|
on Big Data |
| By: | Kriesch, Lukas (Justus-Liebig-Universität Gießen); Losacker, Sebastian (Justus-Liebig-Universität Gießen) |
| Abstract: | The emergence of large language models and the exponential growth of digitized text data have revolutionized research methodologies across a broad range of social sciences. News articles are an important source of digitized text data in this context. News data is crucial for the social sciences as it provides real-time insights into public discourse and societal trends, helping to understand various social phenomena and dynamics. However, most research involving news data is conducted at the national level, as geographically more granular news data is often unavailable. In this paper, we address this gap by providing insights into how news articles can be geolocated and how the texts can then be further analyzed. More specifically, we collect data from the CommonCrawl News dataset and clean the text data for further analysis. We then use a named-entity recognition model for geocoding, linking news articles to geographic locations. Finally, we transform the news articles into text embeddings using SBERT, enabling semantic searches within the news data corpus. In the paper, we apply this process to all German news articles and make the German location data, as well as the embeddings, available for download. As a result, we compile a dataset containing text embeddings for about 50 million German news articles, of which about 70% include geographic locations. The process can be replicated for news data from other countries, as we provide all code and workflows. |
| Keywords: | News data; Natural language processing; Geography |
| JEL: | C45 O33 R12 |
| Date: | 2026–09–25 |
| URL: | https://d.repec.org/n?u=RePEc:hhs:lucirc:2025_004 |
| By: | Kriesch, Lukas (Justus-Liebig-Universität Gießen); Losacker, Sebastian (Justus-Liebig-Universität Gießen) |
| Abstract: | In the transition toward more sustainable energy use, heat pumps play a crucial role. However, large-scale installation of heat pumps has not yet occurred in many countries, often due to locked-in energy technologies and low levels of public acceptance. Given that the diffusion of heat pumps will strongly depend on end-user acceptance, it is important to understand how heat pump technologies are perceived by the public. In this paper, we study the sentiment of news articles about heat pumps to reveal changes in public attitudes toward these technologies over time. We develop a unique dataset encompassing more than 30, 000 German news articles about heat pumps, published between 2018 and 2023, building on CommonCrawl data. We identify news articles about heat pumps using a large language model, and then analyze the sentiment of each article using a pretrained sentiment analysis algorithm. This data is used to identify coherent topics within the corpus of heat pump-related articles, with some topics receiving very positive sentiment scores and others negative. From a longitudinal perspective, our findings reveal a sharp decrease in the mean sentiment of heat pump articles in early 2023, related to the debate about a new heating law during that time in Germany. Our results also show that the sentiment score gradually recovers from this shock. |
| Keywords: | Energy transition; Heat pumps; News data; Natural language processing |
| JEL: | O33 Q42 Q58 |
| Date: | 2026–09–25 |
| URL: | https://d.repec.org/n?u=RePEc:hhs:lucirc:2025_008 |
| By: | Maya Kodeih; Aliaa Alnaggar; Mucahit Cevik |
| Abstract: | Monetary-policy announcements and central-bank communications play a central role in foreign exchange markets, yet their qualitative, unstructured form makes their forecasting value difficult to quantify. While prior research has largely focused on sentiment extracted from financial news, comparatively little is known about the relative contribution of different dimensions of monetary-policy communication. Existing studies primarily evaluate whether textual information improves overall forecasting performance but provide limited insight into which communication channels drive such improvements. To address this gap, this paper introduces a statistical attribution methodology that decomposes monetary-policy communication into interpretable channels and quantifies their incremental forecasting contribution under false-discovery-rate control. Monetary-policy news is transformed into structured communication signals using large language models (LLMs) and temporal feature engineering. These signals are evaluated using rolling-window experiments with tree-based machine-learning models. The results show that monetary-policy communication contains measurable predictive information. Attribution analysis shows that predictive value is concentrated in a small subset of signals, with communication timing providing the strongest individual feature-level contribution, targeted communication-activity measures also contributing positively, and LLM-derived sentiment providing complementary information at the group level. The findings indicate that communication-based forecasting value extends beyond sentiment alone and that attribution, rather than aggregate accuracy alone, is central to evaluating news-derived signals. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.32773 |
| By: | Kenneth L. Judd; Karl Schmedders |
| Abstract: | Deep learning is increasingly presented in economics as a new paradigm for solving dynamic models, superseding the classical projection methods of numerical economics. We argue that this dichotomy rests on a misclassification. The neural-network solvers emphasized in this literature are least-squares projection methods with an adaptive nonlinear parameterization. Placing deep learning inside the projection family reorients the agenda. The issue is not a contest between paradigms, but suggests a single toolkit using a variety of approximation methods, residual criteria, and evaluation designs to be chosen on the merits for the model at hand. In low and moderate dimensions, linear-basis and sparsegrid methods are often more accurate, faster, and easier to verify than a trained neural network. In high dimensions, an appropriate choice for the space of permissible functions will often avoid the curse of dimensionality. |
| JEL: | C60 |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35806 |
| By: | Anna B. Schmidt |
| Abstract: | This work studies how Bayesian machine learning methods can be used for large-scale demand estimation with many product categories. I compare two model classes, a latent factorization model and a mixed logit model and two Bayesian estimation approaches, Markov Chain Monte Carlo (MCMC) and Variational Inference (VI). The analysis combines a simulation study with an application to supermarket scanner data. The results show that the latent factorization model benefits from information across categories and improves its predictive performance as the dimensionality of the choice environment increases, whereas the mixed logit model does not exhibit the same pattern. MCMC delivers the highest predictive accuracy but is computationally intensive. VI achieves slightly lower predictive performance while substantially reducing runtime. In the empirical application, VI also outperforms the mixed logit benchmark. These findings highlight a trade-off between accuracy and computational feasibility in multi-category demand estimation. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.08409 |
| By: | Blankespoor, Brian; Dasgupta, Susmita; Wheeler, David |
| Abstract: | Air pollution imposes severe health and economic costs worldwide, yet ground-based monitoring data are difficult and expensive to collect, limiting the information available for policy analysis. This paper develops machine learning models to estimate daily air quality using satellite observations, meteorological data, emissions inventories, and geographic information. Data sources include IQAir ground station reports; atmospheric pollutant concentrations from the European Space Agency’s tropospheric monitoring instrument, TROPOMI; and meteorological data from the Copernicus ERA5 Land database. The paper estimates separate random forest models for fine particulate matter (PM2.5), ozone, nitrogen dioxide, and carbon monoxide, using a random forest algorithm that quantifies variable importance. The results are robust: TROPOMI dominates predictions for fine particulates (PM2.5), nitrogen dioxide, and carbon monoxide, with the Emissions Database for Global Atmospheric Research and meteorological variables providing secondary contributions. For ozone, the meteorological variables are most influential, followed by TROPOMI and the Emissions Database for Global Atmospheric Research. Geographic features matter across all pollutants—elevation, latitude, and coastal proximity consistently rank high, while Köppen-Geiger climate zones vary in importance by pollutant. Temporal tracking is validated by correlating daily monitored versus predicted pollutant levels across urban areas using IQAir ground station data. Median intertemporal correlations exceed 0.69 for all pollutants, with upper quartiles above 0.82 and lower quartiles above 0.53. Although the pilot uses data from April to November 2025, TROPOMI has been available since 2018 and the Copernicus ERA5 Land database provides matching coverage, enabling daily estimates from April 2018 onward on a global terrestrial grid at 0.1-degree resolution. |
| Date: | 2026–09–14 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:wbrwps:11448 |
| By: | Beáta Horváth (Magyar Nemzeti Bank (Central Bank of Hungary)); Tamás Berki (Magyar Nemzeti Bank (Central Bank of Hungary)); Lívia Réka Ónozó (Magyar Nemzeti Bank (Central Bank of Hungary)); Csanád Temesvári (Magyar Nemzeti Bank (Central Bank of Hungary)) |
| Abstract: | This paper proposes a sentiment-based financial stress indicator for Hungary based on occurrence of financial crisis related terms and explores its relationship with macrofinancial variables related to financial uncertainty. Sentiment analysis was conducted on textual data from two major Hungarian news portals covering the period from 2005 to 2020. The terms used in the analysis carry negative connotations, making them suitable for capturing financial instability and stress. The index successfully identifies stress periods that are not adequately reflected by other available sentiment indices related to Hungary. Overall, we propose a novel type of index that enables timely, even real-time monitoring of financial stress situation in Hungary across different time frequencies. |
| Keywords: | sentiment; financial stress; text analysis |
| JEL: | C32 E44 G01 |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:mnb:wpaper:2026/3 |