Reviewing the World Bank’s Analytical Country Classification: An Update


This page in:

On July 1, we updated the analytical country classification, which groups economies of the world into four categories based on 2012 GNI per capita estimates: low income, lower-middle income, upper-middle income, and high income. This has prompted some questions related to the review of this classification scheme, which we announced late last year and for which we solicited and received your feedback. I thought it would be useful to post an update.

First let me reiterate an important point that I made in my original post, and appears to have been misunderstood by some readers: the focus of this review is the analytical country classification, which is provided purely for the purposes of analysis. Possible uses include the calculation of aggregate indicator values for each of the four groups, or tracking the change in the classification of specific countries over time. Please note that this classification system is not used by the World Bank for resource allocation purposes.

Our work on the analytical classification so far has gathered perspectives of various users inside and outside the Bank, and has identified three main areas of priority focus that we will examine further. These are:

  • Relative vs absolute groupings. Some users prefer a system which maintains constant, “absolute” thresholds over time, while others would prefer a relative method of classification. Methods for creating relative groupings range from simple country rankings to statistical techniques such as cluster analysis.
  • Use of other variables instead of – or as well as – GNI per capita. Some users have expressed interest in a classification based on various measures of welfare, poverty, or economic and social progress, while others find the current measure fit for their purpose.
  • Adjusting the current methodology to take account of changing circumstances, including the availability of improved data. Options include a change to the method used to convert GNI per capita to a common currency, particularly to consider the use of the new purchasing power parity estimates that will be released by the International Comparison Program in December 2013; methods to reduce the sensitivity of the classification to revisions of estimates of GDP or population size, and to exchange rate movements, especially for countries that are close to each threshold value; and improvements in the method used to maintain the threshold values constant in real terms.

Going forward, we intend to examine each of these issues more closely, including conducting empirical work to examine the impact of any change. It is clear from our initial work that the analytical classification is very important to many more users than we originally knew, so we will proceed deliberately and carefully, and will aim to post findings of the work as it progresses. Users should be aware that the review will take some time, especially since we will need to fully evaluate the results of the 2011 ICP following the expected release of new benchmark purchasing power parities in December 2013. Consultations with users remain central to us as we move ahead, thus I would welcome your feedback on the priority areas identified so far, and any other comments on whether the analytical classification is fit for your purpose; while we have tried to identify as many users as possible in our initial review, we do not know all the uses to which the classification is put. I encourage you to post your comments on this blog, or write to either myself ([email protected]) or the project manager for this review, William Prince ([email protected]).


Shaida Badiee

Co-Founder and Managing Director, Open Data Watch

Join the Conversation

Will Prince
July 12, 2013

The original analytical high-income threshold of $6,000 in 1987 prices that was created in 1989 was agreed to by the Bank's Executive Board in order to have all the countries previously classified as "industrial" classified as high income. Updating that threshold each year with the international inflation rate gets you to $12,616 in 2012. It's important to stress Shaida's point that the analytical classifications have no bearing on the World Bank's lending operations. When consulted by users we advise they apply these classifications flexibly and consider other factors if they are going to use them for any purpose, such as in their allocation decisions, since (as noted on the classification website) "GNI does not, by itself, constitute or measure welfare or success in development." No single measure does. You ask if there is a rationale for these thresholds? One purpose for the Data Group is as a way to group countries for presentation in publications, such as World Development Indicators. Are there better ways to group countries than the method we are using? That's what our review is all about. As such, we look forward to the results of the 2011 round of The International Comparison Program (ICP) to be released in December (, as a careful analysis of the new purchasing parity power data will inform our decision on future classification schemes.

M Ravallion
July 13, 2013

The fact that the $6,000 cut-off in 1989 ($12,616 today when updated for international inflation) assured that all the countries previously classified as “industrial” were re-labeled “high income” does not get us very far in understanding the meaning and relevance today of the $12,616 threshold, or even the relevance of $6,000 in 1989.
So I am none the wiser on the answer to my original question: Why $12,616? Your website claims that there is a “stable relationship with poverty and infant mortality.” That would be interesting to see, but users of your website can find no documentation. Every year (for many years) these income classifications are revised with essentially the same minimal and puzzling documentation. And believe me the attention these classifications get is not just from “analytic users.” They have huge influence.
If there is really nothing to these classifications that the Bank has produced for decades then it would be good to finally tell users. And, as I say in my reply to Shaida’s post, while the review you refer to is welcome, it is surely unwise to keep updating and publishing these classifications when you have not yet sorted out how they should be done.

Martin Ravallion
July 12, 2013

Thanks for the update. We are still in the dark about the basis for these influential country classifications that you and your team keep publishing. For example, as I asked in my CGD post with reference to the “High-Income Country” threshold, why $12,616? (
It seems a stretch to suggest that your country classifications are only intended for “analytic” purposes. (And, as an aside, I am not sure what those purposes may be.) You must know that these income classifications are watched closely by (governmental and non-governmental) aid agencies across the globe. For example, the UK government uses them in its foreign aid decisions, which has led to much concern; see for example, Jonathan Glennie’s post just yesterday regarding India, on the Guardian’s Poverty Matters blog. (
If there is some rationale to these thresholds then please tell us. If not then we deserve to know that too, so your well-intentioned data users don’t put more weight on the numbers than they can reasonably bear.
The review you describe (as last year) is welcome. But maybe you should seriously consider not publishing these classifications until you have sorted out how they should be done, in your ongoing review.
Martin Ravallion