Empirical Results Research Articles

Breathing life into equity: How air pollution influences corporate pay gap

Using a dataset encompassing 19,229 Chinese listed firm-year observations from 2014 to 2020, this study investigates the relationship between air quality and the pay gap between executives and employees. Our empirical results indicate that companies tend to reduce the pay gap in response to air pollution. These results exhibit robustness through various methodological approaches: employing thermal inversion as an instrumental variable (IV) for IV regressions, adopting quantile regressions, utilizing alternative metrics for assessing air pollution and corporate pay gap, expanding control variables, excluding firms undergoing reorganization, and applying firm-level clustering. This effect is especially strong in corporations with greater public scrutiny, strong corporate governance, and fewer financial constraints. Moreover, such strategic adjustments in compensation policies not only help firms retain valuable human resources, but also ultimately enhance their green innovation and ESG performance, as diminished pay gap contribute to fostering increased employee engagement and motivation within the work environment. Our research contributes to environmental management by demonstrating how air quality can be a pivotal factor in corporate compensation strategies. The findings provide actionable insights for executives and policymakers, advocating for the integration of environmental considerations into core business frameworks.

Public Energy R&D Spending and Green Energy for Sustainable Development: COP28 Perspective of G7 Economies

Achieving sustainable development goals (SDGs) is one of the top priorities for countries worldwide, and the recent COP28 has focused mainly on achieving sustainable development via clean energy investment financing. However, countries and regions still struggle to find various policy-level indicators to achieve sustainability engagement. In this context, finding the right combination of policy indicators becomes crucial. Therefore, the current study seeks to identify the key drivers of sustainability for the G7 economies by utilizing data from 1990–2019. This study employed the cross-sectional autoregressive distributed lag (CS-ARDL) panel time series method of estimating primary results and a novel sustainable development index as the dependent variable. The study revealed that public energy R&D budgets, increasing income levels, improving the political risk index, and promoting clean energy are important for achieving sustainable development in the G7 countries and that the depletion of natural resources has an adverse influence on sustainability. These empirical results are supported by various economic and environmental theories, i.e., the environmental Kuznets curve, the innovation theory of sustainability, the Dutch disease hypothesis, and the political economy of development and energy transition theory. Based on these findings, energy investment is recommended, especially for clean energy projects, and increasing income levels via sustainable economic activities to achieve sustainability in the G7 economies.

Divert More Attention to Vision-Language Object Tracking.

Multimodal vision-language (VL) learning has noticeably pushed the tendency toward generic intelligence owing to emerging large foundation models. However, tracking, as a fundamental vision problem, surprisingly enjoys less bonus from recent flourishing VL learning. We argue that the reasons are two-fold: the lack of large-scale vision-language annotated videos and ineffective vision-language interaction learning of current works. These nuisances motivate us to design more effective vision-language representation for tracking, meanwhile constructing a large database with language annotation for model learning. Particularly, in this paper, we first propose a general attribute annotation strategy to decorate videos in six popular tracking benchmarks, which contributes a large-scale vision-language tracking database with more than 23,000 videos. We then introduce a novel framework to improve tracking by learning a unified-adaptive VL representation, where the cores are the proposed asymmetric architecture search and modality mixer (ModaMixer). To further improve VL representation, we introduce a contrastive loss to align different modalities. To thoroughly evidence the effectiveness of our method, we integrate the proposed framework on three tracking methods with different designs, i.e., the CNN-based SiamCAR (Guo et al. 2020), the Transformer-based OSTrack (Ye et al. 2022), and the hybrid structure TransT (Chen et al. 2021). The experiments demonstrate that our framework can significantly improve all baselines on six benchmarks. Besides empirical results, we theoretically analyze our approach to show its rationality. By revealing the potential of VL representation, we expect the community to divert more attention to VL tracking and hope to open more possibilities for future tracking with diversified multimodal messages.

FEVERLESS: Fast and Secure Vertical Federated Learning Based on XGBoost for Decentralized Labels

Vertical Federated Learning (VFL) enables multiple clients to collaboratively train a global model over vertically partitioned data without leaking private local information. Tree-based models, like XGBoost and LightGBM, have been widely used in VFL to enhance the interpretation and efficiency of training. However, there is a fundamental lack of research on how to conduct VFL securely over distributed labels. This work is the first to fill this gap by designing a novel protocol, called FEVERLESS, based on XGBoost. FEVERLESS leverages secure aggregation via information masking technique and global differential privacy provided by a fairly and randomly selected noise leader to prevent private information from being leaked in the training process. Furthermore, it provides label and data privacy against honest-but-curious adversaries even in the case of collusion of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$n - 2$</tex-math></inline-formula> out of n clients. We present a comprehensive security and efficiency analysis for our design, and the empirical results from our experiments demonstrate that FEVERLESS is fast and secure. In particular, it outperforms the solution based on additive homomorphic encryption in runtime cost and provides better accuracy than the local differential privacy approach.

A binary choice model for adoption of an emerging travel mode with unique service features

In the era of emerging technologies, the transportation system is witnessing the introduction of innovative mobility services, such as autonomous vehicles, which possess unique service features that cannot be seen from conventional travel modes. To facilitate the understanding of the behavioral impacts and the adoption of innovative mobilities, a novel binary weibit model with an oddball alternative (BW-O) is developed for the binary choice between conventional and emerging mobilities. The BW-O model explicitly considers the unprecedented (or unique) service features of emerging travel modes while retaining the closed-form choice probability. This study empirically illustrates the application of the BW-O model in the mode choice context. The desirable properties of the BW-O model compared to the existing binary choice models are discussed both theoretically and empirically. In the binary mode choice problem with an emerging travel mode, the unique service features of the emerging mode can lead to the “oddball” effect and “superstar” effect, which play a critical role in the travel behavior and mode adoption. The BW-O model inherently captures both effects by considering a higher perception variance for the emerging mode and asymmetric choice probabilities between different modes. Thus, as revealed by the empirical results, the BW-O model outperforms the basic binary weibit model in terms of both model fit and predictive power. The developed BW-O model is not only applicable to the mode choice problem in transportation systems, but also opens a door for more general class-imbalanced binary choice contexts where an alternative has additional attractiveness and asymmetric choice probability.

Time-frequency Co-movement and Cross-quantile Connectedness of Exchange Rates: Evidence from ASEAN+3 Countries

This study investigates the time-frequency co-movement and cross-quantile connectedness of exchange rates. Using wavelet coherence and cross-quantile methods, we examine ASEAN+3 countries’ time-frequency co-movement, quantile spillover effects, and network connectedness of the exchange rate markets. Our empirical results are as follows: significant co-movement heterogeneity exists across countries over different frequency bands. Moreover, the Chinese Yuan (CNY), Japanese Yen, and South Korea Won are desirable sources of diversification for other currencies across different investment horizons. CNY and JPY exhibit good regional safe haven currency attributes in different investment horizons. Overall, these findings suggest ways for currency authorities to maintain exchange rate stability and investor portfolio decisions.

Sharpness-Aware Lookahead for Accelerating Convergence and Improving Generalization.

Lookahead is a popular stochastic optimizer that can accelerate the training process of deep neural networks. However, the solutions found by Lookahead often generalize worse than those found by its base optimizers, such as SGD and Adam. To address this issue, we propose Sharpness-Aware Lookahead (SALA), a novel optimizer that aims to identify flat minima that generalize well. SALA divides the training process into two stages. In the first stage, the direction towards flat regions is determined by leveraging a quadratic approximation of the optimization trajectory, without incurring any extra computational overhead. In the second stage, however, it is determined by Sharpness-Aware Minimization (SAM), which is particularly effective in improving generalization at the terminal phase of training. In contrast to Lookahead, SALA retains the benefits of accelerated convergence while also enjoying superior generalization performance compared to the base optimizer. Theoretical analysis of the expected excess risk, as well as empirical results on canonical neural network architectures and datasets, demonstrate the advantages of SALA over Lookahead. It is noteworthy that with approximately 25% more computational overhead than the base optimizer, SALA can achieve the same generalization performance as SAM which requires twice the training budget of the base optimizer.

An Entrained Flow Biomass Gasification Technology with the Fluidized Bed Concept for Low-carbon Fuel Production

For accomplishing the vision of building of a net-emission zero society, low-carbon fuels such as e-fuel, Sustainable Aviation Fuel (SAF), and chemical products generation, plays a significant role. Especially, among the low carbon fuels, SAF is the most crucial. Ammonia and electric aircraft are under way as alternatives to fossil-derived jet fuel (kerosene). However, none of these have yet found their way to commercialization. Among the SAF production technology, biomass gasification and Fischer-Tropsch (FT) synthesis technology are one of the lowest CO2 emission processes, according to Life Cycle Assessment (LCA) analysis. However, at present there are some issues about biomass grinding, tar, ash treatment, gas purification and wastewater. We, at Mitsubishi Heavy Industries (MHI) R&D laboratory have developed entrained bed gasification, adopting the fluidized bed concept in which biomass particles are recirculated in the gasifier without fluidizer. Principally desired outcomes of this development was to reduce the grinding power, simplify the structure by using atmospheric pressure process, and lower tar concentration while maintain suitable temperature to prevent ash from melting inside gasifier. Empirical and actual results showed that the developed gasifier can obtain high carbon conversion ratio and low tar as compared to the traditional gasification processes for low-carbon fuel production and produce reliably stable syngas for low-carbon fuel synthesis with single gasifier. Based on the pilot plant operation results, effectiveness of moderate (1223 – 1323 K) temperature gasification with this concept has been demonstrated too. As a conclusion, development needs and expectation of gasification technology for contributing to carbon neutral society are mentioned.

Conflicting Results and Statistical Malleability: Embracing Pluralism of Empirical Results

Abstract Conflicting results undermine making inferences from the empirical literature. So far, the replication crisis is mainly seen as resulting from honest errors and questionable research practices such as p-hacking or the base-rate fallacy. I discuss the malleability (researcher degrees of freedom) of quantitative research and argue that conflicting results can emerge from two studies using different but plausible designs (e.g., eligibility criteria, operationalization of concepts, outcome measures) and statistical methods. I also explore how the choices regarding study design and statistical techniques bias results in a way that makes them more or less relevant for a given policy or clinical question.

Understanding how digital infrastructure construction could promote sustainable development

Progress made in environmental protection may negatively affect regional socioeconomic development, reducing or impairing the ability of local economic systems to defend against external shocks, culminating in weakened economic resilience. Implementing carbon abatement programs without risking economic resilience is therefore an urgent problem for the international community, especially for its emerging national economies. Digital infrastructure construction (DIC), as a driving force of technological progress and structural transformation, may offer a viable solution to that problem. Here, we used county-level data for China, the biggest emerging economy in the world, to investigate whether and how Smart City Pilot policy (SCP, a policy close to DIC) has lowered regional carbon emissions and influenced economic resilience. We find that the SCP could significantly promote carbon abatement goals, and this result is robust under several tests. Further, chain-based mediating effect analysis revealed that the SCP's beneficial impact could have arisen by first promoting innovation and then productivity, and public concern acting as a stressor that pressures officials to engage in environmental governance. Furthermore, our results show the SCP is capable of bolstering regional economic resilience, and could even offset or negate the adverse effects of strict environmental regulation on that resilience. However, the positive effects of the SCP on carbon abatement in China's developed counties, or ones located in a resources-based city, were not significant. Altogether, our empirical results highlight how DIC can serve as a way to help achieve sustainable development, but more studies on its contributing role are clearly needed.

The Impact of Agglomeration on CO2 Emissions in China's Transport Sector: A Spatial Econometric Analysis

The long-term processes of urbanization and industrialization have led to the agglomeration of population and industry, fostering economic development while introducing opportunities and challenges for carbon reduction in transport. This paper integrates the Stochastic Impacts by Regression on Population, Affluence, and Technology Model with the Spatial Durbin Model to assess the effects of population agglomeration and industrial agglomeration on transport carbon dioxide emissions. The empirical results show that a 1% increase in population agglomeration decreases local transport carbon dioxide emissions by 1.7065% and generates a spillover effect of 1.0542% in surrounding areas. In contrast, industrial agglomeration increases regional transport carbon dioxide emissions by an average of 0.3309% without significant spillover effects. Furthermore, economic agglomeration exhibits an N-shaped relationship with transport carbon dioxide emissions, reflecting the dual influences of the "economic effect" and the "congestion effect". Mechanism analysis reveals that both types of agglomeration can modulate the impact of infrastructure development on transport carbon dioxide emissions, suggesting that effective infrastructure planning can help alleviate the negative environmental impacts. This study provides a spatial mode for understanding the synergistic effects of population planning, industrial development, and environmental improvement, offering significant reference value for policymakers in the decision-making related to low-carbon transport development.

The Influence of Urban New Energy Development Orientation on New Energy Technology Innovation in Firms

Unlike existing energy policies that adopt a macro-level perspective, this paper examines the impact of China’s new energy demonstration cities (NEDC) policy on micro-agents. Using longitudinal data covering Chinese publicly listed firms in new energy industries, the empirical results show that the NEDC policy significantly increases firm’s new energy patents in pilot cities compared to those in non-pilot cities. We examine the potential channels via industrial upgrading effect, relative scale effect, direct incentive effect, and factor aggregation effect. In addition, we find that the NEDC policy performs better in areas with more governmental new energy attention and higher infrastructure levels, and state-owned businesses are more proactive.

SEA++: Multi-Graph-Based Higher-Order Sensor Alignment for Multivariate Time-Series Unsupervised Domain Adaptation.

Unsupervised Domain Adaptation (UDA) methods have been successful in reducing label dependency by minimizing the domain discrepancy between labeled source domains and unlabeled target domains. However, these methods face challenges when dealing with Multivariate Time-Series (MTS) data. MTS data typically originates from multiple sensors, each with its unique distribution. This property poses difficulties in adapting existing UDA techniques, which mainly focus on aligning global features while overlooking the distribution discrepancies at the sensor level, thus limiting their effectiveness for MTS data. To address this issue, a practical domain adaptation scenario is formulated as Multivariate Time-Series Unsupervised Domain Adaptation (MTS-UDA). In this paper, we propose SEnsor Alignment (SEA) for MTS-UDA, aiming to address domain discrepancy at both local and global sensor levels. At the local sensor level, we design endo-feature alignment, which aligns sensor features and their correlations across domains. To reduce domain discrepancy at the global sensor level, we design exo-feature alignment that enforces restrictions on global sensor features. We further extend SEA to SEA++ by enhancing the endo-feature alignment. Particularly, we incorporate multi-graph-based higher-order alignment for both sensor features and their correlations. Extensive empirical results have demonstrated the state-of-the-art performance of our SEA and SEA++ on six public MTS datasets for MTS-UDA.

Assessing performance of water supply operation of water companies in China: A union dynamic multi-activity network data envelopment analysis

Urban growth and the increasing urban water demand highlight the importance of performance analysis in the water supply industry. Both water supply companies (WSCs) and water supply and sewage companies (WSSCs) produce and deliver potable water for city dwellers. Water authorities use benchmarks to monitor the two types of companies’ operations. The internal operation process and heterogeneities of observations were not considered for relative performance evaluation by conventional DEA models. To overcome these limitations, this study proposes a union dynamic multi-activity network data envelopment analysis (DMANDEA) model to extend static network DEA to a more general case for dynamic efficiency assessment of the water supply operations of water companies. For the first time, the model allows for the decomposition of the water supply operation into water production (WP) and water consumption (WC) substages and performance comparisons for water companies by considering their heterogeneity simultaneously. Specifically, by using a non-radial directional distance function (DDF), the inefficiencies for substages and the entire system can be both assessed dynamically. This model is applied to a sample of 246 water companies in China in 2016–2018. The overall, period, and substages’ group efficiencies, meta efficiency, and technology gap ratio (TGR) for 179 WSCs and 67 WSSCs are all provided. The empirical results reveal that: (1) the average group efficiencies of WSCs and WSSCs are 0.4567 and 0.6611, respectively. The meta efficiencies of WSCs and WSSCs are 0.4523 and 0.4703, respectively. The TGR of WSCs (0.994) is greater than that of WSSCs (0.7067). (2) WSSCs perform better than WSCs do, which suggests that economies of scope do exist in WSSCs. However, WSCs possess the best technology for water supply operations because of their high TGR. (3) The average meta efficiency for all the observed water companies (0.4572) is not optimistic, which correlates more with inefficiency in the WP stage. The causes stem from the natural monopoly, lack of competition, and relatively lower water tariffs. The Chinese water supply industry requires diversification of ownership and operation modes, and gradual increases in water tariffs. (4) Water companies in the eastern region outperform those in other regions. Regional cooperation, such as technology sharing, the transfer of managerial experience, and information exchange, needs to be emphasized.

Leveraging multimodal deep learning for natural disaster event classification and its damage severity analysis through social media posts

Accurate and timely information is essential for coordinating an effective disaster response. Traditional methods have struggled to efficiently categorize disaster events and assess damage severity due to the variety and complexity of data sources. Previous research has focused on specific tasks, such as information gathering or humanitarian assistance, but has not adequately addressed the assessment of disaster damage severity. This paper proposes a hybrid learning model to improve disaster event classification and damage severity identification. The model combines image and text data in a cooperative way, using ResNet50 to extract features from images and a LSTM with attention mechanism to learn sequences from text. This combination allows for a more contextual and informative representation of the input data. Compared to existing approaches, the proposed multimodal approach achieves significantly better results in disaster event classification. Apart from the proposed model also shows promising outcome for damage severity of disaster. These advancements are especially important for real-world applications such as disaster management and response coordination, where accuracy and reliability are essential. The comprehensive methodology and empirical results presented in this paper demonstrate the effectiveness and potential of using hybrid learning models to leverage multimodal data for unique and sophisticated analytical tasks in disaster scenarios

Understanding impulse buying in interest-based e-commerce: the role of content creativity

PurposeThe popularisation of user-generated content has boosted the integration of social media with online retail, especially in the emerging context of interest-based e-commerce (IBEC). As the content brings new incentives for consumers’ impulsivity in consumption, it remains unclear understanding of the mechanism of impulse buying in IBEC from the current literature. This study aims to examine how content creativity encourages impulse buying intention though attitude, and whether the effect is moderated by consumers’ online shopping experience.Design/methodology/approachThis study administered a two-wave online survey to a purposive sample of 286 consumers with shopping experience in Douyin. The hypotheses and a moderated-mediation effect were tested using structural equation modelling in Mplus8.3.FindingsThe empirical results revealed that content creativity, including dual dimensions of relevance and divergence, can drive positive attitude towards platforms through the direct effect of subjective norm and mediation effect through perceived usefulness, which induces an urge to buy impulsively. The mediation effect is further moderated by consumers’ online shopping experience in IBEC.Originality/valueThis study expands the stream of impulse buying research in retailing by proposing a context of IBEC and bringing empirical evidence of the relationship between content and consumers’ urge to buy. The study provides strategic insights for retailers to achieve marketing value by managing consumers’ impulse purchases in IBEC.

The political and economic environment promoting growth in American wealth inequality

AbstractWealth inequality around the world is high and rising. In this paper, we argue that wealth inequality in the United States has been exacerbated by an environment favourable to the ultra‐wealthy. Three distinctive changes in the economic and political landscape have fostered the increase in inequality. First, campaign finance law abruptly changed after the Citizens United Supreme Court decision. This decision gave the ultra‐wealthy an outsized influence on American politics. Second, an increase in competition for executive talent drove CEO salaries to new heights. Demand increased for executives with general leadership skills as corporations grew larger, increasing competition in the market for executive talent. Third, because tax policy in the United States is subject to lobbying influences, those at the top of the wealth distribution are favoured, leading to further concentration in wealth. Our empirical results are consistent with the view that these changes resulted in an increase in disparity, shifting wealth from the 99% to the 1%.

The Mediating Impact of Organizational Innovation on the Relationship Between Fintech Innovations and Sustainability Performance

The paper explores the impact of digital payment systems, blockchain technology, and AI/machine learning on innovation and sustainability in financial organizations. As part of the analysis, the study has adopted an explanatory research design and has used SmartPLS in order to analyze the data collected from 230 professionals of different fields through a structured questionnaire. The results show positive effects of digital payment systems and blockchain technology on organizations’ innovations with the impact of digital payments being the most pronounced. Empirical results suggest that these technologies are important to improve sustainability performance, depending on measures of internal consistency and discriminant validity among the proposed constructs. Al, also machine learning, has the highest relevance with environmental sustainability, thereby underlining the importance and work of such measures. Based on the Resource-Based View (RBV) theory, the study also explains the need for the organization to assimilate these innovations to enhance the organizational operations, customer satisfaction, and compliance with the laws. The study highlights fintech’s potential to address environmental issues and enhance societal goals, but geographical limitations may obstruct its transportability.

Hybrid Algorithm Selection and Hyperparameter Tuning on Distribute Machine Learning Resources: Hierarchical Agent-based Approach

Algorithm selection and hyperparameter tuning are critical steps in both academic and applied machine learning (ML). These steps are becoming increasingly delicate due to the extensive rise in the number, diversity, and distributed nature of ML resources. Multi-agent systems, when applied to the design of ML platforms, bring about several distinctive characteristics, such as scalability, flexibility, and robustness, just to name a few. This article proposes a fully automatic and collaborative agent-based mechanism for selecting distributed ML algorithms and simultaneously tuning their hyperparameters. Our method builds upon an existing agent-based hierarchical ML platform and augments its query structure to support the aforementioned functionalities without being limited to specific learning, selection, and tuning mechanisms. We have conducted theoretical assessments, formal verification, and analytical study to demonstrate the correctness, resource utilization, and computational efficiency of our technique. According to the results, our solution is algorithmically correct and exhibits linear time and space complexity in relation to the size of available resources. To further verify its correctness and demonstrate its effectiveness and flexibility across a range of algorithmic options and datasets, the article also presents a series of empirical results on a system composed of 24 algorithms and 9 datasets. The findings not only highlight the efficiency and scalability of the proposed approach, but also show its flexibility and openness to responding to the dynamic and distributed ML ecosystem.

Growing beyond trade: Export shocks and the spatial expansions of bank branching

AbstractThis paper empirically examines the impact of tariff reductions for exporters on banks' location choices with geo‐referenced information on subsidiaries and branches of China's state‐owned and joint‐stock banks. We construct prefecture‐level tariff barrier changes faced by exporters from 2001 to 2007 and identify the impact by exploiting within‐bank cross‐prefecture changes in bank branching patterns. The empirical results suggest a positive and significant pass‐through from export expansion to banks' geographic expansion. Specifically, a 10% export tariff reduction would induce a 1.88% increase in the local number of bank branches. This result holds under a battery of robustness tests. Mechanism tests suggest that increased credit demand in the corporate sector, both at the intensive and extensive margins, is the main channels that explain the baseline patterns. The findings in this paper offer valuable insights into the relationship between internationalisation and financial development.