|
on Computational Economics |
| By: | Jędrzej Maskiewicz (Quantitative Finance Research Group, Department of Quantitative Finance, Faculty of Economic Sciences, University of Warsaw); Paweł Sakowski (Quantitative Finance Research Group, Department of Quantitative Finance, Faculty of Economic Sciences, University of Warsaw) |
| Abstract: | The paper explores the use of Deep Reinforcement Learning (DRL) in stock market trading, focusing on two algorithms: Double Deep Q-Network (DDQN) and Proximal Policy Optimization (PPO) and compares them with Buy and Hold benchmark. It evaluates these algorithms across three currency pairs, the S&P 500 index and Bitcoin, on the daily data in the period of 2019-2023. The results demonstrate DRL's effectiveness in trading and its ability to manage risk by strategically avoiding trades in unfavorable conditions, providing a substantial edge over classical approaches, based on supervised learning in terms of risk-adjusted returns. |
| Keywords: | Reinforcement Learning, Deep Learning, stock market, algorithmic trading, Double Deep Q-Network, Proximal Policy Optimization |
| JEL: | C4 C14 C45 C53 C58 G13 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:war:wpaper:2025-14 |
| By: | Lutfu Sua; Haibo Wang; Jun Huang |
| Abstract: | Unpredictability of renewable energy sources coupled with the complexity of those methods used for various purposes in this area calls for the development of robust methods such as DL models within the renewable energy domain. Given the nonlinear relationships among variables in renewable energy datasets, DL models are preferred over traditional machine learning (ML) models because they can effectively capture and model complex interactions between variables. This research aims to identify the factors responsible for the accuracy of DL techniques, such as sampling, stationarity, linearity, and hyperparameter optimization for different algorithms. The proposed DL framework compares various methods and alternative training/test ratios. Seven ML methods, such as Long-Short Term Memory (LSTM), Stacked LSTM, Convolutional Neural Network (CNN), CNN-LSTM, Deep Neural Network (DNN), Multilayer Perceptron (MLP), and Encoder-Decoder (ED), were evaluated on two different datasets. The first dataset contains the weather and power generation data. It encompasses two distinct datasets, hourly energy demand data and hourly weather data in Spain, while the second dataset includes power output generated by the photovoltaic panels at 12 locations. This study deploys regularization approaches, including early stopping, neuron dropping, and L2 regularization, to reduce the overfitting problem associated with DL models. The LSTM and MLP models show superior performance. Their validation data exhibit exceptionally low root mean square error values. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.03109 |
| By: | Philippe Goulet Coulombe (University of Quebec in Montreal); Massimiliano Marcellino (Bocconi University); Dalibor Stevanovic (University of Quebec in Montreal) |
| Abstract: | We study the nowcasting of U.S. state-level fiscal variables using machine learning (ML) models and mixed-frequency predictors within a panel framework. Neural networks with continuous and categorical embeddings consistently outperform both linear and nonlinear alternatives, especially when combined with pooled panel structures. These architectures flexibly capture differences across states while benefiting from shared patterns in the panel structure. Forecast gains are especially large for volatile variables like expenditures and deficits. Pooling enhances forecast stability, and ML models are better suited to handle crosssectional nonlinearities. Results show that predictive improvements are broad-based and that even a few high-frequency state indicators contribute substantially to forecast accuracy. Our findings highlight the complementarity between flexible modeling and cross sectional pooling, making panel neural networks a powerful tool for timely and accurate fiscal monitoring in heterogeneous settings. |
| Keywords: | Machine learning, Nowcasting, Panel, Mixed-frequency, Fiscal indicators |
| JEL: | C53 C55 E37 H72 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:bbh:wpaper:25-04 |
| By: | Nikolaos Giannakis (Democritus University of Thrace); Periklis Gogas (Democritus University of Thrace); Theophilos Papadimitriou (Democritus University of Thrace); Jamel Saadaoui (University Paris 8); Emmanouil Sofianos (University of Strasbourg) |
| Abstract: | This study aims to predict currency, banking, and debt crises using a dataset of 184 crisis events and 2896 non-crisis cases from 79 countries (1970-2017). We tested eight machine learning methods: Logistic Regression, KNN, SVM, Random Forest, Balanced Random Forest, Balanced Bagging Classifier, Easy Ensemble Classifier, and Gradient Boosted Trees. The Balanced Random Forest had the best performance with a 72.91% balanced accuracy, predicting 149 out of 184 crises accurately. To address machine learning’s black-box issue, we used Variable Importance Measure (VIM) and Partial Dependence Plots (PDP). International reserve holdings, inflation rate, and current account balance were key predictors. Depleting international reserves at varying inflation levels signals impending crises, supporting the buffer effects of international reserves. |
| Keywords: | Currency crises, banking crises, debt crises, international reserve holdings, inflation, machine learning, forecasting |
| JEL: | F G |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:inf:wpaper:2025.6 |
| By: | Anton Korinek (University of Virginia and Centre for the Governance of AI); Jai Vipra (University of Virginia and Centre for the Governance of AI) |
| Abstract: | This paper examines the evolving structure and competition dynamics of the rapidly growing market for foundation models, focusing on large language models (LLMs). We describe the technological characteristics that shape the industry and have given rise to fierce competition among the leading players. The paper analyzes the cost structure of foundation models, emphasizing the importance of key inputs such as computational resources, data, and talent, and identifies significant economies of scale and scope that may create a tendency towards greater market concentration in the future. We explore two concerns for competition, the risk of market tipping and the implications of vertical integration, and use our analysis to inform policy remedies to maintain a competitive landscape. |
| Keywords: | Artificial Intelligence, economic concentration, vertical integration, AI regulation. |
| JEL: | D43 O33 L86 L40 L41 K21 |
| Date: | 2024–10–02 |
| URL: | https://d.repec.org/n?u=RePEc:thk:wpaper:inetwp228 |
| By: | Elliot Beck; Michael Wolf |
| Abstract: | Accurately forecasting inflation is critical for economic policy, financial markets, and broader societal stability. In recent years, machine learning methods have shown great potential for improving the accuracy of inflation forecasts; specifically, the random forest stands out as a particularly effective approach that consistently outperforms traditional benchmark models in empirical studies. Building on this foundation, this paper adapts the hedged random forest (HRF) framework of Beck et al. (2024) for the task of forecasting inflation. Unlike the standard random forest, the HRF employs non-equal (and even negative) weights of the individual trees, which are designed to improve forecasting accuracy. We develop estimators of the HRF's two inputs, the mean and the covariance matrix of the errors corresponding to the individual trees, that are customized for the task at hand. An extensive empirical analysis demonstrates that the proposed approach consistently outperforms the standard random forest. |
| Keywords: | Exponentially weighted moving average, Linear shrinkage, Machine learning |
| JEL: | C21 C53 C31 E47 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:snb:snbwpa:2025-07 |
| By: | Antonicelli, Margareth; Drago, Carlo; Costantiello, Alberto; Leogrande, Angelo |
| Abstract: | This study examines income inequality across Italian regions by integrating instrumental variable panel data models, k-means clustering, and machine learning algorithms. Using econometric techniques, we address endogeneity and identify causal relationships influencing regional disparities. K-means clustering, optimized with the elbow method, classifies Italian regions based on income inequality patterns, while machine-learning models, including random forest, support vector machines, and decision tree regression, predict inequality trends and key determinants. Informal employment, temporary employment, and overeducation also play a major role in influencing inequality. Clustering results confirm a permanent North-South economic divide and the most disadvantaged regions are Campania, Calabria, and Sicily. Among the machine learning models, the highest income disparities prediction accuracy comes with the use of Random Forest Regression. The findings emphasize the necessity of education-focused and digitally based policies and reforms of the labor market in an effort to enhance economic convergence. The study portrays the use of a combination of econometric and machine learning methods in the analysis of regional disparities and proposes a solid framework of policy-making with the intention of curbing economic disparities in Italy. |
| Keywords: | Income Inequality, Regional Disparities, Machine Learning, Labor Market, Digital Divide. |
| JEL: | C23 C38 C45 O15 R11 R58 |
| Date: | 2025–05–05 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:124910 |
| By: | Philippe Goulet Coulombe; Massimiliano Marcellino; Dalibor Stevanovic |
| Abstract: | We study the nowcasting of U.S. state-level fiscal variables using machine learning (ML) models and mixed-frequency predictors within a panel framework. Neural networks with continuous and categorical embeddings consistently outperform both linear and nonlinear alternatives, especially when combined with pooled panel structures. These architectures flexibly capture differences across states while benefiting from shared patterns in the panel structure. Forecast gains are especially large for volatile variables like expenditures and deficits. Pooling enhances forecast stability, and ML models are better suited to handle cross-sectional nonlinearities. Results show that predictive improvements are broad-based and that even a few high frequency state indicators contribute substantially to forecast accuracy. Our findings highlight the complementarity between flexible modeling and cross-sectional pooling, making panel neural networks a powerful tool for timely and accurate fiscal monitoring in heterogeneous settings. Nous étudions le nowcasting des variables budgétaires des États américains à l’aide de modèles d’apprentissage automatique (machine learning) et de prédicteurs à fréquence mixte, dans un cadre en panel. Les réseaux de neurones intégrant des variables continues et des identifiants catégoriels surpassent systématiquement les alternatives linéaires, en particulier lorsqu’ils sont combinés à des structures en panel mutualisé. Ces architectures permettent de capter les différences entre les États tout en tirant parti des régularités partagées. Les gains de prévision sont particulièrement importants pour les variables volatiles comme les dépenses et les déficits. Le regroupement des données améliore la stabilité des prévisions, et les modèles d’apprentissage automatique sont mieux adaptés pour traiter les non-linéarités transversales. Les résultats montrent que les améliorations prédictives sont généralisées et que même quelques indicateurs infranuels spécifiques aux États contribuent de manière significative à la précision des prévisions. Nos résultats soulignent la complémentarité entre la modélisation flexible et le regroupement transversal, faisant des réseaux de neurones en panel un outil puissant pour un suivi budgétaire rapide et précis dans des contextes hétérogènes. |
| Keywords: | Machine learning, Nowcasting, Panel, Mixed-frequency, Fiscal indicators, Apprentissage automatique, Panel, Fréquences mixtes, Indicateurs budgétaires, Prévisions à court terme |
| JEL: | C53 C55 E37 H72 |
| Date: | 2025–05–27 |
| URL: | https://d.repec.org/n?u=RePEc:cir:cirwor:2025s-15 |
| By: | Sona Benecka |
| Abstract: | This paper proposes a novel framework to the forecast of disaggregated producer prices using both machine learning techniques and traditional econometric models. Due to the complexity and diversity of pricing dynamics within the euro area, no single model consistently outperforms others across all sectors. This highlights the necessity for a tailored approach that leverages the strengths of various forecasting methods to effectively capture the unique characteristics of each sector. Our forecasting exercise has highlighted diverse pricing strategies linked to commodity prices, autoregressive behavior, or a mixture of both, with pipeline pressures being especially pertinent to final goods. Employing a mixture of a wide range of models has proven to be a successful strategy in managing the varied pricing behavior at the sectoral level. Notably, tree-based methods, like Random Forests or XGBoost, have shown significant efficacy in forecasting short-term PPI inflation across a number of sectors, especially when accounting for pipeline pressures. Moreover, newly proposed Hybrid ARMAX models proved to be a suitable alternative for sectors tightly linked to commodity prices. |
| Keywords: | Disaggregated producer prices, forecasting, inflation, machine learning |
| JEL: | C22 C52 C53 E17 E31 E37 |
| Date: | 2025–03 |
| URL: | https://d.repec.org/n?u=RePEc:cnb:wpaper:2025/2 |
| By: | Bachoc, François; Bolte, Jérôme; Boustany, Ryan; Loubes, Jean-Michel |
| Abstract: | Despite growing empirical evidence of bias amplification in machine learning, its theoretical foundations remain poorly understood. We develop a formal framework for majority-minority learning tasks, showing how standard training can favor majority groups and produce stereotypical predictors that neglect minority-specific features. Assuming population and variance imbalance, our analysis reveals three key findings: (i) the close proximity between “full-data” and stereotypical predictors, (ii) the dominance of a region where training the entire model tends to merely learn the majority traits, and (iii) a lower bound on the additional training required. Our results are illustrated through experiments in deep learning for tabular and image classification tasks. |
| Date: | 2025–05–21 |
| URL: | https://d.repec.org/n?u=RePEc:tse:wpaper:130552 |
| By: | Rincón Briceño, Juan José (Universidad de los Andes) |
| Abstract: | Economic decisions are made with high uncertainty about the current and recent past economic activity, due to the limited and imperfect available information. Therefore the following question arises: how can the accuracy of Colombian economic activity nowcasting be enhanced compared to traditional forecasting methods? This paper demonstrates: (a) using a risk-averse customized loss function that accounts for the agent disutility and penalizes directional discrepancies provides a useful alternative for assessing model performance by ensuring more accurate nowcasts, maximizing both precision and economic relevance. And (b) during periods of abrupt shocks and high volatility, such as the COVID-19 (2020–2021) and the post COVID-19 subsequent years (2022-2023), machine learning models outperform traditional nowcasting models |
| Keywords: | Colombian economic activity; nowcast; forecast; Random forests; LSTM. |
| JEL: | C45 C52 C53 E32 E37 |
| Date: | 2025–06–06 |
| URL: | https://d.repec.org/n?u=RePEc:col:000089:021388 |
| By: | Nurbanu Bursa |
| Abstract: | Emerging economies, particularly the MINT countries (Mexico, Indonesia, Nigeria, and T\"urkiye), are gaining influence in global stock markets, although they remain susceptible to the economic conditions of developed countries like the G7 (Canada, France, Germany, Italy, Japan, the United Kingdom, and the United States). This interconnectedness and sensitivity of financial markets make understanding these relationships crucial for investors and policymakers to predict stock price movements accurately. To this end, we examined the main stock market indices of G7 and MINT countries from 2012 to 2024, using a recent graph neural network (GNN) algorithm called multivariate time series forecasting with graph neural network (MTGNN). This method allows for considering complex spatio-temporal connections in multivariate time series. In the implementations, MTGNN revealed that the US and Canada are the most influential G7 countries regarding stock indices in the forecasting process, and Indonesia and T\"urkiye are the most influential MINT countries. Additionally, our results showed that MTGNN outperformed traditional methods in forecasting the prices of stock market indices for MINT and G7 countries. Consequently, the study offers valuable insights into economic blocks' markets and presents a compelling empirical approach to analyzing global stock market dynamics using MTGNN. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.01945 |
| By: | Qiang Chen; Tianyang Han; Jin Li; Ye Luo; Yuxiao Wu; Xiaowei Zhang; Tuo Zhou |
| Abstract: | Can AI effectively perform complex econometric analysis traditionally requiring human expertise? This paper evaluates an agentic AI's capability to master econometrics, focusing on empirical analysis performance. We develop an ``Econometrics AI Agent'' built on the open-source MetaGPT framework. This agent exhibits outstanding performance in: (1) planning econometric tasks strategically, (2) generating and executing code, (3) employing error-based reflection for improved robustness, and (4) allowing iterative refinement through multi-round conversations. We construct two datasets from academic coursework materials and published research papers to evaluate performance against real-world challenges. Comparative testing shows our domain-specialized agent significantly outperforms both benchmark large language models (LLMs) and general-purpose AI agents. This work establishes a testbed for exploring AI's impact on social science research and enables cost-effective integration of domain expertise, making advanced econometric methods accessible to users with minimal coding expertise. Furthermore, our agent enhances research reproducibility and offers promising pedagogical applications for econometrics teaching. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.00856 |
| By: | Shovon Sengupta (SUAD_SAFIR - SUAD - Sorbonne University Abu Dhabi, BITS Pilani - Birla Institute of Technology and Science, Fidelity Investments); Tanujit Chakraborty (SUAD_SAFIR - SUAD - Sorbonne University Abu Dhabi); Sunny Kumar Singh (BITS Pilani - Birla Institute of Technology and Science) |
| Abstract: | Forecasting consumer price index (CPI) inflation is of paramount importance for both academics and policymakers at central banks. This study introduces the filtered ensemble wavelet neural network (FEWNet) to forecast CPI inflation, tested in BRIC countries. FEWNet decomposes inflation data into high- and low-frequency components using wavelet transforms and incorporates additional economic factors, such as economic policy uncertainty and geopolitical risk, to enhance forecast accuracy. These wavelet-transformed series and filtered exogenous variables are input into downstream autoregressive neural networks, producing the final ensemble forecast. Theoretically, we demonstrate that FEWNet reduces empirical risk compared to fully connected autoregressive neural networks. Empirically, FEWNet outperforms other forecasting methods and effectively estimates prediction uncertainty due to its ability to capture non-linearities and long-range dependencies through its adaptable architecture. Consequently, FEWNet emerges as a valuable tool for central banks to manage inflation and enhance monetary policy decisions. |
| Keywords: | Inflation forecasting Wavelets Neural networks Empirical risk minimization Conformal prediction intervals |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05056934 |
| By: | Walenta, Danilo C.; Sturm, Timo; Scholz, Yven; Buxmann, Peter |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:dar:wpaper:154893 |
| By: | Kyungsu Kim |
| Abstract: | In this paper, I explored how a range of regression and machine learning techniques can be applied to monthly U.S. unemployment data to produce timely forecasts. I compared seven models: Linear Regression, SGDRegressor, Random Forest, XGBoost, CatBoost, Support Vector Regression, and an LSTM network, training each on a historical span of data and then evaluating on a later hold-out period. Input features include macro indicators (GDP growth, CPI), labor market measures (job openings, initial claims), financial variables (interest rates, equity indices), and consumer sentiment. I tuned model hyperparameters via cross-validation and assessed performance with standard error metrics and the ability to predict the correct unemployment direction. Across the board, tree-based ensembles (and CatBoost in particular) deliver noticeably better forecasts than simple linear approaches, while the LSTM captures underlying temporal patterns more effectively than other nonlinear methods. SVR and SGDRegressor yield modest gains over standard regression but don't match the consistency of the ensemble and deep-learning models. Interpretability tools , feature importance rankings and SHAP values, point to job openings and consumer sentiment as the most influential predictors across all methods. By directly comparing linear, ensemble, and deep-learning approaches on the same dataset, our study shows how modern machine-learning techniques can enhance real-time unemployment forecasting, offering economists and policymakers richer insights into labor market trends. In the comparative evaluation of the models, I employed a dataset comprising thirty distinct features over the period from January 2020 through December 2024. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.01933 |
| By: | Heinisch, Katja; Scaramella, Fabio; Schult, Christoph |
| Abstract: | Accurate macroeconomic forecasts are essential for effective policy decisions, yet their precision depends on the accuracy of the underlying assumptions. This paper examines the extent to which assumption errors affect forecast accuracy, introducing the average squared assumption error (ASAE) as a valid instrument to address endogeneity. Using double/debiased machine learning (DML) techniques and partial linear instrumental variable (PLIV) models, we analyze GDP growth forecasts for Germany, conditioning on key exogenous variables such as oil price, exchange rate, and world trade. We find that traditional ordinary least squares (OLS) techniques systematically underestimate the influence of assumption errors, particularly with respect to world trade, while DML effectively mitigates endogeneity, reduces multicollinearity, and captures nonlinearities in the data. However, the effect of oil price assumption errors on GDP forecast errors remains ambiguous. These results underscore the importance of advanced econometric tools to improve the evaluation of macroeconomic forecasts. |
| Keywords: | accuracy, external assumptions, forecasts, forecast errors, machine learning |
| JEL: | C14 C53 E02 E37 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:iwhdps:318189 |
| By: | Mohammadhossein Rashidi; Mohammad Modarres |
| Abstract: | Predicting the price that has the least error and can provide the best and highest accuracy has been one of the most challenging issues and one of the most critical concerns among capital market activists and researchers. Therefore, a model that can solve problems and provide results with high accuracy is one of the topics of interest among researchers. In this project, using time series prediction models such as ARIMA to estimate the price, variables, and indicators related to technical analysis show the behavior of traders involved in involving psychological factors for the model. By linking all of these variables to stepwise regression, we identify the best variables influencing the prediction of the variable. Finally, we enter the selected variables as inputs to the artificial neural network. In other words, we want to call this whole prediction process the "ARIMA_Stepwise Regression_Neural Network" model and try to predict the price of gold in international financial markets. This approach is expected to be able to be used to predict the types of stocks, commodities, currency pairs, financial market indicators, and other items used in local and international financial markets. Moreover, a comparison between the results of this method and time series methods is also expressed. Finally, based on the results, it can be seen that the resulting hybrid model has the highest accuracy compared to the time series method, regression, and stepwise regression. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.01402 |
| By: | Amy Wenxuan Ding (EM - EMLyon Business School); Shibo Li (Indiana University [Bloomington] - Indiana University System) |
| Abstract: | Scientists are interested in whether generative artificial intelligence (GenAI) can make scientific discoveries similar to those of humans. However, the results are mixed. Here, we examine whether, how and what scientific discovery GenAI can make in terms of the origin of hypotheses and experimental design through the interpretation of results. With the help of a computer-supported molecular genetic laboratory, GenAI assumes the role of a scientist tasked with investigating a Nobel-worthy scientific discovery in the molecular genetics field. We find that current GenAI can make only incremental discoveries but cannot achieve fundamental discoveries from scratch as humans can. Regarding the origin of the hypothesis, it is unable to generate truly original hypotheses and is incapable of having an epiphany to detect anomalies in experimental results. Therefore, current GenAI is good only at discovery tasks involving either a known representation of the domain knowledge or access to the human scientists' knowledge space. Furthermore, it has the illusion of making a completely successful discovery with overconfidence. We discuss approaches to address the limitations of current GenAI and its ethical concerns and biases in scientific discovery. This research provides insight into the role of GenAI in scientific discovery and general scientific innovation. |
| Keywords: | Scientific discovery, Generative artificial intelligence, Large Language models, ChatGPT |
| Date: | 2025–03–20 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05053017 |
| By: | Dejkam, Rahil (E.ON Energy Research Center, Future Energy Consumer Needs and Behavior (FCN)); Madlener, Reinhard (E.ON Energy Research Center, Future Energy Consumer Needs and Behavior (FCN)) |
| Abstract: | Energy poverty, a multidimensional socio-economic challenge, significantly affects the welfare of many people across Europe. This paper aims to alleviate energy poverty by exploring sustainable energy practices and policy interventions, using pilot household survey data collected within an EU project in Portugal and Denmark. A novel multidimensional energy poverty index (MEPI) is developed to assess energy poverty through different dimensions—such as heating and cooling comfort, financial strain, access to energy-efficient appliances, and overall health and well-being. In a next step, for selecting features, machine learning techniques, including recursive feature elimination and random forest analysis, are employed. These methods help to reduce the number of irrelevant and mutually correlated predictors. Subsequently, a logistic regression model is used to predict energy-poor households based on selected socio-economic and policy-related factors. The logistic regression model results indicate that sustainable energy-saving behaviors and supportive government policies can indeed effectively mitigate energy poverty. Furthermore, to analyze the impact of the determined features, the shapley additive explanations (SHAP) method is being utilized. Finally, the main findings are further evaluated via scenario simulation analysis. |
| Keywords: | Multidimensional Energy Poverty Index (MEPI); Thermal Discomfort; Sustainable Energy-saving Practices; Logistic Regression; Recursive Feature Elimination-Cross Validation (RFE-CV) |
| JEL: | C60 C83 |
| Date: | 2024–10–01 |
| URL: | https://d.repec.org/n?u=RePEc:ris:fcnwpa:2024_001 |
| By: | Pal, Hemendra |
| Abstract: | This study investigates the impact of the Russia- Ukraine conflict on Brent Crude commodity pricing using World Bank time series data. The conflict’s influence on global oil and gas markets, characterized by intricate supply and demand dynamics, is analyzed through advanced time series techniques and machine learning modeling. Univariate models such as Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) are employed to discern temporal patterns in Brent Crude prices. Additionally, Seasonal Autoregressive Integrated Moving Average (SARIMA) and Exponential Smoothing State Space (ETS) models are utilized to capture complex seasonality and trends in the data. Moving beyond traditional methods, multivariate models are leveraged to comprehensively grasp the multifaceted impact of the conflict. Principal Component Analysis (PCA) and Factor Analysis are applied to uncover latent variables influencing Brent Crude pricing in the context of global trade disruptions, inflation, and diplomatic negotiations. These extracted components are then integrated with ensemble machine learning algorithms, including Random Forest, Extra Tree Classifier, Gradient Boosting, K-Nearest Neighbors, and Decision Trees. The fusion of multivariate time series analysis and machine learning empowers a holistic understanding of the conflict’s intricate repercussions on commodity prices. The analysis reveals that not only direct factors related to geopolitical tensions but also indirect economic data are crucial in determining Brent Crude prices. Factors such as declining industrial demand for precious metals like silver, disruptions in vehicle production due to supply chain breakdowns, reduced demand for automotive auto-catalysts, weak copper demand from China, and unexpected changes in steel consumption have contributed to the observed fluctuations in Brent Crude prices. Through a comprehensive exploration of time series data and advanced machine learning modeling, this research contributes to a a clearer understanding of the complex connections between the crisis in Russia and Ukraine and the price of commodities globally. The findings offer valuable insights for policy-makers, industry stakeholders, and investors seeking to navigate the complex landscape of commodity markets during periods of geopolitical instability. |
| Keywords: | Brent Crude Prices, Univariate Models, Multivariate Models, Ensemble Machine Learning, PCA, SARIMA, ETS |
| JEL: | C15 C32 C38 C45 C51 C53 C55 O57 |
| Date: | 2023–08–15 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:124770 |
| By: | Sunyaev, Ali; Benlian, Alexander; Pfeiffer, Jella; Jussupow, Ekaterina; Thiebes, Scott; Maedche, Alexander; Gawlitza, Joshua |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:dar:wpaper:154738 |
| By: | Fourie, Jurgens; Steenkamp, Daan |
| Abstract: | We identify South African business cycles using the algorithm of Bry-Boschan and show that the identified turning points are very similar to those from other approaches. We demonstrate that South Africa has a very volatile business cycle that makes it particularly difficult to predict turning points in the economic cycle. South Africa’s business cycle is characterised by relatively long downswings and short upswing phases with low amplitude. We find that the South African Reserve Bank (SARB)’s Leading Indicator does not substantive improve predictions of the business cycle relative to GDP itself. We assess the performance of a range of potential leading indicators in identifying economic downturns and consider whether alternative indicators and estimation approaches can produce better predictions than those of the SARB. We demonstrate that using a larger information set produces substantially better business cycle predictions, especially when using machine learning techniques. Our findings have implications for the creation of composite leading indicators, with our results suggesting that many of the macroeconomic variables considered by analysts as leading indicators do not provide good signals of GDP growth or developments in the South African business cycle. |
| Keywords: | business cycle, forecast, leading indicator, economic downturns |
| JEL: | E32 E37 |
| Date: | 2025–05–07 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:124709 |
| By: | Hongyang Yang; Likun Lin; Yang She; Xinyu Liao; Jiaoyang Wang; Runjia Zhang; Yuquan Mo; Christina Dan Wang |
| Abstract: | Enterprise Resource Planning (ERP) systems serve as the digital backbone of modern financial institutions, yet they continue to rely on static, rule-based workflows that limit adaptability, scalability, and intelligence. As business operations grow more complex and data-rich, conventional ERP platforms struggle to integrate structured and unstructured data in real time and to accommodate dynamic, cross-functional workflows. In this paper, we present the first AI-native, agent-based framework for ERP systems, introducing a novel architecture of Generative Business Process AI Agents (GBPAs) that bring autonomy, reasoning, and dynamic optimization to enterprise workflows. The proposed system integrates generative AI with business process modeling and multi-agent orchestration, enabling end-to-end automation of complex tasks such as budget planning, financial reporting, and wire transfer processing. Unlike traditional workflow engines, GBPAs interpret user intent, synthesize workflows in real time, and coordinate specialized sub-agents for modular task execution. We validate the framework through case studies in bank wire transfers and employee reimbursements, two representative financial workflows with distinct complexity and data modalities. Results show that GBPAs achieve up to 40% reduction in processing time, 94% drop in error rate, and improved regulatory compliance by enabling parallelism, risk control insertion, and semantic reasoning. These findings highlight the potential of GBPAs to bridge the gap between generative AI capabilities and enterprise-grade automation, laying the groundwork for the next generation of intelligent ERP systems. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.01423 |
| By: | Dreoni Ilda (European Commission - JRC); Serruys Hannes (European Commission - JRC); Manso Luis (European Commission - JRC); Tudo Jose; Amores Antonio F (European Commission - JRC) |
| Abstract: | Consumption taxes are a crucial revenue source for EU Member States, yet they also potentially have non-negligible impact on income distribution. The EU's tax-benefit microsimulation model, EUROMOD, has recently been extended to simulate consumption taxes (CT) across all 27 EU countries allowing researchers and practitioners to examine carefully their design and assess trade-offs. The CT simulation uses consumption patterns derived from Household Budget Survey (HBS) microdata, which are imputed into EUROMOD's input data using the European Union Statistics on Income and Living Conditions (EU-SILC) microdata which contains detailed socio-demographic and socio-economic information. The imputation process employs a statistical matching procedure that joins HBS (the donor survey) with EU-SILC (the recipient survey) using a predictive mean matching method. Expenditure data are integrated into the recipient survey using a multi-stage procedure that involves the use of estimated probit and linear regression models combined with a distance-hot deck approach for the final observation mapping. This methodology offers enhanced results compared to traditional approaches such as only regression-based or distance-based. The imputation performance in distributional terms and the macro validation of the resulting datasets are thoroughly examined. We assess the impact of potential distortions from the statistical matching process by conducting a set of exploratory and comparative analyses, and also by using an administratively matched dataset for Czechia from 2019 to 2021. Our findings in this specific case indicate that, on average, the majority of imputed expenses are exactly the same when comparing the original HBS data with the matched SILC data that includes fitted expenditures. |
| Date: | 2025–04 |
| URL: | https://d.repec.org/n?u=RePEc:ipt:taxref:202502 |
| By: | Ulrike Famira-Mühlberger (WIFO); Thomas Horvath; Thomas Leoni (University of Applied Sciences Wiener Neustadt); Martin Spielauer (WIFO); Viktoria Szenkurök (WIFO); Philipp Warum (WIFO) |
| Abstract: | Europe's demographic shift is putting increasing pressure on long-term care (LTC) systems and raising concerns about the sustainability of LTC financing. This paper analyses Austria's LTC system, particularly its universal long-term care allowance (LTCA), and uses a dynamic microsimulation model to project LTCA expenditure under four scenarios up to the year 2080. Using pooled data from the Survey of Health, Ageing and Retirement in Europe (SHARE), we estimate care needs and prevalence rates across all seven care allowance levels. This enables us to project both public spending and individual lifetime costs, disaggregated by sex and education. Although total LTCA expenditure is projected to rise due to population ageing, scenario comparisons show that compositional shifts – such as higher educational attainment, which is linked to lower care needs, and gains in healthy life expectancy accompanying mortality im- provements – can significantly mitigate cost growth. The projected total expenditure increases range from 29 percent in a scenario where increasing life expectancy – as assumed in official population projections – is neglected, to 185 percent in a scenario accounting for rising life expectancy but no future health gains. The findings also highlight the impact of longevity and education on the distribution of individual lifetime costs. Beyond its policy implications for LTC planning, the study demonstrates the advantages of dynamic microsimulation in capturing individual-level heterogeneity, offering a significant improvement on traditional macrosimulation approaches. |
| Date: | 2025–05–26 |
| URL: | https://d.repec.org/n?u=RePEc:wfo:wpaper:y:2025:i:705 |
| By: | Stefano Fasani; Valeria Patella; Giuseppe Pagano Giorgianni; Lorenza Rossi |
| Abstract: | This paper investigates the macroeconomic effects of a belief distortion shock—an unexpected increase in the wedge between household and professional forecaster inflation expectations. Using survey and macro data alongside machine-learning techniques, we identify this shock and examine its effects within and outside the ZLB, while conditioning on the degree of inflation disagreement. The shock increases unemployment during normal times, whereas it reduces it in the ZLB, when the monetary stance is accommodative. Inflation disagreement instead dampens the expansionary effects of the shock. A New Keynesian model with belief distortion shocks replicates these dynamics and reproduces the inflation disagreement empirical patterns. |
| Keywords: | Inflation, Belief Distortion Shock, Inflation Disagreement, Households Expectation, Machine Learning, Local Projections, New Keynesian model, Monetary Policy, ZLB |
| JEL: | E31 C22 D84 C32 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:lan:wpaper:423478673 |
| By: | Lin, Ziqi (Rachel) |
| Abstract: | In this paper, a kernel Extreme Learning Machine (KELM) model based on vector weighted average algorithm is proposed for the prediction of national tax revenue ratio, which provides a new way of thinking and method for tax revenue prediction. By analyzing the correlation between each index and tax share, it is found that gasoline price and life expectancy are significantly positively correlated with tax share, while fertility rate and birth rate are significantly negatively correlated. The model shows excellent predictive performance on both training set and test set, with an R² of 0.995 in training set and 0.994 in test set, indicating that the model has excellent generalization ability. In addition, the root mean square error (RMSE) of the training set and the test set are 0.185 and 0.177, respectively, and the relative prediction deviation (RPD) is 14.234 and 13.178, respectively, which further verifies the high accuracy and stability of the model. Scatter plots of actual predicted versus actual values show that the model is able to accurately capture trends in tax shares with little prediction error. In summary, the optimized KELM model proposed in this paper not only has excellent performance on known data, but also has good expansion ability, and can be effectively applied to the tax share prediction of unknown data, providing a reliable tool for relevant policy making and economic analysis. The research of this paper provides a new technical path for the field of tax forecasting, which has important theoretical significance and practical value. |
| Date: | 2025–05–29 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:ymjw9_v1 |
| By: | Grebe, Leonard Nils |
| Abstract: | The replication crisis in financial economics highlights significant challenges to the credibility of empirical research, particularly in the study of stock market anomalies. This dissertation aims to enhance the consistency of previous findings by revisiting event-driven and seasonal value effects. Across six individual studies, the findings suggest that conflicting results do not necessarily indicate biases or model misspecifications but rather reflect the influence of underlying factors such as dataset composition, methodological choices, and evolving market conditions. Using a combination of event studies, regression analyses, meta-analyses, and artificial intelligence (AI) modeling, this research explores the impact of study design on research outcomes. The first focus is on replicating investor risk adjustments through established event study methodologies applied to three recent edge case events. The first event study validates established financial theories regarding the systematic risks associated with regulatory institutions. The second challenges the evaluation of cyber risks in the context of unintended software outages. The third event study contributes new insights to the identification of sustainable companies. In summary, these studies reveal systematic patterns in market reactions to event-driven value effects while highlighting that individual edge cases may challenge prior findings, suggesting the presence of additional underlying factors. Unlike the widely standardized methodology used to examine event-driven value effects, seasonal stock market anomalies, such as the day-of-the-week effect, are characterized by heterogeneous study designs, which often result in inconsistent findings. A primary study and a meta-analysis confirm that methodological choices significantly influence the observed weekly patterns of day-dependent returns. The findings suggest that dynamic market conditions, as reflected in divergent study designs, contribute to these inconsistencies. As a result, the final study introduces the "Uncertainty Structure Hypothesis" (USH), identifying market uncertainty as a key factor shaping weekly patterns in daily returns and offering an additional explanation for the replication crisis. In conclusion, this research underscores the importance of data selection, methodological choices, and the integration of advanced techniques such as meta-analyses and AI-driven nonlinear models. By investigating the drivers of the replication crisis, the dissertation enhances the reliability of financial research. Importantly, the findings suggest extending theoretical frameworks to better include the complexities of dynamic and uncertain market environments. |
| Date: | 2025–05–20 |
| URL: | https://d.repec.org/n?u=RePEc:dar:wpaper:154892 |