In line with European requirements, the Office for National Statistics (ONS) introduced annual chain-linking (ACL) into the UK national accounts from the Quarterly National Accounts First Release for 2003 Q2. In order to allow ongoing comparison between the Scottish series and those for the UK and other EU countries, the Scottish Executive has now published a chainlinked Quarterly Scottish GDP series from 1995 Q1 to 2003 Q3. The Fixed Base Index Approach The old "fixed base" index was calculated by collecting data from each industry to reflect the volume growth in Gross Value Added (GVA). These data were then converted to indices referenced to 100 and weighted together to form an aggregated index of total GVA. The weights were derived from each industry's relative share of GVA from the Scottish input-output tables in the base year (latterly 1995), and were updated on a 5-yearly basis. Limitations of the 'Fixed-base' Methodology There are a number of limitations to the old approach. These include: - For some industries, over certain periods, the share of overall GVA changes rapidly. This means that for periods between each 5 yearly rebasing such industries' contribution to overall GDP could become out of date. This can lead to over or under weighting of individual industries' contributions and, subsequently the GDP estimates may not accurately reflect actual growth in the economy. This is most likely during periods of rapid technological change and more general economic restructuring.
- Five yearly rebasing can lead to significant revisions to previously published estimates.
- The most common quarterly proxy indicator for GVA is turnover. Turnover is assumed to have a stable relationship to GVA over time. However this assumption becomes less valid as time passes.
- When the fixed base year is updated, choosing the year to link indices based on the previous set of weights to indices based on the new weights (the link year) is a subjective process and the optimum link year varies across industries.
New Annual Chain-linking Methodology The major difference between the old fixed base methodology and annual chain-linking is that the weights will now be updated every year instead of every 5 years. The series of annually re-weighted growths is then 'chainlinked' to produce a continuous series. Benefits of Annual Chain-linking Benefits of the new approach include: - Annual weights will be more up-to-date and therefore more relevant.
- The most important benefit of annual chain-linking is that more relevant weights are used in constructing estimates of economic growth. Fixed base aggregation uses weights that are updated at five yearly intervals. In recent times, particularly since 1995, this is not sufficient. Analysis of values shares shows that the relative proportions in 1995 are not a sufficient indicator of relative proportions 4 years later.
- Rebasing revisions will occur to a much lesser extent. After the initial introduction of annual chain-linking, most remaining revisions will occur as a result of an improvement in the quality of input data for the production of indices or weights.
- Annual chain-linking reduces the inaccuracies caused by the assumption of a stable relationship between GVA and turnover. When annual weights are used, the assumption has to be maintained only from one year to the next, but only from the level at which annual chain-linking is implemented.
- For annual chain-linking, every year is automatically a link year, so no subjective choice is required.
The main drawback to annual chain-linking is that except for estimates after the latest base year (i.e. the latest year for which input-output tables are available - currently 2000) indices no longer aggregate to totals (often referred to as lack of additivity) due to the fact that for earlier years a different scaling factor is applied to each series. This is already the case with fixed base estimates but is not so apparent due to the less frequent rebasing involved with that approach. Effects of Chain-linking on GDP Growth Estimates ONS published their initial findings on the effects of chainlinking at UK level - "Economic Trends No 593 - April 2003". Their findings showed that annual chain-linking had a negative effect on the annual growth of GDP in basic prices from 1998 to 2000, and then a positive effect in 2001. In the same way as the UK estimates, the first chainlinked Scottish GDP estimates for Scotland for 2003 Q3 include the combined effects of chainlinking, and of some revisions for the usual reasons of improved data availability, revised UK sources and revisions resulting from updated returns by individual companies and sectors. However, detailed analysis shows that the effects attributable to chainlinking are slightly larger in Scotland, but generally in line with what was predicted for the UK GDP, i.e. a dampening effect on the period from 1998 to 2000, and a positive effect post 2000. This is mainly due to the fact that the updated weights for Electrical & Instrument Engineering fell in 2000, reducing the negative effect of the major decline experienced in Scotland post 2000, while, simultaneously the increasing strength of the service sector had a greater positive impact on GDP with increasing weight between 1995 and 2000. |