the "longitudinal formulation," addresses the policy-relevant question of what the taxpayer costs are of adding or subtracting another immigrant. To put it simply, the longitudinal formulation calculates all current and future taxes and all current and future government spending attributable to a new immigrant and his or her descendants. This approach forces the analyst to make explicit assumptions about the future course of many uncertain events. To illustrate with just one example, some assumption has to be made about the timing and magnitude of the resolution of the federal budget deficit when the baby boom generation retires. Because there is no way of avoiding making some assumptions and the inherent uncertainties cannot be dismissed, simulations under alternative assumption scenarios are recommended.

The other approaches that Lee and Miller analyze are all variants of cross-sectional annual budget computations. For a given year, they all ask some counterfactual questions—what if I took away one immigrant ("immigrant only"), one immigrant-headed household, including any native-born children in that household ("immigrant household"), and one immigrant household and any living descendants of this household, including native-born children who head households ("concurrent descendants")? The immigrant-only computation ignores all costs (especially schooling) and taxes associated with native-born children who are living in the immigrant's household. Such costs would certainly seem a consequence of immigration.

The immigrant household concept includes these costs and all others attributable to native-born second-generation children as long as they remain in the immigrant household. However, it ignores all costs and especially all taxes paid by the second generation when they form households of their own. These taxes are also the consequence of immigration. To remedy that problem, the descendent generation approach adds all government benefits received and taxes paid by the second and third generation as long as the first-generation immigrant from whom they descended is still alive. These three approaches can all be computed from annual cross-sectional experiments that are meant in part to account for some aspect of the bias inherent in the cross-sectional annual calculation. Although the authors still favor the longitudinal formulation, they argue that, among the cross-sectional annual budget estimates, the concurrent descendants approach is probably the least biased among the cross-sectional alternatives.

The basic bottom line of the concurrent descendant formulation is that immigrants are a net taxpayer benefit to native-born households. This net benefit takes place exclusively at the federal level and not at the state level. Consequently, residents of some immigrant-intensive states (such as California) experience higher taxes due to immigration.

Lee and Miller also provide an interesting comparison among the three annual budget approaches. They demonstrate that the most frequently used methodology—immigrant households—is the only one that produces a negative immigrant fiscal impact.