Policy makers are showing increasing interest in employment and training programs as a means of reducing welfare costs, despite the evidence that welfare costs are reduced little by these programs. This paper examines the potential for increasing the welfare savings associated with such programs by targeting women for whom the attendant welfare reductions are likely to be the largest. It reanalyzes data from five programs and computes welfare impacts as if the programs had been selectively offered. The results suggest that targeting job-search assistance programs is not likely to be effective, but taht there is a potential for improving the effectiveness of subsidized employment and training programs. Copyright 1989 by MIT Press.

Until recently, estimates of demand functions for public goods were obtained (either with aggregate or micro survey data) using single equation estimation techniques. However, demand estimates may be biased when in dividuals' choices of communities are dependent upon the quantity and quality of public good provided. This paper spells out the nature of this bias (called Tiebout bias) and suggests an improved maximum-likelihood estimation technique. The technique is applied to a data set involving local public education in Michigan. Copyright 1987 by MIT Press.