Poverty has indeed fallen in India, declares the Planning Commission from time to time, opening up the debate on poverty measurement all over again. A visual look at the debate-filled history of counting the poor in India.
Where it all began
The history of counting the poor in India can be dated back to the 19th century. The earliest effort to estimate the poor was Dadabhai Naoroji’s “Poverty and Un-British Rule in India” in which he estimated a subsistence-based poverty line at 1867-68 prices. Using the diet prescribed to “supply the necessary ingredients for the emigrant coolies during their voyage living in a state of quietude” (ibid, p 25), which includes “rice or flour, dhal, mutton, vegetables, ghee, vegetable oil and salt”, he came up with a subsistence cost based poverty line, ranging from Rs. 16 to Rs. 35 per capita per year in various regions of India (Mehta & Bhide).
Planning Commission – measuring consumption
The Planning Commission has been estimating the number of people below the poverty line (BPL) at both the state and national level based on consumer expenditure information collected as part of the National Sample Survey Organization (NSSO) since the Sixth Five Year Plan. The latest available data from such surveys is from NSSO conducted in 2004-05.
Since 1971, the Planning Commission has based its classification on the cost of calorie consumption in rural and urban India. In 1999-2000, the NSSO switched to a method known as the Mixed Reference Period (MRP) measuring consumption of five low-frequency items (clothing, footwear, durables, education and institutional health expenditure) over the previous year, and all other items over the previous 30 days. Until 1993-94, NSSO measured all consumption across a 30-day recall period. This made comparison of poverty estimates across previous exercises irrelevant.
Till the 1990s, consumption expenditure was determined based on the National Accounts Statistics. The National Accounts estimates consumption of a certain good by considering factors such as total production, imports, exports and consumption by specific parties such as government or business. The Planning commission used the National Accounts estimate of consumption as a “control” total. In other words, if the national accounts estimate was greater than the survey estimate of consumption, the Planning commission would multiply the total expenditure of each household by that ratio before calculating the number of poor. This substantially reduced the number of people below poverty line especially when the difference between the two estimates is high.
While the two estimates were comparable in the 1960s, the difference between them has continued to grow over the years. During the 1990s, the national accounts estimate of mean consumption grew much more rapidly than the survey. By 2005, the survey estimate of consumption was around 2/3rds of the national estimate.
Over the years, researchers and practitioners have argued against the ‘scaling’ of poverty estimates based on the National Accounts, contending that the survey estimates are no less accurate. Specifically, the following challenges were highlighted: The differential definition of consumption between the National Accounts and survey approach, the differences in timing of the two exercises, and the heavy reliance of the national accounting practice on various rates and ratios that link observable but irrelevant quantities to the unobservable and relevant ones. The practice was eventually abandoned by the Planning commission during the 1990s.
The Tendulkar report – counting essential goods and services
The fundamental difference between the Tendulkar Committee’s approach and the Planning Commission’s approach was that the Tendulkar panel moved on the premise that people didn’t just have to spend on food but also other services such as health, education, durable goods, and entertainment. The committee also reduced the calorie intake requirement in 1973-74 from 2100 to 1776 calories in urban areas and 2400 to 1999 calories in rural areas per person per day. The contention was that people in rural areas consumed far less calories for the same income today compared to what they did in the early 70s.
This was one of the key criticisms against the committee’s report. FAO reduced the calorie intake norm to 1770 as a Minimum Dietary Energy Requirement (MDER) for a person engaging in “light physical activity.” An example of this kind of activity is “a male office worker in urban areas who only occasionally engages in physically demanding activities during or outside working hours.”
The Tendulkar report also overlooked inclusion of the aged, destitute, primitive tribal groups, the disabled, single women, widows, and pregnant and lactating women in the category of poor – groups that the Supreme Court has already directed the Government of India to automatically include.
Ministry of Rural Development – the BPL survey
Beginning with the 1992 BPL survey, the Ministry of Rural Development (MoRD) has been organizing a survey/census every 10 years for identifying below poverty line households in rural areas. The census results are used by the MoRD and other departments to target beneficiaries for the various programmes they offer.
The first BPL census (1992) used income as a way of classifying the poor, and guidelines were issued to assess the annual income of the family. This approach was criticized because it was based on the self-reported annual income of rural households: income was not straight forward to explain or to measure, especially for rural households where many are self-employed in agriculture (Deaton & Kozel, 2004).
The 1997 census changed the criteria from income to consumption. However, some of the criteria for exclusion were very stringent. For instance, possession of a ceiling fan excluded families from being part of the BPL list; so did two hectares of land, which could be very unproductive in a few regions of India.
The 2002 census was based on a Score Based Ranking (SBR) of each household, indicating their quality of life. Families were identified as poor based on 13 criteria. The aggregated score based on these indicators were compared with cut-off scores of the states to identify the poor. The BPL census approach came under severe criticism from all quarters for its inherent complexity and lack of transparency in identifying the poor. There was lack of clarity in the criteria, drawbacks in how the scoring and aggregation was done, and increased probability of wrong selection. For instance, parameters such as availability of clothes were not directly verifiable or observable. In addition, the survey used factors such as provision of toilets, housing and education to identify the poor. It was feared that this will disincentivize the rural families from investing in these needs for the fear of being excluded from the BPL list. The census was not operationalized until 2006 due to a stay order passed by the Supreme Court on a writ petition filed by the People’s Union for Civil Liberties (PUCL).
The NC Saxena Committee report
Given the criticism for the 2002 approach, the N C Saxena Committee report proposed an improved method for the next BPL census (SECC 2011), which involved the continuation of the scoring method in a simplified form. Instead of 13 indicators with a scale of 0 to 4 for each, the report proposes just five indicators (essentially focusing on community, land ownership, occupation, education, and old age or illness), with an aggregate score ranging from 0 to 10. The report recommended an automatic inclusion criterion for the most vulnerable sections of society (E.g. homeless people, persons with disability) who would all automatically get BPL cards. The report recommended a cut-off line for determining BPL status as `700 for rural areas and `1000 in urban areas since the existing basis of `365 per capita per month in rural area and `539 per capita in urban areas resulted in people consuming “just about 1,820 kcalories (kcal)” of food as against the norm of 2100-2400 kcal (rural-urban).
While the NC Saxena report was commended for proposing an approach that was simpler compared to the confused methodology of the 2002 BPL census, it was criticized for continuing to impose pre-defined “caps” on the BPL coverage for a particular state or area. Implementing caps can result in exclusion errors among the poor who deserve to get the BPL card. Also, given that a particular score can be obtained through various combinations of indicators, the proposed method was unsuitable for participatory implementation or verification.
Planning Commission vs. Ministry of Rural Development
Another recurring pattern has been the wide differences between the MoRD and Planning Commission estimates – the estimates arrived at by the two approaches have always been very varied, sometimes by a factor of 2.
When the NC Saxena report was released by the Ministry of Rural Development, the Planning Commission refused to acknowledge the results of the study, contending that fixing the percentage of BPL is “beyond the scope of the present committee and is being handled by a separate committee” and that fixing the rural poverty level at 50% would have “tremendous financial implications and once granted cannot be reduced.” Though the NC Saxena report was released prior to the Tendulkar Committee report, the recommendations of the Saxena committee were unrepresented in the Tendulkar Committee recommendations.
In a Joint statement by the Deputy Chairman, Planning Commission and Ministry of Rural Development on October 3, 2011, it was confirmed that “the present state-wise poverty estimates using the Planning Commission methodology will NOT be used to impose any ceilings on the number of households to be included in different government programmes and schemes.” This is a departure from the census of the earlier years (and the NC Saxena report) where the number of poor identified by the state is capped by the Planning Commission estimates.
The World Bank approach
The World Bank uses the “money metric” approach, whereby it converts the “one dollar per day” international poverty line into local currencies using “purchasing power parity” conversion factors. It then uses national household surveys to identify the number of persons whose local income is lower than the national poverty lines. Both the dollar a day and $1.25 measures indicate that India has made steady progress against poverty since the 1980s, with the poverty rate declining at a little under one percentage point per year. This means that the number of very poor people who lived below a dollar a day in 2005 has come down from 296 million in 1981 to 267 million in 2005. However, the number of poor people living under $1.25 a day has increased from 421 million in 1981 to 456 million in 2005. This indicates that there are a large number of people living just above this line of deprivation (a dollar a day) and their numbers are not falling.
There have been many criticisms against the World Bank’s approach to measuring poverty.
Sanjay G. Reddy and Thomas W. Pogge, in their paper “How NOT to count the poor”, outline three reasons why theWorld Bank’s approach to meauring poverty is “neither meaningful nor reliable”:
“The first problem is that the Bank uses an arbitrary international poverty line that is not adequately anchored in any specification of the real requirements of human beings. The second problem is that it employs a concept of purchasing power “equivalence” that is neither well defined nor appropriate for poverty assessment. The third problem is that the Bank extrapolates incorrectly from limited data and thereby creates an appearance of precision that masks the high probable error of its estimates. It is difficult to judge the nature and extent of the errors in global poverty estimates that arise from these three flaws.” [G. Reddy and Thomas Pogge]
Firstly, the Bank’s method is unreliable because its results are excessively dependent on the chosen PPP base year. The Bank compares the consumption expenditure of a person in one country and year with that of another person from another country and year, by using national CPIs that deflate or inflate the two national currency amounts into “equivalent” amounts of a common base year, and then using PPPs for this base year to compare the resulting national-currency amounts. PPPs of different base years and the CPIs of different countries each weigh prices of underlying commodities differently, as they reflect distinct global and national consumption patterns. As a result, comparisons over space and time together are path dependent: if they are undertaken in different ways they may lead to different results.
Secondly, consumption patterns vary from country to country for reasons of tastes, as actual consumption patterns are strongly influenced by prices and by the existing income distribution.
Existing “broad-gauge” PPPs are inappropriate for use in poverty assessment as they lack appropriate focus in their informational base. The problem is compounded by the fact that the information used is aggregated in a manner that compounds the distortions inherent in the use of inappropriate information.
Thirdly, the Bank’s estimates of global poverty involve errors due to measurement problems associated with the data used under the Bank’s preferred approach.