The sfcr_random() function to facilitate the inclusion of random variation in the models was introduced in this version. This function is a wrapper around the built-in rnorm(), rbinom(), and runif() functions. The main benefit of using sfcr_random() is that it smartly guesses the size of the sfcr_baseline() and sfcr_scenario() models they're inserted, as well as the size of the sfcr_shock().
The sfcr_random() function should only be used inside sfcr_set()s. If used outside this function, it will only print a message saying that it should not be used in this way.
The disallowance of exogenous series introduced in v0.2.0 broke the models that used functions like rnorm() to add random variation. This version undo this change, allowing exogenous series to be passed along sfcr_baseline() again. Nonetheless, this utilization is discouraged since it generates unexpected behavior at the sfcr_scenario() level. Warning messages are raised about these issues.
Prior to v0.2.1, the inclusion of exogenous series at the sfcr_baseline() level would not be transmitted to the sfcr_scenario() level since the previous infrastructure of the model only replicated the final values of the sfcr_baseline() object to fill in the matrix used to generate the initial values at the sfcr_scenario() level.
This version re-evaluates the expressions created with sfcr_set() and passed as the external argument to the sfcr_baseline() function at the sfcr_scenario() level. In this way, if a user passes creates a random shock to influence the evolution of endogenous variables at the sfcr_baseline() level, this behavior will be reproduced at the sfcr_scenario() level. To avoid complications with different lengths between the two objects, the utilization of sfcr_random() is recommended.
Note that if an exogenous series is passed to the sfcr_baseline() function, it will result in an error if the lengths of the models do not coincide, it will generate an error. If the lengths coincide, the model will run, but the exogenous series will be replicated from the start, which is almost never the intended behavior. That's why this functionality will be removed in later versions of the package when the sfcr_random() function becomes widely adopted.
Constrained the utilization of exogenous series to avoid unexpected behavior and undetected bugs in the models.
More specifically:
sfcr_baseline() to disallow the utilization of exogenous series.sfcr_scenario() to make sure that the length of the series matches with the length of the shock.sfcr_portfolio() function that calculates the matrix of portfolio parameters from a limited vector of parameters by imposing the adding-up and symmetry constraints.First public release.