|
Abstract:
|
Supply-Use (SU) balancing is a data reconciliation exercise used by National Statistics Institutes (NSIs) to reach a "best estimate" of GDP from the many data sources used to calculate the three GDP measures and their components. SU balancing consists of adjusting the initial component estimates such that the adjusted estimates are consistent with the National Accounts identities. Such adjustments must be allocated to the components according to the relative reliability of their source data. Currently, SU balancing is usually done manually; this is challenging and time consuming. However, enabled by recent developments in available computing power, many NSIs are developing software programs to automate it. The paper describes the proposed UK approach, explaining how and why it differs from other NSIs, and summarises UK results and priorities for further work. The proposition is not to seek a best linear unbiased estimator or maximum likelihood, as sufficient metadata is unavailable, but to improve balancing efficiency while at least not worsening results as measured by current quality measures (adjustment size). Results show that the automatic solution delivers smaller adjustments.
|
