A priori information: Let us consider the earlier example of consumption expenditure. We know that  β₁=  0. 13β₁,  since  β₁ =  0.75 and  β₂=  0.1. In other words, the rate of change ofcbilsumption with respect to weal& is 0.13  tines the rate of  change with respect to  income. Then the earlier regression becomes  

where X_i = Y_i  + 0.13 W_i. Once we obtain β₁, we can estimate β₂ using  this estimate.Where does one obtain such a priori information? We obtain  it either from an earlier study where the problem of multicollinearity was less severe, or alternatively  frorn'theory.  For example,  if we are estimating  a Cobb-Douglass production function and we expect constant returns  te scale  to prevail then we can impose  the restriction that β₁ + β₂ + 1.  Note that many times  this cannot be done since-the  objective of many studies is to test the theory. For instance, if the objective is  to find out if there exists constant returns to scale or not, then obviously  this is not a solution we can use.