Letting the Data Speak: Consumption Spending, Rural Distress, Urban Slow-Down, and Overall Stagnation

A view of busy Sarojini Nagar Market, in New Delhi on December 5, 2019. Photo: Sushil Kumar Verma

The National Statistical Office (NSO) brings out a survey, once every five years, on the distribution of consumption expenditure in the rural and urban areas of the country. The last survey report available pertains to the year 2011-12 (68th Round). A draft Report for the year 2017-18 (75th Round) is available with the Ministry of Statistics and Programme Implementation, but has continued to retain its ‘draft’ status for some time, without being released. In November 2019, the newspaper Business Standard came out with a sequence of articles detailing some disturbing trends in consumer spending derived from a leaked copy of the draft 2017-18 NSO Report. The Government of India has decided to ‘junk’ the draft Report, citing poor data quality as a reason.
Several economists and scholars have since protested vehemently against suppression of the report, and sought its release in the public domain. The following article reviews some results derived from an analysis of some of the data in the leaked draft NSO Report of 2017-18. The results, as will be seen, present a most unfavourable picture of tendencies in welfare and poverty indicators related to consumer spending over the period 2011-12 to 2017-18. These results, Economist, S. Subramanian, points out, lend themselves to the speculation that the withholding of the 2017-18 draft Report has less to do with its allegedly dubious data quality than with the depressing message on the economy it communicates.

"Data! Data! Data! I cannot make bricks without clay."
                                                                – Sherlock Holmes, in Arthur Conan Doyle,
"The Adventure of the Copper Beeches", The Adventures of Sherlock Holmes, 1892.

 

One of the finest features of our country, of which we can be legitimately proud, is our post-Independence record of the statistical base of India’s economy and society. This data base is a vast and extraordinary enterprise which had its origins in the exemplary vision of — among other actors — the late, great physicist, statistician and institution-builder Professor P. C. Mahalanobis. Most sadly, and like other institutions that have witnessed swift decay in the course of the last few years of our democracy, the autonomy and independence of our data-generating agencies have also been compromised. We have had misleading messages put out on the growth rates of Gross Domestic Product and on the record of open defecation, which have since been called out by responsible critiques of the methodology underlying GDP growth statistics, and by the actual statistics themselves on open defecation. Information on a record level of unemployment was withheld before its delayed release in the form of the Periodic Labour Force Survey’s Report for 2017-18. Crime statistics on the incidence of lynching have not been released. There is lack of transparency on how the fiscal deficit is being managed, in terms of easy and complete access to information on tax revenues and on the extent and composition of cuts in government spending. And so on.

The latest major case of data-opacity to which the country has been exposed is in the matter of the draft Report of the National Statistical Office’s survey on consumption spending in India.

The latest major case of data-opacity to which the country has been exposed is in the matter of the draft Report of the National Statistical Office’s (NSO) survey on consumption spending in India. The consumption distribution data, which are released once every five years, are the single most important source of information we have for tracking poverty and inequality. The Ministry of Statistics and Programme Implementation (MoPSI) has, however, decided not to release the draft Report for 2017-18, on grounds of the alleged poor quality of its data. It would have been a different matter if the Government had released the data while expressing its own reservations on the quality of the data. But simply suppressing the draft Report in question is an altogether different matter. The motive for censorship would also inevitably be called into question if it turned out that the NSO’s 2017-18 draft Report reflects an unflattering picture of tendencies in welfare, poverty, and inequality indicators relating to consumer spending. This indeed appears to be the case, as suggested by analyses of a leaked copy of the 2017-18 draft Report carried out by a number of independent commentators, including, in particular, Somesh Jha and Abhishek Waghmare in the Business Standard, Varun Krishnan in The Hindu, and Pramit Bhattacharya and Sriharsha Devulapalli in the LiveMint financial daily. 

In what follows, I also undertake a processing of some of the data in the 2017-18 draft Report, to get a sense of what message it conveys on certain crucially important features of our economy. In an earlier article published in The India Forum1, I had reviewed aspects of changes in rural welfare, poverty, and inequality, over the period 2011-12 to 2017-18, employing the National Sample Survey Office (NSSO) Report on consumer spending for 2011-12 (68th Round) and Tables T3 and T4 of the National Statistical Office draft Report for 2017-18 (75th Round), which, as already pointed out, has been ‘junked’ by the Government of India on grounds of the allegedly poor quality of its data. In the present article, and in the interests of completeness of record, I repeat some of these exercises for urban India, employing the previously mentioned data sources. I attempt to present a consolidated and comparative rural-urban picture of welfare, poverty, and inequality tendencies over the period 2011-12 to 2017-18. What we find is a deterioration of welfare in the rural areas, a slowing-down of improvement in the urban areas, and stagnation, if not worsening, at the aggregate all-India level.  I also comment on the judgement that the data in the junked Report are of such poor quality as to be virtually un-usable.

Good quality data, the need for their free dissemination in an environment of easy and democratic universal access are integral aspects of both good politics and good public policy.

I should perhaps warn that though I have tried my best to keep the discussion as non-technical and transparent as I could, there may nevertheless be elements of it which are a bit difficult for the non-specialist to completely comprehend. May I beg the reader’s indulgence for this deficiency, with the assurance that what really matters for an understanding of the article is the gist of its meaning, not the details of its numbers? Above all, this article is about the vital importance of good quality data for the formulation of sensible public policy, the need for their free dissemination in an environment of easy and democratic universal access, and the significance of maintaining the unfettered independence of our data-generating agencies as one integral aspect of both good politics and good public policy.     

1. 2011-12 to 2017-18: Changes in Aspects of Consumption

In Sections 1.1 to 1.3 I shall deal with some preliminary disaggregated statistics on consumption spending in both the rural and the urban areas of the country, for each of the years 2011-12 and 2017-18.

1.1 The Cumulative Density Functions for 2011-12 and 2017-18

The cumulative density function (cdf) of a distribution just presents information on the cumulative proportion of the population below each specified consumption expenditure level, when these expenditure levels are arranged in ascending order. Information on the cdfs and changes in them for both rural and urban India are presented in Table 1. If the cdf of one distribution lies everywhere below that of another, then we have a case of ‘first order stochastic dominance’, which is a signal of the unambiguous welfare superiority of the first distribution vis-à-vis the second. In the case of rural India, we have an unambiguous worsening of welfare from 2011-12 to 2017-18, as reflected in the first order dominance of the 2011-12 distribution over the 2017-18 distribution. Such an unambiguous deterioration is not reflected by the cdfs for urban India, but neither do we have evidence of an unambiguous improvement: the two cdfs intersect, with the 2017-18 distribution lying below the 2011-12 distribution for the most part, but above it at higher levels of expenditure. The picture can be visualised from the numbers in Table 1, which shows that while the 2017-18 urban cdf lies below the 2011-12 cdf upto an expenditure level of about Rs.4500 per person per month, this outcome is reversed for higher levels of expenditure of Rs.4500 and above. 

Table 1: Cumulative Density Functions of Consumer Expenditure for Rural and Urban India in 2011-12 and 2017-18

Table 1a: RURAL

Selected Levels of Per Capita Consumer Expenditure

 

(in Rs., at 2011-12 Prices)

Cumulative Proportion of Population Beneath Each Specified Consumption Level (2011-12)

 

Cumulative Proportion of Population Beneath Each Specified Consumption Level (2017-18)

 

600.0413.0461
900.2487.2825
1200.4984.5498
1500.6857.7399
1800.8070.8522
2100.8750.9130
2400.9145.9394

 

Table 1b: URBAN

Selected Levels of Per Capita Consumer Expenditure

 

(in Rs., at 2011-12 Prices)

 

URBAN

Cumulative Proportion

 of Population

 Beneath Each

 Specified Consumption Level (2011-12)

 

URBAN

Cumulative Proportion

 of Population

 Beneath Each

 Specified Consumption Level (2017-18)

 

600.0101.0056
900.0684.0419
1200.1808.1297
1500.3051.2449
1800.4217.3164
2100.5242.4691
2400.6109.5639
2700.6827.6449
3000.7410.7123
3300.7879.7674
3600.8262.8116
3900.8548.8469
4200.8775.8750
4500.8956.8970
4800.9103.9145
5100.9222.9284
5400.9320.9396
5700.9401.9485
6000.9470.9558

Source: Computations based on data in the 2011-12 NSO Report on Consumer Expenditure (68th Round) and Tables T3 and T4 of the 2017-18 NSO draft Report on Consumer Expenditure (75th Round)

1.2 Decile-Wise Mean Per Capita Consumer Expenditure Levels for 2011-12 and 2017-18

In Table 2 we have information on the average (real) level of consumption expenditure for each decile of the rural and urban populations, from the lowest to the highest, for each of the years 2011-12 and 2017-18. While in rural India, every single decile registers a decline in its average level of spending, and overall average spending has declined by 8.8 per cent, this pattern is almost completely inverted in urban India. Table 2 shows that in the urban areas every decile from the first to the ninth has a higher level of average spending in 2017-18 compared to 2011-12, with the rate of increase fairly systematically declining as we go up the decile ladder. However, the highest decile registers a fairly large proportionate decline in average consumption. Given that the tenth decile accounts for nearly 30 per cent of the total urban expenditure, the large decline in its average level of spending combined with the relatively modest rates of increase registered by the lower deciles has spelt an overall increase of just around 2 per cent in the aggregate average per capita consumption level. (I might add that ‘real’ levels of consumption expenditure have been obtained by employing the price deflator data on the Consumer Price Index of Industrial Workers furnished in Table T4 of the 2017-18 draft NSO Report). Subramanian and Lalvani (2018) report in Table 1 of their paper2 that between 2004-05 to 2011-12, real average per capita consumption expenditure in the urban areas increased for every decile group and by about 32 per cent overall, whereas between 2011-12 and 2017-18 we have a very modest aggregate increase of just 2 per cent, signalling a considerable slow-down in urban consumption growth.

Table 2: Decile-Specific Real Mean Consumption Levels: Rural and Urban India

(in Rs., at 2011-12 prices)

Decile

RURAL

2011-12

Means

RURAL

2011-12

Means

per cent Change

URBAN

2011-12

Means

URBAN

2017-18 Means

per cent Change
1594.86587.01- 1.32804.71900.29+ 11.88
2780.76745.10- 4.571118.091262.50+ 12.92
3900.87869.76- 3.451362.691519.16+ 11.48
41016.70978.92- 3,721624.861769.27+ 8.89
51138.251082.13- 4.931887.652037.89+ 7.96
61272.661197.42- 5.912180.522345.66+ 7.57
71434.251338.09- 6.712547.942707.44+ 6.26
81647.311521.59- 7.633062.853219.95+ 5.13
91987.641808.67- 9.003892.603989.67+ 2.49
103526.282912.02- 17.427815.957071.68- 9.52
All1429.961298.49- 8.802629.652682.35+ 2.00

Source: Same as Table 1

1.3 Inequality in the Distribution of Consumer Expenditure: 2011-12 and 2017-18

Table 3 carries information on the ordinates of the estimated Lorenz Curves for the distribution of consumption expenditure in 2011-12 and 2017-18. The Lorenz curve is a simple graphical device for estimating the extent of inequality that obtains in a distribution. It is derived by plotting the cumulative income/expenditure share against the cumulative population proportion, arranged in ascending order of income/expenditure, for every population proportion from 0 per cent to 100 per cent. Thus, the Lorenz curve will enable us to tell the income share accounted for by the poorest 10 per cent, the poorest 20 per cent, the poorest 30 per cent, and so on, of the population. If expenditure is perfectly equally distributed, then the poorest 10 per cent of the population will account for exactly 10 per cent of the total expenditure; the poorest 20 per cent will account for exactly 20 per cent of the total expenditure; and so on. The resulting Lorenz curve will then be represented by ‘the line of perfect equality’, which is just the diagonal of the unit square in which the Lorenz curve is drawn. But typically, distributions are unequal: the

Figure 1: The Lorenz Curve

 

Source: Wikimedia Commons

poorest 10 per cent of the population would account for less than 10 per cent of the total expenditure, the poorest 20 per cent for less than 20 per cent of the total expenditure, and so on. The Lorenz curve for an unequal distribution would thus be a rising curve that lies beneath the line of perfect equality. The lower the curve is, that is, the further away it is from the line of equality, the greater the extent of inequality in the underlying distribution. Indeed, the area enclosed by the Lorenz curve and the equality line is a very natural and plausible measure of inequality: this is the area marked A in Figure 1, which features a typical Lorenz curve. The well-know Gini coefficient of inequality, widely employed by economists to measure inequality in a distribution, is just twice the area marked A in Figure 1. If one Lorenz curve lies everywhere above another, the first distribution is said to ‘Lorenz-dominate’ the second one, and we can infer that inequality in the first distribution is unambiguously lower than in the second.

Table 3 suggests that in both rural and urban India, we have a case of Lorenz dominance of the 2017-18 distribution over the 2011-12 distribution, signalling an unambiguous decline in inequality over the reference period. This is reflected in a diminution in the Gini coefficient of inequality, from 0.2872 to 0.2581 in the rural areas, and from 0.3685 to 0.3298 in the urban areas. The move towards greater equality in the rural areas has been secured by what was described in my earlier India Forum article as a particularly harsh form of ‘levelling down’. That is, inequality has been reduced by reducing everybody’s income, but by proportionately more for the richer than the poorer sections of the population. This is scarcely the sort of ‘equalisation’ that any sensible egalitarian would seek.

Table 3: Ordinates of the Lorenz Curve

Cumulative Population Share.1.2.3.4.5.6.7.8.91.0

Cumulative

Expenditure Share

(2011-12):

RURAL

.0416.0962.1592.2303.3099.3989.4992.6144.75341.0

Cumulative

Expenditure Share

(2017-18):

RURAL

.0448.1027.1691.2436.3265.4186.5216.6385.77681.0

Cumulative Expenditure Share (2011-12):

URBAN

.0305.0732.1252.1865.2580.3412.4386.5551.70271.0

Cumulative Expenditure Share (2017-18):

URBAN

.0338.0805.1369.2027.2792.3670.4716.5912.73891.0

Source: Same as Table 1

Such systematic 'levelling down' has, however, not been the case in the urban areas. As we have seen, all but the richest (tenth) decile have experienced positive rates of growth in their respective levels of mean consumer expenditure, and the rates of increase have generally declined as we move up the deciles. However, the reduction in inequality has not been painlessly secured: the tenth decile has sustained a negative rate of growth in average spending. It should be noted that the average consumption of the tenth decile, at Rs.10,174 per capita per month at current prices (or approximately Rs.40,000 for a family of four per month) suggests that we are not speaking of millionaires here! And it is probably this class that has been hardest hit by the GST experiment.

1.4 A Consolidated Account of Changes in Welfare, Inequality, and Poverty

In Table 4, we have a summary picture of welfare indicators and how they have changed between 2011-12 and 2017-18. For purposes of comparison, I have juxtaposed the urban statistics with the rural statistics (which latter are drawn from Table 4 of my earlier article, previously alluded to, in The India Forum). Over the period 2011-12 to 2017-18, each of the following welfare indicators has registered a decline in rural India, and an increase in urban India: mean per capita consumption expenditure; the quintile expenditure statistic advanced by Kaushik Basu (which is just the average expenditure of the poorest 20 per cent of the population, and can be interpreted both as a welfare and as a money-metric poverty indicator); the average income of the below-median half of the population; and Sen’s  welfare index of 'distributionally adjusted' mean consumption expenditure.

A negative rate of growth in the rural areas—and for the first time in decades—is naturally cause for worry.

Mean consumption is the simplest summary statistic we can have on the magnitude of consumer spending. That it should display a negative rate of growth in the rural areas—and for the first time in decades—is naturally cause for worry. As we have seen, a disaggregated analysis of fractile-specific expenditure trends reveals that in rural India, every single decile registered a decline in its average level of spending. This pattern, as it happens, is almost completely inverted in urban India: Table 2 shows that every decile from the first to the ninth has a higher level of average spending in 2017-18 compared to 2011-12, with the rate of increase fairly systematically declining as we go up the decile ladder. However, the highest decile registers a fairly large proportionate decline in average consumption. Overall, average per capita consumption expenditure has increased very modestly, by just around 2 per cent, suggesting a considerable slow-down in urban consumption growth.

It is worth reiterating a simple point: all the welfare indicators we speak of have registered a modest, but nevertheless positive rate of change in the urban areas, in contrast to the uniform picture of deterioration revealed in the rural areas.

Inequality has declined unambiguously in both the rural and urban areas, as the relevant statistics on the Gini coefficient of inequality in the distribution of consumption expenditure reveal. As we have noted earlier, the rural reduction in inequality has been achieved by a proportionately greater reduction in the average expenditure levels of the richer deciles vis-à-vis the poorer ones: all decile means have declined. In the urban areas, however, we have modest and declining rates of increase in the average consumption levels of the poorest nine deciles; but an across-the-board improvement (which economists would typically call a ‘Pareto improvement’) is denied by the performance of the richest decile, whose mean expenditure level has declined quite considerably.

Table 4: Summary Welfare and Distributional Statistics for Rural and Urban India: 2011-12 and 2017-18 (All Monetary Values are in 2011-12 Prices)

 

2011-12

Rural

2017-18

Rural

Nature of Change (Rural India), with Comments

2011-12

Urban

2017-18

Urban

Nature of Change (Urban India), with comments
Welfare      
Mean Consumption (Rs)1429.961304.07

(-) 8.80 per cent: Deterioration.

Negative Growth after decades.

2629.652682.35

+ 2.00 per cent

Mild Improvement.

Quintile Consumption Level (Rs)687.81669.64

(-) 2.64 per cent:

Deterioration.

961.401081.40

+ 12.48 per cent

Improvement

Mean Consumption of Poorest Half of Population (Rs.)886.32851.23

(-) 3.96 per cent:

Deterioration.

1357.501497.74

+ 10.33 per cent

Improvement

Sen’s Welfare Index (Rs)1019.26967.49

(-) 5.08 per cent:

Deterioration.

1660.621797.71

+ 8.26 per cent

Improvement

Inequality      
Gini Coefficient.2872.2581

(-)10.13 per cent:

Reduction in Inequality, but achieved through ‘Levelling Down’

.3685.3298

(-) 10.50 per cent

Reduction in Inequality without Pareto Improvement in Consumption Levels

Poverty      

Headcount Ratio H

(Rangarajan Committee Poverty Lines of Rs.972 [Rural] and Rs.1407 [Urban])

31.15 per cent35.10 per cent

(+) 12.68 per cent:

Deterioration.

26.69 per cent20.83 per cent

(-) 21. 96 per cent

Improvement

Aggregate Headcount A (in millions of persons)270.43322.31

(+) 19.15 per cent:

Substantial Deterioration

93.9179.31

(-) 15.55 per cent

Improvement

Poverty Gap Ratio.0658.0746

(+) 13.37 per cent:

Deterioration

.0669.0466

(-) 30.34 per cent

Improvement

Squared Poverty-Gap Ratio (FGT-2).0208.0233

(+) 12.01 per cent:

Substantial Deterioration

.0242.0156

(-) 35.54 per cent

Improvement

 Source: Same as Table 1

The systematic decline of all fracticle-specific mean consumption levels in the rural areas has meant an increase between 2011-12 and 2017-18 in the headcount ratio (HCR) of poverty (employing the Rangarajan Committee’s recommended rural poverty line of Rs.972 per person per month) from 31.15 per cent in 2011-12 to 35.1 per cent in 2017-18, as well as in the aggregate numbers of people in poverty, from 270 millions to 322 millions. We also have increases registered in more sophisticated measures of poverty such as the poverty-gap index (which takes account of the shortfall of the average expenditure of the poor from the poverty line) and the squared poverty-gap index (which additionally takes account of inequality in the distribution of expenditure among the poor). 

In the urban areas, the increases, even if modest, in the expenditure levels of the poorer deciles, in combination with a decline in inequality, have meant that all poverty indicators, contrary to the performance in the rural areas, have displayed an improvement. This is true for the headcount ratio (HCR) (employing the Rangarajan Committee’s recommended urban poverty line of Rs.1047 per person per month at 2011-12 prices), which has declined from 26.69 per cent in 2011-12 to 20.83 per cent in 2017-18; the aggregate headcount (which has declined from nearly 94 millions to 79 millions); the poverty-gap ratio; and the squared poverty-gap ratio. It may be noted that if we employ the poverty line recommended by the Tendulkar Committee, which is lower than the Rangarajan Committee’s poverty line, then the urban headcount ratio declines by about 33 per cent from 13.44 per cent in 2011-12 to 9.02 per cent in 2017-18. According to the estimates of Subramanian and Lalvani (2018)3, the headcount ratio (employing the Tendulkar Committee poverty line) declined by about 50 per cent from 26.74 per cent in 2004-05 to 13.44 per cent in 2011-12. This, combined with our earlier observation on the much sharper increase in average consumption expenditure over the 2004-05 to 2011-12 period, as compared to the increase over the 2011-12 to 2017-18 period, suggests that improvements in welfare and poverty in urban India have definitely slowed down over the latest 6-year period vis-à-vis the previous 7-year period.

What can we say about tendencies at the aggregate all-India level? As we have seen, average per capita consumption expenditure in the rural areas has declined by about 8.8 per cent over the period 2011-12 to 2017-18, while it has increased by only 2 per cent in the urban areas over this period: given the preponderant weight of rural India in aggregate population, this would suggest an overall decline in per capita consumption expenditure. The all-India poverty headcount ratio (the weighted average for the rural and urban areas, using population weights) has increased marginally from 29.87 per cent to 30.92 per cent, while the aggregate headcount has increased more substantially from 364.34 millions in 2011-12 to 401.62 millions in 2017-18. At best, therefore, the overall picture that emerges for India is one of stagnation in welfare and poverty levels, leaning on the side of deterioration.

Taken together, these results should explain a major part of the title of this article: 'rural distress, urban slow-down, and [charitably] overall stagnation'.  

2. Concluding Discussion 

How plausible are the results reviewed in the foregoing account? It is customary to assess the issue of plausibility in terms of other broadly known and acknowledged facts of the economy. There is a wide consensus of opinion among professionals now that the demonetisation experiment was an unmitigated disaster, with grave implications for the living standards of a vast majority of the population, especially in the context of a highly cash-dependent rural economy. There is also wide consensus on the view that a hastily and shoddily implemented GST regime has caused much hardship. There is little doubt that growth in Gross National Product (GDP) has slowed down considerably—necessitating a downgrading of officially projected growth rates by institutions such as the International Monetary Fund, the World Bank, various credit-rating agencies, and recently by the Government of India itself. Agrarian distress has been rampant—presumably the numerous protest foot-marches undertaken in the recent past by farmers to the capital city were not done just for the heck of it. The Index of Industrial Production has had no good news to offer. The unemployment rate revealed by the latest (and also late-to-be-released) Periodic Labour Force Survey for 2017-18 is unprecedentedly high. And even a generally pliant media has been unable to suppress disquiet over recessionary trends and slackening aggregate demand across a range of sectors and products, from automobiles to biscuits.

Even a generally pliant media has been unable to suppress disquiet over recessionary trends and slackening aggregate demand.

Is this then an endorsement of the complete reliability of the data in the junked 2017-18 NSO draft Report? No, it is not. That would have to await a detailed study and critique of the shortcomings, if any, in aspects of survey methodology that are specific to the 2017-18 Report. What is being asserted, though, is that there is little prima facie evidence to suggest that the Report is deeply flawed, and that its data quality is so seriously deficient as to render it unusable for analysis. And yet, it is this latter judgement that has prevailed with the Ministry of Statistics and Programme Implementation. Nor is such a castigation of the Report exclusively an official response to it: it seems to be shared by a few other ‘non-official’ commentators as well.

What are the supposed defects of the junked draft Report? From what one can gather, there is the issue of a discrepancy between survey estimates of consumption and National Accounts Statistics (NAS) estimates, with the former displaying considerably lower levels than the latter. But this is an old problem, and one which is by no means peculiar to the 2017-18 survey. Indeed, it is widely known that for long there has been a hiatus, and an increasingly diverging one, between the NSO survey estimates and the NAS estimates of consumption. A particularly drastic resolution of the problem was advanced by the Government in the mid-1980s, when poverty statistics were computed by uniformly scaling up the survey estimates of consumption across all fractiles of the population by the ratio of the NAS mean consumption to the NSO mean consumption. This result, as expected, yielded very pleasing magnitudes and trends of poverty. Why ‘as expected’? Because it is a mathematical certainty that if you raise all consumption levels for any given distribution of consumption, the poverty rate will fall: it is a truism concerning averages and dispersions around the average. But the truism is not a very helpful one when the objective of the poverty analyst is, presumably, to verify what has been happening to poverty on the ground, not to verify the consistency of arithmetic!

The ‘adjustment’ of survey consumption figures by pro-rating with respect to NAS figures is a peculiar hybrid marriage of two data sources, and the issue was squarely addressed in the Indian context as far back as the late-1980s and early-1990s. The definitive case against such ‘adjustment’ was made by the late Professor B. S. Minhas4, and the practice was mercifully laid to rest. It is therefore strange that we are beginning again to hear a call for such ‘adjustment’, apparently in the light of other country-experiences in this regard. As for international opinion on the subject, it cannot hurt to defer to Branko Milanovic, one of the most universally respected experts on the assessment of global inequality. The following excerpt from a study by Lakner and Milanovic5 is telling:

Addressing jointly top income underreporting and the national accounts discrepancy. The underreporting of top incomes in household surveys and their discrepancy with national accounts are closely connected issues. It is reasonable to expect, and there is some empirical evidence to corroborate it, that the discrepancy between surveys and national accounts is not distribution neutral and is largely due to non-participation of the rich in household surveys (Mistaenen and Ravallion 2003; Korinek et al 2006). Deaton (2005) points out that because national accounts consumption tracks money rather than people, national accounts data are more likely to capture large transactions. Using Indian tax record data, Banerjee and Piketty (2010) find that a significant part of the discrepancy between consumption growth in national accounts and household surveys can be accounted for by underreporting of the rich. Finally, it could be argued that household surveys offer a good approximation to the bottom 90 per cent of the distribution (thus, however, ignoring any underreporting of incomes among the very poor).

The reliability of the 2017-18 consumption data has also been called into question by citing increased sales/utilisation figures for certain products and services such as cars, two-wheelers, the internet facility, mobile phones, smartphones, and airline travel; and increases registered by the sales data of companies such as Hindustan Unilever, Nestlé, Brittania Industries and Venky’s (which specialises in poultry products). To do it justice, the draft 2017-18 NSO Report suggests a decline in rural, not urban, spending. It is hard to imagine that the Report should be pilloried for failing to accurately reflect the alleged bucolic reality of the rural poor whizzing around in their BMWs, jetting to and from their exotic holiday destinations, and making whoopee with beauty products and packaged foodstuff.

To do it justice, the draft 2017-18 NSO Report suggests a decline in rural, not urban, spending.

Then there is the alleged problem of underreporting of consumption by the poor. On this, India's Chief Statistician and the Secretary of the Department of Drinking Water and Sanitation say:

'It all boils down to basic behavioural economics. Although field enumerators attempt to elicit genuine responses, people respond favourably to incentives, either direct or perceived. A genuine response based on a sense of nationalism is diminishing.'6

Apparently, the poor understate their expenditure levels in order to avoid being graduated out of eligibility for Government benefits and subsidies. As it happens, the impugned NSO Report reveals higher rates of growth in urban consumption for the poorer fractiles of the population than for the richer fractiles; to repeat, it is rural India that has displayed a reduction in average consumption. This makes it hard to quite digest the notion that the real problem has to do with the distressing and suddenly acquired mendacity and anti-nationalism of poor villagers; and that it must be a pitiable survey, then, that takes no stock of such villainy.

Blaming the data, delaying the release of data, altogether suppressing the data—and eventually being caught out by the data: these are neither seemly nor statesmanlike responses of a Government to its own poor performance. The country’s economists and statisticians do not need to be protected from poor quality data: they must be expected to have both the competence and the honesty to pronounce adversely on the data if the quality is indeed poor. But for that to happen, they must first, last, and all the time be allowed access to the data!

[S. Subramanian is a former Indian Council of Social Science Research (ICSSR) National Fellow and a former professor of the Madras Institute of Development Studies. Subramanian is an elected fellow of the Human Development and Capability Association (HDCA). He has worked extensively on measurement and other aspects of poverty, inequality, and demography, and on topics in collective choice theory, welfare economics, and development economics. In 2015, he was appointed as a member of the Advisory Board of the World Bank's Commission on Global Poverty under the Chairmanship of Sir Anthony Atkinson. He lives and works in Chennai. He can be contacted at [email protected]].

Note:

The author is indebted to C Rammanohar Reddy, Editor of The India Forum, for permission to draw substantially on an article of the author’s which was earlier published in that journal-magazine.

References:

1. Subramanian, S. 2019. What is Happening to Rural Welfare, Poverty, and Inequality in India?, The India Forum, November 29. [https://www.theindiaforum.in/article/what-happened-rural-welfare-poverty-and-inequality-india-between-2011-12-and-2017-18]. Return To text.

2. Subramanian, S and Lalvani, M. 2018. Poverty, Growth, Inequality: Some General and India-Specific Considerations, Indian Growth and Development Review, Issue. 11, Vol. 2, pp. 136-151. Return to Text.

3. See footnote 3. Return to Text.

4. Minhas, B.S. 1988. Validation of Large Scale Sample Survey Data: Case of NSS Estimates of Household Consumption Expenditure, Sankhya, Series B, Vol. 50, Part 3, Supplement. Return to Text.

5. Lakner, C and Milanovic, B. 2013. Global Income Distribution From the Fall of the Berlin Wall to the Great Recession, Policy Research Working Paper 6719, The World Bank Development Research Group, Poverty and Inequality Team, December. [http://documents.worldbank.org/curated/en/914431468162277879/pdf/WPS6719.pdf]. Return to Text.

6. Srivastava, P and Iyer, P. 2019. Surveys measuring impact of govt programmes have become less reliable, The Indian Express, December 4. [https://indianexpress.com/article/opinion/columns/between-the-lines-of-a-survey-6149205/]. Return to Text.


Download PDF [465 KB]


Post a Comment
  1. Comments will be moderated.
  2. Comments that are abusive, personal, incendiary or irrelevant cannot be published.
  3. Please write complete sentences. Do not type comments in all capital letters, or in all lower case letters, or using abbreviated text. (example: u cannot substitute for you, d is not 'the', n is not 'and').
  4. We may remove hyperlinks within comments.
  5. Please use a genuine email ID and provide your name, to avoid rejection.