Cecil Stoughton/LBJ Presidential Library

Lyndon Johnson campaigning in Illinois in 1964, the year he declared ‘war on poverty’

This administration today, here and now, declares unconditional war on poverty in America. I urge this Congress and all Americans to join with me in that effort. It will not be a short or easy struggle, no single weapon or strategy will suffice, but we shall not rest until that war is won.
—Lyndon Johnson, State of the Union Address, January 8, 1964

Some years ago, the federal government declared war on poverty, and poverty won.
—Ronald Reagan, State of the Union Address, January 25, 1988

Lyndon Johnson became president in November 1963. In January 1964 he committed the United States to a war on poverty. In August he sought and gained authority to expand the war in Vietnam. Of course, the War on Poverty was only a figure of speech—a political and economic promise, not a war from which young men would return in body bags. Nonetheless, most Americans look back on the two wars as kindred failures. Both have had an exemplary part in the disillusionment with government that has been reshaping American politics since the 1970s. Asked about their impression of the War on Poverty, Americans are now twice as likely to say “unfavorable” as “favorable.” In one poll, given four alternative ways of describing how much the War on Poverty reduced poverty, 20 percent chose “a major difference,” 41 percent chose “a minor difference,” 13 percent chose “no difference,” and 23 percent chose “made things worse.”1

Legacies of the War on Poverty is a set of nine studies, edited by Martha Bailey and Sheldon Danziger, that assess the successes and failures of the diverse strategies that Johnson and his successors adopted to reduce poverty. The chapters are packed with evidence, make judicious judgments, and suggest a higher ratio of success to failure than opinion polls do.

Before discussing specific anti- poverty strategies, however, I must note one major gap in Legacies. The War on Poverty was more than just a bundle of programs; it was Johnson’s bid for a place in history. He announced an “unconditional” commitment to do whatever was necessary to raise the incomes of the poor. He also realized that no one really knew how to eliminate poverty without resorting to politically unacceptable methods, like just sending checks to everyone who was poor. When he said that “no single weapon or strategy will suffice,” he was warning Congress and the country that success would require trial and error. When he added that “we shall not rest until that war is won,” he was promising that even if some of his early initiatives failed he would not cut and run but would instead try something new.

Johnson also knew that he would have to leave the White House before success was achieved, although he did not know that he would be gone in only five years. In addition, he knew that a State of the Union Address could not bind his successors to continuing his efforts. Winning a war on poverty therefore depended on his ability to persuade Congress and his fellow citizens that eliminating poverty was a moral imperative. If he could do that, future presidents and legislators would pursue the War on Poverty as a matter of political self-interest. Otherwise, poverty would persist.

Johnson was driven from office in 1968, not by the failure of his War on Poverty but by the failure of his war in Vietnam. The next two presidents, Richard Nixon and Gerald Ford, were Republicans who had never worried much about the poor. Yet despite that fact, the War on Poverty continued after Johnson went back to his ranch. Democrats retained control of Congress for another twelve years, and many of them remained committed to reducing poverty.

As a result, some of today’s most important antipoverty programs, such as food stamps, Supplemental Security Income (a guaranteed minimum income for the elderly and disabled), and Section 8 rent subsidies for poor tenants in private housing, were either launched or dramatically expanded between 1969 and 1980. Had Johnson not put poverty reduction at the heart of the Democrats’ political agenda in 1964, it is hard to imagine that congressional Democrats would have made antipoverty programs a political priority even after Republicans regained control of the White House. This is the big story about the War on Poverty, which provides the setting within which the war’s specific programs need to be assessed.

Given that the War on Poverty was a commitment to eliminating it, the most obvious measure of the war’s success or failure is how the poverty rate has changed since 1964. Bailey and Danziger argue that just looking at changes in the poverty rate is a “simplistic” approach to assessing the War on Poverty, and in one sense they are right. If you want to know how well programs like Head Start or food stamps worked, or how many full-time jobs they created, the reduction in poverty over the past half-century is not a sensible measure.


But Bailey and Danziger’s argument is more fundamental. They object to using trends in poverty as a measure of the war’s success because the prevalence of poverty depends not just on the success or failure of policies aimed at reducing it but also on other independent economic and demographic forces, like the decline in unskilled men’s real wages and the rising number of single-parent families. They are right about this. But Johnson’s promise to eliminate poverty was not contingent on favorable or even neutral economic and demographic trends. His promise was “unconditional,” because he wanted his country to make a moral commitment to end the suffering that poverty causes.

Bailey and Danziger also give a second reason for not using the poverty rate to measure the War on Poverty’s success, which is that the official poverty rate is probably misleading. That too is true. But any assessment of the war’s political legacy requires a detailed discussion of just how misleading the official poverty rate really is, why its flaws have been allowed to persist for decades, and how their persistence undermined political support for efforts to reduce poverty.

The Census Bureau publishes a table every September showing its estimate of the “official” poverty rate for the previous calendar year, along with the rate in every prior year back to 1959. Figure 1 (see below) shows these estimates. They indicate that 19 percent of Americans were poor in 1964. Five years later, in 1969, the official rate had fallen by roughly a third, to 12.1 percent. Had the poverty rate continued to fall by a third every five years, it would have been 5 percent in 1979 and 2 percent in 1989. Had that happened, Johnson’s claim to a place in history would have gotten a big boost.


According to Figure 1, however, there was no clear trend in poverty after 1969, either up or down. The official rate rose in the wake of recessions, reaching 15 percent in 1983, 1993, and 2010–2012, and it fell during recoveries, dropping to 11 or 12 percent in 1973, 1979, 2000, and 2006. If you believe Figure 1, therefore, the War on Poverty got off to a promising start between 1964 and 1969 but then turned into a stalemate.

Before accepting that conclusion, however, you need to ask where the numbers in Figure 1 come from and whether you should believe them. The Census Bureau derives the numbers from a large household survey called the Annual Social and Economic Supplement (ASEC), which tries to interview the “householder” in a representative sample of residences. The “householder” must be one of the people who owns or rents the residence. If a married couple owns or rents the residence, either partner can be the householder. The Census Bureau also asks who else lives in the household, and whether they are related to the householder by birth, marriage, or adoption. Everyone related to the householder is considered part of the same “family.” Those who live alone or with nonrelatives are considered “unrelated individuals.”

The survey also includes detailed questions about how much money each household member received during the previous calendar year from different sources, such as self-employment, wages, or unemployment benefits. “Family income” is the total pre-tax money income of everyone in the householder’s family. To decide who is poor, the Census Bureau compares each family’s total income to a poverty threshold that depends on the size of the family and the ages of its members. If a family’s total income is below its poverty threshold, the bureau counts all its members as poor. Taken together, these thresholds are known as the poverty line.

The Census Bureau raises the poverty line every year by the same percentage as the increase in the Consumer Price Index for All Urban Consumers (CPIU). The CPIU does not try to measure changes in what people need, or why they think they need it. Nor does it measure changes in what Americans mean when they talk about poverty. The CPIU just measures changes in prices. But if we want to know how close America has come to eliminating the kind of poverty that existed in 1964, the official measure is supposed to provide an answer. However, the world has also changed in other ways that the official poverty count either ignores or mismeasures. As a result, Figure 1 does not actually tell us much about changes in the kind of poverty that Lyndon Johnson promised to eliminate. Four changes are especially important when we try to measure changes in the poverty rate since 1964.

Cohabiting couples. Imagine two twenty-five-year-olds who are romantically involved, live together, and each earned $12,000 in 2013. If they were unmarried, the Census Bureau would have classified them as unrelated individuals, with poverty thresholds of $12,119 each. Since their incomes were only $12,000, the bureau would have counted them both as poor. They would each have needed at least $12,199, bringing their total household income to at least $24,238, for the bureau to stop counting either of them as poor.


Had they been married, however, the bureau would have taken a more upbeat view of their economic situation, classifying them as a family of two with a poverty threshold of $15,600. As a result of this change they would both have been above their poverty threshold instead of below it. According to the Census Bureau’s most recent data, 11 percent of all opposite-sex couples who lived together in 2012 were unmarried.2 We don’t have such a figure for 1964, but it was probably only 1 or 2 percent. The assumption that cohabiting couples need more income than married couples has therefore raised the official poverty rate. This increase in the poverty rate would make sense only if the absence of a marriage license increased a couple’s expenses by 55 percent (from $15,600 to $24,238). The Census Bureau has never tried to defend that assumption, presumably because it is a byproduct of rules set by the Office of Management and Budget, which the Census must follow whether it likes them or not.

Noncash benefits. Noncash benefits now provide many low-income families with some or all of their food, housing, and medical care. Such programs were either tiny or nonexistent in 1964, and their growth has significantly reduced low-income families’ need for cash. However, the Office of Management and Budget (OMB) does not allow the Census Bureau to incorporate the value of these benefits into the recipients’ poverty thresholds. The President’s Council of Economic Advisers estimates that even if we ignore Medicare and Medicaid, food and housing benefits lowered the poverty rate by 3.0 percentage points in 2012.3

Medical care is by far the most expensive of today’s noncash benefits, and Medicaid and veterans’ benefits now pay for most of the big medical bills that poor families incur. However, incorporating these programs’ value into poverty calculations is more difficult than incorporating food and housing subsidies. Most of what Medicaid spends on the poor is for “big ticket” items, like nursing homes, heart surgery, and cancer treatments, that poor families have never been able to pay for out of their own income.

Before Medicaid was created, the poor sometimes got such care from state and municipal programs or from doctors and private hospitals that offered “uncompensated” care. Medicaid coverage has undoubtedly made such care available to many poor families that previously went without it, saving some lives and improving many others. But it has not had the same effect as food stamps or rent subsidies on poor families’ nonmedical standard of living. When a poor family gets food stamps or a rent subsidy, it spends less of its cash on food and shelter and has more to spend on the phone bill, fixing the family car, or taking a child to McDonald’s for her birthday. Medicaid frees up far less money for such uses than food stamps or a rent subsidy, because poor families without Medicaid cannot afford to set aside enough money for major medical emergencies. They know that if they need expensive care they will somehow have to get it free or else do without.

Of course, Medicaid also pays some relatively small medical bills that poor families without Medicaid pay out of their own pockets. The best estimates I have seen suggest that in 2010 Medicaid reduced the average poor family’s out-of-pocket medical spending by about $500.4 That does not mean, however, that the introduction of Medicaid in 1965 reduced recipients’ medical bills by the 1965 equivalent of $500, leaving them with more money for everything else. The patchwork of subsidies and free care that existed before 1965 meant that even then poor patients often paid less than the market price for the care they received. I have not been able to find any evidence on how large those savings were. But if the introduction of Medicaid improved poor families’ access to health care without reducing their out-of-pocket medical spending, we should not think of it as having raised their overall standard of living in the same way that the introduction of food stamps or rent subsidies has.

Refundable tax credits. As part of its effort to reform welfare by “making work pay,” the Clinton administration persuaded Congress to expand the Earned Income Tax Credit (EITC) between 1993 and 1996. By 2013 the EITC provided a refundable tax credit of $3,250 a year for workers with two or more children and earnings between $10,000 and $23,000. Because the official poverty count is based on pre-tax rather than post-tax income, these tax “refunds” are not counted as income, even though the working poor often view the checks as the highpoint of their year—the one time when they can afford to live like other Americans.5 According to the Council of Economic Advisers, treating refundable tax credits like other income would have reduced the poverty rate by another 3.0 percentage points in 2012.6

Price changes. Using the Consumer Price Index to adjust the poverty thresholds for inflation pushed up the threshold for a married couple with two children from $3,142 to $23,624 between 1964 and 2013. All the other thresholds rose by the same multiplier (about 7.5). Whether $23,624 bought the same standard of living in 2013 that $3,142 bought in 1964 is an almost unanswerable question. If no new goods or services had been introduced since 1964, if the quality of existing goods and services had not changed, if poor people still wanted the same mix of goods and services as in 1964, and if the prices of all goods and services had risen by a factor of 7.5, almost everyone would agree that multiplying the 1964 poverty thresholds by 7.5 was the right way to correct for inflation. Reality, however, does not meet any of these requirements. Many things that were for sale in 2013 did not exist in 1964, the quality of goods and services available in both years changed at different rates, and prices of identical goods and services also changed at different rates. As a result, even economists cannot agree on how much the value of a dollar has changed.

That said, there is a fairly broad consensus among economists that the CPIU has overstated the cost of maintaining a constant standard of living over the past fifty years, although they disagree about the size of the bias. The most widely used alternative to the CPIU is the chain-price index for Personal Consumption Expenditure (which I will call the “PCE index”). The Commerce Department’s Bureau of Economic Analysis constructs this measure to calculate changes in the total value of all the consumer goods and services produced in the United States each year. The PCE index is therefore the largest single influence on government estimates of economic growth. If the poverty thresholds had risen in tandem with the PCE index rather than the CPIU since 1964, the 2013 poverty line would have been 20 percent lower than it was, and the 2013 poverty rate would have been about 3.7 percentage points lower than it was.7


Figure 2 provides a first approximation of how correcting the 2013 poverty rate for noncash food and housing benefits, refundable tax credits, and upward bias in the CPIU would change the 2013 poverty rate. With these corrections the official poverty rate falls from 14.5 to 4.8 percent, making the 2013 rate roughly a quarter of the 1964 rate (19.0 percent). If we were to lower the poverty threshold for cohabiting couples to match that for married couples the 2013 poverty rate would have fallen even more.

The estimates in Figure 2 are not exact. More important, their combined effect may be smaller than the sum of their separate effects, making the drop in the “true” poverty rate smaller than Figure 2 suggests. But even if the true poverty rate was 6 or 7 percent in 2013, it would have fallen by about two thirds since 1964, putting it considerably closer to what Lyndon Johnson had promised in 1964 than to what Ronald Reagan had claimed in 1988.8

Fixing these flaws in the official poverty rate helps reconcile trends in poverty with trends in more direct measures of material well-being. Today’s poor live in less crowded housing, are more likely to have a complete bathroom and air conditioner in their residence, have bigger TV screens, are more likely to have a telephone, and more likely to have a cell phone. Nonetheless, most of the poor are still beset by constant financial anxiety. In part, that is because the poverty line was set so low in 1964. Linking the poverty line to the Consumer Price Index let it rise a little every year, but not much. Using a more realistic price index keeps the poverty line closer to its real 1964 level, ensuring that those we count as poor are more like those we counted as poor fifty years ago, but in both periods those just above the poverty line have suffered from many of the same problems.

Another reason the poor so often feel beleaguered, anxious, and depressed may be that what is often called “relative poverty” has not changed. Over time, any society’s definition of poverty adjusts up or down depending on how much income those in the middle of the distribution have. There is quite a bit of evidence that Americans need an income at least half that of families near the middle of the distribution in order to buy the things they need to hold up their heads in public. In such a world, the only way to reduce the number of people who feel and act poor will be to reduce the number with incomes less than half the 50th percentile (the median). If we adjust for noncash benefits, taxes, and changes in family size, incomes at the 10th percentile were 39 to 40 percent of incomes at the 50th percentile in both 1967 and 2012.9 It follows that a bit over 10 percent of American families had incomes less than half the median in both years. Using a relative measure not much had changed, even though the absolute poverty rate that Lyndon Johnson promised to reduce has fallen dramatically.

Both liberals and conservatives tend to resist the idea that poverty has fallen dramatically since 1964, although for different reasons. Conservatives’ resistance is easy to understand. They have argued since the 1960s that the federal government’s antipoverty programs were ineffective, counterproductive, or both. Since the 1970s they have cited the stability of the post-1969 poverty rate to support those judgments. To them, the suggestion that poverty has fallen sounds like a suggestion that the War on Poverty succeeded.

Liberals hear the claim that poverty has fallen quite differently, although they do not like it any better than conservatives do. Anyone, liberal or conservative, who wants the government to solve a problem soon discovers that it is easier to rally support for such an agenda by saying that the problem in question is getting worse than by saying that although the problem is diminishing, more still needs to be done. The equation of “bad” with “worse” is so tight in American political discourse that when I tell my friends or my students that “there is still a lot of poverty, but less than there used to be,” they have trouble remembering both halves of the sentence. Some remember that there is still a lot of poverty. Others remember that there is less than there used to be. Few remember both.

Although I have argued that the absolute poverty rate has declined dramatically since President Johnson launched his War on Poverty in 1964, it does not follow that the programs he launched between 1964 and 1968 caused the decline. I argued that food stamps, rent subsidies, and refundable tax credits all had a role in the decline, but food stamps did not become a national program until the end of the Nixon administration, the fraction of poor families receiving rent subsidies grew quite slowly, and refundable tax credits remained tiny until 1993. The growth of these programs was nonetheless inspired partly by Johnson’s earlier success in convincing much of the Democratic Party that poverty reduction was a political and moral challenge they could no longer ignore. The successes and failures of specific anti-poverty programs will be the subject of a second article, which will appear in the next issue.