the society is counted as being twice as impor- 

 tant as income going to other segments of the 

 distribution. (Its impact already shows up 

 once in the total figures.) 



The optimizing behavior is 



Maximize (PI t + 1.0 PI p ) 



Policy 1 yields $1,000,000,000 



+ $200,000,000 = $1,200,000,000 

 m Policy 2 yields $ 800,000,000 

 ^ + $500,000,000 = $1,300,000,000 



Policy 3 yields $1,010,000,000 



+ $100,000,000 = $1,110,000,000 



By choosing policy 2 here we still maximize 

 the weighted value, even though the total in- 

 crease in income is smaller. Here we see the 

 importance of establishing weights and priori- 

 ties on multiple goals. With this weighting, we 

 choose the policy with the lowest total in- 

 crease. 



In situation C, extra consideration is to be 

 given to income reaching the lowest quintile, 

 but not to the extent shown in situation B. 

 An extra premium or weight of 0.6 is given, 

 so that any income to the lowest quintile will 

 count slightly over half again as much as in- 

 come going to other segments of the distribu- 

 tion. The optimizing behavior is 



Maximize (PI t + 0.6 PI p ) 



m Policy 1 yields $1,000,000,000 

 T + $120,000,000 = $1,120,000,000 



Policy 2 yields $ 800,000,000 



+ $300,000,000 = $1,100,000,000 

 Policy 3 yields $1,010,000,000 



+ $ 60,000,000 = $1,070,000,000 



Policy 1 is the optimizing behavior even 

 though it neither maximizes total increase in 

 personal income nor maximizes the increase 

 in personal income to the lowest quintile. 4 



4 More properly, these examples should show a 

 weight or premium of 1.0 in A, 2.0 in B, and 1.6 in C, 

 because in this particular example a value of 1.0 is in- 

 cluded in the increase of Pit for any and all goals. The 

 a term then becomes the "extra" premium. To make 

 the example directly comparable with a later one this 

 distinction is ignored. This point is discussed in 

 Marglin (1967, p. 24). Marglin's analysis is the basis 

 for much of the theory used in developing the deci- 

 sion model, and should be consulted for more thor- 

 ough treatment of this topic. 



The Forest Service must be able to specify 

 the order of priorities of its goals — that is, 

 establish a weighting procedure both on the 

 national level and on the more limited level of 

 specific project areas. Without this, there is no 

 basis for judgment of the optimality of its ac- 

 tions to achieve the goals specified. If we al- 

 ways had available a perfect measure of wel- 

 fare, benefits, or satisfaction, we would not 

 need to determine the weights as a separate 

 process, because they would be automatically 

 reflected in the measured value. The impor- 

 tant distinction between weight and value 

 arises because our measuring capability is in- 

 adequate. 



The Need for Flexibility in Goal Ranking 



Flexibility during the initial goal formula- 

 tion stages of planning is essential. Much of 

 the information required for determining goal 

 priorities will not become available until ex- 

 perimental programs are implemented. It is a 

 difficult task, however, to maintain such flexi- 

 bility without opening the flood gates to spe- 

 cial interest pleading. Unless clear agreement 

 on the priority of wilderness, for example, is 

 arrived at in advance, then every timber cut- 

 ting project may become the focal point of 

 debate. This is counterproductive and limits 

 the ability of the decisionmaker to implement 

 an agreed-upon program. 



This process of establishing weights or pri- 

 orities in advance of project planning is not 

 observable in much of the decisionmaking go- 

 ing on in firms and agencies today. Lowi 

 (1969, p. 147) argues that this is no accident. 

 This failure in bureaucratic decisionmaking is 

 the direct result, Lowi claims, of the growth 

 of "interest group liberalism." He makes an 

 important distinction between involvement of 

 the citizenry in the bargaining process and 

 project planning on the one hand, and, on the 

 other hand, a limited aspect of public involve- 

 ment, logrolling. There is a world of differ- 

 ence between bargaining on the stakes of a 

 particular case, which is logrolling, and bar- 

 gaining on the rules and criteria applicable to 

 the decision. Decisionmaking should proceed 

 from firmly established rules, based on pre- 

 determined weights and priorities. The deci- 

 sions that result from rules arrived at by con- 



7 



