bill passed by the Congress in December 1995, would provide the greatest savings over the next six years--$22.3 billion. Substantial savings would also be realized in Medicaid (which is closely linked to the SSI program), and smaller savings would occur in the AFDC program. That bill would exempt certain groups of aliens--chiefly, refugees who have been in the United States for less than five years and legal aliens who have amassed 40 quarters of employment in the United States--from the cutoff. A grace period would allow aliens who were already receiving benefits to continue receiving them for one more year. Emergency Medicaid benefits would continue to be available to aliens. The bulk of savings would come from SSI and from the regular (nonemergency) Medicaid program. When estimating savings, the Congressional Budget Office (CBO) takes into account a potential increase in the rate at which immigrants would become citizens in order to obtain benefits.

Another option would be to continue deeming for immigrants until they were naturalized. Immigrants generally are not permitted to become citizens until five years after they enter the United States, and many are never naturalized at all. This option--which appeared in President Clinton's budget proposals for fiscal year 1997--would apply to future applicants for SSI, Food Stamp, and AFDC benefits. The President proposes several exemptions, chiefly for immigrants who are more than 75 years old or have worked in the United States for more than 20 quarters. Again, some Medicaid savings would occur because of that program's links to the SSI and AFDC programs. The option would save $5.1 billion over the 1997-2002 period. As under the previous option, CBO assumes that some immigrants would be spurred to become citizens if the plan was adopted.

The option that would offer the least savings over six years--$2.0 billion--would permanently extend the deeming period to five years for all four programs. Since most legal immigrants cannot become citizens until five years after they enter the country, changes in naturalization rates would not affect this estimate.

There are several arguments, aside from savings, in favor of these options. First, some supporters of these measures question immigrants' commitment to the United States if they do not become citizens, and thus contend that they are not entitled to assistance. Second, these options would promote more responsibility among immigrants' sponsors. Third, restricting public assistance might speed immigrants' integration into the American economy and culture. Finally, some people worry that allowing immigrants to collect welfare benefits encourages an influx of people with few skills who may compete for jobs with lowskilled citizens.

There are several arguments for not adopting these options. Legal immigrants enter the country with government permission, many pay taxes, and some can be called to serve in the armed forces. Therefore, opponents of these restrictions argue, legal immigrants who are needy also deserve welfare benefits. Second, removing benefits would lower the living standards of vulnerable immigrants, including children, the elderly, and the disabled, many of whom eventually become citizens. Finally, since it is unconstitutional for states to use immigrant status in determining eligibility for state-run programs, adopting these measures would probably increase the cost of states' programs providing general assistance.

At the beginning of fiscal year 1991, 29 states had EA programs, which focused on assistance after natural disasters or crises threatening family living arrangements. During the late 1980s and early 1990s, federal spending on EA programs ranged from about $125 million to $175 million a year. Now, however, all but two states have emergency assistance programs and many states have expanded assistance into new service areas, such as emergency foster care or family preservation activities that might diminish the need for foster care. As a result, federal spending jumped sharply to $940 million in 1995 and is expected to reach more than $2 billion in 2002 under current law.

This option would limit federal EA spending to $1 billion a year. That amount falls between spending levels before the upsurge and estimated spending for the current fiscal year. Limiting EA spending would save an estimated $555 million in 1997 and

$5.4 billion during the 1997-2002 period. The limit could be accomplished either by setting a program cap or by converting the program to a block grant. Funds could be distributed among states in proportion to current spending, or a different mechanism could be used so that states that have not already increased their spending would not be placed at a disadvantage.

Much of the increased EA spending appears to represent a shifting of state spending to the federal government, rather than an increase in services to needy families. In addition, a new Family Preservation and Support Services program, enacted in the Omnibus Budget Reconciliation Act of 1993, entitles states to $930 million in federal funds during the 1994-1998 period to meet some of the same needs of children that the EA expansions cover. Nonetheless, under this option many states would receive less funding than they have in the recent past. Moreover, if states did not offset reduced federal funds, some of the most vulnerable children--for example, victims of abuse--could be hurt. Also, if family preservation services are successful in avoiding placement of children in foster care (and good information on whether there is such a relationship is not available), any reduction in EA funding could increase federal spending on foster care.

The Child Support Enforcement program collects child support payments from noncustodial parents on behalf of families receiving Aid to Families with Dependent Children (AFDC). Those payments are largely used to offset federal and state costs for AFDC. Amounts up to the first $50 in monthly child support collected, however, are paid to the AFDC family, without affecting the level of AFDC benefits. In essence, that policy means that AFDC families for whom noncustodial parents contribute child support get as much as $50 more a month than do otherwise identical families for whom such contributions are not made.

Eliminating the $50 child support payment to AFDC families would save the federal government $120 million in 1997 and $910 million through 2002. Stopping such payments would end the differential

treatment of AFDC families that depends on whether the noncustodial parent pays child support. Administrative complexity would also be reduced.

Nevertheless, the child support payment provides an incentive for custodial parents to make an effort to obtain support. If the payment was eliminated, recipients of AFDC would be no better off when noncustodial parents paid child support than when they did not, perhaps reducing recipients' cooperation in seeking such payments. Noncustodial parents also might reduce their child support payments if this option was enacted, although new enforcement tools such as the withholding of wages might make it difficult for many to do so. In either case, the well-being of the children in families receiving AFDC would be adversely affected.

The federal government, in conjunction with the states, provides several programs for financially needy families. The programs, including Aid to Families with Dependent Children (AFDC), Job Opportunities and Basic Skills (JOBS), Emergency Assistance (EA), and related child care programs (AtRisk, transitional, and AFDC child care), are administered by the states subject to federal laws and regulations. The federal government, through matching funds, pays more than one-half of the expenses of those programs. States, however, set their own benefit levels and eligibility criteria subject to federal limitations. Therefore, total federal spending is in large part a result of decisions made by state governments.

In 1995, the federal government spent more than $12 billion on AFDC benefits and an additional $2 billion on AFDC administration. Federal expenditures on child care programs associated with AFDC amounted to more than $1 billion, and EA and the JOBS program cost about $950 million each. Thus, those programs and related expenses totaled about $17 billion.

The option considered here--similar to provisions contained in recent legislation (H.R. 4) passed by the Congress but vetoed by the President and the plan put forward by the National Governors' Association (NGA)--would end the specific entitlement provided

by the federal government to AFDC, JOBS, EA, and related child care by replacing those programs with a block grant to the states. States would be required to spend funds from their block grants on needy families with children, and each state would design its own programs subject to some federal constraints. For example, the federal government would give money to a state on condition that the latter spend on its welfare programs a minimum proportion (75 or 100 percent) of its previous spending on AFDC and related programs. If a state spent below that threshold, the federal government would reduce its block grant the following year by the amount that the state underspent.

Federal spending for these welfare programs would no longer be determined by an open-ended entitlement based on states' benefit levels, eligibility criteria, and changes in the economic conditions of potential recipients. Instead, the Congress would control spending by annual budgetary legislation. In the past, as the number of recipients in AFDC and related programs increased, or as states raised their benefit levels, federal spending has grown accordingly. From 1990 to 1995, in fact, federal expenditures on AFDC benefits increased about 20 percent, from more than $10 billion to slightly more than $12 billion (somewhat faster than growth in the consumer price index). Although states would no longer receive automatic increases in federal assistance if caseloads grew, many governors believe states would save money because their increased control over their welfare programs would enable them to reduce people's dependence on welfare.

There are at least two methods of setting the size of a state's block grant. The first provides each state with funds equal to the amount it spent in 1995 on the programs being replaced by the grant. The second allows each state to receive the average level of its yearly spending on such programs during the 1993-1995 period. In addition, block grants could be fashioned to include increased amounts based on population growth.

If the states' initial grants were equal to the level of AFDC spending in 1995 and state spending was required to be 75 percent of its previous level, savings would total about $5 billion during the 19972002 period. If the grants were equal to the average

level of spending for each state during the 1993-1995 period, the federal savings would be about $9.5 billion between 1997 and 2002. Federal savings would be higher if states were required to spend at least 100 percent of their previous spending. The higher savings would come from lower federal payments for Food Stamps. With a 100 percent maintenance of effort on the part of the states, cash benefits to needy families would be higher, and households would qualify for fewer Food Stamp benefits. Over the 1997-2002 period, those additional savings would amount to about $400 million, with most occurring after 2000.

Supporters of block grants argue that freeing states from federal regulations would give them the flexibility to tailor their programs to their particular populations and economies. Administrators could also coordinate service delivery better among multiple programs and between state and local governments. Furthermore, state-based programs would give those who are actually delivering services to families sole responsibility for achieving positive outcomes. Many governors believe that would increase their ability to move people off the welfare rolls and reduce the administrative burden of providing benefits. Moreover, allowing states the freedom to experiment with different types of programs could provide information on the effectiveness of various approaches to welfare reform.

Opponents of this option fear that fiscal pressures on states during an economic downturn or recession might lead them to cut back their support of needy families without a penalty. Under the present system, a reduction in state welfare spending automatically brings about a reduction in federal spending in that state; that would not be true under a block grant, thus increasing the incentive to reduce state spending. At the same time, federal support might decline under the block grant structure since the programs might be considered less of a federal responsibility. Furthermore, the current system already allows for substantial flexibility. States wishing to experiment with modifications to those programs can apply to the Department of Health and Human Services for a waiver. Since 1990, about 40 states have received waivers for one or more AFDC rules, although state administrators have complained that the process for obtaining waivers is slow and cumbersome.