Based on the above scenario, was my reliance on time as a my independent "X" variable completely inappropriate as a test for overhead reasonableness, and if so, is there a better or more appropriate prediction model available for testing/predicting historical cost trends over time? Thanks. -- Mike
In various forecasting applications (e.g. wage rates) “time” is sometimes used as an independent variable, not because time causes costs to change, but rather that time is a suitable “surrogate” for other variables such as inflation, business cycles, etc. This presupposes a regular pattern (e.g. a sustained increase in wages over time).
Open full Question Details
One may use regression with time as the independent variable for individual elements of cost in the overhead pool using the same philosophy as above. For example, as one might use time for direct labor (rates) as a surrogate, one might also use time as a surrogate for indirect labor such as supervisory or administrative labor rates. However, other elements within the indirect cost pool are subject to factors other those captured by time. Given the fundamental requirement of an overhead rate to have a causal/beneficial relationship between the pool and the base across which they must be allocated, it is consequently more proper to regress a total pool against some base other than time.