The association of an ERUF value for a batch program with a batch job with an elapsed throughput time is very dangerous as well. For instance, you may do tests that conclude one ERUF unit is 1 second elapsed in batch jobs. However, the next day, higher priority online jobs are using more CPU cycles, so the ERUF ratio may stretch to 2 or 3 seconds, thus making your batch job take 2 or 3 times longer to complete (in elapsed time).
The ERUF value for batch jobs can really only be used to compare the complexity of a new batch job with the complexity of any existing batch jobs that you have.
