Causality Perspective Research Articles

Economic activities, dry bulk freight, and economic policy uncertainties as drivers of oil prices: A tail-behaviour time-varying causality perspective

This paper examines the dynamic impact of economic activity, Baltic dry bulk and economic policy uncertainty on oil prices. Drawing on demand-side economic theory, we focus on how economic activity and economic policy uncertainty drive oil prices and subsequently contribute to the formation of market equilibrium. In this context, we first introduce a unidirectional quantile and time-varying Granger-causality mechanism for the drivers of oil prices. Our novel rolling window quantile Granger causality approach reveals that economic activity drives crude oil price fluctuations and highlights their important role in market imbalances. In particular, economic activity and the Baltic exchange rate index emerge as essential drivers of oil prices, confirming our hypothesis of a demand-side effect. Moreover, our robustness wavelet quantile correlation (WQC) analyses reveal that economic activity promotes oil price fluctuations along both long-term equilibrium trends and short-term dynamics. In contrast, economic policy uncertainty has a dampening effect on oil prices. More broadly, economic activity is positively correlated with the Baltic exchange dry index, highlighting its broader impact on global trade indicators. Moreover, we use a bivariate Quantile on Quantile regression method to identify the equilibrium searches of different thresholds and estimate the robustness of the validity of the WQC findings. Analyzing the quantile and frequency connectedness spillovers, we observe that shocks from economic activity are particularly pronounced in the tail quantiles, reinforcing the demand-side narrative. Overall, the increased risk in extreme tails underscores the increased interaction between economic activity, oil prices and economic policy uncertainties. Our findings have important implications for policymakers and highlight the need for adaptive strategies to manage potential oil price shocks. Thus, by adopting a demand-side perspective, policymakers can better anticipate and respond to oil price fluctuations, promoting more resilient economic policies.

The persuasive impact of athlete racial advocacy on individuals’ cognitive responses: evidence from survey experiments in Japan

ABSTRACT Research question A growing body of research in sport literature links athlete racial advocacy to social justice. However, less is known about how athletes drive individual-level cognitive outcomes and contribute to the achievement of social justice through their advocacy. This study examines the persuasive impact of athlete advocacy on individual cognitive responses about racial discrimination from a causality perspective. Research methods This study employed survey experiments, specifically randomized controlled trials based on questionnaire surveys, within the context of Japanese athletes’ racial advocacy. Two original experiments were performed with 2,251 Japanese adults recruited from an online panel, using treatments with information on athlete advocacy as the intervention. Results and Findings The findings highlight that exposure to athlete racial advocacy can slightly facilitate individuals’ cognitive responses about racial issues, yet its persuasive power is fairly limited. Using multiple treatments that manipulated the fit between advocate athlete and advocacy content, our data also present experimental evidence that the persuasive power of athlete racial advocacy remains consistent across advocates’ racial backgrounds. Implications This study provides empirical and causal evidence of the persuasive power of athlete advocacy. Our findings contribute to the literature on the influence of athletes as advocates and sport for social justice.

Model-agnostic counterfactual reasoning for identifying and mitigating answer bias in knowledge tracing

Knowledge tracing (KT) aims to monitor students’ evolving knowledge states through their learning interactions with concept-related questions, and can be indirectly evaluated by predicting how students will perform on future questions. In this paper, we observe that there is a common phenomenon of answer bias, i.e., a highly unbalanced distribution of correct and incorrect answers for each question. Existing models tend to memorize the answer bias as a shortcut for achieving high prediction performance in KT, thereby failing to fully understand students’ knowledge states. To address this issue, we approach the KT task from a causality perspective. A causal graph of KT is first established, from which we identify that the impact of answer bias lies in the direct causal effect of questions on students’ responses. A novel COunterfactual REasoning (CORE) framework for KT is further proposed, which separately captures the total causal effect and direct causal effect during training, and mitigates answer bias by subtracting the latter from the former in testing. The CORE framework is applicable to various existing KT models, and we implement it based on the prevailing DKT, DKVMN, and AKT models, respectively. Extensive experiments on three benchmark datasets demonstrate the effectiveness of CORE in making the debiased inference for KT. We have released our code at https://github.com/lucky7-code/CORE.

A Causality Perspective: Investments in Green Energy Business vs. Conventional Business

Abstract At the level of the European Union numerous sustainability programs are being created. Considering these the main role f the research is to explore the potential of green energy production along with its limitations. Studies have found that by combining multiple sources of green energy and harnessing technological advances, there is the possibility to believe that one can reduce human’s dependence on fossil fuels. The aim of this study is to create a stock market analysis in order to identify future investment trends by assessing whether there is greater interest in investing in renewable electricity companies. The current paper explores the impact of investing in renewable electricity companies versus investing in the stock market of fossil fuel-based energy companies. The dataset includes closing stock price data for two major companies, one of which is the British-Dutch company SHELL.L Global (known in the oil industry as well as the energy sector) and the other company is Vestas Wind Systems (a Danish company that manufactures wind turbines for renewable energy parks around the world). This dataset covers the period 3rd of January 2011 to 24th of November 2023 and is used together with time series statistical tests. The purpose is to highlight the main objective of this article: to demonstrate through Granger causality whether investment in the renewable electricity market affects investment in the conventional energy market. This result will shed light on whether long-term investments in fossil fuel-based businesses are worth investors’ attention going forward. While the major purpose of the paper is the research mentioned above, there is also a secondary objective, that of carrying out a descriptive analysis of the two listed companies and their evolution in order to identify the key periods that led to significant fluctuations in the series temporal.

Topology Inference for Network Systems: Causality Perspective and Nonasymptotic Performance

Topology Inference for Network Systems: Causality Perspective and Nonasymptotic Performance

Green environments reimagined through the lens of green finance, green innovations, green taxation, and green energies. Wavelet quantile correlation and rolling window-based quantile causality perspective

Unarguably, a compromised environment portends negativities to humankind. Hence, we explored the potential of green finance, green innovations, green taxes, and green energy for green developments in South Africa (SA). Annual time-series data spanning 1996–2022 and cutting-edge estimators exposed the predictive effects of the under-listed variables on green environments. This includes the wavelet quantile correlation, time-varying Granger causality, and rolling window-based non-parametric quantile causality. The novel insights underscore the vital role of green finance in ensuring green growth, irrespective of some notable divergences over time. Likewise, green innovations, green taxes, and green energies produced notable environmental enhancing effects with underlining asymmetric effects across the time and quantiles of the distributions. The outcomes of the recursive evolving window algorithm and the rolling window-based quantile correlation portray that green finance, green innovations, green taxes, and green energies are significant positive enablers of green growth in SA. SA should forestall all the factors leading to the observed divergences as a policy option. This could be achieved through improved green finances and consistent promotion of all relevant green initiatives. These steps will aid the green revolution and enable the emergence of green development in the republic.

Evolutionary model and risk analysis of ship collision accidents based on complex networks and DEMATEL

To manage and control ship collision risk, this paper aims to develop a quantitative risk analysis approach based on complex networks and Decision-making Trial and Evaluation Laboratory (DEMATEL), to study the evolution of ship collision accidents. Firstly, through the search, selection and analysis of accident investigation reports of ship collisions in the world, 46 Risk Influential Factors (RIFs) were identified, and a network model for the evolution of ship collision was established. Secondly, a risk analysis of the evolution model for ship collision accidents was conducted by utilizing the topology characteristics of the complex network theory. Thirdly, a robustness analysis was used to identify 10 key RIFs involved in ship collisions from a global network perspective, and the grey relation analysis technique was employed to verify the key RIFs. Finally, the DEMATEL method was utilized to identify 11 key RIFs from a perspective of causality, and corresponding risk control measures were proposed to mitigate those key RIFs. The research results provide theoretical support for identifying the key RIFs and controlling the risk transmission of ship collision, and have practical significance for preventing the occurrence of ship collision accidents and ensuring the safe navigation of ships.

Class-incremental learning with causal relational replay

In Class-Incremental Learning (Class-IL), deep neural networks often fail to learn a sequence of classes incrementally due to catastrophic forgetting, a phenomenon arising from the absence of exposure to old knowledge. To alleviate this issue, conventional rehearsal methods, such as experience replay, store a limited number of old exemplars and then interleave with the current data for joint learning and rehearsal. However, the networks following this training scheme might not successfully reduce forgetting due to the lack of direct consideration of relations between samples of previously learned and new classes. Drawing inspiration from how humans learn by noticing the similarities and differences between classes, we propose a novel Class-IL framework called Relational Replay (RR). RR learns and recalls relations between images across all classes over time. To ensure these relations remain intrinsic and robust to forgetting, we incorporate causal reasoning to RR, resulting in Causal Relational Replay (CRR). CRR analyzes these relations using a causality perspective, aiming to identify intrinsic relations rooted in the images’ semantic features, serving as the cause of these relations. Our proposed method shows a competitive performance compared to the state-of-the-art rehearsal methods in Class-IL with clear and consistent improvements in the majority of settings on standard benchmark datasets.

Analysis of market risk volatility and warning in carbon trading market

A good risk warning model is an effective way to guarantee the stable operation of the carbon market and exert its emission reduction function. This paper takes the China's pilot carbon market as the research object and measures the carbon market risk by ARMA-GARCH-VaR. Then, influential factors are selected as non-market characteristics from the perspective of causality and incorporated into the risk warning model, and SMOTE and SMOTEENN are introduced to deal with the unbalanced samples. Results show that (1) The risk fluctuations of Beijing and Shenzhen carbon market are large, and the risk fluctuations of Guangdong carbon market are relatively smooth in the early period and larger in the recent period. (2) Air quality, average daily temperature, CSI 300, power coal, natural gas, European carbon market, exchange rate, and Baidu index are all important factors in the risk volatility of carbon market. (3) The Classification Decision Tree (CDT) has the best performance. CDT has the best performance in Guangdong carbon market, SMOTE-CDT has the best performance in Hubei carbon market, and SMOTEENN-CDT has the best performance in Shanghai carbon market, in which SMOTEENN-CDT has a 76% Recall. The non-equilibrium treatment improves the overall Recall, and the risk warning model after non-equilibrium treatment can avoid missing important risk points. This paper can provide a certain degree of reference for the improvement of carbon trading market and risk management.

Graph Contrastive Invariant Learning from the Causal Perspective

Graph contrastive learning (GCL), learning the node representation by contrasting two augmented graphs in a self-supervised way, has attracted considerable attention. GCL is usually believed to learn the invariant representation. However, does this understanding always hold in practice? In this paper, we first study GCL from the perspective of causality. By analyzing GCL with the structural causal model (SCM), we discover that traditional GCL may not well learn the invariant representations due to the non-causal information contained in the graph. How can we fix it and encourage the current GCL to learn better invariant representations? The SCM offers two requirements and motives us to propose a novel GCL method. Particularly, we introduce the spectral graph augmentation to simulate the intervention upon non-causal factors. Then we design the invariance objective and independence objective to better capture the causal factors. Specifically, (i) the invariance objective encourages the encoder to capture the invariant information contained in causal variables, and (ii) the independence objective aims to reduce the influence of confounders on the causal variables. Experimental results demonstrate the effectiveness of our approach on node classification tasks.

De-biased Attention Supervision for Text Classification with Causality

In text classification models, while the unsupervised attention mechanism can enhance performance, it often produces attention distributions that are puzzling to humans, such as assigning high weight to seemingly insignificant conjunctions. Recently, numerous studies have explored Attention Supervision (AS) to guide the model toward more interpretable attention distributions. However, such AS can impact classification performance, especially in specialized domains. In this paper, we address this issue from a causality perspective. Firstly, we leverage the causal graph to reveal two biases in the AS: 1) Bias caused by the label distribution of the dataset. 2) Bias caused by the words' different occurrence ranges that some words can occur across labels while others only occur in a particular label. We then propose a novel De-biased Attention Supervision (DAS) method to eliminate these biases with causal techniques. Specifically, we adopt backdoor adjustment on the label-caused bias and reduce the word-caused bias by subtracting the direct causal effect of the word. Through extensive experiments on two professional text classification datasets (e.g., medicine and law), we demonstrate that our method achieves improved classification accuracy along with more coherent attention distributions.

The greatest contribution to medical science is the transformation from studying symptoms to studying their causes-the unrelenting legacy of Robert Koch and Louis Pasteur-and a causality perspective to approach a definition of SLE.

The basic initiative related to this study is derived from the fact that systemic lupus erythematosus (SLE) is a unique and fertile system science subject. We are, however, still far from understanding its nature. It may be fair to indicate that we are spending more time and resources on studying the complexity of classified SLE than studying the validity of classification criteria. This study represents a theoretical analysis of current instinctual SLE classification criteria based on "the causality principle." The discussion has its basis on the radical scientific traditions introduced by Robert Koch and Louis Pasteur. They announced significant changes in our thinking of disease etiology through the implementation of the modern version of "the causality principle." They influenced all aspects of today's medical concepts and research: the transformation of medical science from studies of symptoms to study their causes, relevant for monosymptomatic diseases as for syndromes. Their studies focused on bacteria as causes of infectious diseases and on how the immune system adapts to control and prevent contagious spreading. This is the most significant paradigm shift in the modern history of medicine and resulted in radical changes in our view of the immune system. They described acquired post-infection immunity and active immunization by antigen-specific vaccines. The paradigm "transformation" has a great theoretical impact also on current studies of autoimmune diseases like SLE: symptoms and their cause(s). In this study, the evolution of SLE classification and diagnostic criteria is discussed from "the causality principle" perspective, and if contemporary SLE classification criteria are as useful as believed today for SLE research. This skepticism is based on the fact that classification criteria are not selected based on cogent causal strategies. The SLE classification criteria do not harmonize with Koch's and Pasteur's causality principle paradigms and not with Witebsky's Koch-derived postulates for autoimmune and infectious diseases. It is not established whether the classification criteria can separate SLE as a "one disease entity" from "SLE-like non-SLE disorders"-the latter in terms of SLE imitations. This is discussed here in terms of weight, rank, and impact of the classification criteria: Do they all originate from "one basic causal etiology"? Probably not.

The Association of Glaucoma with Ischemic Stroke and Functional Outcome after Ischemic Stroke from the Perspective of Causality

Introduction: Glaucoma may be related to ischemic stroke (IS) and poor outcomes after IS in observational studies, while the causal association remains unclear. Methods: We obtained single-nucleotide polymorphisms (SNPs) related to glaucoma from the genome-wide association study (GWAS) conducted by the FinnGen consortium. The GWAS included a total of 13,614 cases and 295,540 controls. The summary-level datasets regarding IS were collected from the MEGASTROKE consortium, including 34,217 cases and 406,111 controls. Furthermore, we acquired summary statistics datasets for functional outcomes following IS from the GWAS meta-analysis conducted by the GISCOME consortium, which involved 6,021 individuals. The genetic association estimates for functional outcomes at 90 days after IS were evaluated by the modified Rankin Score (mRS), including 3,741 cases with good functional outcomes (mRS = 0–2) and 2,280 subjects with poor functional outcomes post-stroke (mRS = 3–6). Inverse variance weighting (IVW) was used as the primary method, complemented by sensitivity analyses for pleiotropy and increasing robustness. Results: Genetically, glaucoma is associated with an increased risk of IS (odds ratio [OR] = 1.08, 95% confidence interval [CI] = 1.02–1.14, p = 0.0039), as well as poor prognosis after IS with adjustment for severity (OR = 1.64; 95% CI = 1.27–2.13, p = 0.0001) and functional outcome after IS (OR = 1.45, 95% CI = 1.12–1.87, p = 0.0038). Through sensitivity analyses, we confirmed the robustness of the results. In addition, we did not identify any causal association between IS, functional outcome after IS, and glaucoma in reverse analysis. Conclusion: Our study provides evidence suggesting potential genetic causal effects of glaucoma on an increased risk of IS, as well as a poor functional outcome following IS. Future studies are necessary to confirm these findings.

A Generalizable Causal-Invariance-Driven Segmentation Model for Peripancreatic Vessels.

Segmenting peripancreatic vessels in CT, including the superior mesenteric artery (SMA), the coeliac artery (CA), and the partial portal venous system (PPVS), is crucial for preoperative resectability analysis in pancreatic cancer. However, the clinical applicability of vessel segmentation methods is impeded by the low generalizability on multi-center data, mainly attributed to the wide variations in image appearance, namely the spurious correlation factor. Therefore, we propose a causal-invariance-driven generalizable segmentation model for peripancreatic vessels. It incorporates interventions at both image and feature levels to guide the model to capture causal information by enforcing consistency across datasets, thus enhancing the generalization performance. Specifically, firstly, a contrast-driven image intervention strategy is proposed to construct image-level interventions by generating images with various contrast-related appearances and seeking invariant causal features. Secondly, the feature intervention strategy is designed, where various patterns of feature bias across different centers are simulated to pursue invariant prediction. The proposed model achieved high DSC scores (79.69%, 82.62%, and 83.10%) for the three vessels on a cross-validation set containing 134 cases. Its generalizability was further confirmed on three independent test sets of 233 cases. Overall, the proposed method provides an accurate and generalizable segmentation model for peripancreatic vessels and offers a promising paradigm for increasing the generalizability of segmentation models from a causality perspective. Our source codes will be released at https://github.com/SJTUBME-QianLab/PC_VesselSeg.

Rethinking the role of attention mechanism: a causality perspective

Rethinking the role of attention mechanism: a causality perspective

Faithful and Consistent Graph Neural Network Explanations with Rationale Alignment

Uncovering rationales behind predictions of graph neural networks (GNNs) has received increasing attention over recent years. Instance-level GNN explanation aims to discover critical input elements, such as nodes or edges, that the target GNN relies upon for making predictions. Though various algorithms are proposed, most of them formalize this task by searching the minimal subgraph, which can preserve original predictions. However, an inductive bias is deep-rooted in this framework: Several subgraphs can result in the same or similar outputs as the original graphs. Consequently, they have the danger of providing spurious explanations and failing to provide consistent explanations. Applying them to explain weakly performed GNNs would further amplify these issues. To address this problem, we theoretically examine the predictions of GNNs from the causality perspective. Two typical reasons for spurious explanations are identified: confounding effect of latent variables like distribution shift and causal factors distinct from the original input. Observing that both confounding effects and diverse causal rationales are encoded in internal representations, we propose a new explanation framework with an auxiliary alignment loss, which is theoretically proven to be optimizing a more faithful explanation objective intrinsically. Concretely for this alignment loss, a set of different perspectives are explored: anchor-based alignment, distributional alignment based on Gaussian mixture models, mutual-information-based alignment, and so on. A comprehensive study is conducted both on the effectiveness of this new framework in terms of explanation faithfulness/consistency and on the advantages of these variants. For our codes, please refer to the following URL link: https://github.com/TianxiangZhao/GraphNNExplanation

Concept-Level Model Interpretation From the Causal Aspect

With the increasing growth of data and the ability of learning with them, machine learning models are adopted in various domains. However, few of machine learning models are able to reason their prediction, which limits their further applications in real-world tasks. With the potential to address this dilemma, model interpretation has become an important research topic because of the ability to provide the underlying reasons for model predictions at the feature level or concept level. Model interpretation at the concept level focuses on exploring the roles of concepts in model prediction, which enables more compact and understandable interpretations. Concept-level model interpretation requires the identification of the concepts that contribute to model prediction and the exploration of the rules underneath these concepts. To achieve the two objectives, we propose a Concept-level Model Interpretation framework (CMIC) from the perspective of causality. CMIC can automatically detect concepts in data and discover the causal relation between the detected concepts and the model's predicted labels. Furthermore, CMIC ranks the contributions of concepts by their causal effect on the model prediction, reflecting the detected concepts' importance. We evaluate the proposed CMIC framework on both synthetic and real-world datasets to demonstrate the quality of the provided interpretation.

General power laws of the causalities in the causal Bayesian networks

The power laws of network geometric properties are widely adopted to characterize the critical phenomena of complex systems, which may not describe the diverse interactions in systems effectively. By comparison, the causality represents the complicated interactions. Therefore, in this work, the power law of the causal effects in systems is studied to reflect the critical states from the perspective of causality. The causal Bayesian networks and effective information are adopted to construct the causal relationships and quantify the causal effects. The study on 20 systems from different fields shows that for a considerable proportion of systems, the causal effects of the factors in systems follow the power laws. For such causal power laws, an explanation based on the principle of maximum entropy is proposed and verified. The causal power laws may imply some critical states of systems, and can provide the basis for the quantification of systems’ states and functions.

The cross-border interaction of financial stress: From the perspective of pattern causality

It remains enigmatic and challenging, yet inspiring, to identify the interactions of financial stress, particularly when integration of international financial markets has intensified. Considering the nonlinear nature of financial markets, this study investigated the cross-border interaction of financial stress from the perspective of pattern causality. A multi-type and dynamic recognition of cross-border interactions was made, which can be characterized as positive, negative, or dark causality. Using the composite indicator of systemic stress (CISS) ranging from October 10, 2006, to July 15, 2022, empirical studies were conducted on the financial markets of China, the U.S., and the Euro area. The results showed that the positive causality of financial stress dominates, followed by dark causality in most cases. The positive causality between financial stress is enhanced in periods of extensive stress. When stress is released, positive causality declines and dark causality increases, implying that interaction structures are hidden and become opaque. In addition, the causality of different financial markets shows some heterogeneity: the positive causality of financial stress is stronger in advanced economies, while the dark causality between China and the U.S./Euro area is more evident than that between other advanced economies. This implies greater uncertainty regarding the cross-border interactions of emerging economies.

Spatio-Temporal Neural Structural Causal Models for Bike Flow Prediction

As a representative of public transportation, the fundamental issue of managing bike-sharing systems is bike flow prediction. Recent methods overemphasize the spatio-temporal correlations in the data, ignoring the effects of contextual conditions on the transportation system and the inter-regional time-varying causality. In addition, due to the disturbance of incomplete observations in the data, random contextual conditions lead to spurious correlations between data and features, making the prediction of the model ineffective in special scenarios. To overcome this issue, we propose a Spatio-temporal Neural Structure Causal Model(STNSCM) from the perspective of causality. First, we build a causal graph to describe the traffic prediction, and further analyze the causal relationship between the input data, contextual conditions, spatio-temporal states, and prediction results. Second, we propose to apply the frontdoor criterion to eliminate confounding biases in the feature extraction process. Finally, we propose a counterfactual representation reasoning module to extrapolate the spatio-temporal state under the factual scenario to future counterfactual scenarios to improve the prediction performance. Experiments on real-world datasets demonstrate the superior performance of our model, especially its resistance to fluctuations caused by the external environment. The source code and data will be released.