Section Three
Additionally, in forecasting demand, forecasters must
choose predictor variables (independent variables) that are
linked with variations in the underlying level of business
activity.^76 This requirement for forecasting is evident in the
experiences of a computer manufacturer that utilizes
computerized productivity models to predict the demand for
human resources. These models, which are based on statistical
correlations between workload and staffing levels, enable the
company to forecast the demand for human resources in its
manufacturing divisions, as well as in marketing and services.
In some other departments, where such relationships are less
systematic, more qualitative techniques are used.^77
An interesting forecasting example is provided by the
case of maintenance workers. In fore-casting the demand for
maintenance workers, the reliability of the equipment to be
repaired is an important predictor variable. With equipment that
is to be maintained, forecasters can use such factors as the
mean time between failure of components as indicators of the
number of maintenance workers needed. In the case of
maintenance work, these estimates of the demand for
maintenance workers can be considered in conjunction with
decisions to purchase equipment. In cases in which a stable
employment level of maintenance is desired, purchases of
equipment can be phased over time so that expected failures
will not occur all at once, necessitating higher levels of