Logit/Probit with numbers different than 0 or 1?

I am trying to find an econometric model that helps me to answer the following. I would like to measure how much the increase in the cost of a production factor (my X) shifts the supply curve of a market "A" (my Y). That is, I want to regress Y on X. I have daily data of the supply curve (the pairs Price, Quantity) and the daily cost of the factor. As a first step, I calculated a dummy variable that takes the value 1 if in day t there was a shift to the left with respect to t-1. Then I calculated the variable that measures the displacement if there was a shift (my Y). Which econometric model could I use since Logit and Probit only apply when the variables are 0 or 1, but in this case, I want to know how much the curve moves? Should I take a subsample of all the dummies that are equal to 1 and work with that? I don't think this is the best answer.

Thanks in advance!

• If you currently don't need something else from us, please upvote my answer so that your question leaves the "unanswered questions" queue. – Alecos Papadopoulos Nov 18 '20 at 14:59
• I did! Sorry for the delay. – angelavtc Nov 19 '20 at 16:41

1 Answer

I don't see how Logit/Probit models are relevant here. They are designed to estimate conditional probabilities, not what you describe.

It appears you postulate a relation

$$S = f(P, C)$$

namely supply being a function of price and cost, and where you treat price as a "control" This could be written as a linear regression model, as a first-order approximation. Why don't you think estimating that model?

• Yes, I agree. My problem was more about the way my "y" variable was modelled, because I have a lot of 0 values and the different positive values (different than 0), but I found the answer. I am going to apply a Tobit model :) – angelavtc Nov 18 '20 at 14:40