There are important applications in which due to known scientific constraints, the regression line must also go through origin (i.e. the intercept must be zero). In other words, the model should read

This model is often called the regression through the origin model. Assuming that

i

i's are independent with distributionN(0,

2

)

N(0,2).