Model in exp-logs

Hi everyone,

I am trying to enter the model into Dynare by exp-logs method as it is mentioned in the Dynare manual, through “exp(repeatedvariable)” for each variable, and “repeating a letter for a variable means log of that variable” but Dynare gives error which basically saying that it does not recognize the repeated variable (say yy) as the log of the variable (y)! Is this a problem with the new version or there is something else wrong?!!

My model is in a log-linearized form and I would assume it should be entered into Dynare through the exp-logs method (starting with “model” for the model section) rather than** linear model **(starting with “model(linear)” for the model section ), is that right?!!

For example if the equation let say is x=y/T then the log-linearized form becomes x= y-T and I think I need to enter that into Dynare like exp(xx)= exp(yy)-exp(TT), am I right?


No. You perform a variable substitution. By writing X=exp(x), the lowercase x is actually log(X) as exp(x)=exp(log(X))=X. Doing a standard linearization with respect to lowercase x is thus effective doing a log-linearization for uppercase X. So if you use lowercase x, you actually mean log(X).

x=y/T becomes exp(x)=exp(y)/exp(T). If you already linearize it yourself, there is no point in performing a variable substitution.

See also [Linear/log-linear models and order of Taylor series approx)