Application Exercise - Choice Models
These exercises are provided as optional practice only. There is no need to submit your work.
Problem 1. A greengrocer is selling apples, bananas, and oranges in Leeds, UK. They surveyed a random sample of their customers and estimated a multinomial logit model. The estimated parameters and attribute values are given below. Calculate the choice probability for each alternative assuming a multinomial logit error term.
| Alternative | Price | Price Parameter | Unusual Shape (Y/N) | Unusual Shape Parameter | Ripe (Y/N) | Ripeness Parameter | Alternative Specific Constant |
|---|---|---|---|---|---|---|---|
| Apple | 0.84 | -0.25 | TRUE | -0.1 | TRUE | 0.3 | 0 |
| Banana | 0.17 | -0.25 | FALSE | -0.1 | FALSE | 0.3 | 0.7 |
| Orange | 0.25 | -0.25 | FALSE | -0.1 | TRUE | 0.3 | -0.9 |
Problem 2. There is a banana shortage and the price increases by 10%.
What would happen to the selection probabilities for each alternative? Compare the results using a) direct and cross elasticity equations with the result b) the change in the price of bananas and re-calculate the choice probabilities.
Why might the two results differ?