NSS, CMIE surveys are not comparable. Studies should not correlate them

Incomplete data, or the lack of data itself, is driving economists to come out with their macroeconomic models to estimate sensational, eye-popping outcomes, whether they be Covid-19-related deaths or the impact of the pandemic on vulnerable populations. Some economists have argued that the pandemic has hit informal and small-scale workers directly, pushing them into poverty. Poverty estimation through the consumption approach has its challenges. Some researchers have attempted to arrive at poverty estimates based on auxiliary data available from the government or from other small sample studies.

Recently, policy advisers and researchers at the IMF and World Bank have also attempted to estimate staff numbers using various assumptions. The huge differences in staffing numbers between the two studies only add to the confusion in the already complicated problems of measuring poverty in India.
The IMF performed the calculation using adjustments for private consumption expenditure from national accounts statistics and also using expenditure incurred by the government under the public distribution system. The World Bank has attempted to estimate employee numbers using CMIE data from the 2015-2019 Consumer Pyramids Household Survey and comparing them with the 2011 NSS consumer spending data and data from other sources such as the National Family Health Survey and the Periodic Labor Force Survey in to relate, the Farm Household Situation Assessment and the All India Debt and Investment Survey. This exercise does not seem to follow the basic principles of statistics.

In fact, many econometric tweaks were likely made to relate the disparate datasets of NSS to CMIE. It should be noted that India uses the NSS Consumption Expenditure Survey to measure poverty and the results from it, conducted in 2017, are not available due to quality issues in the data collected.

The World Bank paper appears to be based on unrealistic and unsustainable assumptions. First, the sample design of the NSS and CMIE surveys is different. The NSS uses a multi-level stratified sample while CMIE uses a rotational sample. In addition, CPHS households have unequal sampling probabilities, since households on the high streets have a higher probability of selection. The basic definition of the household also differs between the two surveys. Unlike the NSS, the CPHS does not perform a listing, instead using projections of household and population growth to construct sample weights. It has been noted by various researchers that the NSS adequately collects information from households at the lower end of the consumption distribution but insufficiently from those at the higher end. However, many doubts have been raised about the representativeness of the CMIE data.

The best of Express Premium
bonus
The history of sun worship in India - a declining cultbonus
This way to Rashtrapati Bhavan: BJP has the numbers but why is it...bonus
Totally illegal, says ex-CJ of Allahabad High Court;  Bulldozer cases in l...bonus

Second, NSS collects information on more than 345 individual items to arrive at estimates of consumption expenditure, while CMIE does so on 114 items. While NSS spending is based on a 30-day recall period for groceries and others over 365 days, CPHS consumer spending is based on a last four-month recall period. Also, attempting to compare using subgroups of spending such as groceries, non-food, and durable goods may not be very helpful, as errors in data collection from the two sources do not necessarily cancel out, but can add up due to different sets of items.

Third, there is a time difference between the data used from the NSS survey, which covers the year 2011, and the CMIE survey from 2015 to 2019. This lack of comparable years for developing the model introduced another error.

Fourth, the change in weights in the CMIE household-level survey using NFHS and other surveys may not accurately reflect the weighting pattern because long-term consumption spending changes may produce required demographic and other changes.

In fact, measuring poverty at the national level serves no political purpose. You have to go down to the state, district, apartment block and village levels to identify pockets of poverty and design and implement specific programs that are needed in each case. India already has a measurement of multidimensional poverty that allows for a better understanding of deprivation. Another initiative is the Emerging Districts program, which expands to the block level and provides direction and location where specific interventions are needed.

Given the structural limitations of the survey, if the World Bank wants to estimate poverty in India at all, it should only use the CMIE data available from 2015 and measure changes in poverty rates. Alternatively, one can wait for the results of the survey, which will be carried out by the National Statistics Office between July 2022 and June 2023. The results are expected to be available approximately one year after the survey is completed.

Kumar is Senior Fellow NITI Aayog, Verma is former DG MOSPI and Srivastava is former Secretary of MOSPI

Leave a Reply

Your email address will not be published.