Populating Quarterly Constant Price Supply and Use Tables

Populating Quarterly Constant Price Supply and Use Tables with Seasonally Adjusted Data Simon Compton Methodology Directorate Office for National Statistics United Kingdom

Structure of Presentation • Describe supply and use tables and their use in the estimation of GDP • Introduce some key processes of GDP compilation: balancing, seasonal adjustment, chain-linking and aggregation • Consider some of the interactions between these processes • Describe development work in the UK

Gross Domestic Product • Main measure of economic activity and growth • Can be measured in three ways: Output = Expenditure = Income or GDP(O) = GDP(E) = GDP(O) Supply = Demand: Domestic Production + Imports = Household Expenditure + Intermediate Consumption + Capital Formation + Exports + Change in Stocks GDP(O) = GDP(E) Domestic Production – Intermediate Consumption = Household Expenditure + Capital Formation + Change in Stocks + Exports Imports GDP(I) = wages and salaries + profits

Compilation and Balancing • Each component can be measured separately, • annually, quarterly or monthly • Results in three different estimates of GDP • Data confrontation exercise, known as ‘GDP balancing’ to reconcile these estimates • Results in one ‘best’ estimate of GDP

Supply and Use Tables • Enables data reconciliation at a more disaggregated level: – GDP(O) and GDP(E) by product – GDP(O) and GDP(I) by industry ONS considering a breakdown into 369 products and 197 industries

Supply and Use Table

Example of an Input-Output Table

Seasonal Adjustment

Supply & Use and Seasonal Adjustment Supply and Use tables need to be populated with seasonally adjusted series: • for data validation • facilitates the balancing analysis • focus on final published outputs

Prices • Adjustment needed to take account of inflation in order to estimate ‘real’ economic growth • Achieved by deflation – dividing the ‘current price’ series by a price index • Need some way of weighting together different series – use prices, to generate constant price series combining different goods and services • Relative prices change over time • Chain-linking – generates chain volume measures BUT not additive

Seasonally Adjusted Series Required for Constant Price Supply and Use Tables • Domestic production and intermediate consumption tables for 197 industries by 369 products in current prices and previous year prices • Expenditure components of GDP by product in current prices and previous year prices • Income components by industry in current prices only

Aggregation and Seasonal Adjustment • Aggregation required for supply and use at a very low level of disaggregation • Problematic, particularly for series from sample surveys where sampling error may account for most variation • Practical problems with seasonal adjustment for a high volume of series

Chain-linking and Seasonal Adjustment • Some research indicates greater potential for bias if seasonal adjustment precedes chainlinking • Research based on the UK method of chainlinking suggests the opposite (better to remove seasonality first) • Seasonality may be reintroduced after chainlinking • More research needed in this area • Careful testing for residual seasonality after chain-linking is advisable

Chain-linking, Aggregation and Balancing • Ideally chain-linking should take place at the most detailed level possible - the level at which deflation takes place. • Balancing at a low level of disaggregation difficult unless an automated procedure • A more analytical, judgemental approach is only feasible at a higher level of aggregation

Order of Processing • • • At the level of deflation express quarterly constant prices in previous year prices (PYPs) Aggregate PYPs to the level of seasonal adjustment Chain-link quarterly PYPs to Chain Volume Measures (CVMs). Seasonally adjust this series Unchain this series to create seasonally adjusted current year and previous year prices Aggregate to the level of publication and chain-link

Transition of Seasonal Adjustment • • • Current system involves seasonal adjustment of many thousands of series This seasonal adjustment incorporates much knowledge about the series Strategy for transition involves: 1. Use of a sophisticated and robust automatic option for seasonal adjustment 2. A prioritised programme of seasonal adjustment review, setting up tailored seasonal adjustment for individual series, where appropriate.

Some complications • • • The need for additional published outputs, sometimes using different classifications or dimensions to those of the supply and use tables The interdependent nature of many parts of the accounts with data being used in two places, or estimates derived from one component used in another Some technical adjustments bridging the gap between producer (‘basic’) and retail (‘market’) prices, which are conceptually difficult for seasonal adjustment and deflation. Some particular conceptual complexities and data issues around series for changes in stocks Likewise for intermediate consumption

Conclusions • • • Populating quarterly constant price supply and use tables with seasonally adjusted data not easy and requires comprise Disaggregated level of seasonal adjustment a particular concern Design requires careful consideration of each component of GDP and order of processing issues New design not yet implemented in the UK Many benefits offered by the new framework with the analytical opportunities it offers.