APPENDIX II

TRENDS  IN FOOD CONSUMPTION AND LIKELY FUTURE OUTCOMES

1.1. It is widely accepted that dietary patterns have changed in India over the last thirty years. In particular, consumption of coarse cereals has declined in favour of rice and wheat. Also, cereals as a whole now account for a smaller share of total food consumption. However, the National Sample Survey (NSS), which is the main source of data for demand analysis, differs considerably from estimates of food availability (i.e. production less seed, feed, waste, industrial use, net exports and change in public, but not private, stocks), complicating accurate assessment of future prospects.

1.2. As against the usual assessment that cereals production and availability has increased significantly over the last thirty years, the NSS shows massive decline in cereals consumption - from 175 kilograms per capita per annum (kpca) in 1972-73 to only 147 kpca in 1999-2000 (Chart 2.1). This is mainly because NSS estimates for earlier years were high and failed to reflect production growth. But with production flattening out during the 1990s, there is decline in both NSS consumption and official availability. Differences persist, with the NSS showing cereals consumption down significantly from 166 kpca in 1987-88 to 147 in 1999-200 as against only marginal decline in official availability from 156 to 155 kpca between the triennia ending (TE) 1988 and 2000. But since official availability also shows large decline after 1997, dropping to 142 kpca in 2001, there is little doubt that per capita cereals consumption is lower now than a decade ago.

1.3. Most of the decline in cereals consumption is accounted for by coarse cereals, which also account for much of the difference between NSS and official data. Coarse cereals production has stagnated since 1970-71, and thus declined in per capita terms. Consequently, official estimates of their per capita availability have dropped steadily from 37 kpca in TE 1973 to 26 kpca in TE 1988 and further to 23 kpca in TE 2000. The NSS indicates even larger decline in coarse cereals consumption – from 50 kpca in 1972-73 to 25 kpca in 1987-88 and to less than 14 kpca in 1999-00 – and shows this occurring across all income classes and in a regionally widespread manner. In 1972-73 NSS coarse cereals consumption had exceeded 36 kpca in Andhra Pradesh, Bihar, Gujarat, Haryana, Karnataka, Madhya Pradesh, Maharashtra, Rajasthan, Tamilnadu and Uttar Pradesh, and accounted for more than 20 per cent of total cereals consumption in all these states. By 1987-88, reduced coarse cereals consumption meant that these provided 20 per cent or more of total cereals in only Gujarat, Karnataka, Maharashtra and Rajasthan. Consumption has declined further in every state, so that in 1999-2000 only these four states reported coarse cereals consumption higher than 12 kpca, or accounting for more than 5 per cent of total cereals consumption. Interestingly, even for these four states, the 1999-2000 NSS estimate of coarse cereal consumption is 10.3 million tonnes, well short of production of nearly 18 million tonnes during that year. Production was over 12 million tonnes in the remaining states but NSS consumption only 3.4 million tonnes. Even allowing for seed, feed and wastage, the official estimates of coarse cereal availability thus exceed NSS consumption by about 10 million tonnes, suggesting that coarse cereals use for purposes other than human consumption, i.e. as feed for animals and poultry or for starch manufacture, might have increased considerably more during the 1990s than is being allowed for in official estimates.

1.4.  Unlike coarse cereals, there has been sustained growth in per capita production of rice and wheat. But availability for consumption has grown less than production because of import substitution and stock accretion, and growth of rice output has decelerated to just about population growth after the 1980s. Also, as for cereals as a whole, NSS consumption estimates for both rice and wheat were higher than official availability in earlier years. Nonetheless, both sources of data show significant increase in per capita consumption of rice and wheat till the late 1980s followed by stagnation or decline (Chart 2.3). Official availability rose from 111 kpca in TE 1973 to 134 in TE 1991 (or growth of 1 per cent per annum) and declined marginally to 133 kpca in TE 2000 before plummeting to 119 kpca in 2001. Against this, the NSS shows consumption growing from 125 kpca in 1972-73 to 138 in 1990-91 (or at 0.6 per cent per annum), followed by decline to 133 kpca in 1999-2000. Despite differences in earlier years, official availability and NSS consumption are in very close agreement in recent years, both averaging 134 kcpa during 1991-2001, and this lends confidence to inferences from these data. One inference is of course that earlier growth in consumption of rice and wheat did not sustain after the late 1980s. Another inference follows from the fact that although both official availability and NSS consumption have averaged the same since 1991, availability in 2001 was 15 million tonnes lower than required to maintain consumption, having earlier exceeded consumption during 1995, 1996 and 1997 by a cumulative 26 million tonnes. This suggests that there may already have been surplus over consumption by 1995 but with considerable build-up of private stocks during 1995-1997, releases from which could have caused the otherwise inexplicable fall of prices in 2000 and 2001 despite declining availability.

1.5. An important aspect of rice/wheat consumption is that over time this has become more equal across income groups. According to NSS data, average rice/wheat consumption amongst the poorest 20 per cent of the rural population was only 76 kpca in 1972-73, less than half of the 180 kpca consumed by the richest 20 per cent. By 1987-88, consumption of the poorest quintile had increased over 40 per cent to 107 kpca, while consumption of the richest quintile remained unchanged at 180 kpca. Similarly, in urban areas, consumption of the bottom quintile increased 20 per cent from 90 to 108 kpca between 1972-73 and 1987-88, as against only 2 per cent increase from 132 to 135 kpca for the top quintile. Rice/wheat consumption has subsequently declined very sharply amongst the rich, especially in rural areas, but not amongst the poor (Chart 2.4). Thus, although total cereals consumption (including coarse cereals) has reduced during the past decade amongst the poor also, the shift out of rice/wheat has been confined mainly amongst the non-poor. However, despite this, rice/wheat consumption declined in all states except Andhra Pradesh, Karnataka and Maharashtra, the three states with largest decline in coarse cereals consumption.

2.1. It is therefore almost certain that cereals consumption did decline during the 1990s, but this does not imply decline in total food consumption. In fact, although data from the National Accounts Statistics (NAS) confirm a 6 per cent reduction in per capita real expenditure on cereals between TE 1990-91 and TE 2000-01, this also shows that during the same period there was 16 per cent increase in per capita real consumption expenditure (at 1993-94 constant prices) on all food items taken together. This is because of increased consumption of non-cereals food, which the NAS estimates as having increased at about 2 per cent per annum in real per capita terms since the mid-1970s with, if anything, a slight acceleration during the 1990s. Between TE 1990-91 and TE 2000-01, the NAS reports increased real per capita expenditure on all categories of non-cereals food except pulses. The reported increase is small (about 3 per cent) for sugars, moderate (around 15 per cent) for edible oils and fruits and vegetables, and large (45 per cent or more) for animal products (i.e. milk and milk products and meat, egg and fish) and for “other food” (including spices, tea/coffee, and consumption in hotels and restaurants).

2.2. These estimates need to be treated with caution since data on production and consumption of non-cereals food, particularly horticulture and livestock, are known to be weak. Confidence in these is further diminished since in many cases NAS estimates are much higher than those obtained independently by the NSS. The NSS reports total expenditure on non-cereals food at only 55-65 per cent of the NAS, but excludes consumption outside households and there are problems relating to choice of reference periods in NSS 55^th round and comparability issues introduced thereby. Nonetheless, if this round is included, rates of growth of total consumer expenditure on non-cereals food from the NSS are very similar to those from the NAS (Chart 2.5). Moreover, because the NSS differs most with official data on low calorie items such as fruits and poultry, there is closer correspondence between these two data sources on calorie intake from non-cereal food than on expenditure on such food. NSS estimates of calorie intake from non-cereals are 75-85 per cent of estimates in Food Balance Sheets derived from official data, and both show similar per capita long-run growth at about 2 per cent per annum, although the 1990s trend in NSS remains sensitive to inclusion of the 55^th round. Thus, official statistics probably overestimate consumption of non-cereals food, but such consumption has undoubtedly increased significantly over the last 30 years and has improved variety and diversity in Indian diets.

2.3.  However, despite increased consumption of non-cereals food, cereals continue to be the most important source of nutrients. According to Food Balance Sheets, cereals currently supply 59 per cent of total calories, 58 per cent of total proteins and 13 per cent of total fats (having come down from  66, 65 and 20 per cent respectively in 1981) and the NSS implies an even larger cereals contribution, about 65 per cent of protein/calorie intake in 1999-2000. Since cereals account for a much smaller share of total food expenditure (33 and 28 per cent in 1999-00 according to the NSS and NAS respectively), this implies that cereals are much cheaper sources of nutrition than other food. Not surprisingly, therefore, the poor are observed to obtain a much larger share of calories from cereals (about 75 per cent for the bottom 40 per cent of the rural population in 1999-00) than the average. Also, diversification away from cereals is increasing the cost per unit nutrient so that the increase in calorie and protein intake is less than the increase in food expenditure.

2.5.  Nonetheless, since NSS estimates of cereals consumption for the 1970s and early 1980s were almost certainly overestimates and because consumption of non-cereals food has increased significantly, it is fairly certain that the nutritional situation did improve during those decades. But the large difference in more recent data is worrying. The latest NSS round in 1999-2000 involved a change in reference periods which is likely to have led to higher estimates of food consumption than in earlier rounds, but even this shows average intake at only 2150 kcal/day, almost unchanged from the earlier NSS quinquennial round in 1993-94. In both these years NSS estimates were less than 90 per cent of the corresponding Food Balance Sheet (FBS) estimate of around 2400 kcal/day. For both 1993-94 and 1999-00, the NSS and FBS agree that rice and wheat supplied about 1250 kcal/day and the much higher total calorie intake in the FBS is entirely on account of items such as coarse cereals, livestock and horticultural products for which there are reasons to suspect overestimation in official availability. Of course, accretion to public stocks of rice and wheat reduced calorie availability by about 125 kcal/day in 1999-2000; but if non-food use of coarse cereals has in fact increased recently and if there is overestimation of non-cereals consumption, the nutritional adequacy of current food availability, after stock accretion, may not be as secure against the norms of the Indian Council of Medical Research as would appear from Food Balance Sheet data.

2.6.  Among the major states, NSS estimates lowest nutrient intake in Assam, Gujarat, Kerala, Maharashtra and Tamilnadu. Total calorie intake averaged less than 2000 kcal/day in each of these during 1993-94 and, except in Kerala and Maharashtra, worsened further in 1999-2000. Two other states where average intake declined during the 1990s and was less than 2100 kcal/day in both 1993-94 and 1999-00 were Andhra Pradesh and Karnataka. Except Assam, these are all states with above average per capita income and below average poverty incidence. Also, most of these are states where diversification of food production and consumption away from cereals has proceeded further than in the rest of the country. That these states still have lowest total calorie intake is in part accounted for by physiological requirements and cultural habits. But this also appears to be related to a deficiency in cereals since, along with Bihar, these seven states have the lowest per capita cereals production among all the major states. This not only emphasises the continued importance of cereals even in relatively rich states, but also the limitations of trade. A production deficit by itself does not imply food insecurity but involves additional costs and effort of getting supplies from elsewhere. Deficit production and relatively long distances from surplus regions lead to relatively high prices, e,g. NSS implicit prices of cereals purchased in the open market were 30-70 per cent higher in the South Indian states of Karnataka, Kerala and Tamilnadu than in Uttar Pradesh. This restrains cereals consumption and keeps total nutrient intake even below that in the poorest states.

2.7. However, nutritional outcomes among children, which in India are generally even worse than in sub-Saharan Africa, are somewhat superior in many of these richer states than in Bihar or Uttar Pradesh despite the lower nutrient intake. This is because health and sanitary conditions allow more efficient absorption, and because distribution is better, including within families and, with superior public distribution, also over months of the year. Nonetheless, although child health indicators show evidence of better food absorption, low calorie intake associated with low cereals production and relatively high cereals prices keeps the share of adult population with chronic energy deficiency (CED) at over 50 per cent in the relatively rich states of Gujarat, Karnataka and Maharashtra, even higher than in Uttar Pradesh. The situation is different in the rich cereals-surplus states of Haryana and Punjab where cereals prices are relatively low. Here, too, per capita cereals consumption has now fallen well below the all-India average but total nutrient intake is amongst the highest because of high consumption of other food and the incidence of CED is low. Even more interesting are Kerala and Tamilnadu where nutrient intake is the lowest but CED incidence is also relatively low, probably because of the better intra-year distribution of cereals and calorie intake permitted by relatively well functioning PDS. It is well known that the relationship between nutrient intake and nutritional outcomes is complex, and is conditioned not only by public health inputs and micronutrient intake but also by social factors and intra-family distribution. But, in addition, it appears that at the present development stage, it would be a mistake to expect much improvement in India’s appalling nutritional outcomes without continued effort at providing assured supplies of cereals either through local production or through a well functioning distribution system which can keep cereals affordable and prevent undue fluctuations in their price.

2.8.  This complex relationship between nutrition outcomes and assured access to cereals is relevant to whether nutritional status can improve through higher consumption of non-cereals food if cereals prices are high. In this context, it is also important to note that, although cereals consumption has declined and consumption of non-cereals food has increased, NSS data show no further reduction during the 1990s of the share of cereals in total food expenditure for any population quartile in either rural or urban India. This share, at current prices for the population average, had fallen steadily from 51 per cent in 1972-73 to 33 per cent in 1990-91 when cereals prices had risen less than other prices, but shows no decline thereafter. The cereals shares in total NSS food expenditure were 37.3 and 25.7 per cent in rural and urban areas in 1999-00, slightly higher than the corresponding shares of 37.2 and 24.5 per cent in 1990-91, with shares even higher during all intervening years for which data is available (Chart 2.7). The NAS also shows a similar trend, with the share of cereals in total private consumption expenditure on food dropping almost steadily from above 40 per cent in the mid-1970s to 26.5 per cent in 1990-91 but fluctuating thereafter within the range 24-29 per cent. This is a consequence of larger increases in cereals prices during the 1990s than of prices of other food and implies that the importance of cereals, which still account for half the food expenditure of the poor, has not diminished in household food budgets during the past decade despite increased consumption of non-cereals.

2.9. Real cereals prices increased sharply during the 1990s, reversing an earlier trend. Computed as ratio of the Wholesale Price Index (WPI) for cereals to the all-commodity WPI, this had declined 27.1 per cent between 1975-76 and 1990-91. However, this increased 8.4 per cent in 1991-92 and, after remaining relatively stable for the next four years, rose 23.2 per cent from 1996-97 to 1999-2000 before declining by 13.4 per cent subsequently. On WPI basis, therefore, real cereals prices increased 33.2 per cent between 1990-91 and 1999-2000, and continue to be 15 per cent higher (Chart 2.8). Implicit cereals prices from the NAS and NSS show similar movement although the post-1996 increase is somewhat higher in the NAS and lower in NSS than in the WPI. NAS consumption deflators show cereals prices 26 per cent higher in 2000-01 as compared to 1990-91, relative to both non-cereals food and non-food.

2.10. This behaviour of cereals prices is noteworthy for a number of reasons. First, farmers have received lower prices for cereals after 1999-2000, but from a high base. Second, high and rising cereal prices during the 1990s, especially in the later half of the decade, has dampened diversification of agricultural output and also prevented reduction of the cereals share in household budgets. Third, and most important, this has had an  adverse impact on the nutrition condition of the poor for whom cereals still account for almost 50% of food expenditure and over 70% of calorie intake. That this was significant has already been suggested by the evidence from Food Balance Sheets which not only show no increase in per capita cereals availability or per capita calorie intake since 1991 but also their actual decline since 1997, reversing the trend during the 1970s and 1980s. Moreover, there is also evidence that the 1990s saw much more erratic outcomes than the earlier trend reduction during the 1970s and 1980s in both child malnutrition, as measured by weight and height for age[1], and acute demographic stress, as measured by sudden changes in fertility and adult mortality in cereals deficit states[2]. The NSS shows a sharp reversal during the 1990s of an earlier increase of cereals and calorie consumption among the poor (Chart 2.9) and, this also supports the general conclusion that the nutritional condition of the poor worsened during the 1990s.

II.3. Interpretations of available data

4.2.  From the above discussion on cereals consumption trends, the following may be noted regarding the available data from official production estimates and the National Sample Survey (NSS):

Recent NSS data show that estimates of food consumption are sensitive to the choice of reference period. Cereals consumption is reported 14 per cent higher with a 7 day rather than 30 day reference period and differences are larger for other food items. This may bias NSS estimates, and has led to problems of comparability of the 1999-2000 data with earlier estimates

The NSS reports massive decline in per capita cereals consumption from 175 kilograms per capita per annum (kpca) in 1972-73 to 170 kpca in 1983, 166 kpca in 1987-88, 161 kpca in 1990-91, 154 kpca in 1993-94 and further to only 147 kpca in 1999-2000. However, any large decline during the 1970s and 1980s appears implausible since there is little doubt that per capita cereals production did increase during those decades, increasing the per capita availability from 135-150 kpca to 150-165 kpca. The NSS appears to have overestimated cereals consumption earlier, and a progressive decline in the degree of this overestimation may explain this.

There is much closer agreement between NSS and production estimates after 1988. For rice and wheat, NSS consumption and official availability averaged similar, at 138 and 134 kpca, during TE 1991 and both were 133 kpca in 1999-2000 and TE 2000 respectively. Coarse cereals consumption at 26 kpca was also the same in both sources during TE 1991. However, while coarse cereals availability declined only slightly to 23 kpca in TE 2000, NSS consumption fell to only 14 kpca in 1999-00. One explanation is that feed use may have increased much more rapidly during the 1990s than is assumed in calculations of availabilty. But, if so, present feed use of cereals is about 10 million tonnes more than officially estimated.

For rice and wheat, although NSS consumption and availability have both averaged 134 kpca since 1990, the time pattern is somewhat different. The two sources agree at both ends of the 1990s, but NSS consumption was significantly below availability during 1995-97, especially for wheat. One possibility is that private stocks were built up in this period of high and rising international prices. This is plausible because later developments in 2000 and 2001, when prices and availability fell together, can be explained only by some combination of a fall in incomes and release from private stocks. But, if so, private stock behaviour was destabilising in the 1990s.

Considering the the National Institute of Nutrition (NIN)’s normative cereals requirement of 157 kilograms per person per year, it is observed that net production of cereals (net of feed, seed and waste) was short of requirement till 1988, but has exceeded the norm in all subsequent years except 1996 and 2001. Net availability or apparent cereals consumption also exceeded normative requirement in all but one year (1993) during 1989 to 1997, and averaged 163 kgms/person/year. But, worryingly, this has fallen below the NIN norm in three of the four years after 1997 for which data is available.

Both the NSS and official data on availability show dietary diversification, with per capita real expenditure on non-cereals food growing. However, the 1990s trend in the NSS is sensitive to inclusion of the 55^th round.

The NSS reports much lower consumption of most non-cereals food items than their assumed availability, especially in case of horticulture and livestock products whose production data are known to be very weak. This affects assessment of the nutritional impact of dietary diversification.

In particular, while average calorie intake during the 1990s is placed over 2400 kcal/person/day on the basis of availability data, the NSS reports only 2150 kcal/person/day, less than in the 1980s. However the Food Balance Sheet data also show that calorie intake at the national level peaked in 1996, after which there has been some decline.

Although the NSS shows some increase in the consumption of rice and wheat by the poorest groups during the 1990s, their overall cereals consumption, including coarse cereals consumption, declined in per capita terms. More importantly, overall calorie intake of the poorer groups declined throughout the 1990s. There is corroborative evidence from data on nutrition outcome indicators which, although they show trend improvement, imply  more erratic behaviour after 1990.  

4.3. On the basis of the above, it is evident that there are significant data problems. Nonetheless, it is clear that there was decline in cereals consumption per capita during the 1990s, especially after the mid-1990s. Overall calorie intake declined despite some diversification of the dietary pattern, and this is associated with a larger decline in cereals and calorie intake in rural than in urban India. Most importantly, there is unambiguous evidence that the nutrition condition of the rural and urban poor worsened in the 1990s, after having improved during the 1970s and 1980s. An obvious question is whether this reflects a slow down after 1990 in poverty reduction and income growth, particularly in rural areas and among labour. However, this is again a matter on which there are differences:

That poverty reduction may have suffered a setback during the last decade is supported by results of nine NSS rounds (Rounds 46 to 54) from 1990-91 to 1998, which showed no poverty reduction, a clear break with the trend in the two previous decades. However, the 55^th NSS round conducted in 1999-00 shows large poverty reduction. Since this round is not comparable to earlier rounds because of reference periods used, it has been argued that this might have overestimated poverty decline so much that trends revealed by the previous nine rounds should be accepted. However, although it is agreed that the 55^th round did underestimate poverty, there is disagreement about its degree. Some analysts have argued that poverty decline during the 1990s may  still have been similar to that in the 1970s and 1980s. There is evidence from 9 of 10 NSS rounds that poverty reduction slowed down, but this is contested.

Similarly, there can be disagreements about agricultural growth. There are large differences between different estimates, mainly on account of fruits and vegetables for which data is poor. The official Index of Agricultural Production, which covers only those fruits and vegetables whose data are reliable, shows agricultural growth at only 1.7 per cent per annum between TE 1991-92 and TE 2001-02. This is much lower than in the previous two decades, and less than population growth. Latest National Accounts estimates attempt more comprehensive coverage of fruits and vegetables. Although these also show  agricultural growth less than in the previous decade, this is put at around 3 per cent per annum during the 1990s, well above population growth. Again, the evidence is that per capita farm incomes fell if crops with inadequate data are excluded, but this too can be contested.

There is greater agreement regarding employment growth, which is known to have decelerated after the 1980s. It is generally agreed that unemployment has increased during the last decade, that there has been a marked slow down in employment growth in the organised sector and that mechanisation has reduced labour absorption in agriculture. Also that, although employment has increased in the unorganised sector in activities such as trade, transport and construction, overall growth of employment (both organised and non-organised) in non-agriculture has halved, from around 4 per cent per annum during the 1970s and 1980s to about 2 per cent during the 1990s; and the casual labour component has increased in both agriculture and non-agriculture. All these would suggest a worsening of labour market outcomes, but differences exist regarding the reasons for the decline in employment growth and about the quality of employment and on wage rates. There is a view that employment growth has reduced mainly because the young are staying longer in education and differences between different sources of wage data are so wide that it is possible to claim both a sharp decline in real wage growth and some improvement. Although most analysts would accept that employment trends have on the whole been less favourable after 1990 than earlier, many would still claim that wage labourers have become significantly better off in both rural and urban areas.

4.4.  Most of these differences in the interpretation of available data on poverty, employment and rural income growth stem from the fact that although a considerable body of data do indicate worse outcomes post-1991, this is difficult to reconcile with the high rates of GDP growth which have been recorded during this period. Either it must be accepted that benefits of this growth have not percolated to most rural areas and to the poor, in which case there is no difficulty in accepting the evidence on cereals consumption and calorie intake. Or, a more complicated construct is necessary to explain actual outcomes. This involves the view that there has been a large change in patterns of production and consumption favouring horticulture and livestock which simultaneously created rural incomes, led to growth of transport and trade opportunities, and explains the consumption shift out of cereals; but that this was not being reflected adequately in either production or consumption data. In an effort to reflect this adequately, the share of fruits and vegetables in total agricultural production has been increased considerably in the latest series of National Accounts, and reference periods were changed in the NSS 55^th round in a way which it was known would lead to higher estimates of food, particularly non-cereals food, consumption. Problems of interpretation have thereby led to changes in the data itself, making scientific assessment difficult.

4.5.  From the Committee’s point of view, the main problem is that unless the 1990s fall in cereals consumption is explained, future projections are difficult. If poorer groups did become better off, the decline in cereals consumption must be attributed to either relative price changes or to shifts in tastes away from cereals. Moreover, NSS results during the 1990s show overall calorie-protein intake falling but not the cereals share in food expenditure. Unless incomes fell, this would imply a voluntary cut in nutrient intake. It has been suggested that energy requirements have declined as result of lower activity. Also that the sense of well-being has shifted from food to non-food even at low levels of food intake. These ideas are difficult to model adequately.   

Next