Thirdly, a method based on data feature analysis was proposed to select the proper forecasting models, and it was used for reconstruct subsequences forecast. Finally, forecast results of reconstructed subsequences were added to obtain final output, and the ensemble forecast output was compared with other models' forecast results. The empirical results showed that the combined forecast model proposed in this paper is superior to the single model, such as BPNN, SVM, ARIMA, and EMD combination model, as well as other multi-scale combined forecast models based on VMD. And the analysis results reflected the external fluctuation characteristics and intrinsic economic meaning of CCFI.

Customer churn prediction is an important content of customer relationship management (CRM). In many real customer churn prediction modeling, the class distribution is highly imbalanced, so that the performance of model is poor and it's difficult to achieve satisfactory results. At the same time, in reality, there are only a small number of labeled samples, and a large number of them are unlabeled, which cause a lot of waste of useful information. In order to solve the two problems above, this study combines the technologies of meta cost-sensitive learning, semi-supervised learning and ensemble method of Bagging, and proposes semi-supervised ensemble based on metacost model (SSEM) for customer churn prediction. This model mainly includes the following three stages:1) Metacost method is used to modify the label of initial labeled training set L, a new training set L_m is obtained, then L_m is randomly divided into model training set L_tr and model verification set V_a; 2) V_a is used to select three base classifiers with the highest classification accuracy, then these classifiers cooperate to selectively label some samples from unlabeled data set U, which are added into L_tr; 3) N base classifiers are trained on the new model training set L_tr, then using them to classify samples in test set, and the final classification results are obtained by integration. The empirical analysis is conducted in two customer churn prediction datasets, and the results show that the performance of SSEM model is superior to the common used supervised ensemble models and the semi-supervised ensemble models.

Aiming at the problem that how to aggregate group DEMATEL evaluation information, this paper takes complex network as the breakthrough point and proposes a new large-scale group DEMATEL decision making method form the perspective of complex network. Firstly, the decision makers are regarded as nodes and a complex network is constructed by calculating the consensus degree of the decision maker's DEMATEL evaluation matrix. Secondly, the network condensation degree and subgroup condensation degree are designed to measure the node weights and subgroup weights. Finally, group DEMATEL evaluation matrix is aggregated by the network density operator and the multiple attribute decision making is implemented by using the DEMATEL method. One numeral example is used to illustrated the feasibility and rationality of the proposed method. The results show that this method can aggregate large-scare group DEMATEL evaluation information more comprehensively and accurately, and the stability of this method is more better.

As green logistic emerged as a new trend, green vehicle routing problem (GVRP) has received wide attention from related fields, but literature reviews on the latest research of GVRP remain rare. Awared of this fact, this paper is intented to review several typical GVRP models and their solving algorithms. Firstly, an elementary GVRP model and several fuel consumption/carbon emission measuring methods are briefly described. Secondly, according to the optimization of environmental benefit and the composition of objective function, GVRP models are classified into tree types, i.e. fuel consumption/carbon emission minimization VRP, comprehensive cost minimization VRP and multi-objective VRP. Each model is discussed from four aspects, namely optimization objective, factors influencing fuel consumption/carbon emission, measurement models of fuel consumption/carbon emission and constraints. Then, some solving methods about GVRP such as exact algorithms, heuristic algorithms and metaheuristic algorithms are briefly introduced, and several widely used metaheuristic algorithms are analyzed. Finally, through presenting new applications of GVRP in just-in-time logistics distribution, cold chain logistics distribution, electric vehicle logistics distribution and joint logistics distribution, this paper points out the growing trend of theory and practical method of GVRP.

With the Internet of Things, big data, artificial intelligence making breakthroughs in the field of security, public video monitoring systems have developed quickly in recent years. The equipment generates massive amount of unstructured data, through analysis and research on pedestrian trajectory of video data, it can be found that the hidden behavior patterns contained which have an important research value. The article uses the multiple object tracking algorithm based on object detection to extract and describe the pedestrian movement trajectory in the surveillance video of subway station and mall exits, and then analyzed the trajectory pattern of pedestrians on the basis of trajectory. Aiming at the characteristics of pedestrian trajectory, a trajectory clustering method based on trajectory similarity was designed and implemented on the basis of point density clustering algorithm. The results showed that the method can effectively extract pedestrian trajectories, and extract trajectory patterns from large types of trajectory data.

To solve collision-free path planning problem for automated guided vehicles (AGV), an improved A^* algorithm with time factor was proposed to reduce turning time. In addition, combining this new approach with time window and priority strategy could help to transfer the concept of the dynamic collision-free path planning for multiple AGVs into practice. Firstly, the improved A^* algorithm was used to statically plan the path of each AGV with least turning times. Secondly, the arrival time and safety time interval of path nodes were analyzed, and priority was dynamically assigned to multiple AGVs based on power and path performance. Combining with the time window model, the collision problem for AGVs is solved and system efficiency is improved. The results of the case study show that this algorithm not only guarantees the path optimality, but also solves the problem of multiple turning times caused by the traditional A^* algorithm. It can effectively realize the system scheduling without repetition and conflict, and proves its good adaptability and robustness in the dynamic environment.

“Blockchain+” business model empowers companies with practical significance, but lacks systematic and quantitative review guidance. This paper takes domestic and foreign literature data as the source, combines descriptive statistics, cooperative analysis, co-occurrence network analysis and cluster analysis to explore the current research status and research hotspots, and taps future research trends and evolutionary trends. It shows:1) Domestic and foreign research started late, but the research situation is gratifying. Domestic research monographs are obvious, but foreign research cooperation is profound; 2) It emphasizes the integration of Internet of things, big data and other emerging technologies, and effectively empower business model innovation research; 3) There are significant differences between domestic and foreign research. Domestic focus on financial research, foreign research explores of multi-industry business model innovation and subversion. Finally, a “blockchain+” business model research framework is constructed. The research ideas and conclusions of this paper are enlightening to academic research and practical application of “blockchain+” business model.

Input-output analysis is a quantitative method of national economic decision-making, and input-output table is its data foundation. At present, multi-regional input-output (MRIO) table plays an important role in analyzing the socio-economic impact of multiple regions and the decision-making of resource and environmental protection. The core work of MRIO table compilation is to determine the inter-regional trade flow matrix of different products, and the establishment of the matrix depends more on the non-investigation mathematical method, the gravity model is an important method. Due to the large number of product data of MRIO table compiling demand, there is often a short time lag in some data samples. Ordinary linear regression or spatial regression was used to estimate the parameters of the compilation year under a general compiling hypothesis, which is the parameters of the compilation year could be replaced by the parameters of the short-term lag year because there is little change in short time for the structure of multi-regional trade. In order to make more effective use of the sample information of the short-term lag data, a hybrid estimation method is proposed in this paper. The results show that a hybrid estimation method can effectively improve the accuracy of prediction in processing short-term lag year data of compiling MRIO table through empirical analysis of trade flows of agricultural products, automobiles and steel among the 28 EU member states and Monte Carlo simulation. A hybrid estimation method combines the advantage of spatial regression with ordinary linear regression, it provides a good idea for compiling MRIO table.

Site visit is an important channel for investors to obtain corporate information outside the financial statements. Existing research has mainly focused on the impact of site visits on the company's stock price. Based on the site visits' event data of the Shenzhen A-share listed company from July 1, 2012 to December 31, 2018, combined with the tone analysis of the question-and-answer text of the activity record, we found that the stock prices of listed companies have significantly positive excess returns among event windows. Moreover, compared with other stocks, stocks with higher excess return during the event windows when the buy-side institutions gathered to investigate were "understocks" companies, "small-cap stocks" companies and "value stocks" companies, which has shown that the excess return of stock price was information premium. Then we found that there was also a positive correlation between the buy-side institutional visits and the next month's stock returns. After a series of robustness tests, the conclusion did not change significantly. Overall, the research results supported the stock price increase caused by the buy-side institutional visiting behavior based on the information premium. This article enriches the research field of the value of buy-side institutional activities, and may be beneficial to investors' decision-making.

This paper examines the one-way trading problem with limited trading opportunities and limited information about future prices (i.e., price range), and constructs the mathematical model via the method of competitive difference analysis. It obtains the optimal robust online trading policy, analyzes the properties of this online policy, and determines the performance guarantee of this online policy relative to the optimal offline policy, i.e., the minimal competitive difference. In addition, this paper points out all the possible worst-case scenarios for the trader. This research can not only guide the one-way trading behavior in reality when the trader needs to control transaction costs by limiting transactions, but also provide a uniform mathematical framework that connects the well-studied time series search problem and the one-way trading problem with unlimited opportunities. This unified framework facilitates the comparison between these different types of problems, enriches the conclusions of existing literatures, and lays the foundation for future researches on more complicated one-way trading problems.

The massive product experience and feedback in open innovation community could provide the inspiration of product innovation for enterprise. However, with the explosive growth of unstructured data, traditional opinion mining technology could no longer meet the real demand in terms of effectiveness and efficiency. To this end, this paper proposes a more efficient and timely method for idea extraction using deep learning algorithms. Specifically, firstly, this paper builds a multi-embedding layer CNN model with Dropout mechanism, called ME-CNN, to enhance the local feature extraction and to identify posts containing creative ideas. Then make full use of the feature of Transformer in capturing long-distance dependencies, and the strength of CNN in capturing local semantic information, we introduce a combined model, called TF-CNN, to achieve valuable sentences excluded non-creative texts. Finally, the hierarchical aggregation clustering method (HAC) is used to cluster creative ideas. Through experiments using real data, the results show that our proposed method performs well in extracting users' creativities, which can help enterprises obtain users' ideas from open innovation community more efficiently, and provide efficient decision support tools for product innovation.

Location-based social network (LBSN) has great research value with the development of location technology and smart terminal technology. At the same time commercial point-of-interest (POI) recommendation has become a research perspective, and with the increase of data volume, it has become a problem that must be solved. However, the current recommendation is more concerned with the recommendation of geographical factors, and the consideration of social relationships is relatively small. In order to improve the effect of recommendation, this paper focuses on the use of friend relationships in social relationships, and proposes a commercial point-of-interest (POI) recommendation method based on the weighted ripple net model. The method builds a preference network based on friends' relationship and interaction information, and simulates water wave diffusion in the network to consider preference enhancements at ripple overlap. With the simulating, it calculates the preference propagation of candidate point-of-interest (POI) in the target user network, and reflects the effect of geographic factors and time factors on preference propagation through the weight of the relationship to ensure the accuracy of the similarity calculation. Thus forming a recommended list of users. Finally, the data experiments on Yelp show that the algorithm can achieve better recommendation results.

Given the extended producer responsibility policy, various producers form a coalition and act as a single entity which is so-called producer responsibility organization (PRO). Considering different recyclability of different producers' end-of-life (EOL) products, differentiated collection strategies are employed for different producers, which arises as an innovatively cooperative recycling practice. Along this line, given the minimum collection rate required by government and a subsidy policy, this paper develops a collection rate decision model based on multi-choice goal programming (MCGP) to explore the characteristic functions for various producer coalitions. Furthermore, by incorporating the different unit recycling costs of different producers' EOL products, we propose a revised Shapley value to allocate the overall cost incurred by PRO. The numerical study shows that our model based on MCGP can improve the collection rate decisions for PRO in the sense that PRO can collect more EOL products with cost no more than the summation of producers' costs in non-cooperative case. On the other hand, compared to the classical Shapley value, the proposed revised Shapley value better reflects the recyclability differences for different types of EOL products, which may lead to the producers' effort on the recyclability improvement of their products in the design and manufacturing stages.

In recent years, discount and deposit pre-sale modes have become increasingly popular during hot sale period. However, consumers face the risk of uncertainty in the valuation of product, which affect pre-purchase behavior and retailer revenue. Therefore, how to choose a suitable pre-sale mode and determine the optimal pre-sale price, final payment, order quantity are important issues that retailer faced. This paper considered the purchase behavior of risk-averse strategic consumers under two pre-sale modes, constructed consumer valuation, utility function, retailer revenue models. And studied the optimal order quantity and pre-sale price decision under two pre-sale modes, and obtained the critical threshold of which pre-sale mode the retailer choose. Through comparison, analyzed the influence of different factors on the optimal profit under the two pre-sale modes. Research shows that facing strategic consumers, the optimal pre-sale price under the discount mode is less than deposit inflation mode, but the optimal order amount under the discount mode is greater than deposit inflation mode. The retailer's order cost, the deposit inflation coefficient and the percentage of the deposit affect the retailer's pre-sale decision.

Based on the channel power theory and literatures of multi-channel marketing, this paper examines the influence of channel diversity on the effects of channel power exercising by testing the moderating effects of manufacturers' channel diversity on the relationship between exercising channel power and dysfunctional channel conflict, channel cooperation. This paper applies multiple regression analysis to test the proposed hypotheses based on the questionnaire data collected from 525 manufacturers in the context of manufacturer-distributor cooperation relationship. The results indicate that channel diversity would strengthen the positive effect of coercive power exercise on dysfunctional channel conflict and positive effect of non-coercive power exercise on channel cooperation; channel diversity would weaken the negative effect of non-coercive power exercise on dysfunctional channel conflict.

Designing a reasonable medicare reimbursement method, and optimizing the resources distribution of primary care, are important ways to realize the "strong primary healthcare" reform and promote hierarchical diagnosis. We developed a game-theoretic model to analyze strategies when two primary care providers (PCPs) provide homogeneous or heterogeneous services, respectively. There are three strategies:not contract which is called baseline, patients contract with separate provider which is called strategy 1 and patients contract with medical treatment alliance which is called strategy 2. We found that contract is not always good for PCPs and patients. When the two PCPs provide homogeneous service, contract induces the medical cost decrease and the patients' perceived value increase. However, when the two PCPs provide heterogeneous service, the benefits of the PCPs and the patients decrease under strategy 1, but the opposite is true under strategy 2. Moreover, contract does not always encourage PCPs to improve service value. It has no incentive effect on PCPs with homogeneous service, but strategy 2 is a better motivator than strategy 1 with heterogeneous service. This is a socially optimal strategy.

Under the uncertain demand environment, for the problem of cyclic pickup, a collaborative optimization model based on VRP and 3D-KLP (3KL-CVRPCSO) was proposed, and a multi-stage algorithm (HPGBT) to solve the model was designed. First, a hybrid particle swarm optimization algorithm based on genetic algorithm and a heuristic orthogonal binary tree search algorithm are used to solve the optimal driving routing of different vehicle types and the optimal quantity of various types of goods loaded in the compartment. In this way, the optimal single-vehicle routing-loading plan of each type of vehicle is determined; and these plans are used as decision variables, and the actual cargo demand is the constraint condition, a new routing-loading collaborative optimization model based on actual demand is established and solved, and the result is that the number of vehicles executed according to the optimal plan of different models of bicycles. By comparing with the optimization results of foreign scholars in authoritative journals and the application of actual cases, the research and verification of two aspects have proved the feasibility and effectiveness of this method. Thus, a new method of logistics vehicle scheduling is established that takes the "routing + loading" composite plan of different models of bicycles as the decision-making unit, takes the actual demand as the constraint, and then uses the optimal combination plan to solve the uncertain demand problem.

Reducing uncertainty through reasonable investment in information handling is very essential for efficient implementation of project under uncertain conditions. In this paper, the reactive project scheduling problem with information handling cost is studied. The objective is to minimize the uncertain cost of project by deciding on the optimal investment in information handling. Firstly, the functional relationship between standard deviation of activity duration and information handling cost is abstracted. Then, the optimization model of problem is constructed and a property of the model is refined. For the NP-hardness of the studied problem, the heuristic algorithms of tabu search (TS), variable neighborhood search (VNS) and their mixed version (TVNS) are developed. Ultimately, the algorithms are tested on a randomly generated standard instances set, and the effects of several key parameters on the optimization objective are analyzed. The results show that the TVNS is better than the other two algorithms in performance, and its search efficiency is effectively improved by the improvement measure based on the property. The uncertainty cost of project decreases with the increase of the influence coefficient of information handling and the resource strength, and increases with the increase of activity weight.

The "double points" policy requires a higher automotive quality, energy conservation and emission reduction from the supply side. This paper, by considering the vehicle price, carbon remission reduction (consumers' low-carbon preference), and consumers' concerns about the mileage range in one charge from the demand side, studied the supply chain decision and coordination issue of a two-tier supply chain consisted by two carmakers and a dealer. The optimization decision and corresponding parameters of automobile supply chain was firstly analyzed, and then the coordination effects of three different supply chain contracts were explored, which was followed by a numerical analysis to verify the theoretical conclusions. The research findings are as follows:1) The improvement of consumers' low-carbon preference and consumers' concern about the mileage range in one charge can promote automobile manufacturers to further reduce their carbon emissions and improve the mileage range in one charge. 2) The increase of new energy vehicles' points ratio can restrain the production of fuel vehicles, but cannot promote the improvement of emission reduction efforts. There is a positive correlation between new energy vehicles' score accounting coefficient and the promotion effects to new energy vehicle industry. 3) Neither cost sharing nor revenue sharing contracts can completely eliminate the "double" marginal effect (which can only achieve Pareto improvement), while the two-part tariff contract can fully coordinate automobile supply chain.

The research on the optimization of low-carbon multimodal transportation path under uncertainty can present important theoretical significance, which can satisfy the practical needs of market, economy and environmental protection. This paper establishes a hybrid robust stochastic optimization model aiming at the problem of path optimization with double uncertainty of demands and carbon trading prices. Afterwards, a catastrophe adaptive genetic algorithm based on Monte Carlo sampling is designed and tested for validity. The multimodal transportation schemes and costs under different modes are compared and the impacts of uncertain parameters are analyzed by a 15-node multimodal transportation network case. The results reflect that:1) Robust optimization with uncertain demands could increase the total cost of multimodal transportation due to the pursuit of stability; 2) The regret value of robust optimization will affect the total cost of multimodal transportation. However, the randomness of carbon trading prices should also be considered comprehensively in the double uncertainty mode; 3) Carbon trading policy is an effective carbon emission reduction policy. The increasing randomness of carbon trading price is likely to reduce the total cost. Based on the above research results, making full use of the incentive effect of carbon trading policy to set the uncertainty of carbon trading prices and comprehensively considering the relationship between the maximum regret value of uncertain demands and costs can effectively improve the efficiency and environmental benefits of multimodal transportation under dual uncertainties.

During the loading and unloading operations of the actual port, the quay crane and truck operate a container in a single cycle at a time, with an empty load rate of 50%, while a dual-cycling operates an import container and an export container, which doubles the efficiency. However, due to the limitation of the ship's stowage plans and the hatch covers, the quay crane cannot always perform efficient dual-cycling operations, and the trucks connected to the quay cranes cannot always achieve two-way heavy load transportation. Therefore, in order to improve port efficiency, a two-stage cooperative optimization model of quay cranes and trucks with dual-cycling operations in same ship-bay is established in this paper. At first, an optimization model of the quay crane loading and unloading sequence that satisfies the constraints of the hatch covers and the number of operation cycles is constructed, and a genetic algorithm (GA) based on matrix coding is designed to solve the problem. Then, a truck path optimization model with the objective to minimize the truck travel time was given and solved by immune genetic algorithm (IGA) based on vector distance. The analysis of the example shows that the model can effectively reduce the number of quay crane operation cycles and truck transportation time, achieve resource reuse and configuration optimization, thereby improving the efficiency of terminal operations.

Based on the covariate balancing propensity score method introduced by Imai and Ratkovic (2014), this paper proposes a new method to combine the GMM-LASSO type estimation method with covariate balancing approach. The proposed method not only utilizes the property of covariate balancing, but also solves the problem of how to select covariates based on data. Also, it is shown that the proposed estimator is consistent and simulations show that under the condition of sparsity, the proposed method indeed can significantly reduce the median of the absolute errors of the average treatment effect. Finally, the proposed method is applied to the data from the Tuscany region of Italy in the early 2000s to study whether temporary work agency mechanism helps workers to find a stable job in the future.

This paper discusses the limitation of exclusion restrictions on sample selection model estimation. We propose a two-stage semiparametric estimator for the sample selection model by the integrated least squares method. In the first stage, we obtain the propensity score by kernel estimation, based on which we construct the optimization objective function by the integrated least squares method in the second stage. Compared with existing semiparametric estimation methods of sample selection model, our method relaxes the exclusion restrictions. Besides, our method improves the estimation efficiency by using infinite number of moment conditions, thus has more practical value in the empirical application.

Crude oil price forecasting received much attention due to its importance and the non-linearity and complexity of crude oil price series. Using the useful information provided by sub-model to generate comprehensive prediction, forecast combination aims to improve the forecasting accuracy. It is significant how to efficiently generate many diverse sub-models and weighting vector. In this paper, we first introduced various feature selection techniques, including filter, wrapper and embedded methods to determine the key factors affecting crude oil prices. Then, individual models are constructed by incorporating feature selection methods with multiple linear regression, artificial neural network and support vector regression model. Finally, a dynamic particle swarm optimization algorithm is proposed. The algorithm can search for the optimal weighting vector and capture the dynamic changes of weighting series. Experimental results show that the proposed dynamic forecast combination model can reduce the computational complexity and improve the forecasting performance.

To demonstrate the effectiveness of the long short-term memory (LSTM) deep neural network model on short-term passenger flow prediction of high-speed railways, the paper presents the characteristics of the departing passenger flow in different stations based on the real-record passenger flow data of Beijing-Guangzhou high speed railway, from January, 2010 to December, 2015. The passenger dataset is standardized and framed for the LSTM model, considering the expectation input format of LSTM layers and the characteristics of the data. LSTM model is fitted with tuning and regulating all the parameters necessary in the model. Then the fitted LSTM model is applied to forecast the short-term departing passenger flow of Beijing-Guangzhou high speed railway. The influence of important parameters in the LSTM model on the prediction accuracy is analyzed, and the comparison with other representative passenger flow forecast models is conducted. The results show that the LSTM model can achieve a better performance compared to other models. One-step forward passenger flow prediction errors valued by mean absolute percentage error (MAPE) are 7.36%, 7.33%, 8.03% respectively for Chenzhou west station, Hengyang east station and Shaoguan station. The parameters in the LSTM model such as the number of network layers, the number of neurons and the input format of the reshaped passenger dataset have a great influence on the prediction accuracy. The convergence speed and prediction accuracy can be improved as the number of network layers and neurons increase reasonably while LSTM model can performance better as the input size of passenger flow is 7, affected by the periodic characteristics of passenger flow.

Although the cross-efficiency method solves the efficiency overestimation problem of the traditional DEA method, it loses the advantage of setting target for improving efficiency. Therefore, by analyzing the evaluation mechanism of cross-efficiency method, this paper identifies the source, which hinders DMU to set the improvement target, and constructs DEA cross-efficiency target setting approaches in three different decision-making contexts, i.e., improving outputs only, improving inputs only, and improving inputs and outputs at the same time, so as to search the target DMU with the optimal cross-efficiency in the scope of production possibility set. On this basis, this paper reveals the rules of target setting for different DMUs, and studies the impact of traditional cross-efficiency method on setting target, and then the target setting approach is further extended to the decision-making context where all DMUs set their targets together. Finally, the validity and practicability of the new approaches are illustrated by case analysis and method comparison.

Based on the user equilibrium theory, this paper firstly presents a variational inequality (VI) model to characterize the equilibrium conditions for the dynamic traffic assignment problem where the users consider route-and-departure-time choice simultaneously. Then, we propose a fast projection gradient method with relaxation and extrapolation to solve the constructed VI model. The theoretical convergence of the new method is established based on the optimization theory. Finally, applying the proposed method to solve the actual dynamic traffic assignment problems on several different scale transportation networks. According to some numerical simulations, the effectiveness of the model and the advantages of the new method are illustrated, and the departure flow pattern of all routes for each tested transportation network can be obtained under user equilibrium state. This work is helpful to achieve and simulate the dynamic traffic assignment rapidly in the transportation network and provide timely and effectively feedback to the traffic management department, which can provide assurance for taking management measures and travel planning.