Is poverty really rising since 2012? Comparing PLFS data with CES data is flawed poverty assessment

The field operations of the survey had to be postponed and carried out later telephonically.

By Manoj Panda & Rohan Bansal

Movements in magnitude of income poverty have been an important dimension for assessing the level of living of the people. Based on historical experience, growth in per capita aggregate income is considered as a sine qua non for poverty reduction. In fact, improvement in living conditions of the poor was one of major grounds for justifying reforms in India.

The data source for poverty analysis in India has been the Consumer Expenditure Survey (CES) conducted by the NSSO periodically. Analysts across the globe accept this database as a representative one, indicating changes in the level of living over time, especially for the low-income section of the population.

The latest year for which CES data are available is 2011-12, when the head count poverty ratio (HCR) stood at 25.7% and 13.7% for rural and urban areas, respectively, based on the Tendulkar Poverty Line. In the absence of CES data for recent years, some researchers have used consumption data from the Periodic Labour Force Survey (PLFS) in 2019-20 to calculate poverty and compared the poverty estimates from this survey with that of the CES in 2011-12. For instance, Mehrotra and Parida (The Hindu, August 4, 2021), have claimed that percentage of poor in India rose from 21.9 in 2011-12 to 25.9 in 2019-20. Such a conclusion has strong implications regarding India’s development process in the last decade and needs to be examined more seriously.

The primary objective of the CES is to obtain estimates of consumer expenditure by different sections of households, while that of the PLFS is to estimate the labour force participation. The questionnaires in the two surveys are designed keeping these objectives in mind.

The PLFS data are more, but not exactly, comparable with the earlier Employment Unemployment Survey (EUS). In 2011-12, when both CES and EUS were conducted, the average monthly per capita consumption expenditure (MPCE) levels reported by the EUS are 7.4% and 6.7% below those in the CES for the rural and urban areas, respectively. The lower MPCE in the EUS results in a higher proportion of poor than the CES for the year 2011-12.

The CES estimates MPCE through a very detailed questionnaire regarding consumption of different items with built-in consistency checks. In the PLFS, on the other hand, household consumption data is collected by asking only a single-shot question on ‘usual MPCE’ which is meant to be used only for the purpose of classification of households. Thus, the fact that consumption data in both the surveys are collected by the NSSO does not necessarily mean that they are comparable to indicate a change in level of living. This, we are afraid, involves comparison of oranges and kinnows, if not oranges and apples, grown by the same farmer.

Further, there is another serious problem in PLFS for 2019-20. PLFS normally covers the period July-June. Several low-income households did not have normal opportunities for income or normal consumption in the last quarter of the survey period July 2019 to June 2020 due to the Covid-19 induced lockdown effects. The field operations of the survey had to be postponed and carried out later telephonically.

The levels and growth rates in CES-based aggregate consumption data are considerably lower compared to the estimates provided by the Central Statistical Office in connection with its national income accounts involving GDP and other macro aggregates. During 1993-94 to 2011-12, while real per capita private final consumption expenditure (PFCE) grew by 4.7% per year according to the CSO data, real MPCE based on CES by the NSSO grew by just above 2%.

Even though the ‘usual MPCE’ from PLFS is not comparable to MPCE from CES, the three rounds of PLFS are comparable among themselves. Hence, if one wants to get a rough indication about recent trends on the poverty front, the ‘usual MPCE’ data in PLFS might be used keeping in mind its limitations.

The poverty estimates are given in accompanying graphic. Since mean MPCE in employment survey is lower than CES, the poverty ratios are higher. The poverty lines used here are obtained by inflating the Tendulkar poverty line for 2011-12 by consumer price index for rural and urban areas. A falling trend is evident over the three rounds confirming that GDP growth continues to have a poverty-reducing effect.

We would like to reiterate that comparison of incidence of poverty with 2011-12 CES must wait for data from another CES round. The PLFS consumption data is not a substitute for the detailed CES data. In the meantime, a comprehensive documentation and careful analysis of the available database on various other components of level of living could be done to understand how the poor people are doing.

But a narrative of rising poverty formed by simple comparison of CES and PLFS data is not warranted. It does not help the cause of higher economic growth in public perception.

Panda is former director, Institute of Economic Growth, Delhi, & Bansal is manager at the monetary policy department, Reserve Bank of India
Views are personal