The information recovery process envisions sample data that are often limited, partial, or ill-conditioned, and the corresponding statistical models that form the basis for estimation and inference may in some cases be underdetermined or ill-posed. The ill-posed aspect may arise because nonstationary or other model specification reasons may cause the number of parameters to exceed the number of data points or the moment conditions to be less than or greater than the number of unknown parameters. In addition, the design matrix implied by nature or society may be ill-conditioned. Consequently, if traditional estimation procedures are used, the solution may be undefined or the estimates that are obtained may be highly unstable giving rise to high variance and low precision of the recovered parameters.