Please note - this is a temporary window. id21 is joining forces with Eldis and therefore the id21 website has been suspended. Soon all id21 content will be available on the Eldis website.

Getting the statistics right in the Indian poverty debate

One fifth of people living on less than US$1 a day are Indian and discussions about poverty dominate Indian politics. Reforms introduced in the early 1990s were followed by a period of high economic growth and many have related this to a fall in poverty. However, both the reforms and the claims that they have led to poverty reduction remain controversial.

India’s official poverty estimates are based on the consumer expenditure surveys conducted by the National Sample Survey Organisation (NSSO). The NSSO conducts large consumer surveys every five or six years and smaller surveys every year. Research from the World Bank examines the relationship between domestic policy and the collection of survey statistics in India during the 1990s and how these statistics have been used to explain poverty rates in the country.

Estimates between the 1993/4 and 1999/2000 surveys showed a reduction in poverty from 37 percent to 27 percent in rural households, and from 33 percent to 24 percent in urban households. Although the government follows these estimates (which are its own), their validity was questioned because of changes in survey design and length in reporting period, which have a large effect on the amount of measured consumption. However, the Indian national accounting estimates of consumption grew more quickly than the survey estimates and suggested an even bigger reduction.

Everyone agrees that economic growth reduces poverty. And those who believe that 1990s liberalisation reforms have led to poverty reduction argue that the national accounts are right. But recent statistics have worried analysts, as the survey figures on consumption are only two-thirds of those reported by national accounts. Why there are such differences is unclear, but the authors suggest:

One likely explanation is that households have changed their consumption habits, away from food products to non-food goods and services that are perhaps less well measured in the surveys.
Where discrepancies between the national accounts and surveys occur on food consumption data, the national accounts are more likely to be wrong.
Fixing bad data is not an acceptable substitute for obtaining clean, credible data.
There is not enough evidence to say which of the reporting periods was more correct.
Different poverty lines, used to set the base level by which poverty is measured, make a dramatic difference to the survey results.
The Market Information Surveys on Households also suggest poverty has fallen, although it uses only a single question on income.

There is no agreement on what happened to poverty in India in the 1990s. There is evidence to suggest that poverty did fall, but official estimates of poverty reduction may have been too optimistic. India’s experience demonstrates the importance of annual consumption or income surveys as insurance against the problems of the five-year surveys, which include changes in market conditions, differences in harvests and new policies.

Recommendations for the future include:

Survey design should change slowly according to the frequency of poverty monitoring, with revisions only once every five years if surveys are run every year.
When major design changes are made, they should be randomly split between the old and the new design to provide a basis for both backward and forward-looking comparisons.
Policymakers need to ensure adequate financing for annual surveys.
High-quality domestic researchers are essential for creating debate among the research community and the press, to help hold policymakers to account for their strategies.

Source(s):
‘Data and Dogma: The Great Indian Poverty Debate’, The World Bank Research Observer 2005, Vol.20, No.2, pp.177-199, by Angus Deaton and Valeria Kozel, 2005 (PDF) Full document.

Funded by: The World Bank

id21 Research Highlight: 27 June 2006

Further Information:
Angus Deaton
328 Wallace Hall
Princeton University
Woodrow Wilson School
Princeton, NJ 08544
United States

Princeton University

Views expressed on these pages are not necessarily those of DfID, IDS, id21 or other contributing institutions. Articles featured on the id21 site may be copied or quoted without restriction provided id21 and originating author(s) and institution(s) are acknowledged. Copyright © 2009 IDS. All rights reserved.

id21 is funded by the UK Department for International Development. id21 is one of a family of knowledge services at the Institute of Development Studies at the University of Sussex. id21 is a www.oneworld.net partner and an affiliate of www.mediachannel.org. IDS is a charitable company, No. 877338.