|
on Big Data |
| By: | Paulo Júlio |
| Abstract: | This paper assesses whether pretrained time-series foundation models and locally trained deep neural networks can make a useful contribution to central bank forecasting in a small, data-constrained economy. We compare TimesFM, Chronos-2, Moirai 2.0, and several locally estimated architectures with conventional statistical and econometric benchmarks and Banco de Portugal official projections. The applications cover monthly inflation, direct quarterly macroeconomic forecasting, and mixed-frequency nowcasting. All headline exercises use historical-vintage information and final data as evaluation truth. Foundation models are generally more competitive than locally trained neural networks and, for selected targets and horizons, match or outperform simple benchmarks and official projections. These gains are not uniform. Official projections remain particularly strong when judgment, granular or advance information, and exceptional economic events matter, while persistent or irregular series can leave little scope for deep-learning improvements. Locally trained and native mixed-frequency models nevertheless add useful signals for selected real and labor-market aggregates. The results therefore favor using deep-learning forecasts as systematic, reproducible complements to official projections rather than as universal replacements. |
| JEL: | C45 C53 E27 E31 E37 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ptu:wpaper:w202612 |
| By: | Carlos Montes-Gald\'on |
| Abstract: | Most empirical models of trend inflation treat it as a latent random walk driven by unobserved shocks, so they cannot say which economic conditions move it. Deep-$\pi^*$ makes the innovations to the trend a possibly nonlinear function of a large number of macroeconomic indicators, estimated with neural networks, so that Shapley values can be used to attribute every movement of the trend to the indicators behind it. Two complementary loss functions define a current trend, which fits inflation today, and a forward-looking trend, which best predicts inflation over the next year. In US data, employment-related variables were the largest contributors to the post-pandemic rise of the current trend. However, the forward-looking trend rose earlier, with real activity, employment, and money and credit as the main drivers, and peaked eight to twelve months before the current trend. Both trends looked through most of the movements in energy and commodity prices. The contributions of the indicators are nonlinear, with sizeable interactions between them. Re-estimated every month during the surge, the model forecast inflation more accurately than the standard UC-SV model and a random walk, and in early 2021, when professional forecasters expected inflation of 3% or less, the forward-looking trend was already signalling an increase in persistent inflation. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.26298 |
| By: | Ebener, Luca; McGregor, Thomas |
| Abstract: | We introduce a novel measure of how European Central Bank (ECB) monetary policy is interpreted by analysts at major investment banks. We construct a “policy interpretation shock” based on the language used in analysts’ reports covering monetary policy decisions by the ECB’s Governing Council (GovC). The main identifying assumption is that changes in the way analysts report on policy communication by the GovC—that is, the change in the policy tone from the preview report to the recap report—reflect unexpected information revealed by the policy event. Using GPT-based sentiment extraction, we show that the tone in these reports closely tracks that of the GovC, often even anticipating shifts in tone. Changes in the policy tone before and after meetings capture how market intermediaries interpret and adjust their assessment of policy decisions. These interpretation shocks are followed by contractionary macroeconomic responses and systematic movements in financial-market prices. These findings underscore the important role of communication in shaping expectations and influencing economic outcomes. JEL Classification: E52, E58, E44 |
| Keywords: | analyst reports, large language models, monetary policy communication |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:ecb:ecbwps:20263298 |
| By: | Arka P. Bandyopadhyay |
| Abstract: | I use a deep neural network (DNN) on a proprietary panel of commercial mortgages spanning 1998--2021 to separate liquidity-constrained default from strategic default -- two competing, observationally similar explanations for delinquency that are difficult to disentangle with standard parametric models. Conditional on delinquency, property income (net operating income, or NOI) is a more important determinant of the default rate than leverage (loan-to-value, or LTV), despite LTV's status as the workhorse variable in the commercial mortgage default literature. I recover borrower motive from otherwise observationally equivalent delinquency states through a joint analysis of default rates across NOI and LTV buckets, and show that the resulting thresholds are stable across the 2008 financial crisis and the 2020--2021 COVID-19 shock to commercial real estate. Because commercial borrowers are firms rather than households, and because responsibility for a missed payment is diffused across an organization rather than borne by a single decision-maker, the moral and reputational frictions that discourage strategic default in the residential setting are attenuated -- making commercial mortgages a setting in which strategic behavior should be comparatively easier to detect, and comparatively costly to ignore for investors, servicers, and regulators. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.23901 |
| By: | Kleban, Yurii; Rude, Britta Laurin |
| Abstract: | This paper examines how invasion-related disruption reshapes the demand for digital skills. The analysis uses a novel dataset of millions of online job vacancies in Ukraine from 2021 to 2025 classified with large language models and mapped to the ESCO taxonomy. The paper links vacancy-level skill data to high-frequency ACLED attack intensity measures and estimate time-series models. The findings document a robust positive association between attack intensity and the digital skill share of job postings, consistent across specifications and alternative attack measures. Within- and between-occupation shifts and remote work adoption jointly drive this increase, with within-occupation reskilling and upskilling alone accounting for roughly half of the total rise. The growing digital skill share partly reflects the relative resilience of digital hiring against a broader contraction in the volume of vacancies. These findings carry concrete implications for dig-ital reskilling policy, labor market monitoring, and recovery and reconstruction planning. |
| Date: | 2026–09–14 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:wbrwps:11446 |
| By: | Onur Polat (Institute of Informatics, Hacettepe University, Beytepe Campus, 06800 Cankaya, Ankara, Turkiye); Rangan Gupta (Department of Economics, University of Pretoria, Private Bag X20, Hatfield 0028, South Africa); Christian Pierdzioch (Department of Economics, Helmut Schmidt University, Holstenhofweg 85, P.O.B. 700822, 22008 Hamburg, Germany) |
| Abstract: | We examine whether and how structural oil market shocks affect expressed public mood. To this end, we combine the daily forward-looking demand, current demand, and supply shocks of Gazzani, Venditti, and Veronese (2024) with the Twitter Sentiment Geographical Index, a daily measure of expressed sentiment built from geotagged tweets, from January 2012 to June 2023. We estimate lag- augmented local projections for a world index, a panel of 126 countries, and, to assess the role of disaggregation, the United States at the country, state, and county levels. Sentiment does not react to any of the shocks on impact. The cumulative responses, however, reveal a gradual adjustment. Forward-looking demand shocks, which capture news about global growth prospects and risk appetite, raise sentiment persistently, with the response building up over about six months, and persisting for up to a year; the sign of this response, in turn, is stable across aggregation levels, samples, and inference methods. Oil supply shocks raise sentiment briefly in the world and U.S. series, after which their point estimates turn negative at horizons of seven to twelve months. |
| Keywords: | Oil market shocks, Expressed sentiment, Subjective well-being, Twitter, Local projections, Disaggregation |
| JEL: | C23 D91 E71 Q43 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:pre:wpaper:202631 |