MENDELSSOHN: USING MARKOV DECISION MODELS 



are easier to implement and easier to explain the 

 rationale to the public. 



As an example of suboptimal, approximate 

 policies, the following nine modified base stock 

 policies were examined: 



3) Policy of base stock size of 0.56 till 1.40, then a 

 base stock size of 0.84. 



4) Harvest till 0.28, harvest 0.28 till 0.84, a base 

 stock size of 0.56 till 2.52, then a base stock size 

 of 0.84. 



Wood River 



Branch River 



1) Base stock policy, base stock size = 0.84. 



2) Policy of base stock size of 0.56 till 2.52, then a 

 base stock size of 0.84. 



5) Base stock policy, base stock size of 0.40. 



6) Base stock size of 0.4 till 1.6, then a base stock 

 size of 0.6. 



Figure 2(a-m). — Optimal policy functions for the Wood River for various assumptions about the relative value of smoothing costs. (See 



text for details). 



41 



