THE PUBLIC INTEREST
Number 137, Fall 1999
"Reducing poverty, not inequality"
According to official statistics, the distribution of income has become increasingly unequal during the past two decades. A common reaction in the popular press, in political debate, and in academic discussions is to regard the increase in inequality as a problem that demands new redistributive policies. I disagree. I believe that inequality as such is not a problem and that it would be wrong to design policies to reduce it. What policy should address is not inequality but poverty.
The difference is not just semantics. It is about how we should think about the rise in incomes at the upper end of the income distribution. Imagine the following: Later today, a small magic bird appears and gives each Public Interest subscriber $1,000. We would all think that this is a good thing. And yet, since Public Interest subscribers undoubtedly have above average incomes, that would also increase inequality in the nation. I think it would be wrong to consider those $1,000 windfalls morally suspect.
Pareto principle vs. Gini coefficient
When professional economists think about economic policies, they generally start with the principle that a change is good if it makes someone better off without in making anyone else worse off. That idea, first suggested by the Italian economist Vilfredo Pareto, is referred to as the Pareto principle. I find it hard to see how one could disagree with such a principle, which is why it is the widely accepted foundation for the evaluation of economic Policies
Not all policies can be evaluated in reference to the Pareto principle. There are policies that make some people better off while making others worse off. The desirability of such a policy depends on how much the gainers gain, how much the losers lose, and the initial income and circumstances of the individuals involved. But that difficult evaluation is not my concern here. I am interested only in evaluating changes that increase the incomes of high--income individuals without decreasing the incomes of others. Such a change clearly satisfies the common-sense Pareto principle: It is good because it makes some people better off without making anyone else worse off. I think such a change should be regarded as good even though it increases inequality.
Not everyone will agree with me. Some see inequality as so intolerable that they regard increasing the income of the wealthy as a 'bad thing," even if that increased income does not come at anyone elses expense. Such an individual, whom I won describe as a 'spiteful egalitarian," might try to reconcile this with the Pareto principle by saving, " It makes me worse off to see the rich getting richer. So if a rich man gets $1000, he is better off and I am worse off. I dont have fewer material goods, but I have the extra pain of living in a more unequal world." I reject such arguments and stick to the basic interpretation of the Pareto principle that if the material well-being of some individuals increases with no decrease in the material well-being of others, that is a good thing even if it implies an increase in measured inequality.
I would note that one can reject spiteful egalitarianism and still favor redistributive policies and tax progressivity. Such redistributive policies reflect an assumption that the social value of incremental income (in economic terminology, the social marginal utility of income) declines as income rises - i.e., that an extra $100 of income means less to a millionaire than to someone whose income is $10,000. Of course, many economists reject such comparisons on the grounds that there is no way to compare how much pleasure two different individuals get from money or from the goods that money buys. But analyses that conclude that all increases in inequality are bad imply something much stronger: that the social value of incremental income to a rich person is actually negative.
More formally, economists and other policy analysts often use the "Gini Coefficient" as a measure of income inequality. (1) The Gini Coefficient measures the concentration of incomes in the nation, with a higher Gini Coefficient value implying more concentration. A feature of the Gini Coefficient is that an increase in the incomes of the rich with no change in the incomes of others will raise the Gini Coefficient. The common procedure of regarding a higher Gini Coefficient as a deterioration of the national condition is equivalent to treating the social marginal utility of high incomes as negative - i.e., that something bad has occurred when the well-to-do become better off.
In rejecting the criticism of inequality per se, and in asserting that higher incomes of the well off are a good thing, I am not referring to the functional arguments that some have offered in defense of inequality. Such analysts argue that an unequal distribution of income may contribute to general economic growth, and therefore to the poors standard of living, by increasing the national saving rate. Alternatively, they contend that inequality is a reflection of Schumpeterian innovation, which eventually helps most, if not all, individuals in the economy. I am also not defending high incomes because the affluent support charitable causes or "high culture." All of this may be true, and even convincing to someone who doesnt care about the well-being of the wealthy or who gives negative weight to their increased well-being. But I am not relying on such arguments here, because I want to stress that there is nothing wrong with an increase in the well-being of the wealthy or with an increase in inequality that results from a rise in high incomes.
The rich get richer
There has been a relatively greater increase in higher incomes in recent years in the United States and in some other countries. Some part of the rise in top incomes reflects the fact that the cut in top marginal tax rates in 1986 caused high-income taxpayers to shift the form of their compensation to taxable cash from fringe benefits and other nonobservable forms of compensation. But there have also been real increases in the pretax incomes at the top. It is important to understand why.
The increase in higher incomes has been the result of four principal factors. First, there are now more individuals with advanced education and enhanced marketable skills, and market forces reward these high skills relatively more than they did in the past. Thus individuals have a strong incentive to acquire these skills and to select occupations in which such skills are rewarded. Second, entrepreneurial activities are on the rise. The creation and growth of new businesses has been an important source of the larger number of individuals with high incomes and significant wealth.
Third, high-wage individuals work increasingly long hours. We all know about investment bankers, lawyers, and other highly paid professionals who are now working 70 or more hours a week, twice the weekly hours of a typical employee. Dora Costa, an economic historian at MIT, has recently reported that this observation is part of a more general trend toward longer working hours for higher paid employees, a reversal of the earlier tendency of those with lower wages to work longer hours. The result: measured inequality has increased.
Finally, declines in the cost of capital, reflecting an improved fiscal outlook and perhaps a decrease in perceived financial risk as a result of lower inflation, translate into higher stock and bond prices, an additional source of increased wealth for those with higher incomes. Each of these four sources of higher incomes for those at the upper end of the distribution is, I would argue, a good thing in itself. They add to the income or wealth of those individuals without reducing the incomes and wealth of others.
The real problem on which national policy should focus then is not inequality but poverty. I have in mind the incomes of those in the bottom decile or quintile of the income distribution. After discussing the problems of measuring poverty, I will consider three possible sources of poverty - unemployment, a lack of earnings ability, and individual choice - and what can be done about them.
Of course, measuring the incomes of the lowest income group is not a simple task. Cash income overestimates the number of the poor. A broader measure that includes in-kind benefits like health care and housing suggests much less poverty. There is also a problem in classifying someone as poor if his income is only temporarily low.
More generally, sociologists who have actually studied the poor directly and spoken with them about their living conditions (a research method that economists use too little) have been puzzled by how the poor could live on so little income. Those who have gained the confidence of the poor discover the answer: the underground economy. The true incomes of many of those with very low measured incomes are actually higher than the data indicate. Such individuals do not report their total income since doing so might reduce their eligibility for cash and in-kind transfers.
This is a major problem for studies of the incomes of the poor. Careful studies of income distribution are most reliable when they look at the wage distribution of the middle classes, an unfortunate fact since the most interesting questions are about the very poor and the very rich, for whom data are simply not very good.
A separate issue that plagues attempts to measure trends in poverty and in income levels more generally is the difficulty of measuring changes in the cost of living. A growing body of research suggests that the consumer price index (CPU and related official measures overstate the rise in the true cost of living and, therefore, understate the rise in real personal incomes. Even if the bias in the CPI is as little as 1 percent a year, the cumulative effect over two decades is to understate the growth of real incomes by more than 20 percent.
These measurement difficulties should make us cautious about attempting to assess changes in the extent of poverty over time. Nevertheless, poverty today is a real and serious problem in the United States and other countries. I will thus consider three sources of poverty and the policies that might be directed to counter them.
Unemployment and poverty
There exists a small, but serious, amount of very long-term unemployment in the United States that creates poverty and hardship. Its extent goes beyond the measured amount of long-term unemployment since most individuals who have been out of work for considerable periods of time in the United States are classified as "not in the labor force" rather than unemployed. But, although this long-term nonemployment is a problem and a source of poverty, it is not a cyclical problem that is amenable to expansionary monetary or fiscal policy.
Current long-term unemployment is very different from the unemployment of the Great Depression when a large fraction of the labor force was unemployed and out of work for a year or longer. The current long-term unemployment is also very different from the cyclical unemployment that we see now in the United States. Most cyclical unemployment spells are short, ending in less than 10 weeks. During such spells of unemployment, the decline in consumption is very small. Unemployment insurance generally replaces more than half of the lost net income of those who receive benefits, and the earnings of second earners in the household of the unemployed help to stabilize total household income. While the unemployed may not have access to formal lines of credit, they are often able to defer payments during part or all of their unemployment spells.
The situation is, of course, different in Europe where unemployment rates tend to rise during recessions but not to fall during a recovery. Cyclical unemployment there becomes long-term unemployment because of the adverse incentives in the European system of unemployment benefits and welfare payments.
Reform of the American unemployment system in the 1980's led to a decline in the rate of unemployment. One important aspect of those reforms was subjecting unemployment benefits to the personal income tax, a reform that obviously did not affect the poor (who do not pay income tax). However, this measure did reduce the very high replacement rates that previously made it possible for some individuals in higher-income households to have more net income by being unemployed than by working.
Lack of earning ability
The most commonly recognized reason for poverty in the United States is the inability of poor individuals to earn more than a very low hourly wage. This low earnings ability, is often attributed to inadequate schooling or training.
It is clear that inadequate schooling can limit an individuals earning ability and that the obvious remedy is more or better schooling. Many economists and educators who are studying how to improve our educational system have concluded that decentralization and competition are essential. Research by Larry Katz and Claudia Goldin of Harvard shows that the historic spread of high-school education and vocational education in the United States reflected decisions of local governments rather than the actions of the states or federal government. Research by Caroline Hoxby and others shows that the quality of local public education today improves (as measured by graduation rates, continued education, post-school wages, etc.) where competition flourishes due to a larger number of school districts or a greater availability of nonpublic (typically parochial) education. The importance of competition has increased interest in vouchers to promote individual choice.
A second reason for low earnings ability is inadequate training. Experience suggests that on-the-job training is best. The German system of formalized apprenticeships appears to allow Germany to escape the high youth-unemployment rates that plague much of Europe; the system may also reduce poverty in later years. In the United States, in contrast, minimum-wage legislation limits the ability of individuals with low skills, low education, and low ability to obtain on-the-job training. Although someone who comes to a job with good ability and
skills can both earn the minimum wage or more and also obtain additional skills through on-the-job training, an employer cannot afford to pax the minimum wage and provide training to those with the lowest skills.
The evidence on government-sponsored training programs for the middle-aged unemployed is very discouraging. For women, participation in training programs raises employment and wages by more than the cost of the training, but the impact on employment and poverty for the trainees is very small. For men, the results are even worse: The gains from training are less than the costs.
The problem of low human capital as a source of poverty is not just a matter of schooling and training, but also low cognitive ability. As I read the evidence, while variations in cognitive ability (IQ) close to the mean score do not have much impact on individuals wage rates, individuals with extremely low levels of cognitive ability (IQ levels below 80) have a very hard time earning a decent wage rate. This is not a fashionable view. Americans like to think that all men and women are quite literally created equal and that education can therefore solve the problem of low human capital and low earnings. Unfortunately, however, very low cognitive ability is likely to be a serious cause of poverty that cannot be remedied by education and training. Only when this is accepted will it be possible to develop appropriate new policies.
Finally, there are those for whom low earnings ability reflects pathologically dysfunctional life styles - drug abuse, alcoholism, and mental illness. Policies that deal with these specific problems, if they are successful, will do much to reduce human suffering as well as to alleviate poverty.
The role of individual choice
Not all poverty can be attributed to involuntary unemployment or to the lack of earning ability. Individual choices, rational or irrational, can lead to poverty. Some individuals who are in poverty may be making considered choices. For example, some individuals may choose leisure (not working or working very little) over cash income even though this leaves them poorer than they otherwise would be. Choosing not to work may be an increasingly important source of poverty. Over time the standard of living that is possible without working has increased for some segments of the population as a result of the rise in the real value of cash and in-kind welfare benefits. Often the real value of these welfare benefits has increased more rapidly than the real value of wages available to low-skilled workers, increasing the likelihood that the appeal of such benefits would exceed the attractiveness of work. This is reinforced to the extent that transfer rules reduce the incentive to work. Reducing such voluntary poverty requires reexamining the structure of welfare programs.
Not all individual choice is properly described as "considered" or " rational," and some individuals may choose poverty in error. In other words, they may think that they are making a rational decision (what economists would call a "utility maximizing" decision) when in fact their facts are wrong. Some of those individuals may think that they will not like work (or the combination or work and the money that it brings) as much as they currently like staying at home but would discover the opposite if they went to work. Moreover, these individuals may not recognize that they will advance in their jobs, shifting over time to more appealing work or at least to higher incomes. A policy of "tough love" that forces such individuals to enter the world of work for an extended period of time may be the best way to overcome this problem.
1. See, for example, the papers discussed at the Federal Reserve conference, "Income Inequality: Issues and Policy Options" (Federal Reserve Bank of Kansas City, 1998), at which an earlier version of the current essay was presented.