## Interview Question: Binary Choices

Sample Question #95 (econometrics)

Suppose I’m interested in modeling what factors drive a stock price "up" or "down." What’s an appropriate econometric model to do this?

Advertisements
This entry was posted in Sample Qs. Bookmark the permalink.

### One Response to Interview Question: Binary Choices

1. Brett says:

ANSWER

"Up" or "down" translates into a discrete choice dependent variable problem, where the lefthand side variable in the factor regression model takes on a natural number that stands for the state of the variable. For instance, we can ignore observations where the price doesn’t move, and code "up" as 1 and "down" as 0, in which case we have a binary choice model.

To model binary choice problems, you can do:

1) simple linear regression model. A lot of candidates mistakenly believe that you cannot use the linear model for discrete choice problems. That’s not true. As long as the theory allows it (i.e., you have a "well-behaving" error term), it’s okay to use the linear model. It’s just that the linear model suffers the obvious problem that the fitted LHS value may be "out of bounds."

2) logit or probit, the standard discrete choice models. Can you explain the difference between logit and probit, and when would you choose one over the other?