Download presentation
Presentation is loading. Please wait.
Published byBritney Dorsey Modified over 8 years ago
1
Contribution to the joint OECD/Eurostat Task Force on “Performing Revisions Analysis for Sub-Annual Economic Statistics” Leendert Hoven Statistics Netherlands Using results from revisions analysis to improve compilation methods: a case study on revisions of Dutch estimates of GDP volume growth
2
Subsequent estimates of GDP growth Quarterly Accounts, flashafter 45 days Quarterly Accounts, regularafter 90 days Yearly estimate, based on QAApril t+1 National Accounts, provisionalJuly t+1 National Accounts, revised provisionalJuly t+2 National Accounts, finalJuly t+3
5
Summary statistics mean revisionmean absolute revision Flash – QA regular 0.060.23 QA regular – NA 1st annual estimate 0.070.15 NA 1st annual estimate – NA provisional 0,100.32 NA provisional – NA revised provisional 0.040,22 NA revised provisional – NA final 0.200.26 Flash – NA final 0.470.58
6
Graph 3. Mean revision and mean absolute revision between consecutive estimates of GDP volume growth, 1990 – 2003
7
Average contribution of components to revision of GDP growth between QA regular and NA final, 1990-1998 production approachweight %contribution % agriculture, hunting, forestry, fishing315 mining & quarrying27 industry15-4 electricity, gas & water2-3 construction5-18 wholesale, retail trade, repair, hotels &restaurants 1430 transport, storage, communications720 financial intermediation & business services 2435 government1110 banking services1012 health & social care, other services101
8
Average contribution of components to revision of GDP growth between QA regular and NA final, 1990- 1998 expenditure approachweightcontribution household final consumption expenditure 4951 government consumption expenditure 24 gross fixed capital formation 23-11 exports minus imports 415 changes in inventories 027
9
Two important findings biggest mean revision between revised provisional estimate and final estimate. Quite large differences between short-term statistics and structural business statistics. Methodological shortcomings (matched pairs approach sts insufficiently allows for effects of population dynamics). within production approach, commercial services and trade have contributed most to the revisions of GDP growth; short-term source situation particularly bad.
10
How can we improve early estimates? improve methodology short-term statistics improve short-term source situation for commercial services and trade more research needed on: ovalidity different assumptions; oeffects of balancing process; o ……………….. o........................
11
Bias correction at the aggregate level? no guarantee that average bias of past period will also hold in the future ignores the underlying data and subaggregates impact of introduction mew methodologies, changes in data collection, sampling or data processing difficult to capture in the short term
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.