AFDC vs TANF names a comparison between two machines for paying cash assistance to families with children, built inside the same statutory frame and engineered to opposite specifications. The first machine, Aid to Families with Dependent Children, began as Aid to Dependent Children under Title IV of the Social Security Act of 1935 and carried the longer name from 1962 onward. The second machine, Temporary Assistance for Needy Families, entered the same Title IV, Part A in 1996 and displaced the first in the same statutory space. A shared address makes the comparison unavoidable, because each program answered the same public question about aid to needy children in parental care. Opposite engineering makes the comparison instructive, because each program answered that question with a different legal commitment, a different funding formula, a different response to economic contraction, and a different allocation of decision authority between the federal government and the states.

The change from one machine to the other arrived through a single statute. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Public Law 104-193, signed on August 22, 1996, repealed the AFDC program, rewrote Title IV, Part A of the Social Security Act, disclaimed an individual entitlement to assistance in the new text, and substituted a fixed federal block grant for open ended federal matching. A change of that scale rewards close reading of the statute itself, and the statute receives a dedicated treatment in the welfare reform act guide. Each element of the substitution carried a consequence that the sections ahead trace in sequence: a right became a grant, a formula became a sum, an automatic response became a series of decisions, and a federally defined program paid at state rates became a state program financed under federal purposes and conditions.

The comparison proceeds on four axes, and each axis poses a question with a determinate answer in law. The entitlement axis asks whether a family that met the governing criteria held an enforceable claim to assistance, or whether assistance depended on state program definition under a statute that disclaimed any individual right. The funding axis asks whether federal money followed state spending at a matching rate, rising and falling with the volume of aid, or arrived as a fixed sum set in advance and detached from that volume. The cyclicality axis asks whether federal spending expanded on its own when employment contracted and need rose, or held at the statutory figure until Congress or a state acted. The discretion axis asks where the operative decisions sat, and how much of the program a state could redesign rather than merely price. Four such questions convert a familiar historical contrast into a decision framework, because each answer follows from the construction of an instrument rather than from the intentions attached to it by any administration.

The evidence base for the four answers combines statutory text, administrative record, and three measured series that recur throughout the sections. The Congressional Research Service recorded an AFDC caseload peak of 5.033 million families and 14.161 million recipients in 1994. The same service computed that the fixed TANF grant, held at a nominal figure set in the middle 1990s, had eroded by 36.2 percent of fiscal 1997 real value by fiscal 2018. The Center on Budget and Policy Priorities measured program reach as a ratio of families receiving assistance per one hundred families with children in poverty, and that ratio moved from 68 in 1996 to 23 in 2015 and 22 in 2018. Such figures discipline the comparison, because a claim about either model can be tested against the caseload the model carried, the real value of the money the model moved, and the share of poor families the model reached.

Two complications guard against a clean break narrative, and the sections give each complication a full hearing before any verdict. State experimentation under federal waiver authority tested time limits, work requirements, earnings disregards, and sanctions inside the entitlement during the late 1980s and the first half of the 1990s, so much of the content associated with the 1996 act had a trial history under the predecessor program. Benefit erosion guarded the other flank of the story, because state set AFDC grants lost roughly two fifths to one half of real value between 1970 and the middle 1990s in the median state, so the guarantee withdrawn in 1996 had a purchasing power far below the guarantee of a generation earlier. A comparison that omits the waiver record misstates where reform came from, and a comparison that omits erosion misstates the value of the entitlement that ended.

The sections take the axes in order, examine waivers and erosion as correctives to the story, and close with a verdict that names the deciding factor in the trade between the two models. That trade has a fixed structure. Automatic expansion when need rises and predictable federal cost in advance are properties that a single funding instrument cannot supply at full strength at the same time, because a promise to match whatever eligible need produces cannot also promise the total the promise costs. Each model purchased one property by surrendering the other, and each verdict about the two welfare models reduces to a judgment about which surrender a legislature ought to accept.

Two Programs, One Title, Opposite Machines

The comparison between Aid to Families with Dependent Children and Temporary Assistance for Needy Families recurs in settings that rarely announce the comparison as a question about funding structure. A student in a social policy seminar encounters two program names in a textbook chapter and asks why a program that lasted for six decades ended and a differently named program began in the same statutory place. A legislative staffer preparing a briefing memorandum reaches for a one page contrast because a hearing witness has invoked the earlier program as a baseline and the later program as a reform. A journalist on deadline, assigned to explain a state decision about cash assistance, reaches for the same contrast because editors expect historical context in a few paragraphs and because readers who remember the debates of the nineteen nineties carry one mental picture while readers who encountered the earlier program in family memory carry another. A policy analyst at a research organization reaches for the same contrast because evaluation requires a counterfactual, and the counterfactual most readily at hand is the program that occupied Title IV, Part A before August 22, 1996. In each setting, the question sounds historical. In substance, the question is architectural.

The recurrence has a specific cause. Arguments about safety net design often proceed by invoking values, outcomes, or intentions while leaving the operating mechanism unstated, and a mechanism left unstated cannot be tested. One participant in a seminar discussion praises responsiveness and means the capacity of a program to expand when need expands without a new appropriation or a new statute. Another participant praises discipline and means the capacity of a legislature to fix a sum in advance and hold total exposure to that sum across a fiscal period. A third participant praises local control and means the breadth of choices a state government may exercise over eligibility, benefit form, and program purpose. A fourth participant praises uniformity and means the degree to which a family in one state and a family in another state, presenting the same circumstances, encounter the same federal rule. Placed side by side without the underlying structures on the table, such positions talk past one another, because responsiveness, discipline, local control, and uniformity are not free standing virtues that a statute can maximize simultaneously in unlimited measure. Each property flows from a particular arrangement of entitlement, funding, and discretion, and each arrangement forecloses another arrangement. The question returns, generation after generation, precisely because the underlying trade was never settled by vocabulary. The trade was settled for a time by statute, first in one direction and then in the opposite direction, and the statutes left two complete and incompatible machines occupying the same legal address in succession.

A precise statement of the two models therefore has to begin with formal identity, stated once and stated exactly, before any comparison of effects. Precision prevents a common error in which a program name functions as shorthand for an era or a political coalition rather than for a defined set of legal obligations. The first model is Aid to Families with Dependent Children, abbreviated AFDC. The program began as Aid to Dependent Children under Title IV of the Social Security Act of 1935, an act signed on August 14, 1935, and the original title of Title IV was Grants to States for Aid to Dependent Children. In 1962, the words families with were added to the program name, and the program became Aid to Families with Dependent Children. The second model is Temporary Assistance for Needy Families, abbreviated TANF. That program was created by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Public Law 104-193, signed on August 22, 1996. The 1996 act repealed the AFDC program, rewrote Title IV, Part A of the Social Security Act, and placed the new program in that rewritten part, codified at Title 42, United States Code, sections 601 and following. Guidance from the Administration for Children and Families within the Department of Health and Human Services describes the codification in the same terms. Two programs, one title, one part, successive occupancy. The shared address explains why the comparison is unavoidable. The opposite design explains why the comparison matters.

The First Model Stated in Full

AFDC, in the form that stood under Title IV, Part A immediately before August 22, 1996, was a federal and state matching entitlement program for needy families with dependent children, built around state plans that satisfied federal requirements. The federal government set categorical and financial eligibility rules in statute and regulation, including rules about the composition of the assistance unit, the treatment of income and resources, and the conditions under which a child counted as dependent and needy for purposes of federal participation. A state that operated an approved plan and applied those rules to an applicant family produced a legal consequence that did not depend on the balance remaining in a fixed federal allotment. A family that met the governing federal and state criteria held an entitlement to assistance under the plan, and the state held an entitlement to federal matching funds for allowable expenditures made under the plan. The House Ways and Means Committee background volume commonly called the Green Book described AFDC law in exactly that reciprocal form, noting that Title IV, Part A entitled states to federal matching funds for state set benefits and administration without a ceiling on total federal matching. The matching structure meant that federal spending followed state spending according to a formula rather than preceding state spending as a predetermined grant. When allowable state expenditure rose, federal payment rose with the expenditure at the applicable matching rate. When allowable state expenditure fell, federal payment fell in parallel. The appropriation language that originated in section 401 of the 1935 act captured the logic in miniature, authorizing for each fiscal year after the first a sum sufficient to carry out the purposes of the title rather than a named and capped total.

Within that federal frame, states exercised substantial but bounded authority, and the boundary is essential to a fair statement of the model. States set benefit levels, which meant that the dollar amount of a grant for a family of a given size differed across states by design rather than by accident or administrative drift. States also set the need standard against which income was measured and administered the program through state and local agencies under the approved plan. State authority, however, operated inside federal eligibility architecture. A state could pay a higher or lower benefit to the class of families the federal statute defined as eligible, and a state could exercise options the federal statute expressly offered, but a state could not redefine the core federal entitlement into a different program with a different population and a different purpose while drawing AFDC matching funds for the result. Federal rules governed matters as concrete as the treatment of earnings, the counting of a parent in the home, and the availability of a fair hearing for an applicant whose claim was denied, a hearing protection that appeared in the state plan requirements of the original 1935 text. The model thus distributed power in a particular pattern. The federal level controlled the definition of the claim and the obligation to match. The state level controlled the generosity of payment and the daily administration of the claim. Neither level controlled total program cost in advance, because total cost emerged from the interaction of eligibility rules, caseload, and state benefit choices during the period in which claims were presented and paid.

The entitlement character of AFDC deserves a statement separate from rhetoric that attached to the word entitlement in political debate during the nineteen eighties and nineteen nineties. In structural terms, entitlement described two linked legal facts. First, an eligible family possessed a claim that a state administering an approved plan was required to honor by payment according to the plan, subject to the rules of the program. Second, a state that made allowable payments possessed a claim against the federal government for the federal share computed under the matching formula, without a statutory allotment that could be exhausted before the end of a period. The arrangement resembled an open ended reimbursement contract more than an annual grant competition. No state applied for a share of a fixed AFDC pot in block grant fashion, and no federal administrator allocated AFDC dollars among states by discretionary judgment. The formula allocated. The formula responded to expenditure that itself responded to applications that met the rules. That chain, from application to eligibility determination to state payment to federal match, constituted the machine.

The Second Model Stated in Full

TANF, as created by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, is a federal block grant program under the rewritten Title IV, Part A, through which the federal government provides fixed allotments to states for operation of state designed programs for needy families within broad federal purposes and specified federal conditions. The Congressional Research Service primer on the TANF block grant describes the program as a broad purpose block grant that helps fund a wide range of benefits, services, and activities addressed to the effects and root causes of child poverty and economic disadvantage. The breadth of purpose marks a structural difference from the predecessor program. AFDC federal matching attached to a defined category of assistance for a defined eligible population. TANF allotments support a state program that a state designs within the purposes stated in section 601 of Title 42, which include assistance to needy families so that children may be cared for in their own homes or in the homes of relatives, promotion of job preparation, work, and marriage as a means of ending dependence of needy parents on government benefits, prevention and reduction of out of wedlock pregnancies, and encouragement of the formation and maintenance of two parent families. A state TANF program may therefore finance cash assistance, and a state program may also finance services and activities reasonably calculated to accomplish the statutory purposes, subject to the rules that attach when federal funds are used for assistance as the statute and regulations define that term.

The funding commitment in the TANF model runs in the opposite direction from the commitment in the AFDC model. Congress fixed the basic block grant total in statute, and the Green Book summary of the transition recorded a fixed block grant based on federal funding for the replaced programs, stated in that summary as 16.5 billion dollars annually through fiscal year 2002 for the basic grant structure described at enactment. For a statement of the model created in 1996, the defining property is fixity at the point of commitment. The federal payment to a state under the basic grant does not rise automatically because more families apply, more families qualify under the state program, or a state elects to pay a higher benefit. The allotment arrives according to the statutory allocation, historically anchored to state expenditure under the predecessor programs during specified base years, and the state plans expenditure within and around that allotment together with state funds counted toward the maintenance of effort requirement that the 1996 act imposed. The maintenance of effort condition, codified among the provisions following section 601 and administered under regulations at Title 45, Code of Federal Regulations, Parts 260 to 284 according to Administration for Children and Families guidance, requires a state to spend a specified proportion of a historic state expenditure level on qualified activities for needy families. The obligation, however, is to spend at a level tied to history, not to match each additional eligible application with new combined spending in the manner of the predecessor formula.

Entitlement language marks the sharpest formal break between the two models, and the break was deliberate in the text of the 1996 act. Title IV, Part A as rewritten contains an express statement that the part does not entitle any individual or family to assistance under a state program funded under the part. The Congressional Research Service legislative history states the consequence directly, noting that the 1996 law ended AFDC and replaced the program with TANF and that TANF ended the entitlement of individual families to aid, a formulation the Green Book summary echoed. A state receiving TANF funds must meet federal conditions that include work participation standards, time limit rules for federally funded assistance, data collection, and penalties for specified failures, and a state must operate under a state plan submitted to the Secretary of Health and Human Services. Compliance, however, does not recreate an individual claim for a family that satisfies a federally defined eligibility test, because no such test governs access in the AFDC manner. A state defines eligibility for assistance under the state program within federal constraints, sets benefit levels, decides the mix between cash assistance and other allowable uses, and manages demand against resources that are finite by statutory construction. When resources tighten, the adjustment falls inside the state program through choices about eligibility, benefit amount, duration, diversion, services, or waiting, rather than through an automatic increase in the federal allotment. That allocation of adjustment is the machine, stated without praise and without condemnation. Praise and condemnation belong to arguments about the machine. Identification of the machine comes first.

Placement of the two statements side by side exposes the namable claim at the center of the comparison, and the claim deserves a name because unnamed claims smuggle conclusions into premises. The claim is as follows: “You cannot have both.” The sentence is short enough to be quoted and exact enough to survive quotation. Automatic response to need and predictable fixed cost are mutually exclusive properties of a single funding structure at a single moment for a single program. A structure that commits the federal government to match allowable state expenditure for every eligible claim presented during a period cannot also promise, in the same commitment, that total federal cost for the period will remain within a sum fixed before claims for the period are known, unless eligibility, benefit levels, or payment itself is curtailed by a mechanism that breaks the automatic character of the response. Conversely, a structure that fixes the federal sum in advance and allocates that sum among states by formula anchored to history cannot also promise that the sum will expand automatically when need expands during the period, unless a supplemental mechanism outside the fixed commitment supplies the expansion and thereby qualifies the fixity that defined the structure. A legislature may build hybrids, contingency funds, borrowing authority, emergency appropriations, and revision procedures, and legislatures did build such devices around both models at various points between 1935 and 1996 and in the statute of 1996 itself. A hybrid does not refute the claim. A hybrid allocates the tension to a particular valve, and analysis consists in locating the valve, naming the authority that opens the valve, and stating the conditions under which the valve opens. Absent a valve that actually operates, one property yields to the other. Every argument about welfare design that invokes both responsiveness and cost predictability without locating a valve has assumed away the decision that a funding structure must make.

Stated in that form, the claim disciplines the familiar debate in which AFDC appears as generosity and TANF appears as restraint, or AFDC appears as loss of control and TANF appears as restoration of control, with the nouns changing according to the speaker. Generosity and restraint describe outcomes under specified conditions. Control describes a location of authority. None of the three terms identifies the funding commitment that produces an outcome or locates authority in a particular institution. The structural claim redirects attention to commitment. Under AFDC, the federal commitment was to a formula. Under TANF, the federal commitment was to a sum and to a set of purposes and conditions. A formula commitment and a sum commitment behave differently when caseload rises, when a recession increases applications, when a state raises benefits, when a state restricts eligibility, and when Congress revisits an authorization. Behavior under stress reveals a structure more reliably than description under calm conditions, and the history of the two programs supplies episodes in which stress arrived from economic contraction, demographic change, litigation, administrative experiment under waiver authority, and fiscal pressure at both levels of government. Later sections take up those episodes axis by axis. The task in the present section is narrower: to fix the objects of comparison so firmly that an episode can be assigned to a structure rather than to a slogan.

Four axes organize the comparison that follows, and each axis names a dimension along which the two machines differ in construction rather than merely in reputation. The first axis is entitlement, meaning the presence or absence of a legally enforceable claim, held by a family against a state program and held by a state against the federal government, when specified conditions are satisfied. The second axis is the funding mechanism, meaning the rule by which federal money moves, whether by open ended matching of allowable expenditure or by fixed allotment under a block grant supplemented by specified additional funds and conditions. The third axis is cyclical behavior, meaning the pattern of how each structure responds across expansions and contractions in employment, earnings, and applications for assistance, and whether the response amplifies, dampens, or leaves unchanged the fiscal pressure that an economic cycle places on families and on state budgets. The fourth axis is state discretion, meaning the scope of choices a state may exercise over eligibility, benefit levels, benefit form, program purpose, and administration, and the corresponding scope of federal prescription, prohibition, and penalty that bounds those choices. Each axis also interacts with the other three, because an entitlement without a funding rule is an abstraction, a funding rule without a discretion rule conceals the actor who adjusts when money and demand diverge, and cyclical behavior is the observable result of entitlement, funding, and discretion operating together under pressure. The comparison therefore proceeds axis by axis for clarity and returns to interaction for judgment, with judgment restricted to structural properties that a statute creates and that evidence about operation can confirm or refute.

A reader who holds the two statements, the named claim, and the four axes in view possesses a decision framework rather than a pair of portraits. A decision framework does not dictate a choice. A decision framework forces a choice to declare the price attached to the choice in the currency of the property surrendered. Selection of an entitlement matching structure purchases automatic response at the price of advance cost predictability. Selection of a fixed block grant structure purchases advance cost predictability at the price of automatic response. Selection of broader state discretion purchases local variation at the price of national uniformity. Prices of that kind cannot be negotiated away by adjectives. The sections that follow examine each axis in the order named, test the two models against the statutory text and the administrative record, and reserve any synthesis until the evidence for each axis stands complete on its own terms.

The Entitlement Axis: A Right Versus a Grant

The first axis of comparison between Aid to Families with Dependent Children and Temporary Assistance for Needy Families concerns the legal character of the benefit itself. Under one model, assistance was a right held by a family that satisfied defined criteria, enforceable against the agency that administered the program. Under the other model, assistance was a grant that a state could offer within a federally funded framework, without the federal statute conferring a corresponding right on any family. The distinction sounds abstract, but its consequences were mechanical. An entitlement determines what happens when an eligible applicant arrives after a budget allocation has been spent, what process an agency owes before aid ends, and what a court may order when an agency refuses to act. A grant structure assigns those same questions to legislative and administrative discretion. The 1996 replacement of the Aid to Families with Dependent Children program with Temporary Assistance for Needy Families reversed the answer on each point, and the reversal began with a single sentence of statutory text placed near the front of the rewritten statute.

Does TANF give families a legal right to assistance?

No. The statute enacted in 1996 stated that Title IV-A created no individual entitlement to assistance under a state program funded under that part. A family that met state criteria held no federal right to enrollment, payment, or continuation on that basis alone.

The Aid to Families with Dependent Children program rested on Title IV-A of the Social Security Act, and its entitlement character grew out of three interlocking features: a federal definition of the protected class, a mandatory state plan structure, and open ended federal matching funds. Congress defined the central eligibility category in federal law. Section 406(a) of the Act defined a dependent child as a child deprived of parental support or care by reason of the death, continued absence from the home, or physical or mental incapacity of a parent, and living with a specified relative. States exercised substantial discretion inside that federal frame. A state set its own standard of need, its own benefit levels, and its own income and resource thresholds, and state benefit amounts varied widely as a result. Federal law nevertheless fixed the outer boundaries of the class. A state could decide how generous its program would be, and a state could decline to participate in the program altogether, but a participating state could not redefine the federally protected category out of existence or attach eligibility conditions that federal law had not authorized. Eligibility was therefore a legal question with a federal floor of meaning, answered by applying federal and state criteria to an individual family, rather than a discretionary judgment about whether aid seemed warranted in a particular case.

The state plan requirement converted that federal definition into a duty owed to each eligible family. Federal funds flowed to a state only under a plan approved by the Secretary of Health, Education, and Welfare, later the Secretary of Health and Human Services, and the statute prescribed what such a plan had to contain. Among the mandatory provisions, section 402(a)(10) of the Act, codified before the 1996 repeal at 42 U.S.C. 602(a)(10), required that all individuals wishing to make application for aid have an opportunity to do so, and that aid to families with dependent children be furnished with reasonable promptness to all eligible individuals. The phrase all eligible individuals carried the weight of the entitlement. A plan that satisfied the statute promised aid to the entire eligible class, not to a subset selected by order of application, by locality, or by the remaining balance in an appropriation. The legislative history behind the reasonable promptness language confirms the point. A House Report from 1949 on the provision, quoted in Jefferson v. Hackney, 406 U.S. 535 (1972), recorded that shortage of funds had led some jurisdictions to refuse new applications or to keep eligible families on waiting lists until recipients left the rolls, and it described that difference in treatment among eligible people as a source of undue hardship inappropriate in a program financed from federal funds. The statutory design answered that practice directly: funds shortage was an administrative problem for governments to solve between themselves, not a defense against an eligible family.

Open ended federal financing made the promise fiscally coherent. The federal government reimbursed each state for a share of its assistance expenditures without an aggregate cap for the fifty states and the District of Columbia, at matching rates that varied with state per capita income between a statutory minimum of 50 percent and a statutory maximum of 83 percent. Because federal reimbursement expanded automatically as the eligible caseload expanded, a state that enrolled every eligible family drew additional federal dollars with each enrollment, and enrollment growth did not require a new appropriation decision before an eligible family could be served. The individual entitlement and the funding structure were two expressions of the same design choice. Neither contained a mechanism for rationing aid among eligible families, and each presupposed that the size of the program would be set by the number of families who met the criteria rather than by a predetermined spending total. A state agency administering the program therefore stood in a specific posture toward an applicant: the agency adjudicated eligibility, and where eligibility was established, payment followed as a legal consequence.

Two Supreme Court decisions from the years 1968 and 1969 illustrate how federal courts policed the boundaries of that entitlement against state rules that narrowed the eligible class. In King v. Smith, 392 U.S. 309 (1968), the Court considered an Alabama regulation that denied aid to children whose mother cohabited with an able bodied man, on the theory that the man counted as a substitute father and therefore as a nonabsent parent, regardless of whether the man was the father of the children, owed them a legal duty of support, or contributed to their support in fact. The Court held the regulation invalid as inconsistent with the Social Security Act, reasoning that Congress had determined that destitute children who were legally fatherless could not be denied federally funded assistance on the fiction of a substitute father. The decision treated the federal definition of dependent child as binding law that a state eligibility rule could not displace. In Shapiro v. Thompson, 394 U.S. 618 (1969), the Court addressed statutes in Connecticut, Pennsylvania, and the District of Columbia that denied assistance to otherwise eligible residents who had not lived in the jurisdiction for one year before applying. The Court held the durational residence requirements unconstitutional under the Equal Protection Clause, because the classification penalized the exercise of the constitutional right to travel between states and the asserted fiscal justifications did not supply the compelling interest that strict scrutiny demanded. Together the two decisions show the entitlement operating from two directions: federal statutory supremacy confined state definitions of eligibility in King, and constitutional doctrine confined state waiting periods in Shapiro.

Goldberg v. Kelly, 397 U.S. 254 (1970), supplied the procedural dimension of the entitlement and carried the furthest reaching holding of the three decisions. The plaintiffs were residents of New York City receiving aid under the federally assisted Aid to Families with Dependent Children program or under the New York Home Relief program, whose benefits officials had terminated or were about to terminate. At the time suit was filed, no prior notice or hearing of any kind was required before termination. Procedures adopted after suit began offered an informal pretermination review, including discussion with a caseworker and review by a supervisor, while a formal fair hearing remained available only after benefits had ended. The recipients argued that the combination denied procedural due process under the Fourteenth Amendment, and the District Court agreed that only a pretermination evidentiary hearing would satisfy the Constitution. On March 23, 1970, the Supreme Court affirmed by a vote of five to three, with one seat on the Court vacant following the resignation of Justice Fortas in 1969. Justice William J. Brennan Jr. delivered the opinion of the Court, joined by Justices Douglas, Harlan, White, and Marshall, while Chief Justice Burger and Justices Black and Stewart filed dissents.

The holding rested on the entitlement character of the benefit. The Court held that welfare benefits were a matter of statutory entitlement for persons qualified to receive them, that such an entitlement qualified as a form of property protected by the Due Process Clause, and that termination therefore required an evidentiary hearing before benefits ended. A post termination hearing, the Court concluded, was not an adequate substitute, because of what termination meant during the interval before review could occur. For qualified recipients, the opinion reasoned, public assistance provided the means to obtain essential food, clothing, housing, and medical care, and termination pending resolution of an eligibility dispute could deprive an eligible recipient of the very means by which to live while awaiting a decision. That reliance distinguished the welfare recipient from a discharged government employee or a taxpayer denied an exemption, and it drove the balancing that followed: the interest of the eligible recipient in uninterrupted receipt of assistance, joined with the state interest in avoiding erroneous termination, clearly outweighed the state concern to prevent an increase in fiscal and administrative burdens. Procedure, under this reasoning, was not a courtesy layered over the entitlement. Procedure was the mechanism that kept an entitlement from being withdrawn on an unverified assertion of ineligibility, and the timing of the hearing mattered as much as its existence.

The opinion also specified the elements that a constitutionally adequate pretermination hearing required, while declining to require a full judicial trial. The recipient was entitled to timely and adequate notice detailing the reasons for the proposed termination, and to an effective opportunity to defend before the decisionmaker, including the opportunity to present arguments and evidence orally and to confront and cross examine adverse witnesses. The recipient could retain counsel, although the state was not required to furnish counsel. The decisionmaker had to be impartial and could not have participated in making the determination under review, and the decision had to rest on evidence produced at the hearing, accompanied by a statement of reasons and of the evidence relied on. Each element answered a failure mode of informal caseworker review: notice prevented surprise, confrontation tested the factual basis of agency records, impartiality separated the accusing function from the deciding function, and a stated rationale made review possible. The hearing that Goldberg required was administrative in form, and its function was evidentiary discipline imposed on the agency before the entitlement could be withdrawn.

Federal regulation then built Goldberg and the fair hearing tradition into the routine machinery of program administration. The Social Security Act required an approved state plan to provide an opportunity for a fair hearing to an applicant whose claim was denied or not acted upon with reasonable promptness, and to a recipient aggrieved by agency action affecting assistance. The governing regulation at 45 C.F.R. 205.10 stated expressly that state hearings had to meet the due process standards set forth in Goldberg v. Kelly, and it elaborated those standards into operational detail. Every applicant and recipient had to be informed in writing, at application and at any action affecting a claim, of the right to a hearing, the method for obtaining one, and the right to representation by counsel, a relative, a friend, or another spokesman. Before discontinuance, termination, suspension, or reduction of assistance, the agency had to give timely notice, defined as notice mailed at least ten days before the date of action, and adequate notice, defined to include the intended action, the reasons for it, the specific regulations supporting it, the right to request a hearing, and the circumstances under which assistance would continue if a hearing was requested. The hearing itself had to be conducted by an impartial official who had not been directly involved in the initial determination, and the claimant had to have an adequate opportunity to examine the case file, present witnesses, establish pertinent facts, and confront and cross examine adverse witnesses.

Aid pending appeal completed the machinery and gave the entitlement its practical continuity. Under 45 C.F.R. 205.10, if a recipient requested a hearing within the timely notice period, assistance could not be suspended, reduced, discontinued, or terminated until a decision was rendered after the hearing, subject to defined exceptions and subject to recovery by the agency if the agency action was later sustained. A recipient who requested a hearing shortly after the date of action could obtain reinstatement of assistance under specified conditions, and where action had been taken without timely notice, a request made within ten days of the mailing of notice triggered reinstatement and continuation until decision when the dispute turned on facts rather than on a change in law or policy. The structure deserves emphasis because it allocated the risk of administrative error. Under the entitlement model, the risk of an erroneous termination fell on the government during the review interval: payments continued, and the agency bore the burden of recovering aid if its position prevailed. A recipient did not have to finance the appeal period from personal resources or persuade a court to restore aid after the fact. The fair hearing system was thus the administrative expression of the same principle Goldberg had constitutionalized, translated into deadlines, notice contents, and payment rules that caseworkers and hearing officers applied case by case.

The Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Public Law 104-193, signed on August 22, 1996, dismantled that structure at its foundation. The Act repealed the Aid to Families with Dependent Children program, rewrote Title IV-A to create Temporary Assistance for Needy Families, and stated the change in the opening section of the new part. Section 401(b) of the rewritten Title IV-A, codified at 42 U.S.C. 601(b) and headed No individual entitlement, provides in full: This part shall not be interpreted to entitle any individual or family to assistance under any State program funded under this part. The sentence is short, and its drafting is deliberate. The provision does not merely omit an entitlement that earlier law had conferred; the provision instructs every interpreter of the statute, including agencies and courts, against reading the part to create one. A separate transition provision of the 1996 Act reinforced the break by providing that, effective October 1, 1996, no individual or family would be entitled to benefits or services under a state plan approved under the former Title IV-A. The statutory basis on which Goldberg had rested, a legislatively created entitlement that counted as property, was removed for assistance under the new program by the legislature that had created the earlier version.

The rewritten state plan provisions matched the new legal character of the benefit. Under section 402 of the rewritten Act, codified at 42 U.S.C. 602, a state qualified as an eligible state by submitting a plan containing an outline of its family assistance program and a set of certifications, rather than the detailed mandatory commitments through which the former plan had promised aid to all eligible individuals with reasonable promptness. Federal law prescribed program purposes, work requirements, time limits, data reporting, and penalties directed at states, and those requirements bound states as conditions of the block grant. The obligations ran between governments. The rewritten part contained no counterpart to the former fair hearing entitlement as a matter of federal right, and no federal rule required a state to continue assistance pending an appeal of a termination decision. Whatever notice, hearing, or review a state provided under the new program rested on state law, state regulation, or state plan design, and the content of such protections varied as a matter of state choice rather than federal command. The procedural change cannot be separated from the substantive change, because the Goldberg hearing had derived its constitutional force from the existence of the entitlement. Removal of the entitlement removed the property interest on which the federal procedural guarantee had depended for assistance under the program, so substance and procedure fell together in a single drafting decision.

The practical consequences of the axis diverge at each position in the system. For a family, the entitlement model meant that satisfying the eligibility criteria was sufficient: application triggered a legal duty, delay was reviewable as a failure of reasonable promptness, denial generated a hearing right, and termination could not precede an evidentiary hearing at which aid continued in the interim. The grant model enacted in 1996 meant that satisfying state criteria established eligibility for consideration within a program whose capacity, categories, and continuation rules the state defined, and a family refused aid because a state had closed intake, narrowed a category, or exhausted an allocation had no federal entitlement claim to present. For a state agency, the entitlement model imposed a caseload that followed eligibility wherever eligibility led, with federal matching funds expanding in parallel, and it constrained administrative action through notice deadlines, hearing obligations, and continuation of aid during appeals. The grant model gave agency and legislature control over program size as a budgetary variable, funded through a fixed block grant rather than open ended matching, and it converted many disputes that had been individual adjudications into questions of program design. For a court, the entitlement model supplied both a rule of decision and a remedial theory: King permitted invalidation of a state eligibility rule that conflicted with the federal statute, Shapiro permitted constitutional review of exclusionary classifications, and Goldberg permitted an order requiring a hearing before termination. The grant model withdrew the statutory premise for such orders in assistance cases, leaving courts to police the boundaries that other law supplied, such as the conditions Congress attached to the block grant itself, state law, and constitutional provisions independent of a property interest in the benefit.

The entitlement axis therefore measures more than a change in vocabulary between two program names. Between 1935 and 1996, federal law built a structure in which a defined class of families held a right to aid, federal courts and federal regulators enforced the incidents of that right, and procedure protected the right at the moment of greatest vulnerability, the interval between an agency decision and its review. The 1996 Act replaced that structure with one in which the federal statute disclaimed the right expressly, funded the program through a fixed grant, and left the incidents of participation to state definition. Any comparison of the two welfare models that begins elsewhere, with benefit levels, work rules, or time limits, addresses features that operate downstream of this prior legal choice about whether assistance is owed to an eligible family at all.

The Funding Axis: A Matching Partner Versus a Fixed Sum

The funding mechanism of a welfare program allocates financial risk between two levels of government, sets the price a state pays for serving one additional family, decides whether federal spending responds automatically to changes in need, and establishes the budgetary consequences that follow when a caseload rises or falls. Aid to Families with Dependent Children and Temporary Assistance for Needy Families rested on opposite answers to each of those questions. The first program operated as an open ended matching grant in which the federal government acted as a partner in every expenditure a state made on behalf of an eligible family. The second program operated as a fixed block grant in which the federal government paid a predetermined sum and left the financial consequences of variation in need, prices, and caseloads to fall where the fixed sum placed them. The money formula defined the incentives that state administrators faced, the exposure that the federal budget carried, and the degree to which the program could expand or contract without a new act of Congress.

Under Aid to Families with Dependent Children, federal payment followed state expenditure as a matter of entitlement structure. A state operated a plan that met federal requirements, determined which families qualified under the program rules, paid aid to those families, and then claimed federal reimbursement for the federal share of the amount spent. The defining feature of the arrangement was the absence of a ceiling on the federal total for the fifty states and the District of Columbia. The Congressional Research Service described the matching grant for the predecessor program as open ended, with no limit to federal funding for the fifty states and the District of Columbia, while noting that limits applied in Puerto Rico, Guam, and the United States Virgin Islands. Because no national cap constrained the aggregate, the federal obligation was a function of state spending on eligible families rather than a figure fixed in advance. When more families qualified and received aid, total spending rose and the federal share rose in proportion to the applicable matching rate. When caseloads contracted, total spending fell and the federal contribution fell alongside the state contribution.

The share that the federal government reimbursed was not uniform across the country, and the variation followed a deliberate equalization logic. Aid to Families with Dependent Children used the Federal Medical Assistance Percentage, the same formula framework associated with Medicaid, to set the federal share for each state. The formula compared per capita income in a state with per capita income for the nation as a whole and moved in the opposite direction from state fiscal capacity. A state with per capita income above the national average received a lower federal share, and a state with per capita income below the national average received a higher federal share. The Congressional Research Service stated the resulting boundaries in its primer on the block grant that replaced the earlier program, explaining that the predecessor program reimbursed states at the Medicaid matching rate, which ranged from a statutory minimum of 50 percent for the states with the highest per capita incomes to a statutory maximum of 83 percent for the states with the lowest per capita incomes. The federal treasury therefore carried at least half of every dollar of aid in every state and carried more than four fifths of every dollar in the states with the least fiscal capacity.

The arithmetic of the matching rate governed the marginal cost of assistance. Consider a benefit payment of one hundred dollars in a state matched at the statutory minimum of 50 percent. The federal government paid fifty dollars of that payment and the state paid fifty dollars. Consider the same one hundred dollar payment in a state matched at 80 percent, a rate within the statutory range toward the upper end. The federal government paid eighty dollars and the state paid twenty dollars. At the statutory maximum of 83 percent, the state share of the same payment fell to seventeen dollars. A state therefore never faced the full price of an additional case. The price of enrolling one more eligible family, paying aid for one more month, or maintaining a benefit at a given level rather than reducing the benefit was a fraction of the total expenditure, and the fraction varied predictably with the matching formula. Federal policy lowered the state price of assistance everywhere and lowered the price most where state resources were thinnest. That price structure did not dictate any particular benefit level, since states retained discretion over payment standards, but each state choice cost the state less than the choice was worth in total program terms, with the difference carried by the federal treasury.

The same arithmetic operated in reverse. A state that reduced spending on aid surrendered federal money in the process. A reduction of one hundred dollars in assistance in a state matched at 50 percent saved the state fifty dollars and withdrew fifty federal dollars from the program in that state. A reduction of one hundred dollars in a state matched at 80 percent saved the state only twenty dollars while withdrawing eighty federal dollars. Economizing therefore carried a measurable opportunity cost, because every state dollar withdrawn from assistance removed more than a dollar of total program resources. The matching design rewarded expenditure with a federal contribution and penalized retrenchment with the loss of that contribution. The formula also meant that federal spending control was indirect. Congress set the matching rates and the program rules, yet the resulting federal total emerged from the sum of state decisions and caseload outcomes, and no fixed line in a federal budget table could state the total in advance with precision.

Caseload shocks illustrate the risk allocation that the matching formula produced. A recession that threw parents out of work or a plant closure that eliminated employment in a locality translated into higher claims for aid, higher state expenditure, and a higher federal reimbursement that arrived through the operation of the formula rather than through emergency legislation. The federal treasury absorbed a share of the shock ranging from 50 percent to 83 percent depending on the state, and the state budget absorbed only the remainder. The absorption was automatic, since the formula required no discretionary decision to release additional federal funds when need expanded. The converse held in a period of falling unemployment and shrinking caseloads, when state expenditure declined and the federal contribution declined with it, returning savings to both levels of government in proportion to the matching rate. The arrangement therefore behaved as a countercyclical instrument at the federal level, expanding federal outlays when economic conditions generated more eligible families and contracting federal outlays when conditions generated fewer eligible families.

Temporary Assistance for Needy Families replaced that entire logic with a grant whose amount was fixed in statute and detached from the volume of assistance provided in any given year. The core of the federal commitment became the state family assistance grant under section 403(a)(1) of the Social Security Act, a basic block grant paid to each state in a predetermined amount. The national total stood at approximately 16.5 billion dollars a year. Precision about the total requires attention to definition, because published figures combine components in different ways. Tables prepared by the Congressional Research Service recorded the state family assistance grant for the fifty states, the District of Columbia, tribes, and territories at 16.567 billion dollars for fiscal year 1997 through fiscal year 2016, with the total reduced beginning in fiscal year 2017 by 0.33 percent to finance research and technical assistance activities connected with the program, producing a recorded total of 16.512 billion dollars for fiscal year 2017 and fiscal year 2018. Accounts of the grant that confine the figure to the state grants alone have used 16.489 billion dollars. The differences among those figures reflect the components included in each accounting rather than disagreement about the central structural fact, which is that the amount was set once and then held constant in nominal terms regardless of the conditions the program encountered.

The basis on which each state share of the fixed total was calculated froze a particular moment of program history into the allocation formula. Under the original formula established by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, the grant for each state rested on the federal share of expenditures in the predecessor programs, principally Aid to Families with Dependent Children together with the Emergency Assistance program and the Job Opportunities and Basic Skills Training program. The Congressional Research Service explained that the formula entitled each state to the greatest of three amounts: the average federal share of expenditures in those programs for fiscal year 1992 through fiscal year 1994, the federal share of expenditures for fiscal year 1994 with an adjustment for states that had amended Emergency Assistance programs in fiscal year 1994 or fiscal year 1995, or the federal share of expenditures in fiscal year 1995. Taking the greatest of the three measures anchored each state allocation at or near the peak of federal spending in that state during the early to middle years of the decade. An allocation derived from spending in fiscal year 1992 through fiscal year 1995 could reflect the caseload levels, benefit decisions, and economic conditions of that period, yet the allocation carried forward into later years without recalculation against the conditions of the years in which the money was actually paid.

Because the grant amount was fixed in nominal dollars, the usual channels through which a funding formula might respond to change were closed by design. The Congressional Research Service stated in its responses to frequently asked questions about the block grant that the basic amount was not adjusted for changes that occur over time, naming inflation, the size of the assistance caseload, and changes in the poverty population among the factors for which no adjustment was made. Each excluded factor corresponds to a distinct form of drift between the grant and the need the grant supports. A rise in the caseload increased the number of families among whom a state might spread the same federal sum. Inflation reduced the purchasing power of every dollar in the grant even in a year when caseload and population held steady. A state whose caseload doubled faced the arithmetic with the same federal allotment held before the increase, and a state whose caseload halved faced the arithmetic with the same federal allotment held before the decrease. Federal payment and program volume were disconnected variables, related only through the spending decisions a state made within a fixed envelope of federal money supplemented by state funds. Capped federal allotments as an instrument also appeared in children’s health coverage, because the Children’s Health Insurance Program of 1997 used capped federal allotments, a related instrument applied to children’s health coverage, and the CHIP program guide traces that design in full.

The principal qualification on the fixed character of the federal grant concerned state money rather than federal money, and the qualification took the form of a maintenance of effort requirement. Where the matching design had induced state spending by sharing each additional dollar, the block grant design compelled a floor of state spending by requiring each state to continue expending its own funds at a specified fraction of a historical baseline. The baseline was state spending in fiscal year 1994 on the predecessor programs, and the required fraction was 80 percent of that baseline as the general rule. The required fraction fell to 75 percent of the fiscal year 1994 baseline for a state that met the federal work participation rates. The Congressional Research Service stated the requirement in those terms in successive editions of its primer and its responses to frequently asked questions. The maintenance of effort requirement therefore preserved a state financial presence in the program without recreating the matching relationship. Spending above the floor drew no additional federal dollars, because the federal grant was already fixed.

What happens to federal funding when welfare caseloads fall?

Under Aid to Families with Dependent Children, federal payments fell when caseloads fell because each payment matched state spending on eligible families. Under Temporary Assistance for Needy Families, the basic federal grant held at a fixed amount, so a smaller caseload left federal money unchanged and available for other allowable purposes.

The grant amount, the maintenance floor, and the related funding provisions of the replacement model receive a dedicated treatment in the TANF block grant structure.

The maintenance of effort floor and the fixed federal grant together redefined the budget identity within which a state operated. Under the matching program, total program resources equaled state spending plus a federal contribution that scaled with state spending, so total resources rose and fell with the expenditure decision itself. Under the block grant, total program resources equaled a fixed federal grant plus whatever state spending the state chose to make at or above the floor, so the federal component stayed constant while the state component varied within a band bounded below by the floor. The difference determined which government bore the cost of the next case. Under the fixed grant, the federal payment for one more family receiving assistance was zero at the margin. A state that served an additional family financed that service entirely from the fixed grant envelope or from state funds. A state that served one fewer family released the full amount of that assistance for other uses permitted by the statute, because no federal reimbursement was withdrawn when spending declined. The marginal price of assistance to the state had risen from a fraction of the benefit under matching to the whole benefit under the block grant, measured against the fixed resources available.

Inflation converted the nominal freeze into a real decline of a magnitude that can be stated precisely because the Congressional Research Service performed the calculation on a consistent basis. The service computed constant dollar values for the state family assistance grant using the Consumer Price Index for All Urban Consumers, the price index maintained by the Bureau of Labor Statistics. On that basis, the Congressional Research Service calculated that by fiscal year 2018 the fixed grant had lost 36.2 percent of its fiscal year 1997 real value. The table behind the calculation recorded the grant at 16.567 billion dollars in fiscal year 1997 and recorded the fiscal year 2018 amount, stated in constant fiscal year 1997 dollars, at 10.564 billion dollars, with the cumulative percentage change given as 36.2 percent. The decline required no vote, no amendment, and no administrative decision in any intervening year. A grant held constant in nominal terms in an economy with positive inflation loses real value by arithmetic necessity. The figure of 36.2 percent therefore measures a structural property of the fixed grant form over a twenty one year span rather than the effect of any single budget action.

The erosion figure also clarifies the difference between federal budget predictability and program resource stability, two properties that a funding mechanism can trade against each other. The fixed grant gave the federal government a predictable obligation, known in advance, repeated from year to year, and insulated from caseload movements that had made the matching total difficult to forecast. The same property produced instability in the real resources available per unit of need at the state level. A state planning assistance, child care support, work related services, or administrative capacity against a fixed nominal grant faced a resource base whose real value declined each year by roughly the rate of inflation, compounded over time, and whose value relative to the caseload moved with caseload changes that the grant ignored. Predictability was not eliminated from the system. Predictability was relocated, granted to the federal payer in full and withdrawn from the state operator in proportion to inflation and caseload volatility.

The conversion named in the comparison between the two models follows directly from those mechanics. The fixed grant converted the federal government from a partner sharing marginal cost into a payer of a fixed sum. A partner sharing marginal cost participates in every decision at the margin, because each additional dollar of program activity draws a federal contribution and each dollar withdrawn surrenders one. A payer of a fixed sum participates in none of the marginal decisions, because the payment is complete before the decisions are made and unchanged after the decisions are made. Under the matching program, a falling caseload cut the federal contribution automatically, and the saving from a smaller caseload was divided between the state and the federal government according to the matching rate. Under the block grant, a falling caseload left federal money unchanged in state hands, and the saving from a smaller caseload accrued entirely within the state program, freeing funds for other uses within the purposes of the statute. Expenditure restraint under matching saved a state only its fractional share and cost the state program the federal share. Expenditure restraint under the block grant saved the full amount of the reduction for redeployment and cost the state program nothing in federal funds.

Several provisions of the block grant interact with the funding structure and receive extended treatment in the companion account of the program design, so the funding axis requires only a passing acknowledgment of the points of contact. Work participation rates influenced the maintenance of effort percentage a state had to meet, as the reduction from 80 percent to 75 percent of the fiscal year 1994 baseline depended on meeting those rates. A caseload reduction credit adjusted the participation standard against which a state was measured. Transfer rules governed movement of funds between the block grant and related programs and therefore affected the composition of spending possible within the fixed total without altering the total itself.

Read as a decision framework, the funding axis poses a compact set of questions that can be applied to either model. The first question asks who pays for the next case, and the matching formula answers that both governments pay in a stated proportion while the block grant answers that the state pays from a fixed envelope. The second question asks who absorbs a shock, and the matching formula assigns the larger share of a caseload shock to the federal treasury while the block grant assigns the shock to the state budget and to the allocation choices made within the fixed grant. The third question asks what happens to federal money when need declines, and the matching formula withdraws the federal contribution in proportion to the decline while the block grant leaves the federal money in place for other statutory uses. The fourth question asks how the value of the federal commitment behaves over time, and the matching formula ties the commitment to expenditure volume while the block grant ties the commitment to a nominal figure whose real value declines at the rate of inflation. On the funding axis alone, the distance between the two models is the distance between a shared price and a fixed sum, between a federal contribution that breathed with the caseload and a federal contribution that held its nominal shape while prices, populations, and caseloads moved around the shape.

The Cyclicality Axis: What Happens When the Economy Falls

The cyclicality axis asks a narrow fiscal question about each model. When employment contracts and household earnings fall, does the financing structure expand on its own, contract on its own, or hold at a level set in advance by statute? The answer follows from the funding formula rather than from the intentions of administrators or the behavior of individual families. Aid to Families with Dependent Children and Temporary Assistance for Needy Families gave opposite answers because the two programs moved federal money through opposite mechanisms. One mechanism tied federal spending to the number of eligible families a state served. The other mechanism fixed federal spending in advance and left the connection between need and resources to separate decisions by states and by Congress. Under the first model, a recession entered the formula directly and spending responded as a matter of arithmetic. Under the second model, the shock arrived outside the formula, and any federal response required an additional act of government. The recession that began in December 2007 supplied the clearest demonstration because a matching entitlement in food assistance operated alongside the fixed grant in cash assistance during the same downturn, in the same states, for overlapping populations.

The design claim for a matching entitlement is that countercyclical expansion is built into the structure. Under Aid to Families with Dependent Children, the federal government reimbursed a share of each dollar a state spent on benefits for families that met the eligibility rules in federal and state law. The federal share varied inversely with state per capita income, so poorer states received a higher matching rate than wealthier states, but in every state the federal commitment was open ended. No appropriation capped the total. If a downturn pushed more families below the income thresholds, the number of aided families rose, state spending rose with the caseload, and federal spending rose automatically at the matching rate. No vote, no supplemental appropriation, and no administrative finding of emergency was required for the money to move. The program functioned as an automatic stabilizer in the fiscal sense of that term. Federal outlays increased during the contraction, when household income and state tax receipts fell, and caseload pressure eased as employment recovered. The mechanism guaranteed responsiveness, not adequacy. The caseload could rise only through the doorway the eligibility rules defined, and a legislature that narrowed the doorway during a downturn could blunt the cyclical response without changing the matching structure itself.

The historical behavior of Aid to Families with Dependent Children followed the mechanism with one instructive complication. A Department of Health and Human Services appendix on program history recorded the general pattern in plain terms, noting that enrollments and benefit outlays generally tended to increase in times of economic recession and decline in times of economic growth. The recession of the early 1980s complicated the pattern because two forces acted at once. Job losses in the downturn that began in 1980 pushed rolls upward, and then the Omnibus Budget Reconciliation Act of 1981 narrowed eligibility at the same moment, restricting the earned income disregard and imposing a gross income limit tied to each state standard of need. Research prepared for the Council of Economic Advisers described the result as a paradox. The severe recession of 1981 to 1982 coincided with a decline in the rate of welfare receipt, and the explanation offered was the 1981 legislation, which reduced eligibility exactly when a large increase in receipt would otherwise have been expected. The episode refines the design claim rather than refuting the design claim. A matching entitlement responds automatically to need as the statute defines need. When Congress rewrote the definition in the middle of a recession, the automatic response operated on the narrower definition, and the measured caseload reflected the new rules as faithfully as the formula reflected the economy.

The recession of 1990 to 1991 and the weak labor market that followed it displayed the mechanism without a simultaneous narrowing of the rules, and the scale of the response was large. The caseload climbed from roughly 3.7 million families in 1989 to a peak of 5.033 million families and 14.161 million recipients in calendar year 1994, figures the Congressional Research Service and the Department of Health and Human Services reported for the program. The timing carries a lesson about lags. The recession ended in the spring of 1991, yet the Congressional Research Service account of the later Emergency Contingency Fund recorded that welfare caseloads peaked in March 1994, almost three years after the recession ended. Eligibility follows earnings and employment with delay, and a weak recovery leaves incomes below the thresholds long after output resumes growth. An entitlement therefore expands late and contracts late. The fiscal response is automatic, but automatic does not mean immediate. The lag meant that aid remained available while household earnings had not recovered, without any official having to certify that hardship persisted, and the same lag kept federal and state costs rising into the recovery.

Does TANF funding increase during a recession?

Federal TANF funding does not increase automatically during a recession because the basic grant is fixed in statute at 16.5 billion dollars per year. Additional federal money reaches states only through separate congressional action, a limited contingency mechanism, or balances a state reserved from earlier grant years.

The fixed grant inverted the fiscal logic of the predecessor program. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 set a basic block grant amount in statute, and the Congressional Research Service recorded that the amount remained the same from fiscal year 1997 through fiscal year 2016, at 16.6 billion dollars in total for the states, the District of Columbia, the territories, and tribal programs, with the state family assistance grant standing at 16.5 billion dollars after a small reduction that began in fiscal year 2017. The amount was not adjusted for inflation, for the size of the assistance caseload, or for changes in the population in poverty. Each state share rested on federal funding received in the middle 1990s, a period of historically high caseloads, rather than on any measure of need in the year the money was spent. A recession therefore produced no signal inside the formula. A state could experience a doubling of unemployment while the federal grant on the state ledger remained identical to the grant received during the preceding expansion. The Brookings Institution summary of the block grant structure stated the contrast directly. Federal expenditures under the old program showed cyclical variation because matching followed the caseload, while federal expenditures under the block grant were fixed and neither adjusted for inflation nor rose and fell with economic cycles or the size of the caseload.

The position of the states inside a fixed grant turned a neutral federal formula into a potentially procyclical one. State revenues fall during a recession because tax receipts contract with the economy, and state constitutions and statutes generally require balanced operating budgets. A state facing a revenue shortfall and a fixed federal grant confronts choices that point toward restraint. The state can draw on balances, shift funds among the purposes the block grant permits, raise revenue, reduce spending on other programs, or restrict assistance through eligibility rules, benefit levels, or time limit enforcement. Nothing in the structure rewards expansion at the moment need rises, and the maintenance of effort requirement, which obliges states to spend a specified share of a historical state spending level, is also fixed in nominal terms and supplies no countercyclical pressure. The predecessor formula had pushed in the opposite direction at exactly the critical moment, because each additional state dollar spent on an eligible family drew down additional federal dollars, lowering the effective state price of expansion during a downturn. The block grant removed that price effect. Each dollar committed to cash assistance during a recession was a dollar unavailable for child care, work programs, or reserves within the same fixed pool. Contraction of aid as need rises is not a required outcome under a block grant, and some states expanded during the downturn that began in 2007. Contraction is, however, a permitted and fiscally natural outcome, because predictability for federal budgeting was the property the structure optimized, and predictability and responsiveness trade against each other in the construction of the formula.

Congress recognized the cyclical exposure at enactment and attached a partial corrective device, the Contingency Fund, and the history of that device illustrates the difference between a stabilizer and a reserve appropriation. The Center on Budget and Policy Priorities account of the fund recorded that Congress created the Contingency Fund in 1996 to address some of the risks states would face from converting an entitlement whose funding rose automatically in recessions into a block grant with a fixed federal funding level. The original appropriation was 2 billion dollars, access depended on state economic triggers and on additional state spending, and the fund was finite. A finite fund can soften a downturn of ordinary depth and duration, but a finite fund exhausts. The Center on Budget and Policy Priorities recorded that the original 2 billion dollars was used up by December 2009 and that over one third of the states relied on the fund during the downturn. Exhaustion in December 2009 meant that the built in corrective reached its limit roughly two years into a downturn whose labor market effects extended far beyond that date. An automatic stabilizer has no corresponding limit, because the appropriation is the formula itself and the federal share is paid for as long as eligible families present themselves and states spend the matching dollars.

The recession that began in December 2007 tested both structures at once, because food assistance retained the matching entitlement form while cash assistance operated under the fixed grant. The Center on Budget and Policy Priorities analysis by LaDonna Pavetti, Danilo Trisi, and Liz Schott measured the divergence over the first two years of the downturn. On the measure that analysis used, Supplemental Nutrition Assistance Program caseloads grew by 45 percent between December 2007 and December 2009, while Temporary Assistance for Needy Families caseloads increased by just 13 percent over the same December to December span, and the number of unemployed persons doubled across the period. The measurement basis matters and the analysis stated the basis explicitly. Pavetti, Trisi, and Schott used caseload data collected directly from the states rather than data reported by the states to the Department of Health and Human Services, in order to reflect families receiving cash assistance consistently over time and across states, because changes in state funding configurations after the Deficit Reduction Act of 2005 had made the federally reported series inconsistent as a count of families relying on cash assistance. The 45 percent and 13 percent figures therefore rest on a consistent December to December comparison constructed for analytical purposes, and the gap between the figures expresses the difference in program structure under a common shock.

A second measurement, on a different basis, produced a close but noninterchangeable figure. The Ninth Report to Congress from the Department of Health and Human Services recorded that caseloads declined slightly throughout fiscal year 2007 and continued to decline from the official start of the recession in December 2007 through July 2008, and that caseloads then began to rise. From the low in July 2008 through September 2009, close to 199,000 families were added to the rolls, representing an increase of 12 percent on the Department of Health and Human Services measure. The 12 percent figure and the 13 percent figure describe related movement from different starting points, over different spans, on different data series. The Department of Health and Human Services figure measures growth from a July 2008 trough to September 2009 in the federally reported caseload. The Center on Budget and Policy Priorities figure measures growth from December 2007 to December 2009 in a state collected caseload constructed for consistency. For the first seven months of the recession the federally reported caseload fell, a movement opposite to the direction of need, because patterns established during the expansion continued while the fixed grant supplied no counterweight. The state level pattern reinforced the structural reading. In 22 states, caseloads responded very little or not at all on that measure, with 16 states recording increases of less than 10 percent and six states recording declines across a period in which unemployment doubled nationally. A fixed grant administered through fifty state decisions produces fifty outcomes, and the distribution of those outcomes, clustered near zero, is itself evidence about the mechanism.

Reserves supplied the one internal buffer the block grant design contained. Federal law permitted states to reserve unused grant funds without a time limit, and the Congressional Research Service explained the rationale in responses to frequently asked questions about the block grant. The ability to reserve permitted flexibility in the timing of spending, including the ability to save funds for unexpected occurrences that might increase costs, such as recessions or natural disasters. Balances did accumulate during the early years of the program, when caseloads fell far below the middle 1990s levels on which grant amounts had been based. Those unobligated balances functioned as a partial buffer in the downturn that began in 2007, because a state holding a balance could finance additional assistance without new federal money. The buffer was partial and uneven for structural reasons. Balances varied greatly among the states, a point made in congressional consideration of reserve policy as early as 2001, when reported balances were described as suggesting that some states might not be as prepared as other states to address the fiscal effects of a downturn without additional federal assistance. A state that committed grant funds to child care or other permitted purposes during an expansion entered a recession with a smaller reserve than a state that left funds unspent, and balances already obligated to contracts and services were not available for new commitments. A rainy day reserve inside a block grant is therefore a stock, accumulated by restraint in good years and depleted by use in bad years. An entitlement stabilizer is a flow, generated by the formula in the bad year itself, and the flow continues for the duration of the eligibility the law defines.

The exception that demonstrated the mechanism arrived as separate legislation. The American Recovery and Reinvestment Act of 2009 created an Emergency Contingency Fund within the block grant, and the Congressional Research Service summary of the fund fixed the essential facts. The fund provided 5 billion dollars, operated for fiscal years 2009 and 2010, and expired on September 30, 2010, with the full 5 billion dollars awarded by the expiration date. The fund reimbursed 80 percent of increased state spending in three categories measured against a base year, namely basic assistance, nonrecurrent short term benefits, and subsidized employment, and each state could qualify for an amount up to 50 percent of the state annual block grant from the emergency and preexisting contingency sources combined. The Center on Budget and Policy Priorities recorded that every state except Wyoming drew on the fund to cover increased costs in the three categories during the downturn. The design of the fund is the analytical point. For the fund to operate, Congress had to appropriate money outside the basic grant, define a reimbursement rate that restored a matching incentive, specify base years against which increases would be measured, and set an expiration date. Each element replicated, temporarily and partially, a property the entitlement had possessed permanently and generally. The 80 percent reimbursement rate once again lowered the state price of expanding aid, and states responded to that price. Absent the incentive and the money, the basic grant produced the modest and uneven caseload movement documented for the same period.

Subsidized employment supplied the most visible product of the emergency fund and the clearest demonstration of what happened when temporary countercyclical financing ended. The Center on Budget and Policy Priorities account of subsidized employment under the fund reported that 39 states, the District of Columbia, Puerto Rico, the Virgin Islands, and eight tribal programs received approval to use 1.3 billion dollars from the fund to create new subsidized employment programs or expand existing programs, and that the fund placed more than 260,000 low income adults and youth in paid jobs during a period of high unemployment. The fund expired on September 30, 2010 while unemployment remained elevated, and the subsidized employment activity financed by the fund ended with the financing. The Center for Law and Social Policy account described a large county program in which the subsidized employment effort came to an end on September 30, 2010 with the expiration of the fund, leaving over 6,000 workers in that county program unemployed again at the point of expiration. The Congressional Research Service recorded that the House of Representatives twice passed bills in 2010 to extend the fund and that none received approval in the Senate. A discretionary stabilizer operates for exactly the period a legislature specifies and then stops, regardless of labor market conditions on the date of expiration, while an automatic stabilizer operates for exactly the period eligibility persists under the formula.

Set against each other across two downturns, the two models produced fiscal responses that matched the construction of each formula. In the recession of 1990 to 1991, the matching entitlement expanded by roughly one third from the 1989 base to the 1994 peak, continued expanding for almost three years after the recession ended, and carried federal and state spending upward together at the matching rate without any emergency legislation directed at the caseload. In the recession that began in December 2007, a severe downturn in which the number of unemployed persons doubled on the Center on Budget and Policy Priorities measure, the fixed grant produced a caseload increase of 13 percent on the state collected December to December measure and 12 percent from the July 2008 low on the Department of Health and Human Services measure, against a 45 percent increase in food assistance caseloads over the December 2007 to December 2009 span. Reserves moderated the difference for states holding balances and failed to moderate the difference for states without usable balances. The Contingency Fund exhausted the original 2 billion dollar appropriation by December 2009. Only a new 5 billion dollar statute moved substantial countercyclical money through the cash assistance structure, and that movement ceased on the date written into the statute. The entitlement converted a shock into spending through arithmetic that no official controlled in the short run, subject always to the eligibility rules a legislature had set and could reset, as the legislation of 1981 demonstrated. The block grant converted a shock into decisions, by state administrators allocating a fixed pool under falling revenues and by Congress deciding whether to legislate an exception. Decisions can be generous or restrictive, prompt or delayed, sustained or brief. Arithmetic has only the quality of operating, and the presence or absence of that quality is the measurable difference between the two welfare models when the economy falls. The employment, caseload, and poverty record behind those fiscal responses receives a separate assessment in the account of welfare reform outcomes.

The Discretion Axis: Who Decides, and How Much

The discretion axis asks where the operative decisions sat in each program. Under Aid to Families with Dependent Children, the federal government defined the right, the covered categories, and many rules for counting income, and the states set the dollar amounts and ran the offices that applied federal definitions to individual families. Under Temporary Assistance for Needy Families, the federal government supplied a block grant, a set of purposes, and performance and reporting conditions, and the states acquired authority to define much of the program a family encountered. Each program divided authority. The division moved, and the movement changed what variation between states could mean.

Did states have any discretion under AFDC?

States held substantial discretion under Aid to Families with Dependent Children. Each state set its need standard, payment standard, and maximum benefit, administered cases, and chose among federally authorized options. Federal law fixed covered categories, eligibility boundaries, earnings rules, hearing protections, and quality control, so discretion operated inside an entitlement structure.

That answer corrects a recurring error in accounts of the 1996 change. A description that treats the earlier program as a uniform national benefit administered by states as clerks misses the largest source of interstate difference in the program as it operated in the early and middle 1990s. Benefit levels were state decisions, and the decisions diverged widely. The Administration for Children and Families, in a February 1996 fact sheet reporting figures for 1994, stated that benefits for a family of three ranged from 120 dollars per month in Mississippi to 923 dollars per month in Alaska, with the median state paying 375 dollars. The table of need standards and maximum benefits for January 1994 in the House Committee on Ways and Means Green Book recorded the same maximum benefits at the two ends of the distribution.

The dollar benefit was the product of more than one state-set figure. Federal law required each state to establish a standard of need for families with children who had no other means of support. A state also determined a payment standard in a separate process, and the payment standard could sit below the need standard. A state could then apply a maximum benefit below both figures. A state could accordingly acknowledge one level of need in its standard and fund a lower level in its payment, and the gap between the two measures was itself a policy choice rather than a federal instruction. The January 1994 Green Book table illustrated the spread in the underlying standards as well. The need standard for a family of three stood at 673 dollars in Alabama and at 975 dollars in Alaska in that table.

Administration added a further layer of variation. The program ran through state agencies, and in many states through county offices that took applications, verified circumstances, computed grants, and conducted redeterminations. Federal statute and regulations framed the questions an office asked, and federal definitions limited the answers that could support a denial. Within that frame, administration involved the organization of offices and the handling of documentary proof.

The limits defined the AFDC side of the axis as sharply as the grants of discretion did. A state could not deny aid to a family that met the federal eligibility definitions and fell within the income and resource limits the state itself had set. The Green Book summary of the basic federal rules expressed the resulting position in terms drawn from court rulings, stating that states were required to provide aid to all persons who were in classes eligible under federal law and whose income and resources were within state-set limits. The state controlled the height of the financial threshold through its standards, subject to federal outer limits, and the state controlled the amount paid. The state did not control the existence of the obligation once a family satisfied the federally defined class and the state financial tests.

The covered categories were likewise federal. Payments were allowed for a needy child who had been deprived of parental support or care because a parent was continuously absent from the home, incapacitated, deceased, or unemployed, and payments were also permitted for the needy caretaker relative with whom the child lived. Deprivation was therefore not a label a state could attach to any hardship a legislature wished to reach, and it was not a label a state could withhold from a family whose circumstances met the federal grounds. Federal law also fixed the outer age boundary, ending eligibility at a child’s eighteenth birthday, with a state option to extend aid to the nineteenth birthday for a full-time student in a secondary or technical program. A state chose between federally specified outcomes, but the menu was federal.

Procedure reinforced the same division. In Goldberg v. Kelly, decided in 1970 and reported at 397 United States Reports 254, the Supreme Court held that a recipient could not be deprived of welfare benefits without a pre-termination evidentiary hearing. The decision treated benefits as a statutory entitlement for persons qualified to receive them, and it required an opportunity to appear before an impartial decision maker, present evidence, and confront adverse witnesses before termination. A state administering AFDC therefore lacked authority to end aid on an untested caseworker determination and leave the family to a remedy after the grant had stopped. The hearing requirement did not set the benefit level or decide the federal categories. It constrained the manner in which a state could act against a person inside those categories.

Federal financial oversight supplied another boundary, one that operated on states rather than on families. The AFDC quality control system measured payment errors in state administration, comparing amounts paid with amounts federal rules required, and counting both overpayments and underpayments in the assessment of error. The Omnibus Budget Reconciliation Act of 1989 established a revised quality control arrangement that imposed penalties on states on a sliding scale reflecting the degree to which a state error rate exceeded the national average, and the arrangement also provided a Quality Control Review Panel for disputes between states and the federal government. Earlier quality control practice had likewise exposed a state to a reduction in federal financial participation when error reduction targets went unmet. The incentive created by that structure ran toward strict application of federal eligibility and payment rules.

Federal rules on earnings show the boundary in detailed form, because the treatment of a working recipient’s income determined both eligibility and grant size month by month. For a long period the earned income disregard allowed a recipient to keep the first 30 dollars of earnings plus one third of the remainder without a corresponding reduction in the grant, and the disregard operated without the short time limit later law imposed. The Omnibus Budget Reconciliation Act of 1981 restricted the 30 dollars and one third disregard to four months, and required a recipient to be off AFDC for twelve months before becoming eligible for the disregard again. The Deficit Reduction Act of 1984 then settled the disregard into a sequence under which the full 30 dollars and one third treatment applied for four months and a 30 dollar disregard applied for the following eight months. A summary prepared by the Office of the Assistant Secretary for Planning and Evaluation recorded that families with incomes greater than 150 percent of a state standard of need were no longer eligible effective July 1, 1981, and that the cutoff was raised to 185 percent in 1984. No state could rewrite those sequences. A state could raise or lower its need standard, which changed the dollar point at which a percentage ceiling operated. The percentage, the disregard amounts, and the durations belonged to Congress.

The work program layer of late AFDC complicates any account that reserves behavioral conditions for the TANF period. The Family Support Act of 1988, Public Law 100-485, enacted on October 13, 1988, created the Job Opportunities and Basic Skills Training program, known as JOBS, in a new Part F of Title IV of the Social Security Act, replacing the earlier Work Incentive program. An Assistant Secretary for Planning and Evaluation history recorded that the Family Support Act required states, to the extent resources allowed, to engage most mothers with no child below age three in education, work, or training under JOBS. The Institute for Research on Poverty summary of the statute described a structure in which the state agency assessed a participant’s educational needs, child care and supportive service needs, work experience, and skills, and then developed an employability plan with the individual describing the activities in which the individual would participate and the services to be provided. Participation expectations, assessments, and plans therefore existed inside the entitlement structure during the final eight years of AFDC.

The significance of JOBS for the discretion axis lies in the direction of the steering. The federal government used a federally created program with federally approved state plans to press recipients toward education, training, and employment, and the state designed and operated the local version with considerable leeway over services and sequencing. The entitlement to cash aid for a family meeting federal definitions stood behind the work program, and neither development transferred the definition of the cash right to the state. Federal behavioral steering and state program discretion were therefore both present before 1996. The distinction lies in how much of the underlying cash program moved into state hands.

The TANF statute enacted in the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 moved the definition itself. States acquired broad authority to set eligibility rules for assistance, including income thresholds, resource rules, treatment of earnings, and conditions attached to receipt, and states retained authority over benefit levels, exercised under a fixed federal block grant rather than under an open-ended federal matching commitment for allowable payments. Because the federal payment no longer rose automatically with allowable state spending on eligible families, a state decision to widen eligibility or raise benefits carried a different budgetary meaning. The state allocated a capped grant and state maintenance funds among competing uses, so discretion over eligibility and discretion over spending became parts of the same decision.

Sanction policy supplies the clearest single illustration of the widened authority. Federal TANF law required a state to reduce or terminate assistance for a family with a member who refused to engage in work required by the state without good cause, and the Congressional Research Service primer on the block grant stated that states were free to determine the sanction amount and whether to reduce or terminate benefits for a family that failed to comply, a termination described as a full-family sanction. States also determined what counted as good cause for noncompliance. The AFDC practice for work related noncompliance had taken a different form. A Brookings Institution account of sanction history recorded that, until the early 1990s, sanctions did not involve terminating a family’s entire AFDC grant, and that the individual who failed to comply, usually the parent, was removed from the grant calculation, producing a lower grant. For the work sanction itself, the structural contrast holds. Under TANF a state could select a partial reduction, a removal of the adult portion, or a termination of the whole family grant, and could set the duration and the steps for curing the sanction.

Time presented a second field in which state authority operated inside a federal outer boundary. Federal law barred the use of federal TANF funds for assistance to a family containing an adult or a minor child head of household who had already received sixty cumulative months of federally funded assistance, under section 408(a)(7) of the Social Security Act, codified at 42 United States Code 608(a)(7). A state could adopt the sixty month point as its own limit, and a state could also set a shorter period. Arizona supplied the shortest example in the pre-2019 record, having enacted a twelve month lifetime limit in 2015 that took effect in 2016, under which assistance ended after twelve months of assistance in a lifetime. A state could also continue aid beyond sixty months with state funds counted toward maintenance of effort, because the federal provision regulated the use of federal funds rather than imposing a direct prohibition on assistance from any source after that point. Federal law further allowed federally funded assistance beyond sixty months for a limited share of families on hardship grounds, capped at twenty percent of the state average monthly caseload. Administration for Children and Families guidance on time limits stated the same twenty percent ceiling and recorded that the definition of hardship was a state decision.

The use of funds widened discretion in a dimension that had no direct AFDC counterpart of comparable breadth. Section 401(a) of the Social Security Act, codified at 42 United States Code 601(a), stated that the purpose of the TANF program was to increase state flexibility in operating a program designed to achieve four ends. The first end was to provide assistance to needy families so that children could be cared for in their own homes or in the homes of relatives. The second end was to end the dependence of needy parents on government benefits by promoting job preparation, work, and marriage. The third end was to prevent and reduce the incidence of out-of-wedlock pregnancies. The fourth end was to encourage the formation and maintenance of two-parent families. In general, a state could use federal TANF funds in any manner reasonably calculated to accomplish that statutory purpose, in the formulation summarized by the Congressional Research Service. The block grant thus financed a field of activities, of which monthly cash assistance to a state defined caseload was one element rather than the whole program.

A feature of the four purposes that a reader can easily miss alters the meaning of that breadth. The first and second purposes speak in terms of needy families and needy parents, and the Government Accountability Office summary of the regulatory framework recorded that spending intended to meet the first two purposes had to be for persons in financial need, citing the federal regulation at 45 Code of Federal Regulations 260.31. The third and fourth purposes contain no equivalent need language, and the Congressional Research Service primer stated the consequence directly. Activities funded to achieve the third and fourth goals were not limited to needy families, could be provided to the population as a whole without a test of economic need, and could reach families that did not have children. A state could accordingly spend TANF funds on a media campaign, a school based program, or a marriage promotion activity serving persons well outside the income band that governed its cash assistance eligibility rules, and the expenditure could satisfy federal purpose requirements without satisfying a need test. The population reachable with block grant money under two of the four purposes was therefore broader than the population eligible for assistance, and a state allocation decision could shift resources between a need tested cash program and activities with no need test at all.

Federal conditions remained, and an accurate comparison must give those conditions full weight. TANF discretion operated under a work participation standard stated in section 407(a) of the Social Security Act, requiring that fifty percent of all families and ninety percent of two-parent families included in the rate be engaged in work, subject to credits that reduced the effective standard. The Congressional Research Service analysis explained that the caseload reduction credit lowered the fifty and ninety percent standards by one percentage point for each percentage point decline in the caseload from the statutory base year. The standard functioned as a performance measure for the state in the aggregate rather than as a rule any individual family satisfied or violated, and a state that failed to meet its effective standard faced a risk of a reduction in federal TANF funds.

Maintenance of effort imposed a spending floor tied to the earlier program. The Office of the Assistant Secretary for Planning and Evaluation overview of AFDC and TANF stated that states had to spend on needy families at least seventy-five percent of the amount of state funds used in fiscal year 1994 on the predecessor programs, rising to eighty percent for a state that failed to meet the work participation requirement. A state could not withdraw its own effort to a nominal level and retain the full federal grant, and the size of the required contribution was anchored to a stated historical year rather than to the size of the caseload a state chose to serve in a later period.

Data reporting imposed a third continuing condition. The Congressional Research Service primer recorded that TANF law required states to provide aggregate counts of families and individuals receiving assistance and detailed information on the full caseload or on a sample of the caseload. That information included demographic characteristics such as age, sex, and educational attainment, the hours of work activities of adults, and the financial circumstances reported. Federal reporting instructions required the state TANF data report to be filed within forty-five days following the end of a quarter, subject to a penalty for late filing. Federal oversight in TANF therefore concentrated on aggregate performance, expenditure categories, and case data, whereas federal oversight in AFDC had reached into the eligibility and payment determination in each case through quality control review.

The practical consequence follows from the accumulation of those allocations rather than from any single provision. Under AFDC, interstate variation was principally variation in generosity and administration within a federally defined right. A family that met the federal definition of a needy dependent child deprived of parental support, and that satisfied the financial tests of the state of residence, held a claim to aid at the amount produced by that state’s standards and payment rules. The claim could be generous or meager, and the difference between 120 dollars and 923 dollars for a family of three in 1994 was a difference in the lived value of the same federal entitlement. The claim could not be redefined away by a state preference for a different category of family, a different treatment of earnings from the federal sequence, a termination without a hearing, or a work sanction that removed the children portion of the grant for a parent noncompliance. Under TANF, by contrast, interstate variation reached the structure of the program itself. One state could set a low income cutoff, a short time limit, a full-family sanction, and a heavy allocation toward nonassistance services under the third and fourth purposes. Another state could set a higher cutoff, adopt the sixty month federal point, use partial sanctions, spend state funds beyond the federal limit, and concentrate its allocation on cash assistance and work supports. Each state operated under the same federal statute, the same four purposes, the same participation standard in nominal form, and the same maintenance and reporting obligations. The programs that resulted could differ in eligibility, benefit, duration, compliance consequence, and use of funds at the same time, which is a difference in kind from a shared program paid at different rates.

What the Simple Story Omits: Waivers and Erosion

The familiar comparison between Aid to Families with Dependent Children and Temporary Assistance for Needy Families tends to be told as a clean break in a single year. Under that telling, a federal entitlement with federal rules operated in one form until 1996, and then Congress replaced that entitlement with a block grant and a different set of rules in a single legislative act. The clean break leaves out two parts of the record that change what the comparison means. The first concerns origins. Much of the policy content associated with the 1996 act had a testing history inside the older program, conducted by states under federal waivers during the late 1980s and the first half of the 1990s. The second concerns the value of the entitlement that ended. Aid to Families with Dependent Children entered its final decade after a long period of benefit erosion, and the guarantee that the 1996 act withdrew had a purchasing power far below the purchasing power attached to the same guarantee a generation earlier. A comparison that omits the waiver era misstates where reform came from. A comparison that omits the erosion misstates what was lost.

Experimentation Inside the Entitlement

The legal instrument for the waiver era sat inside the Social Security Act itself. Section 1115 of the act, codified at 42 U.S.C. 1315, authorized the Secretary of Health and Human Services to waive compliance with specified requirements of a state plan to the extent and for the period necessary for a state to carry out an experimental, pilot, or demonstration project that the Secretary judged likely to assist in promoting the objectives of the program. House Report 112-677, a committee report of the House of Representatives, traced that waiver authority to the Public Welfare Amendments of 1962, Public Law 87-543, and described the earlier waivers as concerned primarily with program administration and service delivery. The significance of Section 1115 for the AFDC and TANF comparison therefore lies less in the existence of the authority than in the scale and direction of the use made of the authority after the middle of the 1980s. Once secretaries began approving waivers that altered core eligibility and benefit rules rather than administrative arrangements alone, the entitlement acquired an internal mechanism for its own revision, and state governments acquired a route to national policy influence that did not run through an act of Congress.

The acceleration proceeded across three administrations of two parties, a sequence documented in the same House committee report and in reports of the Congressional Research Service on welfare waivers. Between 1987 and 1989, during the administration of President Ronald Reagan, fifteen waiver applications for welfare reform were approved for fourteen states. During the administration of President George H. W. Bush, another fifteen applications from twelve states were approved. The administration of President William J. Clinton then approved waivers at a far larger volume. Between January 1993 and August 1996, according to the Congressional Research Service report Temporary Assistance for Needy Families (TANF): Welfare Waivers and the House committee report that drew on the same administrative record, a total of eighty three waiver applications from forty three states and the District of Columbia were approved. Waiver experimentation was not the project of a single state or a single region, and the delegation of design authority was not a concession extracted by one administration from a reluctant department.

The content of the waivers traced the outline of the statute that followed. The Department of Health and Human Services summary in Welfare Indicators and Risk Factors, prepared by the Office of the Assistant Secretary for Planning and Evaluation, described the changes tested under Section 1115 authority during the 1990s as including expanded earned income disregards, family caps, education and adult oversight requirements for underage single mothers, increased work requirements with stronger sanctions for failure to comply with those requirements, time limits on benefits, and expanded access to transitional benefits such as child care and medical assistance. The Congressional Research Service account added texture to two items on that list. Waivers that addressed earnings did so by disregarding a greater share of earnings in the determination of eligibility and benefit amounts than federal law allowed, and some waivers also permitted families to hold more savings and retain eligibility than federal law allowed. Waivers that addressed work did so by expanding work requirements to more mothers receiving assistance than federal law required, particularly to mothers with very young children. An Institute for Research on Poverty discussion paper, cataloguing an August 1996 department fact sheet, grouped waiver provisions under headings that included requiring work, time limited assistance, making work pay, and encouraging parental responsibility, and listed devices such as community service work in exchange for benefits, self sufficiency plans with goals and deadlines, and one time payments in lieu of continuing assistance. Read as a set, the waiver catalogue and the 1996 statute share a vocabulary of instruments to a degree that coincidence cannot explain.

Two state examples give the general pattern a concrete form. Wisconsin obtained federal approval for a demonstration called Work Not Welfare, a pilot that Education Week reported in November 1993 after the Clinton administration gave the state approval to launch the experiment. The pilot operated in two counties and began in January 1995. Applicants signed contracts pledging to work for benefits, and within a month an applicant had to begin employment, education, or training. Cash benefits ended after two years, and a family that left through the time limit was ineligible to return to welfare for three years, while food stamps and medical assistance remained available outside the cash limit. Wisconsin under Governor Tommy G. Thompson functioned in that period as the most frequently cited laboratory for time limited assistance, and the two year limit at the center of Work Not Welfare anticipated the structure, if not the exact duration, of the sixty month federal limit on assistance from federal TANF funds that Congress enacted in 1996. New Jersey supplied the parallel example for the family cap. The state reform enacted under Governor James J. Florio, built around the Family Development Program, included a provision denying an increase in the grant for a child born to a family already receiving assistance, and an account of the experiment in The Public Interest in the fall of 1996 recorded that the plan took legal effect in October 1992 with a ten month grace period for the family cap provision, so that the first births affected by the provision occurred in August 1993. The family cap later appeared as an express option that states could adopt under TANF rather than as a federal mandate.

The waiver process imposed disciplines that a simple devolution account tends to miss. A waiver was permission, not a transfer of sovereignty. The Secretary approved a defined project, for a defined period, and the federal department expected rigorous evaluation of the demonstration as a condition of approval, a requirement recorded in the evaluation literature of the Office of the Assistant Secretary for Planning and Evaluation. States proposed, the federal executive disposed, and the terms of approval converted state proposals into federally legible experiments with comparison groups, outcome measures, and reporting obligations. Legislators in 1996 were therefore not speculating about wholly untested devices. A body of state demonstrations, federally approved and federally evaluated, sat behind the statutory text, even where the evaluation findings were mixed, preliminary, or contested among analysts.

The analytical consequence for the AFDC and TANF comparison follows directly. The 1996 act changed the fiscal and legal architecture of cash assistance in ways no waiver had done, because no waiver could convert an open ended matching entitlement into a fixed block grant, repeal the individual entitlement to assistance, or impose a federal time limit as a condition attached to federal funds in every state. Against that architectural change, the program rules that attracted the most public attention had a longer and more distributed authorship. Time limits had operated in Wisconsin counties before a federal time limit existed. The family cap had operated in New Jersey before Congress debated whether to require, permit, or forbid such caps nationally. Expanded earnings disregards and expanded work obligations had been tested in dozens of states under Democratic and Republican secretaries alike. The origin story of the 1996 reform therefore runs through state capitols, through waiver applications drafted by state agencies, and through approval memoranda in the Department of Health and Human Services, as well as through the House and Senate. Codification captures the relationship more accurately than invention. Congress in 1996 selected from a menu that state experimentation had populated, generalized several items on that menu into national law, and attached those items to a new funding structure.

An Entitlement Reduced in Value

The second omission concerns the substance of the right that AFDC conferred. An eligible family held an entitlement in the sense that a state with an approved plan had to furnish assistance to families that met the conditions of the plan, and federal matching funds followed eligible expenditures without a fixed national ceiling. That legal form said nothing about the adequacy of the payment. States set benefit levels under AFDC, federal law imposed no minimum grant for a family of a given size, and federal law supplied no automatic adjustment that preserved the purchasing power of a grant against inflation. A legislature could leave a nominal grant unchanged for years, and inflation converted legislative inaction into a benefit reduction without any vote to reduce benefits appearing in the record. Erosion was therefore the predictable output of the structure rather than an accident that befell the structure from outside. The measurement of that erosion requires care, because the standard figure is a median state figure for a specified family size, expressed in constant dollars. The figure describes the maximum benefit available to a family with no earnings in the state standing at the middle of the distribution of state benefit levels.

On that measure, the decline was large by any accounting. An analysis of the AFDC program published in The Future of Children, a journal of the Princeton University and Brookings Institution research community, reported that the median state grant for a family of three with no earned income stood at 366 dollars per month in 1994 and that the median state grant had declined in value by 47 percent between 1970 and 1994. Chapter 7 of the House Ways and Means Committee Green Book record, reproduced in the indicators reporting of the Department of Health and Human Services, supplied the parallel computation for a larger family across the full span to repeal. The maximum AFDC payment for a family of four in the median state fell by more than 50 percent in real value from 1970 to 1996, from 910 dollars to 450 dollars expressed in 1996 dollars. The Center on Budget and Policy Priorities, in the analysis of the 1970 to 1996 period that later summaries of TANF benefit trends repeatedly cited, found that AFDC benefits fell by more than 20 percent in every state but one and by more than 40 percent in two thirds of the states, after adjustment for inflation. The estimates differ in family size, in endpoint, and in unit of measurement, and the estimates agree on direction and on rough magnitude. Roughly two fifths to one half of the real value of the cash guarantee disappeared between 1970 and the middle of the 1990s, and the disappearance occurred in nearly every state rather than in a handful of low benefit jurisdictions.

The median figure also conceals the width of the distribution, and the width was itself a structural property of a program in which each state set a grant without a federal floor. The range and the median for a family of three in 1994, as reported by the Administration for Children and Families fact sheet of February 1996, were stated in the discretion discussion above. Expressed against the poverty threshold, the same fact sheet reported that AFDC benefits in all fifty states and the District of Columbia fell below the Census Bureau poverty threshold, varying from 12.2 percent of the threshold in Mississippi to 75.0 percent in Alaska, with a median of 38.3 percent. A family in the highest benefit state and a family in the lowest benefit state held the same federal entitlement in legal form and held guarantees of radically different value in economic substance.

Food stamps softened the arithmetic without reversing the conclusion. Food stamp benefits followed a national schedule and received adjustment for food price inflation in almost all years, a property that AFDC grants lacked, so the combined package declined less sharply than the cash grant alone. An International Monetary Fund working paper on welfare reform in the United States reported that the maximum level of combined AFDC and food stamp benefits offered by the median state declined by 26 percent in real terms between 1972 and 1992. The February 1996 department fact sheet reported a comparable relationship, with the combination of AFDC and food stamps for a mother and two children with no earnings declining by 29.6 percent between 1972 and 1994. The level mattered as much as the trend. The National Research Council reported that the median of the maximum combined AFDC and food stamp benefit across the states in January 1994 equaled 69 percent of the poverty threshold. The Future of Children analysis reached the same proportion for a family of three in the median state in 1994, combining a maximum AFDC grant of 366 dollars with a maximum food stamp allotment of 295 dollars to reach a total income equal to 69 percent of the federal poverty level. In most states, on each of these accounts, a family that received the maximum cash grant and the maximum food stamp allotment for its size remained below the federal poverty line. The entitlement guaranteed inclusion in a program. The entitlement did not guarantee an income at the threshold that federal statistics used to define poverty.

Criticism of AFDC before repeal came from several directions at once, and the directions did not share a remedy. The Center on Budget and Policy Priorities documented the erosion described in this section and treated the falling real value of benefits as a central defect of the program. A different line of criticism concerned the treatment of two parent families. Coverage of children in two parent families ran through the AFDC Unemployed Parent component, which federal regulation at 45 C.F.R. 233.100 conditioned on a definition of an unemployed parent built around the principal earner. Under that definition, the principal earner qualified as unemployed by being employed less than 100 hours a month, with a narrow allowance for a month above that line where the excess work was intermittent and temporary. California guidance from the Department of Social Services stated the same test, under which both parents had to live in the home and the principal earner had to be unemployed or employed for less than 100 hours per month. A principal earner who crossed from ninety nine hours of work in a month to one hundred hours placed family eligibility in jeopardy, so the program attached a large eligibility loss to a small increment of work, and a family could face a choice between added earnings in a two parent household and continued eligibility that a one parent household did not face in the same form. Defenders of the rule treated the hours test as a targeting device that confined the two parent component to families whose principal earner lacked substantial employment. Critics of the rule treated the hours test as a marriage penalty embedded in regulation. The disagreement concerned incentives and targeting, and both positions described the same mechanism accurately.

A third line of criticism concerned administration, and the mechanism again lay in the incentive structure facing states rather than in any single rule of eligibility. AFDC administration operated under a federal quality control system that measured payment errors and exposed states to fiscal sanctions tied to error rates. A Washington State Department of Social and Health Services history recorded that the Omnibus Budget Reconciliation Act of 1989 established a revised AFDC quality control system imposing penalties on states on a sliding scale reflecting the degree to which a state error rate exceeded the national average, with overpayments and underpayments both entering the error calculation. The asymmetry in that arrangement drew sustained attention in Congress. Payment errors in approved cases generated a measured rate and a sanction, while wrongful denials and terminations sat outside the original center of the measurement system, and legislation introduced in the One Hundred First Congress, H.R. 2243, proposed an improved methodology for measuring state performance on inappropriate denials and terminations and an extension of quality control to that category of error. The premise of that proposal was that a sanction system concentrated on payment accuracy gave state administrators a reason to resolve doubtful eligibility questions against approval, because an approved case with an error counted against the state in a visible way that a denied application did not. Whether that pressure explained any particular volume of denials in any particular state was a matter of dispute among administrators and analysts. The structural exposure was not in dispute. A state bore a fiscal penalty for one class of mistake and bore no equivalent federal fiscal penalty, under the original design, for the opposite class of mistake.

The two movements of this section converge on a single correction to the simple story. The waiver era showed an entitlement program generating, through federal permission, many of the instruments that defined the successor program, which complicates any account that assigns authorship of reform to Congress alone or to the states alone. The erosion record showed an entitlement program delivering a guarantee whose real value legislatures had allowed to contract for roughly two decades, without any single enactment announcing a reduction of that size. AFDC at repeal was a program under criticism from the Center on Budget and Policy Priorities for inadequacy, from proponents of the 100 hour rule critique for penalizing two parent work at the margin, and from members of Congress concerned with quality control design for rewarding caution at the application door. TANF exchanged that structure for a fixed grant with broad state discretion over program rules and a weak structural response to changes in need. The comparison that the evidence supports is therefore a comparison between two imperfect structures. One structure conferred a right of assistance whose value had been allowed to shrink. The other structure conferred flexible funding whose response to need was weak by design. Neither side of that comparison corresponds to a lost ideal, and neither side corresponds to a simple fall from adequacy into discretion. The honest contrast lies between forms of imperfection, specified in law, measurable in dollars, and traceable in the waiver files and benefit tables that preceded repeal.

AFDC vs TANF: The Four-Axis Comparison Table

Axis	Aid to Families with Dependent Children	Temporary Assistance for Needy Families	Practical consequence
Entitlement	A family meeting federal and state criteria held an enforceable right to assistance	The federal statute disclaimed an individual entitlement and left assistance to state program definition	An AFDC claim could be pressed at a fair hearing and in court, while a TANF claim rested on the rules a state chose to write
Funding mechanism	Open ended federal matching paid between 50 and 83 percent of allowable state spending	A fixed block grant of about 16.5 billion dollars a year, unadjusted for inflation or caseload	Federal money rose and fell with state spending under AFDC, and held at the statutory figure under TANF regardless of volume
Cyclical behavior	Federal spending rose automatically as the eligible caseload rose during a downturn	The federal grant held flat, and added money required separate congressional action or a finite contingency fund	AFDC operated as an automatic stabilizer, while the TANF response depended on decisions by states and by Congress
State discretion	States set benefit levels and need standards inside federal eligibility categories and counting rules	States set eligibility, benefit levels, shorter time limits, sanctions, and the allocation of funds across four statutory purposes	Variation under AFDC was principally variation in generosity, while variation under TANF reached the structure of the program itself

The Verdict: Name the Deciding Factor

The comparison between Aid to Families with Dependent Children and Temporary Assistance for Needy Families reduces to a single trade that neither program escapes. An entitlement financed by open ended federal matching responds automatically when need rises, because eligibility defined in law generates claims, claims generate state spending, and state spending within the matching formula generates federal payment without a fresh appropriation and without a ceiling that a surge in eligible families can exhaust. A block grant financed by a fixed federal allotment delivers predictable cost for exactly the opposite reason, because the federal commitment is stated as a sum rather than as a share of whatever eligible need produces, and a surge in eligible families changes the demand placed on a fixed sum rather than the size of the sum itself. Automatic response to need and predictable fixed cost are mutually exclusive properties of a funding structure. Each property follows from the definition of the instrument, and each instrument excludes the other property by the same definition. Drafting ingenuity can soften the edge of the trade or shift the cost of the trade between levels of government, and drafting ingenuity cannot produce both properties at full strength from one instrument, because a payment rule cannot both promise to match whatever eligible need requires and promise in advance the total that the promise will cost.

Administration operates inside boundaries that the funding structure sets. Under the entitlement structure, a state during a recession faced rising claims that the federal matching formula partly financed, and state funds supplied the unmatched share of every additional dollar of assistance. Under the block grant structure created by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, a state during a recession faced rising claims against an allotment that the recession did not enlarge, and held discretion to redirect effort toward services or reserves because federal law attached the money to broad purposes rather than to an individual entitlement. Each incentive pattern is the funding structure expressed as behavior at the state level.

Hybrids Buy Part of Each Property at a Stated Price

A hybrid instrument purchases part of each property at a stated price, and honesty about hybrids requires naming the price alongside the purchase. A block grant with a contingency mechanism retains predictable cost for the base allotment and gains a bounded form of responsiveness, and the price is that responsiveness lasts only while contingency funds last, triggers only under conditions written in advance, and ends by appropriation or exhaustion rather than by a decline in need. A capped entitlement retains the legal form of an individual claim and gains a ceiling on exposure, and the price is that the claim weakens or fails at precisely the moment when claims press hardest against the cap. Neither hybrid refutes the trade, and the legislative value of a hybrid lies in making the partial surrender on both sides explicit.

The Emergency Contingency Fund episode of 2009 to 2010 supplies compact evidence of what a bolted on response costs and how such a response ends. The American Recovery and Reinvestment Act of 2009 created the fund inside the Temporary Assistance for Needy Families structure with an appropriation of five billion dollars, and the Congressional Research Service recorded that the fund reimbursed states, territories, and tribes for eighty percent of increased costs in basic assistance, nonrecurrent short term benefits, and subsidized employment during fiscal years 2009 and 2010. The Administration for Children and Families recorded that the full five billion dollars was obligated, with more than one billion dollars supporting subsidized employment. The fund then expired on September 30, 2010, and the Congressional Research Service recorded that measures to extend the fund passed the House of Representatives twice during 2010 and received no approval in the Senate. A contingency mechanism can deliver genuine countercyclical help while money remains available, delivery depends on a separate political decision rather than on eligibility alone, and termination follows the calendar and the appropriation rather than the level of need.

The Verdict on the Deciding Factor

The verdict follows directly once the deciding factor is named. If the deciding judgment is that assistance should expand automatically when need rises, then the entitlement structure is the only structure of the two that performs that expansion as a property of the design rather than as a favor of a subsequent appropriation. A legislature selecting that judgment accepts the attached costs with open eyes. Federal exposure is open ended, because the matching commitment follows eligible claims wherever eligible claims lead, and budgetary control operates through eligibility rules and benefit levels rather than through a total. A second cost belongs to the entitlement form itself. An entitlement can erode by neglect without a vote to cut, because a legislature that leaves benefit levels and eligibility thresholds unadjusted while prices rise has reduced the real value of the guarantee while preserving the legal shell of the guarantee. Automatic response arrives bundled with exposure and with vulnerability to quiet erosion, and a defended preference for the entitlement ranks automatic response above both hazards.

If the deciding judgment is instead that predictability of cost and state discretion matter more, then the block grant is the structure that delivers both properties as properties of the design rather than as aspirations in report language. The federal total is stated in advance, and the Congressional Research Service recorded that the basic block grant total established in the 1996 law was sixteen point six billion dollars, unchanged from fiscal year 1997 through fiscal year 2016 without adjustment for inflation, caseload change, or change in the poverty population. State discretion follows from the same fixed form, because money appropriated for broad statutory purposes can be allocated among cash assistance, work activities, child care support, and other allowable uses according to state judgment within federal conditions. A legislature selecting that judgment also accepts attached costs with open eyes. Countercyclical response is surrendered, because a fixed allotment does not expand when unemployment expands, apart from separate contingency provisions that a legislature may or may not supply and fund. A federal guarantee to any individual family is surrendered as well, because the block grant creates a funding relationship between governments and creates no individual entitlement that a family can enforce against a state.

The verdict is therefore a defended structural conclusion rather than a partisan one. The comparison decides which structure does what, with confidence grounded in definitions and in the operating record of both programs. The comparison does not decide which set of properties a polity ought to prefer, because preference between automatic response and predictable cost is a value judgment about the purpose of assistance, the proper location of discretion, and the acceptable distribution of risk among the federal government, state governments, and families with low incomes. A structural verdict that names the value judgment performs a useful service, locating disagreement at the deciding factor rather than dispersing disagreement across disputes about mechanics that evidence and definition have already resolved.

The Dependency Argument at Full Strength

The dependency argument, in the form that shaped the debate of the 1990s and continued afterward, holds that an open ended entitlement to cash assistance alters behavior in ways that deepen the condition the assistance was meant to relieve, and that a well designed program must therefore attach obligations to assistance and limit the period during which assistance can substitute for earnings. Charles Murray stated an early and influential version in Losing Ground, published in 1984. Murray argued that social welfare programs as implemented in the United States tended to increase poverty rather than reduce poverty, because program rules created incentives that rewarded short sighted behavior and penalized the conduct associated with escape from poverty over the long term. Generosity itself, routed through rules that changed the relative rewards of work, marriage, and childbearing, could produce outcomes opposite to the outcomes that generosity intended, and a program could fail by incentive design even when appropriations rose.

Lawrence M. Mead supplied a second version that differed from the Murray version in remedy while sharing the diagnosis that permissive assistance fails recipients. In Beyond Entitlement: The Social Obligations of Citizenship, published in 1986, Mead argued that welfare programs must require work as well as offer support if recipients are to be integrated into society rather than merely subsidized, and that the central defect of the welfare state was permissiveness rather than size. In The New Paternalism: Supervisory Approaches to Fighting Poverty, published in 1997, Mead developed the administrative form of the argument, describing paternalism as a social contract under which receipt of benefits carries obligations in return, enforced through administrative oversight by case managers who both help clients and hold clients to expectations. Federal authority sets standards, states administer demanding programs, and assistance becomes a reciprocal institution in which support and obligation arrive together. On that account, the entitlement defect was the absence of enforceable expectations, and the block grant attracted support from adherents of the Mead position because the 1996 law paired state discretion with federal work participation standards and a general five year limit on federally funded assistance.

The strongest form of the dependency argument connects a behavioral claim, an institutional claim, and a fiscal claim. The behavioral claim is that weakly conditioned assistance, available for an indefinite period, reduces the urgency of employment and allows a spell of need to harden into a durable status. The institutional claim is that government possesses both the right and the responsibility to structure assistance so that the durable status becomes difficult to occupy, through work requirements, time limits, sanctions for noncompliance, and close supervision that treats recipients as citizens bearing obligations alongside rights. The fiscal claim completes the structure. An open ended matching entitlement finances the durable status automatically, so a legislature troubled by dependency confronts a funding rule that expands support for the very pattern the legislature seeks to discourage. A fixed block grant with participation standards reverses the fiscal signal, rewarding transition toward employment. That is the dependency argument at full strength, attributed to named proponents, and entitled to an answer on evidence rather than dismissal by label.

The Adequacy Argument at Full Strength

The adequacy argument, in the form made by named analysts and researchers critical of the block grant, holds that a fixed allotment combined with broad state discretion predictably produces a program that reaches a shrinking share of families in poverty, loses real value across time without any legislated decision to reduce assistance, and responds weakly when a recession multiplies need. Analysts at the Center on Budget and Policy Priorities developed the reach evidence that anchors the argument. Ife Floyd, LaDonna Pavetti, and Liz Schott, in analysis published by the Center on Budget and Policy Priorities under the title TANF Reaching Few Poor Families in December 2017, documented the contraction in the share of poor families receiving cash assistance under the block grant. As recorded in Article 236 of the delivered series, drawing on the Center on Budget and Policy Priorities ratio of families receiving assistance for every one hundred families with children in poverty, the ratio fell from sixty eight in 1996 to twenty three in 2015 and twenty two in 2018. A decline of that magnitude, on the adequacy account, reflects design rather than accident, because a fixed grant gives states no additional federal resources for serving additional families.

Kathryn J. Edin and H. Luke Shaefer gave the adequacy argument a household level account in $2.00 a Day: Living on Almost Nothing in America, published in 2015. Edin and Shaefer documented one point five million households, including about three million children, surviving on cash incomes of two dollars per person per day, and the authors traced the rise in such extreme cash poverty to the weakening of the cash assistance safety net after the 1996 law replaced an entitlement with a block grant. A household budget without cash income translates the aggregate ratio into rent unpaid, food obtained through irregular channels, and survival strategies that substitute for a predictable monthly grant. On the adequacy account, a program that reaches twenty two or twenty three families for every one hundred families in poverty has not merely become more selective in a neutral sense. Selectivity has crossed into absence for a large share of the population that the predecessor program would have reached, and absence during a recession compounds the failure because need and program contraction move in opposite directions.

The adequacy argument at full strength also contains an institutional claim about accountability. A federal guarantee disciplines both levels of government, because eligible families hold a claim that administrators must honor and legislatures must finance. A block grant relaxes that discipline. States can tighten eligibility through procedural requirements, diversion practices, and time limits shorter than the federal maximum, and each decision can occur administratively or incrementally without a single vote that announces a reduction in the national commitment to families in poverty. The fixed nominal total performs a similar reduction on the federal side, since a grant held at sixteen point six billion dollars across two decades delivers less assistance per poor family as prices rise. The two arguments collide at a genuine point of disagreement. The dependency argument treats contraction in receipt as evidence that incentives and obligations are working when contraction accompanies employment and self support. The adequacy argument treats contraction in receipt as evidence of failure when contraction accompanies persistent poverty and diminished reach. Caseload numbers alone cannot arbitrate between those readings, because the same decline supports either reading until evidence about income, employment durability, and material hardship enters the account.

Guidance by Reader Type

Recommendation by reader type follows from the structure of the evidence. The student should study Aid to Families with Dependent Children first, because the design choices inside Temporary Assistance for Needy Families become legible only as answers to properties of the predecessor program, and a student who begins with the block grant encounters time limits, participation standards, and fixed funding as freestanding features rather than as deliberate negations of entitlement status, federal matching, and federally defined eligibility. After the predecessor program is understood on entitlement terms, the student should study the text of the 1996 act itself, reading the block grant provisions as a record of choices about funding, purpose, and discretion rather than treating later commentary as a substitute for the statutory language. A wider path through the surrounding statutes appears in the US legislation study guide. During that sequence, a student can keep your statute notes, citations, and case chronologies together free on VaultBook rather than scattering the comparison across loose pages. After the reading, a student can practice and revise US government and civics material on ReportMedic to test command of the four axes.

The congressional staffer should study the funding mechanics and the contingency question, because funding mechanics and contingency design mark the location where legislative choice actually operates. Eligibility rhetoric and program naming attract public attention, and the operative decisions concern the matching formula or the fixed total, the adjustment or nonadjustment of the total across time, the conditions and financing of contingency assistance during downturns, the maintenance of effort obligations placed on states, and the fiscal consequences attached to participation standards. A staffer who commands those mechanics can identify the exact clause that determines whether assistance expands during a recession, whether inflation erodes the federal commitment by default, and whether a contingency fund exists as a funded instrument or as an authorization without money behind the authorization.

The researcher should study the waiver record and the state variation that both programs produced, because the waiver record and state variation constitute the usable evidence for separating structural effects from administrative and economic effects. Waivers granted under the predecessor program allowed states to test time limits, work requirements, and earnings disregards before the 1996 law nationalized several features of the waiver era, and careful attention to waiver design permits the researcher to ask which outcomes followed the tested feature and which outcomes followed the strong labor markets of the period. State variation under the block grant offers a second laboratory, less controlled and far broader, in which states selected different benefit levels, time limits, and sanction regimes under a common federal funding form. Disciplined comparison across states can identify which outcomes track state policy choices and which outcomes track the fixed funding structure that every state shared.

The journalist should study the caseload series and the reach series and the difference between caseload decline and poverty decline, because conflating caseload decline with poverty decline is the standing error in coverage of the subject. A caseload can fall because employment rose and family income rose with employment, because eligibility narrowed, because procedural barriers discouraged application or continuation, because time limits terminated assistance regardless of income, or because several of those forces operated together. A poverty rate can fall, rise, or hold across the same period for reasons that include wages, hours, other transfer programs, and the business cycle, and no honest account can infer the poverty movement from the caseload movement without independent evidence. The ratio of families receiving assistance for every one hundred families with children in poverty belongs in coverage precisely because the ratio joins the two series and exposes divergence between them.

What the Comparison Settles and What the Comparison Leaves Open

The closing assessment separates settled structure from unsettled judgment. The comparison settles the structural properties, and the structural properties are not in dispute among serious accounts of either program. An entitlement with open ended matching expands federal spending automatically when eligible need expands and creates an enforceable individual claim, at the price of open ended exposure and vulnerability to erosion through unadjusted benefit levels and eligibility rules. A block grant with a fixed total holds federal cost predictable and confers broad state discretion, at the price of weak automatic response during downturns and the absence of a federal guarantee to any individual family. A hybrid purchases partial responsiveness or a partial ceiling at a stated price, on terms that need can neither extend nor enlarge. The dependency argument and the adequacy argument, stated at full strength by Murray, Mead, analysts at the Center on Budget and Policy Priorities, and Edin and Shaefer, disagree about the human consequences of assistance, and neither argument overturns the structural description that the funding definitions supply.

The comparison cannot settle the value judgment, and refusal to disguise that limit counts as a strength of the comparison rather than a defect. Evidence can establish how many families received assistance, how benefits changed in real terms, how states allocated flexible funds, and how a contingency fund performed before expiration, and evidence cannot establish whether a society ought to rank an automatic guarantee above predictable cost or predictable cost and state discretion above an automatic guarantee. That ranking rests on convictions about obligation, risk, federalism, and citizenship that citizens and legislators bring to the evidence rather than derive from the evidence alone. Every other question in the comparison has an answer that research can approach, and the deciding question has an answer that judgment must supply. The deciding factor, stated in one sentence, is whether assistance must expand automatically when need rises even at the price of open ended federal exposure, or whether predictable federal cost and state discretion matter more even at the price of surrendering automatic expansion and any federal guarantee to an individual family.

Frequently Asked Questions

Q: What is the difference between AFDC and TANF?

Aid to Families with Dependent Children was a federal and state matching entitlement under Title IV, Part A of the Social Security Act, under which a family that met federal and state eligibility criteria held an enforceable right to cash assistance and the federal government reimbursed a share of allowable state spending without a ceiling. Temporary Assistance for Needy Families replaced that program in 1996 as a fixed federal block grant. The TANF statute disclaimed an individual entitlement, paid each state a predetermined sum, imposed a sixty month limit on federally funded assistance, attached work participation standards, and gave states broad authority over eligibility, benefit levels, sanctions, and the use of funds across four statutory purposes. The difference is therefore architectural rather than cosmetic, covering legal status, funding formula, cyclical behavior, and the location of decision authority in a single substitution.

Q: Was AFDC an entitlement unlike TANF?

AFDC was an entitlement in two reciprocal senses. A family that satisfied the governing federal and state criteria held a legal claim to assistance that an agency could not deny for lack of funds, and a state with an approved plan held a claim to federal matching payments for allowable expenditures. Courts enforced the incidents of that claim, and procedure protected the claim at termination. TANF reversed the legal character of the benefit. Section 601(b) of the Social Security Act, as rewritten in 1996, stated that the part created no individual entitlement to assistance, and the federal payment ran to the state as a block grant rather than to a family as a right. A state could define the terms of participation, and a family held only the rights that state law and the federal conditions on the grant supplied. The contrast on the entitlement axis is therefore complete: a right held by a family under one model, and a grant held by a state under the other.

Q: Did AFDC funding rise automatically unlike TANF?

Federal funding under AFDC rose automatically with need as the statute defined need, because the federal government matched state spending on eligible families at rates between 50 and 83 percent, and the appropriation carried no ceiling on the total. A rise in the caseload raised state expenditure, and the federal reimbursement followed the expenditure through the formula without a new appropriation or an emergency finding. TANF funding followed the opposite rule. The basic grant was fixed in statute at about 16.5 billion dollars a year and was not adjusted for inflation, for caseload change, or for change in the population in poverty. Additional federal money during a downturn arrived only through a finite contingency mechanism, through balances a state had reserved, or through separate legislation, as the Emergency Contingency Fund of 2009 demonstrated. Automatic movement belonged to the matching structure alone.

Q: Which gives states more control, AFDC or TANF?

TANF gave states substantially more control, although AFDC discretion was broader than a simple federal program description suggests. Under AFDC, states set benefit levels, need standards, and payment standards, and the resulting grants diverged widely, from 120 dollars a month for a family of three in Mississippi to 923 dollars in Alaska in 1994, with a median of 375 dollars. Federal law, however, defined the covered categories, the counting of income and resources, and the procedural rights of recipients, and federal quality control reviewed individual case determinations. Under TANF, state authority reached the structure of the program itself. States set eligibility thresholds, benefit levels, time limits shorter than the federal sixty month point, sanction amounts up to termination of the full grant, and the allocation of funds among cash assistance and services under the four purposes. Control under AFDC was control over generosity inside a federal right, while control under TANF extended to the design of the program a family encountered.

Q: What did Goldberg v. Kelly give AFDC recipients?

Goldberg v. Kelly, decided by the Supreme Court in 1970, gave AFDC recipients a constitutional right to an evidentiary hearing before benefits were terminated. The recipients in the case faced termination with no prior notice or hearing, and a formal fair hearing was available only after benefits had ended. The Court held that welfare benefits were a matter of statutory entitlement for persons qualified to receive them, that termination pending resolution of a dispute over eligibility carried grievous consequences for a family dependent on the payment, and that due process therefore required notice and an opportunity to be heard before the payment stopped. The hearing had to permit confrontation of adverse witnesses and an impartial decision maker. The decision converted the entitlement from a statutory promise into a procedurally protected interest, and the protection applied at the moment of greatest vulnerability, the interval between an agency decision and review of that decision.

Q: Did states test AFDC changes before TANF?

States tested many of the defining features of the 1996 reform inside AFDC under federal waiver authority. Section 1115 of the Social Security Act permitted the Secretary of Health and Human Services to waive specified state plan requirements for experimental, pilot, or demonstration projects. Waiver use accelerated across three administrations, and between January 1993 and August 1996 alone, according to the Congressional Research Service, eighty three waiver applications from forty three states and the District of Columbia were approved. The projects tested expanded earnings disregards, broader work requirements reaching mothers of very young children, sanctions for noncompliance, time limits on benefits, family caps, and transitional supports such as child care. Evaluation accompanied approval as a condition of the demonstrations. The waiver record therefore supplied a trial history for the statute that followed, and several devices that appeared as national TANF features or state options under TANF had operated earlier as approved experiments inside the entitlement that TANF replaced.

Q: Which model responds better to a recession, AFDC or TANF?

AFDC responded to a recession by construction, because the matching formula converted a rise in eligible families into higher state spending and higher federal reimbursement without any new decision by Congress. The caseload climbed from roughly 3.7 million families in 1989 to a Congressional Research Service peak of 5.033 million families and 14.161 million recipients in 1994, and federal and state costs rose together at the matching rate. TANF responded only through decisions. During the downturn that began in December 2007, the fixed grant produced a caseload increase of 13 percent on one state collected measure, against a 45 percent increase in food assistance caseloads between December 2007 and December 2009 under a matching entitlement serving overlapping populations. The original 2 billion dollar Contingency Fund was exhausted by December 2009, and substantial countercyclical money moved only under a new 5 billion dollar statute in 2009. On the cyclical axis, the entitlement supplied arithmetic and the block grant supplied discretion.

Q: Which should a student study first, AFDC or TANF?

A student should study AFDC first, because the design choices inside TANF become legible only as answers to properties of the predecessor program. Time limits read as a negation of an entitlement with no federal duration limit, participation standards read as a negation of a program whose work rules sat inside a guaranteed grant, and fixed funding reads as a negation of open ended matching. A student who begins with the block grant encounters each feature as a freestanding rule and misses the deliberate character of the substitution. After the predecessor program is understood on entitlement terms, the productive sequence runs to the text of the 1996 act itself, read as a record of choices about funding, purpose, and discretion, and then to the measured record of caseload, reach, and erosion under each model. Sequence matters because the comparison is architectural, and architecture is understood from the earlier structure outward.

Q: When did TANF replace AFDC?

TANF replaced AFDC through the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Public Law 104-193, signed on August 22, 1996. The act repealed the AFDC program, rewrote Title IV, Part A of the Social Security Act, and placed the TANF block grant in the rewritten part, codified at 42 United States Code sections 601 and following. The replacement operated through state plans submitted under the new statute, and the transition from the matching entitlement to the block grant was completed across the states during fiscal year 1997, with state plans under the new program in place no later than July 1, 1997. The date therefore carries two layers: a signing date in August 1996 that ended the legal life of the old program as the governing model, and an administrative transition across the following months during which states began operating cash assistance under block grant rules, purposes, and funding.

Q: Did AFDC have a time limit on benefits?

AFDC carried no federal lifetime time limit on benefits. A family that continued to satisfy the categorical and financial criteria of the program remained eligible for assistance, and eligibility ended through changes in family circumstances, income, or composition rather than through exhaustion of a statutory clock. Duration on the rolls could extend across years, and that open duration formed part of the criticism directed at the program during the debates of the 1990s. Two qualifications complete the picture without altering the answer. State waiver demonstrations during the late 1980s and the first half of the 1990s tested time limits inside the entitlement under Section 1115 authority, so the device had a trial history before 1996. The 1996 act then nationalized a duration rule for the replacement program, barring use of federal TANF funds for a family containing an adult beyond sixty cumulative months, subject to a hardship allowance for a limited share of the caseload.

Q: How did work requirements differ between AFDC and TANF?

Work requirements existed under AFDC and hardened under TANF, and the difference lies in obligation, coverage, and consequence. AFDC work policy ran through the Work Incentive Program and then the Job Opportunities and Basic Skills Training Program created by the Family Support Act of 1988, which required states to operate a work and training program with participation expectations and broad exemptions, including exemptions that shielded many parents of young children. The entitlement itself continued underneath the work program. TANF attached work to the funding structure. Federal law set work participation standards of 50 percent for all families and 90 percent for two parent families, measured state performance each year, and exposed a state to a fiscal penalty for a missed standard, softened by the caseload reduction credit. States in turn required participation as a condition of assistance and sanctioned refusal by reduction or termination of the grant. Obligation under AFDC was a program layered on a right, while obligation under TANF helped define the terms on which assistance existed at all.

Q: What was the federal matching rate under AFDC?

The federal matching rate under AFDC varied by state and by expenditure type. For assistance payments, the federal government reimbursed a share of allowable state spending that ranged from 50 percent to 83 percent, computed under a formula that moved inversely with state per capita income, so a poorer state received a higher federal share than a wealthier state for the same dollar of aid. Administrative costs were generally matched at 50 percent. The rate structure carried the central incentive of the model. Each additional dollar a state spent on an eligible family drew down federal money at the applicable rate, which lowered the state price of generosity and of expansion during a downturn, and each dollar a state withdrew from assistance surrendered the federal contribution attached to that dollar. No ceiling capped the total federal payment, so the rate applied across the full volume of allowable expenditure a state chose to make under an approved plan.

Q: Did states keep the federal money when TANF caseloads fell?

States kept the federal money when TANF caseloads fell, because the basic grant was fixed in statute and did not vary with the number of families a state assisted in a given year. A decline in the caseload reduced state spending on cash assistance without reducing the federal payment, and the released funds remained available for other uses permitted by the statute, including child care, work supports, and services under the four purposes, or for reserve balances carried into later years. The maintenance of effort requirement continued to oblige a floor of state spending measured against a fiscal year 1994 baseline, but the federal grant itself stayed whole. That arithmetic produced the fiscal signature of the block grant during the late 1990s, when caseloads fell sharply from the AFDC peak: total federal resources per remaining case rose even as the number of assisted families contracted, and allocation choices inside a fixed envelope replaced the matching question of how much total spending a caseload would generate.

Q: Did AFDC benefits keep up with inflation?

AFDC benefits did not keep up with inflation, because federal law set no benefit floor and no indexation rule, and each state determined the dollar grant under the need and payment standards of the state plan. States frequently left nominal grants unchanged across long stretches, and prices rose against the frozen figures. The Congressional Research Service and the analyses summarized in the erosion record place the loss in the median state at roughly two fifths to one half of the real value of the maximum benefit between 1970 and the middle of the 1990s. The February 1996 fact sheet figures show the endpoint of that process in cross section: a median grant of 375 dollars a month for a family of three, equal to a median of 38.3 percent of the poverty threshold. The entitlement therefore guaranteed inclusion in a program and a procedurally protected claim to the state set amount, and the amount itself was exposed to a quiet reduction that required no vote and no statutory amendment.

Q: What are the four purposes of TANF, and did AFDC have equivalents?

The TANF statute states four purposes for the use of block grant funds. The first purpose is assistance to needy families so that children may be cared for in their own homes or in the homes of relatives. The second purpose is the ending of dependence of needy parents on government benefits through promotion of job preparation, work, and marriage. The third purpose is the prevention and reduction of out of wedlock pregnancies. The fourth purpose is encouragement of the formation and maintenance of two parent families. AFDC had no equivalent list. The predecessor program carried a single animating purpose, aid to needy dependent children in parental or relative care, expressed through eligibility categories rather than through a set of spending purposes. The contrast matters for the use of money: TANF funds could lawfully flow toward activities serving the second, third, and fourth purposes far from cash assistance to a family, while AFDC federal matching attached to payments and administration under a defined benefit program.

Q: Who set benefit levels under AFDC and TANF?

States set benefit levels under both programs, and the identity of the decision maker is one of the few continuities across the 1996 substitution. Under AFDC, each state established a standard of need, a payment standard that could sit below that standard, and any maximum benefit, and the federal government matched the resulting payments without imposing a national floor or a national schedule. The outcome in 1994 was a range from 120 dollars a month for a family of three in Mississippi to 923 dollars in Alaska, with a median of 375 dollars. Under TANF, states retained authority over the dollar grant, exercised inside a fixed federal envelope and alongside state authority over eligibility and duration. The budgetary meaning of a state decision changed even though the identity of the decision maker held constant. A higher benefit under AFDC drew additional federal matching money, while a higher benefit under TANF consumed a larger share of a fixed grant and competed directly with every other use of that grant.

Q: What is the caseload reduction credit, and did AFDC have an equivalent?

The caseload reduction credit is a TANF device that lowers the work participation standard a state must meet by the number of percentage points by which the state caseload declined against a statutory baseline year, with declines attributable to eligibility restrictions excluded from the calculation. The baseline stood at fiscal year 1995 under the original design and moved to fiscal year 2005 under later amendment. A state whose caseload fell by thirty points against the baseline faced an effective all families standard of twenty percent rather than the statutory fifty percent. The credit rewarded caseload decline itself, regardless of whether decline reflected employment, diversion, sanction, or time limit exits. AFDC had no equivalent. The predecessor program imposed program participation expectations through the Job Opportunities and Basic Skills Training Program, and federal oversight measured state administration through quality control, but no mechanism converted a shrinking caseload into a reduced federal standard, because the entitlement design treated caseload size as an outcome of eligibility rather than as a performance target attached to funding.

Q: What does an open-ended matching grant mean?

An open ended matching grant is a funding instrument under which a higher level of government promises to reimburse a fixed share of whatever a lower level of government spends on defined allowable purposes, with no ceiling on the total reimbursement. Three properties follow from the definition. Federal exposure cannot be stated as a single figure in advance, because the total emerges from the sum of state spending decisions and caseload outcomes. The price of the aided activity falls for the spending government, because each dollar of service costs the state only the unmatched fraction. Responsiveness arrives automatically, because a rise in eligible need that raises spending raises the federal contribution through the same formula. AFDC embodied the instrument in cash assistance, with federal shares between 50 and 83 percent for assistance payments. A fixed block grant is the opposite instrument on each property: exposure stated in advance, no price effect at the margin, and no automatic movement when need changes.

Q: Which program gave recipients stronger rights in court, AFDC or TANF?

AFDC gave recipients stronger rights in court by a wide margin, because the entitlement supplied both a substantive claim and procedural protection for that claim. Federal courts enforced federal eligibility rules against conflicting state definitions, confined state waiting periods, and applied the holding of Goldberg v. Kelly that benefits could not be terminated without a prior evidentiary hearing. A recipient could frame a grievance as the denial of a right created by federal law, and the federal matching claim of the state reinforced the reciprocal structure. TANF weakened each element. The rewritten statute stated that the part created no individual entitlement, state definitions governed the terms of participation within broad federal purposes, and litigation shifted toward the conditions attached to the federal grant, such as reporting and maintenance obligations owed between governments. A family under TANF held the rights that a state chose to confer and the process that state law supplied, rather than a federal claim enforceable against the administering agency.

Q: What should an exam answer include when comparing AFDC and TANF?

A complete exam answer compares the two programs on four axes and supports each contrast with a statute, a figure, or a case. The answer names the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 as the instrument of substitution, states the entitlement contrast between an enforceable family claim and a statutory disclaimer of individual entitlement, and explains open ended matching against a fixed grant of about 16.5 billion dollars a year. Measured evidence strengthens each axis: the Congressional Research Service caseload peak of 5.033 million families in 1994, the 36.2 percent erosion of the fixed grant in real terms by fiscal 2018, and the Center on Budget and Policy Priorities reach ratio falling from 68 families per one hundred in poverty in 1996 to 22 in 2018. A strong answer adds the waiver record and benefit erosion as correctives to a clean break story, cites Goldberg v. Kelly on procedure, and closes by naming the deciding trade between automatic expansion and predictable cost.