A brief overview of the Indian poverty debate

June 9, 2008

By Alita Nandi, Sanhati. Open for comments.

Click here to read the technical version of this article [PDF, English, 120KB] »

The official poverty estimates published by the Planning Commission showed a decline in absolute poverty levels from 36% in 1993-94 to 26% in 1999-00. In the early 1990s various liberalisation policies had been introduced in India and India had started to experience higher growth rates (compared to pre-liberalisation period). The question that became important at this juncture was, “Did the advantages of this high economic growth reach all echelons of society, in particular the ‘poor’?” And so the official reports at this time showing a reduction in absolute poverty levels created a stir. Some old issues about poverty measurement and some new ones were brought into the foreground and heavily debated and discussed. Here I attempt to trace out the key issues of this debate.

The first issue that became a bone of contention was the methodology used for the construction of the poverty line itself. Since the 1960s the official poverty estimate is based on a head count measure of absolute poverty. The basic idea has been to count as poor all persons who cannot afford to consume a basket of goods (food, fuel, housing, clothing and health) that is necessary for surviving and working. The exact method to identify the poverty line (or poverty threshold) is based on recommendations of the “Report of the Task Force on Projections of Minimum Needs and Effective Consumption Demand”, 1979 (1979 Task Force). The poverty lines are estimated primarily using consumption expenditure data collected by the National Sample Survey Organisation (NSSO). The recent poverty debate flared up around the estimates based on the 55th round of the NSS, 1999-00.

The first poverty line that was estimated using this definition was done using data from the 28th round of the NSS, 1973-74. This data showed that in rural areas persons with a monthly per-capita expenditure (MPCE) of Rs. 49.09, on an average, consumed food that resulted in a daily calorie intake of 2400. Similarly, in urban areas persons with MPCE of Rs. 56.64, on an average, consumed food that resulted in daily calorie intake of 2100. Of course in addition to the food items all these persons also consumed non-food essential items such as fuel, transportation, health care, housing, etc. Using information from nutrition experts the 1979 Task Force had estimated that in order to survive and work, persons in rural and urban areas would need to consume at least 2400 and 2100 calories per day, respectively. The daily calorie intake levels of 2400 and 2100 are also called Required Daily Allowance (RDA). So, it can be said that all those with MPCE less than Rs. 49 in rural areas and Rs.56 in urban areas during the period 1973-74 could not consume the minimum basket of goods necessary for surviving and working and hence are poor. In other words, MPCE of Rs 49 and Rs 56 were set as the rural and urban poverty lines, respectively for 1973-74.

In years after 1973-74, the poverty lines were calculated by a slightly different method. In these later years the cost of purchasing the consumption baskets corresponding to the 1973-74 poverty lines (which ensured RDA) was calculated at current prices for rural and urban areas. These estimated costs became the poverty lines for that year. The problem with this updating is as follows.

In 1973-74, on an average, persons in rural areas with MPCE of Rs. 49 not only consumed food that ensured the RDA but also consumed other basic necessities such as fuel, housing, clothing. However, in later years the consumption pattern of persons may have changed and all those who could consume that particular consumption basket may not do so. That would imply that their daily calorie intake may be different from the RDA. The NSS data in later years showed that those persons whose MPCE was equal to the (price-updated) poverty line actually consumed food that resulted in a daily calorie intake that was lower than the RDA. Why might this happen? Two answers have been offered in the literature. First, this could be because people’s preferences had changed in favour of goods with lower calorie intake. A second answer is that as prices for non-food essentials increased but incomes did not increase (or increased but at a slower rate), people started cutting down on some of the food items. Patnaik (2006) and Mehta (2004) favour the second explanation.

For instance, here is what Mehta (2004) has to say on the matter:

Thus, in 1973-74 the expenditure group at poverty line in rural areas spent 80% of its total expenditure on food. In 1999-2000, the expenditure group at rural official poverty line spends only 65% of its total expenditure on food…These [consumption pattern in rural India for the expenditure class at the official poverty line for 1999-00] figures show how fallacious is the argument advanced by some that the calorie intake has reduced because even at poverty group people prefer to get their calories from more expensive food.

Patnaik (2006) makes a similar point:

The food share of total expenditure will fall also when people are getting worse off and their income and expenditure is stagnant or falling. This is because under such conditions of stagnant or falling income, the minimum unavoidable non-food spending on fuel for cooking and lighting, debt service, health-care, transport to work and so on, which are becoming more expensive, will force reduction in the food expenditure and raise hunger….This proposition receives strong support from the fact that at the official poverty lines the share of spending on fuels and on ‘miscellaneous goods and services’ has risen further between the 55th and 61st Rounds while there is decline in real spending on food.

So, the official poverty lines after 1973-74 do not necessarily ensure the RDA for persons with MPCE equal to the poverty lines. In 1999-00, 27% of the rural population and 24% of the urban population have per-capita expenditures below the (rural/urban) poverty line but as Mehta (2004) and Patnaik (2006) show around 75% of the rural population and 55% of the urban population have an average daily calorie intake which is less than the RDA. In 1993 an expert group was set up to address some criticisms levelled against the existing poverty line estimation method. This group presented its recommendations in the “Report of the Expert Group on Estimation of Proportion and Number of Poor”, 1993. This report shows that the 1993 Expert Group was aware of this issue but chose a fixed consumption basket as opposed to a fixed calorie norm definition of the poverty line to ensure comparability across the years. See Table 1 for rural poverty estimates provided by Patnaik (2006) using the fixed calorie norm definition that Patnaik (2006) and Mehta (2004) favour.

A second issue that generated a lot of debate was related to the way consumption questions were asked in the 55th Round of the NSS. As we have seen consumption data collected by the NSSO is used to estimate the poverty lines and so any problem with this data will have a direct impact on the poverty estimates. The poverty estimates for 1999-00 which had been computed using data from the 55th Round of the NSS was severely criticised because of the way the food consumption questions were asked in that year. Typically respondents are asked about their consumption expenditures on different items over a certain period of time, such as the last seven days, or last month, or last year. This time period is call the recall period. In the 50th Round (1993-94) food items and intoxicants were asked using a 30-day recall period and five low frequency items (clothing, footwear, durable goods, education and institutional medicine) were asked twice, using a 30-day and a 365-day recall period. In the 55th Round the food items and intoxicants were asked twice, using a 7-day and a 30-day recall period, and the five low frequency items were asked using only a 365-day recall period.

What are the implications on poverty estimates of asking consumption questions in this way? First, since the same recall period is not used for the same items in both rounds the poverty estimates in the two rounds may not be comparable. Using shorter recall periods for high frequency items increases consumption expenditure reports as it reduces memory loss. Again using longer recall periods for low-frequency items is also likely to increase reports of consumption of these items. This is because such items are generally purchased in long intervals and shorter recall periods would miss these purchases. Second, since questions about consumption of some items were asked twice using the two recall periods, it is possible that respondents may have reported consumption of these items in such a manner that there is consistency between the two reports. In that case which recall period elicited the truth? For example, if first asked to report the amount of milk purchased in the last month and then asked for a recall period of seven days did respondents simply divide the first answer by 4 to arrive at the second?

Deaton, Sundaram and Tendulkar and Sen and Himanshu try to address these problems and provide different poverty estimates (see Table 2). None of these provide any definitive answers as each of these estimates is based on a particular set of assumptions but give an idea of the range of possible poverty estimates. In Table 1 Patnaik (2006) also provides the poverty estimates after correcting for this data collection issue for 1990-00. After correcting for the recall period error, the decline in absolute poverty between 1993-94 and 1999-2000 ranges from 7.1 percentage points (Deaton, 2005) to about 2.8 percentage points (Sen and Himanshu, 2005).

Another criticism levelled against the official poverty estimates in general is that these estimates have not been derived from income poverty lines but from expenditure poverty lines. It is quite possible that many persons have expenditure levels above the poverty threshold but are only able to do so with heavy borrowing. If these borrowings are for consumption, and not investment purposes, then in the long term these persons are likely to fall below the poverty threshold.

One point that should be noted here is that based on recommendations of the 1993 Expert Group, rural and urban poverty lines are estimated separately for each state (to take into account the price differentials across states). I have abstracted from discussing how the corresponding national poverty lines are computed as that does not add anything to the understanding of the poverty debate.

So, what do the official poverty estimates tell us?

The official poverty line of 1973-74 tells us that persons with these MPCE levels on an average do consume the RDA. And as average daily calorie intake is lower for lower MPCE classes all those with MPCE less than the poverty line consume calories less than the RDA and hence are identified as “poor”.

The official poverty lines in the years after 1973-74, tell us that persons with these MPCE levels can consume, at the current price level, a consumption basket which ensures RDA and those who have MPCE levels less than these poverty lines cannot consume that consumption basket and hence are identified as “poor”. These lines do not tell us who actually does consume the RDA.

Given that the poverty lines based on the “fixed consumption basket” and the “fixed calorie norm” definitions provide different information about the level of deprivation in the economy and given that both can be computed from the same dataset, it would be very useful if two poverty estimates based on these two definitions were provided. However, the “fixed consumption basket” should include essential non-food items, in addition to, essential food items if it is to measure whether people can afford to consume goods that are absolutely essential for living.

This review of the Indian poverty debate is based on the following articles. Thanks to Dipankar for his useful comments.

1. Deaton, Angus and Valerie Kozel. “Data and Dogma: the great Indian poverty debate,” World Bank Research Observer, 2005, 20 (2), 177-199.

2. Mehta, Jaya. “Poverty in India”, 2004, http://gamma.nic.fi/~otammile/povindia.htm
Also printed in Alternative Economic Survey 2003-2004

3. Patnaik, Utsa. “Poverty and Neo-Liberalism in India” Based on Rao Bahadur Kale Memorial Lecture delivered at Gokhale Insitute of Politics and Economics, Pune, February 03, 2006. December 2006

4. Planning Commission, Government of India “Report of the Task Force on Projections of Minimum Needs and Effective Consumption Demand”, 1979

5. Planning Commission, Government of India “Report of the Expert Group on Estimation of Proportion and Number of Poor”, 1993

6. Sen, Abhijit and Himanshu, “Poverty and Inequality in India: Getting Closer to the Truth,” The Great Indian Poverty Debate, Edited by Angus Deaton and Valerie Kozel, Macmillan India. 2005.

Table 1

Table 1: Comparing poverty estimates (% poor below the poverty line) using the “fixed calorie” and the “fixed consumption basket” norms[1]
Official Poverty Estimates (based on the “fixed consumption basket” norm) Poverty Estimates based on “fixed calorie norm”[3] Official Poverty Estimates (based on “fixed consumption basket” norm), adjusted for 2400 calorie norm in 1973-74[3]
1973-74 56.4 72 72
1977-78 53.1 65.5 63
1983 45.7 70 54
1993-94 37.3 74.5 49.2
1999-00 27.4 74.5 (77.5)[2] 39
2004-05 29.5 N.A. 41.5
[1] Source: Patnaik (2006)
[2] Adjusting for recall period problems
[3] Patnaik (2006) shows that it is possible that the official 1973-74 rural poverty line of Rs. 49 is not consistent with daily calorie intake of 2400 (the RDA) but only with that of 2200. A poverty line of Rs. 56 is consistent with daily calorie intake of 2400. The estimates in column (3) and (4) use Rs. 56 as the poverty line in 1973-74.

Table 2

Table 2: Comparing Poverty Estimates (% poor below the poverty line) using Different Methods to Correct for the Recall Period Problem [1]
Official (rural) Official (urban) Deaton (rural) Deaton (urban) Sundaram and Tendulkar (rural) Sundaram and Tendulkar (urban) Sen and Himanshu (rural) Sen and Himanshu (urban)
1993-94 37.3 32.4 37.3 32.4 34 26 31.9 29.2
1999-00 27.0 23.6 30.2 24.7 29 23 29.1 26.1
[1] Source: Deaton and Kozel (2005)

1 Comment »

One Response to “A brief overview of the Indian poverty debate”

  1. Samanth Says:
    June 13th, 2008 at 15:15

    Dear Sir,

    One of the critical deficiencies of in the intellectual debates on poverty is the invisibility (atleast to people outside of academia and the government) of comparative figures of poverty reduction between states and within a state under different regimes.

    Would be helpful if you could also publish statistics of comparative poverty rates between West Bengal on the one hand and Maharashtra, Gujarat, Punjab and Tamil Nadu on the other. This way we can atleast form a partial judgement of how “idelogical formulations” have actually effected people’s lives.

    Thank you and looking forward to seeing the data.


Leave a comment