|
on Big Data |
| By: | Filipović, Slađana; Kozarević, Safet; Mujić, Haris |
| Abstract: | Digitalization has expanded insurers' databases, making machine learning (ML) a valuable tool for identifying key rsik factors in casco insurance pricing. While Generalized Linear Models (GLM) remain the standard, modelling claims frequency and severity separately is essential for accurately assessing the impact of each feature. Extreme gradient boosting (XGBoost), an ensemble ML method, can complement traditional GLM by improving predictive accuracy through feature selection. This study investigates a proposed hybrid XGBoost-GLM approach using a real-world dataset of 116.205 casco policies from a Bosnian insurer covering the period 2011-2023. The results show that XGBoost-based feature selection did not outperform the baseline GLM, which included all initial features without any selection. Deviance analysis confirmed that the baseline GLM with the full initial set of features achieved lower deviance and represented the optimal specification for the frequency-severity models. The modelling process was conducted in the R software environment (R Core Team, 2025). |
| Keywords: | casco insurance pricing, GLM, feature selection, machine learning, XGBoost |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:esconf:343875 |
| By: | Hamid Boustanifar; Sasan Mansouri |
| Abstract: | Researchers increasingly use generative large language models (LLMs) to convert corporate text into empirical variables. We examine the extent to which LLM-based textual measures are invariant to model choice using thirteen measures, including sentiment, management clarity, uncertainty, answer specificity, and climate and political risk. Seven LLMs from different providers score earnings call transcripts of S&P 500 companies on these constructs. Cross-model rank correlations average only 0.52, and transcript-level differences common across providers account for only 34% of total score variation. Cross-model disagreement does not predict subsequent analyst or market disagreement, consistent with a substantial model-specific component rather than common ambiguity in the underlying disclosure. Model choice significantly affects downstream inference, with coefficient magnitudes, signs, and statistical significance varying substantially across models. Averaging across providers makes transcript rankings more stable for most constructs, but score levels remain sensitive to the models included in the ensemble. LLM-generated variables should therefore be treated as model-contingent measurements and validated across providers. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.31013 |
| By: | Tamkin Nuriyev (Central Bank of Azerbaijan Republic); Aygun Garayeva (Central Bank of Azerbaijan Republic); Gulzar Tahirova (Central Bank of Azerbaijan Republic) |
| Abstract: | Using 800, 000 transaction-level customs records from January 2018 to February 2026, the paper constructs a trade-weighted Imported Food Price Index (IFPI), covering 34 items from the consumer basket with significant import dependence. The index is developed using the Fisher ideal methodology to provide a timely measure of external food price pressures. The results indicate that the IFPI leads official food Consumer Price Index (CPI) by approximately two months, with a maximum correlation of 0.81, highlighting its potential usefulness as an early indicator of domestic food inflation. Building on this, the paper develops a forecasting framework for the IFPI by combining non-parametric Binary Segmentation and Hidden Markov Models with a regularized machine-learning ensemble. The model employs an ensemble approach that combines Histogram-based Gradient Boosting Regression Tree, Random Forest, and Extreme Gradient Boosting, alongside rigorous time-series cross-validation. The optimized ensemble achieves a 58% out-of-sample R² relative to a random walk benchmark, vastly outperforming traditional linear Autoregressive Distributed Lag (ARDL) (13.60%) and Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) (0.18%) baselines. The forecast results are intended to be incorporated into broader inflation forecasting models to improve short-term projections. |
| Keywords: | Import price index; Fisher Ideal index; Food price inflation; Machine learning forecasting; Hidden Markov models. |
| JEL: | C43 C53 C55 E31 F14 |
| Date: | 2026–01–16 |
| URL: | https://d.repec.org/n?u=RePEc:aze:wpaper:2601 |
| By: | Francisco Blasques (Vrije Universiteit Amsterdam); Nepomuk Dunz (World Bank); Erik Feyen (World Bank); Charl Jooste (World Bank); Siem Jan Koopman (Vrije Universiteit Amsterdam); Gabriele Mingoli (Aarhus University) |
| Abstract: | Predicting macro-financial crises in developing economies is a critical but difficult task, complicated by significant data challenges. Machine learning techniques have shown improved predictive performance relative to traditional parametric models, yet this often comes at the cost of lower model explainability, limiting their use for policy makers. To strike a balance between predictive performance and explainability, this paper proposes a Mixture-of-Experts framework that combines ensemble models and traditional logistic regression to generate country-level crisis risk scores which can be decomposed into sub-risks, such as fiscal, financial, and balance of payments, to aid understanding of the drivers of risk within and across countries. Leveraging a global panel dataset spanning from 1990 to 2025 for 161 countries, the Mixture-of-Experts framework is shown to meaningfully outperform standard crisis prediction models in the literature. This paper therefore offers a practical and flexible framework for technical audiences developing early-warning systems in the presence of data gaps. |
| JEL: | C45 C82 G18 O11 |
| Date: | 2026–09–27 |
| URL: | https://d.repec.org/n?u=RePEc:tin:wpaper:20260074 |
| By: | Erik Heitfield; Mallick Hossain; Katie Merritt |
| Abstract: | Using fine-grained data on 465, 000 Florida home sales over twelve years and semiparametric machine learning methods, we examine how home prices respond to weather-related risk factors. After controlling for geography, home characteristics, and transaction features, we find that home prices are negatively related to property-level expected weather losses derived from catastrophe models, insurance premiums, and exposure to recent hurricanes, but with notably different magnitudes. Expected weather losses show the strongest association with prices and are consistent with markets rationally capitalizing anticipated future losses into current property valuations. Insurance premiums have a statistically significant but economically more modest negative effect on home prices, suggesting that housing market participants may view premium differences across properties as more transitory. Recent hurricane exposure has only a marginal effect on house prices, suggesting that hurricane risk is well-understood in Florida so that realized weather events convey little new information about future losses. |
| Keywords: | catastrophe modeling; home insurance; hurricanes; natural hazards; weather |
| Date: | 2026–09–18 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedgfe:103789 |
| By: | Matthew Bone; Fabian Stephany; Maria del Rio-Chanona |
| Abstract: | Employers are increasingly using large language models (LLMs) to automate their hiring process. This paper investigates the risk of monocultural biases, in which the widespread deployment of large language models homogenizes biases across the labor market, leading to greater systemic exclusion for certain demographic groups. For ten LLMs, we measure hiring biases across their base and post-trained versions to identify which stage, pre-training or post-training, lead to monocultural biases. We find that, compared to their base models, post-trained models are 3.6% less likely to callback older applicants. This negative shift occurs in eight of the ten models that we evaluate. Post-trained models have much more correlated decisions than base models which is likely driven by human capital traits like skills or college major. However, greater consensus among models increases global systemic exclusion rates from 5.6% to 17.3% and exacerbates demographic inequalities, with intersectional systemic exclusion rates ranging from 12.2% to 21.7% for post-trained models. We find that this inequality is primarily driven by age-based discrimination that is exacerbated in post-training. These results indicate that while post-training techniques may improve models' abilities to select the best applicants, they may raise systemic inequality risks for those at the margin by uniformly introducing new biases. |
| Date: | 2026–08 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.22169 |
| By: | Martin Feldkircher (Oesterreichische Nationalbank, Foreign Research Division); Márton Kardos (School of Culture and Society - Center for Humanities Computing); Kristoffer Laigaard Nielbo (School of Culture and Society – Center for Humanities Computing) |
| Abstract: | Central bank press conferences are not merely information releases — they are structured narratives. We study whether the shape of sentiment within a statement, not just its average tone, carries policy-relevant signals. Constructing sentiment arcs for ECB press conferences along three dimensions — monetary policy stance, economic outlook, and uncertainty — we assess their predictive content for policy rate changes and inflation expectations. Our findings show that arc shape robustly predicts rate decisions beyond lexicon-based benchmarks: it is not merely whether a statement sounds hawkish or economically optimistic on average, but how these sentiments are sequenced, emphasized, and sustained across the statement, that carries the policy signal. Arc features also shape how professional forecasters update inflation expectations, pointing to a receiver-side effect distinct from the direct policy signal. Our results are robust to a range of additional exercises, including different sample splits as well as controlling for the content of the statement. They also carry over to a short sample of Fed press conferences. These findings suggest that the sequencing and emphasis of policy language across a statement is a first-order feature of the policy signal, not a second-order refinement. |
| Keywords: | monetary policy; introductory statement; sentiment arc |
| JEL: | C55 C88 E52 E58 D83 |
| Date: | 2026–09–28 |
| URL: | https://d.repec.org/n?u=RePEc:onb:oenbwp:281 |
| By: | Hammed A. Olayinka; Saheed O. Olayemi |
| Abstract: | We develop a Dynamic Spatial Panel Bayesian Additive Regression Trees model with Horseshoe shrinkage (DSP-BART-HS) for high-dimensional spatio-temporal panel data. We jointly evaluate the model against a comprehensive suite of structural spatial econometrics, non-parametric machine learning methods, and small-area estimators across nine data-generating scenarios spanning 200 replicates each, including irregular spatial topologies, dense policy effects, and non-linear individual-level interactions. DSP-BART-HS is the best or statistically indistinguishable from the best estimator in every scenario. Conventional region-time-aggregate comparators suffer severe performance degradation -- trailing by a factor of three or more -- whenever individual-level non-linearity drives outcome variance, a limitation this framework's tree-ensemble design directly addresses. The model also maintains strong predictive accuracy under a zero-training-region spatial holdout via its spatial diffusion mechanism. We demonstrate practical utility across two U.S. county-level panel applications -- intergenerational economic mobility and geographic income inequality -- under genuine unseen-region, random, and temporal holdouts, with results validated through repeated-split uncertainty quantification and paired significance testing against every comparator. One clear limitation emerges: under temporal extrapolation specifically, a lighter per-region autoregressive specification (MTS-CAR-X) achieves a robust, and consistent advantage. DSP-BART-HS nonetheless establishes strong predictive performance, automated variable selection, and flexible inference for hierarchical spatio-temporal panel data. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.00072 |
| By: | Jawad Addoum; Vitaly Meursault; Justin Murfin |
| Abstract: | Stock and bond investors treat different aspects of firm information as value-relevant, generating distinct mappings of text to returns (“dictionaries”). Comparing stock and bond responses to earnings calls, bondholders emphasize operations and downside risk (relative to growth and innovation). While consistent with asset payoff differences, dictionary differences follow the investor cohort and not the payoff structure. Among bonds that are de facto equities but priced by bondholders—junior debt for which firm value is less than senior debt face value—bond investors continue to interpret information through a creditor’s lens. Information ignored by one dictionary but not the other generates price underreaction. |
| Keywords: | Cross-asset returns; Natural Language Processing; Machine Learning; Market Segmentation |
| Date: | 2026–09–21 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedpwp:103784 |