Manpower Programs as Markov Chains

A discussion of how manpower training programs can be described as Markov processes and of the utility of such a conceptual description. Although programs may have the same name and organizational structure, they may be quite different in operation and effectiveness. Markov chains provide a simple, satisfactory way of describing differences between programs. In describing the dynamics of a training program by a Markov chain, the program is viewed as a set of transitions from one state to another. Programs typically consist of a finite number of progression levels and outcomes that can serve as the state descriptors: (1) no service, (2) intake, (3) counseling only, (4) in-training, and (5) employment. Although movement of people from one state to another occurs continually, it is possible to review a program only periodically (say once each month) and determine for each state the proportion of people who have moved to another state. The movement rates may differ greatly for different programs even though resource and personnel allocations are similar. By computing the numerical transition rates between categories in programs, significant interprogram differences and similarities become apparent and the reasons for them can be further explored. (See also RM-5739.)