Global Optimization of Economic and Social Policy

Background

The U.N and its various agencies have during the last decades been heavily involved in the formulation of goals of economic and social policy (such as the millennium goals, see the UN, 2000) and the design of policy aiming at realizing such goals. In order to meet them, many countries are adjusting their energy policy, educational policy, and environmental policy.

There was a time when economists thought that global optimization of economic and social policy could be accomplished by the stratagem of maximizing some over-arching global “utility function.” This was the approach of Jan Tinbergen, who wrote a classical treatise on the subject in the 1950s. Fortunately, our understanding of economic and social policy has advanced considerably since then. International organizations do no longer insist on the assumptions of economic equilibrium that hampered Tinbergen´s efforts. Indeed, the International Labor Organization (an arm of the United Nations head-quartered in Geneva) explicitly recognizes that the goals of “decent work” that the ILO itself has formulated are certainly not always attained. Instead, ILO has called for the calculation of the shortcomings of the various countries from this (equilibrium) ideal (see ILO, 2001).

The present paper draws attention to a recent extension of data envelopment analysis that permits the presence of a convex social welfare function of all outputs (all policy goals), see Cooper, Thore, and Tarverdyan (2010). The new approach seems to meet the call by Jan Tinbergen for the construction of a model of economic policy that is empirically implementable, and that delivers a numerical assessment of the success of policy. Starting out from Tinbergen´s work, the paper gives a mathematical account of the new model. While neoclassical equilibrium is still possible, the new idea is to allow for the possibility of non-achievement of the equilibrium ideal. Neoclassical equilibrium is viewed as a matter that needs to be tested. In the case of optimum failure, the deviation from equilibrium will be computed.