With the landmark 1996 welfare reform law up for reauthorization this year, both Republicans and Democrats are taking credit for the astounding fall in welfare rolls. Between March 1994 and July 2001, they fell 59 percent from their historical high of 5.1 million families. But with all the self-congratulatory backslapping about welfare reform's success, you'd never know that the decline in the caseload stopped last July, that caseloads rose 2.3 percent between July and December 2001--and that they have probably risen more since then.

In some states the increases have been much larger than 2.3 percent. Between December 2000 and December 2001, Nevada's caseload rose 44 percent, Indiana's and Montana's 21 percent, and South Carolina's and Arizona's 20 percent. In the hundreds of thousands of words written in the past few months about reauthorization of the 1996 welfare reform law, these rises have gone largely unnoticed. Even the administration's most recent report to Congress does not mention them. The convoluted way in which welfare cases are counted hides the real story.

In May, the Department of Health and Human Services issued a press release reporting "no increase in the national welfare caseload" between October and December 2001. But caseloads hit their low last July, not October. From July to December, the caseload nationwide rose 0.7 percent, increasing from 2.083 million to 2.098 million.

Even this correction understates the size of the increase. Between July and December, according to the HHS data, New York state's caseload fell 17 percent and Minnesota's 11 percent. These reported declines must have come as a surprise to state officials. The 28,004 New York families that HHS thinks left welfare were simply shifted from the state's basic welfare program to its Safety Net Assistance program, because they had hit the federal time limit for benefits. (Since December, another 15,000 families have entered New York's safety net program.)

Similarly, in October 2001, Minnesota shifted 4,682 two-parent families from its federally funded welfare program to a separate state program--probably to avoid the federal welfare law's stringent work requirement for two-parent families. Counting these families would show a one percent increase in Minnesota's caseload since July, rather than an 11 percent decline.

In all likelihood, caseloads in some other states have also been artificially reduced by such transfers and other accounting devices. But adding just these New York and Minnesota cases back into the totals would make the national rise about 2.3 percent in the last six months of the year.

The reason caseloads are growing, of course, is the weakening economy. Between December 2000 and December 2001, the unemployment rate for single mothers rose 60 percent, from 5 percent to 8 percent. By April of this year, unemployment for single mothers was at 8.6 percent. Welfare increases tend to lag behind rises in unemployment, so even if there is no further growth in joblessness, expect rolls to continue rising.

Until now, though, the welfare reauthorization debate has proceeded as if caseloads would never rise again. Indeed, most major bills now before Congress take advantage of the financial savings from the earlier caseload declines to fund their favored programs: The Republicans want to redirect a good portion of the surplus to mandatory workfare-like programs, while the Democrats want to use the surplus to expand child care. Observers expect a compromise bill to pass sometime in the fall--as long as neither side overplays its hand.

But the workfare vs. child care debate ignores the looming possibility that more families, not fewer, will need the safety net of basic welfare assistance. If caseloads rise high enough, funding for public assistance will be pitted against spending for either the Republican work mandates or the Democratic child care expansions, or both. Many states could find themselves under enormous pressure to restrict access to welfare, reduce benefits and push families off the rolls.

The answer is to give states much broader authority to bank their welfare surpluses for the almost inevitable rainy day. When the House passed a reauthorization bill in mid-May, there was no discussion of rising caseloads and the programmatic and budgetary dislocations that could result. By the time the Senate takes up reauthorization later this year, the extent of caseload rises will be more visible to Washington policymakers. If the new welfare law is to protect families needing assistance, both parties need to rethink the rosy scenarios built into their proposal.