Demand Estimation

There are three methods for estimating a demand function

1. Consumer interviews
2. Market experiments
3. Regression analysis

A discussion of each of them follows:

Under this method, consumers are interviewed with regard to their consumption habits. Interviews could be conducted on the census basis or sample basis. Under the former, all past and prospective consumers are interviewed, while under the latter, only a subset of them, called the sample, are interviewed. The interviews could be planned orally or through pre designed questionnaire, depending upon the complexities of the problem. These interviews, called surveys, aim at obtaining the relevant information on a variety of variables, useful for estimating the demand function for the product under study.

Market experiments provide an alternative method to estimate the demand function. It has two versions; actual and simulated. Under the actual experiment, shops are opened in different localities (places) and then consumers’ reactions are observed and recorded. Different localities would include consumers with varying levels of income, caste and religion, sex, age group, tastes and preferences etc. Further during the experiments, various could be tried to elicit consumers’ reaction to price changes. If such an exercise is carried out with sufficient care with regard to the sample of locations and probable prices, the researcher should have no difficulty in coming out with a demand function, indicating quantities that consumers would demand at various levels of incomes, prices, and other relevant variables in the function.

The market simulation method, also called consumer clinic or laboratory experiment technique, involves providing token money to a set of consumers and asking them to shop around in simulated market. The prices of various goods, their quality, packaging etc vary during the experiment to observe consumers’ reactions to such changes. This generates information which could be sufficient to estimate the demand function.

The most used method for demand estimation followed by economists is the regression method. The method involves four steps:

1.Identification of variables which influences the demand for the good whose function is under estimation.
2.Collection of historical data on all the relevant variables
3.Choosing an appropriate from for the function.
4.Estimation of the function

Speci0fication of casual variables comes from the underlying economic theory of demand. For example, if we were to estimate the demand function for groundnut oil in India, the relevant causal variables would be national income at constant pries, price of groundnut oil, prices of vanaspati ghee and pure ghee (substitute items) and prices of eggs, fish, meat , gram flour and vegetables (complementary items). By a prior reasoning, demand for groundnut oil might be affected by several other variables also, such as consumers’ tastes and preferences, distribution of population into rich and poor or between South Indian and North Indian and consumers’ expectations about future price of ground nut oil. However, one should note that a model is a simplified version of a true structure, and the model builder faces a lot of constraints such as availability of data costs of their collection and time constraints within which one would like to complete the work demand of demand estimation. For these reasons, a model builder might be content with the important causal variables only.