Probability >1 ?

[Francib93]

Hi guys and thanks for your help. I'm a newbie to EViews so my question may be extremely simple.

I want to model the probability that a firm exports as a function of a set of regressors:
SIZE (Firm size, in thousand employees)
RD INT (Ratio R&D Expenditures over turnover, in percentage)
NORD (a dummy equal to 1 if the firm is located in the North of Italy and 0 otherwise)
By using a sample of 2'172 Italian firms I estimate a Linear Probability and a Probit model whose results are shown in the following tables.



(I just put the LS table since my problem is here)

The professor now ask me to compute this:

Based on the results, compute the probability that a firm exports when it is located in North, it has 5'000 employees and has an RD INT equal to 6%.

My problem is that my probability is higher than 1 and I don't understand why. Am I doing something wrong or the table is wrong? Can you help me with the computation?

Y = 0,21 + (0,018 * 0,06) + (0,059 * 5) + 0,589 = 1,09 !!

I have also troubles understanding the meaning of the coefficients. Since SIZE is in thousand of workers, if I increase by 1000 workers my EXPORT is going to increase by 0,059. (Right?)
RD_INT is a percentage so if I increase by 1% my ratio my EXPORT is going to increase by 0,01% or 1%?

Thanks a lot!!!

[startz]

You've found the problem with a linear probability model, nothing requires the probability to be between zero and one. Using probit fixes the problem.

[Francib93]

Thank you so much! I understand now.

What about the comments on the coefficients? Are they right even if they are not really meaningful?

[startz]

You have the interpretation of the size coefficient right. The other one depends on whether an increase of one percent is coded as 1.0 or as 0.01.

[Francib93]

Thanks a lot really! :D