A new method to reconcile inconsistent data with an application to 2017 regional EU NUTS2 trade-linked world input–output tables
Entropy-based estimation techniques are commonly utilized for reconciling inconsistent economic data, drawing on information theory. While prevalent for integrating supply-use or input–output datasets from disparate sources into coherent national account systems, we demonstrate that entropy methods overweight deviations in small figures compared to large ones. This can yield undesirable reconciled estimates given the magnitude variance of typical economic data. Particularly problematic are analyses of trade shocks (e.g. Brexit), which inherently involve high variation in trade flows.
To address this limitation, we advance a root mean square weighted error and percentage error (RMSWEPE) approach for balanced reconciliation, computationally efficient for large datasets. Applying RMSWEPE, we generate 2017 regional EU28 IO tables showcasing the methodology's suitability for constructing consistent regional inter-country trade accounts.
Our advancement provides better estimates of reconciled figures across variable datum magnitudes, benefiting a broad field of research that reconciles and analyzes data with considerable magnitude variation.
Authors
Specifications
- Publication title
- A new method to reconcile inconsistent data with an application to 2017 regional EU NUTS2 trade-linked world input–output tables
- Publication date
- 22 October 2026
- Publication type
- Article
- Publication language
- English
- Magazine
- Economic Systems Research
- Issue
- 2026
- Product number
- 5516
