I'm a student and I want to determine the demand for a product (and consequently determine what will happen if the price of the product increases). I don't have access to any businesses that can provide the data for me (price and demand at that price?

How do I estimate the demand curve or function by surveying people? What I currently have in mind is to ask people what amount of the product they'll demand at a given price and extrapolate this data for a city or a region.

Additionally, what'd be a good sample size for such a survey?

  • $\begingroup$ Is this a project? Advanced course or a dissertation? If it is a dissertation question, you should consider using simulateneous equation modelling that requires more than just quantity and price. It is hard to answer your question without knowing the context within which your results are assessed/used. please tell us more about the context and your resource limitations. $\endgroup$ – london Feb 15 '18 at 12:14
  • $\begingroup$ This is for a small project I'm doing. $\endgroup$ – WorldGov Feb 15 '18 at 15:31

If it is a small project, you may benefit from reading this

At most, you can estimate price elasticity of demand using your survey data to make predictions/forecasts. For this you need to estimate a demand function as @EconJohn suggested, but you need to include additional variables such as income, price of a substitute or complement etc. to hold them constant (i.e. ceteris paribus). Note that the demand function does not allow you to draw the downward sloping demand curve (like the one in textbooks). It is very hard, in practice, to draw actual demand curves using survey data. Even when you have the most complete data, without including the suply function in your estimation, you face so called identification problem. In short, econometric estimation of demand curve does not allow you to draw a downward sloping curve of the type you see in textbooks.

The link above has a good explanation about this point, and it can be used as a guide for your project.


If you are designing a survey to calculate demand for a product you would need to ask just 2 questions to your sample population:

1) How much of product X do you consume per month

2) How much do you pay for each unit of product X

These questions are important because it allows you to build a demand curve using simple linear regression. $$p=\beta_0+\beta_1 q+\epsilon$$

where $p$ is price and $q$ is the amount of product X.

Using this framework you can obtain appropriate data from your survey for calculating demand for a given product.

Hope this helps.

  • 3
    $\begingroup$ This should really be done using a structural model that takes into account simulataneity bias in the estimates and should include more than just $q$ and $p$ as variables. However, if the goods are supplied to the market by a monopolist that has a price-setting power, then the above equation is fine. $\endgroup$ – london Feb 15 '18 at 12:11
  • $\begingroup$ Thanks for the answer. I'm just wondering: with the two questions you've given, I'll only get one q (the amount of a product someone buys) and one p (the price he or she pays). How will this give me the demand curve? I specifically want to know what quantity will be demanded at particular price points. $\endgroup$ – WorldGov Feb 15 '18 at 15:33
  • $\begingroup$ @WorldGov if you recall how your basic demand curve looks this should be ample data. $\endgroup$ – EconJohn Feb 15 '18 at 18:36
  • $\begingroup$ -1: This is incorrect because the proposed regression suffers from "simultaneous equations bias." This is in fact the classic situation of the bias that comes up in most treatments of the topic. For example, see here encyclopedia.com/social-sciences/… or here en.wikipedia.org/wiki/Endogeneity_(econometrics) $\endgroup$ – jmbejara Mar 8 '18 at 0:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.