CHAPTER 2

POVERTY

POVERTY HAS BLIGHTED EVERY society in every era. Deuteronomy tells us: “There will always be poor people in the land. Therefore I command you to be open-handed toward your brothers and toward the poor and needy in your land.”1 Likewise, Jesus said, “For you always have the poor with you.” (He immediately added “and whenever you wish you can do good to them.”)²

Before the Industrial Revolution, almost everyone was poor by today’s standards. Elites and most of the poor themselves took pervasive poverty as a simple, inevitable, and lamentable fact of social life like illness, early death, deep anxieties, and other forms of human suffering. Until then, economic growth was minimal so few could hope to escape poverty’s grasp, at least in this world. Religion often provided solace along with hope for a better life in the hereafter (leading Marx to deride it as “the opium of the people”—though only after calling it “the heart of a heartless world”). The state’s role was mainly to press the poor to work (often as its own cannon fodder, slaves, and taxpayers), minimize and isolate their vagrancy and violence, prevent them from threatening property, and repress threats to the powerful.³

Popular revolutions, the spread of democracy and suffrage, economic growth, and the gradual development of welfare states in the West transformed poverty into a social issue, an object of meliorative public policy. In the United States, federal solicitude for the poor focused first on veterans’ pensions after the Civil War, and then in the New Deal era, on the elderly, the unemployed, and needy mothers of dependent children. (Some states had already established programs directed toward the poor.) Even then, concern focused mostly on the jobless, not poor people generally.

Antipoverty policy entered a new era when President Lyndon B. Johnson launched what he called a “War on Poverty” in his 1964 State of the Union address. This campaign led to the immediate creation of an Office of Equal Opportunity, establishing community action agencies at the local level,* and instituting over time a wide variety of antipoverty programs. Such programs were adopted at all levels of government. For centuries, of course, religious and other voluntary organizations had sought to alleviate poverty at the local level. Today, even putting aside the political motives, the humane impulses that favor antipoverty programs are obvious: a moral revulsion against persistent destitution in a society as wealthy as ours; a desire to reduce the suffering, and improve the future prospects, of helpless children; and a conviction that poverty lies at the root of many of our most disturbing social problems, including family disintegration, crime, drug addiction, ill health, domestic violence, communal breakdown, and reduced social mobility.

Some fifty years—two or three generations later—tax professor Edward Kleinbard assesses where we stand despite those efforts. “The United States has the highest poverty rate, the greatest income inequality, and the greatest wealth inequality of any major developed economy in the world. Our parents’ incomes play a larger role in our personal economic fortunes than is true for other peer countries.”⁴ Future progress depends, among other things, on understanding the nature, magnitude, sources, and consequences of poverty and of limited mobility in America.* At the same time, synthesizing the relevant social science research, policy design, and program implementation should also deepen our appreciation of how hard it is to define poverty in a rigorous way, how complex its diagnosis remains, how resistant it has been in the face of a half-century of diverse policy approaches, but also how much progress has been made in reducing it.

I do not review either the history of poverty or the moral issues that surround it. Large literatures exist on both of these topics—especially the moral issues that have long preoccupied many religious writers, novelists, social philosophers, and other commentators.⁵ Instead, I proceed on two assumptions: Alleviating poverty is an urgent social imperative, and the social science that I synthesize here provides the best basis for understanding poverty and thus the soundest foundation for remedying this ancient scourge.

This is the longest chapter in the book, as befits its importance, its complexity, and the enormous amount of social science research devoted to it. Accordingly, I hope to help the reader by identifying its major themes in advance. The measure of poverty is highly controversial, but by all accounts it has declined significantly since 1965 (and sharply in 2015). The standard of living of the poor has improved substantially, but certain subgroups—particularly jobless black men and mothers and children in female-headed households—are worse off in some important respects. Poverty has many causes—bad luck, bad choices, family breakdown, educational deficits, dysfunctional and self-reinforcing cultural patterns, social and neighborhood isolation, immigration, discrimination, mass incarceration, and economic dislocations among them—and most of those causes are also its effects, producing complex, tragic feedback loops. (One cause, immigration, can cut both ways.) Antipoverty programs vary in their effectiveness, with the elderly benefiting most and being affected least by the work disincentives built into many of the programs. Remedies for poverty do exist, but they all entail difficult trade-offs that must be seriously considered.

The chapter proceeds in four parts, beginning with the context of antipoverty policy. This includes the distinction between poverty and inequality; the definitions of poverty; its measures; its incidence among certain subgroups; intergenerational mobility; and poverty trends over time. A second part analyzes its many complex causes. The third part discusses existing antipoverty programs—their nature, size, and effectiveness. The final part discusses how policy makers can best seek to reduce poverty going forward.

CONTEXT

Although the chapter focuses on poverty, I begin by briefly discussing inequality in America. Because poverty and inequality are closely correlated as a factual matter, people often use these terms interchangeably, as if they are essentially the same conditions, pose the same problems, and yield to the same approaches. In truth, however, they are distinct phenomena. Treating them as if they were the same engenders confusion and can lead to incoherent policy. The rest of the chapter, then, analyzes poverty as such.

Inequality

Rigorous historical and empirical study of inequality only became possible in the second half of the twentieth century when economists Simon Kuznets (in the United States) and Anthony Atkinson (in the UK) developed the necessary databases, analytical tools, and methodology.6 Although researchers have had access to aggregate federal income tax data for about a century, household survey data did not become available until after 1960 and more refined data until even more recently. Today, public debate about inequality is more intense, partly spurred by a long, highly technical, even abstruse, economic analysis of inequality by Thomas Piketty, a French economist previously known only to the professional cognoscenti. Improbably, Piketty’s tome became a best-selling book in the United States when an English translation was published in 2014.7

Social concern about inequality is widespread—more among liberals, who point to its unfairness and social corrosiveness, than among conservatives, who stress its inevitability in a competitive economic system and oppose many proposals for combating it (often because they think those policies will increase it). Today’s concern reflects recent evidence that inequality has grown significantly since the 1970s. A 2015 analysis by Emmanuel Saez, a frequent Piketty collaborator, finds that pretax family market income* inequality rose during the Clinton-era (1993–2000), Bush-era (2002–7), and Obama-era (2009–14) expansions. In these three periods, Saez argues, real average income grew 20.6, 16.1, and 8.4 percent, respectively, yet the income of the top 1 percent (those with incomes above $423,000 in 2014) grew much more, capturing 45 percent of the total gains during the Clinton expansion, 65 percent during the Bush expansion, and 58 percent during the Obama expansion.⁸ Saez claims that this spread continued to grow even in 2013–14, as real recovery from the Great Recession progressed, when average family income rose 4.8 percent (the highest annual increase since 1999). In 2014, he finds, the bottom 99 percent’s income grew substantially (3.3 percent), but the top 1 percent’s incomes grew even faster (10.8 percent). The spread also widened a bit further down the income distribution: The share of the top 10 percent (incomes above $121,400), driven mainly by the top 1 percent, was the highest ever except for 2012.⁹ More recent analyses find that this tax data overstates inequality by not including nontaxable income received by lower-quintile groups.¹⁰ Inequality is also much lower when measured on an after-tax basis and taking government transfers into account, which neither Saez nor the official poverty statistics do. Inequality, thus measured, has actually narrowed since 2007 even though income grew at a negligible rate until very recently.11 And in 2015, income gains were much larger—in percentage terms—for households in the tenth percentile of income than for those in the ninetieth.¹²

Inequality (whether it is widening or not) has many causes, and economists and other social scientists are by no means agreed about their relative importance, policy remedies, and future trends.¹³ Political scientists Martin Gilens and Benjamin Page emphasize what they term “economic-elite domination” and “biased pluralism” that stack the political decks against the poor.¹⁴ For economists, the usual suspects include: a growing gap between the share of national income going to labor and capital; wage pressures from global competition; technological change; higher returns to college education and advanced skills; the growing fraction of single-parent-headed households and rising divorce levels; the wage gap between high- and low-profit employers;¹⁵ alcohol and drug abuse; the effects of residential segregation; weaker labor unions; and more assortative mating (reversing a pattern of nonwork by the wives of high-earning men).¹⁶ Before the 2015 figures came out, these mating changes alone accounted for an estimated 26 percent increase in household income inequality;¹⁷ if marriage patterns remained the same as in 1960, today’s inequality would be much lower.¹⁸ The 2015 Economic Report of the President finds that if income was shared in the same proportion as in 1973, and if productivity gains and female workforce participation stayed at 1995 levels, middle-class households’ mean income would have doubled in the last twenty years (up $51,000).¹⁹

Since the 2011 Occupy Wall Street protests, politicians and other opinion leaders have used election campaigns and mass media to make inequality reduction a higher priority.²⁰ A 2015 poll finds that a strong majority of Americans want the government to do more to reduce income and wealth inequality, a view that crosses income and partisan lines, but not uniformly.²¹ Yet, a number of studies show that Americans, a highly competitive people, favor some level of economic inequality, reflecting economist Richard Easterlin’s finding that happiness depends on being wealthier than one’s reference group (and H. L. Mencken’s quip defining wealth as “any income that is at least $100 more a year than the income of one’s wife’s sister’s husband”).22 Relative mobility, after all, is a zero-sum game.²³

Americans tend to oppose explicit redistribution, believing that better education is the answer.²⁴ The parties differ sharply on the priority of reducing inequality: 90 percent of Democrats, but only 36 percent of Republicans, want their presidential candidate to focus on it.²⁵ Compared with Europeans, Americans find given levels of inequality less objectionable, partly because we are much more likely than they are to believe that individual effort is more important than luck in causing people’s economic fates.²⁶ Americans also disagree about how much deprivation and suffering actually exist,* a question analyzed later in the chapter.

President Obama calls inequality reduction “the defining challenge of our time.”²⁷ Left-populist politicians like Senators Elizabeth Warren and Bernie Sanders argue that government policies have increased inequality, especially tax, regulatory, and other advantages enjoyed by wealthy business interests. They argue that policy changes could reduce or eliminate these unwarranted advantages.²⁸ Some conservatives also blame the government for the level of inequality. Wall Street Journal editorials often denounce “crony capitalism” (e.g., the Export-Import Bank and subsidies to agribusiness) and “corporate welfare” (e.g., many inefficient and poorly targeted tax breaks). Conservatives also contest some liberals’ claims about inequality as well as many of their specific proposals for reducing it. Ben Bernanke, the former Federal Reserve Board chairman (and no populist), also laments current levels of inequality, but ascribes them less to public policy than to “deep structural economic reasons including globalization, technological progress, demographic trends, and institutional change in the labor market and elsewhere.”²⁹ As Bernanke’s list implies, liberals and conservatives disagree sharply about how unjust these inequalities are and how much they are needed to enhance economic efficiency.

Poverty and inequality overlap significantly but, as noted earlier, they are not the same. Poverty is a condition of deprivation that causes suffering among those who endure it (and among many of us who observe it), and it is conventionally assessed against some measure of well-being. In contrast, inequality is measured by the amount of dispersion within a relevant population with respect to income, wealth, or some other aspect of well-being. Both poverty and inequality exist in even the wealthiest societies; in this sense, both are matters of degree.

We also distinguish between inequality and inequity. In ordinary discourse, we often characterize inequalities as inequitable or unfair—and they often are, by almost all accounts. In principle, however, they are distinct ideas. Inequality is an empirical fact whose conventional measures are discussed below; even Piketty says that it is “not necessarily bad in itself.” By contrast, inequity is not a fact but a moral judgment about the fairness or justness of a particular level of inequality. Almost everyone, of course, agrees that certain inequalities are unjust, but which ones remains a morally contested claim—although less contested where children and some other groups are concerned. In judging the fairness of any particular inequality, most people want to know which choices, behaviors, or factors contributed to it.30 Even if we could answer such questions empirically, reasonable people would surely disagree about how to assess and weigh the normative factors. Notably, Americans view some inequalities as more acceptable than others, countenancing economic inequalities more than political ones.³¹ Even so, many Americans deplore our current levels of economic inequality.

Complicating analyses of inequality is a separate, intensely empirical dispute about the effectiveness of various approaches to reducing it. Policies meant to redistribute wealth could end up reducing the total wealth created by society and thus available (in principle) for redistribution. Economist Arthur Okun called this “the big tradeoff” between equity and economic efficiency. He likened it to a leaky bucket used to carry wealth from the rich to the poor, with some leaks caused by incentive effects reducing productive effort by both rich and poor, and other leaks due to the administrative costs of taxing and transferring wealth from one group to another.32 Whether these leaks could be offset through well-crafted antipoverty policies is a hard question that I address at length later in the chapter.

Not surprisingly, analysts disagree about the magnitude and location of these incentive effects and tax-and-transfer costs, about the factors that create them, about how various policies aimed at reducing inequality would actually work, and even about how such policies have worked in the past! After all, an extraordinarily complex mélange of factors affects economic outcomes, so even the most careful analysis cannot isolate each factor’s causal contribution to inequality. Another factor contributing to this causal uncertainty is “rational expectations”: one’s propensity to anticipate policy changes (most of which will take effect in the future) and adjust one’s own behavior to reduce the new policy’s costs to oneself (e.g., estate taxes on the wealthy, which they can minimize or avoid through astute tax and estate planning).

The key point is that efforts to reduce inequality to help the “99 percent” might possibly make them worse off, depending on these incentive and cost effects.³³ Another hard policy choice is whether to reduce inequality by raising the incomes of the worst off, lowering those of the well-off, or both.³⁴ A recent simulation by Brookings tax experts found that raising the top individual tax bracket to 50 percent and redistributing all the proceeds to the bottom 20 percent of households would reduce the Gini coefficient (discussed just below) by only a trivial amount—even assuming no behavioral changes.³⁵ I explore some of these issues later in the chapter.

Different dimensions of inequality produce different results. For example, consumption is more equally distributed than income, and income is more equally distributed than wealth (net worth). The incomes of families are also more equal than those of individuals, and pretax market income is more unequal than post-tax income. Including fringe benefits in income makes it more unequal, while including means-tested noncash benefits to the poor makes incomes more equal.36 For inequality trends over time, use of particular inflation adjustments and other methodological choices will affect the conclusions.³⁷ Definitional and methodological choices also affect conclusions concerning the trajectory of the “middle class” or any other class. This is evident in varying interpretations of Census Bureau and Pew data released in December 2015.³⁸

Analyses of economic inequality usually focus on the “Gini coefficient,” which measures the dispersion of incomes across society; it ranges from zero (all people have equal income) to 1 (all national income goes to only one person).* By this measure (there are several others),³⁹ the United States is an outlier among the world’s rich nations. Among the thirty-four relatively wealthy nations in the Organization for Economic Cooperation and Development (OECD), the United States has the second highest Gini coefficient (the highest is Mexico; in some comparisons, Chile is also higher than the United States).⁴⁰ When other non-Gini measures mentioned in the previous paragraph are used instead, inequality in the United States is somewhat lower and less of an outlier—but it is still large in comparative terms.

These comparisons, however, mask a very significant, underappreciated fact about inequality: It has been increasing in almost all OECD countries, including many like Sweden and Norway with strong traditions and policies of social democracy. Today, the average income of the richest 10 percent in OECD countries is almost ten times that of the poorest 10 percent, up from seven times in the 1980s.⁴¹ In the United States, this ratio rose from 15.1 in 2007 to 18.8 in 2014.⁴² These OECD trends strongly suggest that many causes of inequality are endemic to post-industrial economies. (Of course, some of them, like the United States, are more unequal than others.) The international differences that remain largely reflect differences in our public policy preferences.

Philosopher Harry Frankfurt recently published a provocative, clarifying essay on the moral status of economic inequality and its relationship to poverty.43 He distinguishes sharply between equality (i.e., whether everyone has the goods that others have, especially money that can be converted to so many other things of value) and sufficiency (i.e., whether they have enough to lead a life that will reasonably satisfy them). He personally favors more equality, believing that more equality will promote various social and political ideals, but he insists that economic equality, unlike sufficiency, is not a moral imperative (i.e., a goal whose goodness is intrinsic, not derived from other things). Morally, he maintains, what matters most is whether we have good lives, not how our lives compare with others’ lives.⁴⁴ Morality is not treating people equally but treating them with respect, which means impartially, not arbitrarily: seeking “outcomes matched specifically to the particularities of the individual,” whether or not those outcomes seem equal.⁴⁵ I might question Frankfurt’s dismissal of comparison in moral analysis, but his goal of sufficiency seems to be the proper goal of antipoverty efforts. If such efforts succeed, equality would probably increase, yet it is sufficiency—assuring that all of us have enough to lead reasonably satisfying lives—that should be the main measure of our success.

Poverty

Definitions. Poverty, as noted earlier, is best understood as a state of deprivation or suffering. It can take many forms: hunger, homelessness, inadequate medical care, emotional stress, social isolation, limited opportunities, the “digital divide,” ignorance, stigma, spiritual emptiness, and other life-limiting conditions. For perfectly understandable reasons, however, officials and the public have sought a specific, numerical definition of poverty that could be calculated each year to determine how poverty is trending. After clarifying these definitions, the rest of the chapter will use the terms “poverty” and the “poor” more loosely not only to include those who meet such definitions but also to include the broader population toward whom anti-poverty programs are typically directed. That the terms are often used in ill-defined and even circular ways is no accident. Indeed, such usage—sometimes careless but often self-conscious and tactical—constitutes an important feature of the poverty debate.

The federal government’s poverty definition was originally devised in 1963–65 by Mollie Orshansky, a Social Security Administration statistician-economist.46 Liberal and conservative analysts criticized it from the very beginning on many grounds, and no one today defends its accuracy.⁴⁷ (Partly to meet these criticisms, and following a National Academy of Sciences report on the subject, the Census Bureau in 2011 introduced a “Supplemental Poverty Measure” (SPM), which among other changes incorporated new family composition patterns and deducted out-of-pocket medical and child care expenses from family income. Leading poverty experts, however, still find it inadequate in many ways.)⁴⁸ Reviewing these critiques brings some of poverty’s measurement difficulties into clear view.

A number of important social changes since 1965 seriously distort—and vastly overstate—the official overall poverty rate, which in 2015 had declined to 13.5 percent. (The SPM rate was 14.3 percent.)⁴⁹ A refined understanding of the poverty rate requires a number of adjustments to the official one. Here are some of the most important:

•  Noncash government benefits are excluded. Including food and housing benefits would reduce the poverty rate by about 3 percentage points. But even those adjustments do not take account of the valuable services that Medicare and Medicaid provide to the poor, services that they would otherwise under-consume.

•  The refundable Earned Income Tax Credit (EITC) is excluded; including it would lower the poverty rate by another 3 percent.

•  The standard consumer price index (CPI) to compare poverty rates over time overstates the cost of maintaining a constant standard of living; correcting for this would have reduced the 2013 poverty rate by 3.7 percent. These first three adjustments alone would have yielded a poverty rate of 4.8 percent, far below the official rate. (That rate declined further in 2015.)

•  The official rate treats cohabiting couples (an estimated 11 percent of opposite-sex couples living together) differently than married ones; treating them the same would lower the poverty rate even more.50

•  The official rate does not consider either regional differences in the cost of living or the episodic nature of much poverty during a given year.

•  Today’s households are smaller, reducing their living costs correspondingly relative to earlier periods.⁵¹ Some measures take this into account, some do not.⁵²

•  Many poor people who work at least part of the year can tap into other “work-based safety net” benefits such as unemployment insurance and the refundable child tax credit.⁵³

When the Congressional Budget Office (CBO) made some (not all) of these adjustments, it found that real median household income actually increased between 1980 and 2010—and not by 15 percent (as in the non-adjusted calculation) but by 53 percent.⁵⁴ (Median income increased significantly in 2015.) If consumption of new products and improved goods and services during that period is included, the increase would be about 2.5 percent a year over the thirty years—a substantial gain.⁵⁵

The official poverty rate is also exaggerated because of vast underreporting of income and government transfers, and over-reliance on Current Population Survey data. An important 2015 study compared Census Bureau data based on responses to its CPS with records of what was actually disbursed to those same respondents. CPS data, it found, sharply understates poor households’ income: “the poverty reducing effect of all programs together is nearly doubled while the effect of housing assistance is tripled,” and “the share of single mothers falling through the safety net [falls] by one-half or more.”⁵⁶ Another leading study on under-reporting finds that “as many as half of the dollars received through Food Stamps, Temporary Assistance for Needy Families (TANF) and Workers’ Compensation [a state program] has not been reported in the [CPS]. High rates of understatement are found for many other government transfer programs and in datasets such as the Survey of Income and Program Participation (SIPP) and the Panel Study of Income Dynamics (PSID). These datasets are among our most important for analyzing incomes and their distribution as well as transfer receipt. Thus, this understatement has major implications for our understanding of the economic circumstances of the population and the working of government programs.”57 Finally, studies of low-income urban communities find that residents receive substantial off-the-books income.⁵⁸

Experts on poverty measures agree that the official poverty measure casts little useful light on either current poverty or poverty trends. Strong arguments exist for focusing not on income levels but instead on consumption levels, which more closely reflect both permanent income and well-being. Bruce Meyer and James Sullivan show that it better identifies those who are less likely to have health insurance and have less education, smaller and cheaper cars, and fewer household appliances and amenities; that it can overcome four biases in the CPI (not factoring in big-box store discounts, substitutions of cheaper products, product quality improvements, and new products); and that mainly for these reasons, a consumption measure substantially reduces the poverty rate, especially for single-parent families and the elderly.⁵⁹

Indeed, studies show large consumption gains by the poor. “Low-income and poverty-level households today,” demographer Nicholas Eberstadt finds, “are better-fed and less threatened by undernourishment than they were a generation ago. Their homes are larger, better equipped with plumbing and kitchen facilities, and more capaciously furnished with modern conveniences. They are more likely to own a car (or a light truck, or another type of motor vehicle) now than thirty years earlier. By most every indicator apart from obesity, their health care status is considerably more favorable today than at the start of the War on Poverty. Their utilization of health and medical services has steadily increased over recent decades.”⁶⁰ Greatly improved living conditions are also reflected in declining crime rates in many low-income neighborhoods.⁶¹ (In 2015, however, homicides in the fifty largest cities rose by almost 70 percent, despite a substantial decline in gun violence.)62 In 2014, the federal government reported that homelessness (which is likely undercounted) had declined by almost a third since 2007.⁶³

Most Americans do not consider these conditions to constitute abject poverty, particularly if informed that the typical poor American has considerably more living space than the average European does.⁶⁴ Consumption-wise, then, today’s poverty problem is not so much absolute material deprivation (Frankfurt’s “sufficiency”) as it is relative poverty—relative both to other Americans and to people in other wealthy countries. Note, however, that while relative poverty may seem an appropriate focus, it means that the poverty rate rises as society becomes wealthier, a logic that may seem counterintuitive.

Subgroups. More fine-grained analyses of subgroups and poverty dynamics provide additional insight into the nature of poverty in America. I first compare child and elder poverty, and then discuss the special case of immigrants. Another subgroup, households with children headed by single women, is central to so many aspects of American poverty that I analyze it extensively below, primarily in connection with family breakdown.

The federal government spends vastly more per capita on the elderly than on children, who are supported mainly by state and local programs: “Whereas aged Americans receive $5.72 in federal spending for each $1.00 received by someone under 19, those under 19 receive $10.11 in state and local spending for each $1.00 received by someone who is 65 or older. To be sure, total federal spending is considerably greater than that of state and local governments, but the imbalance of public spending on the young and the old is less extreme than federal budget statistics suggest.”⁶⁵ Although the official poverty rates for children and the elderly in 2015 were 19.7 and 8.8 percent, respectively, those rates do not include government transfers and tax credits. Common adjustments to the official rates lowered child and elder poverty to about 15 and 2.6 percent, respectively.⁶⁶ (In 2015, these rates presumably declined even further.) Despite seniors’ inflation-indexed Social Security benefits, those with low earnings histories, especially women, remain poor due to lower benefits.

Many immigrants who are poor have nevertheless achieved high, sometimes dramatic, income mobility: they left much poorer countries and earned much more here than they did there. Our poverty index defines them as poor by our standards* even if their income here would make them solidly middle-class in their home countries—presumably their main socioeconomic reference point. Writing almost twenty years ago, economist Robert Lerman explained that our poverty statistics obscure this important fact, analogizing to Germany’s experience after its 1989 reunification when it in effect absorbed a large low-wage country: “The U.S. did not absorb a country equal to 20% of its initial population, but the high levels of immigration to the U.S. in the 1980s and 1990s raise a similar compositional issue. Immigrants who settled in the U.S. since 1980 made up about 7% of the 1996 U.S. work force and over 17% of the work force without a high school degree. Moreover, over 90% of these recent immigrants came from low wage countries.”⁶⁷

Intergenerational Mobility. The most damning fact about American society today is the high correlation between the social status of parents and that of their grown children. The likelihood that children born into poverty will escape it, so celebrated in our self-congratulatory Horatio Alger mythology, is not particularly high, casting doubt on our ability to fulfill the American dream. Intergenerational mobility is among the most crucial of all social indicators—more so than inequality and more so than national poverty measures at any given point in time.

As the earlier quotation from Edward Kleinbard indicates, the United States has lagged badly on this measure. By the most common measure of income mobility (the extent to which a father’s income can predict his son’s income) or indeed by any measure, we are less mobile than most other OECD countries. One study finds that 40 percent of grown sons born here to fathers in the lowest quintile remain in the lowest quintile. (Christopher Jencks, a leading poverty researcher whose work is discussed below, notes that even with perfect mobility, the figure would be 20 percent; the Scandinavian figure is about 30 percent.)68 This low mobility is consistent across all income groups, applies to education level, and has changed little in the last fifty years, a period when our economy vastly expanded. As we shall see, research can readily identify which factors correlate with poverty, but determining causality is much harder.

Trends. The difference between the poverty rate in 1965 and today matters not just to the poor. It also is central to the debate over what the War on Poverty actually accomplished. As the previous discussion showed, this question remains controversial for at least three reasons.

•  Good faith disagreements exist among poverty experts about how poverty should be measured, and the different methods produce quite different results. What is clear, however, is that the poverty rate has declined substantially during this period according to the official measure (19.5 percent in 1964 and 13.5 percent in 2015, a decline of about 30 percent).⁶⁹ As we have seen, the decline in poverty is much more impressive if one includes government transfers and measures it according to actual standard of living.

•  Despite this progress in reducing poverty (however measured), those households with children who live in extreme poverty (defined as less than $2 a day for at least one month per year) were arguably worse off in 2012 than they were in 1996 when TANF was enacted.⁷⁰ However, a 2016 analysis of child poverty rejects this claim⁷¹ and the USDA announced a significant decline in food insecurity in 2015, the first such decline since 2010.⁷² Presumably, the 2015 decline in the child poverty rate further reduced extreme child poverty.

•  Determining precisely which factors caused how much decline in poverty is even more controversial than how to measure it. The combined effects of immense global changes—many poverty-reducing, some not—probably dwarf the causal significance of particular U.S. government policies, yet the sheer size and diffusion of global effects obscure the relative contributions of each. The same multiplicity of factors makes the causal effects of domestic policy changes hard to determine. That said, federal (and some state) transfer and tax credit programs clearly helped to reduce poverty significantly.73 I discuss this further below. Three groups have benefited most. The elderly (particularly “young” ones under the age of seventy-five) have enjoyed substantial increases in Social Security payments while also benefiting from traditional defined-benefit pensions and real estate and stock market gains.⁷⁴ Working-poor parents’ income has been boosted by the EITC. And food stamps and child care, energy, and housing subsidies have also lifted many recipients out of poverty. (Medicare, Medicaid, and other low-income health-care benefits clearly enable them to lower their out-of-pocket expenses, but quantifying this cost avoidance is difficult.)

•  Assessing the second-order effects of antipoverty policies—how they affected recipients’ work effort, independence, family life, and other behaviors—is as difficult as it is important. Virtually all analysts acknowledge the possibility that these perverse incentives can blunt poverty-reduction efforts to some degree. Studies of programs like unemployment compensation and disability insurance have detected such effects.⁷⁵ The size of work disincentives is debatable but they clearly contribute to both programs’ looming insolvencies.⁷⁶

In the following analysis of poverty’s causes, the discussion and data will often focus on blacks. (I explain elsewhere why I say “blacks” instead of “African Americans”; any such term is unsatisfactory in some respects).⁷⁷ The focus on blacks is emphatically not because they account for most of America’s poor. In fact, whites constitute the largest group of poor people (about 19 million). Blacks do not have the highest poverty rate (the rate for Native Americans and Alaskan natives, 27 percent, is slightly higher). Nor are most blacks poor (indeed, more than 75 percent of them were not poor in 2015). And black African immigrants do well here; 35 percent of those aged twenty-five and older had a college degree or better in 2013; 15 percent have advanced degrees;78 and their income is sharply higher than for U.S.-born blacks.⁷⁹ Blacks have reduced the education gap more than have poor people generally, leaving the gap between poor and rich children nearly double that between black and white children.⁸⁰ But this is a mixed blessing, as it also reflects the growing plight of working-class (and nonworking) whites: worsening marital stability, household structure, substance abuse, morbidity and mortality, children’s school readiness, incarceration rates, and employability.⁸¹ Their decline means that while the racial gap in these measures remains deeply troubling, the class gap is even wider. Although the categories significantly overlap, of course, class is now a stronger predictor of disadvantage in many important areas of social well-being.⁸² An even stronger predictor is being born to an unmarried mother and an absent father.

Blacks’ prominence in poverty analysis, then, has many causes. Slavery, Jim Crow, racial hatred, and other forms of oppression—what Gunnar Myrdal famously called “An American Dilemma”⁸³—have isolated them, impeding their climb out of poverty more than probably any others. Also, their share of the poverty population (25.8 percent according to the 2007–2011 Census) is more disproportionate than that of any other large, well-defined group.⁸⁴ (Native Americans are a small group, with much higher out-marriage rates. Hispanics, as explained in chapter 5, are a far less discrete, identifiable group.) Blacks are the largest single component of the long-term poor and of long-term welfare dependents.⁸⁵ That poverty tends to be equated with blacks may also reflect their vastly disproportionate presence in public housing, the iconic image of poverty. Indeed, blacks constitute 48 percent of public housing residents nationwide, almost quadruple their share of the population. In many large cities, the percentage exceeds 90 percent.⁸⁶ They also commit a vastly disproportionate share of violent crimes, disproportionately against other poor blacks. For all these reasons, any analysis of poverty and of possible remedies for it must place blacks at the center, while recognizing that other groups, including whites, are well-represented there.

POVERTY’S CAUSES

This too is an ancient question and one on which strong disagreement persists, even in recent decades when a great deal of social science research has been devoted to answering it. Indeed, the more we have learned about poverty, the more analysts seem to disagree about the mix of poverty’s causes, the effectiveness of various antipoverty policies, or both. One might call some of this disagreement political or ideological in the sense that one can usually predict how people with particular worldviews will explain poverty’s causes and cures.

Liberals tend to emphasize the social structures, inequalities, and discrimination that cause and entrench poverty, as well as the bad luck (discussed below), job loss, or divorce that casts many people into poverty and helps keep them there. Conservatives tend to assign more explanatory weight to moral, cultural, and familial decay—“the breakdown of social order and authority,” in poverty scholar Lawrence Mead’s phrase—and to the perverse incentives created by many existing policies and institutions. Each side denounces special interests but focuses on different ones: Liberals blame corporate rapacity and influence over government, while conservatives blame labor unions, self-aggrandizing government bureaucracies, and individuals’ bad choices. Restrictionists like the Federation for American Immigration Reform blame undocumented and even some legal immigrants. These and other conflicting explanations are hardly surprising: The data are often ambiguous about causes, and people exhibit a strong propensity to interpret it in ways that render it consistent with their preexisting social and moral frames, a process of cultural cognition discussed earlier.87

Some research has found IQ and other standardized test differences among subpopulations due to some combination of genetic and environmental causes which together could help account for poverty. These findings remain highly speculative as to both the relative contributions of each cause and the quality and significance of the underlying data. These questions, however, are anything but academic: They arouse the deepest concerns about the potentially toxic effects of such findings on our social, moral, and political arrangements. Indeed, those who have dared to present them are invariably pilloried as racists.88 Researchers today seem to believe that both environment and genes probably play some role in cognitive differences and thus in poverty, but there agreement ends.⁸⁹

The main proximate cause of poverty is under- and unemployment by working-age, non-disabled family heads. (This presumes the same hourly wage they now earn or could earn given their work-relevant credentials.) Poor families with children work far fewer hours (less than half as many) than non-poor families with children. If the poor (or their partners) worked full-time, holding all else constant, the poverty rate would drop about 40 percent. ⁹⁰

In the rest of this section, I present nine factors—we might call them more ultimate causes—whose effects on work effort and thus on poverty are fairly well substantiated by social science studies. They are (1) bad luck; (2) family and community breakdown; (3) disappearing jobs; (4) educational deficits; (5) isolation; (6) discrimination; (7) bad choices; (8) incarceration; and (9) the culture of poverty. Tragically—for it makes finding effective remedies much more difficult—they overlap considerably, so I discuss them in no particular order. The discussion contains a blizzard of empirical findings and statistics drawn from published studies, yet some key concepts cannot be defined with precision and the evidence is often weak, conflicting, dated, and context-specific. Interactions among key variables are often difficult to identify and interpret, and we cannot rule out the possibility that another, unidentified variable is doing some of the causal work. Indeed, even causality’s direction is often uncertain: We cannot always tell what is cause, what is effect, and under what conditions each is true. (One example is the relation between family breakdown and poverty, discussed just below. Another is poor people’s legal problems.⁹¹) Randomized controlled trials (RCTs), common in hard science research, are usually (not always, as we shall see) impossible to design and conduct by social scientists, but they are considered the gold standard.

These preliminary considerations help explain why poverty’s causes are complex—but they also suggest why much of what we think we know may be wrong. Those with political or ideological axes to grind contest these causes, but so do leading social scientists who use rigorous, objective data to study poverty. Their disagreements should chasten any claims of certainty about poverty’s causes.92

Bad Luck. Some people are much luckier than others. It is not simply that some of us are born with better parents, greater intelligence, happier dispositions, stronger constitutions, and in a favorable birth order.⁹³ It is also that what happens to us later on may have little or nothing to do with those initial endowments or with the kinds of bad choices that I shall discuss below. Impoverishing misfortunes can come in many forms—debilitating illness, depression, job loss, uninsured disaster—and can overtake anyone at almost any moment, rendering them poor for a considerable period of time. Divorce often impoverishes women. A 1996 study of such women found that in the first year after divorce, the average wife’s standard of living decreased by 27 percent; the husband’s improved by 10 percent!⁹⁴ Even if these differences have narrowed since then, poverty can still be just a bad divorce (or divorce lawyer) away, especially for women.*

Family and Community Breakdown. As the family is the essential core of any society, the steady decline of two-parent family households in the United States is probably the single most important social trend of the last half-century. In 1965, Daniel Patrick Moynihan famously predicted that this decline would be most precipitous in black families,⁹⁵ but even he under-estimated the trend: 72 percent of black babies are now born out of wedlock, triple the 1965 rate. Regardless of race, children with no father at home are now four to five times more likely to be poor than children of married parents.96(Many of their unmarried mothers do have live-in partners for certain periods.) Children in female-headed households account for well over half of all poor children.⁹⁷ Indeed, Harvard’s Equality of Opportunity Project finds that the single best predictor of low upward mobility in an area is the fraction of children with single parents.⁹⁸

In a tragic convergence, moreover, family decline now also blights the white working class (a hard-to-define group, often described as blue-collar workers). The rates of divorce, non-marital births,⁹⁹ and single-parent households among working-class whites now are equal to or exceed those of black families five decades ago. For instance, the percentage of white children living with a single parent in 2014 was 25 percent—greater than the 22 percent of black children in the same predicament in 1960.¹⁰⁰ And, contrary to conventional wisdom, family breakdown is worse in some red, more religiously observant states, particularly in the South, than in many blue, relatively secular ones.¹⁰¹ (The correlation is muddy, with some striking exceptions and uncertain causal patterns.)¹⁰² Strikingly, this weakening of families has coincided with enormous economic growth and much greater antipoverty spending at all governmental levels. Some other wealthy societies (e.g., Sweden, Switzerland) also have high divorce rates,¹⁰³ but the effect on poverty there is less extreme, perhaps because of more generous child subsidies.

Fortunately, the poverty-related news isn’t all bad. As noted earlier, the poverty rate declined sharply in 2015 among all demographic groups. Teenage pregnancy has declined by more than 50 percent since 1991, including in 2015. High school completion is up, with the black-white gap closing and stabilizing over the last twenty-five years at around 85 percent. The uptick in divorce rates seems to have stabilized, albeit at a high level. Crime has dropped in low-income communities as well as wealthier ones. The crime rate in 2015 was about half of what it was in 1991, with violent crime falling by 51 percent and property crime by 43 percent.

A related but distinct aspect of family and community breakdown is the phenomenon sometimes referred to as “disappearing black men.” (Before discussing this problem, it is worth nothing that most African American men are not poor, out of work, destined to spend time in prison, or unmarriageable.)104 A New York Times analysis published in April 2015 found that “more than one out of every six black men who should be between 25 and 54 years old has disappeared from daily life”; incarceration and early deaths overwhelmingly cause this gap. Higher incarceration rates account for almost 600,000 of the 1.5 million missing black men from ages 25 to 54 (demographically the prime-age years)—almost 1 in 12 black men compared with 1 in 60 nonblack men in that age group. But as we shall see, these disappearances began well before the big rise in incarceration rates. The other main cause of the disappearance is higher mortality. The Census counts about 900,000 fewer working-age black men than black women. Homicide, the leading cause of death for young black men, plays a large role, but men in this group also die more often from heart disease, respiratory disease, and accidents than other demographic groups do, including black women. There are only eighty-three non-jailed black men for every one hundred black women in this age group. (For whites, the comparable number is ninety-nine non-jailed men for every one hundred women.)¹⁰⁵

Such “disappearances” and demographic disparities distort social dynamics in these communities, entrenching emotional as well as economic poverty. Orlando Patterson, a leading scholar of black Americans, finds these patterns rooted in the culture of slavery, which severely deformed male behavior, relationships between men and women, and thus those between parents. Writing almost twenty years ago, Patterson noted that “Afro-American men are not only far behind their Euro-American counterparts but also significantly worse off than Afro-American women” with respect to life expectancy, suicide rates, crime, and education and skill levels, including in the professions and hard sciences.¹⁰⁶ (Black women, however, suffer a particularly appalling disadvantage: Their maternal mortality rate is more than triple the rate for whites and Hispanics, probably due to worse health conditions and worse care.)¹⁰⁷

These disparities corrode family relations in many ways. Black men, who are less likely to marry in the first place, experience more divorce and separation, which in turn harms their children in many ways, including future economic mobility.108 And when black men’s marriages dissolve, they are also far less likely to remarry. In fact, black women typically spend only 22 percent of their lives being married; white women are married about twice as long.¹⁰⁹ Andrew Cherlin, a prominent sociologist of marriage, describes it as “just a temporary stage of life for blacks, preceded by a lengthening period of singlehood and followed by a long period of living without a spouse.” Many unmarried mothers cohabit with men,¹¹⁰ but these relationships, Cherlin shows, tend to be relatively short.¹¹¹ Black couples are far more likely than other groups to have very different educational attainments—another potential source of marital discord—and this mismatch has even worsened over time with the increase in black college graduates. Indeed, 45 percent of black women with a college degree had a current sexual partner whose education amounted to a high school diploma or less. (Black men with college degrees also had less well-educated partners.)¹¹² Given black women’s severely constrained choices, then, we should not be surprised to learn that those with advanced degrees are about as likely to be single mothers as are white women with high school diplomas.¹¹³ (Although the definition of “marriageable men” has changed due to women’s growing economic independence, black women still face the greatest shortage of such men.)¹¹⁴

Patterson shows that educated, prosperous black men are no more likely to marry than their poorer counterparts; indeed, they are less likely than poor blacks to have ever married.¹¹⁵ (But foreign-born blacks are different; 52 percent are married, compared with only 28 percent for U.S.-born blacks.)¹¹⁶ Contributing to this reluctance to marry, Patterson finds, are very different expectations about marriage, fidelity, sexual morality, and sexual practices between black men and women. For example, 44 percent of black husbands say they have been unfaithful to their wives, far higher than the rates for white and Latino men and for all groups of women. Black families experience much higher levels of child neglect and abuse than other families, and the quality of conjugal relationships is worse in other ways.* Nor are these dysfunctional family and cultural patterns outliers: According to Patterson’s estimates, about 35 percent of blacks experience them.

This distorted marital bargain encourages the abandonment of wives, sexual partners, and children. Not surprisingly, the effects on children are devastating. The Fragile Families and Child Wellbeing Study, a decades-long research project run by Princeton University and Columbia University, studies five thousand children born to unmarried parents in large U.S. cities between 1998 and 2000. Although most unmarried fathers express good intentions at and shortly after the time of birth, only 35 percent of these couples are still together five years later and less than half of these are married. Fathers’ involvement with the children declines over time; by age five, only half the fathers have seen their child in the previous month. By age five, 30 percent of the children have had two or more “father figures” in their home.¹¹⁷ Growing up in such chaotic family settings is bound to affect the school performance of many of these children. This may help to explain why, after controlling for the usual socioeconomic variables, Harvard economist Roland Fryer found that a black-white gap in performance, non-existent in kindergarten, appears by the third grade, when blacks are 20 percent less likely than white children to do multiplication and other standard functions. He has also confirmed the importance of peer group (“acting white”) effects: The popularity of black and Hispanic pupils drops as their GPAs rise; for whites, it is the opposite.¹¹⁸

All of these factors—economic and physical “disappearances” of black men, low marriage rates, family and custodial chaos—contribute to the persistence of disproportionate black child poverty, even as a recovering economy lowers child poverty rates. Among female-headed families generally, the child poverty rate was 45.8 percent in 2013 compared with 55.4 percent in 1991, before welfare reform.119 A Pew analysis of Census data finds that while the child poverty rate for other ethnic groups has declined since 2010, the rate for black children has remained high and disproportionate at 38 percent. Indeed, the number of them in poverty (4.2 million in 2013) may now be larger than for white children—even though there are three times as many white children as black children.¹²⁰ Robert Rector of the Heritage Foundation, among others, has probed how marriage affects child poverty, especially among blacks. Analyzing how the child poverty level would have changed if marriage rates among parents had simply continued at the 1960s rates, Rector pairs single mothers with the single men of similar age, race, and education whom they might have married (but also might have divorced at some point). Had they done so, he finds, child poverty in these hypothetical families would have plummeted 82 percent, from 37.1 percent to 6.8 percent. Since single mothers tend to be less educated than married ones, he controls for their education level and finds that marriage would still lower the poverty rate 75 percent—and even more for black children. (Births to teens are not a large part of this problem: Only 7.7 percent of unmarried mothers are younger than age 18.)¹²¹

Most of the other drivers of poverty that I shall discuss here correlate strongly with, and are likely caused by, children growing up in single-parent, fatherless homes, and the correlation is larger here than in most other countries.¹²² Indeed, Raj Chetty’s research on locational factors, discussed below, confirms other studies’ findings: Growing up in such families is the single best predictor of later poverty.¹²³

Disappearing Jobs. Jobs, particularly full-time ones, are the single most important route out of poverty. Only 2.9 percent of full-time workers were poor in 2014, compared with 14.3 percent working part-time, and 24.2 percent of those who did not work at all during the year.¹²⁴ Yet even with a very low unemployment rate (5.1 percent in September 2015), the labor force participation rate in the United States, at 62.4 percent, was the lowest in almost forty years. Our aging demographics can explain only some of this non-work. In 2000 and 2007, years when the unemployment rate was also this low, prime working-age people were more likely to work. Wage growth has also been sluggish.125 Even in a tight labor market, it seems harder for poorly educated people to find steady work and family-sustaining wages.

The sociologist William Julius Wilson is a leading proponent of the idea that social ills and unemployment among the poor (especially blacks) is the result of “deindustrialization”—the disappearance of traditionally well-paid, often unionized blue-collar manufacturing jobs caused by permanent structural changes in the economy since the 1970s. Indeed, Wilson suggests that the disappearance of such jobs and the “disappearance” of so many black men from family and community life and into prisons are intimately connected. Particularly affected were rust belt cities. Long-term unemployment has increased since the 1970s: In 1979, 8.6 percent of the unemployed had been out of work for more than six months; in August 2015, this figure was 26.1 percent despite a much larger economy.¹²⁶

In Wilson’s book, The Declining Significance of Race: Blacks and Changing American Institutions, and in more recent work,¹²⁷ he finds that the main reason for joblessness is not racial discrimination (although it still exists) but the mismatch between employers’ needs for workers with skills (now including teamwork, flexibility, and other social skills that computers lack)¹²⁸ and proximity (low-income workers now live even farther from metropolitan-area jobs than in the past).¹²⁹

Wilson’s diagnosis, however, is challenged (or augmented) from several directions. Some analysts emphasize structural social practices—family wealth, social and neighborhood networks, and institutional selection—that “lock in white advantage” even without racial bias.¹³⁰ Law professor Joseph Fishkin finds the greatest barriers to equal opportunity (“bottlenecks”) in assumed but discredited notions of merit, talent, luck, and human development—notions he claims compound one another, entrenching and institutionalizing existing advantages of the more privileged.131 Decades ago, sociologist Roger Waldinger pointed out that although many low-skill and unionized jobs have indeed disappeared, many more (non-unionized) jobs have been created, noting that black immigrants and other minorities have still managed to find work, with those who can’t find work presumably return home for not working.¹³² As discussed later in the chapter, a surprisingly small percentage of working-age unemployed men cite lack of jobs as the reason. Immigrants’ higher employment might reflect greater docility, wariness of unions, acceptance of poor work conditions, or better social networks (discussed below). But it could also reflect deeper cultural differences. Indeed, sociologists find that fellow workers including immigrant blacks also criticize many U.S.-born blacks for their job habits.¹³³

Patterson insists that the community and cultural breakdown and behavioral pathologies discussed above—the lack of a stable, co-parenting family environment for children—explain black male workers’ marginality better than Wilson’s factors.¹³⁴ As we saw, many low-skill mothers who were on welfare before TANF pushed (and the EITC helped pull) them into the workforce have found jobs even through multiple recessions.¹³⁵ Gender also complicates the unemployment analysis: Growing up poor predicts adult joblessness more for men than for women.¹³⁶ Finally, Mead argues that the decline in well-paid jobs for the low-skilled matters less than their lack of self-discipline, which impedes their getting and holding the available low-paid jobs, and explains employers’ doubts about hiring them.¹³⁷

Educational Deficits. Years of formal education and skills training (“human capital”) strongly predict wealth and income. In 2014, almost 30 percent of people aged 25 and older without a high school degree, but only 10 percent of those with some college but no degree, were poor.¹³⁸ Although high school graduation rates have risen, reaching a record 81 percent in 2013,* more than a third of those graduates do not get further education.139 Many of them remain poor, confirming that today’s economy demands more education and training to lift people out of poverty, that schools’ standards have fallen, or both.

The average wage premium for having a college degree rather than only a high school degree is huge—74 percent and slowly rising, nearly the highest in the OECD. Yet educational mobility in the United States is low: Only 30 percent of Americans today have achieved more education than their parents did, one of the OECD’s lowest levels. (This low educational mobility between generations, moreover, sharply reverses past trends when the United States led the world in graduation rates and mobility.)¹⁴⁰ Moreover, global competition and technological advance increasingly require job competencies that neither high schools nor most college programs provide. Since 2010, the ManPower Group’s annual survey of hiring managers has found that the hardest segment of the workforce to fill is not web developers, engineers, or nurses, but rather skilled trade workers such as electricians and machinists.¹⁴¹ On both the national and state levels, today’s skilled trade workers are typically 45 years of age or older, which suggests an even larger impending skills gap as baby boomers retire.¹⁴² Even if business owners’ confidence in the economy strengthens, their inability to locate qualified candidates may leave many good jobs unfulfilled.¹⁴³ Much of the public blames these educational deficits on inadequate schools, but the deficits exist before U.S. children even begin school. International comparisons support the view that poverty is the main factor¹⁴⁴—although defining and then controlling for poverty makes such comparisons difficult.

Isolation. Poor people are isolated in a number of ways that can hold them in poverty.* The digital divide deprives them of information and connections of all sorts.¹⁴⁵ A related but even more important form of isolation is geographic: They tend to live in areas of low economic growth, modest wage levels, and high unemployment. (As one expert puts it, “those with get up and go have already gotten up and gone.”) Recent research by economists led by Raj Chetty has confirmed what an earlier experiment, Moving to Opportunity (MTO), had seemed to deny—that the quality of one’s neighborhood (controlling for other factors) independently and significantly affects one’s economic prospects. The MTO study had found that lower-class families given vouchers enabling them to move to better neighborhoods did no better in most respects (including employment) and in some ways did worse after the move (particularly young men). But when Chetty’s group controlled for the age at which they moved to the new neighborhoods, and hence the amount of time they had lived there and gained its advantages, the results were much more encouraging—although they took many years to materialize. As one commentator put it, “the net present value of the extra earnings that will eventually accrue to a child who moved at age 8 is $99,000, meaning that for a family with two children, the program yields $198,000 in extra earnings.”146 Reinforcing and extending this “neighborhood matters a lot” finding is another study by the group that tracked five million movers over seventeen years. It showed that the earlier a family moved to a better neighborhood, the better the children’s long-term outcome (up to age twenty-three).¹⁴⁷* Conversely, researchers also find that the negative effects of staying in bad areas persist beyond the children’s generation.148 And a new study, which controls for the likelihood that those who entered the MTO voucher lottery were already more highly motivated parents, finds that the observed positive effects of moving are actually understated.¹⁴⁹

The social isolation of many poor people is not limited to geography, and it severely restricts their mobility and opportunity. Consider a recent summary of the research findings:

Friends and relatives can lend money, pool risk, mind children and bring news of job openings. Researchers from the London School of Economics found that when a group of Bangladeshi women were given business training and free livestock, not only did they move up the income ladder, but their friends’ lot improved too. A year later the friends’ consumption had risen by almost 20%, and they claimed to have become savvier about business as well. The downside is that not having the right friends can cement hardship. The more concentrated the poverty, the less helpful social networks tend to be. In Atlanta, living in a poor neighborhood decreases the chance of having a friend with a job by almost 60%, and [the chances] of having a friend who had been to university by over a third. A global survey conducted in 2014 by Gallup … found that 30% of people in the poorest fifth of their country’s population had nobody to rely on in times of need, compared to 16% of the richest fifth.… [O]ne American study found that in poor neighborhoods, three-quarters of jobholders found work through friends.¹⁵⁰

Upward mobility, then, depends on a complex social process consisting of relationships, networks, and institutions that can help one gain human capital by earning trust, acquiring experience, and then using it to advantage. One’s “embeddedness” is measured by “density,” the extent to which individuals in one’s group of contacts also know each other. Network sociologist Mark Granovetter has shown how this density can be both an advantage and a disadvantage, a conclusion confirmed by an ethnographic study.¹⁵¹ A very dense network, he shows, may provide emotional support and friendship, but can make it harder to find a job or other kinds of opportunity because this very density tends to limit access to information from outside the group. In a paradox he calls “the strength of weak ties,” people seeking to rise may benefit more (at least economically) from mere acquaintances than from friends because the former can pass on information from other networks of which they are members.152 Granovetter also shows that information generated by weak ties is often crucial to finding a job.¹⁵³

Orlando Patterson’s work on social networks provides additional, and dismaying, insight into the role of social isolation in perpetuating poverty among blacks—even as a large black middle class has emerged against great odds.¹⁵⁴ In work published in 1998, Patterson took on what he called “the Myth of the ’Hood.” He found that blacks’ social networks were smaller, denser, and included remarkably few kinsmen; indeed, almost half of them had no kinsmen in their networks. In a “truly startling finding,” no relationship whatsoever existed between network density and education level; Patterson inferred that more education did not expand their ties much. Unlike other groups for whom employed people had much less dense ties than retired people, blacks’ networks were about as dense among retirees as among those still working. In other groups, lower-class people had much denser networks than higher-class ones. Among blacks, network density had little relationship to class. And while marriage for other groups was strongly related to the proportion of kinsmen in other groups’ core networks, black marriages had little or no effect on network composition. Indeed, blacks had many fewer kinsmen in their networks than any other group; 91 percent reported none in their closest circle of contacts.

These patterns, Patterson concluded, put blacks in the “worst possible position. Their networks are smaller than those of nearly all other groups, and they are denser and hence their range of contacts narrower than those of other Americans. What is more, they do not enjoy the main benefits that usually come with dense networks—the trust, security, and support of kinsmen. For, contrary to conventional wisdom and ideology, their networks have the smallest proportion of kinsmen of all native-born Americans.”155 As if this were not isolation enough, Patterson adds another dimension—marital and familial—that reinforces the opportunity-limiting density of black networks: The odds that a black woman will marry a black man are 27,444 times greater than that a non-black woman will marry a black man!¹⁵⁶

Discrimination. Being an ethno-racial minority correlates strongly with being poor. Discrimination against minorities and poor people surely contributes to their poverty. But the meaning of discrimination depends on context. When we condemn discrimination, we usually mean treating people differently for reasons that should be socially and morally irrelevant, that can harm them in unacceptable ways, and that should thus be presumptively illegal. For clarity, I call this “invidious” discrimination.

What actions constitute invidious discrimination? Conceptually, I distinguish three types of discrimination. In intentional discrimination, the decision maker (say, an employer) treats A differently than B because of a characteristic of A’s that the law (or society, usually both) holds should be irrelevant to the decision. Antidiscrimination law terms this invidious discrimination “disparate treatment.” It identifies a number of such characteristics—race, religion, national origin, disability, illegitimacy, gender, age, (increasingly) sexual preference, and others—and protects them from such discriminatory decisions in employment and some other settings. Depending on the characteristic at issue, the law protects it by subjecting such decisions to different levels of judicial review, with the highest level (“strict scrutiny”) applying to race and the lowest (“rationality”) applying to age. Intentional discrimination based on these characteristics is difficult to prove; people are loath to recognize it in themselves, much less admit it, in a society that professes egalitarian ideals.

In statistical discrimination, decision makers treat A and B differently because they predict, based on what they think is a statistically justified generalization about the racial (or other) groups to which A and B belong, that they will behave or perform differently. The accuracy of this generalization depends on the quality of the information on which it is based and how many exceptions to the generalization there are. Condemnation of such generalizations as “stereotypes” raises three important questions: How accurate is it as a general proposition; how accurate is it as applied to the individual in question (e.g., a job applicant); and how legitimate is using such a stereotype? Here, too, context is crucial. Stereotypes are inescapable and valuable because they economize on information in the countless daily situations when we need to predict how one will behave but cannot easily find out whether one fits the relevant stereotype or is an exception to it. But because stereotypes are sometimes false and unjustly harm individuals in certain situations, the law weighs various factors to decide when they may and may not be used.157

Unintentional discrimination occurs when government and private actors act in ways that are facially neutral and not motivated by bias. But in a society as unequal and diverse as ours, innocent decisions and actions that fall most heavily on protected minorities may reflect and reinforce existing inequalities. Antidiscrimination law terms this “disparate impact,” punishing it even though the actor lacked invidious intent. Much rides on this distinction between treatment and impact. Title VII of the 1964 Civil Rights Act, which bars employment discrimination, punishes both, and the U.S. Supreme Court, after decades of uncertainty, recently held that Title VIII’s ban on discrimination in housing also punishes both.¹⁵⁸ But although the distinction is crucial in most other legal areas, the Court has left it quite muddy.¹⁵⁹

For present purposes, the key point is this: Poverty is not a protected characteristic in any of these three discrimination situations. Americans (and, I suspect, all other societies) feel great sympathy toward the poor but view exclusion on the basis of poverty (“classism”) as normatively and legally acceptable in most contexts (exceptions to this include imposing a fee on divorce petitioners unable to pay,¹⁶⁰ and poll taxes¹⁶¹), while viewing racial discrimination of any kind as normatively and legally unacceptable, even in many private realms.¹⁶²

This difference also makes it harder to prove racial bias legally. Neighborhoods, for example, are almost always class-identified. The non-poor regard this class differentiation in neighborhoods (strongly reinforced by public housing location policies) as desirable and natural, and evidence from the MTO study shows that many poor people agree with them. Because race and income are so highly correlated, classists may behave much as racists do, seeking to avoid the same people. These same decisions may have been motivated by classism rather than racism. Racists can easily invoke a classist rationale for their decisions, thus escaping both the moral censure and the illegality that racism would arouse. Another complication arising from this overlap is the problem of asymmetric information between whites who already live there and minorities who want to do so. When government subsidizes low-income people to move into middle-class communities, both classists and racists may assume that the newcomers are lower-class unless they can somehow signal a higher status and residents can easily ascertain the truth of the matter.

Whatever the precise mix of causes for class and racial isolation in neighborhoods, it surely contributes to the concentration of poverty. This concentration has grown significantly since the Great Recession, with more people living in “extremely poor neighborhoods (i.e., census tracts where 40 percent or more of the population lives below the federal poverty line).”163 So entrenched is this isolation that typical middle-income blacks (with some geographic variations) actually live in lower-income neighborhoods than poorer whites do!¹⁶⁴ As I discuss further below, such neighborhood effects practically ensure the projection of inequalities into education, employment, culture, personal networks, freedom from crime, and the many other opportunities, amenities, and freedoms that are so closely related to location. (Locational effects on future mobility are smaller for better-off children.) All these inequalities can perpetuate poverty. But again, conflating prohibited racism and normatively approved classism makes these issues harder to disentangle and thus to resolve.

The relationship between discrimination and poverty is further complicated by the experience of immigrant groups, some of which have been resilient enough to weather even the harshest forms of discrimination. Thomas Sowell, an economist who has written widely about the relative progress of various groups, reports that although the interment of the Japanese in the United States reduced the internees’ income in the immediate postwar years, they achieved economic parity with Americans by 1959; in Canada, where the internments were even more damaging economically, their rebound was “spectacular.” In 1959, only sixteen years after the repeal of the virulently racist Chinese Exclusion Act, Chinese family income in the United States (some of which reflected previously affluent refugees from the Communist takeover) approximated the national average.165 Sowell documents many other examples of groups—Jews and Indians, notably—succeeding here and in other destinations where they faced severe discrimination.¹⁶⁶

The remarkable advances of so many immigrant groups in so many unwelcoming (to say the least) societies complicates any simple explanation of poverty. Could it be that anti-immigrant animus, which so cruelly and perversely disadvantaged these groups in the short run, actually strengthened their economic and other social positions in the long run? In some paradoxical way, might hostility and discrimination promote the survival and prosperity of the very group (if not always of each member) that the destination societies hope to exclude? These are intriguing questions suggested, but certainly not answered, by this cross-cultural evidence. To ask them, of course, is certainly not to countenance discrimination against the poor, which inflicts undeserved suffering on many innocent, hardworking people. But such questions do suggest that discrimination cannot fully explain persistent poverty.

Indeed, it is impossible to know how much of today’s poverty is due to racial discrimination. Black poverty is “over-determined,” a product of many different, powerful causes whose separate contributions cannot be disentangled. (Ironically, as Wilson points out, the same decline in discrimination that enabled many blacks to move to middle-class neighborhoods and jobs also deprived those left behind in the ghettoes of their salutary influence.) Further confounding our understanding are public attitudes about race relations. On the one hand, Americans—surely influenced by intense coverage of white-on-black violence and Ferguson-type conflagrations—believe that race relations are worsening despite much evidence (not just the election and reelection of a black president) that racism has declined dramatically, that a majority of blacks think there has been a lot of progress, and that two out of three blacks (in a CBS/New York Times poll conducted after these violent events) consider race relations in their own communities generally good.167 Perhaps this is akin to the frequent finding that the public has a very low opinion of Congress but a much higher opinion of their own representative, especially if they know the representative’s name. (This gap has declined in recent years as disaffection with Congress has reached record levels.)¹⁶⁸

Bad Choices. Many people are poor for a reason that they sometimes acknowledge to themselves and others: They have made bad choices in the past that continue to harm them. Those who make this point are often accused of “blaming the victim,” lacking compassion, and ignoring structural causes of behavior. But any analysis that fails to take the role of bad choices seriously misses an important clue to what causes some poverty and what might reduce it.¹⁶⁹ By “bad choices,” I mean decisions that squander or foreclose opportunities to better one’s socioeconomic prospects in the future. Some bad decisions, such as narcotics or alcohol abuse, are self-destructive. Many are not only antisocial but illegal; if detected and punished, they will foreclose many paths out of poverty. Most bad choices, however, are simply shortsighted; they sacrifice the real possibility of future, durable gains for more immediate but transitory ones. Common examples include chronic truancy, dropping out of high school, alcohol abuse, gambling, habitual indiscipline, ignoring schoolwork in favor of TV or video games, excessive borrowing and spending, domestic violence, parenting children whom one cannot afford or do not know how to care for, cultivating self-destructive habits and hanging out with antisocial people; and quitting jobs or training programs from which one could acquire useful skills or experience.

People are drawn to bad choices for various reasons. It may be difficult to envision alternative courses of action. Alternatives may be imaginable but seem too distant, hard, or downright unattractive. One may hope to avoid hard choices, thinking that one can enjoy short-term pleasures without actually sacrificing longer-term opportunities. One may make bad choices because one is too present-oriented, discounting future possibilities very heavily.* Their deeper causes—impulsivity, low intelligence, limited imagination, lack of empathy, habituation, and emotional or neurological conditions—surely apply differently to different people and in different combinations. The stresses of living with severe scarcity, according to recent research, can affect brain function in ways that lead to bad choices.¹⁷⁰

Incarceration. A June 2015 report by the Economist magazine presents the now-familiar but disturbing contours of this problem:

No country in the world imprisons as many people as America does, or for so long. Across the array of state and federal prisons, local jails and immigration detention centres, some 2.3m people are locked up at any one time. America, with less than 5% of the world’s population, accounts for around 25% of the world’s prisoners. The system is particularly punishing towards black people and Hispanics, who are imprisoned at six times and twice the rates of whites respectively. A third of young black men can expect to be incarcerated at some point in their lives. The system is riddled with drugs, abuse and violence.¹⁷¹

The report goes on to note both the recent reductions in incarceration and the severe constraints on achieving further declines. The popular nostrum of releasing or diverting those convicted of minor offenses or those above a certain age, while desirable, will actually have little impact on incarceration.¹⁷² Those imprisoned for mere drug possession, the low-hanging fruit for de-incarceration efforts, now account for about 3 percent of state prisoners, who constitute 87 percent of U.S. prisoners. And even the imprisoned drug offenders were usually plea-bargained down from more serious trafficking offenses. People whose only offense is using drugs almost never go to prison and rarely spend much time there.173 The core problem is far more daunting: Almost half of prisoners are violent offenders; another 12 percent are sex offenders. An Urban Institute study confirms that even limiting incarceration to these serious criminals would hardly make a dent in the problem.¹⁷⁴ U.S. incarceration rates are much higher than in Europe partly because our incidence of violent crime and recidivism is so much higher.

Prisoners, especially for long periods, rupture their ties to legitimate sources of upward mobility, perhaps permanently. Their access to good jobs is severely limited—hence the “ban the box” movement to limit employers’ information about applicants’ criminal records,* and President Obama’s proposals to ease their reentry.¹⁷⁵ Convicts’ family and communal relationships are severed or strained, their friendships attenuated, their trustworthiness cast in doubt, their skills atrophied, and their emotional lives blighted. Recidivism and a continuing life of crime are strong possibilities—and magnify and extend these poverty-inducing effects. Indeed, more than half of those released from prison are arrested for a new crime within three years of release. Even if they were not poor before imprisonment, they are likely to be poor thereafter. Still, the timing and trajectories of family dissolution and poverty suggest that incarceration is not a leading driver of poverty. Even when black non-marital births soared in the 1960s and early ’70s, black incarceration remained quite stable—at about one-tenth of today’s rate! Much of the damage to black families and communities, then, had already been done before black incarceration rates began their rapid rise.¹⁷⁶ Still, the historical relationship between incarceration and poverty remains uncertain. For present purposes, then, the crucial points are that (1) that incarceration is a cause of poverty, not just a consequence of it, but (2) it is not a principal cause, and (3) release of minor offenders would not affect poverty much.

Culture of Poverty. I have been discussing factors that are known to contribute to poverty. Some of them, like bad luck and invidious discrimination, are ultimate causes in the sense that if we could somehow eliminate them, we could end the amount of poverty attributable to those particular factors. But, welcome as that would be, ending such discrimination would have a relatively minor impact on poverty because it contributes far less to it than the other causal factors I have discussed: family and community breakdown, disappearing jobs, educational deficits, isolation, bad choices, and mass incarceration. Those factors are themselves symptoms or manifestations of the more fundamental (“root”) causes—both social and individual—that drive them.*

Many poverty researchers maintain that what causes, compounds, and perpetuates most or all of the factors discussed above (except invidious discrimination) is an underlying “culture of poverty.”† Culture (of any kind) is notoriously difficult to define, but in this case it usually denotes widespread, entrenched, self-reinforcing patterns of despair, present-orientation, inability to grasp what limited opportunity there is, and self-destructive behavior. The Moynihan Report imbroglio made this phrase highly controversial in the United States, yet the post-1965 proliferation of cultural studies programs in leading universities signaled its growing legitimacy, while more systematic studies of poor communities provided empirical support. Even so, antipoverty activists often resist this because they see it as “blaming the victim,” and an excuse for society to throw up its hands and do nothing. Although they are wrong to dismiss culture, its conceptual blurriness can breed misunderstanding.177 Culture does resist change and usually changes slowly, although examples of fairly rapid change do exist (as with attitudes toward same-sex marriage). And it is shaped by even more stable factors: geography and climate, natural resources, entrenched institutions, history, and others.¹⁷⁸ Interestingly, however, opinion surveys indicate that poor and non-poor Americans view poverty, welfare, and future progress in much the same way, views that have not changed much in the last thirty years.¹⁷⁹

Culture’s salience in explaining much of what goes on in poor communities is widely accepted today. As William Julius Wilson puts it, “it’s more than just race.”¹⁸⁰ The culture of poverty posits a milieu in which the causes of poverty and dependency discussed above are widespread and entrenched. In particular, the bad choices described above have a kind of coherence to participants in this culture; in that sense, these perverse choices seem rational. These choices, reinforced and validated by other aspects of the culture, create a “poverty trap” from which escape is especially difficult.¹⁸¹ Changing the culture and removing the trap pose the single greatest challenge to government, policy makers, and to the poor themselves.

CURRENT ANTIPOVERTY PROGRAMS

According to the Congressional Research Service, the federal government spent $848 billion in fiscal 2015 on programs with an explicit focus on low-income people or communities, a very large increase over the fiscal 2008 (just before the Great Recession) level of $561 billion. (These numbers exclude three spending program categories: (1) social insurance programs such as Social Security, Medicare, and Unemployment Insurance; (2) tax provisions other than the refundable portions of the EITC and the Child Tax Credit; and (3) state and local spending on such programs.)182 The common assertion that the federal government’s social welfare spending lags far behind that in OECD welfare states is false. It ignores the fact that a far greater share of that spending occurs through off-budget, tax-expenditure modalities—what political scientist Christopher Howard calls the “hidden welfare state.”¹⁸³ Taking this into account, a recent comparison of social welfare spending in the OECD states as a share of their GDPs shows that “only the French spend more than Americans, while the alleged welfare-addicted Scandinavians and Europeans spend less on average.” Our approach has some advantages but seems to produce worse outcomes in alleviating poverty.¹⁸⁴

For this reason, Christopher Jencks’s exhortation “to come clean about which parts of the War on Poverty worked and which ones do not appear to have worked, and stop supporting the parts that appear ineffective” is particularly wise and welcome.¹⁸⁵ But as Jencks well knows, this is easier said than done; rigorous studies are hard to come by.¹⁸⁶ In the rest of this section, I marshal the best evidence on how effective antipoverty programs are. To set the stage, I must clarify some terms, distinctions, techniques, and constraints.

Terminology, Distinctions, Techniques, and Constraints. Broadly speaking, antipoverty programs are financed in three ways. Entitlement programs are those in which a statute confers eligibility on various groups of people based on specific criteria. They may be universal or means-tested. People who satisfy the statutory criteria are entitled to claim the benefits specified in the law, and Congress is morally if not legally obliged to appropriate the funds necessary to satisfy those claims. It always does so. Social Security and Medicare, the two largest, are universal programs in the sense that they cover virtually everyone above a specified age. These two programs contribute enormously to poverty prevention and alleviation, but as noted earlier, they are not counted as antipoverty programs. Indeed, their political popularity depends on the widespread belief that their beneficiaries have financed their own benefits by contributing to a centrally administered “trust fund,” a self-financed social insurance fund. In fact, this is untrue. Most of those who die before retirement age get far less than they paid into the system through their payroll taxes, while today’s beneficiaries receive far more in benefits than they paid in.187 Other important entitlements include unemployment compensation (though many unemployed do not qualify for it) and veteran’s benefits. The most important means-tested ones are Medicaid, a joint state-federal program providing health care to the poor; the Food Stamp program (now Supplemental Nutrition Assistance Program, or SNAP); Supplemental Security Income (SSI), which provides income support for the aged, blind, and disabled; the refundable portions of the EITC and Child Tax Credit; and the low-income subsidy for Medicare Part D.

Discretionary programs constitute the largest group of antipoverty programs, including almost all of those to be discussed below. Congress must reauthorize them periodically, and their funding requires periodic (often annual) appropriations. These programs are means-tested. Section 8 housing vouchers ($17.9 billion in 2013) and Head Start ($8 billion) are the largest examples—and enjoy more political support than other such programs.

Some tax programs are designed to reduce poverty. Authorized in the Internal Revenue Code, they are created by two powerful congressional committees (House Ways and Means; Senate Finance). They do not need appropriations and once enacted are more or less permanent. Most of them are targeted at either businesses (e.g., depreciation allowances) or middle- and upper-class people (e.g., deductions for medical insurance and home mortgage interest), but some target the poor. The EITC is means-tested and targets parent-earners and some noncustodial workers (Congress is considering expanding eligibility for this second group) on a sliding income scale; in 2014, it phased out at incomes around $52,000 for a married couple with three or more children. The Child Tax Credit is also means-tested but does not begin to phase out until the income level reaches $110,000, so relatively little of the benefit ($21.6 billion in 2013) goes to the poor. (Most of them are exempt from federal income taxes anyway but must still pay federal payroll tax, which takes a large bite out of any on-the-books wages they earn.)

Some other programmatic distinctions merit brief mention. One is between programs that affirmatively confer benefits on the poor (the traditional form of antipoverty programs) and policies that instead incentivize self-help by the poor—for example, by exempting low-wage workers from requirements. Another distinction concerns the specific form that benefits take. Some antipoverty programs provide cash. The main examples are the EITC and TANF (which replaced traditional welfare in 1996). Others provide “near-cash” for the purchase of approved categories of goods. The major examples here are SNAP and Section 8 housing vouchers.

Few other antipoverty programs give benefits directly to the poor. The vast majority of programs instead pay middle- and upper-class intermediaries, many of them professionals, to provide their services to the poor. Providing services rather than cash (or near-cash, as with SNAP)—what Moynihan mocked as “feeding the horses to feed the sparrows”—is by far the dominant approach to our way of fighting poverty—and a dubious one at that. Indeed, “of the $800 billion spent on poverty programs in 2012, less than $150 billion was distributed in cash income, if one includes as cash benefit the tax rebate under the EITC. That is a grand total of 18 percent of the whole. The rest was … money paid to providers of various kinds, most of whom have incomes well above the poverty line.”188

Antipoverty programs also differ in their relationships to various levels of government and the private sector. Federal programs are almost never operated solely by Washington. (Social Security is a rare exception.) Instead, the federal government usually funnels money to states (and through the states sometimes to localities), which then operate the programs under regulations issued by the federal and state governments. Implementation of federal programs is almost entirely a state and local responsibility, a fact that usually reshapes programs in important ways not always envisioned by Washington.¹⁸⁹ States also have their own antipoverty programs in policy areas where they traditionally bear primary responsibility: public health, labor standards (including minimum wage regulation), housing, and so forth.

A final distinction concerns the identity of the beneficiary groups. As noted earlier, and despite the effusive rhetoric about children being our future, programs for the elderly are much better funded than those targeted at poor families with children. Those favoring the working poor are much better funded than those for the nonworking poor. (In this context, “working” usually includes those who are in job training or educational programs.) Notably, the 2016 presidential campaign paid little attention to the nonworking poor.190

Studying the effectiveness of a policy is replete with methodological difficulties.¹⁹¹ What does it mean to say that a policy is effective or ineffective? In a previous book, I devoted a long chapter to analyzing this very question, which of course is essential to any such discussion.¹⁹² Readers who want the more complex, nuanced, and qualified analysis will find it there. Here, I present a much shorter and conclusory summary of that analysis.

All policies succeed in some respects (after all, the money benefits someone) and fail in others. Program advocates’ promises may bear little resemblance to actual performance. Mere common sense or inferences drawn from a program’s intentions or political support are poor guides to program evaluation; only systematic analysis and evidence can do that. A policy may fail for many reasons—a flawed theory or design, costs that exceed benefits, muddled implementation, erratic enforcement, or political compromises that undermine its efficacy either at its inception or as it plays out in the real world. Another common reason for program failure is unforeseen behavioral responses to its incentives—as when many people with private health insurance dropped it once Medicaid expansion make them eligible, or when public pensions reduce private retirement savings. Moral hazard—greater risk-taking when a program insures against loss—is a common effect that increased the huge mortgage losses that helped cause the Great Recession.¹⁹³

Public policies should not be held to impossible standards. Instead, they should be assessed according to four obvious and minimal criteria, as well as some other factors.

•  Its benefits should exceed its costs. The standard tool for determining this is cost-benefit analysis. CBA does not demand that policies be perfect, only that they be more beneficial on net than alternative policies, including the status quo. This is simple rationality, which is why every president from Ronald Reagan to Barack Obama has mandated its use to analyze proposed government regulations. Some critics oppose the very idea of using CBA to analyze programs designed to protect values such as human life and health, biological diversity, and environmental purity that are difficult to measure, much less price; some argue that the very efforts to do so are immoral. Some criticisms of CBA are more technical or methodological in nature; many of these do not object in principle to using CBA but rather to particular applications of it. Such disagreement is desirable and will improve the method. In the end, CBA—when conscientiously conducted and attentive to its own potential flaws—is probably the best that we can do to assess the efficiency merits of policies. What other decision modes are better? Good intentions, intuition, caprice, political power, abstract philosophical theories, or coin flips are not acceptable ways to make sound decisions. As environmental economist Bjorn Lonborg has put it, CBA “is a for more effective and moral approach than basing decisions on the media’s roving gaze or the loudness of competing interest groups.” Adapting Winston Churchill’s quip about democracy, CBA is the worst decision tool except for all the others that have been proposed so far. A proper CBA should at least shift the burden of proof to those who would challenge its verdict on effectiveness.

•  The cost-effectiveness criterion (or the “most bang for the buck” principle) means that whatever a program’s purposes are and whatever level of benefits it actually produces, those purposes and this level should be attained at the lowest possible cost. The flip side of this principle is that at any given level of cost, the program should produce as many benefits as possible. One reason that a program may not be cost-effective is that it uses the wrong tool. The government has many in its tool shed: subsidies, taxation, regulation, tort law, market incentives, competition, and many others. The best choice for a given policy goal will usually depend on a number of factors, so answering this question can be difficult. For example, if the policy seeks to raise the income of the working poor, almost all economists believe that the EITC is ordinarily a far more effective tool for doing that than raising the minimum wage. I discuss this below.

•  Target efficiency means that a program should target its benefits on those who will benefit most from the policy.194* Although this is just common sense, policy makers often fail this test, even when they want to reduce poverty. They may lack information about key facts, such as how much program resources will produce desired policy outcomes. Political factors often sacrifice targeting to other legitimate goals such as greater universality (e.g., Social Security retirement) or a desire to keep higher-income people in the program mix (e.g., public housing in New York City where more than 10,000 over-income families remain despite a waiting list of more than 300,000 income-eligible ones).¹⁹⁵

Disputed normative judgments may also compromise targeting. We do not always want antipoverty programs to target the most disadvantaged of the eligibles. We may predict that the neediest people will not use the benefits wisely or effectively, or that they can be independent without the government’s help, or are needy only because of some sort of self-destructiveness or other personal failing. The once-colloquial, morally loaded terms “deserving” and “undeserving” poor are now derided in much public discourse. President Obama, visiting a federal prison in July 2015, observed, “There but for the grace of God [go I]” but also strongly condemned violent criminals.196 He surely recognized that this distinction was fundamental to Americans’ thinking about which low-income groups most merit society’s solicitude. Indeed, those who live in close proximity to the culture of poverty (and perhaps narrowly escaped from it) probably draw this distinction more easily than the rest of us. But however one comes out on this moral debate, the key point about target efficiency is this: Many of our costly programs fail its test. The huge federal student loan program whose benefits flow largely to the better off is but one example;¹⁹⁷ indeed, Hillary Clinton’s proposed fix to the program may exacerbate this.¹⁹⁸

•  “Invisible” or “silent” victims. A program may make people worse off who are unaware of this fact. Policy makers will not attend to the interests of these cost-bearers unless they make a special effort to anticipate and assess the program’s impacts on them. These costs ordinarily gain little recognition because they are small at the individual level and because the causal link between the policy and these costs is difficult to trace, even or especially by its victims; they often cannot organize politically to protect their interests even though the benefits of doing so may be quite large when aggregated across all of them. They include, for example, those who will not get a job opportunity or benefit, or retain one, because a new regulation, higher minimum wage, or litigation-averse employer chills job creation, causes layoffs, or affects firms’ locations.¹⁹⁹ Many poor people do not receive their Medicaid benefits because its rigid, inadequate reimbursement rates induce doctors not to participate. And many restaurants respond to minimum wage increases by prohibiting tips for servers.²⁰⁰

Inherent, Severe Constraints on Designing Effective Programs. The first constraint is political, by which I mean the difficulty of gaining Congress’s support for antipoverty programs. The poor are but a small minority with little in common other than their poverty. Many are dispirited, do not vote even if they are U.S. citizens, and are notoriously difficult to organize, especially at the national level. Accordingly, they exert relatively little pressure on Congress.201 They are heavily concentrated in urban centers and Appalachia, which dilutes their voting power nationally.* As noted earlier, most Americans still believe that self-help (including disciplined frugality) can achieve the American dream and thus that many of the poor are largely responsible for their low condition. The upward mobility of so many immigrants, ostensibly with little or no government assistance, fortifies this view. Antipoverty programs’ reputation as inefficient and fraud-prone reinforces political resistance to creating or expanding them.

One kind of trade-off that invariably constrains antipoverty programs arises out of the intersection of market forces and the public’s distinction between the deserving and the undeserving poor, one that Americans commonly make, at least implicitly, but that may apply differently in specific cases. Consider the EITC, discussed below. Most economists favor it for the working poor and agree that some “benefit reduction rate” must be imposed as income rises in order to phase out the subsidy. They recognize the trade-off that this entails between work incentives, target efficiency, the cost of the subsidy, and income supplementation for low-wage workers. Another example of this trade-off was President Nixon’s proposal of a negative income tax (NIT) in the early 1970s, which took the form of a family assistance plan that coupled a guaranteed minimum income with work and training requirements for the employable. The government funded rigorous experimental research in various cities to see what an NIT’s effects would be under different program designs. Key researchers in these experiments explain some of the trade-offs they found:

Economic theory unambiguously predicts that extending welfare to a new group will reduce their work effort. But the effect of different tax rates is ambiguous. In particular, lower tax rates are something of a two-edged sword: they clearly encourage people who are not working to take a job; but they also keep working families, who would have become ineligible under a higher tax rate, on welfare longer and thereby extend all the negative work incentives to those families. Which effect would predominate—more work by those currently on welfare or less work by those newly affected by it? Economic theory could not say, and no reliable evidence existed to answer the question.… The bottom line from the studies was that the NIT reduced rather than increased overall work effort.202

The NIT had another important effect: It increased marital breakups, probably by enabling unhappily married women to get by on their own. Upon learning of this effect, Senator Moynihan immediately dropped his support for the proposal, which doomed it.²⁰³

Particular Antipoverty Programs. The fact that the actual poverty rate, properly measured, has declined substantially over the last fifty years does not tell us how effective specific antipoverty programs have been. After all, Christopher Jencks notes, the rate “could have declined despite the War on Poverty, not because of it.”²⁰⁴ In assessing the most important programs explicitly targeted at poverty (in no particular order), I draw on his recent review of a collection of War on Poverty assessments²⁰⁵ and on other systematic evaluations.

Temporary Assistance for Needy Families (TANF)

Probably no social policy faces greater implementation obstacles than programs seeking to move unmarried, unemployed, poorly educated, welfare-dependent mothers to self-supporting work while assuring decent care for their children. TANF, which spent $17.3 billion in 2016, is no exception. But due to a combination of an unusual bipartisan coalition, encouraging state-level experimentation with various approaches, a robust economy, and state creation of work-readiness and childcare supports, TANF swiftly produced a rapid decline in the welfare caseload as welfare mothers entered the work force. But even TANF’s supporters worried about what would happen when a recession occurred, which it did beginning late in 2000 and again in 2008.

As it turned out, the former welfare recipients who TANF helped have held on to most of their gains, even after years of the Great Recession.206 A former head of New York City’s Human Resources agency and long-time administrator of TANF and other welfare-to-work programs summarized the data: “labor force participation among never-married mothers in the U.S. increased from 59.9 percent in 1995 (the year before [TANF] was passed) to a peak of 73.8 percent in 2001.… [L]abor force participation among this group was still 69.9 percent in 2013. This increase in labor force participation led to substantial declines in the official measure of poverty among never-married mothers. In 1995, the poverty rate among this group was 51 percent. By 2001 it declined to 38.5 percent and even after the Great Recession, the poverty rate among never-married mothers remains below that of 1995 (43 percent in 2013).”²⁰⁷

Against these successes are some shortcomings.

•  Although TANF’s provisions significantly raised child support payments by fathers from $12 billion to $32 billion between 1995 and 2012, that situation, Sawhill notes, “remains dismal. In 2011, only 35 percent of all single mothers received any child support, and among single mothers living in poverty, the proportion was lower: only 29 percent. Payments to these mothers are typically small; even excluding mothers who get no payments, the average benefit to an unmarried mother is around $4,500 per year …, and the average payment for low-income mothers is even smaller. Obviously, this is a drop in the bucket of what it takes to raise a family.”²⁰⁸ Brookings researchers propose that delinquent fathers care for their children while the mothers work, in lieu of cash payments.²⁰⁹ The practical obstacles to this seem insuperable.

•  The take-up rate for TANF-eligible families is far below those for SNAP, EITC, and AFDC (before its replacement by TANF); indeed, almost twenty years after its enactment, their participation rate is less than 33 percent. Finding out why should be a top priority. TANF did seek to substitute work income for AFDC benefits and has made some progress; female-headed families’ poverty rate is lower under TANF than it was under AFDC.210 More troubling, some states may have made it hard to apply and diverted some of their TANF block grants (more than $11 billion of the $16.5 billion, by one account)²¹¹ to other uses.²¹² If so, federal policy makers should rectify this.

•  So-called disconnected mothers—those who receive neither work-related income nor TANF benefits—are very poor, have few job prospects, and are growing in number and as a share of the low-income population.²¹³ And as noted earlier, “extreme poverty” (less than $2 per day cash income for at least a month) may be higher now than in 1996 when TANF was enacted.

•  TANF’s cash benefits are not indexed to inflation, so their value and adequacy have eroded substantially. In the median state, according to a congressional study, the maximum benefit for a family of three was 42 percent below the median state’s maximum AFDC benefit for such a family in 1988 in real terms.²¹⁴

Food Stamps (SNAP)

SNAP, which is administered by the U.S. Department of Agriculture (USDA), had 46.5 million recipients (in about half that number of households) in 2014, at a cost of $74.1 billion. The program expanded greatly during the Great Recession, but that crisis explains only about half of SNAP’s growth; the rest seems to reflect eased federal standards and states’ more aggressive enrollment (they bear few of the costs). Along with unemployment insurance, SNAP is our main countercyclical entitlement program, and it is well targeted on the nutritional and income needs of households that include children, seniors, or people with disabilities. Studies confirm its positive effects on the maternal and child health of recipients.²¹⁵ It made more than $2 billion in improper payments in 2014, but its error rate was actually lower than that of most other large antipoverty programs.²¹⁶

As a result of SNAP’s rapid expansion and relatively weak enforcement of work rules, Congress ordered the USDA in 2014 to seek to increase participants’ employment, earnings, and work incentives. In response, the USDA plans to commission research on the SNAP Employment and Training Program’s effectiveness, noting that research has been “relatively scarce.”217 An older study found that this employment program failed on all measures of employment, earnings, and general welfare cash assistance, and had a minute effect on reducing food stamp use.²¹⁸

Earned Income Tax Credit

The EITC was not part of the original War on Poverty. It was enacted in 1975 after Congress failed to enact a family income maintenance plan. In 2014, it provided more than $65 billion in refunds to 26.7 million low-income working families (an average of $2,400 per family). It is generally regarded as one of the most successful social policies for low-income families, and many other countries have now adopted it. Because the EITC is designed to increase work effort, it is politically popular. (Some of the benefit goes to employers, but the greatest benefit by far goes to the workers.)²¹⁹ Indeed, Congress has expanded eligibility and benefits several times since the Great Recession, and more than half the states now provide EITC supplements.²²⁰ The EITC draws many single mothers into the work force, and is also far better targeted at low-income workers with children than are minimum wage increases (discussed later) because the EITC is tied to household income, not a worker’s wage, and it varies with family size, so almost all benefits go to parents. It exemplifies what Mead approvingly calls “help and hassle” programs, providing new benefits but only alongside work requirements. He does question, however, whether the EITC in fact raises work effort rather than simply raising the incomes of those who are already working at a given level.²²¹ A July 2015 analysis has a less skeptical view than Mead’s; it predicts that expanding the EITC by $1,000 per recipient would increase employment of single mothers by 7.3 percent, substantially reducing family poverty.222

For all of its advantages, the EITC has some major flaws. Its effective marginal tax rate at the benefit phase-out level is high, sharply reducing work incentives as that point. It made $17 billion in improper payments in 2014 (exceeded only by Medicare), an error rate of more than 27 percent due to some combination of governmental errors, honest beneficiary errors, and outright fraud.²²³ The program provides little support to other single workers or single noncustodial parents; increasing it would be quite costly.²²⁴ Due to its complexity, many eligible families do not apply for the program, and the form of the benefit—a lump-sum payment once a year—helps beneficiaries less than a monthly wage supplement would (although many recipients have preferred the former).²²⁵ Some also question whether expanding the EITC would be cost-justified in terms of how much work effort it might engender in a period of historically low labor participation.²²⁶

Head Start

Head Start, which began in 1965, is a very popular, politically well-connected program that serves almost one million preschoolers, costing about $8 billion a year. The evidence on its effects is disputed. Supporters usually cite the Perry Preschool in Ypsilanti, Michigan, which opened in the early 1960s. For many years, researchers followed a pool of poor black three-year-olds who were divided into those randomly selected into the program and those not admitted. Studies found that any difference in test scores between the two groups vanished over time. A 2010 federal government study of Head Start found, similarly, that participation has no significant impact on measureable outcomes (e.g., academic, social-emotional, or health status at the end of the first grade). When the government then extended the study to third graders, it found the same “fadeout” of benefits. (Nobel economist James Heckman’s research may help to explain why; it suggests that even Head Start begins years too late to help disadvantaged children in school.) Preschool programs in other countries have also had mixed results, with no clear patterns explaining why some succeed more than others.

Nor is Head Start alone in these disappointing results. The Even Start Family Literacy Program failed to improve child and parent reading, at a cost exceeding $1 billion before it was finally defunded; the Scared Straight programs that bring at-risk youngsters into prison to learn about what might await them actually seem to increase their criminal behavior; and 21st Century Community Leaning Centers fail to improve academic outcomes but increase various forms of disciplinary problems, yet receive more than $1 billion a year in federal funds. A July 2013 Brookings research report on pre-K programs using random assignments finds little or no success,227 and an RCT of Tennessee’s statewide voluntary pre-K programs in 2015 found negative effects by the end of first grade.²²⁸ (In Canada, a study of Quebec’s universal child care program in 2015 found a large increase in its use but a deterioration in a host of measures of child well-being.)²²⁹

Still, pre-K education has strong advocates among liberal groups, citing the success of the Chicago Child-Parent Center Program²³⁰ and long-term positive effects of Perry and Head Start. A follow-up of Perry pupils at age twenty-seven found that they were less likely to have been arrested for any crime, had higher incomes, were less likely to smoke, and had other better outcomes. Several studies found that those who attended Head Start did better on high school completion and some other social indicia compared with siblings who did not enroll. Heckman has also found long-term positive effects on adult behavior and income. One skeptic, noting that these studies lack the methodological rigor of the RCTs that found that any positive effects soon dissipate, asks a sensible question about the studies finding delayed effects: “If Head Start has little to no effect when participants are children, how can the program suddenly become effective in adulthood?”²³¹ Jencks’s answer to this question is that Head Start may somehow help youths develop character traits that only affect outcomes later in life.²³² After thirty years of skepticism about Head Start, he “more or less became a convert: no persistent test score effects but non-trivial positive effects on social and economic outcomes in adulthood.”233

Title I

The program was a core element of the original War on Poverty and has been periodically reauthorized ever since. Under Title I, Washington distributes more than $14 billion annually to about 66,000 local education agencies (LEAs) serving 33 million students. Its purpose is to reduce inequalities in the resources available to LEAs, inequalities that arise because public education in the United States is almost entirely a local responsibility funded mostly by each local community through local taxes that reflect that community’s wealth.

Jencks summarizes the evidence on Title I’s effectiveness: The funding “helped reduce spending disparities between rich and poor districts in the 1960s and 1970s but not after 1990. Nor does Title I seem to have reduced expenditure differences between schools servicing rich and poor children in the same district. Finally there is no solid evidence that Title I reduced differences between rich and poor school districts on reading or math tests. If there was any such impact, it was small.” He also notes that Title I did give the federal government more leverage over LEAs when it came to desegregation and, more recently, to increasing accountability for test results. A November 2015 review by the Brookings Institution of Title I is quite critical:²³⁴

[I]ts funding per student is quite low, averaging about $500 to $600 a year. And there is little evidence that the overall program is effective or that its funds are used for effective services and activities. Large proportions of school principals report using Title I funds for teacher professional development, which many studies have shown to be ineffective and which teachers do not find valuable. Other services on which principals spent Title I funds include after-school and summer programs, technology purchases, and supplemental services, which also have been shown to be ineffective, and class-size reductions, which are unlikely to be of the size needed to generate effects found in previous research.

Upward Bound

This relatively small program ($265 million in 2014) was created in 1965 as part of the War on Poverty. It provides supplemental academic and support services and activities to help about 61,000 low-income high school students, disproportionately females and blacks, become college-ready. The final evaluation, which used RCTs, found that two years after they were scheduled to graduate from high school, the program had no effect on any of the twenty-one effectiveness measures, except for the high school dropout rate, which, surprisingly, was slightly higher for those who, prior to random assignment, indicated a “high expectation” of attending college. A follow-up study seven to nine years after scheduled high school graduation found that the program had no effect on the likelihood that the high-expectations students would attain a bachelor’s degree, certificate, or license, and it lowered their probability of attaining an associate’s degree. Unexpectedly, the program appears to have given a small boost to students with lower expectations, though it was unclear which post-secondary credential they obtained.235

Other Family-Centered Services

After a very detailed review of the formal assessments of many other federal programs intended to improve the prospects for poor families and children, Heritage Foundation analyst David Muhlhausen delivers a grim report: “Federal social programs largely fail at improving the academic and cognitive abilities of infants, toddlers, and other children. Overall, these programs do little to move people out of poverty. And forget about becoming economically self-sufficient.”²³⁶ The studies also found that these programs “fail to stop kids from having sex, get couples married, or keep marriages together.”* The most notable exception, he found, was welfare-to-work programs, and even there the positive effects were small. Some special services were beneficial but none succeeded in moving participants into full-time employment beyond what TANF’s work and job search requirements (and, to a lesser extent, the time limits) did after 1996.237

College Financial Aid

Federal grants and loans to college students mushroomed from $800 million in 1963–64 to $157 billion in 2010–11 (in constant dollars). The Pell Grant program, targeted at low-income students, cost $31.9 billion in 2013. Most federal financial aid, however, goes to the non-poor; 57 percent of college students now receive federal aid, up from 47 percent only four years ago. In July 2013, the Consumer Financial Protection Bureau estimated that outstanding student loans, almost all guaranteed by the federal government, had reached $1.2 trillion—a 20 percent rise since the end of 2011 and almost twice the 2008 level.

Although these programs are supposed to close the opportunity gap by enabling more children from low-income families to attend and complete college, Jencks reports, “[t]hat has not happened. In the mid-1990s Congress broadened federal financial aid to include middle-income families, reducing low-income students’ share of the pot. But the college attendance gap had been widening even before that.” The gap between college dropout rates of high- and low-income freshmen also increased between 1979 and 2001.²³⁸ Two-thirds of low-income freshmen and about half of middle-income freshmen in four-year colleges drop out. Poor targeting—a major failing of Title I at the pre-college level as well, as we just saw—is not the only serious flaw in student financial aid programs. Others include little marginal effect on the decision to attend college; much fraud; growing student debt; high and rising loan delinquency; encouraging tuition and fee increases; and little internal assessment of program effectiveness.239 “As an experiment in social policy,” Jencks concludes, “college loans may not have been a complete failure, but they cannot be counted as a success.”²⁴⁰ On the other hand, the efforts by the University of California system to attract lower-income students have been unusually effective and surely can provide lessons for other states and institutions.²⁴¹

Job Training

The GAO studied forty-seven programs designed to improve job information, training, and the hiring of unemployed or underemployed workers. The programs were spread across nine agencies and spent $18 billion in 2009. Only five of them had assessed their impacts and only some of these demonstrated positive effects; these effects tended to be small or short-term.²⁴² The United States is far behind OECD countries in assessing such programs, and the results on cost-effectiveness, such as they are, are underwhelming.²⁴³

According to Robert La Londe, a leading expert on these policies, job search assistance programs are more likely than other employment programs to yield positive short-term impacts. The mixed results for other programs depend on the targeted groups. Classroom training consistently generates modest short- and medium-term benefits for economically disadvantaged adult women, but just as consistently fails to produce gains for youths. For adult males, the results vary greatly across studies, partly because of methodological differences and sample sizes that are too small to measure program impacts with the same precision. On-the-job training programs have received less study. Some researchers find short-term benefits but with some doubt as to how long they persist. For reasons that are not altogether clear, private employers seldom participate in these programs, even though they might obtain substantial wage subsidies for hiring disadvantaged workers. A study of the Dayton Target Jobs tax credit during the 1980s suggested that these economically disadvantaged workers cannot overcome the labor market stigma of being classified as such.244 An exhaustive analysis of the studies on the fifty-year-old, $1.8 billion annually, Job Corps program finds few, if any, positive effects.²⁴⁵

Social Security Retirement

Jencks echoes other poverty analysts in saying that Social Security was “the simplest, least controversial, and most effective anti-poverty program of the past half-century”—even though, strictly speaking, it is not targeted toward the poor. In 1964, about 30 percent of the elderly were poor. Today, the poverty rate for the elderly is only two-thirds of the rate for eighteen- to sixty-four-year-olds and half the rate for children. Its administrative costs and error rates are low. Social Security has many critics among small-government conservatives, but the main functional critiques—that it displaces private savings for old age, pays too low a return, is vulnerable to both political meddling and long-term insolvency, and has become less progressive due to life expectancy differences and rising retirement ages—have had little resonance with the public.²⁴⁶

Social Security Disability (SSDI)

This program, part of the Social Security system, has alleviated much poverty by providing income replacement during periods when workers cannot work due to disability. But at the same time, it contributes to poverty through incentives to reduce work effort, in effect operating as a parallel unemployment insurance scheme—although one with badly backlogged claiming and payment processes. SSDI recipients have tripled since 1990 despite a much healthier working-age population. Much of this expansion occurred among young people, and little was caused by the Great Recession. The workers-to-disability recipient ratio has plummeted from 134:1 to only about 16:1 (or, by another measure, 11:1). Recipients are now much younger: In 1960, only 6 percent were in their thirties or early forties; by 2011, more than 15 percent were that young. More than 15 percent are now granted disability for “mood disorders” and another 29 percent for musculoskeletal and connective tissue conditions—both of which, being almost impossible to disprove, encourage fraud. A recent analysis of the program finds that although the expansion is mainly due to demographic changes (more baby boomers in their disability-prone years, more women in the work force), it is also driven by program changes that invited more fraud-prone claims, and by business cycle factors.247 Poor work incentives and program controls, a dysfunctional appeals process, and eased eligibility also play a role, with 58 percent of recipients now qualifying because of non-medical impairments.²⁴⁸

Supplemental Security Income (SSI)

This program, established in 1974, provides monthly cash payments to low-income people who are elderly, blind, or disabled. Although the asset criterion for eligibility has been restricted over time, more than 8 million people receive SSI benefits totaling in excess of $54.8 billion in 2013; 86 percent have disabilities, mostly mental. It constitutes the only source of income for almost 60 percent of the recipients, and substantially reduces the poverty gap for many recipients.²⁴⁹ Like other antipoverty programs, improper payments are large: Totaling more than $5.1 billion, with an error rate of 9.2 percent, it is the sixth largest among federal programs.²⁵⁰

Housing

According to a comprehensive 2015 analysis on which this part is largely based,²⁵¹ the federal government spends around $40 billion annually for means-tested housing programs, plus another $6 billion or so annually in tax expenditures on the Low Income Housing Tax Credit (LIHTC).* (This is well over twice the level of federal spending on either cash welfare or the Title I compensatory program in education and five times what it spends on Head Start.) These programs seek to make decent housing more affordable for the poor. In-kind housing programs, the theory goes, generate more housing consumption than would similarly costly cash transfers. That is, if the poor were given an equivalent value in cash, they would (except in the highest rent cities) probably spend less of it on housing than the government wants, and more of it on other things.

Public housing tends to be occupied by very poor people. Half of the households in public housing earn less than $10,000 a year, and only 4 percent are two-parent families with children. Much the same is true of those receiving housing vouchers, which subsidize rents in private apartments: Almost half of tenants earn less than $10,000 a year, and single parents rent some 900,000 units. Unlike recipients of TANF, who are subject to time limits on benefits, public-housing residents can remain indefinitely—and many do. The median “tenure” for both public-housing and voucher units is more than nine years; in New York City, the nation’s largest public-housing system, it is seventeen years.252 (For comparison, the median renter had lived in his home for 2.2 years.)

Two features of housing subsidies differentiate them from the other major antipoverty programs.

•  Public housing is located in particular neighborhoods rather than being attached to beneficiaries wherever located. Voucher recipients can use them most anywhere and do improve their housing conditions, but most still remain in low-income neighborhoods.

•  Targeting is crucial since only a small percentage of poor people receive these valuable benefits. Fewer than one-quarter of income-eligible households receive HUD benefits, but those who do can receive subsidies worth roughly $8,000 per year; in more expensive cities, they can be worth $12,000 per year or even more. More than 6.5 million households were on wait lists for these benefits in 2012.

The 2015 study summarizes the limited evidence on how these subsidies affect the poor:

[T]here is surprisingly little good evidence about the effects of existing programs on the behavior and well-being of participating families. We say “surprisingly” both because these programs consume significant amounts of government resources each year (and so are important), and because the excess demand for these program services (fewer than one out of four income-eligible families in the U.S. participates in such a program) would seem to offer numerous opportunities to carry out studies with truly comparable comparison groups.

The best available evidence suggests that increasing housing consumption without improving neighborhood conditions may have little detectable impact on conventional measures of human capital accumulation of children and may reduce labor supply of working-age adults. But these questions are hardly settled; there is considerable room for further evidence on the effects of housing subsidy receipt on families and children. Indeed there is virtually no evidence about the effects of what is currently our largest low-income housing subsidy program, the Low Income Housing Tax Credit program. There is more robust evidence about the importance of neighborhoods, which suggests that exposure to less-distressed neighborhood conditions can improve health outcomes, and, as may be suggested by more recent research, among young children to boost their long-term labor market success as adults.

POLICY REFORMS

The poverty rate in the United States, correctly defined, is much lower than it was fifty years ago, and antipoverty programs have played an important role in that progress. This is a great achievement but leaves us with a kind of paradox. The reduction was in the low-hanging fruit that could most easily be picked off the hard-to-reach branches of poverty. Cash, near-cash, and social services of various kinds could be dispensed to those poor people who were in the best position to take advantage of them. These benefits helped to lift many of them (or their children) into the safer though still-fragile precincts of the working class or, in some cases, beyond.

But the millions who remain behind must rest their hopes for better living conditions and social advancement elsewhere. For some, especially immigrants who come here in middle age or later without much opportunity to employ their old skills, to acquire new marketable ones, or to learn English well enough to enter the wider market economy, aspirations may focus on the next generation, which has a better chance of doing so. Indeed, for the vast majority of low-skill immigrants, full integration into American life has always been a multigenerational process. This may help explain the somewhat surprising finding that Latinos (and blacks) are more optimistic about the future than whites generally.253 Many other poor people, including some single and childless individuals, noncustodial parents, and the homeless, live mostly hand-to-mouth. They survive on food from soup kitchens and clothing from local social service agencies. More than 5 percent of households average seven months a year of hunger.²⁵⁴ The poorest depend on begging, petty crime, or SSI for cash, on nocturnal trips to shelters and brief stays with family and friends. Many of the elderly barely get by on Social Security, SSI, food stamps, and help from children. They do not expect upward mobility, only a dignified sustenance.

Other poor people need more, and perhaps different, help if they are to escape poverty. They include adolescents who do poorly in school, have little interest in it, and are unlikely to find relief in an unwelcoming job market; other young people neither in school nor in the job market;²⁵⁵ those with little or no family support and no useful networks outside their family. They are often dropouts from high school, community college, or vocational programs; mothers trying to raise small children alone with little or no child support (especially those who are also disconnected from both TANF and the job market); other young working-age individuals who can barely survive on their SSI checks and may not qualify for other benefits; criminals recently released from prison with no place to go; and others with poor prospects.

One source of help to the poor in America is private charity—on a scale that dwarfs such contributions in other wealthy societies where antipoverty efforts are almost entirely government-funded.256 Alexis de Tocqueville emphasized this distinctive feature of American society in the Jacksonian era, but it goes back much earlier to colonial times and our religious roots.²⁵⁷ In 2015, Americans donated an estimated $373 billion to more than one million charities, with 80 percent of the total given by individuals and bequests. This was the largest amount on record and an increase of 4 percent over 2014.²⁵⁸ The categories in which contributions are reported obscures how much goes directly or indirectly to the poor, but it is surely immense.* Perhaps surprisingly, poor people, who tend to be more religious, contribute more of their household incomes than rich people do; both groups give away much more than middle-class people do.²⁵⁹ For all the poverty alleviation that private charity on such a vast scale may do, however, no one should imagine that it could solve the problem of poverty in America, so the question remains: What should the federal government do about it?

In the remainder of this chapter, I canvass possible government strategies and policies for managing and ameliorating poverty. I think of strategies here as general, somewhat theorized approaches to either reducing poverty or minimizing its adverse effects on both the poor and the non-poor. Policies are the specific instruments for implementing strategies. (Many policies, of course, are already in place.) The strategies are not pure types—they merely serve as rubrics for the more specific policy discussion—and are less discrete and more eclectic than they appear to be. Indeed, because the strategies mirror our social diversity and complexity, they propel antipoverty policy in directions that sometimes seem inconsistent, even incoherent.* Three preliminary cautions are in order.

•  As already noted, policy makers do distinguish between the “deserving” and “undeserving” poor, although seldom explicitly. Rather, the distinction is implicit in how they define the policy problem and which policies they favor. (When using these terms below, then, I shall not enclose them in scare quotes.) But to target benefits on those most likely to use them to escape poverty is not nearly as straightforward as it sounds. In practice, it is hard to define, apply, and publicly justify the idea of deservingness. As is often the case, the most needy may not be the most deserving, and vice versa. Only 14 percent of Americans support giving cash to poor, able-bodied adults without dependent children. The least deserving group, in the public’s view, is a family with no adults who are continuously working (or, presumably, in training or school), and with no elderly or disabled member.²⁶⁰ According to journalist Eduardo Porter, these judgments about deservingness help explain why the very poorest families receive a much smaller percentage of government antipoverty support than they did in the early 1980s.²⁶¹

•  Serious policy proposals must be analyzed far more systematically than is possible here. Ideally (but probably seldom in practice), the analysis would clarify the proposal’s often conflicting goals; marshal and scrutinize the relevant facts; predict its downstream political and policy effects; consider all trade-offs that the proposal entails; compare its benefits and costs to alternatives including the status quo, with special attention to costs borne by invisible victims (especially likely with poor people); and be rigorously tested for effectiveness on a pilot basis. (A promising development in this last respect is the growing interest of for-profit companies in “social impact bonds” through which they can invest in research-tested pilot programs that may yield monetizable cost savings from promising innovative approaches to old problems.)262 All of this requires deep immersion in a program’s operational details and political context by analysts with no axe to grind. My goal here is to identify and clarify such choices rather than to firmly advocate solutions—except those that are most compelling.

•  Good policy requires good data, which is exceedingly hard to come by. RCTs and other high-quality studies often find that hopeful-sounding policy innovations—like the NIT discussed earlier—are unsuccessful (or worse), wasting the government’s time, money, and credibility.²⁶³

Any antipoverty approach that involves spending (i.e., almost all of them) must address how it can compete with other budgetary demands. The 800-pound gorilla in this room is the sizeable, demographically driven increase in the future costs of Social Security and Medicare, as well as more interest on a larger national debt.²⁶⁴ Those two programs are far more popular than antipoverty programs, and the debt interest is implacable. Any funding mechanism also raises a set of design issues: Who will bear the burden initially and then after the cost is passed on to others; the behavioral incentives and changes it might elicit; the government entity and level that imposes it; how difficult is it to administer; and other factors. No program can succeed without good answers to these questions, of course, and by saying little about them I in no way minimize their importance.

My purpose here is to discuss plausible strategies and policies, not how to pay for them. (Significantly, the sequestration provisions enacted in 2011, implemented starting in 2013, and relaxed only slightly for domestic programs in 2016, have seriously constrained the budgets of many domestic and defense programs, but they exempt almost all programs targeting the poor.)²⁶⁵ Nevertheless, a few points about the revenue side deserve brief mention.

•  Tax reform is both possible and desirable, though politically difficult. Eliminating or better targeting some of the hugely inefficient, poorly targeted tax expenditures that mainly benefit the better-off could raise substantial revenue.266 For example, about 70 percent of the benefits of the popular 529 College Savings program go to households making more than $200,000 a year, costing the government $1 billion in foregone revenue over the next decade.²⁶⁷ Congress could raise tax rates on the “1 percent” and give the proceeds to the poor through one of the approaches discussed here. Liberals often propose returning these rates to their much higher 1970s levels. Economists, however, note that the effective rates for that elite then were actually far lower than the nominal rates were, and indeed not so much higher than today’s.²⁶⁸ They also disagree sharply about the details of such a change and its effects. The top 1 percent (average income $1.3 million; more than a third comes from capital gains) already pay more in federal income taxes than the bottom 90 percent, and its share, now more than 35 percent of tax revenues, has grown significantly (even after the Bush tax cuts that favored the rich relative to other groups and relative to prior law).²⁶⁹

•  A 2015 Brookings study finds that even a large tax increase on the top 1 or even 5 percent would have only an “exceedingly modest effect on overall inequality”—an important reality check on optimistic predictions about how much redistribution this popular proposal would actually yields.²⁷⁰

•  Some key tax policy questions are: How high can the rates be raised without adversely affecting work and investment incentives and perhaps even lowering tax revenues? How much new legal tax avoidance would the super-rich undertake? How much of the new revenue would end up being redistributed to the poor rather than, say, be used for a perennially popular “middle-class tax cut”? Spread out over the entire poverty population, how much difference would it make? And how desirable would it be to reduce high-end inequality even if it does not raise all that much revenue?

Several other sources of funding are less controversial, at least in principle. The amount of fraud, waste, and abuse in federal programs is vast. If rectified—more enforcement could be highly cost-effective—the savings alone would be enough to substantially reduce or eliminate poverty. To cite just one egregious example: In 2012, the Treasury Department failed to collect either the $780 billion that Americans already owed in delinquent loans, fines, and penalties, or the additional $300 billion they already owed in back taxes. (This included $1 billion owed by federal employees and $757 million owed by federal contractors!) Improper payments under Medicare and Medicaid alone exceeded $77 billion in 2014.271 And “simply” ending clearly ineffective programs and redirecting their funds to poverty reduction would eliminate poverty, at least from a purely fiscal perspective (and not taking into account any adverse work incentives).

With all these clarifications in mind—and leaving the question of political feasibility to the pundits—I now present nine antipoverty strategies, along with some policy proposals that might be used to implement them. (I do not discuss separately here the numerous specific proposals to improve existing antipoverty programs, but I have mentioned some of them already [for example, TANF].)²⁷² They are: (1) encourage macroeconomic growth; (2) redistribute wealth directly to the poor; (3) help “disconnected mothers”; (4) supplement or insure wages; (5) enhance human capital; (6) strengthen families; (7) dismantle existing barriers to economic opportunity; (8) facilitate residential mobility; and (9) improve health.

Two omissions are worth noting here. First, I do not include a strategy to combat class bias, discussed earlier. Classism does indeed harm some poor people, but, as explained there, almost everyone finds it acceptable, including (or especially) the working poor whom classism often targets. Indeed, as they live closest to underclass dysfunction, they probably fear it the most. Second, I do not separately discuss Opportunity, Responsibility, Security, an admirable, ideologically broad-based report issued in December 2015 by a joint AEI/Brookings working group. My analysis draws heavily, often explicitly, on some of the scholars and analysts who produced it, and I am guided by the same three values underscored in the title of their report.

Macroeconomic Growth

Although poverty experts seldom discuss macroeconomic policy, most of them recognize that balanced economic growth* is the single best strategy for reducing poverty. For that reason, it is worth a brief mention.

By increasing demand for goods and services, economic growth expands the supply of jobs at decent wages that rise with productivity, strengthens poor people’s incentives to develop their human capital, and does not require convincing voters to support special programs aimed at the poor—programs that are hard to enact, fund, and (as we just saw) implement effectively. No economist doubts this. They also agree that economic growth is driven primarily by demand, investment, productivity, and innovation, and that the government has some role to play in fostering it—for example, by regulating pollution and other externalities, producing public goods that the market will not adequately supply, running an efficient tax system, and maintaining competitive markets through antitrust law, freer trade, disclosure, and other policies. They also agree that all levels of government erect too many barriers to starting new businesses, and that such barriers seem to be increasing.²⁷³ Even so, the United States is rated the most competitive large economy in the world.²⁷⁴

Economists famously disagree, however, about which regulatory techniques, deregulatory measures, public goods, competition policies, and tax regimes are most conducive to such growth. These debates reflect differences over what the government’s role should be, over what causes economic phenomena,* and over the likely effects of particular interventions. To vastly over-simplify long-standing, highly technical debates: Keynesians (or neo-Keynesians) believe that government spending should be used to maintain the demand that will produce such growth, and that the Federal Reserve should provide the liquidity to sustain it. In contrast, monetarists argue that the Fed’s chief role should be to maintain a rate of monetary growth that is stable and predictable, so that the “real” (non-financial) economy can function efficiently without monetary distortions. Conservative and liberal economists differ sharply about how efficient markets are, which types of regulation will improve that efficiency, and which policies—particularly tax law²⁷⁵—will strike the right balance between promoting efficient growth and ensuring fairness. There is no consensus on these issues, and little point for present purposes in further analyzing the competing arguments.

Redistributing Wealth Directly

One might think that the most straightforward way to reduce poverty is to have the government write checks to the poor. Alaska, Texas, and Wyoming have long practiced what one proponent calls “conservative socialism,” distributing proceeds from state-controlled mineral extraction.²⁷⁶ As we shall see in a moment, many of our foremost analysts—both liberal and conservative—have advocated this policy in some form. I shall describe the most plausible approaches and then explain why they would be very difficult, perhaps impossible, to sell politically. Another approach is to increase benefits and participation rates under existing antipoverty programs.

One way to achieve redistribution directly through the Treasury is to reduce federal taxes on the poor. Although few poor households now pay federal income tax (43 percent of all households do not pay it), the vast majority of poor workers do pay a federal payroll tax of 12.4 percent of covered earnings for Social Security.277 Making the tax even more regressive, it exempts earnings above $126,000 (in 2017). The Medicare tax is 2.9 percent of wages, and has no wage cap.* These payroll taxes are actually twice as high for workers as they seem: Although the employer nominally pays half the tax, standard economics teaches us that the actual incidence of the employer’s half comes out of wages, effectively reducing the worker’s pay by that amount, so the worker bears the full tax in the end. A strong case can be made for reducing the resulting fiscal hit on low-income workers, but the best way to do that is probably through a refundable tax credit like the EITS or an NIT discussed earlier and next.†

A much more ambitious approach would be some form of a universal basic income (UBI) grant to all Americans. Prominent policy analysts and scholars—including leading libertarian conservatives—have proposed such a grant, which could take the form of a guaranteed income. In 1962, Milton Friedman, a Nobel laureate in economics and conservative guru, proposed an NIT,²⁷⁸ an idea also supported by Friedrich von Hayek, another Nobel Prize winner in economics:

The assurance of a certain minimum income for everyone, or a sort of floor below which nobody need fall even when he is unable to provide for himself, appears not only to be a wholly legitimate protection against a risk common to all, but a necessary part of the Great Society in which the individual no longer has specific claims on the members of the particular small group into which he was born.

Hayek added, however, that such a scheme must be designed so as to “not destroy the market order or infringe on the basic principles of individual liberty.” Because he would apply it “only to those who themselves obey liberal principles,” it probably would not include the undeserving poor. 279 Many liberals have advocated a guaranteed income for the poor—most notably George McGovern, who proposed a “demogrant” in his 1972 presidential campaign. Legal scholars Bruce Ackerman and Anne Alstott proposed a scheme in which the government would pay $80,000 to every qualifying American at the verge of adulthood.²⁸⁰ Even the prominent conservative Charles Murray has endorsed a guaranteed income of $10,000, which would wholly replace all income-support programs.²⁸¹

Under most such proposals, the payments would be virtually universal, extending to all citizens, and would replace current federal means-tested programs. In 2012, Robert Rector listed seventy-nine such programs, costing $927 billion. About half of this sum goes for medical care. Another 40 percent is allocated for cash, food, and housing, and the remaining 10–12 percent is for programs that seek to help the poor become more self-sufficient. “If converted to cash,” Rector told the House Budget Committee, “means-tested welfare spending is more than sufficient to bring the income of every lower-income American to 200 percent of the federal poverty level, roughly $44,000 per year for a family of four.”²⁸² This fact, together with the fact that many poor people do not now receive benefits from these programs and all are beholden to the legions of bureaucrats who now regulate and dispense them, constitutes a powerful argument for cashing out these programs and using some or all of the proceeds to fund a guaranteed income that will eliminate poverty (at least as officially defined).

In the United States, direct redistribution schemes of this sort are political non-starters.* As we have seen, even a minimal guaranteed family income plan could not pass Congress in the early 1970s, and it would be no more palatable today. Indeed, the elderly, a crucial voting bloc, increasingly oppose such redistribution.283 Congress has not enacted a new low-income entitlement since the Child Tax Credit (CTC) in 1997. The only programs that provide cash (or near-cash) directly to the poor are those that claim to be insurance programs to which recipients contributed (Social Security retirement); provide basic nutrition (SNAP); assist the putatively deserving poor (SSI and EITC); are time- and amount-limited, and hedged with eligibility restrictions (TANF); and provide a refundable tax credit to poor working families with children (CTC). Henry Aaron, a senior Brookings analyst, sees most Americans as “commodity egalitarians”—they will support programs that prevent poor people from going hungry, living in decrepit housing, or going untreated—but only a tiny minority are “cash egalitarians” who support providing money to those without cash. Thus, “we far more willingly vote refundable tax credits to pay for health care than we do refundable tax credits for low incomes—besides, troublesome though administration of the health credits may be, it is child’s play next to administering large transfers of cash.”²⁸⁴ Past experience supports this assessment.

A UBI would require some tough policy choices—for example, which conditions should be imposed (e.g., work or training effort), whether higher-income people should be included, whether the payment itself should be taxable, the scheme’s effects on state programs, what happens to recipients who still run out of money, and especially how it affects work incentives. (Americans work many more hours per year than Europeans do.) Even if a UBI were limited to the working poor, work effort would likely decline. The many whose work is repetitious, boring, physically demanding, dirty, dangerous, or poorly paid would need ever more subsidies just to be sustained at any given income level. Finally, as the Economist points out in a broad review of such proposals, a UBI would wreak havoc on immigration policy: “The right to an income would encourage rich-world governments either to shut the doors to immigrants, or to create second-class citizenries without access to state support.”²⁸⁵

A more politically viable approach to redistribution would be to increase benefits under existing antipoverty programs. There are at least four ways to do this.

•  Congress could raise benefit levels under programs like EITC, CTC, TANF, and SSI. Grants or larger tax subsidies for children, as the Century Foundation proposes, might be more attractive politically.286

•  It could index cash and near-cash benefits to inflation (or some portion of it), just as it did for Social Security decades ago.

•  Using a conditional cash transfer (CCT) approach, it could grant additional cash benefits conditional on improved behaviors regarding preventive health measures, children’s school attendance, and other desirable conduct. A controlled trial of CCT in several cities reported in September 2016 that it had few significant effects even with the addition of proactive guidance of the families, and it reduced work effort somewhat.²⁸⁷

•  The programs themselves could surely increase participation rates among eligible poor people, which are often surprisingly low. A recent congressional study of take-up rates in 2012 found that only 70 percent participated in SNAP, 65 percent in WIC, and a mere 28 percent in TANF.²⁸⁸ Take-up rates respond to bureaucratic efforts, as large state-to-state variations and some studies of policy changes show.²⁸⁹ Even so, many who know they are eligible still fail to apply because they don’t want to depend on government.²⁹⁰ This is not true of Medicare and Social Security. TANF is long overdue for reauthorization (it has been on continuing resolutions since 2010), which should be the occasion for its reform.

A final approach to redistribution is to create new entitlements to social provision or social insurance that relieve the poor of having to spend what little money they have on such goods, shifting those costs to non-poor taxpayers. The recent expansion of Medicaid did this, and proposals for more universal child care support and free community colleges would do so as well. More flexible work arrangements that enable more poor people to both work and care for their children would also enhance income. And eliminating the digital divide is a relatively cheap and uncontroversial way to enhance opportunity.291 As always, however, much turns on the policy details.

Disconnected Mothers

As noted previously, a large and growing number of single mothers are disconnected from TANF work (they may be unemployable), EITC (the same reason), child support (the fathers are too poor or elude enforcement), or simply fall through the safety net for other reasons. A surprising—and encouraging—finding, however, is that “most” of the $2-a-day poor (not including government transfers) “have a recent history of work”²⁹² and so may be less disconnected than one might assume. Poverty analysts like Rebecca Blank have proposed ways to help these disconnected adults and their children, usually by temporarily waiving TANF and SNAP work requirements while conducting case-by-case reviews of whether and how they can be made employable, and if not, designing arrangements that will support them without undermining the work rules.²⁹³ In addition, programs to increase paternal child support, which has declined significantly in the last decade, can be strengthened by linking them to benefits under other antipoverty programs, including job training for non-custodial parents.²⁹⁴

Wage Supplements and Insurance

There are three main ways to augment earnings from work: raise the minimum wage; increase EITC benefits; and insure against wage losses.

Minimum Wage. The federal minimum wage (as well as state and local ones) can be raised. This approach has galvanized public support in recent years, as unions and other advocates for low-income workers mounted successful campaigns in various cities and states to raise it (sometimes across the board, sometimes for particular groups such as fast-food workers or chains of a particular size) to $15 an hour, while pressing Congress to raise the national minimum to that level. Minimum wage increases, however, are not well targeted on the working poor, as noted earlier. A 2010 study found that less than 25 percent of those who receive the minimum wage are poor or near-poor, while 63 percent live in households with incomes above twice the poverty line; 42 percent are in households above three times the poverty line.295

Numerous studies of the income and employment effects of minimum wage increases, including recent reviews,²⁹⁶ find negative employment effects on low-skill worker employment (especially entry-level; teen employment declined sharply between 2000 and 2014);²⁹⁷ on consumers of the products and services they produce; and on small businesses. Such increases also encourage employers to substitute machines for labor. David Card and Alan Krueger’s 1994 study, the major exception,²⁹⁸ was strongly criticized for its data and methodology.²⁹⁹ In 2014, the CBO reviewed the many studies on the minimum wage, analyzing two options. The largest was a raise to $10.10 an hour (much lower than jurisdictions have been enacting since then). The CBO found that this would move 900,000 workers above the (officially defined) poverty line while reducing employment by at least 500,000 jobs.³⁰⁰ The job losers, only 0.3 percent of the workforce, are a highly vulnerable group, disproportionately minorities and high school dropouts.³⁰¹ Small minimum wage increases may have less adverse effects on jobs, but those as large as some jurisdictions are now adopting, especially if focused on specific sectors and firm sizes, may lead to larger long-term effects. Although harder to predict, they are also likely to have a net negative effect on low-skill workers.³⁰² A recent research summary predicts much larger longer-term (if the increase is inflation indexed) job losses: in the restaurant industry, one hundred times more than in the short term, and much lower future job growth for young, unskilled workers.³⁰³ (President Obama has also ordered an increase in overtime pay under the Fair Labor Standards Act, a step that, like minimum wage increases, will likely have mixed effects on the poor.)304 Finally, a 2015 review finds that the minimum wage reduces inequality somewhat at the lower end of the wage distribution, but the impact is “almost negligible for males” and “the net effects on inequality are less than half as large as previously claimed.”³⁰⁵

EITC. Several proposed reforms have considerable bipartisan support.³⁰⁶ Including noncustodial single workers (Mead would condition this on their working full time and paying child support, if any)³⁰⁷ would create few, if any, of the job-displacement or relocation risks of a higher minimum wage. In addition, government could subsidize employers to pay higher wages or make it easier for unions to organize workers—but both approaches have a number of efficiency and other disadvantages that cannot be analyzed here.

Wage Insurance. Some leading economists propose adding protection against wage loss to existing safety net programs. Existing protections are limited to the disabled (SSDI) and to older workers displaced by foreign competition and earning less than $50,000. Thus, ample room for expansion of such protection exists—probably through a progressive form of payroll tax, together with some restrictions for limiting moral hazard.³⁰⁸

Enhancing Human Capital. Permanent escape from poverty means, at a minimum, that one must possess the knowledge, analytical ability, and training to be a productive citizen who can earn an adequate income in the market. Our educational system is supposed to provide this. Traditionally, it led the world but that day is long past, except in post-secondary education, where the United States is peerless. Interested readers will find a vast body of research on our system presented in a recent book by leading educational policy experts, on which I shall draw here.³⁰⁹ My limited purpose is to summarize some findings on the education of low-income children, and to describe some programmatic approaches that reformers are now investigating.

•  Only seven of the seventy-seven educational interventions evaluated by RCTs (without major study limitations) commissioned by the Institute for Education Sciences (IES) since 2002 were found to produce positive effects—despite the fact that most of them had shown promising results in earlier research. And statistics like these likely overstate the fraction of interventions found to be successful due to publication bias—i.e., the tendency for researchers and journals to dismiss or reject studies that find no effect.310

•  The public has a low opinion of our public schools: “They know that only somewhat more than 70 percent of students graduate from high school within four years after entering ninth grade. Only about one-fifth of the public gives the nation’s schools an A or a B on the five-letter scale traditionally used to grade students [but grade their own children’s schools higher].”³¹¹

•  Our failures in educating children have continued despite enormous increases in per-pupil expenditures and reductions in class size. Results are so consistently negative that a broad scholarly consensus has emerged that simple increases in financial resources or reductions in class size across the board, without more fundamental changes, yield few benefits for student performance. Of course, perhaps more compelling than the academic debates is the fact that the United States has tried these policies over the past half century with little to no obvious results; the country has dramatically increased expenditures and reduced pupil-teacher ratios, yet [National Assessment of Educational Progress] scores of U.S. seventeen-year-olds have been constant.… Just about as many high-spending states showed relatively small gains as showed large ones.… [The research does not hold that money never matters].… How money is spent is much more important than how much is spent.³¹²

• Although low-income children enter school with many disadvantages in terms of vocabulary, reading skills, home stability, health, stress, and parental support, these factors do not fully explain their low academic outcomes. What goes on within the schools is also crucial.

•  “[P]erformance differences between the United States and other countries [cannot] be attributed simply to ethnic heterogeneity or poverty. The United States is not the only country with a diverse population.… [T]he proficiency rates of white students and of the most advantaged socioeconomic subgroup of children of college-educated parents in the United States fail to keep pace with the proficiency rates of all students in many countries across the globe.”313

•  The quality of teachers can make a difference to academic and later-life outcomes for students, especially low-income ones: “One study, focusing entirely on a disadvantaged group of students in Gary, Indiana, finds that the difference in student performance gains with a highly effective teacher compared to an ineffective teacher was one full year of learning for each academic year: the top teachers produced gains of 1.5 grade levels, while the bottom teachers produced gains of 0.5 grade level in an academic year. Again, these impacts of teachers were recorded for precisely the group of students that are the focus of those emphasizing the centrality of socioeconomic issues. Furthermore, the well-documented differences in teacher effectiveness indicate that three to five years in a class with a top teacher versus an average teacher is sufficient to erase the average achievement gap between poor and better-off children. That is, regardless of the source of any preexisting achievement differences, effective teachers can eliminate them. Family background is not fate.”³¹⁴

•  These facts are widely appreciated by educational reformers, who have devised many programs to address them. Earlier, I discussed some of them, such as Head Start and other early childhood, pre-K programs. The most promising approaches for elementary and secondary education, such as KIPP, grow out of the alternative (often charter) schools and parental choice movements that have gained considerable traction. In a multiyear study of 41 urban areas in 22 states, a Stanford research center found that compared with matched peers, charter school students experienced greater growth in math and reading.³¹⁵ These gains were particularly pronounced among blacks, Hispanics, low-income, and special education students, as charter students typically receive 25-plus days of extra learning.316 A RAND Corporation study even found that charter high schools outpaced regular district schools in graduation rates (by 15 percent) and college attendance rates (by 8 percent).³¹⁷ On the other hand, recent studies of statewide programs in Louisiana and Indiana providing vouchers for private schools found lower academic performance compared with similar students that remained in public schools.³¹⁸

•  Despite the demonstrated effectiveness of some of these alternative school programs, they still account for a tiny fraction of the nation’s students. Another study ascribes the limited reach of such reforms to the fierce campaigns waged against them by the teachers’ unions: “Opposition to change is not the result of intense conflict among the public at large. Differences of opinion do not loom large between parents and nonparents, old and young, affluent and those of more moderate means, renters and homeowners, those from different religious and cultural backgrounds or among groups of varying ethnic background, or even between Democrats and Republicans. By far the deepest cleavage over education policy is between the members of the teaching profession and the general public.”³¹⁹

More low-income students who manage to complete high school go on to college than ever before (45 percent in 2012, up from 28 percent in 1970). But this good news is subject to some important caveats.

•  Many of those who are qualified for admission to high-ranked private universities do not apply (this is also true, to a lesser extent, of more affluent ones) and end up enrolling in lower-ranked public ones, even though the former (because of financial aid packages) would have cost them less than the latter. (Research shows that they could be better informed about this choice at trivial cost.)³²⁰

•  The vast majority of them do not graduate—only 20 percent, about the same as in 1970.³²¹ (This includes those who enrolled in two-year colleges.) Indeed, only 55 percent of all students who matriculate in U.S. colleges graduate322—the highest dropout rate in the developed world.³²³ Also troubling, a prominent study finds that college students actually don’t learn much there.³²⁴ The problem is not just affordability. Consider a Kalamazoo, Michigan, program that makes college tuition-free for the city’s high school graduates, of whom nearly 90 percent from the classes of 2006–08 were eligible. Even so, only 40 percent of the program recipients earned a post-secondary credential within six years of high school graduation, and only 21 percent of those who enrolled in community colleges earned a post-secondary credential.³²⁵ This suggests that President Obama’s plan to make community college free for two years³²⁶ may help low-income students get to campus, but keeping them there is another thing altogether. Indeed, studies suggest that comparable students are more likely to graduate from four-year colleges than from community colleges.³²⁷

•  The immense for-profit college-and-vocational training industry enrolls primarily older, female, low-income job-seekers. Typically, these schools get the vast majority of their revenue from federal student loans, and they account for nearly half of the defaults. When Corinthian Colleges, one of the largest, filed for bankruptcy in 2015 after repeated criticism for maladministration and poor teaching, the government decided to forgive loans to their students, estimating that if all 350,000 of them over the last five years received forgiveness, it could cost $3.5 billion.³²⁸ Under pressure, some other private companies may forgive some loans as well.³²⁹ Corinthian’s failure may be just the tip of the iceberg of for-profit educational fraud.³³⁰ Far more extensive loan forgiveness will place even more political pressure on the federal loan system on which so many low-income people depend. More and better information to applicants about these schools and the economic payoffs to their students is essential, but designing and enforcing such disclosure systems is difficult.³³¹ A more radical approach would bar government loans to students attending schools with very low graduation rates, but this would deny many low-income students access. Making grants for only the first year and thereafter providing loans based on continued performance should also be tried.

Some private initiatives to train unskilled youngsters for good jobs show more promise. Working with high schools, some major corporations, hospitals, and other entities have instituted a program, known as P-Tech, that is designed to transform the idea of vocational education. It is a six-year program, based on a company-developed curriculum combining conventional high school subjects with science, technology, and math training that leads students to an assosciate’s degree at no cost to them, with mentoring and paid internships along the way. According to the Economist, more than seventy small and large companies have already adopted, or are working with schools to adopt, this model—so far with considerable success.332 Such employer-led approaches should be expanded.³³³ Some classroom-linked, specific-field certificate programs do fairly well on job-readiness preparation with weak students; cities vary considerably in their use of them.³³⁴ Finally, reformers are showing much more interest in apprenticeship programs, drawing on the experience of other countries that have used them more extensively than the United States to prepare academically limited poor people for jobs.³³⁵

Other approaches to human capital focus on altering the behavioral patterns of at-risk youngsters, including their decisions to drop out of high school, get involved with drugs or gangs, and get into confrontations that can escalate to violence. In three large RCTs, researchers found that certain techniques designed to reduce the tendency of poor young men to respond unthinkingly to dangerous social stimuli in high-stakes situations did succeed in reducing their criminality and dangerous choices.³³⁶

Strengthening Families. Of all antipoverty strategies, this is surely the most important. As noted earlier, the single best predictor of low economic mobility is the fraction of children with single parents. Research also finds that “Children of married parents also have higher rates of upward mobility if they live in communities with fewer single parents.”337* But this strategy is also the most challenging. Although family is the basic unit of any society, its dynamics differs among cultures, resists outside intervention, seems impervious to conventional policy tools, and for all its ubiquity and “familiarity” remains poorly understood.

In this discussion, I consider three different approaches to strengthening families: delay their formation through better choices, encourage marriage, and provide services to at-risk families. (Needless to say, much of the earlier analysis of poverty’s causes and possible remedies is highly relevant to the flourishing of families, so I shall try to avoid repetition.)

Delaying family formation means not having children until after completion of what Sawhill and Ron Haskins call the “success sequence” consisting of three key choices that, singly and especially together, would greatly reduce the poverty rate.

•   Work effort is clearly the most important. This was discussed earlier.

•  Having children only when married and after age twenty-one.

•  If every household head at least completed high school, the rate would drop even further. (GED holders do not do as well.)³³⁸

Newer simulations using 2013 data yielded similar findings.³³⁹ Even among those who follow the success sequence, whites tend to do better across a broad range of poverty- and middle-level incomes.³⁴⁰

Poor people obviously have more control over some of these choices than others. High school completion and delaying children are more clearly matters of choice. Marital stability is some combination of rational choice (whether, when, and whom to marry) and unpredictable developments. Sawhill and Haskins know, of course, that full-time work is not always simply a matter of choice, but they note three facts relevant to this point:

•  Lack of available jobs is not the main reason why the poor work so much less than the non-poor do. An update of their analysis based on 2014 data yields a remarkable finding: for nonworking heads of households who do not receive disability income, a surprisingly small fraction (only 11 percent of females and 20 percent of males) gave as the reason for not working that they were “unable to find work.” The most important reason women gave for not working was “taking care of home or family,” while for men it was being “ill or disabled.” (Note, again, that the study excluded those who said that they received SSDI or SSI, as well as adults younger than age twenty-five, or older than fifty-four.)341

•  Given how antipoverty programs are structured, “disincentives to work are large. As incomes rise, various forms of assistance are scaled back and payroll and other taxes begin to take a larger bite out of people’s earnings.”³⁴²

•  Choices about work affect the other three choices (and probably additional ones as well): “Young people who know that they are going to have to work would be more likely to finish school. Those who aspire to be stay-at-home mothers for an extended period would be more likely to delay having children until they are married since the government would no longer subsidize them to be full-time mothers. And those required to work would have less time to care for additional children and might plan their families accordingly. Indeed, serious work requirements may be more of an incentive to finish school, delay childbearing until marriage, and limit the size of one’s family than all the combined government programs directly aimed at these objectives.³⁴³

The “success sequence”—if young low-income people could somehow be induced to follow it—would enable many or most of them to avoid or escape poverty. Of course, altering behavior shaped by youthful impulsiveness and poor judgment, sexual and maternal urges, inter-personal power struggles, and cultural signals is exceedingly difficult, as many social reformers have found.* Our society has a huge stake in propagating the importance of the success sequence to one’s prospects in life, while designing policies to encourage people to make those choices—certainly the three that are wholly within their control. And we have made great progress on one of these choices. “Teen pregnancy rates,” Haskins reports, “have declined every year except two since 1991; over that period the rate has declined more than 50 percent, saving billions of dollars in taxpayer spending and improving other serious problems including school failure and poverty.”344 The cause(s) of the decline are not certain, but the abstinence education that Congress mandated in 1996 clearly was not one of them.³⁴⁵ Even so, Haskins notes, we still have the highest rate among advanced economies.

Sawhill’s 2014 book, Generation Unbound: Drifting into Sex and Parenthood without Marriage, establishes a powerful nexus between poverty reduction and carefully planned and deferred parenthood. Echoing the American College of Obstetricians and Gynecologists,³⁴⁶ she urges that at-risk teens and women be educated about sexuality, contraception, abortion and its alternatives, and the dire consequences of premature family formation. Indeed, even after a decades-long drop in teenage births and the lowest number of unwanted pregnancies (defined as unwanted now or within the next two years) since the 1980s,³⁴⁷ nearly half of each year’s 6.6 million pregnancies are still unintended. The abortion rate for poor women has risen nearly 20 percent in the last twenty years, while dropping nearly 30 percent for well-off ones.³⁴⁸ Yet a simple remedy for unwanted births exists: delaying childbearing through long-acting reversible contraceptives (LARCs), in the form of IUDs and hormonal implants, which are twenty times more effective than the pill and other common methods, and are safe. Indeed, in only five years, Colorado has used LARCs to raise the median age of first births to women in its poorest areas from before age 21 to after age 24, thus helping young women to finish school and gain a foothold in the job market.349 Even so, fewer than 10 percent of sexually active women use the devices—partly because many doctors fail to recommend them³⁵⁰ and partly, one presumes, because many women are reluctant to make a long-term commitment to not getting pregnant.³⁵¹

A second way to strengthen families—encouraging marriage—is less tractable to public policy. Indeed, it runs counter to one of our (and Europe’s) strongest and most troubling trends: non-marital childbearing, discussed earlier. The second Bush administration’s Health Marriage Initiative designed to encourage and sustain marriage was evidently ineffective.³⁵² And while Robert Rector, Sawhill, and others note the substantial “marriage penalties” in our tax code and welfare programs, these appear to have little effect on marriage one way or the other. (They would also be fiscally very costly to eliminate.)³⁵³

The third approach is to provide discrete support services to low-income households. Public and private programs of this kind have existed for generations; they are numerous and remarkably diverse. Taken together, they seek to address every cause of poverty discussed above—and then some (e.g., religious approaches). One private-sector success, celebrated by management guru Peter Drucker, is the Salvation Army’s ability to discourage bad choices.³⁵⁴ A governmental success, the Nurse-Family Partnership program sends nurses to visit poor women with first pregnancies, counseling them on nutrition, avoidance of risky behavior, breast-feeding, conversational teaching, and affection—and continues their visits until the child reaches age 2. According to three RCTs, the program improves many social outcomes and is cost-effective.³⁵⁵ But relatively few other service programs have been rigorously evaluated.

Dismantle Barriers to Opportunity. Perversely, the same governments that claim to fight poverty—including through civil rights law mandating equal opportunity in employment, credit, and other areas—often help to entrench it by erecting ill-advised barriers to self-help and small-scale entrepreneurship. (The United States ranks forty-sixth in the world in ease of starting a business!)356 One area in which this occurs is occupational licensing rules, which often serve little or no purpose except to protect existing businesses from competition and innovation by new, low-income entrants. The portion of the U.S. workforce covered by state licensing grew from less than 5 percent in the early 1950s to 25 percent in 2008—largely because of the number of occupations requiring a license, rather than due to employment growth within certain heavily licensed fields such as health care and education.³⁵⁷ The Institute of Justice, a conservative public-interest organization, has challenged many of them—laws requiring hair braiders to first obtain a cosmetology license requiring 1600 hours at a private cosmetology school, laws limiting vending or car service opportunities to a monopoly firm, laws barring the sale of caskets, laws requiring florists and interior designers to pass a licensing exam, and many others.³⁵⁸ President Obama’s Council of Economic Advisers recently expressed alarm about the growth of these restrictions. Today, it found, they cost millions of jobs nationwide and cost consumers over $100 billion annually. The Council proposes more efforts to dismantle them.³⁵⁹

Several other kinds of barriers raise closer policy questions where legitimate competing interests exist on both sides—unlike the absurd anti-competitive licensing mandates that have no plausible public-interest justification just discussed. One example is employer rules requiring job applicants to agree to credit checks or to disclose any criminal record (discussed earlier). Like many other progressives, Joseph Fishkin strongly opposes such practices, viewing them as “opportunity bottlenecks.”³⁶⁰ But past criminal activity or chronic credit problems sometimes helps predict an applicant’s character and likely job performance, information that reasonable employers might legitimately consider relevant. Another example of a close question is the use by some institutions of standardized tests on which poor people, especially blacks, do worse on average. The merits of such tests can certainly be debated (some institutions have abandoned them while others find them essential) and Fishkin’s distinctions based on a practice’s arbitrariness, legitimacy, and strictness are helpful in assessing its use. But policy makers must consider the possibility that banning them may induce employers and institutions to make worse decisions by resorting to cruder proxies (address, for example) that turn out to be more unfair to the poor.

Residential Mobility. In the earlier discussion on isolation as a cause of poverty, we saw the importance of the geographical enclave factor in entrenching poverty, now confirmed by the Chetty group’s research. So convincing are these findings that efforts are under way in some low-income neighborhoods to implement the Chetty strategy of subsidized out-migration through more tenant- and landlord-friendly Section 8 vouchers and other affordable housing programs.361 Like previous efforts,³⁶² these will surely encounter fiscal, zoning, classist, environmental, political, and other barriers that exemplify the competing interests at stake.³⁶³ In July 2015, the Obama administration adopted a new remedy in the form of a HUD rule designed to induce, if not force, local communities to integrate housing economically, not just racially.³⁶⁴ Whether the next administration continues this rule and how it implements it, of course, remains to be seen. On the other hand, some evidence suggests that even poor people who are mobile usually move to neighborhoods that are little better than the ones that they left.³⁶⁵ A New York study suggests that improving the neighborhoods where they already live (often criticized as gentrification) may be a better approach than trying to move them.³⁶⁶ Finally, programs could encourage the jobless in declining regions to move to job-growth areas in other states—what Henry Olson proposes as a “new Homestead Act.”³⁶⁷

Improve Health. Poor people are much less healthy than better-off Americans, whether the index is acute and chronic diseases, mental health, or longevity.³⁶⁸ Unsurprisingly, poor health correlates with (and presumably causes and is caused by) joblessness, family chaos, parental neglect, and other conditions associated with poverty. A partial list of reasons for this would include higher rates of smoking, diabetes, obesity, substance abuse, high blood pressure, inactivity, accidents, and violence. Indeed, the Health Inequality Project reported in 2016 that these behaviors are far more detrimental to life expectancy among the poor than either levels of income inequality or access to health care.369 (Poor people use fewer health services even when they have insured access!)³⁷⁰

This is why public and private programs have long focused on improving poor people’s health. Changing entrenched behaviors, however, is notoriously difficult. Solutions—other than expanding access to health care (which of course addresses symptoms more than causes)—are far from clear. Unless effectiveness seems very likely on the program’s face, the most promising approaches should at a minimum be rigorously evaluated and resources targeted accordingly.

CONCLUSION

Poverty is an immensely complex, tangled, and recalcitrant problem. Among the wealthiest societies in the world, America is also uniquely charitable, innovative, and animated by a can-do spirit, yet we have failed to solve the poverty issue. This does not mean that we have not made significant progress. Properly measured, the poverty rate is far below what it was when the War on Poverty began a half-century ago, especially for the elderly. Moreover, the standard of living of most people who still live in poverty is much higher than it was then. Through a combination of government transfer programs and economic growth, the United States has managed to abolish the worst forms of material deprivation for the vast majority of the poor.

But the hardest-core poverty has proved (by definition) resistant to our best efforts—and to the trillions of dollars that we have devoted to the task. The evidence casts serious doubt on the hope that more money for more programs would cure it. Yes, better teachers could make a big difference for poor children and higher pay might attract more of them, but they cannot remedy the domestic chaos that brings so many already severely disadvantaged students into the nation’s classrooms. Tragically, many of the reinforcing causes of this chaos—family breakdown, incarceration, educational deficits, and bad choices—have worsened, especially for poor black men. Their disadvantages will surely be deepened by the many forces transforming America: global competition; mechanization; robotics; fewer decent-paying unskilled jobs, with highly motivated immigrants competing for them; more children raised in non-marital households; the decline of many urban cores (and the rise of others that the poor can no longer afford); waning religiosity; and less confidence that government can solve these problems.

Americans should take some comfort in the real progress that has been made in reducing poverty. But just as the costs and difficulties of solving the last 10 percent of almost any social problem far exceed those necessary to reach that point, eliminating the poverty that remains will be much, much harder to accomplish.

* Christopher Jencks, a leading scholar on poverty policy, rightly states: “The community action programs that challenged the authority of elected local officials … might have been a fine idea if they had been privately funded, but using federal money to pay for attacks on elected officials was a political disaster.” http://www.nybooks.com/articles/archives/2015/apr/01/war-poverty-was-it-lost/. They have little anti-poverty significance today, and I shall say no more about them.

* Outside the United States, progress in eliminating the most abject poverty has been dramatic in some areas, particularly China, but the levels of absolute deprivation remain high. Rakesh Kochhar, “A Global Middle Class Is More Promise Than Reality,” Pew Research Center, July 8, 2015.

* Saez defines family market income to include salaries and realized capital gains but not government transfers. As we shall see, these transfers are a very important component of overall income for poor families.

* It is not clear which way this would cut. Some psychological studies suggest that smaller differences in well-being may only make the remaining ones more objectionable.

* In the United States and most if not all other societies, wealth is distributed much more unequally than income, while consumption is distributed much less unequally.

* They also increase the number of poor people to the extent that they depress wages for low-income American workers. As discussed in chapter 3 labor economists intensely debate this question.

* On the other side, economist Robert Frank explains that recognizing one’s own good luck seems to increase one’s generosity to others with less of it. http://www.theatlantic.com/magazine/archive/2016/05/why-luck-matters-more-than-you-might-think/476394/

* For law professor Robin Lenhardt, the dire conditions of black subordination deprived marriage of the dignitary value that white America—and the Supreme Court, most recently in its Obergefell decision upholding same-sex marriage—celebrates, and help explain why blacks are the least-married group in America. For blacks, she argues, non-marriage is more dignity-enhancing and should be supported by social policy. Lenhardt, “Race, Dignity, and the Right to Marry,” 84 Fordham Law Review 53 (2015).

* This increased high school completion rate may help to explain why the National Assessment of Educational Progress, the leading measure of student performance, has remained fairly flat in recent years, as many relatively marginal students remain in school longer and take the tests.

* High-income Americans are also isolated from the mainstream culture, according to Charles Murray’s “bubble quiz.” http://www.aei.org/publication/lessons-from-the-bubble-quiz-1/?utm_source=paramount&utm_medium=email&utm_content=AEITODAY&utm_campaign=040516

* Less encouraging is research on the effects of the Harlem Children’s Zone, the most celebrated effort to flood a low-income neighborhood with improved social services and targeted programs. A Brookings Institution analysis of the research found that the charter schools in the Zone significantly improved student performance but that the other community programs in the Zone had no effect. http://www.brookings.edu/research/reports/2010/07/20-hcz-whitehurst

Other neighborhood effects can cause or entrench poverty. Children growing up in poor neighborhoods are more likely to ingest toxic levels of lead. http://www.cdc.gov/nceh/lead/acclpp/final_document_010412.pdf. A Dutch study finds suggestive evidence that even after controlling for the usual socioeconomic and individual factors, living in an undesirable neighborhood causes cellular changes that accelerate aging, illness, and death. Nicholas Bakalar, “Neighborhood May Age You,” New York Times, June 30, 2014, D4.

* Unfortunately, new research suggests that self-control may also accelerate the aging process, especially for the poor. “No good deed goes unpunished,” The Economist, July 18, 2015, p. 67.

* New research finds that this policy actually reduces their job prospects. Jennifer R. Doleac and Benjamin Harrison, NBER Working Paper No. 22469, July 2016.

* It is quite wrong to argue that we cannot reduce poverty without first eliminating these root causes. More commonly, as with crime control and health care, policy can only deal with symptoms unless and until it is able to understand and solve (if possible) the root causes. Indeed, root causes are themselves products of other, even more fundamental and more elusive ones. This is why insisting that we address root causes rather than symptoms, while understandable, can produce misguided policies, inaction, and lost opportunities for reform.

† Mass incarceration is an important example of this phenomenon. It both causes poverty, as discussed above, and fosters its own distinctive culture inside prison walls, which cultivates norms and behaviors that replicate and deepen poverty. See Loic Wacquant, “Deadly Symbiosis: When Ghetto and Prison Meet and Mesh,” 3 Punishment & Society 95 (2001).

* In my view, an antipoverty program’s costs should also be borne progressively through a progressive tax system. The classic debate on this is Walter J. Blum & Harry Kalven, Jr., The Uneasy Case for Progressive Taxation (1953).

* Indeed, the Food Stamp (SNAP) program could not survive without the support of more powerful agricultural interests eager to expand consumption of their products—an old political fact dramatized most recently in 2014.

* This last refers to two George W. Bush administration programs. Building Strong Families provided counseling in relationship skills, emotional support, and other social services to encourage marriage by low-income unmarried couples with or expecting a child. Supporting Healthy Marriage used counseling and other services to keep married parents of low-income families together. Unsurprisingly both failed to show any significant results; marrying the father of her child may only magnify her problems. Muhlhausen did not review findings about Head Start’s impact on adults, discussed above, because he finds these studies to lack the same scientific rigor as evaluations based on random assignment.

* The same analysis notes that most federal spending on housing (roughly $195 billion of an estimated $270 billion) subsidizes homeowners through the tax code. The nontaxation of imputed rent also subsidizes them by roughly $600 billion. The vast majority of these subsidies go to non-poor households.

* The largest category by far is religion, though it is a steadily declining share of the total. Second largest is education, with presumably most of that going to elite institutions. Human services is third, with more than 15 percent of the total. Aid to the poor is probably a goal of other categories, including health, foundations, and religion.

* Such pragmatism and opportunism are endemic to American culture, institutions, and politics. As the political scientist Nelson Polsby warned, “The complexity of the American political system may as well be directly acknowledged.… [It] stymies proposed reforms based on false analogies with simpler systems.”

* By “balanced growth,” economists generally mean growth that produces solid job and wage growth in all economic sectors at a low rate of inflation and without unsustainable asset bubbles or other distortions.

* For example, they agree that productivity has stalled since the late 1990s but disagree about why.

* The benefit side of both programs is progressive, flowing disproportionately to low-income people—so these payments, conventionally viewed as taxes, might instead be viewed as prices for a deferred benefit. On the other hand, such people tend to die much earlier, roughly offsetting this benefit.

† This would require many low-income workers who need not file tax returns now to begin doing so.

* Finland, with the agreement of all major political parties, plans to inaugurate a guaranteed income in 2016, providing every adult citizen about $900 a month with few if any strings attached—while largely eliminating its social welfare bureaucracy. In June 2016, on the other hand, 77 percent of voters in Switzerland rejected a proposal to provide a basic monthly income of 2,500 Swiss francs (US $2560) to adults and 625 francs to children under 18 regardless of employment status (http://www.nytimes.com/2016/06/06/world/europe/switzerland-swiss-vote-basic-income.html).

* This last conclusion about communities is not robust because the Chetty et al. study relied on IRS data lacking a number of potentially important uncontrolled variables. Email to author from Christopher Jencks, December 5, 2015.

* Their Brookings colleague, economist and poverty analyst Henry Aaron, is skeptical about their success sequence claims: “So, the bottom line is that a) we don’t know how to change the relevant behaviors; and b) even if we could get the people being studied to have acted differently from the way they did the results would not be as they were for the people who actually did behave differently. [Sawhill and Haskins’] response would be: ‘look we tried everything we could to control for individual differences and this is what came out; you are really saying that the sort of thing we are attempting can’t be done.’ Which is just what I am saying.’ ” Email to author dated November 8, 2015. Jencks agrees: “We’ve tried this for decades. It has helped a little, not a lot.” Written comment to author dated November 30, 2015.