If you have a first order condition:
E([1+R]p^t[U'[Ct+1]])=U'[Ct]
Note: p^t is some constant discount factor, U' denotes Utility partially derivated on [Ct] or [Ct+1] ) (marginal utility in time period t and time period t+1.
My professor claims that you can do this:
Divide by U'[Ct] on both sides, and rewrite to:
E[1+R]S[t+1]=1 where S[t+1]= p^t[U'[Ct+1]]/U'[Ct]
However, my intuition says that this is illegal, as U'[Ct] was not in expectations form to begin with, and he cannot just write it in, nor is it possible to remove the expectations of S[t+1] unless S[t+1]is constant, which it is not , as he calls it the stochastic discount factor.
Expectations of two variables that are not constant: E(x)E+cov(x,y)
Is it "legal" to do what he does? If so why?
E([1+R]p^t[U'[Ct+1]])=U'[Ct]
Note: p^t is some constant discount factor, U' denotes Utility partially derivated on [Ct] or [Ct+1] ) (marginal utility in time period t and time period t+1.
My professor claims that you can do this:
Divide by U'[Ct] on both sides, and rewrite to:
E[1+R]S[t+1]=1 where S[t+1]= p^t[U'[Ct+1]]/U'[Ct]
However, my intuition says that this is illegal, as U'[Ct] was not in expectations form to begin with, and he cannot just write it in, nor is it possible to remove the expectations of S[t+1] unless S[t+1]is constant, which it is not , as he calls it the stochastic discount factor.
Expectations of two variables that are not constant: E(x)E+cov(x,y)
Is it "legal" to do what he does? If so why?