Gaaitzen de Vries (University of Groningen) | Rebasing sector GDP time series

28 May 2015.

Gross value added (VA) in current and constant prices by sector is normally available from the National Accounts (NA). As these accounts are compiled according to the UN System of National Accounts, international comparability across countries is high, in principle. However, national statistical offices frequently update their methodologies. In the NA, VA series are periodically revised which includes changes in:

  • Changes in base year of the prices used for calculating volume growth rates.  The main change is the reference period for the individual price and volume indices, which used to be equal to the old base year and are now equal to the new base year.For example, Nigeria revised its base year from 1990 to 2010.
  • The coverage of activities, for example after a full economic census has been carried out and “new” activities have been discovered. For example, new activities, such as the movie industry and telecommunications (mobile telephony etc.), were included in the recent GDP revision for Nigeria.
  • Changes in the methods of calculation. For example, new methods and procedures in the UN system of national accounts were implemented with the recent GDP revision for Nigeria. These methods and procedures differ substantially and in many respects in different UN Systems of National Accounts (e.g. between the 1993 UN SNA and the 2008 UN SNA).

The general approach in constructing the Groningen Growth and Development Centre (GGDC) 10-Sector Database (version 2014); is to start with VA levels for the most recent available benchmark year, expressed in that year’s prices, from the NA provided by the National Statistical Institute or Central Bank. Historical NA series are subsequently linked to this benchmark year. This linking procedure ensures that historical growth rates of individual series are retained although absolute levels are adjusted according to the most recent information and methods.

The VA series in current (nominal) and constant (volume) prices at the ten sector level are constructed as follows. First, the nominal time series are constructed by linking historical data to the benchmark level estimates, see Equation (1). Growth rates of the price deflator are calculated at the sectoral level, using the nominal and volume data from the same source. This results in continuous series of nominal VA data and price deflators at the sectoral level, see Equation (2). Finally, sectoral volume series of VA are constructed by deflating the nominal values, see Equation (3). This approach allows one to choose the base year, in the GGDC sector database the constant prices are now expressed in 2005 prices.

For each sector back casted values from the benchmark years are calculated using:

calc

Estimating volume data from separately constructed series of VA and price deflators has a number of advantages. In cases where statistics at the detailed sector level are missing, we assume that growth trends of the aggregate sector are representative of the underlying detailed sectors. It is more reasonable to make this assumption for price developments than for volume growth rates. In addition, this method allows us to add information on price developments from external sources, such as the consumer price index, when this is not available from primary sources.

Value added series are periodically revised. Typically, the statistical office makes an effort to revise the value added series backwards as well. Say, a benchmark revision from 2000 to 2010 would be accompanied with a new value added time series from 2000 onwards. The statistical office can use additional data to improve upon the backward extrapolation presented above. We therefore prefer the new value added time series and then link these backwards (in this example for 1999 and earlier).

However, we in the recent update of the GGDC 10 sector database the GGDC team has have been reluctant to revise the series for Nigeria following its benchmark revision. One reason for this is that the revised series are not extrapolated backwards by the statistical office. The new series are for 2010 to 2013. The break in the series is large and therefore it would have been helpful to see how earlier GDP estimates would be re-estimated taking the new sources and methods into account. Has growth been underestimated in historical series and if see by how much and in which sectors? A second reason is that the revision has been so comprehensive that it might well be that some further substantial revisions will happen in the near future. For now, the revised data for Nigeria are presented separately in the GGDC sector database. The methods presented in this note can be used to create time series based on the new benchmark estimates or incorporate new revisions in other countries.