Levy often explained how he had approached the problem of unemployment. He told me that he employed people because he expected to make a profit. . . . spending $1,000 for rent, merchandise, and wages with the expectation of getting $1,100 back. He put $1,000 into circulation but wasn't sure where the other $100 came from. He set out to find the answer. He believed that private enterprises would employ all available workers if they found it profitable to do so.

One person's public good is another's public bad and so, perhaps, the public goods problem could be more generally described as the public commodities problem,1 in which disagreement about the basic goal of a spending program complicates the decision of how much to spend to achieve that goal. Although public goods spending is a continuous variable, it often has a binary goal such as win a war, deter crime, provide transportation, or reduce poverty. To decide how much spending is necessary, society must answer both a normative and a positive question: Normatively, should the government adopt the goal of this spending program? Positively, given this spending program's goal, what is the optimal level of spending?2 If the answer to the first question is yes, it may be desirable that the actual level of spending be set at the level that would be optimal given that goal, but spending may not be set at that level. This paper uses the median voter theorem to demonstrate that those who do not believe the government should pursue the goal and those who believe that the government can achieve the goal with relatively less spending can form a coalition to keep spending at a level below which most supporters (and possibly most citizens) believe is objectively the optimal amount of spending. Thus, even if voters have rational expectations about the amount necessary to achieve a goal, disagreement about whether or not to pursue the goal can cause underfunding bias.

* Direct correspondence to Michael Handel, The Jerome Levy Economics Institute, Bard College, Annandale-on-Hudson, N.Y. 12504. I thank Christopher Winship, Christopher Jencks, Lee Rainwater, Michael Piore, Lawrence Mishel, Jared Bernstein, John Schmitt, Ruy Texeira, Harry Holzer, David Autor, seminar participants at the Labor Lunch of the Management Department of the Wharton School, University of Pennsylvania for helpful discussions and comments. This paper was presented at The Macrodynamics of Inequality in the Industrialized and Developing Countries conference held at The Jerome Levy Economics Institute on October 28, 1999.