# In the case of E_{t-1}

Dear colleagues,

I have a question for the case of E_{t-1} when the other equations are based on time t information.
For example, to estimate a LucasFuerst LP model,
one needs to include the optimality condition based at the time t-1 information as follows,

E_{t-1} U_{c,t} /P_t ] = b E_{t-1} [R_t U_{c,t+1} / P_{t+1} ].

How can I let dynare implement this?
OR Is there any circumventing way?

Jiho

There are two elements:

1. if the equation is conditional to the information in t-1, it must be setting one of the variables before knowing the innovations of period t. This variable should belong to period t-1, because, in Dynare, the variables are tagged with the period in which they are decided upon. In summary, after this change, at least one of the variables should be in t-1
2. Once, this is done, you can just shift forward in time the equation: (add one period to each variable). It is then conditional to the information in period t and you can write it as usual in Dynare

Best,

Michel

Dear Michel,

When using LP model, I regard next period deposit (D_{t+1}) as a control.
Thus, in Dynare, I changed the household’s intertemporal condition from

1/(C*P) = beta * R /(C(+1)*P(+1)*m); in standard CIA model

to

1/(C(+1)*P(+1)) = beta * R(t+1) /(C(+2)*P(+2)*m(+1)); in LP model.

But, I received the following message.

“There are 7 eigenvalue(s) larger than 1 in modulus
for 8 forward-looking variable(s).”

I do not know why. I think that the other conditions are not affected by this change of information structure.