about sensitivity analysis

[luckysinger]

I want to know how the standard variance of consumption fluctuation varies when the value of beta vary from 0.9 to 0.99 by every 0.01. And I also want to know how the standard variance of consumption varies whent the value of delta vary from 0.01 to 0.5 by every 0.1.
Can some directions in the dynare help me to complete the missions above.

In other words, I want to know how the variances of some key variables vary with the changes of the deep parameters, can dynare do it with some simple directions.

[pakocica]

Dear Lucky Singer,

there are more ways how to get the results you need. The choice of the right way depends on your model and your computational facilities. The most exact way is, I think, to use some features of the Dynare preprocessor. Namely, you can use double cycle to run your model with all the combinations of the two parameters you mention. Dynare will store the results in files and you need to make a simple Matlab procedure to summarize those results for different values of the parameters.

Well, the above way is precise, but can take you some time to handle it unless you are familiar with Matlab and Dynare enough. It requires a lot of computation time unless you use a supercomputer too. An alternative option is to use at least second order approximation of your model and use some of the features build in Dynare (or develop them yourself if you don't find what you need there).

Lucky Singer, does this help? Should you need more help, specify your model and its desired output more in detail, please.

Good luck!

Pavel

[otb]

You can do a loop around stoch_simul and simulate the model for each value of your parameter. You need to get the resulting standard deviations from the mat files in oo_ and store them in a matrix as you do the loop, otherwise the results are erased in each iteration.