This study compares regularized linear, tree-based ensemble, boosting, and feed-forward neural-network approaches for forecasting country-level annual GDP growth using a global panel covering 213 countries from 2010 to 2023. A chronological out-of-sample evaluation is employed, with models trained on observations through 2021 and tested on 401 country-year observations from 206 countries during 2022–2023. Ridge regression, Random Forest, Histogram-Based Gradient Boosting, AdaBoost, and a multi-layer perceptron (MLP) are evaluated using root mean squared error (RMSE). Ridge regression records the lowest observed test RMSE (5.50), followed by Histogram-Based Gradient Boosting (6.28), Random Forest (6.75), AdaBoost (7.13), and the MLP (8.12). AdaBoost is treated as a classical boosting-based machine-learning method, whereas the MLP serves as the neural-network benchmark. The baseline models use one-year lags, while AdaBoost and the MLP use one-to-three-year lag windows. Consequently, the observed ranking reflects differences in both model specifications and information sets and should not be interpreted as a strictly like-for-like comparison or evidence of statistical superiority. Within this annual panel, predictor set, and 2022–2023 evaluation period, greater model complexity is not associated with lower forecast errors. The findings highlight the practical importance of regularization, model parsimony, and chronological evaluation in country-level macroeconomic forecasting.
Early numeracy is central to school readiness and educational equity, yet national policies may be implemented unevenly across geographically and institutionally diverse early childhood education (ECE) settings. This study examined the implementation of Indonesia's 2024 early literacy-numeracy policy in urban, rural, and remote settings in West Nusa Tenggara and explored the feasibility of the LENTAS (Locally Contextualized Early Numeracy with Technology Support) model. A convergent concurrent mixed-methods design combined policy and document analysis, a teacher-parent survey, interviews with 15 key informants, focus groups with 50 parents or caregivers, and an embedded single-group pre-post pilot involving 60 children and six teachers across three ECE sites. Of 900 survey questionnaires distributed, 662 complete responses were included in the final analysis; the remaining 238 were incomplete or not returned. Findings showed an implementation gradient favoring the urban site, where teacher preparation, materials, infrastructure, and family-engagement mechanisms were stronger. Rural and remote implementation was constrained by limited professional support, unreliable infrastructure, restricted access to learning materials, and weaker cultural and linguistic alignment. LENTAS integrated culturally familiar play-based activities, offline-first technology, structured parent engagement, and continuous teacher mentoring. Descriptively, the proportion of children meeting the count-to-ten criterion increased from approximately 75% to 85% in Mataram, from 60% to 78% in East Lombok, and from 50% to 80% in remote Sumbawa. These preliminary changes should not be interpreted as causal effects because the pilot was small and lacked a comparison group. Equitable implementation therefore requires differentiated, sustained support for teachers, locally relevant resources, offline-capable technology, family-community partnerships, and stronger local coordination.
This study investigates the effects of green growth (GG) and green technological innovation (GTI) on environmental degradation, measured by per-capita CO₂ emissions (LED), across the BRICS economies (Brazil, Russia, India, China, and South Africa) over the period 1990–2023. Motivated by the urgent need to balance fast-paced economic development with sustainable ecology in high-growth emerging economies, presents three key innovations: (i) the analysis of GG and GTI is conducted concurrently with the EKC dynamics in the context of the integrated PMG-ARDL approach; (ii) a newer dataset (1990–2023) compared to previous studies focusing on BRICS countries; and (iii) cross-sectional dependence tests and model robustness verification through Panel DOLS and FMOLS. Unit root testing shows mixed order of integration between I(0) and I(1), which fulfills PMG-ARDL requirements, while Kao residual co-integration test confirms the existence of long-term equilibrium among variables. The long-run estimation results show that GG and GTI have a significant impact on lowering CO₂ emissions by 0.38% and 0.65% per unit increase, respectively, which supports the Porter Hypothesis and decoupling paradigm. FDI also has a significant impact on lowering emissions (0.32%), which aligns with the Pollution Halo Hypothesis. Contrarily, fossil fuel use and population expansion contribute to worsening environmental quality. More importantly, the significant positive value of GDP and the negative value of GDP² validate the EKC hypothesis in BRICS.The error-correction term (−0.680) indicates that 68% of short-run deviations are corrected per period. Granger-causality tests reveal unidirectional causality from GG and GTI to LED, and bidirectional causality among GDP, FDI, fossil fuels, population growth, and LED. These findings advocate for coordinated policy frameworks that prioritise green technology investment, renewable energy transition, and carbon pricing across BRICS member states to attain Sustainable Development.
Social media platforms have become central arenas for identity formation and community interaction, yet their role as instruments of disciplinary power within religious communities remains underexplored, particularly in non-Western contexts. This study examines how digital social interactions within a youth Muslim religious community function as a panoptic mechanism that shapes members' self-presentation practices and collective identity on Instagram. Grounded in Michel Foucault's theoretical framework of panopticism, discursive power, and subjectivation, the study employs a qualitative approach within a critical constructivist paradigm. Digital ethnography was used as the primary research strategy, combining digital observation of the community's official and personal member accounts with semi-structured interviews conducted with six active members of the MM community (pseudonym), a youth-based Islamic organization in South Tangerang, Indonesia. Data were analyzed using thematic analysis following Braun and Clarke's framework. Findings reveal three interconnected dynamics. First, religious knowledge held by community advisors and leaders operates as a dominant discursive authority that disciplines members' digital behavior both overtly and covertly. Second, members internalize communal norms of piety, resulting in self-censorship, cautious content selection, and the reproduction of symbolic religious narratives on their Instagram accounts. Third, members actively negotiate spaces of personal freedom through the use of alternative accounts, selective audience features such as "close friends," and strategic content curation. These findings demonstrate that Instagram functions simultaneously as a site of disciplinary surveillance and a contested space for identity negotiation. This study contributes empirically and theoretically by demonstrating that digital panopticism in religious communities is not a replication of classical surveillance models, but a discursively constructed process in which religious knowledge itself becomes the primary instrument of normalization.
Firm value remains a central indicator of corporate performance. However, empirical evidence on how internal financial decision-making processes contribute to firm value remains limited, particularly in manufacturing firms operating in emerging economies and in studies based on primary managerial data. This study addresses this gap by conceptualizing financial decision-making quality as a multidimensional managerial capability and examining its effect on perceived firm value. A quantitative research design was employed using survey data collected from managers and financial executives of manufacturing firms. Financial decision-making quality was measured through investment analysis, financing structure decisions, and financial risk evaluation, while perceived firm value was assessed using indicators related to competitiveness, growth prospects, and sustainability. The data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results indicate that financial decision-making quality has a positive and statistically significant effect on perceived firm value. Firms with higher-quality financial decision-making processes tend to report stronger perceived firm value. However, the findings should be interpreted with caution due to the cross-sectional design and the use of perception-based measures, which may be subject to respondent bias. This study contributes to the literature by providing direct empirical evidence on the role of managerial financial decision-making quality as an internal determinant of perceived firm value in emerging economy contexts. The findings highlight the importance of strengthening managerial financial capabilities as a strategic lever for enhancing firm performance.
