Bangladesh's recent promotion to lower middle- income status by the World Bank has created extensive discussion in the media on its implications and path forward for the economy. For instance, Dr. Biru Paksha Paul carried out an interesting exercise, based on different growth rates, to understand when Bangladesh can graduate to the next income category. A further extension to such an analysis is to understand how sensitive these country classifications are to methodological aspects. For policy-makers to view this upgrade as only a reflection of economic progress in Bangladesh, stability of these classifications from a methodological viewpoint is of utmost priority. This writer argued that these are not stable. Policy analysts in Bangladesh need to test stability of the World Bank's methodology using relevant economic perspectives.
First, as is well known, the World Bank uses the Atlas method to calculate Gross National Income (GNI) per capita in order to smoothen fluctuations in exchange rates. The method uses the IMF's unit of account - Special Drawing Rights (SDR) - to calculate international inflation, called SDR deflator. This SDR deflator is calculated as a weighted average of the GDP (gross domestic product) deflators of Japan, the United Kingdom, Euro area and the United States. The weights represent each country's currency in one SDR unit. However, the rise of the Chinese Renminbi in international monetary transactions might lead to a new member in the SDR basket in the near future to medium term. This implies a substantial adjustment in the SDR deflator and a mechanical revision of Bangladesh's per capita GNI without any changes in domestic economic conditions.
Second, note that the Atlas method uses current price levels while computing GNI per capita. For low-income countries which experience high inflation, a real GNI per capita measure gives more accurate information on economic activities. Naturally, one needs to ask how different economic growth rates are based on Atlas method from price-adjusted measures. A look at the figure below shows stark difference in growth rates between GNI per capita using Atlas method and that expressed in constant local currency units (LCU). What is most alarming is that the two growth rates take opposing direction in certain years such as 2006 and 2014.
Figure 1 Growth rate of GNI Per Capita
 Source: World Bank Word Development Indicators (2014)
Third, country classification thresholds have remained fixed in real terms over time. The global price level, as measured by the World Bank's operational threshold, increased at a much slower pace than average per capita nominal income since 1987 (the base year for the current classification system). This resulted in a downward trend of the absolute thresholds relative to average world income. This raises concerns about using these absolute thresholds. Because if we assume that the average world income will continue to rise, there could be a time when the high-income threshold falls below the average world income, which is counter-intuitive. Similarly, it is possible for no country to be classified as a low-income country even though there may still be millions of people in poverty. China and India's rise to middle-income status confirms this.
Fourth, the relevant literature has acknowledged that the market exchange rates underestimate purchasing power in low-income countries. This is because non-traded goods and services are undervalued in these economies. A look at cross-country ranking of GNI per capita based on Atlas method shows that Bangladesh ranks 181 out of 213 economies. Comparing with GNI per capita based on purchasing power parity (PPP) adjusted method shows that Bangladesh's rank is 171 - a substantial 10-place difference. This should not be surprising given that the world per capita income based on PPP method is only 1.5 times that of Atlas method while Bangladesh's PPP-adjusted national per capita income is more than 3 times its Atlas-based measure (Source: Gross national income per capita 2014, Atlas method and PPP, World Bank). It thus becomes essential for Bangladesh policy-makers to compare the time-series behaviour of PPP-adjusted GNI per capita with the one derived from the Atlas method.
Finally, can a recently-promoted country be quickly 'demoted' again? It is quite possible for a country to experience a temporary higher-than-equilibrium growth rate due to one-time shocks to economic factors. This implies a country can temporarily get bumped up in country classifications only to be taken back down again (in case it is very close to the threshold). Figure 2 below shows the number of demotions by the World Bank since the current classification system was adopted in 1989. Total number of downward revisions is 78. Furthermore, four countries were demoted thrice, while seven countries faced this unfortunate phenomenon twice (Source: Analytical Classification of Income, World Bank).
One might argue that these revisions reflect worsening economic conditions and has nothing to do with methodological issues. But then why were countries like Albania (1995-1997 and 2010-2012), China (1996-1998), Malta (1999-2001) and Lesotho (1995-1996) revised up and then down in the space of just two to three years? Is it not the case that a country has to be above (or below) an income threshold for three consecutive years to be recognised as part of that income group? Or is that guideline only used when a country first makes the leap (or fall)? If indeed this three-year 'wait-and-see' approach is used for first-time changes only, there needs to be strong economic justification for not following the rule for subsequent revisions. Â
 Figure 2 Year-wise Demotion in country classification
Source: Historical Classification of Income, World Bank (Base year of current classification system is 1987)
Ultimately, the World Bank's use of GNI per capita for country classification is understandable given its high correlation with most economic variables as well as lack of better alternatives. Bangladesh's promotion indeed reflects impressive strides in major economic factors. But the issues raised above create tension for policy-makers when making macro-level inferences and decisions based on this recent promotion.
The writer is a macroeconomic analyst and student of Master of International Economics and Finance, Johns Hopkins University, Washington D.C., USA.
Email: shaque4@jhu.edu