USBIG Discussion Paper No. 55, February 2003
Work in progress, do not cite or quote without author’s permission
What Defeated a Negative Income Tax?: Constructing a Causal Explanation of a Politically
Controversial Historical Event
Leland G. Neuberg
Department of Mathematics and Statistics
111 Cummington Street Boston, MA 02115
thDraft of a Paper to Be Presented at the 29 Annual Meetings of the Eastern Economic
Joint Program of the Eastern Economic Association and the United States Basic Income
Crowne Plaza Times Square Hotel
February 21-23, 2003
WHAT DEFEATED A NEGATIVE INCOME TAX?: CONSTRUCTING A CAUSAL EXPLANATION OF A
POLITICALLY CONTROVERSIAL HISTORICAL EVENT
In 1969 the Nixon Administration proposed to reform welfare with a negative income tax (NIT) that made none of those on welfare worse off in income terms. Three years of intense congressional struggle ensued at the end which the Nixon NIT welfare reform died. This paper studies that conflict with the political methodologist's logistic regression models of votes on bills in Congress, and the political historian's analysis of primary and secondary textual sources. Employing the results of its study, the paper constructs a narrative that embeds a causal explanation of the defeat of the Nixon NIT proposal in its story. That narrative focuses on the political ideas of key liberal Senator Fred Harris (D, Oklahoma) because previous narratives have neglected them. The paper concludes with a critique of some of the causal claims embedded in previous narratives of the defeat of the Nixon NIT proposal.
WHAT DEFEATED A NEGATIVE INCOME TAX?: CONSTRUCTING A CAUSAL EXPLANATION OF
A POLITICALLY CONTROVERSIAL HISTORICAL EVENT
During the 1964 presidential campaign, Goldwater's politically conservative economic advisor Milton Friedman proposed that a negative income tax (NIT) replace all government programs aimed at alleviating poverty. (Appendix One sketches the mathematics of an NIT.) Goldwater lost to Johnson in a landslide. Nevertheless, at the suggestion of Robert Lampman of the politically liberal Institute for Research on Poverty at the University of Wisconsin at Madison, economists at the Office of Economic Opportunity (OEO) began planning an NIT in the summer of 1965. In the fall of 1965 and again in the summer of 1966. OEO presented an NIT plan to the Bureau of the Budget but "the plan never was taken seriously by the Johnson Administration" (Levine 1975, 16). An NIT "was not regarded as a serious proposal that could be enacted in less than a decade" (Levine 1975, 20).
Some who had worked on NIT plans under Johnson at OEO stayed on under Nixon and continued to do so at OEO and at the Department of Health, Education, and Welfare (HEW). During the 1968 presidential campaign Nixon had promised to "reform welfare". After the election advocates of various welfare reform plans,
1 including those who favored an NIT, among his advisors began jockeying for position. Daniel Patrick Moynihan,
Nixon's closest White House domestic policy advisor, at first favored a family allowance system similar to those in Europe, but soon embraced an NIT. He also gained frequent access to, and some influence on, the President. In the Spring of 1969 Nixon decided on a plan containing an NIT, and presented that Family Assistance Plan (FAP) to Congress in August 1969. "In retrospect one can ask why Mr. Johnson turned a deaf ear on a proposal that in its basic mechanics was similar to the one that President Nixon was to endorse a few years later" (Williams 1972, 4).
Williams gives three answers to his question. First, "the negative income tax lacked a well-placed, articulate and tenacious advocate in the inner councils of the President, a crucial role that Daniel P. Moynihan was to assume in the Nixon administration" (1972, 5). In fact, under Johnson "Wilber Cohen, then undersecretary of HEW, was relatively unenthusiastic about the negative tax" and "in effect held a veto in the Executive Office over any new income maintenance plans" (1972, 5). "Second, the costs of the war in Vietnam were moving to their peak point so that the White House was hardly interested in major and expensive new initiatives" (1972, 5). "Third, congressional and public opinion toward the negative tax was quite hostile" (1972, 5).
Certainly Moynihan had Nixon's ear in a way that no similar NIT advocate had Johnson's, and this partially explains the change of NIT fortunes. Moynihan confirms that the Nixon Administration anticipated reduced war costs: "the prospect of a 'peace-and-growth' dividend still seemed viable, and it was assumed that somewhere down the road major initiatives in social policy would be made possible by the availability of these funds" (1973, 75). Why, though, did Nixon decide to commit anticipated future war cost savings to FAP? Also, Williams offers no evidence of a difference of congressional and public opinion on an NIT at the beginning of the two administrations. If there was such a difference, what change between the summers of 1965 and 1969 made Congress and the public more open to an NIT, and so made such a plan look politically feasible, or even necessary, to Nixon? Or did differences in the plans presented to Johnson and proposed by Nixon make the Nixon plan more popular?
What apparently put "welfare reform" on the political agenda at the opening of the Nixon Administration was a two faceted revolt of the urban black poor in the mid to late 1960s. First, unprecedented numbers of poor black women surged onto the welfare rolls in cities across the country. From 1950 to 1960 the number of families receiving welfare in the United States rose by only 17%. From December 1960 to February 1969, however, their numbers rose 107%. As Piven and Cloward put it: "Fully 71 percent of the huge welfare increase during the 1960s took place in the four years after 1964. It was truly an explosion" (1971, 187). Burke and Burke argue:
The explosion in family benefit recipients put welfare, a subject typically shunned by the White
House, on the agenda of President-elect Nixon.... The welfare explosion angered taxpayers and
put severe pressure upon state treasuries, especially in such states as Illinois, California,
Pennsylvania, and New York. Their Republican governors wanted relief from Washington and
from their party's president-to-be. (1974, 41)
Thus Nixon came under immediate pressure from states and their taxpayers to provide some form of fiscal relief from the mushrooming cost of welfare.
Second, urban disorders involving largely poor black males erupted with mounting frequency. Moynihan notes:
In 1965 there had been four major riots and civil disturbances in the country. In 1966 there were
twenty-one major riots and civil disorders. In 1967 there were eighty-three major riots and civil
disturbances. In the first seven months of 1968 there were fifty seven major riots and
disturbances.... Mayors, governors - presidents - took it as given that things were in a hell of a
shape and that something had to be done. (1973,102-3)
Thus it fell to Nixon to somehow restore domestic tranquility.
However, Moynihan also observes that "to compound the difficulty of devising a viable response there was then beginning to be voiced a generalized sentiment among the white working-class that it was being discriminated against, even exploited, in the interest of lower-class minorities" (1973, 104). The civil rights movement victories and years of Johnson's Great Society programs had engendered a reaction of resentment among some working-class whites. They began to see "lower-class minorities", with government help, threatening their jobs, schools, neighborhoods, etc. - i.e., their places above those minorities in the social order.
So FAP became the something that had to be done largely because holdover NIT advocates from the Johnson Administration joined other such advocates in the new Administration to persuade Nixon that an NIT approach to "welfare reform" could deal with the immediate political situation that he faced. Moynihan stresses:
The events leading to and from the proposal of FAP have a conceptual unity that admits
of separate treatment as a long range development in social policy. The proposal was made,
however, as part of an over-riding short term strategy to bring down the level of internal violence.
This is a matter to be dealt with many years hence, if ever. (1973, 12)
FAP would provide fiscal relief for states by federalizing a large portion of welfare costs. FAP would placate angry black males by making those with families eligible for benefits, and so restore domestic tranquility. FAP would
lessen white working-class resentment of "lower-class minorities" by making those white workers eligible for benefits, and requiring all who receive benefits (except for female household heads with small children) to work in order to receive them.
Friedman saw an NIT as a long term rationalization of income transfer policy rather than as a short term response to a political situation. He wanted an NIT to replace "our present collection of welfare measures" including "old age assistance, social security benefit payments, aid to dependent children, general assistance, farm price support programs, public housing" (Friedman 1962, 193) which he saw as poorly aimed at alleviating poverty. He also cautioned that:
Like any other measure to alleviate poverty, it reduces the incentives of those helped to help
themselves, but it does not eliminate that incentive entirely, as a system supplementing income
up to some fixed minimum would. An extra dollar earned always means more money available
for expenditure. (Friedman 1962, 192)
So Friedman recognized that any income transfer program would probably discourage work to some extent.
In contrast to Friedman's proposal, the Nixon Administration did not seek a comprehensive revision of all income transfer programs. Since programs like old age assistance and social security benefits enjoyed wide popular support, it steered clear of them. FAP aimed only at the areas where the political conflict had arisen: aid to dependent children and general assistance. To curb political resentment among the white working class, FAP added a work requirement to its NIT. So FAP differed from Friedman's NIT proposal in its lack of comprehensiveness and its work requirement - ways that would make it more broadly popular. For example, Moynihan reports that a Harris Survey found 92 percent support for the FAP work requirement (1973, 269).
So an NIT began as a proposal of a politically conservative academic economist to make government efforts to alleviate poverty more efficient by targeting funds more precisely at those with lower incomes. The .
research arm of a politically liberal federal anti-poverty agency under Johnson sought to make the proposal more concrete, but foresaw little near term political interest in it. Nevertheless, NIT proposals continued life at OEO and HEW under Nixon - partly in the hands of holdover Johnson Administration researchers. As Nixon's chief domestic policy advisor, Moynihan came to favor an NIT as part of "welfare reform" and was influential with the President. Nixon planners shaped their NIT to deal with the tumultuous domestic political situation that the Administration faced: mounting welfare costs, urban disorders, and resentment among elements of the white working class.
This narrative explanation of what put an NIT on the political agenda centers on reasons why the Nixon Administration proposed FAR Given his position in the Nixon Administration, Moynihan 1973 serves as a primary source for my story. Why would he credit the element of containing domestic disorders in Administration thinking if
2it were not so? Moynihan often criticized Piven and Cloward yet his account of why the Administration proposed
3FAP is consistent with their view that the State expands poor relief to cope with the unrest of the poor. This
agreement among usual political combatants strengthens the credulity of Moynihan's account. Finally, Burke and Burke 1974 is another primary source for my story. As journalists who closely covered the Nixon Administration welfare initiative with no obvious political axes to grind, the Burkes also reinforce the Moynihan account.
This paper uses the usual political historian's tool, careful scrutiny of historical texts and documents, to help construct a narrative explanation of the 1970-72 FAP congressional defeat. However, to aid the construction it also uses the conventional political science arbiter of cause and effect relations, multiple logistic regression models, to analyze FAP congressional votes. I begin by drawing on earlier narratives to help model the 91st Congress House floor and Senate Finance Committee FAP votes. Then the results of these models help to structure a narrative of the 91st Congress FAP defeat. Next I model the 92nd Congress floor votes and use those results to help structure a narrative of the 92nd Congress FAP defeat. I conclude by distinguishing liberals from conservatives in the FAP battles and critiquing earlier narratives of the FAP congressional defeat.
THE 91ST CONGRESS FAP BATTLE
Once its Chairman Wilbur Mills (D, Arkansas) and ranking Republican John W. Byrnes (Wisconsin) decided to modify and back FAP, the House Committee on Ways and Means formulated and reported a bill by a 21 to 3 vote on March 11, 1970. Mills led the House floor debate for FAP where his bill passed 243 to 155 on April 16, 1970. However, in the Senate Finance Committee, ranking Republican Williams (Delaware) firmly opposed the FAP welfare reform approach and Chairman Long (D, Louisiana) played an ambiguous role. A modified version of the House bill suffered a 10 to 6 Senate Finance Committee defeat (with Long voting for the bill) on November 20, 1970 and no version of FAP ever reached the Senate floor during the 91st Congress.
From Previous Narratives to a Model
What variables should one include in a model that attempts to explain the April 16, 1970 House floor FAP vote? Previous narratives are a logical source for such explanatory factor candidates. Though they often differ with him on the importance of various factors and the way that they came into play, most earlier narratives employ factors first identified by Moynihan as those that might explain the defeat of FAR He first identifies three such factors explicitly:
The congressional response to a proposal such as Family Assistance may be analyzed
formally in terms of the two parties, or of the two political tendencies, generally labeled liberal
and conservative, but in reality the major question was how the South would respond. (1973,
So the three factors are: section, political party, and political tendency.
Later Moynihan implicitly suggests a further explanatory factor candidate. By the spring of 1970 the antiwar movement was relatively quiescent as Nixon proceeded with troop withdrawals. However, on April 30, 1970, during the Senate Finance Committee FAP hearings, Nixon announced the U. S. move
into Cambodia and the antiwar movement, including its congressional wing, sprang back to life. About FAP at that point, Moynihan remarks:
A rush of public support could have forced a decision, and the decision, in those circumstances,
would probably have been favorable. But no such surge of support took place, first because of
the war, which greatly aroused the old distrust and hatred of the president ... After a point the
war receded. (1973, 498-9)
Certainly feelings against Nixon ran extremely high at various points during the war so that mistrust of Nixon engendered by the Vietnam War is a fourth explanatory factor candidate.
An indicator of a vote for FAP is the outcome variable of a model of the floor vote. How, though, can one make operational the explanatory factors into variables? Indicators will make operational section and political party. SOUTH indicates a state that the Census Bureau so designates and REPUBLICAN indicates a Republican. The lobbying group Americans for Democratic Action (ADA) recorded votes on a selection of Vietnam war measures during each Congressional session. ANTIWAR indicates whether a member voted the anti-administration antiwar position on the one such measure ADA recorded during the 1970 House session, and so reasonably makes operational Vietnam-engendered mistrust of Nixon.
More difficult to make operational is political tendency. ADA was "liberal" in some sense, and recorded votes on a number of measures (including those on Vietnam) that it deemed important during each Congressional session. It labeled the proportion of those measures on which a member voted the ADA position as his or her "liberal quotient". So if LIBERAL is ADA's "liberal quotient", then LIBERAL makes operational political tendency. (Appendix Two details the data sources and measurement of each variable employed in the paper's modeling.)
The difficulty with LIBERAL as an operational version of political tendency is that a scalar measure of a tendency to vote with ADA does not really capture in an analytic fashion what the terms "liberal" and "conservative" mean. In particular, LIBERAL does not capture the connotations of those terms as the political discourse in the period under study bandied them about. So should LIBERAL prove a helpful explanatory factor in a model, some burden to carefully analyze the liberal/conservative distinction in the context of the FAP conflict will remain.
Finally, what prior expectations for signs of coefficients in a model including the operational variables do previous narratives and political logic suggest? Moynihan's narrative suggests that the signs of the coefficients of SOUTH and ANTIWAR will be negative. With FAP's income redistribution toward the poor, a positive sign for the coefficient of LIBERAL is plausible. With LIBERAL and other variables fixed, political logic suggests that Republicans would follow the lead of their President so that a positive sign for the coefficient of REPUBLICAN is plausible.
House Floor FAP Vote Model Results
Table 1 presents results from using maximum likelihood coefficient estimation in a single equation multiple logistic regression model, with all explanatory variables entered linearly, of the April 16, 1970 House FAP floor vote. Descriptively, the sign of each coefficient estimate is what previous narratives and political logic suggest. To interpret the results causally one must at least assume that the model functional form is correct, the list of operational variables includes all that are of causal importance, and nominal are indistinguishable from actual p-values.
INSERT TABLE 1 HERE
That the p-value for the ；2 test is <.0001 means that variability in explanatory variables explains variability in log (odds of a vote for FAP), in the sense of "explain" that statisticians employ, at any conventional ;-level. That
the coefficients of LIBERAL, ANTIWAR, and REPUBLICAN differ from zero at any conventional a-level while that of SOUTH does not, suggests that those explanatory narratives that stress too much the role of the South in defeating FAP may be in error.
Since each explanatory variable ranges from 0 to 1, it makes sense to informally compare，estimated
coefficient，values. Such a comparison estimates LIBERAL has about four times the effect on log (odds of a vote for FAP) as the next most influential explanatory variable. Hence the results of the logistic regression model of the
stHouse floor FAP vote suggest that a narrative account of the 91 Congress defeat of FAP should center its
explanation primarily on political tendency.
Senate Finance Committee FAP Vote Models and Results
Moynihan's original narrative portrayed the first Senate Finance Committee FAP vote as the dramatic climax of the 91st Congress FAP battle, and the center of political controversy in subsequent narratives is often the activities of that Committee. As part of the FAP conflict some members of the Senate submitted bills that provided more income redistribution toward the poor than did FAP (Moynihan 1973, 451-3). Sponsorship of one of these bills might have led a Senator to oppose FAP so that such sponsorship is a plausible fifth explanatory factor in a model of the November 20, 1970 Senate Finance Committee vote. Let EGALITARIAN indicate sponsorship of a bill more generous to the poor than FAP.
A fresh reading of the hearings and debates surrounding reform suggests that in fact the
critical issue was work, not welfare. Congressmen divided, in the main, not over the preferred
cost or extent of welfare, but over whether employable recipients
should face serious pressures to work in return for support. Only a small minority on the right
opposed welfare, or even its extension, as such. The real battle was over what
kind of work requirements should be attached to the new benefits that most members wanted.
Reform died, in essence, because conservatives and moderates demanded more onerous
requirements than liberals would accept. (1986, 104)
So Mead claims that concern on the part of moderates and conservatives for a strong work requirement killed FAR Hence, in seeking to explain the Senate Finance Committee first FAP vote, an explanatory factor candidate to add to the others is: concern for a strong work requirement.
Mead does not present evidence for how each member of the Senate Finance Committee felt about the FAP work requirement. However, using quotations, his narrative does make the case - presumably as best the historical evidence allows - that many in Congress wanted a strong work requirement. His close reading of the Committee's 1970 FAP hearings identifies six members expressing concern for a strong work requirement: Bennett, Curtis, Fannin, Hansen, Long, and Talmadge. So let WORK indicate Mead-identified concern for a strong work requirement. Then WORK is a plausible way to make operational concern for a strong work requirement.
Let LIBERAL, SOUTH, and PARTY be as in the House FAP floor vote model. Let ANTIWAR now be the proportion of votes, on the two Vietnam war measures that ADA recorded during the 1970 Senate session, that a member cast for the antiwar anti-administration position. Table 2 contains results from modeling the November 20, 1970 Senate Finance Committee vote. Logistic regression results are all from single equation models that use maximum likelihood coefficient estimation with all explanatory variables entered linearly.
Descriptively, the signs of the SOUTH, REPUBLICAN, and WORK coefficient estimates in the multiple logistic regression model are as previous narratives or political logic suggest. However, those of the LIBERAL and ANTIWAR coefficient estimates are the opposite of what previous narratives suggest. Further, the logistic regressions of FAP vote on LIBERAL and ANTIWAR singly also have coefficient estimates with signs opposite those suggested by previous narratives.
INSERT TABLE 2 HERE
With only 17 observations, one could not expect too much of inferential value to come out of the modeling. Even for such a small number of observations, though, the Table 2 results are disappointing. Except for the 2 x 2 table result that shows that both members who had their own more generous bills did not vote for FAP, no 13--value from any model is even close to statistically significant at the conventional .05 ;-level. The closest another p-value
comes is .285 for the WORK coefficient estimate in the multiple logistic regression model. So modeling provides little help in understanding the first Senate Finance Committee vote.
Moynihan says that the White House liaison with the Senate Finance Committee in the period leading up to its November 20, 1970 vote indicated that Chairman Long was "in 'mental turmoil,' his mind 'sowed with seeds of doubt' that FAP would correct the problems of the present system" (1973, 499). Burke and Burke stress that "Russell Long fiercely opposed Mr. Nixon's plan" (1974, 151). Piven and Cloward assure us that "Chairman Long and others had made it abundantly clear that they would organize a filibuster should the bill ever reach the floor of the Senate" (1979, 346-7). So Long was evidently either undecided about or opposed to FAP.
Table 3 tabulates the November 20, 1970 Senate Finance Committee FAP vote on which much of the conflicting interpretation in earlier narratives of the 91st Congress FAP defeat centers. A first puzzle for anew narrative is: why does political tendency do a good job of explaining the House floor vote but break down in explaining the Senate Finance Committee vote? Table 3 raises two further puzzles for such a narrative. Why did liberals like Harris and McCarthy vote against a bill designed to improve the lot of the poor? Why did Long, who apparently did not favor FAP, vote for it? Earlier narrators embed different answers to these questions in their stories. INSERT TABLE 3 HERE
To resolve the three puzzles a new narrative will have to focus attention on aspects of the historical record neglected by earlier narrators. Do the different FAP versions faced explain the different FAP vote outcomes? What alternatives to FAP did liberal senators offer? How did a liberal senator like Harris develop his position on FAP in the Senate Finance Committee hearings and explain his November 20, 1970 vote? Finally, given that A precedes B in time is a necessary condition for A to cause B, the new narrative will have to pay as close attention to the chronology of events as possible.
The Original Administration FAP
The original Administration FAP contained a federally financed NIT for all families with children. The NIT floor income (guarantee level) was $1600 per year for a family of four. The plan disregarded the first $720 of earned income for that family and used a negative tax rate of 50%. Hence that family would receive some support up to $3920 in annual earnings (see Appendix One).
The bill also required states to add payments above $1600 up to the levels of their Aid to Families with Dependent Children (AFDC) payments, for all who qualified for AFDC by the criteria prevailing at the enactment date. In addition, "[o]ne of the least noticed features of Family Assistance was that it mandated the AFDC-U [Unemployed Father] program in all states" (Moynihan 1973, 467) - an expansion from the twenty three states where it existed in August 1969. Finally, the federal government would provide at least 50% of the funds for the supplements and/or enough to assure that a state would reduce its total welfare expenditures by at least 10% from what prevailed at the enactment date.
So FAP provided more income for some of the poor, and appeared to provide no less for those eligible for
AFDC, than did existing welfare-tax arrangements. The "working poor" got support for the first time. In areas like the South where wages and welfare benefits were low, the $1600 FAP figure provided income increases for large numbers of the poor. Welfare recipients in the urban North received no income increases from the bill because they already received more than $1600. However, they also suffered no direct losses of benefits from it and the bill
consolidated the gains secured by the welfare rights movement in the previous period by mandating prevailing eligibility criteria and benefit levels that states were not otherwise legally obliged to maintain. Finally, the fiscal
4plight of each state would improve under FAP.
So the original FAP targeted income redistribution at the welfare conflict that threatened political stability. Federal funds would ease the fiscal stress that the states faced from the expansion of the rolls led by the welfare rights movement. Law would legitimate income gains and expanded eligibility that the
states and welfare bureaucracies of the urban north had supplied in response to that movement. To quell the resentment with which some of the "working poor" had reacted to that movement, FAP would also include benefits for them.
To meet the complaint voiced by congressional conservatives that welfare paid an increasing number of people not to work, FAP included a "work requirement". Able-bodied adults would have to register for work and be willing to accept "suitable work" or the family would suffer a $300 (of $1,600) reduction in benefits. However, the bill exempted mothers with children under school age and those with a man in the house who registered for work and was willing to accept "suitable work".
FAP also required the federal government to provide child care as a precondition for single mothers' working. Quadagno maintains that "numerous gaps in day care planning rendered it ineffective. One problem was inadequate funding" (1990, 17). Burke and Burke think that "the scarcity of day care assured most mothers immunity against the work rule" (1974, 136). Moynihan argues that the FAP "work requirement" was less coercive than the one in the Work