Great Lakes Institute of Management

This study investigates the dynamic and asymmetric connectedness between four crude oil benchmarks (Brent, WTI, INE, Murban) and three climate risk indexes (Physical Risk Index, Transition Risk Index, and U.S. Climate Policy Uncertainty Index). Addressing a critical gap in the literature, which often relies on linear models and average connectedness, we employ the quantile-on-quantile connectedness method to capture non-linear, asymmetric, and state-dependent spillovers, particularly under extreme market conditions. Our analysis reveals that climate risk indexes are predominantly net receivers of shocks from oil markets, with connectedness intensifying sharply during periods of market stress, political conflict, or sudden climate events. The findings highlight that systemic risk is significantly elevated at extreme quantiles, demonstrating that linear models may substantially underestimate true systemic risk during critical junctures. Methodologically, this research demonstrates the efficacy of quantile-on-quantile connectedness in revealing tail-risk effects. Empirically, it provides the most comprehensive comparison to date of connectedness across diverse crude oil benchmarks and climate risk indexes. The results offer crucial insights for investors seeking resilient portfolios, and for policymakers and regulators in designing macro-prudential oversight frameworks that recognize the non-linear and state-dependent nature of climate-financial contagion, emphasizing the need for flexible policies and continuous monitoring.

We study how financial stress originating in the United States (US), Advanced Economies (ADV), and Emerging Markets (EM) relates to movements in Economic Policy Uncertainty (EPU). Using monthly data for 2000-2024, we estimate horizon-specific responses to 4 EPU to one-standard-deviation innovations in 4 Financial Stress Index (FSI) via Jord & agrave; (2005) local projections with four lags and standard controls 4 Bloomberg Commodity Index (BCOM), 4 Federal Funds Rate (FEDFUNDS), 4U.S. Dollar Index (DollarIdx), 4CBOE Volatility Index (VIX). Across regions, the impact coefficient is negative,indicating that stress shocks are associated with an immediate reduction in the month-over-month change of EPU. Beyond impact, responses are small in magnitude, yielding limited persistence; cumulative effects over six months are modest and typically encompassed by wide confidence bands. Taken together, the evidence suggests that policy-uncertainty index adjusts quickly to stress realizations, with little systematic propagation at monthly horizons. This transience is most consistent with information/communication channels whereby policy guidance and rapid market repricing compress subsequent uncertainty innovations, while allowing for regional heterogeneity.

Surface grinding wheels are employed to manufacture components with high precision and finishing. They are frequently used to improve the finish on flat, curved and angled surfaces. The process is capable of grinding metals, ceramics or composite workpiece materials. Condition monitoring of surface grinding wheels is important for maximum performance and extended life. This research aims to forecast the life of a grinding wheel using machine learning models trained on images captured after each grinding pass. The experimental system comprises a surface grinding machine, a DSLR camera and appropriate lighting to record images of the grinding wheel. The wheel is split into eight parts, and the pictures are taken at the end of each machining pass until the grinding wheel is loaded. Surface roughness generated on the workpiece at each machine pass is maintained throughout the grinding cycle. The recorded images are further segmented along the cutting face of the wheel. Relevant statistical image features, such as entropy, skewness, standard deviation, kurtosis and the number of embedded particles, are extracted using image clustering methods. They are correlated and labelled correctly using surface roughness measurements on the workpiece. The features are labelled as the initial, intermediate, and final stages. The conditions of the grinding wheel are trained, and models were tested using machine learning techniques such as classification and regression tree and support vector machines. The results show a strong correlation between the extracted image features and surface roughness during grinding. The support machine with a cubic kernel achieved the highest predictive accuracy of 91.16%. The model was also tested on data from another experiment conducted on the same wheel, achieving 83% classification accuracy.

Dry machining of medium-carbon steels plays an important role in sustainable manufacturing; however, high tool wear and thermal instability pose challenges. The study aims to evaluate the kinematic-tribological performance of EN8 steel during dry milling and compare up-milling and down-milling to trade-off tool life and surface finish. The experiments were conducted using a central composite design (CCD) as part of response surface methodology (RSM), with 36 runs to evaluate interactions among spindle speed, feed rate, and depth of cut. Down-milling outperformed up-milling, achieving 12.4% less tool wear, 45.9% better surface finish, and a 47 °C lower peak temperature from cutting. The above benefits are attributed to the unique kinematics of chip formation during down-milling, which offers lower friction at entry and better heat dissipation, contrasting with the high-friction ploughing phase of up-milling. Grey relational analysis (GRA) found that down-milling with a mid-range cutting speed (22.31 m/min) and a low feed rate (25 mm/min) provided a multi-objective optimum. The findings support the existence of a kinematic-tribological coupling, providing a solid single approach to optimising the dry machining of harder materials.

The highly competitive nature of the online food delivery (OFD) market faces a serious retention problem, with acquiring new users typically being much more expensive than retaining existing users. Traditional prediction methods that rely primarily upon static transactional metrics such as recency and frequency are often unable to capture the psychological 'disconfirmation' which occurs prior to churn. To fill this gap, this study proposes a framework based on Expectation-Confirmation Theory (ECT). Unsupervised K-Means clustering was employed to classify a simulated and filtered dataset with 1500 customer records containing behaviour, geography, etc. This framework also couples sentiment analysis from BERT, allowing it to identify psychological "silent" attrition. Heterogeneous cohorts, which exhibit different psychological antecedents (utilitarian versus hedonic), were identified. The empirical results of our analyses demonstrated that Random Forest Classifiers with segment-specific features outperform baseline transactional models (F1 = 0.76) with an F1 Score of 0.89. The visual analytic interface developed provides a holistic view of the consumption process than traditional prediction models, including prescriptive, automated segment-based mitigation strategies. Our findings contradict the assumption that the "frequency-loyalty" model applies to all users. High-frequency discretionary users are found to be elastic in terms of retention and will experience significant churn. By utilising the automated action log, managers can plan targeted, highly efficient retention strategies rather than blanket discounting approaches.