If India’s economic reforms since the 1990s have led to "considerabledisruption of earlier livelihood patterns" according to economists ProfessorAngus Deaton and Jean Dreze, the official data on the number of those who livebelow the poverty line, for instance, doesn’t reflect that reality. Instead,these numbers reinforce a sense of optimism in government that a dramaticreduction has occurred on the poverty front, especially in rural India, due toreforms. However, using exactly the same data, quite a few economists in Indiafrom a Leftist standpoint totally debunk this progress on poverty reduction.
During the 1990s, the Planning Commission’s estimates indicate the share ofIndia’s rural population living in poverty declined by 50 million to 27.1 percent. This dramatic progress on the war on poverty is expected to gathermomentum with this share declining to 21 per cent by the end of the Tenth Planin 2006-07. This sort of discourse -- which attributes all of this to thefavourable impact of reforms on growth and targeted interventions, among otherthings -- is totally rejected by Leftist economists like Professor Utsa Patnaikwhose estimates are as high as 74.5 per cent in rural India
How can the same data support radically different estimations of ruralpoverty? Before we get into this statistical snake pit, it is necessary toindicate how the government goes about crunching its optimistic poverty numbers.The basic data for this purpose are the large-scale household surveys ofconsumption expenditure conducted by the National Sample Survey Organisation (NSSO)conducted every five years (1993-94, 1999-2000 with the next one due in 2005-06)and the National Accounts Statistics (NAS). NSSO and NAS consumptionestimates are different with the latter being on the higher side.
The first step is to define the so-called poverty line. The official line isthat those who cannot afford a standard of Rs 49 per capita at 1973-74 prices or2,400 calories per day constitute the rural poor. This line, of course, isupdated for inflation using an all-India price index or state andsector-specific price indices since the mid-1990s. Since the poverty lineremains fixed and updated for inflation, this standard goes up to Rs 328 in1999-2000 and the share of rural population who cannot afford such expendituresjoin the ranks of the reserve army of the poor according to the PlanningCommission.
Official estimates have always underestimated rural poverty for variousreasons. Till the early 1990s, the Planning Commission believed that the NASnumbers on aggregate consumption were not only higher but also showed a lesserratic pattern than those of the NSSO. Nevertheless, it also reckoned that the NSSOdata on size distribution of household consumption were good and applied this toNAS data. To make them comparable, this entailed scaling up the NSS data on sizedistribution of consumption expenditures -- a method which was strongly objectedto by well-known economists like BS Minhas.
This scaling up artificially pulled up a large number of people over thepoverty line and was responsible for the dramatic reduction in poverty tillthen. But faced with such criticism from Minhas, the Planning Commissionabandoned scaling up and implemented the recommendations of the Expert Group onEstimation of Proportion and Number of Poor in 1993. But there are still a few,like Surjit Bhalla, whoprefer the NAS for being more reliable than the NSSO andby scaling up the latter show an even faster decline in poverty than thePlanning Commission did in 1999-2000!
To be sure, there is a full-fledged industry that is still thriving churningout poverty estimates in the country besides Bhalla. Various other economistslike A Vaidyanathan, Suresh Tendulkar, to name a few, have put out their ownestimates using different poverty lines and price indices than the PlanningCommission and have exercised their preference for NSSO data despite all itslimitations. But what is common to these studies are estimates of rural povertythat are a tad higher than the official one. They also show a less sharperdowntrend in rural poverty than the official one! Deaton thus confirms 70 percent of the official reduction in India’s rural poverty in the 1990s.
Official estimates show a much faster decline in rural poverty thanks to ‘contaminated’data in 1999-2000. NSSO’s estimates on overall consumption are, no doubt,lower but the reference period for recording consumption was inexplicablychanged. Earlier five-yearly surveys like in 1993-94 were based on a 30-dayrecall period but in 1999-2000 a questionnaire incorporating a 7-day recallperiod was simultaneously canvassed -- contaminating the entire data in theprocess. As the consumption estimates using the 30-day period were upwardlybiased due to the simultaneous presence of the 7-day reference period, theresult was a much sharper reduction in poverty
Naturally, all of this triggered a flurry of data adjustments in the povertyindustry to enable comparisons between 1993-94 and 1999-2000. Utsa Patnaik, forher part, uses this same contaminated data as she adopts a direct approach toderive her higher poverty estimate in 1999-2000. As the NSSO data also presentsthe calorific equivalent of average diets in various expenditure groups, shedirectly observed that seven-tenths of the rural population consumed less than2,400 calories per day -- the original subsistence standard defining theso-called poverty line set way back in 1973-74!
To be sure, a lot of economists might quibble against Utsa Patnaik’smethods but the issues she raises won't go away. Foremost among them is therationale for assuming a constant poverty line and updating it using variousprice indices for all time to come which only show less poverty. The upshot isthat behind this seemingly neutral exercise of crunching out poverty numbers,the government would like to believe that reforms have done their job on thepoverty front, while critics like those from the Left only point to the seriousdisruption of livelihood patterns due to economic reform.
N.Chandra Mohan is an economic commentator
.jpg?auto=format%2Ccompress&fit=max&format=webp&w=376&dpr=2.0)
