Videos and questions for Chapter 3a of the course "Market Analysis with Econometrics and Machine Learning" at Ulm University (taught by Sebastian Kranz)

### Choice as Utility Maximization and Random Utility Specification

Could we in principle estimate the utility function as a linear regression via ordinary least squares with a given discrete choice data set?

What will we assume for the distribution of the random utility shocks $$\varepsilon_{nj}$$ in a conditonal logit model?

### R illustration conditional logit model

If we increase the variance what happens for consumer 3's choice probabilities of both products in our example?

Which product will consumer 3 pick given the utility values U from the example in the video?

You may remember from your microeconomic classes different classifications of utility functions, like ordinal utility functions or cardinal utility functions.

What does a cardinal utility function in general represent?

### Maximum likelihood estimation

That's all for the videos and slides.

In the RTutor problem we will apply the conditional logit model for a discrete choice data set about heating systems in Californian houses. Besides just exploring the preferences and trade-offs between investment and operational cost, we will use our estimated model to make a prediction of how an investment subsidy to an environmental friendly heating system will affect resulting market shares.