Regression model

[Irene Guzman]

Hello everyone! I recently joined the forum to get some answers to complete my thesis work.
My thesis wants to study the impact of variables on fuel prices. The model has several independent variables and fuel price as the y variable.
the data I have are monthly for the last 20 years.
After setting to stata the "regress" command, from which I get a high R^2, I determine the error part which, however, does not have a normal distribution.
I think it is a nonlinear multiple regression model, but other than OLS, I would not know which regression to follow.
At the moment I have a basic knowledge of Stata, and Econometrics, so I do not know what I should do in this case to get a model that is considered valid.
If anyone has any suggestions I would be grateful!

[Carlo Lazzaro]

Irene:
welcome to this forum.
First, please post your queries following the FAQ recommenadtions.
If the residual are heteroskedastic, you can invoke the -robust- option for standard errors.
An OLS is linear in the parameters, not in the variables. Therefore, you can have, say, BetaX² while you cannot have Beta²X.

[Jeff Wooldridge]

Using y = log(fuel price) is a good start — for many reasons.