Why the World Bank's International Comparison Program has limited use for India

  • Blog Post Date 20 February, 2015
  • Perspectives
  • Print Page

Preliminary results from the World Bank’s International Comparison Program, which seeks to compare the economies of 199 countries across the globe, were released recently. In this article, Ranjan Ray, Professor of Economics, Monash University, highlights several features of the exercise that limits its usefulness for a diverse country such as India, and makes recommendations for the next round.

The recently released preliminary results of the 2011 exercise of the International Comparison Program (ICP) (World Bank 2014) have attracted considerable attention from the media and economists. The ICP seeks to provide estimates of the Purchasing Power Parity (PPP) of a country’s currency with respect to a numeraire currency (currency of the reference country), namely, the US dollar. The objective is to make the Gross Domestic Products (GDP) comparable across countries, which in turn can be used to compare the size of the economy, 1 economic welfare of the populations (by looking at per capita estimates), productivity, investment potential, incidence of poverty etc.2

In this column, I point to several features of the ICP exercise that limit its usefulness for large and diverse countries such as India and China, and draw attention to empirical evidence in support of this claim (Ray 2014).

The ICP debate

Although the ICP PPPs have come with the warning that they should not be used in poverty comparisons, that is the area which has seen one of the biggest applications of the PPP estimates. This has led to a fruitless but media-catching debate among experts on what is the ‘correct set’ of PPPs to use in poverty comparisons While the 2005 ICP exercise, published in 2008, led to headlines such as ‘The world is poorer than what we thought’, the 2011 ICP led to claims that ‘The world is richer than what we thought’! Yet, the poor have gone nowhere all this while! While the poverty number crunchers engage each other on the ‘correct estimates’ of world poverty, the poor continue to endure sub-human existence.

The figures for India from ICP 2011 show that the ratio of its PPP to exchange rate, known as the ‘Price level Index’ (PLI), has hardly changed over the period between ICP 2005 and ICP 2011, even though with the high growth rates recorded in India and the consequent increase in prices of non-tradeable items, the PLI should have moved towards one (in other words, PPP should have moved closer to exchange rate)3. That itself raises doubts on the validity of the latest PPPs. The failure of the PPP to move closer to the exchange rate in the case of India raises doubts on the validity of the latest PPPs.

Limitations of the ICP exercise for India

  1. The idea of a single PPP that holds true for all regions and for all the expenditure classes in India robs it of much of its policy usefulness. It overlooks the large regional variation in prices and expenditures in India. Evidence on this is provided in Majumder, Ray and Sinha (2014a) which reports widespread spatial variation in prices across states, while Majumder, Ray and Sinha (2012) extend that evidence to report price variation between rural and urban areas. The former study also provides evidence that suggests that the picture of spatial price variation in India extends to that of spatial real expenditure variation.
  2. The rural-urban differences in prices within India have ramifications for bilateral PPP calculations between India and other countries. For example, Majumder, Ray and Sinha (2014a) found that the rural to rural PPPs between India and Vietnam are quite different from the urban to urban PPPs, such that the overall India to Vietnam PPPs that do not take these sub-national PPPs into account can yield misleading estimates.
  3. The idea that the PPPs hold for all the expenditure classes, especially for the rich and poor alike, is another unrealistic assumption that is particularly limiting in welfare applications. The poor have access to a limited basket of items often with inferior quality; this is not picked up by the prices at retail outlets that are used in ICP calculations since such retail outlets are generally not accessible to the poor and are unrepresentative of their consumption and prices paid by them. In calculations based on unit values of a selection of mostly food items from India’s National Sample Surveys (NSS) and Vietnam’s Living Standard Survey (VLSS), reported in Majumder, Ray and Sinha (2014b), the PPPs between India and Vietnam have been shown to vary sharply across expenditure classes.
  4. The concept of multilaterally determined PPPs involving 200 or so countries, which straddle a wide spectrum of development, is not of much use for bilateral comparisons between India and other Asian countries as their spending habits are far removed from that of the ‘star’ country, USA, used as the numeraire in ICP calculations. In PPP calculations involving India, Indonesia and Vietnam reported in Majumder, Ray and Sinha (2014c), it is shown that the introduction of a third country has a strong effect on the bilateral PPP/ PLI estimates. If the trilateral PPPs differ from the bilateral PPPs in the context of Asian countries - India, Indonesia and Vietnam that have some similarity in their food habits, one can well imagine how much the multilateral PPPs in ICP calculations will differ from their bilateral counterparts. With the rise of trade within emerging economies, especially, within Asia, the usefulness of anchoring all ICP calculations to a global norm that is heavily geared towards advanced market economies is questionable.
  5. Finally, there is little robustness check on the ICP numbers and no estimates are provided of the standard errors of the PPPs; hence, we are unable to assess the preciseness of the estimates. One explanation that is often provided for the vastly different set of PPPs between the 2005 and 2011 ICP exercises is the quality of price data supplied by the participating countries, especially India and China. If that is the case, what is the guarantee that in the next ICP, if there is to be one, another set of price information will not be provided that will help to torpedo the latest PPPs? In ongoing work, my co-authors and I (Majumder, Ray and Sinha 2014c), have shown that even moderate adjustment to the expenditure figures in Indonesia to account for the weekly recall in the SUSENAS (National Socioeconomic Survey) - unlike the monthly recall in Indian and Vietnamese expenditure surveys - can affect the PPPs that will alter, even reverse, the rankings of the three countries.4

Recommendations for the next ICP

First, the governance structure of the ICP should change to give greater representation to the emerging economies and their statistical bodies in the highest levels of decision making on matters such as methodology and data collection. It is remarkable that while the OECD (Organisation for Economic Cooperation and Development), the Eurostat, International Monetary Fund (IMF), the US Bureau of Labour Statistics and, even Statistics Austria, had representatives on the ICP 2011 TAG, there is no such representation from either India or China. India should insist on having a representative from the National Sample Survey Organisation (NSSO) or the Ministry of Statistics and Programme Implementation in the highest level of decision making in the ICP, namely, in the TAG. While some input from academic researchers with expertise in price index numbers is desirable, we should strive to move away from an excessive reliance on them as is currently the case.

Second, there should be availability of sub-national PPP s both within and across countries and those that are specially targeted at the poor. For this, the ICP needs to require the participating countries to improve the collection of the information on prices paid by the poor.

Third, there needs to be a change in focus from global multilateral to regional PPPs. Bilateral PPPs between countries linked geographically, culturally, and by trade, should be calculated and made available along with the multilateral PPPs. This requires less centralisation in the global office of the World Bank and more power and resources to the regional organisations such as the Asian and the African Development Bank. In my view, the Asian Development Bank (ADB) did a much better job of calculating PPPs in the Asia Pacific region than ICP 2005 and ICP 2011 (ADB 2007).

Fourth, there needs to be a change of methodology to accord greater weight to changing preferences within and across countries, through the use of price indices that are based on regional expenditure patterns and regional prices in accordance with utility theory. This requires improvement in household budget surveys and greater synchronisation between surveys conducted in different countries. This is one area where the World Bank and the ICP can take a lead by seeking uniformity in the quantity and quality of information provided by the participating countries. India’s NSS provides a model for other countries to follow since it contains data at state level, further disaggregated between the rural and urban sectors, which is quite comparable across the states. This provides an additional reason for someone from the NSSO to be on the highest echelon of decision making in the ICP.

Finally, there is considerable scope for greater transparency in the ICP by making all the data and related information readily available on the internet. For example, the information on prices at the level of basic headings on which the ICP PPPs are calculated is not readily available. It should be possible for independent researchers to readily replicate the ICP estimates, benchmark them against alternative approaches, and examine possibilities for improvements in the ICP methodology and data collection.

The US$40 million that was reportedly spent in conducting the 2011 ICP exercise requires more accountability and greater transparency to merit similar or even larger commitments by the participating countries in future ICP exercises in collecting and supplying price information. The ICP needs to demonstrate that it is much more than an intellectual exercise indulged in by some Western academic economists. Future ICP exercises should be more meaningful for all the participating countries and for their citizens. India has a big role to play in steering future ICP exercises in such a direction.


  1. A working definition of the PPP is provided in World Bank (2013, p.19): “it represents the number of currency units required to purchase the amount of goods and services equivalent to what can be bought with one unit of currency of the base or reference or numeraire country”.
  2. GDP is the total value of an economy’s domestic output of goods and services. GDPs are not comparable across countries because they are expressed in national currencies and valued at national price levels. But, after PPPs are applied, the GDPs get converted to a common currency and are revalued at a uniform price level. As a result, (after applying PPPs) differences between the GDPs reflect only differences in the volumes of goods and services.
  3. PPPs deviate from market exchange rates due to a host of factors that range from short-term factors such as capital and interest rate movements to longer-term causes such as the fact that while the exchange rates are determined by tradeable items, PPPs are dependent on a wider basket of items that affect cross-country living standard and real income comparisons. In developing countries such as India, the main reason for the difference between PPP and exchange rate is the presence of the inexpensive non-tradeable items that enter the calculations of PPP but not of the exchange rate. As a country develops, the non-tradeable items become more expensive and this drives the PPP upwards towards the exchange rate.
  4. The ‘recall period’ is the period over which the respondent is asked to quantify his/ her purchases. The 55th round of the National Sample Survey (NSS) in India showed that the choice of a particular recall period can have a significant effect on poverty estimates.

Further Reading

  • ADB (2007), ‘Purchasing Power Parities and Real Expenditures, 2005 International Comparison Program in Asia and the Pacific’, Economics and Research Department of Asian Development Bank, Manila, December.
  • Majumder, A, R Ray and K Sinha (2012), “Calculating Rural-Urban Food Price Differentials from Unit Values in Household Expenditure Surveys: a Comparison with Existing Methods and a new Procedure”, American Journal of Agricultural Economics, 94(5), 1218-1235.
  • Majumder, A, R Ray and K Sinha (2014a), “Spatial Comparisons of Prices and Expenditure in a Heterogeneous Country: Methodology with Application to India”, Macroeconomic Dynamics, Forthcoming.
  • Majumder, A, R Ray and K Sinha (2014b), “Estimating Purchasing Power Parities from Household Expenditure Data Using Complete Demand Systems with Application to Living Standards Comparison: India and Vietnam”, Review of Income and Wealth, Forthcoming.
  • Majumder, A, R Ray and K Sinha (2014c), ‘A Unified Framework for the Estimation of Intra and Inter Country Food Purchasing Power Parities with Application to Cross Country Comparisons of Food Expenditure: India, Indonesia and Vietnam’, Monash University, Discussion paper, 31/14.
  • Ray, R (2014), “International Comparison Program of the World Bank: How Meaningful is the Exercise for India? “, Economic and Political Weekly, 30 August, 70-75.
  • World Bank (2013), ‘Measuring the Real Size of the World Economy: The Framework, Methodology, and Results of the International Comparison Program (ICP)’, World Bank, Washington, DC.
  • World Bank (2014), ‘Purchasing Power Parities and Real Expenditures of World Economies: Summary of Results and Findings of the 2011 International Comparison Program’, World Bank, Washington, DC.
No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter