Livestream e-commerce enables continuous consumer engagement across the day, yet the role of time-of-day in shaping consumer behavior remains underexplored. This study examines whether shopping during nighttime hours is associated with differences in impulsive engagement, post-purchase outcomes, and consumer retention. Using 982,746 session-level observations from Douyin livestream commerce, we compare consumer behavior during a daytime alertness window (10:00–18:00) and a nighttime low-alertness window (22:00–06:00). Descriptive analyses, Welch’s t-tests, and multivariate regression models are employed to evaluate circadian differences across engagement, sales, return behavior, and retention-related measures. Results show that nighttime livestream sessions generate significantly higher impulsive engagement and sales volume but are associated with lower retention and engagement stability. In contrast, return rates do not increase meaningfully for nighttime purchases once economic and contextual controls are introduced, indicating that heightened impulsivity does not translate into higher regret-driven returns. These findings suggest that time-of-day systematically shapes both short-term commercial performance and longer-term engagement outcomes in livestream commerce. By introducing circadian timing as an explanatory dimension, this study extends applied economic analyses of digital markets and offers implications for platform strategy, performance evaluation, and time-sensitive policy considerations in 24-hour online retail environments.

Show more

This research explored the factors that determine housing prices and their regional differences in Beijing's second-hand housing market using complete transaction data from 2024 covering all 16 administrative areas, adopted the Hedonic Price Model and carried out separate Ordinary Least Squares (OLS) regressions for different sub-markets, the results showed a clear two-part difference in how prices were formed which was described as being "resource-driven" compared to "residence-driven", in central urban areas, housing prices were mainly influenced by proximity to top-quality public resources and there was a large price increase for apartments having three or more bedrooms, on the other hand, in suburban zones, market evaluations concentrated more on fundamental living features showing a definite preference for south-facing directions and medium to large sized units, moreover, the additional impacts of important structural elements varied significantly among these regions, the discoveries offered strong empirical proof to back location-specific regulatory measures for the existing housing supply in mega-cities such as Beijing and also provided useful information for market players including house buyers, builders and policymakers enabling them to make better-informed decisions in line with the main value influences in each market section.

Show more

Time series of financial asset returns typically exhibit pronounced non-ergodicity and heavy-tailed, spiky distributions. Traditional mean–variance models and expectation-based Deep Reinforcement Learning (DRL) approaches struggle to effectively capture distributional shifts induced by exogenous logical shocks. To address this challenge, this paper proposes a risk-constrained distributional reinforcement learning framework that integrates large language model–based logical reasoning with dynamic graph dependencies (LLM-G-DRL). At the perception level, rather than relying on conventional sentiment polarity classification, this study introduces large language models that support Chain-of-Thought (CoT) reasoning (e.g., DeepSeek-R1) to construct a Bayesian logical belief updating mechanism. This mechanism maps unstructured financial texts into high-dimensional latent logical states embedding causal transmission paths, thereby correcting predictive biases arising from exclusive dependence on historical price and volume data. At the structural level, to characterize the nonlinear contagion of systemic risk, the framework abandons the assumption of static adjacency matrices and employs dynamic graph attention networks (Dynamic GAT) to reconstruct time-varying topological dependencies among assets, enabling explicit modeling of risk propagation channels. At the decision-making level, the portfolio optimization problem is formulated as a Constrained Markov Decision Process (CMDP). Implicit Quantile Networks (IQN) are adopted to approximate the full probability distribution function while preserving higher-order moment information. Furthermore, based on Lagrangian duality theory, a hard constraint on Conditional Value at Risk (CVaR) is introduced, transforming tail-risk control into a dynamic penalty mechanism governed by dual variables. Theoretical analysis demonstrates that, through a closed-loop design combining “logical priors, structural contagion, and distributional decision-making,” the proposed framework exhibits stronger mathematical robustness than traditional point-estimation models. It effectively identifies and avoids tail losses under extreme market conditions, offering a statistically interpretable new paradigm for intelligent asset allocation in non-stationary markets.

Show more

In the context of the digital economy and the rapid restructuring of financial business models driven by new productive forces, implementing digital transformation strategies has become a critical measure for commercial banks to enhance service efficiency and adjust credit systems. This study uses longitudinal data of Chinese commercial banks from 2011 to 2023 and employs text analysis methods for empirical research. By constructing an evaluation system for the degree of bank digitalization based on text mining techniques, the study explores the mechanisms through which digital transformation influences the development of consumer credit. Furthermore, a comprehensive consumer finance evaluation index is developed using the information entropy method, providing a quantitative tool to assess both the progress of bank digitalization and the development of consumer credit. The findings indicate that the digitalization of banks significantly promotes the growth of consumer credit, primarily by improving bank profitability, expanding credit issuance, and strengthening risk management effectiveness. The study also finds that these effects vary across regions with different levels of economic development and among different types of banking institutions. Robustness tests and instrumental variable methods confirm the reliability of these conclusions. This research not only clarifies the internal mechanisms through which bank digital transformation affects consumer credit from a micro-level perspective but also provides empirical support for commercial banks in formulating transformation strategies and enhancing consumer finance services.

Show more