[Econometrics] Why is a log dependent variable used in linear regression?

When DV and IV are in a nonlinear relationship, it is hard to analyze using OLS(Ordinary Least Square) regression method.

 

* By using a log DV, the relationship between DV and IN can be presented in a linear line. It is important to remember that the main reason for using the log of DV is to impose a constant percentage effect of IV on DV. 

 

 

< Summary of Functional Forms Involving Logarithms>

In the log-level model, 100β1 is called semi-elasticity of y with respect to x. In the log-log model, β1 is the elasticity of y with respect to x.