Gross Domestic Product (GDP) estimates are some of the most heavily requested and used data published on data.worldbank.org . And as many users notice, the estimates are sometimes revised, occasionally resulting in large changes from previously published values. Why do revisions happen, what information do we publish about those revisions, and where do you find it?
Let’s take an example: the case of Ghana. In November 2010, the Ghana Statistical Service improved their national accounts series by incorporating new data sources and better estimation methods, classifications and standards, as well as “re-basing” the volume estimates. As a result, the estimated level of GDP for 2006 was revised upwards - compared to previously published estimates, the increase was some 60 per cent. [The details of this revision are provided in an information paper  on the Ghana Statistical Service website .]
The Ghana case is particularly striking because of the size of the revision, but it’s worth noting that dramatic changes in GDP levels are less likely to occur if countries revise their national accounts methodology and incorporate new source statistics more frequently. Additionally, the older the base year (i.e. the pricing period used for constant price or “real terms” calculations), the more outdated the weights, resulting in lower quality estimates of GDP growth and volumes. In fact, the freshness of the base year of the national accounts is used as a key indicator of statistical capacity in the World Bank’s Bulletin Board on Statistical Capacity (http://bbsc.worldbank.org ).
I know that some data users compare our latest data with previous versions - they may want to check, for instance, if their previous analysis still holds with the revised data. In fact, we use quality assurance procedures to do the same, before we update our World Development Indicators database; it helps us review any major changes for accuracy, and understand the reasons for any major revisions. To give you a peek “behind-the-scenes”, this is what my screen looks like when I am doing this for GDP data for Burundi.
As a result of this and other review processes, we are able to document those countries with revised GDP estimates in the primary data documentation section of the World Development Indicators publication (http://issuu.com/World.Bank.Publications/docs/9780821389850/429 ); we also provide the national accounts base year for each country. We’re working hard to move this and other documentation to data.worldbank.org , and to improve the information available – as we do, it would be very useful for us to know what would be most useful to you. Please take a look at what we have, and let us know.