Section Threecould experiment with transition probabilities based on
movements for one year prior to that, an average of the two
previous years, an average of the three previous years, and so
on, and compare these fore-casts with what actually happened.
In this manner, the forecaster can determine the particular
number of years that produces transition probabilities that will
result in the most accurate fore-cast. Of course, any use of
historical data assumes that certain conditions of the past will
remain in the future, which may not be true.
In addition to forecasts of an organization’s internal supply of
human resources one year into the future, Markov analysis can
be used for longer-range forecasts. Heneman and Sandver
have pointed out that, through applications of matrix algebra,
forecasts for multiple years into the future can be developed.
Additionally, they have explained that such forecasts can
incorporate new hires in the various jobs for individual years.^63
Iterative approaches provide an alternative for extending
Markov analysis forecasts several years into the future. With an
iterative approach, the forecasted distribution of employees
across job categories is used as the input for the next year’s
employee distribution, which is then used with the transition
probabilities to derive the next forecasted distribution. This
process is then repeated for a forecast of another year into the
future and can be performed on computerized spreadsheets.
Forecasters must bear in mind that the validity of the transition