ABSTRACT This paper examines the growing importance of digital-based financial inclusion as a form of organising development interventions through networks of state institutions, international development organisations, philanthropic investment and fintech companies. The fintech–philanthropy–development complex generates digital ecosystems that map, expand and monetise digital footprints. Its ‘know thy (irrational) customer’ vision combines behavioural economics with predictive algorithms to accelerate access to, and monitor engagement with, finance. The digital revolution adds new layers to the material cultures of financial(ised) inclusion, offering the state new ways of expanding the inclusion of the ‘legible’, and global finance new forms of ‘profiling’ poor households into generators of financial assets.
Large Language Models (LLMs) have demonstrated strong general capabilities, yet their deployment in finance remains challenging due to dense domain-specific terminology, stringent numerical reasoning requirements, and low tolerance for factual errors. We conduct a controlled empirical study showing that in specialized vertical domains, performance is largely determined by the quality and difficulty/verifiability profile of post-training data. We introduce \textbf{ODA-Fin-SFT-318k}, constructed via multi-stage distillation and verification to produce high-quality Chain-of-Thought supervision, and \textbf{ODA-Fin-RL-12k}, curated for hard-but-verifiable tasks that balance reward precision and task diversity. Using standard SFT and RL pipelines, we show that high-quality CoT distillation establishes a robust foundation during SFT, while difficulty- and verifiability-aware sampling improves RL generalization. Evaluated on nine benchmarks spanning general financial tasks, sentiment analysis, and numerical reasoning, our ODA-Fin-RL-8B consistently surpasses open-source state-of-the-art (SOTA) financial LLMs of comparable size. We release our ODA-Fin-SFT-318k and ODA-Fin-RL-12k datasets, along with trained models to advance data-centric financial AI research.
Financial time series forecasting is central to trading, portfolio optimization, and risk management, yet it remains challenging due to noisy, non-stationary, and heterogeneous data. Recent advances in time series foundation models (TSFMs), inspired by large language models, offer a new paradigm for learning generalizable temporal representations from large and diverse datasets. This paper presents the first comprehensive empirical study of TSFMs in global financial markets. Using a large-scale dataset of daily excess returns across diverse markets, we evaluate zero-shot inference, fine-tuning, and pre-training from scratch against strong benchmark models. We find that off-the-shelf pre-trained TSFMs perform poorly in zero-shot and fine-tuning settings, whereas models pre-trained from scratch on financial data achieve substantial forecasting and economic improvements, underscoring the value of domain-specific adaptation. Increasing the dataset size, incorporating synthetic data augmentation, and applying hyperparameter tuning further enhance performance.
Conditional Autoencoders (CAEs) offer a flexible, interpretable approach for estimating latent asset-pricing factors from firm characteristics. However, existing studies usually limit the latent factor dimension to around K=5 due to concerns that larger K can degrade performance. To overcome this challenge, we propose a scalable framework that couples a high-dimensional CAE with an uncertainty-aware factor selection procedure. We employ three models for quantile prediction: zero-shot Chronos, a pretrained time-series foundation model (ZS-Chronos), gradient-boosted quantile regression trees using XGBoost and RAPIDS (Q-Boost), and an I.I.D bootstrap-based sample mean model (IID-BS). For each model, we rank factors by forecast uncertainty and retain the top-k most predictable factors for portfolio construction, where k denotes the selected subset of factors. This pruning strategy delivers substantial gains in risk-adjusted performance across all forecasting models. Furthermore, due to each model's uncorrelated predictions, a performance-weighted ensemble consistently outperforms individual models with higher Sharpe, Sortino, and Omega ratios.
Language Models (LMs) struggle with complex, interdependent instructions, particularly in high-stakes domains like finance where precision is critical. We introduce FIFE, a novel, high-difficulty benchmark designed to assess LM instruction-following capabilities for financial analysis tasks. FIFE comprises 88 human-authored prompts and employs a verification system with chainable, verifiable constraints for fine-grained reward signals. We evaluate 53 models (proprietary, open-weight, open-source) in a zero-shot setting. Our key findings reveal a clear performance hierarchy: the top open-weight model (76.1 strict / 79.5 loose) surpasses the leading proprietary system (65.9 strict / 70.5 loose), while the best open-source models lag significantly (45.5 strict / 48.9 loose). However, even top-performing models struggle with FIFE's complex requirements, failing to achieve perfect compliance. We release our dataset and code as an open-source resource to promote research in Reinforcement Learning for the financial domain.
Carrying out green technological innovation is a necessary way for enterprises to realize high-quality development, and government fiscal and tax incentive policy is an important initiative to promote enterprises' green technological innovation. This paper selects the A-share listed enterprises on the Shanghai and Shenzhen Stock Exchanges from 2018 to 2022 as research samples to empirically test the impact of fiscal and tax incentive policies on the green innovation performance of enterprises. The results of the study show that fiscal and tax incentive policies enhance the green innovation performance of enterprises, with both government subsidies and tax incentives significantly promoting the green innovation performance of enterprises. The mechanism test finds that the fiscal and tax incentive policies enhance the green innovation performance of enterprises mainly through the fulfillment of ESG responsibilities, as reflected explicitly in the fact that the fiscal and tax incentive policies can better enhance awareness of enterprise environmental responsibility, promote the fulfillment of corporate social responsibility, and improve the corporate governance system. Further test results show that both fiscal and tax incentive policies in state-owned and non-state-owned enterprises significantly promote corporate green innovation performance. Moreover, fiscal and tax incentive policies in Central China significantly promote corporate green innovation performance. In contrast, government subsidies in Eastern China significantly promote corporate green innovation performance, but the promotion effect is lower than that in Central China. This paper expands on the role of fiscal and tax incentive policies in influencing the green innovation performance of enterprises, which is of great significance in helping the government to change the direction and focus of fiscal and tax incentive policies promptly in order to improve the efficiency of those policies and better promote the green innovation of enterprises.
The demands of controlling when meeting cutting-edge technology are quite high given its underlying principles, its prospective character, flexibility, but also the desire for transparency, ethics, and responsibility. Through controllers (expert accountants), in their roles as collaborators, reminders, relationship managers of top management, smart technologies will be truly put to good use as business intelligence tools, as trusted allies (digital assistants, AI copilots, AI generative chatbots, interactive dashboards with AI inserts). Of course, there will be obstacles, a certain amount of distrust related to the “black boxes” regarding creation, operation, possible reactions. Hence the multiplication of searches to find something safer, with fewer unknowns regarding the purpose, risk levels, possible discriminations. This is how we arrived at XAI — explainable artificial intelligence, but also at HITL — complex models in which human judgment is integrated. The two systems also have their limits (especially regarding the balance between accuracy and explainability), but it is certain that the degree of trust, openness, and understanding of users (towards algorithms, models, artificial intelligence in general) through these tools will further increase. Basically, both tools suggest the same thing: if employees are directly involved and helped to understand something from the arguments, from the behavior of machines (whether it is about machine learning models, neural networks, or deep learning), then there will be an interactive collaboration between specialists and machines that is particularly beneficial to each productive or functional segment, but also to the entire organization.
The economy of a region is assessed based on its economic growth, which is a measure of the success of development in the region. Economic growth is a long-term increase in per capita production that reflects the dynamics of the economy. Evaluating economic growth is an important metric to assess the progress of a region, where the economy of a region will be better if its economic growth is faster. This study aims to analyze four factors that influence economic growth, namely: 1) the influence of the labor force, 2) the influence of years of schooling, 3) the influence of population density, and 4) the influence of the Human Development Index (HDI). This research is a quantitative study using secondary data from BPS for the period 2018-2023 and panel data regression analysis with Eviews 12. The results showed that: 1) the labor force has a significant positive effect on economic growth, 2) length of schooling has a significant positive effect on economic growth, 3) population density has a significant negative effect on economic growth, and 4) HDI has a significant positive effect on economic growth.
In the fast-evolving domain of artificial intelligence, large language models (LLMs) such as GPT-3 and GPT-4 are revolutionizing the landscapes of finance, healthcare, and law: domains characterized by their reliance on professional expertise, challenging data acquisition, high-stakes, and stringent regulatory compliance. This survey offers a detailed exploration of the methodologies, applications, challenges, and forward-looking opportunities of LLMs within these high-stakes sectors. We highlight the instrumental role of LLMs in enhancing diagnostic and treatment methodologies in healthcare, innovating financial analytics, and refining legal interpretation and compliance strategies. Moreover, we critically examine the ethics for LLM applications in these fields, pointing out the existing ethical concerns and the need for transparent, fair, and robust AI systems that respect regulatory norms. By presenting a thorough review of current literature and practical applications, we showcase the transformative impact of LLMs, and outline the imperative for interdisciplinary cooperation, methodological advancements, and ethical vigilance. Through this lens, we aim to spark dialogue and inspire future research dedicated to maximizing the benefits of LLMs while mitigating their risks in these precision-dependent sectors. To facilitate future research on LLMs in these critical societal domains, we also initiate a reading list that tracks the latest advancements under this topic, which will be continually updated: \url{https://github.com/czyssrs/LLM_X_papers}.
Purpose: This paper explored how investors in PSX and PMEX make investment decisions that either contribute to the success of PSX and PMEX or seem irrational to them because of the same behavioral factors.
Design/Methodology/Approach: This qualitative study is aiming at the examination of new features in behavioral finance by shedding light on the representative bias, availability bias, and anchoring bias as they relate to decision making among individual investors in Pakistan. This research will select Stock Broker's Individual Investors in PSX and PMEX to be the unit of analysis for the research of heuristic components.
Findings: The analysis of this qualitative research explored that investor in this study use behavioral factors to make their investment decision making, but not have negative experiences using these heuristics. However, their experiences suggest that investors tend to rely on past information from companies which shows that representative heuristic effect investors decision making as well as availability heuristic in investment decision making make investors to take decision irrationally by listening to rumors and tips from broker, friends and family.
Implications/Originality/Value: This qualitative study is aiming at the examination of new features in behavioral finance by shedding light on the representative bias, availability bias, and anchoring bias as they relate to decision making among individual investors in Pakistan
Abstract Decarbonization is often misunderstood in financial studies. Furthermore, its implications for investment opportunities and growth are even less known. The study investigates the link between energy indicators and Tobin's Quotient (TQ) in listed companies globally, finding that the carbon content of energy presents a negative yet modest effect on financial performance. Furthermore, we investigated the effect carbon prices in compliance markets have on TQ for exempted and non-exempt firms, finding that Energy efficiency measures yield greater effects in the latter group. Conversely, it is also true that carbon prices marginally reduce TQ more in non-exempt firms. This implies that auction-mechanisms create burdens that companies are eager to relinquish by reducing emissions. However, reducing GHG yields positive effects on TQ only as long as it results in energy efficiency improvements.
Utilizing three recent waves of Demographic and Health Surveys data from nationally representative samples, this study employs the Alkire and Foster methodology to gauge the Multidimensional Poverty Index in Ethiopia. Examining various factors including living standards, healthcare, and education access, analysis extends to subpopulation groups. By employing an ordered probit model after data restructuring, trends and determinants of multidimensional poverty at national and sub-population levels are assessed. Key factors impacting multidimensional poverty include location, household head’s demographics (sex, literacy, and age), family size, land area, and region of residence. Despite a notable decrease in households in multidimensional poverty, vulnerability to poverty is on the rise. While multidimensional poverty remains predominantly rural, vulnerability in urban households escalates. Empirical evidence supports growing economic disparity in Ethiopia. Regional disparities are evident, with Somali and Afar regions being the hardest poverty hit. Household size demonstrates a non-linear effect on poverty. This study underscores practical and theoretical implications for poverty alleviation strategies.
As a useful tool for managing ambiguous and inconsistent data, the Single Value Neutrosophic Set (SVNSs) is an extension of both Fuzzy Sets (FSs) and Intuitionistic Fuzzy Sets (IFSs). In the field of information theory, metrics like similarity, entropy, and distance are important. Although a number of entropy measures for SVNSs have been put forth and used in real-world situations, both academic research and real-world applications have pointed out certain drawbacks. Additionally, the Similarity Measures (SMs) is a useful instrument for determining how similar any two fuzzy values are to one another. The distance between the values allows the current SMs to evaluate the similarity. However, due to a few characteristics and intricate value operations, there are irrational and nonsensical cases. To deal with these preposterous cases, this paper proposed a parametric similarity measure in view of three parameters m1,m2,m3 in which decision makers can obtain the appropriate SMs by changing parameters with different decision styles. Furthermore, we analyze some existing SMs from a mathematical perspective and demonstrate the success of the proposed SMs using mathematical models. Ultimately, we apply the suggested SMs to resolve the Multi-Attribute Decision-Making (MADM) problems. We learn from the correlation and analysis that the suggested SM outperforms certain other SMs that are based on the SVNSs.
Fraud across the decentralized finance (DeFi) ecosystem is growing, with victims losing billions to DeFi scams every year. However, there is a disconnect between the reported value of these scams and associated legal prosecutions. We use open-source investigative tools to (1) investigate potential frauds involving Ethereum tokens using on-chain data and token smart contract analysis, and (2) investigate the ways proceeds from these scams were subsequently laundered. The analysis enabled us to (1) uncover transaction-based evidence of several rug pull and pump-and-dump schemes, and (2) identify their perpetrators' money laundering tactics and cash-out methods. The rug pulls were less sophisticated than anticipated, money laundering techniques were also rudimentary and many funds ended up at centralized exchanges. This study demonstrates how open-source investigative tools can extract transaction-based evidence that could be used in a court of law to prosecute DeFi frauds. Additionally, we investigate how these funds are subsequently laundered.
Automated market makers (AMMs) are a new prototype of decentralised exchanges which are revolutionising market interactions. The majority of AMMs are constant product markets (CPMs) where exchange rates are set by a trading function. This work studies optimal trading and statistical arbitrage in CPMs where balancing exchange rate risk and execution costs is key. Empirical evidence shows that execution costs are accurately estimated by the convexity of the trading function. These convexity costs are linear in the trade size and are nonlinear in the depth of liquidity and in the exchange rate. We develop models for when exchange rates form in a competing centralised exchange, in a CPM, or in both venues. Finally, we derive computationally efficient strategies that account for stochastic convexity costs and we showcase their out-of-sample performance.
We develop a methodology that utilizes deep learning to simultaneously solve and estimate canonical continuous-time general equilibrium models in financial economics. We illustrate our method in two examples: (1) industrial dynamics of firms and (2) macroeconomic models with financial frictions. Through these applications, we illustrate the advantages of our method: generality, simultaneous solution and estimation, leveraging the state-of-art machine-learning techniques, and handling large state space. The method is versatile and can be applied to a vast variety of problems.
Fraud Prevention in Village Government through Individual Morality
Purpose: The aims to determine impact of village apparatus competence, whistleblowing, internal control, financial reporting observance in preventing fraud with individual morality as moderating variable.
Method: Quantitative approach was employed by distributing questionnaires to twenty four villages.
Results: The competence of village apparatus cannot influence fraud prevention. Whistleblowing, internal control and financial reporting compliance have effect on fraud reduction. Individual morality does not moderate the competence of village apparatus, whistleblowing and the internal control for fraud prevention.
Novelty: Individual morality can moderate compliance with financial reporting.
Contribution: To prevent fraud, it is necessary to increase the competence of village apparatus, whistleblowing, internal control and compliance with sustainable financial reporting.
Pencegahan Kecurangan di Pemerintah Desa melalui Moralitas Individual
Tujuan: Untuk mengetahui pengaruh kompetensi aparatur desa, whistleblowing, pengendalian internal, kepatuhan pelaporan keuangan dalam pencegahan kecurangan dengan moralitas individu sebagai variabel pemoderasi.
Metode: Pendekatan kuantitatif dilakukan dengan menyebarkan kuesioner ke dua puluh empat desa.
Hasil: Kompetensi aparatur desa tidak dapat mempengaruhi pencegahan kecurangan. Whistleblowing, pengendalian internal dan kepatuhan pelaporan keuangan berpengaruh terhadap pengurangan kecurangan. Moralitas individu tidak memoderasi kompetensi perangkat desa, whistleblowing, dan pengendalian internal untuk pencegahan kecurangan.
Kebaruan: Moralitas individu dapat memoderasi kepatuhan terhadap pelaporan keuangan.
Kontribusi: Untuk mencegah terjadinya kecurangan, diperlukan peningkatan kompetensi aparatur desa, whistleblowing, internal control dan kepatuhan terhadap pelaporan keuangan secara berkelanjutan.