This project is read-only.

cholesky decomposition and stress scenario

Topics: Kalman Filtering, Models, Times Series
Sep 17, 2014 at 4:58 PM
Edited Sep 17, 2014 at 4:58 PM
Hi Mr : Jaromir

can you please explain me this syntax. it is about running stress scenario simulation with two hypothetical shocks on macroecomics variables (gdpgap- exchange rate).

The first hypothetical scenario assumes a temporary shock on the output gap. In particular, from the beginning of the forecast horizon, the GDP gap declines by 1.5 percentage points in every quarter during the first year. Given the assumption that the potential output growth is 6% (using hp filter), this hypothetical scenario is equivalent to 0% growth for the actual output during the first year. After one year, the actual GDP growth returns to its trend and grows on average by 6%.

In the second scenario, I consider a nominal shock on the GEL/USD exchange rate. For some reason, the exchange rate depreciates by 10% in the first quarter of the forecast horizon. For the rest of the periods, the exchange rate depreciates on average by 1.2%, as in the baseline case.

here the syntax introduced in matlab :
j=0; % j=1 stochastic basleline scanario, j=2 stress with GDP, j=3 stress with exchange rate
for j = 1:3 
for k = 1:Reps
    for i = n+1:n+h
       %% structural shocks simulation
       if j == 1
           eps =randn(1,10)*chol(sigma);   % stochastic baseline scenario
       elseif j ==2 
           if i<=n+4 
            v = randn(1, 10);
            v(8) = -0.8;       %GDP shock
            eps = v*chol(sigma); 
            eps = randn(1,10)*chol(sigma);  
           if i<=n+1 
            v = randn(1, 10);
            v(9) = 2.6;    %ex. rate depreciation shock
            eps = v*chol(sigma); 
            eps = randn(1,10)*chol(sigma);  
i don't understand why puting (-0.8% and 2.6% )
Sep 17, 2014 at 6:58 PM
I'm sorry, but this question and piece of code has nothing to do with IRIS, whatsoever. You will have to ask the author of that code. I can't help you...
Marked as answer by jaromirbenes on 9/17/2014 at 10:58 AM
Sep 17, 2014 at 7:47 PM
thank you jaromir.