Nutritional Deprivation at the Household Level in Rural India:Causes and Concerns

The aim of the paper is to quantify the proportion of undernourished households in rural India without relying on any particular calorie cut-off point. For that, mean RDA (Recommended Dietary Allowance) has been estimated at the household level, after adjusting for age and gender distribution of the sedentary household members. The two NSS rounds that pertain to the years 2004-05 and 2011-12 are used here. The results confirmed that it was among the lowest expenditure group that seems to have reported the highest increase in RDA between 2004-05 and 2011-12. An investigation of the determinants of calorie deprivation leads us to the finding that poor ST and OBC households, regular wage and self-employed in non-agriculture and Christians have the greater probability of being calorie deprived.


Introduction
With the celebrated reduction of rural poverty, a commensurate fall in the level of undernutrition may not come about in reality. The growing divergence between hunger and poverty and the public policy inertia with regard to undernutrition (Basu and Das, 2014;Rao, 2016) has contributed to a number of studies in the area. However, none of them are satisfactory in that it often revolves around arbitrary calorie norms and its likely correlates.
As is well-established, undernutrition stems from imbalanced diets and from the perceived deficiencies of macro and micro nutrients. Since low socio-economic status of a household is inextricably linked to undernutrition, an understanding of the level and proximate causes of undernutrition forms the bedrock of this paper. The recent undernutrition levels in rural States have been improved but the temporal improvement is far from satisfactory. For example, the chronic energy deficiency of adults (proportion of adults with body mass index less than 18.5) which was at 40 per cent in 2000-01, fell to 35 per cent in 2011-12 (Radhakrishna, 2005Dreze, 2007, NNMB, 2012. With the significant strides in poverty reduction, a modest improvement was only discernible in case of the outcome indicator, which was reflected by the body mass index of the population. Hence, it is necessary to look at the input indicator that accommodates the nutritional intake of the household; the deficiency of it surmounts to the nutritional deprivation.
There is a consensus among the studies that poverty ratios in rural areas seem to have declined. The divergence between hunger and poverty suggests that there will be some problems either with the measurement of poverty or with the measurement of undernutrition or with both the measures. To get around the issues relating to the measurement of undernutrition, we have estimated the calorie requirements without relying on any calorie thresholds and then arrived at the nutritional deprivation of households. Our estimation has largely been based on the age and gender adjusted requirements of nutrients by the rural population. Such an innovative approach is followed on account of the fact that food and nutritional intake vary with the age and gender differences due to the differences in activity status, metabolism rates and other physiological factors. Apart from this, there is lingering and unequal distribution of food to the old and girl child within the family.
The monthly per capita consumption expenditure is adopted here as a proxy for the economic status of the household. The nutritional status of the household is also conditioned by household types, occupation categories, religious factors, among others.

Rural Transformation in India
Rural India has dramatically changed since the mid-2000s, enough to accommodate the urban settings, which has now been christened as 'Rurban' (Gupta, 2015).  Kumar et al (2011) found that with one per cent increase in the share of rural non-farm employment, the rural poverty would be reduced by 0.5 per cent. In another study, Lanjouw and Shariff (2004) pointed out that poverty rates can also be reduced by the growth of non-farm sector and its trickle-down effects on agricultural wage rates. These studies have well-recognised the potential role of non-farm sector in generating employment opportunities in the rural sector and thereby increasing food and nutritional security of the country.

Brief Description of Database
This study makes use of the NSS unit record data that pertains to 2004-05 and 2011-12 years. The unit record data give a sample of approximately 80,000 and 60,000 rural households, respectively, in these two rounds. A stratified multi-stage design was used for these surveys. The First Stage Units (FSUs) comprise the 2001 Census villages in the rural sector and Urban Frame Survey (UFS) blocks in the urban sector. The Ultimate Stage Units (USUs) are the households in both the sectors. Within each district of a State/UT, two basic strata were formed: (1) rural stratum comprising all rural areas of the district and (2) urban stratum comprising all urban areas of the district.
The unit record data provide information relating to household and demographic features, in addition to household production and consumption of various food items in terms of quantity and expenditure. Broadly speaking, there are five different sources of household consumer expenditure which can be met via purchase, home-grown stock, receipt in exchange of goods and services, transfer receipts including gifts, loans and charities and the remaining expenditure of the household is met by way of free collection.
The NSS unit record data report the socioeconomic and demographic features of all sample households under review. However, the quantity figures of food items of some households are found to be missing or unreported. At the same time, household characteristics are given. For the sake of uniformity, we have dropped these households for whom it is not possible to estimate the nutrients consumption. Thus, ten households that belong to the States of Andhra Pradesh, three from Maharashtra and one household from Kerala have been dropped in 2004-05. Similarly, two households that belong to Madhya Pradesh and Arunachal Pradesh have been excluded in 2011-12. By doing this elimination procedure, we tried to avert the problem of missing calories at its best.
For estimating the macro nutrients, the quantity figures of food items of each household given in the NSS unit record data are multiplied by the nutrient contents. This is further divided by household size and by 30 in order to get the daily consumption of nutrients per person. The information on the nutrient content of each food item is sourced from the publications of the National Institute of Nutrition (Gopalan et al, 2000 water quality and so on (Himanshu, 2010).

State-wise Trends in MPCE and Calorie Intake:
The Monthly Per Capita Consumption Expenditure (MPCE) is the total household monthly consumption expenditure adjusted for household size. In unit level data, MPCE is given in paisa and some adjustments have to be made. Table 1 brings out the median income of households that belong to rural areas. The use of median suggests that around 50 per cent of rural households are lower than the average MPCE and close to 50 per cent of rural households are higher than the average MPCE. Thus, the median divides the total households into two equal parts. per cent), Kerala (8.9 per cent) and Haryana (9.9 per cent) in this regard. In most of the backward States, the income growth was not substantial to help improve the nutritional intake in those States.
Having these issues, Rajasthan and West Bengal have made headway in terms of income growth.
In these States, around 70 per cent of rural households had an average income of above 1 000. Income has contributed positively to the calorie consumption in most of the leading States. However, this was not the case with backward States, where calorie intake has registered a decline in the States of Rajasthan, Madhya Pradesh, Bihar and Uttar Pradesh.  Any deprivation measurement can be either unidimensional or multidimensional. The unidimensional measures based on FGT indices are more common in the poverty and nutrition literature. The FGT indices are based on the monotonicity and transfer principles; the incidence of undernutrition violates the monotonicity axiom but it is captured by the depth of undernutrition, let alone the transfer principle. Both the axioms are incorporated when the severity of undernutrition is used. The monotonicity axiom looks at the increase in undernutrition as a result of lower calorie achievement levels. On the other hand, the transfer axiom proposes a decrease in undernutrition levels when food is transferred from the richer household to the poorer household (Alkire and Foster, 2011).
The Foster-Greer Thorbeck indices (1984) that measure the incidence, depth and severity of undernutrition are computed here for all macro nutrients. These indices are the most reliable ones when the deprivation is unidimensional in nature. The FGT index can be specified as follows: Where Q R is the minimum required calories (RDA), Q E = estimated calorie intake of the household and n is the total number of households. The sigma symbol refers to the summation of all households which consume less than minimum requirement.
When α = 0, the formula shows the Head Count Index which represents the proportion of households whose calorie consumption fall below the minimum requirement. This simple measure discards the depth of undernourishment. When α=1, the Proportionate Gap Index can be calculated. It measures the average distance from the minimum requirement, but it is insensitive to the distribution among the undernourished. When α =2, the FGT2 index can be calculated. The index takes into account inequality among the undernourished and shows the severity of undernourishment by assigning greater weights to those households which are far from the minimum required calories. Thus, FGT2 index incorporates the idea 'relative deprivation' , as measured by outcome inequality among the deprived households. Table 3, there is a significant prevalence of calorie undernutrition which ranges from 32 to 60 per cent during 2011-12. The prevalence of calorie deprivation was more or less stagnant at 45 per cent. On the other hand, fat deprivation has declined more precipitously than the protein deprivation. The depth of calorie deprivation was up from 12 to 37 per cent during 2004-05 to 2011-12. At the same time, the depth of deprivation of both proteins and fats declined and it was more pronounced in respect of fats. As far as the severity of undernutrition is concerned, the severity of calories and proteins increased over the period 2004-12.

State-level Estimates:
The State-level estimates exhibit an interesting pattern; the status of Southern and Western States is dubious, given their higher head count ratio of nutrients. Thus, Andhra Pradesh, Kerala, Karnataka, Tamil Nadu, Gujarat and Maharashtra are historically notorious for calorie deprivation and this pattern is consistent with the studies by Sharma (2015), Jha and Gaiha (2003) and Meenakshi and Vishwanathan (2003). The use of calorie thresholds and age-gender adjusted nutritional norms do not produce contrasting results when we look at the case of some leading States, already noted above. These leading States due to better infrastructure, better healthcare and good sanitation facilities have improved the ways of absorption of nutritional intake. Furthermore, how recall method is able to trace out the consumption of home-away cooked meals, processed foods and beverages in its totality is doubtful in the case of developed States. Hence, the nutritional deprivation in these States does not matter for the policy makers.
The prevalence of calorie deprivation was as low as 32 per cent in Punjab and Rajasthan. The reasons are yet to be explored for the lower calorie deprivation in Rajasthan, despite the poor health status and widespread illiteracy of the State (Sagar, 2010).
As for the head-count index, calorie deprivation was higher in the States of Gujarat (60 per cent), closely followed by Tamil Nadu States. Another interesting pattern that emerges here is fat deprivation, declined more sharply than that of protein deprivation.
When we delve deeper into the Statewise comparison, it can be seen that fat deprivation is higher at 3 per cent in Kerala, Gujarat and Tamil Nadu. Overall, fat deprivation has declined in all the States. This matches with the fact that fat consumption has not ratified any declining trend for rural India as the per capita consumption of edible oils steadily improved (Deaton and Dreze, 2009; Gupta, 2012).  When 2004-05 and 2011-12 are considered, the depth of calorie deprivation has increased in all the States without exception (Table 4). In some States namely, Haryana, Assam, Odisha, Rajasthan, Uttar Pradesh and West Bengal, the depth of calorie deprivation has increased by four times between the two periods.
There are some gainers and losers if one looks at the depth of protein deprivation. The depth of protein deprivation increased from 2 to 5 per cent over the period. In other States such as Haryana, Punjab, Bihar, Madhya Pradesh, Odisha, Rajasthan and Uttar Pradesh, the depth of protein deprivation has marginally increased. On the other hand, the depth of fat deprivation has virtually declined in all the States and it is not important to be considered.   A glance at Table 5 shows that the severity of calorie deficiency was as high as 56 and 49 per cent in Odisha and Assam, respectively. In Punjab, Haryana and Rajasthan, the severity of deprivation in terms of all macro nutrients was found to be low.
A marginal improvement in the severity of protein deprivation was recorded in the States of Haryana, Maharashtra, Madhya Pradesh and Rajasthan. In all backward States, except Madhya Pradesh and Rajasthan, the severity of fat deprivation has considerably increased between 2004-05 and 2011-12. In 2011-12, the severity of fat deprivation was much higher at 18 and 13 per cent in Odisha and Assam, respectively.
When the severity of deprivation is considered, it marks an impressive performance of the Southern States over time. Except in Kerala, the severity of fat deprivation was as high as 14  deprivation has halved between the periods while a modest improvement was perceptible in case of Rajasthan.

Income as a Way Out of Calorie Deprivation:
How much income is needed for the rural households to escape from the label of being calorie deprived? This section tries to answer this question, by comparing the average income of calorie deprived and calorie non-deprived households. If the recommended dietary allowance of calories is greater than that of the derived calorie intake from the quantities of food items, then the household will sink into the situation of being calorie deprived. On the other hand, if the recommended dietary allowance of calories is less than or equal to the daily calorie intake of households, then these households can be counted as calorie non-deprived. States such as Assam and Bihar and Odisha. The relationship between income and calorie deprivation is dubious in these States. The highincome growth period of 2011-12 suggests that income-augmenting policies have worked better in the States of Punjab, Haryana and Rajasthan.

Factors Influencing Nutritional Deprivation:
The previous studies have estimated the calorie deprivation of rural India by accommodating a set of norms such as 1800, 2100, 2200, 2400 and 2700 (Meenakshi and Viswanathan, 2003;Suryanarayana and Silva, 2008;Gupta and Mishra, 2013;Mishra, 2010). These studies are largely misleading owing to the inclusion of arbitrary norms. In an exceptional study by Sharma (2015) the RDA has been calculated after adjusting for age, gender and occupation of the households. As the RDA requirements of different households is different, the present study also offers an errorfree approach in this regard by working out the age, gender and occupation adjusted calorie requirements and finally measures the calorie deprivation if the RDA surpasses the calorie intake of the household. The methodological difference of the work lies in the inclusion of calories (proteins/fats) derived from those food items such as pan, ganga, toddy, country/foreign liquor, beer and other intoxicants. Our presentiment is that the exclusion of these unhealthy food items is likely to intensify the nutrient deprivation in rural India.
The binary logistic regression method has been used to study the income-calorie nexus. The CD i is the dependent variable and monthly per capita consumption expenditure, household size and land ownership are the independent variables of interest. The CD i is equal to one if the household is calorie deprived and otherwise zero is recorded.
To get an insight into the probability of being calorie deprived across socio-economic and demographic groups, a logit model has been fitted as: ln(π i /1-π i ) = α+ βX i +ε i ---------- (1) Where (π/1-π) is called the odds ratio. The estimated probability (π) is obtained as follows: Where x is the predictor variable and e is the base of natural logarithm with a value of 2.7183.
In case more than one explanatory variable is included, then the model becomes:

Rationale Behind Selection of Variables
Income: The impact of monthly per capita consumption expenditure, a proxy for income, is well-documented in literature. As MPCE goes up, calorie deprivation also decreases. However, there are chances that when income increases, households may not apportion all of their income on calories. As a result, the impact of income on calories would be less influential.
As given in Table 9, the logit model shows that below median households are severely calorie-deprived, with the odd ratio of 1.68. It implies that income effect works in the opposite direction on calorie deprivation.
Household Size: As is well-known, the larger households often utilise price discounts when purchased in bulk quantities (Meenakshi et al, 2000). The relationship between household size Journal of Rural Development, Vol. 37, No. 1, January -March : 2018 and calorie consumption can be either positive (Kaicker and Gaiha, 2013) or negative (Gaiha et al, 2010). This could be due to the higher (lower) proportion of adults in the household relative to the dependent population such as the aged and children.
In both small and medium-sized households, calorie deprivation has significantly gone down. The opposite results hold for medium-sized households which were perceptible in case of protein deprivation.
Land Ownership: Land ownership exerts a negative influence on calorie deprivation. The coefficient for land ownership is significant but negative. It implies that when a household has access to land, its calorie deprivation decreases. This goes in line with the argument that selfproduction of cereals contributes to an increment in calorie intake (Basu and Basole, 2012). Table 9 shows that the below median MPCE households are at a greater risk of calorie deprivation. It displayed the odd ratio of 1.68.
Among the social groups, ST households followed by OBC households face more calorie deprivation than SC households. The households which belong to regular wage and self-employed in non-agriculture are the most deprived categories. Though calorie deprivation has declined among casual labour in agriculture/non-agriculture, the results were not significant. Among the religious groups, Christians face more calorie deprivation than all other religious groups for whom a decline in calorie deprivation was reported.
A quick perusal of Table 10 shows that medium-sized households are more likely to be protein deprived. Also, SC households have improved their protein intake which was not the case with calorie intake. Not surprisingly, the regular wage earning households and households which are self-employed in nonagriculture are the most calorie and protein deprived households. This could be due to the heavy job pressure, less time available for cooking and regular skipping of meals to reach the offices in time.

Concluding Remarks
The aim of the paper was to quantify the proportion of undernourished households in rural India without relying on any particular calorie cutoff point. To do so, median RDA has been estimated at the household level, after adjusting for age and gender distribution of the sedentary household members. In order to pin down the nutritional deprivation, an attempt has been made to estimate the nutrients consumption derived from the quantity figures laid out in NSS unit record data and the nutrient contents drawn from the nutritive value of Indian foods. The two NSS rounds that pertain to the years 2004-05 and 2011-12 are used here.
Across the States, it can be seen that the median RDA of calories has increased. This was the case with proteins except the fats where the increased disease burden may not induce the fat consumption any longer. Our results confirm that the argument in favour of any further reduction of calorie requirements cannot be granted by any reason.
Given the depth and severity of calorie deprivation, the stagnant incidence of calorie deprivation is not a cause for celebration. The paper also highlighted the notable income difference between calorie deprived and calorie non-deprived households in a handful of States such as Punjab, Haryana, Rajasthan, Uttar Pradesh and Bihar that points to the worsening income inequality in these States. The stagnant calorie deprivation could be due to the plausible omission of pregnant women and children and the conventional treatment of all rural households as sedentary ones. Although a number of government programmes are in operation, income growth is not sufficient for the well-being of rural households in general and for the nutritional adequacy in particular.
An investigation of the determinants of calorie deprivation leads us to the finding that poor, ST and OBC households, regular wage, self-employed in non-agriculture and Christians have the higher probability of being calorie deprived. Among the backward States, higher incidence of nutritional deprivation persists among Assam, Madhya Pradesh, Odisha and West Bengal. The legacy of higher calorie deprivation in leading States is not a cause of worry. Lower the nutritional intake, higher will be the absorption level in Southern and Western States due to their improved ways of living. This is further reflected in the lower incidence of stunting and underweight among children in these States (Mishra and Mishra, 2009).