BACK TO SCHEDULE REGULAR

Session 37

Balancing Accuracy and Timeliness

5 June 2026
10:30 – 11:45
ŠIBENIK III

Presentation title
Real time indicator for household consumption in Sweden
Demand for timelier and more frequent consumption statistics has grown sharply in recent years, while financial institutions have begun publishing their own indicators ahead of official releases.

Read more Read less

To keep official statistics relevant and trustworthy in this rapidly changing data landscape, Statistics Sweden has developed a weekly household consumption indicator using transaction data from card acquirers, card issuers and other payment service providers (PSPs).

Operating withing an official statistics framework that safeguards quality and transparency, the cornerstone of user trust, the indicator delivers near real‑time insights into household consumption trends and patterns, making the weekly series particularly valuable for detecting sharp turning points and enhancing the timeliness and responsiveness of economic policymaking during periods of rapid change. The methodological framework developed by Statistics Sweden demonstrates that a governed partnership with private institutions can responsibly leverage new data sources without compromising quality, while improving timeliness and closing critical information gaps for users. Other countries seeking to strengthen short-term economic monitoring may benefit from these results and adopt similar models.

The paper will present an overview of the development and implementation of the weekly consumption indicator—illustrating how official statistics can enhance trust and shape the future. The paper will focus especially on the methodological approach, including data integrations techniques, statistical modeling and validation strategies.

Main author / Presenter
Thorsten Holzer, Tod Salomonsson Segerby, Björn Forssell
Statistics Sweden

Read more Read less

Thorsten Holzer has been working at Statistics Sweden since 2018. He has been working with economic short-term statistics, mostly with the monthly indicator on household consumption. In the past year, has been taken part in an EU-financed grants-project with the aim to bring forward statistics on sales in retail trade, other services and household consumption. The project has tested to produce weekly indicators based on transaction data from card acquirers, card issuers and other payment service providers (PSP). Thorsten has a master’s degree in social science and a bachelor’s degree in statistics. Björn Forssell is the Head of the Innovation, Business Sector Production and Research Section at Statistics Sweden, within the Department of Economic Statistics and Analysis. He also serves as the Swedish Coordinator for Short term Business Statistics towards Eurostat, overseeing national contributions and ensuring alignment with European statistical standards. Tod Salomonsson Segerby is a Statistician working at the Data department at Statistics Sweden. Since joining in 2022, he has focused on economic and labour market statistics, with a particular emphasis over the past year on developing the weekly indicator of household consumption as part of an EU funded project that advances statistics on retail trade, other service industries, and household consumption using data from card acquirers, issuers, and other payment service providers. In addition to this work, he serves as a specialist on digitalised annual reports. Tod holds an MSc in Economics.

CO-AUTHORS:

Presentation title
The monthly GDP volume index as a business cycle indicator
In our study, we suggest applied methods and data sources for calculating the monthly GDP volume index in Hungary.

Read more Read less

The development of this indicator is important because although decision-makers receive a quarterly picture of the economy for t+30 days, intra-quarter movements remain invisible. In a rapidly changing world, where shocks can occur from one moment to the next, an indicator with a higher frequency would be necessary, even if its accuracy may not reach the quarterly indicator.

Since our goal was not only to decompose the quarterly time series back in time, but also to extrapolate the decomposed time series forward in time with monthly indicators, we were looking for a method that could be easily applied and results a robust GDP estimation in a short time. The most available methods optimize the extrapolation by taking the growth rate of the monthly indicator compared to the previous month as a basis. However, based on our analysis, this causes a large distortion on yearly basis calculated growth rate of the decomposed time series, which errors already appear cumulatively in the case of GDP. For this reason, it was necessary to develop a method that minimized the errors of year-on-year indices. We chose the Denton method, however, we modified the function so that the time series to be decomposed approximates the year-on-year dynamics of the indicator series.

Using the modified Denton method, we managed to calculate the added value at average prices of 2021 of each national economy branch in a production side approach for the period between Q1 2015 and Q3 2024, and then after chain-linking, we determined the GDP at average prices of 2021. Based on the indicator series and monthly value added data of branches, the monthly GDP can also be estimated.

The method we have presented is suitable for producing and and provides input for forecasting the monthly GDP, which can be a useful economic indicator for analysts and decision-makers.

Main author / Presenter
Klaudia Máténé Bella
Corvinus University of Budapest

Read more Read less

I am an economist and obtained my PhD in 2022 from the Doctoral School of Economics and Business Informatics at Corvinus University of Budapest. Between 2016 and 2025, I worked at the Hungarian Central Statistical Office in various departments, including the National Accounts Department, the Methodology Department, the Sectoral Statistics Department, and the Service Statistics Department, first as an expert and later as a head of section and head of department. Since 2025, I have been a full time Assistant Professor at Institute of Data Analytics and Information Systems at Corvinus University of Budapest, teaching statistics and econometrics. My research focuses on GDP and consumption analysis, the development of business cycle indicators, as well as the study of innovation, transportation and road accident data.

CO-AUTHORS:

Presentation title
Strengthening the Role of the official Job Vacancy Rate through Improved Timeliness of Flash Estimates
Job Vacancy Statistics (JVS) are key leading indicators of the business cycle and play a crucial role in structural labour market analysis, particularly for assessing mismatches between labour demand and supply.

Read more Read less

The job vacancy rate is included among the Principal European Economic Indicators (PEEIs), which provide timely and high-quality information for monitoring short-term economic developments in the European Union (EU) and the euro area. Official JVS are produced on the basis of vacancy surveys conducted under the European Regulation on Labour Market Business Statistics (2025/941), which requires the transmission of Flash estimates within 45 days after the end of the reference quarter and final estimates within 70 days.

Alongside official indicators, experimental statistics on labour demand based on non-traditional and innovative data sources—such as online job advertisements (OJAs)—are increasingly being developed. Recently, Eurostat disseminated experimental results on the most demanded occupations using OJA data from its Web Intelligence Hub combined with Labour Force Survey data to break down official job vacancy statistics by occupation. While these web-based sources offer valuable insights due to the high granularity of information on occupations and skills, they still face significant quality challenges that currently limit their use for official statistical production.

This paper focuses on improving the timeliness of the Italian official production of job vacancy statistics by proposing an alternative method for flash estimation aimed at further reducing production time. At present, the Italian flash job vacancy rate is produced around t+43 days using a process largely aligned with that of final estimates and based on calibration, with simplified editing and imputation procedures. The proposed approach replaces calibration with a method based on year-on-year variations computed from microdata referring to enterprises observed both in the current quarter and in the corresponding quarter of the previous year. By eliminating the calibration step, the processing time is reduced by approximately two days, allowing the inclusion of additional questionnaires received closer to the release deadline, which may enhance the quality of the estimates.

The paper presents a comparative assessment of the new proposed flash method, the current flash approach, and the final benchmark estimates, including an analysis of revision measures. Enhancing timeliness strengthens the leading indicator role of the job vacancy rate and represents a quality dimension in which official statistics can improve rapidly, remaining competitive with emerging experimental statistics without increasing costs for NSI’s or respondent burden for enterprises.

Main author / Presenter
Emilia Matera, Annalisa Lucarelli
Istat

Read more Read less

Emilia Matera is a senior researcher responsible for the Survey on Vacancies and Hours Worked at the Italian National Institute of Statistics (ISTAT). She is in charge of data collection and all aspects related to the proper conduction of the survey, paying particular attention to maintain an adequate response rate while ensuring the timeliness and the dissemination of the results obtained. Annalisa Lucarelli is senior researcher and responsible for the production of quarterly indicators on job vacancies and hours worked at the Italian National Institute of Statistics (ISTAT). Building on the ESSnet projects on Online Job Advertisements (OJAs) launched in 2018, she currently plays a key role in the Italian Eurostat-funded project aimed at developing experimental indicators based on OJA data.

CO-AUTHORS:

Presentation title
Balancing Accuracy and Timeliness: A Quality Framework for Truncated EE-MRIO CO₂ Footprints
Environmentally extended MRIO (EE-MRIO) models such as FIGARO are critical for producing climate-policy-relevant CO₂ footprints, but their complexity often constrains timeliness and complicates quality assurance.

Read more Read less

This paper introduces a quality-oriented framework that combines value truncation with Generalised RAS (GRAS) redistribution to operationalise Eurostat’s most comprehensive footprint variant while preserving core accounting identities.

Eurostat’s most detailed footprint model generates more than 1 billion records of CO2 footprint flows every year. The combination of truncation and GRAS avoid costly data storage as well as computation time and memory, enabling efficient computation with sparse operations, and can be fully implemented in R. Small footprint flows are truncated according to defined thresholds, then rebalanced using GRAS to ensure row and column sums remain consistent with published totals.

Quality is evaluated through multiple indicators: Weighted Absolute Percentage Error (WAPE), maximum absolute and relative deviations, and preservation of structural properties (row/column sums, zeros). On EU FIGARO-based CO₂ footprints, a 1-ton threshold reduces non-zero records by ~97 % and file size by an order of magnitude, while keeping WAPE near 1.3% across aggregates and most detailed breakdowns. A threshold of 100Kg implies a WAPE near 0.3% but file size is more than 2.5x bigger than the 1-ton threshold. A threshold of 10 tons generates unacceptable local biases, showing the accuracy–size trade-off and supporting an evidence-based operating point .

The framework supports systematic QA by logging adjustments, providing convergence diagnostics, and offering risk-based exceptions for flows where small values are policy-critical. It aligns with ESS quality principles of coherence, comparability, and cost-effectiveness, and promotes fitness for use by accelerating footprint production without undermining reliability.

By documenting trade-offs transparently and quantifying impacts, the method fosters user confidence while advancing innovative approaches in environmental statistics.

This contribution addresses innovating methodological and quality assurance frameworks by showcasing a novel quality assurance methodology that balances computational efficiency with statistical soundness in the production of EE-MRIO-based CO₂ footprints.

Main author / Presenter
Pedro Martins Ferreira
Eurostat, Unit E2

Read more Read less

Pedro Martins Ferreira is a statistical officer at Eurostat working on environmental footprints, EE-MRIO modelling and quality assurance of EU climate indicators. He is currently part of Unit E2 (Environmental Statistics and Accounts), where he focuses on CO₂ and GHG footprints, quarterly emissions estimates and Physical Energy Flow Accounts (PEFA). He previously led the Integrated Global Accounts team and played a key role in turning the FIGARO project into an official Eurostat production process, as well as in integrating carbon footprint calculations into inter-country input–output frameworks. He has over 20 years of experience in official statistics, with a strong background in input–output analysis, trade methodologies, and statistical production systems.