1. Run check_next.m with write_policy.m in which an auxiliary shk 
is specified in the constant term.

The code solves the stationary model and solves a model in which
the next-period policies are explicitly specified. The two
solutions should be the same.

2. Run main.m with no change in parameters and the second-order
approximation of the policy functions and with write_policy.m
specified with shk in the constant term

There should be no deviation from the initial state.

3. Run main.m with no change in parameters and the first-order
approximation of the policy functions.

There should be no deviation from the initial state.
