I'm building a simulation with an a priori elasticity estimate, and building a simulation to see how many samples I need until I converge to the correct elasticity estimate based on that simulated data. How do I calculate the elasticity of the simulated data?

Some more detail: Let's assume a priori elasticity of -2. I randomly generate a dataset where each "person" has 10% chance of buying the product. Each person gets a random value between .95 and 1.05 that changes the price for them (let's call this p). Then I calculate how likely each person is to buy the product with 10% * p-2 = z. Now I randomly generate whether this person buys the product with IF(rand()*z>0.9,1,0).

Now I have a column of price change, original expected probability of buying (10%) for everyone, and a column of 1's and 0's representing which people bought the product.

Now, I want to know how many "people" I need to simulate before my data converges to the population elasticity (or close to it) with some level of certainty.

I expect the number of samples required for convergence to be related to the width of the price change factor. That is, if p is [0,2] it would require less samples than if p is [0.95,1.05] or [.99,1.01].

  • $\begingroup$ You want to recover own-price elasticity of demand from a binomial choice model? $\endgroup$ – BKay Apr 9 '18 at 20:49
  • $\begingroup$ @BKay yes, exactly! Specifically, I want to know how many simulated samples to achieve a specific standard error for a given width of random price change factor. $\endgroup$ – Caleb Apr 11 '18 at 0:40

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.