Calibration for annual frequency

[cmarchionatti]

Hello. I am using a calibrated log-linearized model applied for a specific region of Brazil. However, most of its data is in annual frequency, and the literature presents most of parameters in quarterly frequency (for example, calvo lottery, Taylor Rule, among others).

How could someone transform a calvo lottery parameter, for example of 0.66 (change in prices every three quarters), to an annual basis? Or the Taylor Rule parameter of the impact of the last quarter interest on its current value into the impact of last year interest to the current value?

I noticed that comparing the moments, for example the correlation between the variables, changes drastically between the same variables but in another frequency. So it is important to make Dynare create the simulations in the same frequency as the available data.

Thank so much!

[jpfeifer]

For Calvo, you can compute the duration in years and map it to typical quarterly values. See

How to do a calbiration?

In a quarterly model with a zero inflation steady state, a four percent interest rate implies \beta\approx 0.99 a ten percent markup implies \varepsilon=11 as the gross steady markup is \frac{\epsilon}{\epsilon-1} five months average price duration is 1.6667 quarters. With \theta=0.4, the average price duration 1/(1-\theta)=1.6667 You need to make the slope of the PC equal, i.e. \frac{\varepsilon-1}{\gamma}=\frac{(1-\theta)(1-\beta\theta)}{\theta} and therefore \gamma =\frac{(\epsilon-1)\th…

For persistences, I would expect the fourth power of the quarterly persistence parameter. See e.g. Monthly parameters calibration from quarterly values - #2 by jpfeifer