The results have reinforced the hypothesis of a significant diversification in the dietary pattern of households in recent years and has mound stark differences in the consumption pattern across different income quartiles. The food demand behavior has been explained using a set of demand elasticity corresponding to major food commodities. The demand elasticity have been estimated using multi-stage budgeting with SQUIDS model and another alternative model, FACES. The study has revealed that the estimated Income elasticity vary across Income classes and are lowest for cereals group and highest for horticultural and livestock products.
The analysis of price and Income effects based on the estimated demand system has suggested that with Increase In food price Inflation, he demand for staple food (rice, wheat and sugar) may not be affected adversely but, that of high-value food commodities is likely to be affected negatively. Therefore, the study has cautioned that if inflation in food prices remains unabated for an extended period, there is the possibility of reversal of the trend of diversification and that of consumers returning to cereal-dominated diet, thus accentuating under- nourishment.
Key words: Food demand, Demand elasticity, SQUIDS model, FACES model, Household food demand JELL Classification: IQ 1, Q Introduction One of the conspicuous outcomes of the economic development India has experienced In recent years Is a marked change In the dietary pattern of Its population. Several studies have shown dietary diversification of Indians towards the high-value food commodities such as milk, meat, fruits, fish, processed food products, etc. And away from the traditional cereals-dominated food basket (Kumar et al. , 2006; 2007).
Rapid arbitration, increased disposable incomes of households, availability of a larger variety of food commodities in the market * Author for correspondence, Email: pkumariari@gmail. Com S This paper is drawn from the research work undertaken as a part of the studies entitled, ‘Befouls and the Poor funded by International Food Policy Research Institute (PRI), Washington, DC, USA and ‘Developing a Decision Support System for Agricultural Commodity Market Outlook’ (NAIF-subproject at NCAA) and growl food processing facilities In the country are some of the predominant factors behind this shift.
Therefore, an analysis of food consumption pattern and its of agricultural products to attain food security in the country. This study is an attempt towards this direction, with focus on the changes in food consumption tatter of Indian households and estimation of the demand parameters of major food commodities. A better understanding of demand elasticity helps to predict future demand of food products under different scenarios of prices and income and could prove worthy for the policy planners on important policy decisions.
The major food commodities included in the present analysis are: cereals, pulses, edible oils, fruits and vegetables, milk, sugar, meat, fish and eggs, as they constitute more than 95 per cent of the total food consumed by the Indian households. The food demand elasticity have been 2 Agricultural Economics Research Review Volt. 24 January-June 2011 estimated using two alternative methodological tools, namely Quadratic Almost Ideal Demand System (SQUIDS) model and Food Characteristics Demand System (FACES) model to enable a comparative as well as a realistic estimation.
The Demand Model For estimating the price and income elasticity of demand for various food commodities, a number of demand models are available. The recent demand studies are centered on complete demand systems which take into account mutual interdependence of a large number of commodities in the budgetary allocations of the consumer. The functional form used in the demand study affects demand elasticity estimates. There are two important requirements for the functional form that are used to estimate income elasticity of food demand.
First, these should be flexible enough to allow income elasticity to differ cross income groups, as the income elasticity of food demand generally fall with rise in income. Second, the functional form should account even if a household has zero consumption of a particular food commodity, since dropping these households from the sample could lead to estimation bias. Linear Expenditure System (Stone, 1954), ND Almost Ideal Demand System (AIDS) (Denton and Mueller, 1980) are the demand models that have received considerable attention among the economists.
These models are generally used for estimating demand equations for a group of commodities and not for commodities at a disaggregate level. Also, these models do not allow increasing or decreasing income elasticity. The normalized quadratic demand system (NUNS) and transcendental logarithmic demand system (TLD), suggested by Swampy and Binger’s (1983), are the models which satisfy all general restrictions of demand theory and also allow the estimation of cross price elasticity thin a group of close substitutes or complements, and do not assume the additive condition.
These models also include linear and squared income terms which allow more flexibility in the response of consumer items to changes in income. The framework, an extended version of AIDS model, is a modified version of the earlier model as it gives away the assumption of linearity in the expenditure function and also accounts for the zero consumption influence while estimating the income elasticity of demand. The modified model assumes that there is a non-linear relationship between income and consumption. Following Blunder et al. 1993), Dye 2000) and Kumar and Dye (2004), the specific functional forms used in the two stages have been discussed in the subsequent section. Data and Methodology The unit level data on dietary pattern and consumer expenditures collected by National Sample Survey Organization (EONS) were used for this study. The household data collected under major rounds of National Sample Survey (INS) covering the years 1983, 198788, 1993-94, 1999-00 and 2004-05 pertaining to 38th, 43rd, 50th, 55th and 61 rounds, respectively were used.
The data referred to the average per capita consumption of all the food and non-food commodities in the sample households. The per capita expenditure was considered as a proxy for income, and therefore, these have been used interchangeably in the study. The sample households were categorized into four expenditure/ income groups. These were; very poor, moderately poor, non-poor lower and non-poor higher (Figure 1).
The Very poor’ class comprised households which have income level below 75 per cent of the poverty line (PL), between 75 per cent of PL to PL were defined as ‘moderately poor’, between PL and 1 50 per cent of PL were grouped as ‘Non-poor lower’ (middle income group) and households having per capita income above 150 per cent of PL were categorized as Non-poor higher’ (high income group). The poverty line for different states corresponding to various INS rounds was used in the study. Figure 1 . Categories of four income groups Kumar et al. Estimation of Demand Elasticity for Food Commodities in India 3 Multi-stage Budgeting Framework with SQUIDS Model A multi-stage (two-stage) budgeting framework is used to model the consumption behavior of households. In the first stage, the model captures household decisions on how much of its total income (expenditure) is to be allocated for food consumption, conditional on consumption of non-food commodities and household as well as demographic heartsickness. In the second stage, the household’s allocation of total food expenditure across different items/groups, biz. Reels (rice, wheat, coarse cereals), pulses, milk, edible oils, vegetables, fruits, meat, fish & eggs, sugar, and other foods is modeled. The specific functional forms used in the two stages are as follows: Stage 1: Food Expenditure Function where, Pi is the price of tit item/group; I is the Stone price index; and Urban is a capita food expenditure (F) as: In (HI) The parameters of the model (a’, b], c’, did and elk) were estimated by imposing the ingenuity (degree zero in prices), symmetry (cross price effects are same across the commodities), and adding up (all the budgetary shares add up to one) restrictions.
The following restrictions are econometrical imposed: Homogeneity: b In (F) -a + yell In (If)+Y In (Pan)+; In (Y)+EBBS … (1) Symmetry: big = b J , Call CIO = CO CO … =CNN where, F is the per capita food expenditure; Y is the per capita total expenditure (income); P f is the household-specific price index for food; and Pan is price index of non-food. The socio-demographic and conditioning variables (vector Z) include education, family size, and urban dummy.
The parameter varies as follows: = + Pl In (Y) Equation (1) was estimated by the ordinary least squares (OILS) method, and homogeneity of degree zero in prices and income was imposed by restricting yell + Y + + 2 Pl In (Y) = O at the sample mean of In (Y). Stage 2: Quadratic-AIDS (SQUIDS) Adding up: The homogeneity and symmetry restrictions are imposed at sample mean.
Adding up restriction is imposed while computing the parameters of the last equation of the model, which is not included in the estimation. Given the quadratic specification of the demand system, a test of symmetry additionally requires that the ratio of the efficient on the food expenditure and the square terms in food expenditure be the same for all items/groups (Blunder et al. , 1993). The predicted value of food expenditure obtained from stage 1 has been used as the explanatory variable.
The income and price elasticity can easily be computed as follows: Food item / group income elasticity: IQ = (CIO + chic In [F]/ w’) + 1 Uncompensated price elasticity In stage 2 of the analysis, the quadratic extension to Denton and Mulberry’s (1980) almost ideal model (SQUIDS) for food demand system was used. This model is quite popular and was adopted recently by Namesakes and Ray (1999) for India food model, by Dye (2000) for fish demand model of Bangladesh, by Kumar and Dye (2004) for fish demand model of India, by Metal (2006; 2007) for cereals, by Shinto and Matter (2006) for spices and by Dye et al. 2008) for fish demand in Asia. The specific functional form of this model for the tit items/groups is as follows: … (2) where, ski is Crookneck delta, which takes the value one for own-price elasticity and zero for cross-price elasticity; and WI is the share of the tit item/group used as a weight in constructing Stone’s price index. Once the expenditure and uncompensated price elasticity are estimated, the compensated own and cross- price 4 elasticity are computed using the Sluts equation in elasticity form; I. E. Where, elasticity. S the compensated (Hickman) price product of expenditure elasticity of individual item/group and food expenditure elasticity with respect to total income (my): ninny = in x NY x IQ where, = Probability that positive consumption of the ‘tit item occurs. It is estimated as the proportion of households consuming the tit item in the sample households during the survey. Food Characteristic Demand System (FACES) In addition to econometric models, Bouts ND Haddam (1992) suggested a non econometric model based on demand characteristics known as food characteristic demand system (FACES).
Several studies have shown that demand elasticity can vary widely across income groups (see Alderman, 1986, for a review) and regions as production environments and tastes change. FACES can be easily used for those households who spend a high proportion of their total income on food, and a large share of their total food expenditure on a low-cost-calorie staple, to avoid going hungry. How will such a low-income household react if the price of this low-cost-calorie staple (say wheat) falls? The household could afford to substitute a part of this staple with some preferred staple (say rice) without going hungry.
A drawback of such a decision, however, is that the diet would still consist almost entirely of bland cereals. The household may instead prefer to continue eating nearly the same amount of wheat as before to meet its energy requirements, and may supplement the monotonous diet with some low-cost-per-kilo meat. If latter is the case, I. E. If consumption of non-staple diet is more important for the household than the superior taste of rice, then the uncompensated own-price elasticity for wheat may be (negative but) very low in absolute value.
Now suppose that a lower price of wheat in the above example prevails and the income of the household goes up on a regular basis. Then, the household may afford a substantial amount of some preferred food item (say meat) in the diet, and may even afford consumption of a relatively superior quality of rice. Suppose that the price of wheat rises (although still remains below the rice price), the household being economically stronger now, does not have to worry about the specter of hunger (a low energy intake), despite increase in wheat price.
The household may even plan to substitute a substantial amount of wheat with rice. Since the household pays more for cereals now, the total consumption of both cereals and meat may be reduced marginally. However, although the total utility goes down, the marginal utilities of “energy’ (calorie intake) and “variety’ (non-staple food consumption) have declined enough so that the loss in utility is minimum by sacrificing some calorie intake from the non-staple food consumption, but recouping some utility from the superior “taste” of his chided commodity.
