Many students memorise the standard answer, that is, when the estimated data is not within range, we are extrapolating so it is not reliable. But why is it unreliable!
Consider an example of monthly expenditure on monthly income, if we have data of 100 people with monthly income ranging from 
10000 and their appropriate expenditure, can we accurately use that regression line to estimate the expenditure of Bill Gates, or Warren Buffet.
Hopefully this gives a bit of insights. 🙂
