source:
August 10, 2013
by Deepankar Basu
The Planning Commission’s latest estimates of the incidence of poverty in India – measured by the head count ratio – have been met with incredulity in most quarters where reason still holds sway and evidence still matters. Using a poverty line of Rs. 816 per capita per month for rural areas and Rs. 1000 for urban areas, the Planning Commission has estimated the that 25.7 and 13.7 percent of the population are poor in rural and urban India, respectively, in 2011-12. The corresponding proportion of the poor was 41.8 percent in rural India and 25.7 percent in urban India in 2004-05. Therefore the new estimates show stunning 16.1 and 12 percentage points decline in poverty in rural and urban India over the past 7 years. No wonder, pro-reforms economists and commentators are ecstatic, falling over each other in highlighting the wonders of economic growth.
The Destitution Line
The outrageously low poverty lines, which deliver these flattering poverty levels, have been rightly called “kutta-billi” (dog-cat) poverty lines by N. C. Saxena, member of the National Advisory Council. Any familiarity with life in contemporary Indiais enough to drive home the point that such low levels of expenditure can only support a sub-human condition of existence. While hidebound conservatives, using arguments bordering on the ridiculous, are willing to defend the measly poverty lines used by the Planning Commission, even simple back-of-the-envelope calculations are enough to show them for what they are: destitution lines.(Economic and Political Weekly, 2008).
Let us take one of the cheaper metros, Kolkata, and do some rough and ready computations to arrive at an approximate amount of expenditure that would be required for a decent, but frugal, life today. For a family of 4, the following quantities of food would be needed for a decent life over the period of a week: 10 kgs of rice, 3 kgs of dal, 1 litre of cooking oil, 5 kgs of potato, 3.5 kgs of green vegetables, 1 litre of milk, 4 eggs, ½ kgs of chicken, ½ kgs of fish, and 1 litre of kerosene (fuel). Notice that the quantities of milk, eggs, chicken, fish and kerosene (fuel) have been deliberately assumed to be far lower thanwhat a decent standard of living would demand. Notice also that I have not added any expenditure on fruits and nuts.
How much would the above food basket – with deliberately chosen low quantities for several items – cost? A simple survey of prices prevailing in the market today suggests that the above quantities of food items can be purchased for Rs. 840 a week or Rs 120 a day. Thus, food expenditure would need Rs. 3600 per month.
Let us add to this the following monthly nonfood expenditures: Rs. 400 for education, Rs. 500 for healthcare, Rs. 500 for conveyance & clothing, Rs. 1000 for rent, and Rs. 500 for miscellaneous expenses (including footwear, mobile phones, consumer services, entertainment, etc.).The total expenditure comes out at Rs. 6500.
Thus, even with unrealistically low levels of expenditures devoted to many food items (like milk, eggs, fish, meat and fruits), and to education, healthcare and rent, the minimum expenditure required in Kolkata for a family of 4 is Rs. 2500 more than what the Planning Commission uses as a “poverty line” (recall that the Planning Commission’s poverty line is Rs. 1000 per capita per month in urban India).
Given that the Planning Commission adjusts poverty lines across regions to account for price variations across states and regions, the poverty line in Kolkata would have likely been lower than the all-India average. On the other hand, the poverty line of Rs. 1000 is for the year 2011-12, so that inflation since then would push up the poverty line. Hence, these two opposing forces might lead to a deficit over the poverty line which is only approximately Rs. 2500. But it seems difficult to deny that, no matter how we adjust the poverty line, there would in fact be a substantial deficit. (Note that a similar back-of-the-envelope calculation by S. Subramanian for Madras in 2004-05 came out with an equally large deficit; see Subramanian (2012) for details).
The absurdity of the Planning Commission’s poverty lines is so pronounced that even the Central Government has distanced itself from them. In the context of its welfare schemes, especially the recently approved National Food Security Ordinance, the government has chosen to effectively use a poverty line that is 85 percent higher than the Planning Commission’s absurdity. The more realistic poverty line used by the government would lead to a head count ratio of 65 percent for the country in 2011-12, about 3 times higher than the Planning Commission’s estimate of 22 percent.
Poverty Line and Trend
A close reading of the Press Note of the Planning Commission for poverty estimates in 2011-12 suggests that even the Delhi mandarins were uneasy about using such a low poverty line. But the Planning Commission’s defense is that the value of the poverty line does not have any impact on the estimation of the trend of poverty. According to the Planning Commission, even if a more realistic poverty line (like the one sketched above) is chosen, it will not change the picture of rapid poverty decline over the last 7 years. This, as we will see shortly, is a widely accepted but erroneous proposition. But first let us carefully read what the Planning Commission has to say on this.
After highlighting the remarkable speed of poverty decline during the 7 year period from 2004-05 to 2011-12, the Planning Commission’s note hastens to assure us that even though a more realistic, and hence higher, poverty line would translate into a higher level of poverty, it would not affect the trend of rapid poverty decline.
“It is important to note that although the trend decline documented above is based on the Tendulkar poverty line which is being reviewed and may be revised by the Rangarajan Committee, an increase in the poverty line will not alter the fact of a decline. While the absolute levels of poverty would be higher, the rate of decline would be similar.” (page 3, Press Note on Poverty Estimates, 2011-12, Government of India, Planning Commission, July 2013)
The proposition that the trend of poverty is independent of the poverty line is, in fact, incorrect (we will see the reason for this below). The choice of poverty lines impacts both the level and trend of poverty. This has been highlighted by the work of several economists, including S. Subramanian (2012) in India. Despite this well-known result in development economics, the fallacious idea has surprisingly wide currency. It has been referred to many times in the past, explicitly or implicitly, by pro-reforms commentators like Jagdish Bhagwati, Surjit Bhalla, Arvind Panagariya, Swaminathan A. Aiyar, and others. Even economists in good standing seem to have fallen for this fallacy: in a recent piece in the Indian Express(“What the poverty numbers don’t say”), the point has been repeated by economist Bhaskar Dutta.
“It is, of course, a tautology that a higher poverty line will imply a greater level of poverty. However, this is a criticism about the estimated level of poverty in 2011-12, and is completely silent about the trend in the incidence of poverty… Unless the change in the distribution of consumption expenditure has been extremely perverse, the dramatic reduction in poverty according to the Planning Commission estimate also guarantees that there would be a sizeable reduction even if the poverty line were set [at] a higher level.”
A Brief Detour into Poverty Line History
To see the fallacy underlying the claim of poverty line independence ofpoverty trends, it would be useful to recall how the Planning Commission computes the poverty line. Following the recommendations of the 2009 Expert Group to Review the Methodology for Estimation of Poverty a.k.a. Tendulkar Committee, the Planning Commission now computes the poverty line by adjusting the 2004-05 urban poverty line of Rs. 578.8 per capita per month (at 2004-05 prices) with price indexes – constructed from quantity and value data collected in the household consumer expenditure surveys conducted by the NSSO– totake account of temporal and spatial variation in prices.
The 2004-05 urban poverty line of Rs. 578.8, in turn, is the price-updated version of the 1973-74 all-India urban poverty line, which is the expenditure level (Rs. 56.56 at 1973-74 prices) that ensured fulfillment of the calorie norm of 2100 Kcal per capita per day in urban India in 1973-74.
At the time when the Tendulkar Committee was formed, the Planning Commission had been using the method recommended by the 1993 Expert Group, one of the many Expert Groups and committees that have been formed over the years for studying these issues in India. The 1993 Expert Group method rested on using the 1973-74 commodity bundle as a “reference bundle” and updating the expenditure required to purchase such a bundle with relevant consumer price indexes. Thus, the Tendulkar Committee applied the 1993 Expert Group methodology to the 1973-74 urban poverty line to arrive at the figure of Rs. 578.8 per capita per month.
“The latest official estimates of poverty following broadly the Expert Group (1993) method and using the uniform reference period (URP) of 30 days indicate that below poverty line (BPL for short) population was 28.3 per cent of the rural population (described as headcount ratio or poverty ratio) and 25.7 per cent of the urban population in 2004-05. These official estimates released by the Planning Commission are based on (a) the 1973-74 rural and urban poverty line baskets originally at 1973-74 prices adjusted for price changes between 1973-74 and 2004-05 (b) a uniform reference period (URP for short) of 30-days for canvassing consumption of all items of current household consumption in NSS and (c) rural and urban size distributions of per capita total consumer expenditure (PCTE for short) data collected during the 61st (quinquennial large sample) round (July 2004 to June 2005) on household consumer expenditure of the National Sample Surveys (NSS).” (Government of India, 2009, page 5).
Having arrived at the urban poverty line, i.e., an expenditure of Rs. 578.8, in 2004-05, the Tendulkar Committee then uses the commodity bundle – including both food and nonfood items – corresponding to an expenditure level of Rs. 578.8 in urban India in 2004-05 as the “reference bundle” to compute poverty lines in rural areas and in different states.
“In the interest of continuity as well as in view of the consistency with broad external validity checks with respect to nutritional, educational and health outcomes, it was decided to recommend MRP‐equivalent of urban PLB [i.e., poverty line basket] corresponding to 25.7 per cent urban headcount ratio as the new reference PLB to be provided to rural as well as urban population in all the states after adjusting it for within‐state urban‐relative‐to‐rural and rural and urban state‐relative‐to‐all‐India price differentials.” (Government of India, 2009, pages 1-2).
Thus, for all its criticisms of the 1993 Expert Group methodology for using an outdated commodity bundle, the Tendulkar Committee also ends up using a fixed commodity bundle. The only difference arises from the specific commodity bundle chosen as the reference bundle and the price indexes used for the process of “updating”.
First,the 1993 Expert Group methodology would recommendusing the 1973-74 poverty line commodity bundle as the “reference bundle”, the Tendulkar Committee instead uses the commodity bundle corresponding to an expenditure level of Rs. 578.8 in urban India in 2004-05 as the “reference bundle”.Second, the 1993 Expert Group used relevant consumer price indexes (CPI), the Tendulkar Committee uses prices indexes constructed from consumption expenditure data collected by the NSSO during its consumer expenditure surveys.
Thus, it is important to note that the Planning Commission’s current methodology, following the recommendation of the Tendulkar Committee, is notvery different from its earlier methodology, which was based on the recommendations of the 1993 Expert Group, in that both use fixed commodity bundles to compute poverty lines. [1]
Problems of Fixing and Variable Poverty Trends
The method of using a fixed commodity bundle, either the 1973-74 poverty line commodity bundle (PLCB)used by the 1993 Expert Group or the 2004-05 urban poverty line basket (PLB) used by the Tendulkar Committee, to compute poverty lines over time is fraught with serious conceptual problems. Changes in tastes and preferences, movements in relative prices and income, changes in public provisioning of education and health care and decline in access to nonmarket sources of food and fuel, will lead consumers to opt for very different consumption bundles at different points in time. Hence, it makes little sense to compute the poverty line as the expenditure required, in currently prevailing prices, to purchase a fixed commodity bundle, the bundle that was chosen in some previous year under very different price, taste and income configurations.
Let us step back a little and recall that the 1973-74 commodity bundle was chosen from actually observed data on consumption expenditure.
“In deriving the poverty lines, it was recognised that human existence required more than just food, and provision for other goods and services also needed to be made. Since there are no a priori norms for these, and in order to avoid arbitrariness, it was felt that the actual expenditure of households shouldform the basis for estimating the necessary expenditure on these goods and services. In order to do so, the NSS household consumption expenditure data for 1972-73 was used.” (Sen, 2005)
This has an important implication with regard to consistency of the method used to construct poverty lines: just as the poverty line for 1973-74 was based on a commodity bundle actually chosen by households in that year, it should be based for other years also on choices that consumers actually make in those years. This simple and intuitive idea can give us a powerful critique of the 1993 Expert Group methodology and also show us how the choice of poverty lines can impact the level and trend of the incidence of poverty.
Following excellent recent workby S. Subramanian (2012), let us choose 1983, 1993-94 and 1999-00 as three comparison years and compute three different series of head count ratios (HCR) for rural India. The first series – with the HCR at 65, 63 and 60 percent for the three comparison years respectively – uses 1983 as the “base year” and “updates” the poverty lines for 1993-94 and 1999-00 with the consumer price index for agricultural labourers (CPIAL); in Table 1, this series appears in the column “Method 1”. The second series uses 1993-94 as the “base year” (coupled with relevant CPIAL updating) and gives 75, 74 and 68 percent as the HCR for the three comparison years; in Table 1, this series is given in the column “Method 2”. The third method uses 1999-00 as the “base year” and results in 80, 80, and 74 percent as the HCR series; in Table 1 this series is collected under the column “Method 3”.
Table 1: Poverty Estimates (Head Count Ratio) for Rural India Using Alternative Methods
|
Method 1 |
Method 2 |
Method 3 |
Consistent Method |
1983 |
65% |
75% |
80% |
65% |
1993-94 |
64% |
74% |
80% |
74% |
1999-00 |
60% |
68% |
74% |
74% |
Source: Subramanian (2012)
Clearly then the choice of the poverty line, governed here by the choice of the base year, has an impact on both the level and trend of the incidence of poverty. The fact that the level of poverty declines in all three computations might be read as evidence in support of the idea that the choice of the poverty line is irrelevant to understanding trends in poverty. That is not true.
But before we see why, let us briefly look at the main problems of the Tendulkar Committee poverty line.
Tendulkar Committee Falls Short
The Tendulkar Committee has rightly criticized the 1993 Expert Group methodology for using an outdated bundle for computing the poverty line in years after 1973-74. But its antipathy to using nutritional norms stops the Tendulkar Committee from going all the way and adopting the consistent method of computing the poverty line.
What do I mean by a consistent method in this context? The consistent method of computing poverty lines, as I have already indicated, is to use each comparison year as its own base year. For each year, NSS data provides a relationship of per capita expenditure and calorie intake. The consistent method would use this empirically observed relationship to choose the poverty line as the expenditure level that allows a household to meet the calorie norm in that particular year. Instead of mechanically using a fixed commodity bundle, this method allows consumers in every year to choose optimal commodity bundles given prices, incomes and preferences prevailing in that year, and also to meet a scientifically determined nutritional norm (like 2100 Kcal per capita per day for urban India).
The Tendulkar Committee does not adopt this consistent method, which, in fact been recommended by the 1984 Report of the Study Group on the Concepts and Estimation of Poverty Line, but instead replicates the 1993 Expert Group methodology with a different reference bundle. Thus, the Tendulkar Committee poverty line falls short of the very potential that was opened up by its own critique of the 1993 Expert Group methodology.
Can We Please Be Consistent?
Returning to the issue of the putative independence of poverty trends from poverty lines, we can show that the consistent methodology gives a very different trend in the incidence of poverty in India. Using the consistent method for computing the poverty line, the HCR comes out as 65, 74, and 74 percent for 1983, 1993-94 and 1999-00 respectively (see the series that appears under the column “Consistent Method” in Table 1). Thus, while the first three methods used above – with 1983, 1993-94 and 1999-00 as base years, respectively, and price-updated poverty lines for other years – throw up a declining trend in poverty, the consistent method – with each year treated as its own reference year – gives us an increasing trend in the incidence of poverty.
The same conclusion holds when we extend the analysis to 2011-12 (the latest year for which we have data from a large scale consumer expenditure survey conducted by the NSSO). Even as the Planning Commission’s fixed commodity bundle method shows that the HCR has declined from 50 to 26 percent in rural India between 1993-94 and 2011-12, consistently computed poverty lines throw up an increasing trend in the incidence of poverty over the same period, rising from 71 percent in 1993-94 to even above 90 percent in 2011-12!
Thus, if the Planning Commission and other conservative economists were to compute poverty lines consistently for every year using the method that was used to compute the poverty line in 1973-74, and not just use some price index to inflate a fixed commodity bundle, then the declining trend in poverty would give place to an increasing one, and all the triumphalism and euphoria surrounding poverty trends in India would melt into thin air.
(I would like to thanks Debarshi Das and S. Subramanian for helpful comments and suggestions).
******************
References
Economic and Political Weekly, 2008. “How Many Poor in the World?”, Editorial, Economic and Political Weekly, 25-31 October.
Government of India. 2009. Report of the Expert Group to Review the Methodology for Estimation of Poverty.Government of India, Planning Commission. Available for download here:
Sen, P. 2005. “Of Calories and Things: Reflections on Nutritional Norms, Poverty Lines and Consumption Behavior in India,” Economic and Political Weekly, October 22, pp. 4611-4618.
Subramanian, S. 2012. The Poverty Line. Oxford India Short Introductions. Oxford University Press: New Delhi.
Endnotes
[1] There is considerable confusion among policy makers, economists and media persons about the details of the Tendulkar Committee poverty line. Part of the reason for this, no doubt, is that the report has not been able to explain its methods very clearly, but part of the reason is that different commentators have read their own beliefs into the Report.
For instance, many commentators, including a member of the Planning Commission, believe that the Tendulkar Committee poverty line has been expressly derived to be equivalent to the international poverty line used by the World Bank, and this international equivalence makes it a correct poverty line. This is incorrect. The Tendulkar Committee itself notes that the international equivalence is only happenstance. “…the new poverty line happens to be close to, but less than, the 2005 PPP $1.25 per day poverty norm used by the World Bank in its latest world poverty estimates.” The important phrase here is “happens to be” (Government of India, 2009, page 8).
Again, another strident pro-reforms commentator believes that “…the Tendulkar committee raised the original 1973-74 rural poverty line by 25.9 per cent and the urban line by 6.4 per cent.” The last part of this sentence is obviously false: the Tendulkar Committee poverty line for urban India in 2004-05, as we have seen, is exactly the 1973-74 urban poverty line “updated” for price increase. In real terms the two are the same.
– See more at: http://sanhati.com/excerpted/7805/#sthash.2Vrckxo2.dpuf