because the goals of encouraging local economic activity and providing jobs to disadvantaged zone residents are typically quite clear. For these programmes, the research shows that the cost per new job per zone resident is quite high, and probably exceeds $40 000 per year.
Clearly, there is room for additional research on the effects of tax abatements and enterprise zones. One of the obvious dangers in evaluating programmes such as enterprise zones is that they are likely to be established in areas with unusually high unemployment or at times when unemployment is at a peak. As a consequence of these extreme starting points, the unemployment rate is likely to go down relative to other areas or over time even in the absence of the programme. This observation makes it particularly important that control groups be used in the evaluation. Ideally, the control group should be randomly chosen. For example, the state might randomly select areas for enterprise zone designation from among a number of areas applying for zone status. The areas not selected could then be used as control groups with which the outcomes in the participating zones could be compared. The use of comparable areas as controls will directly address the challenge of determining what would have happened in the zone in the absence of the tax abatement programme. However, it does not address the other thorny analytical question of whether the new jobs in the zone represent net new jobs for the metropolitan area or simply jobs that were transferred from nearby areas.