Asset allocation

Hi, I am struggling to figure out whether Dynare can handle asset allocation (or portfolio choice).
For example, in the neoclassical growth model, agents save in capital only, so as in the classical Diamond’s OLG model. Basically I want to introduce risk-free saving in these classical model, whose return is known in prior and is in the agent’s information set. For instance, the budget constraint becomes:
where c_t, s_t, b_t and w_t stands for consumption, saving in risky capital, saving in risk-free bond and wage at time t, respectively.
the next period’s budget constraint thus is:
where re is the return on risky capital because the economy is subject to stochastic productivity, rf is the risk-free return. I use rf(-1) because rf is known in prior.

There are two Euler equations related to asset allocation:
(1). Uc=betaUc(+1)(1+re(+1))
(2). Uc=betaUc(+1)(1+rf)
where Uc stands for the marginal utility of consumption.

I find Dynare (I use Dynare 4.3.0) can not return steady state for me. It seems there is indeterminacy of asset allocation because in steady rf=re and the value of s and b (steady state value of s_t and b_t) can not be determined. When I remove the risk-free saving, allowing investment in risky capital only, Dynare returns steady state well given a calculated initial guess and all go well in simulation.

Can Dynare deal with two or more assets? How can I handle this indeterminacy problem if I want to introduce risk-free saving? Many thanks to any reply. I think someone raise similar question here but surprisingly got no reply:

It’s not a problem of Dynare, but of first order perturbation techniques. Currently, the only way would be to use Dynare’s capabilities to implement e.g. the Devereux/Sutherland algorithm.

Many thanks to your reply. Sorry but could you recommend their most relevant papers that develops the algorithm? I should read all their 2007a, 2007b,and 2007c papers but time is limited and I have to solve this problem asap. I appreciate if you could recommend the most relevant one. Thanks.

I would start with

Thank you very much. I will start from there.