On January 14, the ZJU Doctoral Innovation Forum and the 2023 ZIBS Doctoral Forum were successfully convened at the ZIBS. The forum aimed to provide a high-quality academic exchange platform to enhance the quality of doctoral education. It covered various disciplines such as business management, applied economics, finance, computer science, and technology, offering doctoral students the opportunity to present their research work and engage in various discussions. Participants from Peking University, Tsinghua University, Zhejiang University, and Renmin University of China attended the forum, sharing their research achievements and promoting interdisciplinary integration through the exchange and collision of ideas. ZIBS Dean Ben Shenglin delivered a welcome address at the event, and the forum was moderated by ZHOU Ziyue, a doctoral student from ZIBS Class of 2023.

During the welcome remarks from Professor BEN Shenglin, he extended a warm welcome to all participants and mentioned the "5i" philosophy of ZIBS which contains the "Interdisciplinary" that echos with this forum inviting students and professors from diverse backgrounds. He also highlighted the original intention of organizing the forum from the perspective of doctoral student training, emphasizing the forum as a crucial platform for academic exchange and expanding scholarly perspectives.

CHENG Yuxiang, from Peking University, presented his research on "Who Should Finance the Supply Chain? Impact of Accounts Receivable Mortgage on Supply Chain Decision". His study focuses on the interaction of enterprise operational choices under bank credit policies in both cooperative and non-cooperative scenarios. He specifically analyzed how short-term loans provided by banks to retailers affect supply chain decisions. Taking into account different bank credit policies and the operational capabilities of retailers and manufacturers, his research revealed the optimal strategies and the fact that using short-term bank loans can improve supply chain efficiency. Moreover, supported by empirical data, his findings showed that financially stronger companies tend to increase orders, and the optimal purchasing decisions depend on the retailer-manufacturer share ratio (RMR). In cases where RMR values are lower, there is a greater inclination to use trade credit, and capital-constrained supply chains benefit more from bank credit.

In the era of data-driven technologies, federated learning, as an advanced approach to distributed learning, plays a crucial role in handling non-independent and identically distributed (Non-iid) data scattered across multiple clients. The report presented by FANG Liyu from Zhejiang University, titled "Federated Learning with Non-IID Data," employs clustering federated learning. This involves utilizing clustering algorithms to group clients with similar data distributions, effectively enhancing training efficiency and model performance. Furthermore, the study delves into personalized aggregation processes, utilizing customized aggregation strategies to optimize the performance of specific client models, thereby strengthening model adaptability. The introduction of federated knowledge distillation techniques facilitates the transfer of model knowledge among clients, overcoming data distribution disparities without sharing actual data, thereby further enhancing the effectiveness and data security of federated learning.

The tendency of entrepreneurs towards social responsibility is increasing day by day, and the effective and recognized ways of performing responsibility need to be discussed. MA Shanzi' study explores the mechanism of entrepreneurs' social responsibility behavior on the promotion of consumers’ willingness to participate in service recovery after artificial intelligence service failure. The mediating role of emotional trust and cognitive trust and the moderating role of the type of service failure have also be focused. The expected empirical result is that both active and reactive entrepreneurs' social responsibility behavior can positively affect consumers’ willingness to participate in service recovery, but their effects will be varied in different situations. At the same time, consumers' emotional trust and cognitive trust play a mediating role in the promotion of consumers' willingness to participate in service recovery through active and reactive entrepreneurs' social responsibility behaviors respectively. When considering the type of service failure, reactive entrepreneurs' social responsibility behavior is more effective in dealing with technical failures, while active entrepreneurs' social responsibility behavior is more effective in dealing with service failure that social norm violations.

With the theme of "The Impact of COVID-19 on Consumer Credit Changes", WANG Renfeng, from Zhejiang University, explores the changes in consumer habits and the trends of forward consumption behavior during the pandemic. It particularly focuses on internet consumer credit based on financial technology. Through empirical analysis, Wang revealed the connection between financial technology consumer credit and traditional credit during the pandemic, as well as the factors contributing to changes in user credit habits. This research provides valuable insights into the relationship between the COVID-19 pandemic and the market's consumption and credit dynamics.

Asynchronous Consensus is a crucial mechanism in distributed systems for handling information asynchrony among nodes. In a presentation titled "Introduction to Asynchronous Consensus," WU Di from Zhejiang University introduced key concepts, the working principles, and the application of asynchronous consensus in distributed computing. The presentation also provided a brief overview of the definition of asynchronous consensus, compared it with synchronous consensus, and highlighted its advantages in the face of challenges like network latency and node failures. The talk delved into the design and implementation of asynchronous consensus algorithms, emphasizing their practical applications in systems such as blockchain technology and distributed databases. Finally, WU Di underscored the importance of asynchronous consensus in enhancing system fault tolerance, scalability, and flexibility, while also outlining future research directions in this field.

"Motivated Extrapolative Beliefs ", introduced by YANG Siyuan, doctoral student from TsinghuaUniversity, investigated the relationship between investors' prior gains or losses and their adoption of extrapolative beliefs. His findings indicated that  investors who have experienced losses are more likely to form overly optimistic expectations, while those who have made profits tend to become overly pessimistic. This finding also confirms the theory of motivated beliefs. Furthermore, he suggested that the interaction between book profits and losses and extrapolative beliefs could lead to serious mispricing of stock prices, with monthly returns reaching up to 1%. These motivated extrapolative beliefs not only affect investors' expectations but are also closely related to their trading behavior. Individual investors are particularly susceptible to these belief distortions, while institutional investors are better able to avoid the phenomenon of overpricing associated with overly optimistic extrapolative beliefs.

ZHANG Qiongwen, from Zhejiang University, focuses on the cognitive perspective and adopts an ambidexterity framework to explore the impact of perceived opportunities and threats of global value chain decoupling on firm innovation response. Her research is based on panel data of 955 Chinese high-tech listed companies between 2017 and 2021. She found that there are different relationships between the ambidexterity perception of opportunities and threats and firm innovation: the balanced ambidexterity perception of opportunities and threats shows an inverted U-shaped relationship with innovation performance, while the combined ambidexterity perception is positively related to innovation performance. Besides, technological decoupling and the level of internationalization act as moderating factors. By applying a comprehensive framework of the ambidexterity perception of opportunities and threats, her research enhances understanding of how firms interpret their external environment and make innovation decisions, providing some insights for firms to adapt and enhance their innovation in the context of reshaping global value chains.

ZHANG Shuhuai, from Tsinghua University, presented his research titled "Trust-building in Charitable Giving: A Field Experiment",  which explores the role of trust in alleviating financial frictions in the charitable donation industry. Through a randomized field experiment design, the study examines how trust-building information increases charitable participation, as well as the treatment heterogeneity across different types of donors and recipients. The results indicate that such strategies could effectively mitigate market failures in the charitable giving sector, enhance industry efficiency and social welfare.

The strategic focus and actions are crucial for a company's growth. ZHANG Yichu, from Zhejiang University, presented a report titled "Duplicity is Always Evil? The Effect of Inconsistency between Attention and Action on SADI Firms' Performance in China". This report focuses on Chinese SADI (Specialized, Refined, Unique, and Innovative) companies, particularly examining how the inconsistency between strategic attention and action influences their performance. The study specifically considers the impact of strategic orientation in terms of specialization, refinement, uniqueness, and innovation capabilities. Using a sample of 1,222 SADI listed companies from 2012 to 2021, the research reveals a significant inverted U-shaped relationship between the inconsistency of strategic attention and action and company performance. Additionally, she suggested that company age and industry concentration are crucial boundary conditions that increase the steepness of this inverted U-shaped relationship. Through innovative theoretical exploration and empirical research, her study provides insights into the advantages and disadvantages of the inconsistency between strategic attention and action in SADI companies under the four-dimensional strategic orientation.

Graph data plays a crucial role in fields like finance and healthcare. In reality, these data are often distributed across different entities, and centralized collection can lead to sensitive privacy issues. Consequently, federated learning, as an emerging paradigm of distributed learning, has garnered widespread attention in recent years. When it comes to processing graph data, the unique structural characteristics of graphs present distinct challenges. ZHOU Pengyang，from Zhejiang University, brought his report on "Federated Graph Learning", mainly introducing the latest developments in federated graph learning at both the graph and sub-graph levels, and discussing how to effectively address these structural challenges, aiming to maximize the value and potential of graph data while ensuring data security.

As a crucial determinant of urban livability, the importance of access to high-quality green spaces has long been recognized for achieving sustainable urbanization. In urban areas, higher values are usually placed in residential properties with higher accessibility to green spaces. ZHU He, from Zhejiang University, using housing transaction data from as many as 3388 residential communities in Shanghai coupled with high- resolution satellite data of urban green spaces, comprehensively examined the relationship between residential property values and the accessibility of both community-owned and public green spaces. Her empirical findings provide valuable guidance for real estate developers and local governments in valuing environmental amenities and urban planning in the context of a residential housing market.