Roger Halliday, Scottish Executive14
Introduction
In September 1999, the Scottish Executive published the first figures from a new model that produces quarterly estimates of growth in Scotland's GDP. The data showed
This article provides technical detail on the methodology and data sources used in the model.
Chart 1: Indices of Output by Sector, Scotland 1994-1999Q2
<<chart1>>
Source : Scottish Executive
Concepts
GDP can be defined as the sum of all incomes in the economy. The latest estimate of the value of Scotland's GDP, using an income approach, is for 1997 and is £56.2 billion. This figure comes from the Office for National Statistics' Regional Accounts. These figures are produced in current prices. As such, the increase in GDP from one year to the next will be due not only to the expansion of the economy, but also because of inflation.
GDP can also be defined as the total value added in the production of traded goods and services. The value added of a business, or of an industry group, is the difference between the value of gross output and that of the intermediate goods and services required to produce the output - i.e. in the manufacture of the goods or provision of the services. This is the basis for the methodology used to calculate GDP using the new quarterly model.
In order to produce growth rates in 'real terms' - these are the most revealing form of GDP figure in the short-term - the new GDP model works in constant prices. Under the 1995 European System of Accounts (ESA95), constant price estimates of an economy's economic activity are termed Gross Value Added (GVA). Whilst there are 3 approaches to the calculation of GDP - namely income, output and expenditure based - the methodology used in the calculation of GVA provides the most robust GDP growth rates.
The new quarterly figures are produced in the form of indices with average value for 1995=100, and it is from these series that growth rates are derived. The series are comparable between industries, and with the indices for the UK produced by the Office for National Statistics.
Methodology
Overview
The new model is based on almost 200 series which are selected as indicators of change in net output15 for sub-sectors across the economy. Examples are deflated turnover and physical measures of a traded good or service. The series are converted into indices and seasonal variation is removed.
GDP growth figures are then calculated using weighted averages of these seasonally adjusted constant prices indices. The weights used are proportional to the contribution of each industry or service to GDP in the base year (currently 1995).
The methodology of estimating growth rates for Scotland is similar to the methodology used by the Office for National Statistics (ONS) for figures on the UK as a whole16.
Producing output based series.
As mentioned above, GDP is calculated using Gross Value Added (GVA). Theoretically, the calculation of GVA requires both gross output17 and intermediate goods and services (inputs) to be estimated and deflated to constant prices. However, due to the large amounts of data that are required, this method is only used for the agriculture sector. Instead, gross output is used as a proxy indicator for GVA.
Therefore, this approach assumes that indicators of change in output are proportional to change in value added. This is generally a reasonable assumption over a period of a few years. In order to use output as a reliable indicator over a longer period, the weight each industry contributes to GDP is updated every five years. This means that current estimates are compiled with weights based upon 1995 data (at 1995 prices), whilst figures for the early nineties and late eighties were calculated on a 1990 basis. (see also section on derivation of weights).
As mentioned above, data on output from businesses surveys arrive in current prices. In order to get estimates of changes across time, these need to be converted to constant prices by sets of deflators. A range of UK deflators is used, in a similar way to ONS in their calculation of GDP. Different deflators are used for different industries; these include disaggregated Retail Prices Index figures, information on consumers expenditure, Producer Price Indices, and Corporate Service Price Indices.
Seasonal adjustment procedure
All indices are examined for the presence of seasonal variation. Where a stable seasonal pattern is identifiable, it is removed using a multiplicative model under the Statistics Canada's X11-ARIMA procedure. This is carried out separately for each detailed (3-digit SIC) industry. Seasonally adjusted estimates are then aggregated for publication.
Derivation of weights
Over time, as industries grow or decline, weights used in previous years may no longer reflect each industry's true share of overall GVA. Therefore, the weights are updated (rebased) every five years. The 1990 and 1995 weights are given in units of parts per thousand and shown in the chart below for the main industrial sectors.
Weights for over 100 industrial groups are taken directly from the Scottish Executive's input-output tables. These weights are then apportioned to each of the 200 or so series which derive GDP, by using information on the UK GDP weights and the proportion of UK employees working in Scotland.
For most industries, the weights used in Scotland are quite similar to those for the UK. However, there are some industries where the weights differ significantly. For example, the hotel sector accounts for 0.87% of UK GDP, but an estimated 1.1% of GDP in Scotland. This reflects the larger number of hotel employees in Scotland proportional to the UK.
Chart 2: Weights Used to Calculated GDP in Scotland, 1990 and 1995
<<chart2>>
Source: Scottish Executive
Quality assurance
Throughout the derivation of GDP, steps are taken in order to minimise any errors present. When data arrive, each of 200 or so series are checked for outliers, inconsistencies and trend changes using a dedicated computer system which continues to analyse the quality of the data at each stage of developing the GDP estimates.
Data sources
A summary of the main sources of data for each sector of the economy is given below.
A full list of sources is available on request.
|
Agriculture Forestry and Fishing |
Gross output at constant (1995) prices is estimated by the Scottish Executive's Rural Affairs Department. |
|
Production |
The Index of Production is derived largely from the quarterly returns of the value of turnover of 1,500 businesses in the production sector in Scotland. This and around 20 constituent sub-series are produced and published each quarter by the Scottish Executive. |
|
Construction |
The Index of Construction is based on data from the Department of the Environment, Transport and the Regions. It covers construction in both the public and private sectors in Scotland. |
|
Services |
Returns of the value of turnover of 3,500 businesses18 each quarter are used to derive series for the following sectors: Retail & Wholesale, Hotels & Catering, Communications and Business services. |
|
Sources for the Financial Sector include: Fee and Commission income series from the Association of Scottish Clearing Banks. Indicators of activity in the Scottish Building Society sector from the Financial Services Authority. |
|
|
Figures of Commission and Management expenses are taken as indicators of activity for the Life and non-life Insurance sectors. |
|
|
Since output proxies for value added in the public sector are more elusive, employee numbers form the basis for indicators of activity for the large parts of Public Administration, Education and Health sectors. |
Available statistics
Estimates are produced separately for 12 industrial sectors of the economy. Further detail is given on 11 separate parts of the manufacturing sector in the Scottish Index of Production. The industries shown are:
|
Agriculture, forestry and fishing |
Construction |
|
Production |
Services |
|
- Mining and quarrying |
- Retail and wholesale |
|
- Electricity, gas and water supply, |
- Hotels and catering |
|
- Manufacturing |
- Transport, storage and communication |
|
of which |
- Financial services |
|
- Refined petroleum products & nuclear fuel |
- Real estate and business services |
|
- Chemicals & man made fibres |
- Public admin., health and education |
|
- Metals & metal products |
- Other services |
|
- Mechanical engineering |
|
|
- Electrical and Instrument engineering |
|
|
- Transport equipment |
|
|
- Food and tobacco |
|
|
- Drink |
|
|
- Textiles, footware, leather & clothing |
|
|
- Paper, printing & publishing |
|
|
- Other manufacturing industries |
Further disaggregated series are not published because the sources of data used to calculate the indices are not robust enough to give reliable estimates of change for very small industrial sectors.
Future work
The development of the current methodology used to generate the Scottish GDP estimates required significant resources. However, these methods will need to be continually assessed and reviewed in order to ensure that the most reliable results are produced.
Over the next few years, this means reviewing techniques for seasonal adjustment, for estimating missing data, and for making revisions to data already released. In addition, alternative data sources for different sectors of the Scottish economy will be investigated and used where appropriate.