Multi-level programming and development policy
CANDLER, W. | NORTON, R.
Most economic policy problems can be decomposed into two related subproblems: the "behavioral" problem of forecasting the economy's reactions to policy changes, and the "policy" problem proper, of choosing among the alternative possible outcomes. Traditionally, optimization models address one subproblem or the other, but not both. This paper presents a new algorithm which enables the simultaneous treatment of both subproblems; it is a modification of the simplex algorithm which permits the simultaneous operation of two distinct objective functions. For illustrative purposes, the procedure is applied to a model of Mexican agriculture. This demonstration application reveals that: (a) the traditional technological (production possibilities) frontier may be quite irrelevant to the policy problem, for it ignores decentralized preferences; and (b) even without precise knowledge of the weights in the "policy objective function," it is possible to use multi-level programming to significantly clarify the nature of the available policy choices.
显示更多 [+] 显示较少 [-]