Twice this year, the chief statistician of India Pravin Srivastava had announced that the report as well as the data on consumer expenditure for 2017-18 will be released on time. The decision to junk the survey obviously was a hasty decision and a panic response after the media leak of the report. But why did the government junk the report?
The official explanation as given in the press release from Mospi is that “there was a significant increase in the divergence in not only the levels in the consumption pattern but also the direction...when compared to other administrative data sources. Concerns were also raised about the ability/sensitivity of the survey instrument to capture consumption of social services by households, especially on health and education". It was also mentioned that the government had set up an expert committee to examine some of these issues and the committee did give suggestions for future surveys.
But that specific committee did not find anything wrong with this survey round. Neither has the government released the report of the expert committee, nor has it provided any other evidence about the nature and extent of divergence with other administrative data.
How valid is the explanation offered by Mospi for junking the survey and stating that they will release the revised report by 2023? While it is difficult to come to a clear conclusion in the absence of any information/report, both these issues are not new and have been examined in detail a number of times in the past by several expert groups including a panel of the National Statistical Commission, which examined it as recently as 2008.
The concern regarding the divergence between the aggregate consumption expenditure reported by similar surveys (which are done every five years) and those obtained as a residual in the National Accounts (Private Final Consumption Expenditure goes into the gross domestic product or GDP calculations) are well known. However, the issue of comparability between CES and other administrative data has never been publically expressed as a problem.
The reasons are pretty clear. Administrative data is generally production data generated by government agencies. Total production may be consumed domestically, exported or even kept as inventory. There is a problem of comparability since the consumption surveys are only reporting consumption by households, whereas, the administrative data has no means of segregating the production data by category of consumers.
There is no reason to assume that household consumption surveys will show the same growth of consumption, for example, of coal, steel, food or automobiles as some of the consumption may be intermediate consumption or final consumption by entities other than households (companies, governments, etc.). This is why any comparison between a household survey and administrative data is spurious.
The other reported concern of ability/sensitivity of National Sample Survey Office (NSSO) survey instruments to capture consumption on social services like education and health also rests on shaky ground. There is no valid reason to make such a claim in the absence of a study. In any case, the same government has released survey reports done on education and health conducted in the 73rd round (2014). Both these surveys on consumption spends on social goods like education and health have also been used by the respective ministries.
Clearly, the objections raised by Mospi this time are not just irrelevant, but have been hurriedly invented on the day the report was leaked since no such concern was raised at any point of time by any official of the ministry before last Friday.
But even if there were any real technical concerns, then there are enough and more reasons to still release the report in the public domain. This is not the first time that questions have been raised about an NSSO survey. Perhaps the most controversial was the 55th round (1999-2000) consumption survey, which showed a sharp reduction in poverty compared to 1993-94. The response of the government was to release the data and it was this which contributed to a vast amount of research on the merits/demerits of NSSO surveys.
The participants in the debate included economist Angus Deaton, who later won the Nobel Prize, but also several others from India. It was this useful public debate which forced the then National Democratic Alliance to accept that the 1999-2000 poverty estimates are not comparable with earlier surveys. But at no point of time did the government prevent access to the data or the full report or attempt to delegitimize the data or the NSSO survey collection methodology.
However, this time, it is different and it seems to be part of the general pattern where the government in power has been trying to either deny access to data or has mounted a campaign to delegitimize unpleasant data. For NSSO, this is the third such instance this year alone.
The first was the attempt to block the release of the Periodic Labour Force Survey (PLFS), followed by subsequent attempts to raise doubts about the findings spearheaded by the top echelons of the government. Then came the episode of raising questions about the enterprise survey report, which found that a significant chunk of India’s registered firms are bogus entities. But in both those cases, the government at least did not block the release of the report and the data. In fact, this is the first time ever in the history of NSSO that a survey has been completed and a report approved for publication, but has been withheld from release. The question that should be asked is: why did the government take this extreme step?
Poverty on the rise
A quick reading of the leaked findings makes it amply clear that the real reason for withholding the release of the report is the revelation that real consumption expenditure on comparable measure has declined between 2011-12 and 2017-18, a large part of which is the period when this government has been in power. According to the leaked report, real consumption expenditure declined by 10% per annum in rural areas and increased marginally by 2% per annum in urban areas, with an overall decline of around 4% per annum for the country as a whole (simply put, Indians were spending less and consuming fewer items).
The immediate implication is that poverty headcount ratios would, in all likelihood, have increased between 2011-12 and 2017-18, as against a sharp decline between 2004-05 and 2011-12. It is this uncomfortable truth that the government did not want to come out at a time when the hype has been that everything is well with the economy.
The seriousness of these numbers can be understood by the fact that the last time real consumption expenditure declined during a six-year period was between 1960 and 1966, when the country was going through a severe crisis including a major food crisis. It also declined for a brief period around 1973-74, but not thereafter. The decline is surprising when compared to both earlier episodes, which were periods of economic uncertainty and crisis, because the period between 2011-12 and 2017-18 had reported growth rates at more than 6% per annum.
While the discomfort within the government is understandable, these numbers are not presenting anything new. Anybody who tracks the economy would have guessed. The surprise would have been if the consumption surveys showed a real increase. NSSO’s own surveys have pointed this out. While the consumption survey of 2017-18 has been junked, the data on consumption survey on consistent and comparable basis has shown that real consumption expenditure in rural areas declined from ₹1,587 per person per month (ppm) in 2014 to ₹1,524 ppm in 2017-18. The decline in urban areas was from ₹2,926 ppm in 2014 to ₹2,909 ppm in 2017-18 (all estimates in 2018 prices).
Similarly, the trend in growth rate of rural wages from the labour bureau has been consistently showing that real wages have stagnated between 2013 and 2018 covering a large part of the period between 2011-12 and 2017-18.
The PLFS data also showed that earnings of regular workers declined in both rural and urban areas between 2011-12 and 2017-18. Rural regular wages declined at 0.3% per annum, whereas urban regular wages declined at 1.7% per annum. At the same time, the fact that unemployment rates rose to the highest ever implies that the nature of growth in the last six years has been accompanied by job losses.
This period between 2011-12 and 2017-18 is an important period in the growth story of India. While it has been characterized by a sharp slowdown in the first three years due to policy paralysis, the last three years have seen two droughts, intensification of agrarian crisis, demonetization, and the introduction of the goods and services tax (GST).
Clearly, it is in everybody’s interest to understand and analyse what happened to individual incomes/consumption. But, more importantly, what happened to poverty and inequality. While all evidence points towards declining incomes at least in rural areas and among the informal workers, what has also come as a surprise from the unreleased consumption survey is the sharp decline in inequality reported in the second part of the article based on the leaked report.
Inequality fell in both rural and urban areas, suggesting that the poor were less affected by the overall distress in the economy than those at the top half of the distribution. It is also interesting to examine this pattern state wise. Preliminary analysis suggests that states with large urban and industrial concentration suffered, whereas some of the rural dominated and agrarian states have managed to buck the trend. It might as well be the case that poverty will still rise as a result of declining consumption but the impact of it was muted by a sharp decline in inequality.
Is it the case that those at the bottom end of the distribution and net consumers were less affected because of lower agricultural prices, whereas surplus farmers witnessed large decline in incomes?
Some of these questions are relevant for understanding the impact of past events such as falling agricultural prices, demonetization and GST. But they are far more important for making sense of the current economic slowdown and the future prospects of recovery. A detailed analysis with the unit level data would have allowed researchers to analyse the reasons for the slowdown and suggest possible remedies in terms of policy.
CESs are also vital to rebasing the consumer price indices which are on an upward curve over the last six months. It is also crucial for rebasing the GDP series and other national account aggregates, most of which are under a cloud. A detailed analysis of the consumption survey would have allowed at least some of the questions to be examined and analysed.
The decision to junk the consumer surveys, which are known internationally for their quality, has not only left a void in our understanding of crucial issues facing India, but has also hurt the credibility of a statistical system built carefully over the years by academic, practitioners and policy makers. The surveys used to enjoy a trust and reputation unparalleled in the world. While the economy may recover sooner or later, it is difficult to bring back the trust in the statistical system.
Himanshu is associate professor at Jawaharlal Nehru University and visiting fellow at the Centre de Sciences Humaines, Delhi.