Household Nutrient Demand: Use of Characteristics Theory and a Common Attribute Model

Abstract

A characteristics model, which assumes goods generate a common set of attributes but no unique attribute, is described. The model yielded two equations which were estimated. One was a set of hedonic price equations in which the price paid for each food purchased was a function of imputed attribute prices. This set of equations was estimated at the household level. Nutrient demand equations were estimated across households. Imputed prices, income, and household characteristics including location, size, education, age distribution, and race affected nutrient demand levels.

Quantification of a diet's value using hedonic analysis: Integrating a health aspect

“…Further, in order to avoid multicollinearity among the explanatory variables, the selection of vitamin C and calcium is based on low partial correlations (i.e. correlation coefficient B0.8) between these nutrients and calories (see also Eastwood et al, 1986). Carbohydrates, proteins and fats are highly correlated with calories.…”

“…In line with Eastwood et al (1986), we abstract from unique characteristics and consider only those common to several food products. Compared with Lancaster's original contribution, the most important extension of the CGCM is that it works with common, not necessarily linear, production functions and it allows for negative marginal utility from some or one characteristic provided that the sum of the marginal utilities of all characteristics remains positive.…”

“…As pointed out by Eastwood et al (1986), the values of utility that the consumer derives from HFD or C 1 , . .…”

“…Because an individual's consumption of each characteristic depends on the quantities of food items purchased (q i ), and considering the restrictions in Equations (3) and (4), the utility level an individual obtains from food items is a function of characteristics, quantities of food items and the consumption of other goods (q 0 ) (cf. Eastwood et al, 1986). Maximising this utility subject to a budget constraint with prices p i and income Y results in: max q 1;...; q n U 0U (HFD(q 1 ; .…”

“…For each p i , Equation (6) is multiplied by the respective quantity, q i , and summed over all n. Subsequently, a consumer's expenditure for his/her food basket is divided by the total quantity (Q) to obtain the unit price of his/ her consumption bundle (Shi, 1992). We hence estimate implicit prices for nutrients not within product categories, but for the entire food basket as in Eastwood et al (1986).…”