Emerging Market Economic Data Cannot be Trusted

Brian Sturgess - August 2013


Speed Read
  • The failure by many national statistics offices to regularly update the base year on which GDP growth is measured casts doubt on the accuracy of much published data.
  • The need to account for changes in the dynamic structure of economies necessitates regular rebasing, ideally every 5 years, but these exercises are costly and are carried out irregularly by most countries.
  • A survey of 187 countries using IMF data found that the average period since the base year was updated was 11 years, but 29% of the non-OECD countries had not rebased for over 16 years.
  • Rebasing after a long period of time can have a dramatic impact on GDP revisions, but the infrequency of the practice means that most published emerging market economic data and cross-country comparisons are largely meaningless.



The Importance of Base Years

The accuracy of much international GDP and economic growth data, particularly for emerging and frontier markets, is poor. This makes many cross-country comparisons of GDP, income per capita and growth rates made with the more developed countries of dubious value. A serious problem undermining the quality of published data is the failure of national statistical offices to update regularly the base year, a procedure known as rebasing. Rebasing involves replacing the old base year used for compiling the estimates of goods and services produced and consumed to estimate GDP with a new base year. The new base year becomes the most recent period from which the structure and level of output and relative prices income across all measured economic activities has been surveyed and it is used to determine the appropriate weights of sectors and sub-sectors in GDP calculations. Later GDP figures depend crucially on the comparison of base year information with subsequent estimates of transactions within monitored sectors.

Upward revisions in GDP frequently arise as a result of updating of base years and these revisions occur often in tandem with the application of a change in methodology, the application of the latest recommendations from the international System of National Accounts (SNA) standards[1] on industry classification and the application of improved surveys of industries or activities. Regular rebasing is necessary in order to ensure that new industries such as mobile telecommunications and the internet are included in the latest GDP estimates. In many developing countries the rapid penetration of mobile phones has occurred after the most recent base year used.

It is also necessary to rebase in order to check that the weights assigned to previously measured sectors – services, agriculture, industry – still reflect their continuing importance in total value added. In Ghana’s recent rebasing from 1993 to 2006, the application of SNA 1993 allowed for a broader coverage of economic activities not accounted for in prior GDP estimates leading to a rise in measured GDP and to a shift in the relative significance of sectors. Services replaced agriculture as the largest sector accounting for 51.1% of GDP in 2010 in the new revised data compared to 36.1% using the old base year.

The additional information generated by rebasing also often leads to better measures of unmonitored traditional activities. In many emerging markets the relative size of the unmeasured informal sector in urban areas and non-cash agricultural activities can be estimated by the application of new methodologies when rebasing is carried out. [2] Despite the benefits in the improved accuracy of economic data frequent rebasing carries a trade-off in cost. Rebasing necessitates the application of considerable resources to generate data from new and expanded surveys in and before the new base year. Jerven and Duncan (2012) note that despite a common belief within Ghana that GDP was seriously underestimated using 1993 as a base year the country’s president explained ‘that the Ghana Statistical Office lacked the resources to calculate these statistics.’ [3]

 

Most countries do not rebase often enough

Recommended best practice by the United Nations Statistics Division is to rebase every 5 years, but an analysis of 187 countries found that most countries rebase relatively infrequently. [4]Calculating the number of years between 2012 and the most recent base year the average lapse across all these countries was 10.6 years, with a median base of 2002, twice the recommended period. The analysis also confirms that as expected given the lack of resources and staff training found in many national statistics offices, rebasing and consequent data accuracy is related to the level of economic development. The mean period since the last rebasing was 7.0 years for the 34 richer OECD countries included in the global group compared to an average of 11.4 years for all other countries. The standard deviation across the OECD group at 3.4 years was nearly half the level for the other group and many of these countries were using base with long and varying vintages such as Albania (1996), Barbados (1974), Bolivia (1990), Brazil (1995), Chad (1995), Nigeria (1990), Indonesia (2000), Pakistan (1999) and Thailand (1988) to give a few examples across continents. The base year dates as published by the IMF are shown in the Table broken down by broad geographical region: Europe, Asia-Pacific, Africa and the Americas. [5]

 

Download the fill list of Country/Base Year data in Excel.

 

Although the data shows that some OECD countries have also been tardy in rebasing, as the Figure demonstrates, generally it is the non-OECD group of countries where most problems are encountered with the accuracy of GDP data. For example, 54% of the non-OECD countries had not rebased for over 11 plus years, 29% had not rebased for over 16 plus years and a sizeable 10% had not done so for more than 20 years.

 




Impact of Rebasing

A failure to rebase frequently means that the published level and sector distribution of GDP drifts further away from reality over time as relative prices and industries change making the information increasingly inaccurate and irrelevant. This drift is particularly serious in the emerging markets that have grown most rapidly in the last decade. Recent experience shows that the impact of updating a base year after a long time lag can be significant. In Ghana, updating the base year led to an upward revision of the country’s estimated GDP by more than 60%. [6] At present Nigeria is undergoing an extensive, albeit delayed, exercise to update its base year from 1993 to 2008 and the figures when released are expected to raise estimates of GDP by 50%[7] making it Africa’s largest economy overtaking South Africa. Whereas these revisions do not make citizens better off overnight they cast considerable doubt on any conclusions drawn about the level and speed of economic development in Africa. [8]

According to Jerven and Duncan (2012):

“A ranking of African economies by GDP levels cannot be taken at face value. Current income tables reflect an uneven application of statistical methods, data availability, and country level. The most recent country level estimates reported in international databases are, in large part, automatic data permutations, and the differences are as likely to reflect statistical methodology as economic reality.”

World Economics analysis of 187 countries shows that the problem of infrequent rebasing extends far beyond Africa and doubts about the accuracy of published GDP data can be applied to most emerging markets. This has serious implications for development policy, for the allocation of foreign aid, for the asset allocation strategies of investing institutions and for the credit ratings of sovereign debt to give just a few examples. Until more resources are allocated to better collection of national income data all of these activities, which rely on accurate GDP data, rest on shaky foundations.

 

References

Jerven, M. (2011), Counting the Bottom Billion, World Economics, December: http://www.worldeconomics.com/Papers/Counting%20the%20Bottom%20Billion_f4168a63-84c1-4bb8-b22b-a2291d757e7a.paper

Jerven, M. and Duncan, M.E. (2012) Revising GDP estimates in Sub-Saharan Africa: Lessons from Ghana, The African Statistical Journal, 15, August, pp13-22.

 

OCED (2002), Measuring the Non-Observed Economy, Paris, France

Sturgess, B.T. (2013), Nigerian GDP to rise by 50 percent on improved measurement, World Economics, April, http://www.worldeconomics.com/Papers/NigerianGDP_db19af62-8a3d-42f9-b8e6-721ebdbd4993.paper

 

 



[1] The SNA makes recommendations on the classification of economic activities to be taken into account when measuring GDP. In the most recent and well publicised rebasing exercise by Ghana the country brought its classification of industrial activity which had been based on 1968 SNA in line with the 1993 SNA recommendations. The latest version is 2008 SNA. The historical versions of SNA are described by the Unite Nations at: http://unstats.un.org/unsd/nationalaccount/hsna.asp

[2] See the recommendations in OCED (2002).

[3] Jerven and Duncan (2012) p20

[4] The data was sourced from the IMF.

[5] There remain some inconsistencies between IMF data and national sources which add to the uncertainty about published GDP data. The Table includes only IMF which was used in this study for consistency and for its wide use in studies. For example, China’s base year is given as 1990 whereas official sources record it as 2005. Differences were also found for Ghana and Nigeria.

[6] See Jerven and Duncan (2012).

[7] See Sturgess (2013)

[8] See Jerven (2011).