trick to meet Blanchard Kahn conditions

[naivej]

Dear all

I have a 2 country NK model with tradable good only, which works OK. But when I add nontradable good sector, I have Blanchard Kahn conditions problem (22 eigenvalues>1 and 20 forward looking variables) . I have just found a trick that fix the problem but I am not sure if I am doing right.

please see line 88 and 90, which are the capital market clear equations for home and foreign countries. All capital in one country equal to the sum of capital used in tradable good sector and nontradable good sector in this country. s stand for foreign country variables, t stand for tradable and n stand for nontradable. At the beginning i use the current period timing and I have the Blanchard Kahn conditions problem (no '1' for all variables in these two equations). I also use model_diagnostics and there is no return. But the problem is fixed if I simply forward this two equation one period ahead (24 eigenvalues>1 and 24 forward looking variables). I guess I need to let dynare know that sector capital are forward looking variables by doing so. Otherwise dynare will treat them as predetermined variables because I only have capital accumulation and capital Euler for overall capital. however, this trick makes sector capital enter model either at t-1 or t+1.

[jpfeifer]

You need to get the economic setup right. Think about how you want to model capital in the two sectors and whether there are frictions. See e.g. http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=3564 and http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=7000

[dmbn]

So if I understand right, I can't just rewrite my jump variables as predetermined variables? It solves the problem with eigenvalues, but apparently it leads to wrong results. I'm I right?

[jpfeifer]

If your jump variables correctly classified as jump variables you cannot simply change their timing.