Everything there was to worry about before. People who track the data about the underclass are like a physician with an overweight, chain-smoking executive who has gone on a diet and lost three pounds, cut down to five cigarettes a day, and taken a daily walk for two weeks. The patient feels great. He thinks everything has turned around. Meanwhile, the physician knows that nothing has really changed yet.

Nothing has really changed yet with the American underclass; that is the thesis to which most of this article is devoted. Why then has the underclass disappeared so completely from the national radar screen? I think the answer is simple and ignoble. What needs to be done? Maybe nothing. Maybe more than we can bear to contemplate.

Outcroppings of an Underclass

Let us first be clear on terms. Underclass is not a synonym for poor or even for disadvantaged. By underclass, I mean a population cut off from mainstream American life--not cut off from its trappings (television and consumer goods penetrate everywhere), but living a life in which the elemental building blocks of a life--productive work, family, community--exist in fragmented and corrupted forms. Most members of the underclass have low incomes, but its distinguishing characteristics are not poverty and unmet physical needs, but social disorganization, a poverty of social networks and valued roles, and a Hobbesian kind of individualism in which trust and cooperation are hard to come by and isolation is common.

Since 1989, I have been using three indicators as a concise way of tracking the underclass: criminality, dropout from the labor force among low-income young males, and illegitimacy among low-income young women.[1] The rationale for each of the three indicators is straightforward.

In the case of illegitimacy: When a large proportion of the children in a given community grows up without fathers, the next generation, especially the young males in the next generation, tend to grow up unsocialized--unready to take on the responsibilities of work and family, often criminal, often violent. The effects of absent fathers are compounded by the correlations of illegitimacy with intellectual, emotional, and financial deficits among the mothers that in turn show correlations with bad parenting practices.[2]

In the case of criminality: The habitual criminal is the classic member of an underclass, living by preying on his fellow citizens. High crime also creates a milieu, demoralizing the law-abiding elements of the community and establishing a predatory ethic that spreads beyond the criminals. Violent crime is the most directly indicative of an underclass.

In the case of dropout from the labor force among young males: One of the basic elements of the social contract is that healthy young men go to work. The economic and social institutions of mainstream society depend on it. When large numbers of young men in a community, not in school, are neither working nor looking for work, those institutions atrophy.

In any individual instance, these measures may be misleading--witness the never-married woman who is a wise and nurturing mother, the criminal who reforms, the young male who is not looking for work because he is living off his trust fund. In the aggregate, however, they are valid--there is no such thing as a thousand randomly-chosen never-married mothers, criminals, or young men out of the labor force who do not live disproportionately in neighborhoods plagued by the characteristics I ascribe to the underclass.

Four Checkpoints

The following discussion tracks each of the indicators of an underclass from mid-century through the most recently published data. When racial breakdowns are available, the graph shows the figures by race, with separate discussion of the black underclass and white underclass. I mark each graph with four checkpoints.

1954. The first checkpoint is the first year for which data on all three indicators are available, and it represents our baseline. Nineteen fifty-four stands halfway through Eisenhower’s first term, the second full year after the end of the Korean War. The poverty line had not been invented, and poverty was not part of the political dialogue. On the contrary: Americans were smugly aware of being citizens of the richest country on earth. Underclass? What’s an underclass?

1964. Lyndon Johnson has just taken office. It is the era of Michael Harrington's The Other America, the Civil Rights Act of 1964, race riots, and the War on Poverty. Nineteen sixty-four also marks the flowering of a radical critique that would set the intellectual terms of debate for two decades: America is a pervasively unjust, racist, oppressive society, as evidenced by the existence of an unacceptably large number of people who are shut out from the mainstream--the people we now call the underclass.

1982. It is the second year of the Reagan presidency and the second coming of the underclass, this time seen in a less idealistic light. The country is at the bottom of a severe recession. Reaganomics is an epithet. Homeless people seem to be everywhere, attracting a steady stream of news headlines decrying the shredding of the social safety net. Nineteen eighty-two is also the year that Ken Auletta publishes a book entitled The Underclass.[3] The word will not come into common use for another three years, but his book signals a growing agreement that a large number of Americans are not just cash-poor, but enmeshed in patterns of behavior that assure continued poverty of many kinds, economic and moral alike. The left still blames poverty and oppression, but fewer voices still argue, as they did in 1964, that more money for the poor is the all-purpose answer.

1997. The fourth checkpoint is the most recent year for which data are available on all the three indicators. Bill Clinton is beginning his second term, easily reelected in a campaign in which the issues were targeted toward the American middle class and the media paid as little attention to poverty as it did in 1954. Underclass? What underclass?

Crime Versus Criminality

Crime is a good place to start our inquiry into the course of the underclass because it so vividly illustrates how a positive trend for the nation as a whole can have quite different implications for the underclass. If we think of crime as it affects the nation as a whole, recent news has been wonderful. After seven straight years of decline, the crime rate is at its lowest in a quarter-century. Progress is more than just statistical. Almost everyone feels safer, especially in big cities. Suppose, however, we ask not how many crimes have been committed, but how many Americans demonstrate chronic criminality. That number is almost surely larger than ever. We don’t notice--because so many of the chronically criminal are in jail.

The Effects of Imprisonment. For the past twenty years, the United States has engaged in a massive effort to take criminals off the street. As of 1997, about 1.8 million people were behind bars. And while that figure includes some white collar criminals and minor offenders, they are a small part of the total. Even in the late 1990s, the American criminal justice system puts most first-time and nonviolent offenders on probation--and, for that matter, still leaves startlingly large numbers of serious offenders on the streets. The great majority of prisoners are there because they have been a menace to their fellow citizens.[4] The figure below shows how the crime rate has tracked with incarceration.

Table 1

Sources: FBI, Crime in the United States, various issues; Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics, various issues; Bureau of Justice Statistics Bulletin, Prisoners in 1997; Office of Juvenile Justice Fact Sheets on the juvenile population in public and private residential facilities.

To see how our appraisal of the crime problem depends on the massive increase in imprisonment, suppose that in 1997 we imprisoned at the same rate we did in 1980, the year that the crime rate hit its all-time high. That year, the nation had 24 prisoners in state and federal institutions per 1,000 reported index crimes. If we had still been imprisoning at that rate in 1997, we would have had a state and federal prison population of 316,200 people. The actual population in those prisons was 1,185,800—a difference of 869,600.[5]

How much crime did not occur in 1997 because those 869,600 extra people were in prison? This raises the vexing question of how much effect incarceration has on crime rates. During the 1960s and 1970s, the dominant school of academic criminologists, philosophically opposed to punishment as a solution to crime, argued that the effects of incarceration were small. They supported their position with the high recidivism rates associated with traditional imprisonment and with a research literature showing only ambiguous and comparatively small deterrence effects from imprisonment.[6]

In the 1980s and '90s, two developments changed this assessment of the effects of prison. First, rising incarceration rates gave researchers a much better environment for measuring the effects of imprisonment by offering a variety of natural experiments. Second, criminologists, disillusioned with the record on rehabilitation, began to give more attention to what is called the incapacitation effect of imprisonment. Incapacitation does not rely on complex psychological theories about why people commit crimes, but on two undeniable facts about crime and incarceration: (1) the criminal who reaches the attention of the criminal justice system tends to have committed many more crimes than his record shows, and (2) such a person may not be deterred by the prospect of jail, and he may not be rehabilitated by his imprisonment, but while he is in jail his crime rate is zero.

The evidence is now heavily weighted toward a substantial effect produced by the aggregate deterrence and incapacitation effects.[7] The most recent and most sophisticated analysis, by economist Steven Levitt, takes advantage of the natural experiment offered by prison over-crowding litigation. He finds an estimated net impact of approximately 15 total crimes prevented from adding one additional prisoner for one year.[8] Of these, approximately six would be reported crimes. Multiply six by the number of people in prison in 1997 who would have been on the streets if we had maintained the 1980 imprisonment rate and one gets 6 869,600, or 5,217,600 extra reported index offenses--in the context of the 13 million reported index offenses in 1997 that went into the calculation of the crime rate. Or we could multiply 15 869,600 and come up with 13,044,000 extra total crimes.

If these numbers seem ridiculously large, note the ways in which they are conservative estimates. They are based on 1980, when the imprisonment rate was already higher than it had been at its low point in 1975. Much more drastically, I ignored everyone incarcerated in local jails and in youth institutions in 1997 (because Levitt’s results were based on offenders in state prisons). That amounts to roughly 667,000 persons--a large fudge factor indeed.[9] If they were included, then the calculation of number of people who are behind bars now, but would not be if we were imprisoning at the 1980 rate, would produce a figure of approximately 1.3 million.[10]

In any case, the purpose of the exercise is not to estimate how many crimes would actually have occurred if we were still imprisoning at the 1980 rate (a host of technical problems put such an estimate out of reach), but to illustrate that the number must be very large. Perhaps the easiest way to judge this assertion is to ask yourself what you think would happen to the crime rate if, tomorrow, we opened to doors of the nation's correctional facilities and let out 1.3 million of their residents. My modest but hard-to-contest basic point: If the nation had not embarked on massive imprisonment, we would no longer be bragging about a falling crime rate. The only uncertainty is how high the crime rate would be. It is a major accomplishment that crime is becoming a more manageable problem for the nation as a whole. It has been achieved not by socializing the underclass, but by putting large numbers of its members behind bars.

Estimating the Increase in Criminality. The imprisonment numbers make a strong case that criminality has increased. They do not tell us the real magnitude of the increase, however, because the standards for putting a convicted felon on probation instead of sending him to prison have changed so much over the course of the last half century.

Another statistic is much more useful for this purpose: the number of adults under "correctional supervision," including not only prisoners, but persons on probation and parole. As a proxy measure of criminality, it has the disadvantage of including many persons who are not chronic criminals, but the major advantage of being unaffected by changes in sentencing practices. The main sources of change over time come from changes in clearance rates (does the criminal even get arrested?) and in prosecutorial practices (once arrested and charged, is the criminal prosecuted and convicted?). Since 1980, the first year that the Department of Justice published the "adults under correctional supervision" number, these changes have been much smaller than in the chaotic decades of the 1960s and 1970s. Since 1991, the current period of falling crime rates, they have been smaller yet.[11]

In the first year for which we have data, 1980, 1.8 million adults were under correctional supervision. At our third checkpoint in 1982, that number had already jumped to 2.2 million. This was nothing, however, in comparison to the increases that were about to come. By the fourth checkpoint in 1997, an additional three and a half million persons had been added, for a total of 5.7 million. During the recent years (1991-97) in which the crime rate has gone down by 17 percent, the number of persons under correctional supervision increased by 25 percent.

Data by race on adults under correctional supervision begin in 1985, when 1.9 million whites and 1.0 million blacks qualified, representing 1.2 percent of white adults and 5.2 percent of black adults.[12] The most recent numbers by race end in 1996. These numbers too continued to rise steadily during the 1991-96 period of falling crime rates: 576,100 more whites and 356,200 blacks. As of 1996, 2 percent of white adults and 9 percent of black adults were under correctional supervision. The proportional increase from 1991-96 was about equal for each race (21 percent for whites, 20 percent for blacks).

In summary: While the precise magnitude of the increase is open to argument, the criminality indicator strongly suggests that the size of the underclass is increasing.

Dropout from the Labor Force

The labor market is another great success story for the country as a whole, with the statistics on unemployment getting the most attention. Unemployment has fallen for five years. For practical purposes, America has had a full-employment economy for at least the last two years. The percentage of Americans in the labor market has also been increasing, led by women, reaching a record 60 percent of all Americans 16 or older in 1997.

In the context of this success, the statistics on dropout from the labor force among young males are both puzzling and one of the most striking indications that the underclass might be expanding. The following figure shows, by total male population from 1953-64 and by race from 1964-97, the percentage of young males, ages 16-24, who are not enrolled in school--and also not in the labor force. At the time of the Current Population Survey interview to which they responded, they said they were neither working nor looking for work.

Table 2

Sources: For 1953-81, Bureau of Labor Statistics (1982), Labor Force Statistics derived from the Current Population Survey: A Databook Volume I, Table C-42; For 1982-97, unpublished data provided by the Bureau of Labor Statistics.

It would be a mistake to think of these young men as staying forever outside the above-ground economy. Many go into and out of the labor market sporadically when they are young. As they grow older, almost all eventually enter the labor force. The problem is that by the time they become regular members of the labor force, they have lost the crucial years in which young men establish job skills, work habits, and a job history that enables them to find a secure niche. A man who reaches his 30s with a minimal education and a poor work history is usually consigned to the margins of the economy for the rest of his life.

Young black men. For blacks, the figure tells a fascinating story on several counts. We know from data about all blacks (not limited to those not enrolled in school) that in the 1950s and early 1960s, black males ages 16-24 had lower labor force absence than did their white peers, despite facing much higher unemployment rates. But by our second checkpoint, in 1964, 9 percent of young black males not in school were out of the labor force, almost twice the white rate. After a brief dip, this number began to rise for blacks, despite a national unemployment rate of less than 4 percent. It continued to rise irregularly into the late 1970s, when it began to move up and down around a mean of about 15 percent. As of our third checkpoint in 1982, more than 18 percent young black males who were not in school were out of the labor force.

For many years the scholarly received wisdom held that young black males were discouraged by a poor job market in the urban areas where labor force dropout was most common. If only we had a full-employment economy, we were told, they would come flocking back into the labor force.[13] Then came the 1980s, which produced tight labor markets in many urban areas. Unemployment rates of young black males in these tight labor markets dropped substantially, but the pool of those remaining outside the labor market remained little changed.[14] The advocates of the discouraged-worker explanation responded to these results by arguing that the tight labor markets had not yet lasted long enough, but clearly young black males were not responding as their theory had predicted.

If the data from the 1980s were inconsistent with the discouraged-worker theory, the data from the 1990s are downright contradictory. Beginning in 1989, despite continued low unemployment in that year, the percentage of young black males not enrolled in school who were not in the labor force began to rise again, to an all-time high--and then continued to rise, breaking through the 20 percent mark in 1992, during the recession, rising again during the boom, and standing at 23 percent at our final checkpoint in 1997, in the face of an economy that had employers taking extraordinary methods to attract low-skill workers from any source possible, including the inner city. The recent increase from 1990-97 has been proportionately equal for black teenagers ages 16-19 (up 57 percent during the period) and for young men ages 20-24 (up 56 percent).

The juxtaposition of this discussion with the preceding discussion of incarceration may have raised a question: Can the recent increase in labor force dropout be an artifact of the large increases in incarceration of young black males? The answer is no. The labor force participation rate is based on the civilian noninstitutional population.

Young white men. In comparison with the black trendline shown in the preceding figure, the white trendline looks reassuringly flat. Only 5 percent of young white males not in school were out of the labor force at the 1964 checkpoint. That figure increased in the late 1960s but then dropped in the 1970s (that bump from 1967-71 looks suspiciously like a combination of counterculture dropouts plus draftees waiting for orders to report to the army). Since 1977, young white males not in school have shown a modest secular upward trend. By 1982, the third checkpoint, 7 percent were out of the labor force. But if recent changes are much smaller in magnitude than the increases in dropout among young black males, they are proportionately worth noticing. White labor force dropout is up 25 percent from 1990-97. This is concentrated among white teenage males ages 16-19, who showed a 33 percent increase. Overall, as of the final checkpoint in 1997, almost 9 percent of the 16-24 year-old group were out of the labor force--about where the corresponding black group was in the mid-1960s.

Unemployment among young males. Is there a happier story to tell about unemployment? For blacks, not much. Black males ages 20-24 saw a big drop in the unemployment rate between 1993 and 1994 and continued reductions through 1995. Unemployment has risen again in the last two years, standing at 20 percent in 1997. The same profile holds for black males ages 16-19, whose unemployment in 1997 stood at 37 percent. These are staggeringly high unemployment rates in today’s tight labor markets.

For young white males ages 16-24 who remained in the labor market, the news has been better, with a fairly consistent decline since 1992 and a 1997 for both 16-19 and the 20-24 age groups. Their 1997 unemployment rates were 14 percent and 7 percent respectively. The net result is that the black-white unemployment discrepancy for young males has been increasing over the last four years of economic boom.

In summary: If trendlines in labor force participation among young males are a valid indicator of the course of the underclass, then it is unquestionably growing.

Illegitimacy

Table 3

Source: National Center for Health Statistics, Advance Report of Final Natality Statistics, various editions

The following figure shows, by race and ethnicity, the illegitimacy ratio--births to unmarried women as a proportion of all live births--from 1950 to 1997.

The black ratio. At the first checkpoint in the Eisenhower years, 20 percent of black children were born out of wedlock.[15] Or to put it in a more positive way, 80 percent of black children were born to married parents. When the Great Society began at our second checkpoint, 1964, the deterioration was still mild, with the percentage growing to 25 percent. Then the roof fell in on the black family. By the third checkpoint in 1982, illegitimacy had become the norm, with 58 percent of all black children born out of wedlock. By the fourth checkpoint, 1997, the black illegitimacy ratio in 1997 was 69 percent, down from its high of 70 percent in 1994. To my knowledge, to have more than two-thirds of a new generation of children born to unmarried women is unprecedented for any large subpopulation of any culture, ancient or modern.

The white ratio. The white illegitimacy ratio was vanishingly small when Eisenhower was in office--less than 2 percent in 1954. It almost doubled between then and the second checkpoint in 1964, passing 3 percent, but the large proportional growth simply reflected the small baseline. But the increases added up, and by the time the underclass was discovered at our third checkpoint in 1982, 12 percent of white children were born to unmarried mothers.

Between 1982 and the final checkpoint in 1997, the increase in white illegitimacy got serious, more than doubling again. This time, the large proportional increase could not be passed off as a function of a small baseline. As of 1997, 26 percent of white children were born to unmarried women, a figure comparable to the black ratio in the mid 1960s. To some extent this reflects a growing Latino population, mostly white, that has a higher illegitimacy ratio than non-Latino whites. But even if we restrict the calculation to non-Latino whites, 22 percent of all white births in 1997 were to unmarried women.

These numbers come at a time when the negative effects of single parenthood are no longer in serious scholarly dispute. The regnant academic position of fifteen years ago ("A single woman can raise children just as well as the two-parent family if she has a decent level of support") has all but vanished, replaced by an increasingly detailed profile of the outcomes of single-parenthood, with the never-married mother usually being the worst of the various single-parent configurations.[16] And yet recent news stories about out-of-wedlock births have had a markedly positive tone. Why?

The trendline for the illegitimacy ratio has leveled off. The national illegitimacy ratio has in fact been effectively flat for four years, and the black illegitimacy ratio has dropped since 1994--a very slight drop, but a drop nonetheless. After decades of unremitting increases, it is appropriate to see these changes as genuinely good news.

The birth rate for unmarried women is down. The last seven years have seen marked reductions in out-of-wedlock birth rates among black women and smaller reductions among white women since 1994. This is also genuinely good news--but it doesn’t necessarily have anything to do with illegitimacy as a social problem.

To understand the reasons for this seemingly self-contradictory statement, think first about why illegitimacy is a social problem as distinct from a problem affecting the lives of individual youngsters. A child born to an unmarried mother faces a variety of heightened risk factors as an individual--for poor school performance and delinquency, for example. But if that child is the only illegitimate child in the neighborhood, illegitimacy represents only a minor social problem for that neighborhood. If instead 30 percent of the children born in a neighborhood are born to unmarried women, the illegitimacy problem has gotten much worse. The sheer number of children facing higher risk factors has multiplied, obviously, and this multiplication of individual effects begins to reach the level of a social problem. But more importantly in the long run, the socialization of the next generation is also affected. To take just one of many examples, but perhaps the most important: in a community where a third of children are born without a father, young boys grow up with more than one example what it means to be a grownup male. The alternative role model is the unconnected male for whom success is defined by sexual conquests and who sees the responsibilities of parenthood as a trap for chumps. If 80 percent-90 percent of the children in a neighborhood are born to unmarried women--and this figure is not atypical for inner cities--then young boys grow up with only one visible example of what it means to be a grownup male, the bad one, and the social problem represented by illegitimacy has become a cris is.

Now consider a chronic source of confusion that afflicts almost every discussion of illegitimacy, the difference between the illegitimacy rate and the illegitimacy ratio. The mathematical distinction is this: A birth rate refers to the number of children born per 1,000 women in the group in question, while a ratio refers to the proportion of children born in a given condition to the group in question. It may not be intuitively apparent, but the two measures are arithmetically independent. If the number of illegitimate babies is represented by a, the number of women ages 14-44 is represented by b, and the total number of babies (both legitimate and illegitimate) is c, then the illegitimacy rate is computed as a/b while the ratio is computed by a/c. The rate can increase while the ratio decreases, or vice versa. The question of what actually happens is an empirical one.

Which of the two statistics is more relevant to whether illegitimacy as a social problem? I argue for the ratio. To see this, imagine a situation in which the illegitimacy rate is rising, but the birth rate to married couples is rising even faster. In such a situation, the social problems associated with illegitimacy are probably going to get smaller, not larger--the prevalence of illegitimacy will have fallen. And prevalence is what the ratio measures. If the ratio of children being born without fathers rises from 10 to 20 percent, we know that something important to the socialization of children has gotten worse, regardless of the size of each year's birth cohort. Thus, in my view, the reason to resist being comforted by a drop in the out-of-wedlock birth rate among teenagers if the illegitimacy ratio is not going down commensurately.

There is a brighter scenario to be drawn from the birth-rate data, however, prominently advocated by my AEI colleague Ben Wattenberg: One may plausibly argue that the reduction in the teenage birth rate presages a drop in the illegitimacy ratio. Insofar as out-of-wedlock births tend to decline as women get older, and insofar as the declining teenage birth rate indicates a delay in the onset of child-bearing, we may reasonably hope to see significant declines in the illegitimacy ratio in years to come.

The argument is plausible, but we must confront the fact that the black teenage birth rate has declined before without producing a drop in the illegitimacy ratio. Specifically, as shown in the following figure, the black teenage birth declined almost without a break for a quarter of a century from the late 1950s through 1983.

Table 4

Source: Advance Report of Final Natality Statistics (National Center for Health Statistics, annual)

The magnitude of the drop in the teenage birth rate was huge (173 births in 1957 to 96 in 1983) compared to the 1990s drop (116 to 90). And yet during that same period the black illegitimacy ratio (superimposed over the black teenage birth rate in the figure) increased without a break from 21 percent to 59 percent. Maybe things will be different this time, but past experience should temper our expectations.

Everybody does it. A familiar argument holds that illegitimacy doesn’t mean what it used to mean. Births to single women now occur throughout society, and therefore the occurrence of illegitimate births no longer reflects behavior within an underclass, or even the lower classes.

This argument draws on a kernel of truth: In proportional terms, illegitimacy has been growing among the affluent and well-educated. For example, the number of never-married women with a college degree who have had a baby was 68 per 1,000 in 1995 (the most recent published breakdown by education). This represents almost triple the rate in 1982 , 28 per 1,000, and presumably reflects a destigmatization of illegitimacy.[17] But of the 849,000 never-married women who had a child in 1995, only two percent were women with a college degree. Ninety-three percent of them were women whose terminal degree was no higher than a high school diploma. Illegitimacy in the United States remains overwhelmingly concentrated among women in the bottom rungs of the social and economic ladder.

Welfare rolls are down. In the public mind, the size of the illegitimacy problem has long been seen as intertwined with the size of the welfare problem. Welfare rolls are now plunging. Therefore the illegitimacy problem must be plunging.

Before I don my pessimistic hat, let me pause for a moment to acknowledge the phenomenal change that has taken place. The following figure shows the history of welfare rolls from 1954 onward, expressing welfare families as a percentage of all American families.

Rarely does a social trendline show such a steep and prolonged increase as we saw in the AFDC caseload from 1966 to 1972, and just as rarely do we see a drop as steep as the one since 1993. And the economy gets only a modest part of the credit--welfare either soared (in the 1960s) or declined modestly (in the 1980s) during the two preceding economic booms. The declines we are observing now are real and large even after taking the economy into account.

What does the large drop in welfare rolls mean for the underclass? The answer depends partly on who is leaving the rolls. If large numbers of chronically dependent women are among them, then a strong argument can be made that life is going to change in communities where the underclass is most concentrated. If the decline is concentrated among women who would be on welfare for short periods of time anyway, then the underclass is going to be largely unaffected.

Table 5

Souce: Bureau of the Census, Statistical Abstract of the United States, various editions

Information on this issue is still fragmentary. Racially, the reductions in the rolls through the most recent published figures (June, 1997), were about the same--down 38 percent for blacks, 33 percent for whites--but the published data cast no light on the length of time that people leaving the rolls had been on welfare. Using more recent unpublished data, the New York Times reported in July, 1998, that whites are leaving the rolls faster than blacks, and that reductions tend to be smaller in large cities than elsewhere. These trends imply that the reductions in caseloads are occurring disproportionately among women who wouldn’t have spent much time in the system anyway and are not part of the underclass.[18] For now, we can say only that some chronic welfare recipients must have left the rolls, but we still know very little about how big that number is.

The more profound question is what difference it makes if single mothers go to work. Is a community without fathers importantly different just because more mothers are earning a paycheck? One logic says yes. Jobs provide regularity, structure, and dignity to family life, even if the father is not around. But we know from recent research that the bad effects of single-parenting persist for women not on welfare. No counterbalancing body of research demonstrates that it is good for children when a single mother works (rather the opposite). I like to think children who see their mothers working for a living grow up better equipped to make their way in the world than children who watch their mothers live off a welfare check, even if there are no fathers in their lives. But that is a hope, not a finding.

While we're waiting to find out, illegitimacy ratios are at near-record levels. On the illegitimacy indicator, we can hope that we have seen the peak of the illegitimacy ratios, and even hope that the ratio will decline in the future. But whatever bad things go with high illegitimacy rates are still going on, and will continue to have reverberating effects well into the next century as the next generation grows up.

Searching for Better News

I have focused on just three topics for assessing the underclass. A reasonable question is whether other indicators might tell a more optimistic story. I will quickly review a few of the obvious ones, comparing the situation in 1982 with the most recent data.

Poverty. Poverty in 1997 was 13 percent of Americans compared to 15 percent in 1982, a modest improvement. The improvement among blacks has been more sizable, with the poverty rate down from 36 percent in 1982 to 27 percent in 1997.[19]

The difficulty with interpreting these statistics is that cash poverty has no necessary relationship with the size of an underclass. In 1959, the first year for which the official poverty line was calculated, a devastating 55 percent of blacks were below the poverty line--and yet in 1959, almost 80 percent of black children were born to a married couple, more than 90 percent of young black males were in the labor force, and crime in the black community was much lower than today and holding steady. Black poverty in 1959 was more than twice as high as in 1997 but the size of the black underclass was only a fraction of its current size. It is a good news that poverty among blacks is down, and it is bad news that the figure still stands as high as 27 percent of the population. Neither datum tells us much about the prospects for the black underclass.

For whites, poverty stood at 11 percent in 1997 compared to 12 percent in 1982. The current figure is higher than it was in the late 1980s or all of the 1970s, and exactly equal to the level of white poverty observed in 1967, more than three decades ago.

Drugs. The good news about drugs involves marijuana and powder cocaine, for which use dropped sharply between 1982 and 1992 (their use has risen modestly since then). But the drugs most closely associated with the black underclass are crack cocaine and heroin. The drug most closely associated with the white underclass is methamphetamines.

Crack did not even exist in 1982. By the last half of the 1980s the black community was suffering from what was commonly called a crack epidemic. Recently, many urban newspapers have reported evidence that the crack epidemic has subsided, but these reductions are not yet confirmed by the nation’s most authoritative source for such data, the National Household Survey on Drug Abuse (NHSDA). Crack was first included in the survey in 1988, when use within the previous year was estimated at 0.7 percent of the population 12 years and over.[20] As of 1997, it was 0.6 percent, a statistically insignificant difference.[21] White crack use has increased substantially. According to the NHSDA figures--and surprisingly, given crack's reputation as a black drug--past-year use is now about three times more common among whites than among blacks.

The number of people who had used heroin during the course of the last year was put at 0.2 percent of the population in 1982, where it remained until 1992, when it dropped to 0.1 percent. Paradoxically, it was in that same year that new users began to rise, a trend that the NHSDA now concludes is statistically significant through 1995. New-user data are not yet reported for 1997. We do know from the 1997 survey that total past-year heroin use rose from an estimated 455,000 in 1996 to 597,000 in 1997, representing the first time since the survey began in 1979 that the percentage figure has been as high as 0.3 percent.

Stimulant use was not reported as of 1982. As of 1990, 1.2 percent of respondents reported past-year use. This figure had been dropping since 1985, and continued to drop through 1994. Since 1994, it has moved in a narrow range, from 0.7 percent-0.9 percent.

In summary: One of the drugs most commonly associated with the black underclass (crack) didn’t even exist in 1982 and in 1997 remains a significant problem, with the possibility, as yet unconfirmed in the survey data, that crack use among blacks has been falling. Use of the other (heroin) has risen significantly. Use of the drug most commonly associated with the white underclass has fallen, while white use of crack has increased.

Low birth weight babies and infant mortality. I pair these indicators because pairing them so vividly illustrates the nature of the problem represented by the underclass. From 1982-97, infant mortality dropped by 41 percent for whites and 30 percent for blacks--major progress. During the same period, the incidence of low-birth-weight babies (under 2,500 grams) rose by 16 percent for whites and 5 percent for blacks. The incidence of very-low-birth-weight babies (under 1,500 grams) rose by 22 percent for whites and 38 percent for blacks.[22] Low birth weight is associated with a variety of problems in later life, ranging from emotional development to cognitive ability. For very-low-birth- weight babies, the risk of subsequent problems jumps substantially, which makes the large increase in very-low-birth weight babies especially significant.

Infant mortality is determined primarily by medical technology and the provision of medical care. On that front, the nation is doing dramatically better. Low birth weight is determined primarily by how well a pregnant woman takes care of herself. Money and sophistication is not required. Rather, the woman must eat nourishing food and refrain from destructive habits such as smoking, heavy drinking, and drugs. That low-birth-weight babies are increasing in an environment of constantly improving medical care (as evidenced by the infant mortality statistics) suggests a significant increase in the proportion of mothers who are getting pregnant and failing to take even rudimentary care of themselves during pregnancy.

Trying to summarize all of these indicators is a problem because they are such a mixed bag. It is particularly difficult to know what to make of the drug data--the journalistic accounts of improvements in the inner city are much more optimistic than the survey data would give reason to conclude. Economically, underclass neighborhoods are probably somewhat more prosperous than they were during the recession of 1991-92. Behaviorally, it is not at all clear that much has changed. The low-birth-weight figures give real cause for concern that things have gotten worse. I can at least assert that no major, face-valid indicators that I have ignored give us reason to think that the bad news about criminality, dropout from the labor force, and illegitimacy is wrong.

Did We Give Up on the Underclass?

In thinking about what ought to be done about the underclass, it helps to have a perspective on what has been tried already. The first step in this process is to discard a script that has wide acceptance, especially on the left. According to this story line, of course we have an underclass because the government quit trying to help. The nation was energized for a brief period of time during the 1960s. We made progress. Then Nixon was elected and put a stop to the War on Poverty. After Nixon came the economic stagnation of the 1970s. Then came Ronald Reagan, who slashed spending on social programs. George Bush didn’t do much one way or the other, and Bill Clinton’s desire for a more activist policy has been frustrated by an uncaring national mood. If asked to draw a trendline showing real government spending on the poor, it seems that many people would draw one that shows the highest spending in the 1960s, the lowest spending in the 1980s, and something in the middle during the 1990s.

The actual trendline is shown in the following figure, which shows federal spending on programs that the Census Bureau labels as "cash and noncash benefits for persons with limited income," using constant 1995 dollars, as reported in a standard table in the Statistical Abstract of the United States for years since 1978 and reconstructed from budget data for the years prior to that. The programs represented in the total include medical services for low-income groups (mostly Medicaid), cash assistance (mostly AFDC, Supplemental Security Income, and the Earned Income Tax Credit), food benefits (mostly food stamps), housing benefits, Head Start, Pell educational grants, social services, jobs and training programs, and energy assistance. Note that the figure does not include two huge expenditures that have grown most rapidly over the last few decades, the portions of the Social Security and Medicare budgets that go to low-income people.

Table 6

Source: Bureau of the Census, Statistical Abstract of the United States, various editions.

When the reform era began under Lyndon Johnson in 1964, the rhetoric was fervent but the programs were brand-new and small in size. In 1964, total federal expenditures on benefits for persons with limited income amounted only to $31 billion (as with all other dollar figures, in constant 1995 dollars). Spending increased substantially during Johnson's tenure, and then still more rapidly midway through Nixon’s first term. The trendline continued to rise steeply through 1977, then flattened as rising inflation counterbalanced increases in nominal budgets. When the underclass was becoming a major issue in 1982 under Ronald Reagan, expenditures had reached $140 billion, quadruple the real expenditures of 1964.

As inflation moderated, increases in nominal budgets once again resulted in increases in real spending. Reagan slowed the rate of increase in spending on the low-income population but, popular wisdom notwithstanding, did not cut such spending. Then comes one of the least-told stories in recent social policy: Under George Bush, real spending on low-income Americans rose more steeply than under any other president. The steep rise continued under Bill Clinton through 1995. In 1996, the third year in which welfare rolls had declined, expenditures on low-income persons finally began to fall--by 1.6 percent.

In 1996, even after the reduction from 1995, expenditures on cash and noncash benefits for persons with limited income definition amounted to $254 billion, more than eight times the amount spent in 1964 and, perhaps even more startling, an increase of two-thirds during 1982-96, a period supposedly was dominated by conservative, no-nonsense policies toward welfare.

Perhaps the increase can be explained merely by the inflated costs of medical care? Some, but not all. Even after excluding medical care expenditures, spending on programs for the low-income population increased by more than five times from 1964-94 and by more than 50 percent from 1982-96.

Perhaps the increase an be explained by an increase in the number of people who need help? No; even when calculated on a dollars-per-person basis, spending increased dramatically. Suppose we take as the numerator total federal spending on the officially defined "cash and noncash benefits for persons with limited income" and use as the denominator the officially defined number of people in the United States who would fall below the poverty line if they received no government income transfers at all. As always, figures are in constant 1995 dollars. In 1964, the federal government spent $726 for each such person. By 1982, that figure had risen to $2,536. By 1996, it was $4,422--sums that, recall, do not count the Social Security and Medicare benefits that go to low-income persons. The United States under Bill Clinton spends much more on each low-income person than Lyndon Johnson or Jimmy Carter could have hoped to spend.

Why the Indifference?

In sum, we have a larger underclass than ever before and are spending more money on it than ever before. And yet the underclass is no longer a political issue. Why?

Part of the explanation is benign: People tend to feel good when things seem to be going in the right direction and bad when they seem to be going in the wrong direction, regardless of the current state of affairs. Crime has indeed fallen, even if the number of criminals has grown. The black illegitimacy ratio has indeed leveled off, even if it is at 69 percent. It is obvious that jobs are available for anyone who wants to work, even if more than a third of young black males do not have jobs. It is natural to feel less urgency about the problem if the prospects for improvement seem good.

Part of the explanation is conflation. Much of the good news we see in the national indicators applies primarily to people who were not in the underclass to begin with. But good news for the nation is taken as being equally good news for all strata of society.

I propose as the rest of the explanation that most Americans became concerned about the underclass primarily because it intruded on their lives. Mainstream America took measures to deal with that intrusion, and succeeded. The underclass is out of mind because it is now out of sight.

To see what I mean, consider the ways in which the underclass directly affected the lives of the "mainstream America"--or, more simply, "we." There were essentially four:

Busing sent children of the black underclass into our schools. It is a truth that the media could seldom bring themselves to say: White parents were much less upset that their children were forced to go to school with black children than that so many black children from inner-city neighborhoods behaved in ways that middle-class parents, white and black alike, found unacceptable. Associating with the children of the underclass was bad for our children, and we reacted as worried parents always react, determined to do whatever is necessary to protect our own children and to hell with high-flown sentiments about the greater social good.

The homeless physically invaded our public spaces. It is no accident that the emergence of the homeless in the early 1980s coincided with the coining of the term underclass. The population of such people had actually been growing throughout the 1970s, for reasons that involved everything from the deinstitutionalization of the mentally ill to increased drug use, but it was not until the 1980s that the people once called bums and vagrants began to understand the new rules: "Don’t worry about vagrancy laws; the police aren’t going to roust you. You may lodge in a doorway on the fanciest street in town, and no one will do anything about it." And so the homeless invaded our public spaces--and while some of us felt guilty about it, almost all of us hated the intrusion.

Public order deteriorated . I refer here not to the homeless or to crime, though both of those represent a deterioration of public order, but to an in-between category. Graffiti is the classic example. It seems a trivial thing in itself but recall how omnipresent it used to be, covering subway cars, urban buildings, highway underpasses, billboards, bus shelters, and the pavilion in the park. It was ugly and scary. The squeegee men were another form of this deterioration, as were knots of menacing teenagers and prostitutes working the streets in what were otherwise "nice" parts of town. These were the broken windows of public order, to evoke the image made famous by James Q. Wilson and George Kelling: small things in themselves but tokens that in aggregate said that the forces of order and civility were no longer in control.[23]

Crime made fear a chronic part of urban life. The living symbol of the underclass, even more than the homeless, was the black urban criminal. If we lived in a major city, our daily movements were as carefully considered as patrols into enemy territory. We followed elaborate preventive strategies. And most of us nonetheless experienced some sort of criminal victimization at first-hand. We felt afraid and besieged.

It took us the better part of three decades, but as of 1999, we have dealt with the intrusions of the underclass. Some of us have dealt with them by moving to safe havens in suburbs and small towns. Others of us have remained in the city but created urban enclaves isolated by concierges, private security guards, private schools, private playgrounds, and various tacit forms of territorial demarcation that staked out our turf.

Public policy also dealt with the intrusions of the underclass--to different degrees in different cities, but effectively overall. Busing is so much a part of our past that the very word has an archaic ring to it. Graffiti yielded to miracle substances that make surfaces hard to spray and easy to clean and to the insight that graffiti artists lose heart if their work disappears the next day. The homeless are not altogether gone, but their numbers have diminished drastically in most cities via a combination of revamped vagrancy laws and homeless shelters. The squeegee men and menacing teenagers--sometimes even the prostitutes--have retreated in the face of zero-tolerance policing.

Most importantly of all, we figured out how to deal with the crime problem. Target hardening helped--in the last two decades it has become more difficult to steal everything from a book to a car. Private security forces also helped, as did the creation of the urban islands in which someone who doesn’t belong is under instant surveillance. Innovations in policing helped. But the key insight was a very old one: Lock 'em up, as we have done in unprecedented numbers. The net result: Not only is crime rarer, it is most conspicuously rarer in those parts of town where we are likely to spend time.

Why is the underclass no longer an issue? It wasn’t the existence of an underclass that bothered us, but that the underclass was in our face. Now it is not--for the most part, and for the time being, so we can forget about it.

How Long Can We Manage the Underclass?

"For the time being" is the crucial hedge. Over the long term, can the United States retain its political and social culture in the presence of a permanent underclass?

The answer is certainly yes if an underclass is sufficiently small. Every civilization has had a fragment of the population that fits the definition of underclass. As long as it is only a fragment, the disorganization and violence of its culture do not spill over into the mainstream. When the underclass is sufficiently small, it is disreputable. Nobody in mainstream society wants to be like them. Few in mainstream society have any qualms about doing whatever is necessary to keep them in their place.

As the underclass grows, mainstream society can continue to isolate and control its direct effects on the rest of the population, given a few conditions. The first condition is the willingness to put large numbers of people in prison, a test the United States has met. The second condition is affluence. The United States is now so rich that the costs of supporting and incarcerating the underclass are easily borne. Earlier, I pointed out how rapidly spending on social programs for the poor have increased. But as a percentage of GNP, the recent increases in spending still leave us about where we were in the late 1970s. Meanwhile, the huge increases in GNP since the early 1980s mean that the raw number of dollars available to spend on competing priorities has also increased. The United States is rich enough to support a substantial underclass without feeling the pinch. This leaves two questions without clear answers.

Cultural Spillover

First question: How much has the culture of the underclass already spilled over into the mainstream? So far, the American underclass has been predominantly urban and black. Urban black culture has been spilling over into mainstream American culture for more than a century now, historically to its great advantage. Urban black culture continues to spill over as much as ever, but during the last three decades it has increasingly been infiltrated by an underclass subculture that celebrates a bastardized code duello, predatory sex, and "getting paid." The violence and misogyny that pervade certain forms of popular music are coordinate with these values. So is the hooker look in fashion and the flaunting of obscenity and vulgarity in comedy. Perhaps most disturbing is the widening expression, often approving, of underclass ethics: Take what you want. Respond violently to anyone who antagonizes you. Despise courtesy as weakness. Take pride in cheating (stealing, lying, exploiting) successfully. I do not know how to measure how broadly such principles have spread, but that they are more openly espoused in television, films, and recordings than they used to be is hard to deny. I am suggesting that among the many complicated explanations for this deterioration, cultural spillover from the underclass is implicated.

Implicated--that’s all. There are many culprits behind the coarsening of American life. It should also go without saying that vulgarity and violence were part of mainstream America before the underclass came along. But these things always used to be universally condemned in public discourse. Now they are not. It is not just that America has been defining deviancy down, slackening old moral codes. Inner-city street life has provided an alternative code and it is attracting converts.

The White Underclass

The converts are mainly adolescents, which makes sense. The street ethics of the underclass subculture are not "black." They are the ethics of male adolescents who haven't been taught any better. For that matter, the problem of the underclass itself is, ultimately, a problem of adolescents who haven’t been taught any better. There are a lot more white adolescents than black ones, leading to the second unanswered question: How fast will the white underclass grow?

National statistics tell us that in the last decade white criminality has not only increased but gotten more violent, that white teenage males are increasingly dropping out of the labor force, and that white illegitimacy has increased rapidly. Anecdotal evidence about changes in white working-class neighborhoods points to increased drug use, worse school performance, and a breakdown of neighborhood norms that recall accounts of three decades ago about decline in black working-class neighborhoods. Systematic documentation of these trends is still lacking.

Looking ahead, much depends on whether illegitimacy among whites has already reached a critical mass. The good news is that the growth in the white illegitimacy ratio has slowed. The bad news is that it stands at 26 percent--22 percent for non-Latino whites--which, judging from the black experience in the early 1960s, may be near that point of critical mass. No one knows whether the subsequent trajectory of events for whites will be the same as it was for blacks. If it is, we can expect accelerated and sustained growth in white crime, labor force dropout, and illegitimacy by the middle of the next decade. European countries with high white illegitimacy ratios certainly offer no comfort. Juvenile crime is increasing rapidly across Europe, along with other indicators of social deterioration in low-income groups.

The Question We Prefer Not to Ask

I have deliberately kept the discussion focused on the underclass as it affects the rest of society because that seems to be the only thing that matters politically. Let me at least make the obvious, omitted point: to be a member of what we call the underclass is an awful way to live a human life. It is awful for reasons having nothing to do food, shelter, clothing, medical care, or any of the other commodities that activist government knows how to provide. I have refrained from making that point because it is not, apparently, salient. We have successfully assuaged our guilt about the underclass by spending money on the commodities. Providing the other requisites of a satisfying life--family, neighborhood, safety and civility, productive work--is too tough. Thinking about how these good things come about would require us to think about how it is that families form and become nurturing centers of life; how neighborhoods form and find willing recruits to perform the functions of neighborhoods; how children grow up understanding the necessity and the dignity of productive work. Thinking about those questions honestly, without self-indulgence, would require us to consider the possibility that a large underclass exists in the midst of the affluence and opportunities of American society only because of destructive public policies that encourage its existence.

Compartmentalized America

But we quite successfully avoid thinking about those questions at all. The most striking aspect of the current situation, and one which makes predictions very dicey, is the degree to which America is culturally compartmentalizing itself. America in 1999 is a place where the local cineplex may play Sense and Sensibility next door to Natural Born Killers. Brian Lamb is a few channels away from Jerry Springer. Formal balls are in vogue in some circles; mosh pits in others. Name just about any aspect of American life, and a case can be made that the country is going in different directions simultaneously, some of them Jamesian, others Hogarthian.

The Jamesian elements are not confined to a cultured remnant. Broad swathes of American society are becoming more civil and less vulgar, more responsible and less self-indulgent. The good news is truly good, and extends beyond the qualities measured by statistics. The bad news is, perhaps, manageable. One way to interpret the nation's success in reestablishing public order is that we have learned how to cope with the current underclass. One may then argue that the size of the underclass is stabilizing, meaning that we can keep this up indefinitely. It requires only that we set aside moral considerations and accept that the huge growth of the underclass since 1960 cannot now be reversed.

Welfare reform and the growing school voucher movement are heartening signs that many are not ready to accept the status quo. But they struggle against a larger movement toward what I have in the past called "custodial democracy," whereby the mainstream subsidizes but also walls off the underclass.[24] In effect, custodial democracy takes as its premise that a substantial portion of the population cannot be expected to function as citizens.

At this moment, elated by falling crime rates and shrinking welfare rolls, we haven’t had to acknowledge how far we have already traveled along the road to custodial democracy. I assume that the next recession will disabuse us. But suppose that our new modus vivendi keeps working? We just increase the number of homeless shelters, restore the welfare guarantee, build more prison cells, and life for the rest of us goes on, pleasantly. At some point we will be unable to avoid recognizing that custodial democracy has arrived. This will mark a fundamental change in how we conceive of America. Will anyone mind?

Notes

[1] The three indicators and their rationales were first described in an article about the British underclass published by the London Sunday Times. An expanded version published in America is C. Murray (1990), "The British underc