In both (R,Q) models, the one with a shortage cost and the one with a service level constraint, we set up Solver so that the multiple k is constrained to be nonnegative. The effect is that the reorder point R will be no less than the mean demand during lead time, and the expected safety stock will be nonnegative.

This seems reasonable, but is it always optimal?

Experiment with the service level in the file Ordering Cameras 2.xlsx. Change the Solver settings to allow the multiple k to be negative; that is, don’t constrain it to be nonnegative. For lower service levels, is it ever optimal to have k negative? Comment briefly why this might or might not be the case and explain the implications for the company.