A major constraint for policymakers devising strategies to protect Indian citizens and businesses from the vicissitudes of the pandemic has been the paucity of quality, high-frequency data. It is unusually difficult to construct policies when there is no clear sense of how this unprecedented crisis has affected household and business activity — from employment
availability to migration to consumption choices. Some datasets from the private sector have been used to proxy for broader indications about prosperity or economic recovery. Economists
and investors have also turned to the Centre for Monitoring Indian Economy’s (CMIE’s) Consumer Pyramids Household Survey, or CPHS. The CPHS’ strong conclusions about the pandemic’s effects, including that 97 per cent of households have lost income because of the pandemic and 10 million were rendered unemployed during the second wave, have been crucial indicators of the depth of the pandemic’s effects and the need for relief and revival efforts.
However, the CPHS itself has been the site of contestation by both left-leaning academics and the government. Jean Dreze and others have criticised the CPHS methodology as being biased towards more elite samples, indicating, for example, that households are surveyed starting from the main streets of urban areas, which might systematically bias the results against the poorer households in back streets. The CMIE has responded to this concern by pointing out that the scale of the survey ensures that households in back streets are represented. Either way, the independent economists’ criticism is well-meaning, and can be used to improve the CPHS. Government economists
have attacked the basis of the CPHS itself, indicating that they do not think it provides an accurate picture of the economy.
As evidence, they cite the puzzling conundrum that employment
in the CPHS appears to be negatively correlated in recent times with output growth. But, as the CMIE points out, they have used nominal and not real output to make this point, rendering it ineffective. They have also argued that, unlike the government-run Periodic Labour Force Survey, or PLFS, the CPHS uses a simple and more stringent definition of employment: Has the respondent been employed that day or the day before? The PLFS asks a question about whether the respondent has had an hour of employment
in the previous seven days. Economists
may disagree about which of these is the better survey question. But this difference does not take away from the broader trends identified by the CPHS. There is no reason, therefore, for it to be ignored by observers of the Indian economy
— as has been urged by government economists.
If the government is concerned about the use of the CPHS data, it should spend less time criticising it and more time working to improve its own efforts to collect and release economic data, particularly large-scale high-frequency data. The PLFS, for example, is produced with such a long lag and so infrequently that it is simply unusable for any practical purpose. India needs to bring its statistical apparatus into the 21st century. The United States’ Bureau of Labour Statistics estimates unemployment using the Current Population Survey, which uses permanent Census employees to contact 60,000 households across that country every month. Appropriately scaled up, this is the mechanism that should be used in India.