Pasar al contenido principal

SME DYNAMICS: LEVERAGING THE DIGITAL FOOTPRINT FOR SUSTAINABLE FUTURES

PID2023-152106OB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2023
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 14
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

AI adoption in Spain (2023-2025): A web-derived dataset based on LLMs

  • Pastor-Merino, Ana
  • Martínez-Barbero, Xavier|||0000-0002-6040-8118
  • Domenech, Josep|||0000-0002-7302-5810
[EN] This article introduces a nationwide dataset that maps how 112,814 Spanish firms communicate and implement artificial intelligence (AI) on their corporate websites in 2023 and 2025, resulting in 225,628 firm-year observations. Using a systemic pipeline based on large language models (LLMs), website text is segmented, semantically filtered, and evaluated with a structured rubric to identify explicit evidence of AI use in internal processes and in products or services. The dataset offers a detailed portrait of AI adoption across regions (NUTS 3), industries, and firm size categories. For each province-sector-size combination, it reports whether firms adopt AI, whether they apply it internally, whether it is embedded in their offerings, and how many firms have valid website content. This multi-dimensional structure enables users to explore territorial patterns, sectoral differences, and size-related disparities in the uptake of AI. By providing indicators for two benchmark years, the dataset supports the study of how AI adoption evolves across the Spanish business landscape. It offers a reproducible and scalable foundation for research on technological diffusion, regional digitalisation, and industry-level transformation, and can be readily extended to future years or adapted to other countries., This work was supported by the Generalitat Valenciana, Spain under Grant CIAICO/2023/272; the Agencia Estatal de Investigacion, Spain (MCIN/AEI/10.13039/501 100 011 033 ) and ERDF/EU under Grant PID2023-152106OB-I00; and FPU24/01116, Ministerio de Ciencia e Innovacion (Spain).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Uncovering Companies Missing from the SABI Database: A Web Scraping Approach

  • Xin-Hui Huang
  • Domenech, Josep|||0000-0002-7302-5810
[EN] This study evaluates the completeness and representativeness of the SABI database, a widely used commercial source for firm-level data in Spain and Portugal, by comparing it to BORME, the official Spanish business register. Using web scraping techniques, we collected and processed approximately 100,000 BORME publications in PDF format, covering the period from 2010 to 2023. These were transformed into a structured dataset comprising over 1.2 million companies, which we then matched against SABI records from the same period. Our analysis reveals that SABI covers only 38.3\% of newly established companies, with significant underrepresentation of younger firms, small enterprises, specific sectors, and certain regions. Furthermore, we find clear evidence of survivorship bias: the longer a company has been dissolved, the less likely it is to appear in SABI. Sectoral and geographic disparities are also substantial, and the coverage is skewed toward firms with higher initial capital and specific legal forms. These findings suggest that SABI represents a non-random subset of the Spanish business population, and caution should be exercised when using it for empirical research. Adjustments for sample bias are recommended to improve the reliability of analyses based on this database., This work was partially supported by Grant PID2023-152106OB-I00 funded by MICIU/AEI/ 10.13039/501100011033 and by ERDF/EU; Grant CIAICO/2023/272 funded by Generalitat Valenciana; and by Ministerio de Educación y Formación Profesional.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Turning Tastes into Trends: The Power of Enotourism in Boosting Wineries&apos, Popularity

  • Baixauli-Gimenez, Jose
  • Debón, A.|||0000-0002-5116-289X
  • Cervelló Royo, Roberto Elías|||0000-0002-8304-4177
  • Domenech, Josep|||0000-0002-7302-5810
[EN] Reputation and intangible assets, such as cultural heritage and regional authenticity, are crucial for wineries in the global wine market. Unlike purely physical products, wine embodies unique cultural and regional associations that shape consumer preferences. A strong reputation at both the winery and regional levels supports premium pricing and fosters brand loyalty, helping wineries thrive in a competitive, fragmented industry. This paper examines the role of enotourism in enhancing winery reputation and popularity. Enotourism offers immersive experiences at vineyards, creating opportunities for storytelling and heritage-sharing that strengthen brand identity. These personalized interactions boost consumer perceptions of wine attributes, such as flavour and origin, driving future purchases. Despite significant research on wine tourism¿s influence on brand loyalty, its impact on online winery popularity remains underexplored. Using data from 1098 wineries, this study analyzes the relationship between enotourism and online visibility, accounting for spatial effects. Findings reveal that enotourism significantly enhances online popularity, especially for larger wineries. Proximity to popular wineries amplifies these effects, offering strategic insights for leveraging enotourism to strengthen market presence and online engagement., The work was supported by the Agencia Estatal de Investigacion [PID2023-152106OB-I00]; Generalitat Valenciana [CIAICO/2023/272]. This work was supported by the Generalitat Valenciana under Grant CIAICO/2023/272; the Agencia Estatal de Investigacion (MCIN/AEI/10.13039/501100011033) and ERDF/EU under Grant PID2023-152106OB-I00; and the Ministerio de Educacion y Formacion Profesional.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Addressing Google Trends inconsistencies

  • Cebrián-Cerdá, Eduardo
  • Domenech, Josep|||0000-0002-7302-5810
[EN] Google Trends reports the evolution of the popularity of internet searches. Its main output is the Search Volume Index (SVI), a relative measure of the popularity of a term computed using a sample of the searches. Due to the sampling, the SVI series are not entirely consistent, as the same query produces different results that can widely change from day to day. This paper investigates the nature of these inconsistencies by modeling and simulating the data-generating process. Simulations are applied to describe how a typical time series is distorted due to the sampling process and to quantify how averaging extractions smoothes the series. Finally, a relationship between term popularity, series dispersion, and the averaged extractions is derived, so recommendations for constructing consistent SVIs can be provided., This work was partially supported by grants PID2019-107765RBI00, PID2023-152106OB-I00, PEJ2018-003267-A-AR and CIAICO/ 2023/272, funded by MCIN/AEI/10.13039/501100011033, ESF Investing in your future , and the Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Datos de investigación Set de datos (Dataset)

Tracking the Energy Transition of Spanish Firms (2023–2025): A Large-Scale Web and LLM-Based [Dataset]

  • Martínez-Barbero, Xavier|||0000-0002-6040-8118
  • Pastor-Merino, Ana
  • Domenech, Josep|||0000-0002-7302-5810
This article introduces a nationwide dataset that maps how 104,553 Spanish firms communicate and implement energy transition practices on their corporate websites in 2023 and 2025. Using a scalable pipeline based on large language models (LLMs), website text is segmented, semantically filtered, and evaluated through a structured rubric designed to identify explicit evidence of: (1) energy efficiency and energy consumption reduction actions, (2) decarbonization and greenhouse gas emissions reduction strategies, and (3) the use, production, or procurement of renewable energy. The resulting indicators are aggregated at the province (NUTS-3), sector (NACE 2-digit), and firm-size level, providing a detailed picture of corporate energy transition patterns across the Spanish economy. To preserve anonymity and statistical robustness, only aggregated cells containing at least four firms with valid website content are included in the public dataset. For each province–sector–size combination, the dataset reports the share of firms associated with each transition dimension together with the number of firms represented in the cell. By covering two years, the dataset enables the analysis of temporal changes in corporate sustainability and energy transition strategies across regions, industries, and firm-size categories. The dataset provides a reproducible and extensible framework for studying energy transition dynamics using web-based evidence and LLM-assisted classification, offering valuable applications for sustainability research, regional analysis, industrial transformation studies, and energy policy evaluation.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Detecting AI adoption at scale: a web mining and LLM methodology

  • Pastor-Merino, Ana
  • Martínez-Barbero, Xavier|||0000-0002-6040-8118
  • Domenech, Josep|||0000-0002-7302-5810
[EN] The recent emergence of Artificial Intelligence (AI) technologies has created multiple opportunities for companies. The growing interest in AI adoption has led to an increase in technological innovations and business applications. At the same time, web mining techniques have been used to monitor technological trends, often through basic natural language processing methods. However, advances in Large Language Models (LLMs) enable a more detailed and in-depth analysis of web content, necessitating the development of new methodologies. This paper proposes a methodology for integrating LLMs into the technology mining of corporate websites, with a focus on Spanish companies. The study uses data collected in 2023 and 2025, applying web scraping techniques to gather textual information. The analysis process involves automated AI detection, manual validation, and evaluation through standard classification metrics. The findings contribute to a better understanding of AI adoption trends and inform future research on tech mining applications for industry., This work was supported by the Generalitat Valenciana under Grant CIAICO/2023/272; the Agencia Estatal de Investigacion (MCIN/AEI/10.13039/501100011033) and ERDF/EU under Grant PID2023-152106OB-I00; and the Universitat Politecnica de Valencia (PAID-01-24, PAID-12-25).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Scraping innovativeness from corporate websites: Empirical evidence on Italian manufacturing SMEs

  • Bottai, Carlo
  • Crosato, Lisa
  • Guerzoni, Marco
  • Liberati, Caterina
  • Domenech, Josep|||0000-0002-7302-5810
[EN] Research in innovation studies usually relies on financial statements, surveys, or patents as primary data sources, although these sources of information show some limitations when applied to Small and Medium Enterprises (SMEs). Our paper explores whether the HTML code of a company¿s website is a further source to better inform innovation policies, under the assumption that how HTML is employed in crafting a corporate website provides insights into the company¿s innovation capabilities. In particular, we leverage HTML tags and their associations to empirically show that the websites of innovative SMEs are different from non-innovative ones both in terms of their size and coding practices. Our findings, based on a sample of Italian companies, indicate that the features of the HTML code of corporate websites reflect unobservable characteristics related to the skills and creativity present in businesses., We thank the Italian Ministry of University and Research (MUR) for sponsoring this work under the Departments of Excellence 2018 2022 funding schema. We greatly acknowledge the DEMS Data Science Lab of the University of Milano Bicocca for supporting this work by providing computational resources. This study was carried out within the project Data Driven Innovation. Measuring its Effects on Industries, Firms and Business Models prot. nr. 2022JHZL37 and received funding from the European Union Next-GenerationEU - National Recovery and Resilience Plan (NRRP) MISSION 4 COMPONENT 2, INVESTIMENT 1.1 Fondo per il Programma Nazionale di Ricerca e Progetti di Rilevante Interesse Nazionale (PRIN) CUP N. H53D23002440006; H53D23002450006.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Does AI boost firm productivity? A web scraping and LLMs approach

  • Pastor-Merino, Ana
  • Martínez-Barbero, Xavier|||0000-0002-6040-8118
  • Domenech, Josep|||0000-0002-7302-5810
  • Vicente, María R.
[EN] Artificial Intelligence (AI) is widely seen as a transformative technology with the potential to reshape firm productivity across sectors. However, empirical evidence on its actual impact at the firm level remains limited, mainly due to the lack of extensive firm-level data. Leveraging web scraping and Large Language Models (LLMs), this study develops a novel framework to detect and quantify AI adoption among 62,525 Spanish firms. The approach distinguishes between firms that adopt AI and those that integrate it intensively across business functions, constructing both binary and continuous measures of AI use. Using entropy-balanced instrumental variable estimations to address endogeneity, we find that AI-adopting firms exhibit approximately 53%¿55% higher sales and 48% higher value added than comparable non-adopters. Moreover, increasing the intensity of AI use is associated with an even larger productivity premium of 74%¿76% in sales and up to 67% in value added, highlighting the substantial returns from deeper AI integration. In dynamic terms, AI adoption is linked to 9%¿13% faster annual sales growth. These findings provide evidence that AI significantly enhances firm performance, underscoring its role as a key driver of productivity growth., This work was supported by the Generalitat Valenciana, Spain under Grant CIAICO/2023/272; the Agencia Estatal de Investigacion, Spain (MCIN/AEI/10.13039/501100011033) and ERD-F/EU under Grant PID2023-152106OB-I00; and the Universitat Politecnica de Valencia, Spain (PAID-01-24) and FPU24/01116, Ministerio de Ciencia e Innovacion (Spain).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A holistic approach to interpretable modelling and forecasting of human mortality by gender and country

  • Debón, A.|||0000-0002-5116-289X
  • Haberman, Steven
  • Piscopo, Gabriella
[EN] Studies from many countries find that gender differences in mortality rates and life expectancy vary by country. The multipopulation Lee-Carter family of models, a widely-used methodology, decompose mortality rates into age, time, and country components, offering valuable insights into mortality trends. We delve into the interpretability of the Lee-Carter multipopulation model, elucidating its ability to capture underlying mortality patterns and project future trajectories. Moreover, we extend our analysis by incorporating machine learning techniques to model the residuals of the Lee-Carter framework. The main contribution of the paper is to introduce these techniques in the context of the multiple population mortality models. Specifically, we employ Random Forest to refine joint mortality forecasts by country, effectively capturing complex nonlinear relationships in residuals and improving predictive performance. In this paper, we revisit these models using new statistical techniques and data sets from the Human Mortality Database. By leveraging advanced computational algorithms, we aim to enhance the accuracy of mortality rate predictions and account for residual patterns that may not be captured by the traditional Lee-Carter approach alone. Through empirical validation and comparative analyses, we demonstrate the efficacy of integrating machine learning into multiple population mortality forecasting, thereby contributing to the refinement and improvement of mortality modeling methodologies., Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This work was partially supported by grants PID2023-152106OB-I00 and CIAICO/2023/272, funded by MCIN/AEI/10.13039/501100011033 and the Generalitat Valenciana, respectively.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Digital footprint approach for the study of competitiveness in wineries

  • Debón, A.|||0000-0002-5116-289X
  • Domenech, Josep|||0000-0002-7302-5810
[EN] Advances in computation and web-scraping techniques allow us to use real-time information from company websites and social media platforms, namely digital footprint indicators. Additionally to their real-time availability, these indicators are easily accessible, making them a potentially practical tool for monitoring a company¿s level of competitiveness. Therefore, this article aims to obtain a multivariate analysis method to accurately predict the wineries¿ competitiveness group, applying the methodology to a sample of Valencian wineries to explore the association between digital footprint indicators and competitiveness. Unsupervised learning techniques were implemented to detect outliers and clusters in observations using financial variables obtained from the Sistema de Análisis de Balances Ibéricos (SABI). Thus, clustering was used to identify groups of wine companies differentiating the sample of companies according to their competitiveness characteristics. Subsequently, footprint indicators were used to create multivariate models to predict the above classification of companies. Also, this methodology permits the study of which digital indicators are essential in this prediction, specifically the presence of words on the web and others regarding online company activities on social networks associated with competitiveness. This research provides practical guidance for developing and incorporating the essential digital indicators, which could be applied in wineries and any study of companies¿ competitiveness., This work was partially supported by grant PID2023-152106OB-I00 and CIAICO/2023/272, funded by MCIN/AEI/10.13039/501100011033 and the Generalitat Valenciana, respectively.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Proyecto fin de carrera (BachelorThesis)

Clasificación de la actividad económica de las empresas: un enfoque basado en aprendizaje automático y contenido web, Classification of companies&apos, economic activity: a machine learning and web content-based approach, Classificació de l&apos, activitat econòmica de les empreses: un enfocament basat en aprenentatge automàtic i contingut web

  • Fernández Navarro, Joan
[ES] La clasificación de las empresas en función de sus actividades económicas es fundamental para el análisis económico, la elaboración de políticas y la investigación de mercados. Los métodos tradicionales de clasificación, basados en datos autodeclarados o en la categorización manual, requieren mucho tiempo y son propensos a incoherencias. Este trabajo propone un enfoque automatizado que aprovecha las técnicas de aprendizaje automático (ML) para clasificar las empresas a partir de datos textuales obtenidos de la web. La metodología incluye el procesamiento del contenido web, la extracción del texto y la generación de embeddings, seguidos de la evaluación de varios modelos de ML. Los resultados experimentales demuestran la eficacia de los modelos de ML en la clasificación de actividades económicas, destacando el potencial de los enfoques híbridos que integran modelos clásicos con contextual embeddings. Este trabajo contribuye a mejorar la precisión de la clasificación a la vez que reduce la intervención humana, ofreciendo una alternativa escalable a las metodologías tradicionales., [CA] La classificació de les empreses en funció de la seua activitat econòmica és fonamental per a l’anàlisi econòmic, l’elaboració de polítiques i la investigació de mercats. Els mètodes tradicionals de classificació, basats en dades autodeclarades o en la categorització manual, requereixen molt de temps i són propensos a incoherències. Aquest treball proposa un enfocament automatitzat que aprofita les tècniques d’aprenentatge automàtic (ML) per a classificar les empresses a partir de dades textuals obtingudes del web. La metodologia inclou el processament del contingut web, l’extracció del text i la generació d’embeddings, seguit de l'avaluació de diversos models de ML. Els resultats experimentals demostren l’eficàcia dels models de ML en la classificació d’activitats econòmiques, destacant el potencial dels enfocaments híbrids que n’integren models clàssics amb contextual embeddings. Aquest treball contribueix a millorar la precisió de la classificació al mateix temps que redueix la intervenció humana, oferint una alternativa escalabre a les metodologies tradicionals., [EN] The classification of companies according to their economic activities is fundamental for economic analysis, policymaking, and market research. Traditional classification methods, based on self-declared data or manual categorization, are time-consuming and prone to inconsistencies. This work proposes an automated approach that leverages machine learning (ML) techniques to classify companies based on textual data extracted from the web. The methodology includes web content processing, text extraction, and the generation of embeddings, followed by the evaluation of several ML models. Experimental results demonstrate the effectiveness of ML models in classifying economic activities, highlighting the potential of hybrid approaches that combine classical models with contextual embeddings. This work contributes to improving classification accuracy while reducing human intervention, offering a scalable alternative to traditional methodologies.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Unpacking the nuances of industrial agglomerations and their impact on manufacturing firm performance in emerging economies, Desentrañando las particularidades de las aglomeraciones industriales y su impacto en el rendimiento de las empresas manufactureras en economías emergentes

  • Villavicencio, Nathaly
  • Puig, Francisco
  • Debón, A.|||0000-0002-5116-289X
[ES] Objetivo Este artículo analiza cómo distintos tipos de localización (clústeres industriales, aglomeraciones urbanas y áreas aisladas) influyen en el desempeño de las empresas en economías emergentes. Asimismo, examina si la capacidad de innovación, entendida como la habilidad de la empresa para adaptar y transformar sus actividades, media en la relación entre localización y desempeño. Diseño/metodología/enfoque ¿ Se utilizan datos a nivel de empresa de 190 pymes ecuatorianas del sector textil y de la confección (2014¿2019). Las localizaciones se clasifican mediante Cocientes de Localización, y la innovación se operacionaliza a través de un índice compuesto derivado de Análisis de Componentes Principales. Las hipótesis se contrastan con ANOVA y modelos de regresión. Resultados Los resultados muestran que las empresas en clústeres industriales superan a las de otras localizaciones, pero la localización por sí sola no garantiza un mejor desempeño. La innovación refuerza la rentabilidad, especialmente en los clústeres industriales, mientras que las firmas aisladas dependen más de estrategias orientadas a la eficiencia. Los efectos de interacción son débiles, lo que refleja tanto el desbalance de datos como el carácter contingente de los vínculos entre localización e innovación. Originalidad/valor Este estudio ofrece uno de los primeros análisis sistemáticos que vincula tipologías de aglomeración e innovación con el desempeño empresarial en América Latina. Reafirma la distinción entre clústeres industriales, aglomeraciones urbanas y áreas aisladas y conceptualiza la innovación como capacidad dinámica en contextos institucionalmente frágiles. Implicaciones prácticas Los directivos deben adoptar modelos de negocio orientados a la innovación para aprovechar los beneficios de la aglomeración. A su vez, los responsables de políticas públicas deben diseñar instrumentos territorialmente sensibles para clústeres industriales, aglomeraciones urbanas y áreas aisladas. Implicaciones sociales Promover la capacidad innovadora fuera de las grandes aglomeraciones puede favorecer el desarrollo regional y reducir la desigualdad., This work was partially supported by grants PID2023-152106OB-I00 and CIAICO/2023/272, funded by MCIN/AEI/10.13039/501100011033 and the Generalitat Valenciana, respectively.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Constructing ESG-driven portfolios using genetic algorithms based on LSTM-predicted data

  • Martínez-Barbero, Xavier|||0000-0002-6040-8118
  • Cervelló Royo, Roberto Elías|||0000-0002-8304-4177
  • Jordán, Jaume|||0000-0003-0400-9136
  • Ribal, Javier|||0000-0002-9355-0145
[EN] Environmental, social, and governance considerations are becoming more relevant for investors, suggesting the need to include ESG metrics in optimizing portfolios to comply with investor requirements. Genetic algorithms offer the flexibility, adaptability, and effectiveness needed to deal with complex multi-objective problems with nonlinear constraints that traditional deterministic methods could not resolve. This paper aims to construct ESG-responsible portfolios that are able to replicate the index's returns and maintain low volatility. We propose a hybrid approach that combines long short-term memory (LSTM) neural networks to predict closing prices and genetic algorithms to solve a multi-objective problem that minimizes tracking error, ESG scores, and volatility. Our analysis uses LSTM-predicted data of the stocks that compose the IBEX 35 and EURO STOXX 50 during 2021 and the first half of 2022. The results show that optimized portfolios present low ESG risk; the returns follow the index closely and have small volatility. This allows investors to replicate the index performance while enhancing the ESG score of the portfolios. Our study addresses the limitations of traditional ESG portfolio optimization by explicitly incorporating predicted returns from long short-term memory networks and leveraging genetic algorithms (GAs) enhanced by NSGA-III.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Is it possible for Google Trends to forecast rural tourism? The situation in Spain

  • Cebrián, Eduardo
  • Domenech, Josep|||0000-0002-7302-5810
[EN] The study aims to assess the potential of Google Trends (GT) data for improving the prediction of monthly rural overnight stays by national residents in Spain. The study uses forecasting models that incorporate Google Trends data and compares their accuracy with traditional time-series benchmark models. The data used spans from January 2012 to February 2020. This comparison allows for a direct evaluation of whether GT variables offer any predictive advantage over classical time-series methods in the context of rural tourism. The results indicate that models incorporating GT variables do not outperform traditional time-series models in predicting rural tourism flows in Spain. Previous studies in other tourism sectors have found that GT data enhances predictive accuracy. The study concludes that Google Trends data may have limitations in predicting rural tourism flows, despite its demonstrated utility in other tourism forecasting contexts. These limitations suggest a need for further investigation into why GT does not improve rural tourism forecasts. The findings suggest that tourism industry stakeholders and policymakers should be cautious in relying on GT data for forecasting rural tourism flows., This work was partially supported by grant PID2023-152106OB- I00 funded by MCIN/AEI/10.13039/501100011033 and by ERDF/EU and by grant CIAICO/2023/272 funded by Generalitat Valenciana.