The release of the factsheet of the Household Consumption Expenditure Survey (HCES) has sparked a flurry of activity, with everyone trying to push a narrative about the trends and causalities. The debate has largely been on the rate of poverty reduction and, more importantly, whether that has been higher under the NDA or UPA. The differences in the poverty estimates are due to the reference period for data collection in the base year, the poverty line considered and the price index used.
Since the inception of HCES, the NSSO has employed a 30-day recall period, known as the Uniform Reference Period (URP), to collect consumption data for all items. In 1983, a 365-day recall period was introduced for durables (termed mixed reference period, MRP). In the ’90s, significant differences emerged between the aggregate consumption expenditure of NSS and that derived from the national accounts. Studies revealed that surveys were systematically underestimating consumption. This led to the adoption of also a seven-day reference period, in addition to 30 days and 365 days, known as the modified mixed reference period (MMRP). In 2011-12, alongside URP and MRP, NSSO adopted the MMRP. The latest HCES fact sheet suggests a shift to MMRP.
Official estimates of the poverty line and poverty ratio were derived using the URP data until 2004-05 as per the Lakdawala Committee’s guidelines, linked with certain calorie intake. There was, however, growing dissatisfaction with this approach due to the absence of any definite evidence linking calorie input with health outcomes. Consequently, the Tendulkar Committee moved away from anchoring the poverty line to a calorie norm.
A comparison of the average consumption expenditure using the three reference periods establishes that MMRP results in improved reporting due to better recall. The Tendulkar Committee, therefore, recommended that as and when the MMRP is adopted, the MRP equivalent per capita expenditure (for poverty line) should be shifted to MMRP. In a subsequent report, the Planning Commission estimated poverty rates using the 2011-12 HCES data based on MRP. However, no official poverty estimation has been built based on MMRP in 2011-12.
In 2012, the Planning Commission appointed another expert group under the chairmanship of C Rangarajan, which recommended that the poverty line should be based on certain normative levels of nourishment, clothing, house rent, conveyance and education, and a behaviourally determined level of other non-food expenses. The poverty lines estimated by this committee exceed those of the Tendulkar Committee, partially due to the latter’s utilisation of MMRP-based consumption data.
Researchers have mostly used the consumer price index to update their base year poverty line. For instance, SBI research, which utilised this to update the poverty line, found a substantial decrease in the poverty rate — from 25.7 per cent and 13.7 per cent in 2011-12 to 7.2 per cent and 4.6 per cent in 2022-23, respectively, for rural and urban areas. NITI Aayog supported this poverty figure being below 5 per cent in India for 2022-23. However, this methodology is questionable as it updates an MRP-based poverty line and applies it to consumption data based on MMRP, resulting in an underestimation of poverty in 2022-23. C Rangarajan and Mahendra Dev also used the CPI to update their MMRP-based poverty line for 2011-12. They report a reduction in the headcount ratio from 30.9 per cent and 26.4 per cent in rural and urban areas in 2011-12 to 12.3 per cent and 8 per cent respectively in 2022-23. There are no comparability issues here since the poverty lines and the consumption data are based on MMRP.
Using the international poverty line of $1.90 at 2011 prices, Surjit Bhalla and Karan Bhasin claimed that the percentage of poor in rural and urban areas is 2.5 and 1.0 respectively. There is no problem of temporal comparability here since both these estimates are obtained through application of a global poverty line adjusted for inflation, to the MMRP-based expenditure data. However, getting comparable poverty estimates for the years for which MMRP-based data are not available would be a challenge.
Another point on data comparability — the 2022-23 survey employs three distinct questionnaires covering food items, consumables and services items, and durable goods, as opposed to a single questionnaire used before. Additionally, there is a transition from single-visit to multiple-visits for data collection which would affect the quality of responses. The sharp increase in non-food items of consumption in 2022-23 could partly be due to better reporting. While complimenting the NSSO for bringing this much-needed change, one must not ignore the loss of temporal comparability. Interestingly, scholars claiming much faster poverty reduction under NDA than UPA have dismissed issues of non-comparability of the latest data with earlier rounds.
The fall in the percentage of poor during 2004-2011 from 37.2 per cent to 21.9 per cent, an annual decline of 2.2 percentage points, works out to be higher than the annual decline of 1.6 percentage points during 2011-22, from 21.9 per cent to about 6 per cent (average figure emerging from the projections). It would nonetheless be rash to pronounce greater poverty reduction during the UPA period than the NDA since the time points of the available data do not coincide with their periods of governance.
Also, as we near the goal of eliminating poverty, the task of reducing poverty by a certain percentage point becomes more difficult. Finally, the impact of anti-poverty measures may not manifest right from the year of their launch. Given this, it is difficult for Congress or BJP to claim a relative advantage in poverty reduction.
Kundu is Professor Emeritus at L J University, Ahmedabad and Rahaman is an independent researcher
