[Economic Report of the President (2003)]
[Administration of George W. Bush]
[Online through the Government Publishing Office, www.gpo.gov]


 
CHAPTER 3


Policies for Dynamic Labor Markets


Although the economy continued to grow in 2002, employment growth did
not keep pace. From December 2001 through December 2002, nonfarm
payroll employment fell by 181,000, a small figure compared with total employment of almost 131 million. During the same months the
unemployment rate hovered between 5.5 and 6.0 percent. The lack of
change in these statistics paints a picture of a labor market that is
static and stagnant. But this picture is misleading: dynamic change
remains the most fundamental characteristic of the U.S. labor market
even today. The conventional misperception stems in part from the
nature of most labor market statistics, which by necessity show the
situation at only a single point in time, and which meld the often very different experiences of individual workers and households into a
single aggregate measure.

A closer look suggests ripples--even crosscurrents--beneath the surface.
The unemployment rate may have changed little in 2002, but the names
and faces of the individual workers who are unemployed do change. What
makes the labor market appear stagnant is that the official payroll
employment and unemployment statistics that are the most visible
indicators of the health of the labor market cannot capture its true
dynamism. For example, in December 2002, 67 percent of unemployed
workers reported being unemployed for 5 weeks or more; this
point-in-time statistic may suggest that people who are unemployed this
month will be unemployed next month. Yet a recent study of employment
flows found that the majority of workers seeking work in any given
month are not the same individuals who will seek work the following
month. Similarly, over the same period in 2002 in which payroll
employment scarcely grew, between 3.5 million and 5 million workers
started new jobs each month, and roughly the same number quit or lost
their jobs. This argues that dynamism, not stasis, is the essence of
the U.S. labor market.

This dynamism and its implications for the design of economic policy
are central themes of this chapter. Within these broad themes, the
chapter discusses the rewards to skill and work generated by the U.S.
labor market and how government policies can foster long-run job
mobility by encouraging skill development and education. The labor
market and the economy as a whole today face multiple challenges: in
the short run, the challenge is to move past the recent downturn in the business cycle; in the long run, it is to address the risks associated
with dynamism: technological change and growth inevitably lead to the destruction of some jobs and to the decline of certain industries. If
the Nation can maintain the dynamism--the flexibility and mobility--of
its labor markets, while providing all workers with meaningful
insurance against unemployment and loss of income, both the cyclical
and the structural economic challenges can be met without impairing
those features of the labor market that foster long-run growth.

Economic downturns are a difficult time for many workers and their
families, as growth in employment slows and unemployment and layoffs
increase. The recent economic slowdown has been no exception. Flexible
labor markets, however, can lessen the impact of a downturn on workers,
and probably have done so since the recent contraction began. Although
the unemployment rate did increase sharply during the contraction of
2001 and persisted at levels near 6.0 percent in 2002, unemployment
remains low relative to the experience in previous recessions since
World War II. Job creation and destruction continued despite the
decline in nonfarm payroll employment.

Workers do not encounter economic difficulties only in recessions: even economic growth, or, more precisely, the structural change that
accompanies that growth, makes some workers worse off. Advances in
technology and the expansion of free trade provide benefits for
consumers and for the vast majority of workers, yet these same changes
do real harm to some workers in some areas of the economy. Workers
displaced by technology or trade may remain unemployed for long
periods or drop out of the labor force altogether. When they again
find jobs, they are likely to earn less than at the jobs they lost.

Government has a role in assisting both those who suffer
disproportionately during times of economic hardship and those who fail
to benefit from, or are harmed by, economic progress. Among other
things, government can provide retraining services and relocation
assistance to those who would benefit from them, and it can reward reemployment, through appropriate provisions in the tax code, in social programs, and elsewhere, to encourage rapid reentry into the work
force. In these and other ways, effectively designed government
policies can help make labor markets work better. However, policies
that fail to recognize the dynamics inherent in these markets can
impair long-term economic mobility and the well-being of workers.
Policies will support labor markets and help them work better if they recognize their dynamism and avoid undermining their ability to reward
work and skill.

Social insurance, through unemployment benefits and similar programs,
is an important mechanism by which government can assist those hurt or threatened by the forces of economic change. Yet policymakers face a
tradeoff when seeking to provide social insurance. Such insurance is
valuable in sustaining the well-being of workers and their families
during periods of unemployment, but it can also distort both their
incentives and their behavior, undermining what the U.S. labor market
does best, namely, reward work and skill and match workers to jobs. In
a static, unchanging world, policies that simply transfer public
resources to those who are temporarily poor would not distort their
behavior or lead to dependency on welfare. But in a world of continuous
change in employment and unemployment, poorly designed policies can inadvertently inhibit upward mobility. Although this tradeoff cannot be entirely avoided, labor market policies are more effective when,
recognizing the dynamism of these markets, they provide social
insurance in a manner that least distorts workers' incentives to stay
employed and to improve their employment situation.

The objective of social insurance is to guard individuals and
households against sharp fluctuations in their standard of living that threaten their well-being. A standard assumption in economics is that
most people would prefer their consumption to be certain and steady
over their lifetime, rather than uncertain and variable. However,
because employment and earnings vary in response to events outside
their control, most people find that their incomes are not certain and
steady. This creates a mismatch over time between their desired
consumption and their actual ability to consume, which they seek to
remedy by smoothing their incomes over their lives. They do this in a
number of ways. One way is by saving part of their income when income is relatively high and by dissaving (that is, drawing down their savings,
or borrowing) when it is relatively low. Another is by purchasing
private insurance policies against unexpected and costly events, such
as large health expenses, disability, or premature death. A third is by relying on informal private insurance mechanisms, such as support from
family members and charities, when times are bad. Finally, the public
safety net, of which social insurance is a vital part, acts as a
backstop in case these private insurance mechanisms prove
insufficient.

For most people, any spells of unemployment that occur during their
working years are temporary. Public insurance programs would ideally
therefore provide assistance only for a similarly limited duration.
However, a well-known problem in insurance markets is that of moral
hazard. Moral hazard arises when people who have insurance against a
given risk have less of an incentive to take actions to minimize that
risk than they would if they lacked insurance. For example, people who
have generous health insurance may consume more medical services than
they really need, because the additional services cost them little or
nothing. Similarly, subsidized flood insurance may encourage building
and rebuilding of homes in flood plains, because the insurer or the government, not the homeowner, pays when the house is destroyed in the
next flood.

Moral hazard in social insurance can take the form of dependency on
welfare. For example, for decades the Aid to Families with Dependent
Children (AFDC) program provided income support for the poor but also generated substantial work disincentives that encouraged people to stay
in poverty and out of the work force. The Personal Responsibility and
Work Opportunity Reconciliation Act of 1996 (PRWORA) transformed this
system into one that acts more like insurance against temporary poverty
and less like a permanent transfer program. PRWORA, the most important
piece of welfare reform legislation in several decades, replaced AFDC
with a new program, Temporary Assistance for Needy Families (TANF), and allowed States to implement innovative provisions in their welfare
programs. (Many States had already implemented welfare reforms before
1996 by obtaining waivers from Federal welfare requirements.)
These changes, combined with time limits on welfare receipt and work requirements as a condition for benefits, quickly led to a large
decline in caseloads. Research has found that these reforms led to
increases in work, earnings, and income and a reduction in poverty.
Other effects included an increase in the marriage rate and a reduction
in the prevalence of single motherhood among women with little
education (many of whom likely would have been welfare recipients had
welfare reform not happened). In addition, States that placed a cap on
welfare benefits, as opposed to increasing benefits if a mother had an additional child while on welfare, saw a reduction in out-of-wedlock childbirths. Welfare reform under PRWORA thus provides a striking
example of how well-designed policies can meet the needs of those
struggling in the face of labor market change while maintaining the
incentives that underlie long-term economic growth.

The remainder of the chapter proceeds as follows. It first discusses
the dynamics of employment and unemployment and provides examples of unemployment policies that are made with a dynamic labor market or with
a static labor market in mind. Second, it discusses the dynamics of participation in welfare and other social assistance programs and
contrasts those programs that are designed with an understanding of
dynamic labor markets with those that are not. Finally, it discusses
policies that support mobility and dynamism in labor markets by
fostering investments in skill. For example, in January 2003 the
President proposed the creation of Personal Reemployment Accounts.
These would provide unemployed workers with up to $3,000 to use for
training, child care, transportation, moving costs, or other expenses associated with finding a new job. Recipients who take a new job within
13 weeks would be allowed to keep the funds remaining in the account as
a reemployment bonus. This would give unemployed workers an incentive
to find work faster.

Employment Dynamics
and Labor Market Policy

Whether from the perspective of the economy as a whole or from that of
the individual worker, labor markets work best when they are fluid and flexible, that is, when workers and employers can change their mutually agreed-upon working arrangements as they see fit, to meet changing
needs. Over the long term, the U.S. labor market has indeed been full
of change. A vibrant economy created over 40 million new jobs between
1980 and 2002. Even though the population of the United States aged 16
and over grew by more than 46 million over the same period, a greater
fraction of Americans are working today than in the past: civilian
employment rose from 59 percent of the population aged 16 and over in
December 1980 to 62 percent in December 2002. Women enjoyed a
particularly large rise in their employment-to-population ratio over
this period: for example, in December 1980, 48 percent of the female population were employed, but 56 percent had jobs in December 2002
(Chart 3-1). Meanwhile the proportion of the male population who were
employed fell slightly, from 72 percent to 69 percent.





Blacks and Hispanics also experienced rapid growth in employment since
1975 (Chart 3-2). Indeed, employment-to-population ratios for both
these groups rose by more (7.4 and 8.0 percentage points, respectively)
than did the ratio for whites (6.3 percentage points). By 2000 the
ratios for Hispanics and whites were almost equal. Unfortunately,
although employment grew faster during this period among blacks than
among Hispanics or whites, the black employment-to-population ratio
remains lower than for whites, having started from a much lower level. (Comparisons of employment by race and ethnicity are somewhat clouded
because the categories of Hispanic, black, and white are not mutually exclusive: some Hispanics identify themselves as white and others as
black. Available data do not allow a comparison of non-Hispanic whites
and non-Hispanic blacks with Hispanics.)

The growth in employment in the 1980s and 1990s opened the door to
increased economic well-being for many more Americans. Workers with a
great deal of education and skill benefit from the greater availability
of jobs, but so do workers with less education and fewer skills: one
study indicates that, at both low and moderate skill levels, more labor
market experience means higher earnings; even entry-level jobs provide
real economic opportunities. For both lower skill and higher skill
workers, real wages grow roughly 5.5 percent a year during the
worker's first 10 years in the labor market.





Why does time spent in employment, even in low-skilled jobs, promote
wage growth? One reason is that labor market experience fosters skill development. In a modern economy, school is not the only place where
skills are learned: family members and employers play a central role
alongside formal education in developing skills. One study estimates
that job mobility, workplace education, and on-the-job learning account
for as much as half of all skill formation.

In the dynamic view of labor markets, job changes are not necessarily
events to be minimized at all costs, but rather are often changes for
the better; for example, job changes can lead to a better matching of
workers to jobs. Job mobility also contributes to skill development and
wage growth. Young workers change jobs often: a study shows that the
typical young worker holds seven jobs over his or her first 10 years of
labor market experience, and one-third of the wage growth that young
workers experience occurs when they change one job for another. Indeed,
two-thirds of lifetime wage growth occurs within the first 10 years of
labor market experience. Together this evidence indicates that this job
search and job tryout process--playing the labor market field--is a
crucial component in the economic progress of young workers.

Job mobility is not limited to the young, of course. Studies show that
one-third of new full-time jobs end within 6 months, and one-half to
two-thirds end within 2 years. Not surprisingly, then, a large fraction
of the work force--roughly one-fifth--have been at their current job for
less than a year. However, once a worker has found a good match--a job
in which the worker's skills are valued by the employer and the worker
is sufficiently compensated, both monetarily and in nonmonetary
benefits and amenities--the job often turns into a long-term employment
relationship, to the benefit of both worker and employer. Recent
studies indicate that such relationships remain common (Box 3-1). The
pattern seems to be that many workers switch jobs several times until
they find the right one, ratcheting up their wages along the way.

Job mobility and labor market experience, especially for young workers,
are an important component of overall income mobility in the United
States. In fact, studies of overall income mobility that include the
benefits of job mobility and experience find much more mobility than do studies that implicitly exclude these sources of income growth. Box 3-2 provides a further description of these two contrasting ways of looking
at income mobility.

Nonwage benefits are also an important indicator of workers' well-being.
For the majority of households, health insurance coverage is linked to employment. But even many households with working members lack health insurance. Data from the Current Population Survey, conducted by the
Bureau of the Census and the Bureau of Labor Statistics, show that out
of a
---------------------------------------------------------------------
Box 3-1. Has There Been a Decline in Long-Term Employment?

The fraction of the work force in long-term employment relationships
has been falling over time. In 1979 over 40 percent of the work force
were in employment relationships that had lasted over 10 years, and
over 25 percent had been in employment relationships that had lasted
at least 20 years. In contrast, a 1997 study found that only about 35
percent of employment relationships had lasted at least 10 years, and
about 20 percent had lasted more than 20 years. However, this decline
in the fraction of long-term jobs is largely the result of the rapid
expansion in employment that has occurred since 1980 rather than a
decline in the number of long-term relationships. Workers who are new
to the labor force have short job tenure by definition. There has been
no increase in the incidence of job loss among workers with long-term employment relationships.
______________________________________________________________________

U.S. population of almost 285 million, 41.2 million lacked health
insurance at any given time during 2001. However, just as the
unemployment numbers fail to capture the dynamics of the labor market,
so, too, these commonly cited estimates of the population without
health insurance fail to tell the whole story. The Census figure
probably overestimates the number of people who go without insurance
for a full year. Data from the Medical Expenditure Panel Survey (MEPS), conducted by the Agency for Healthcare Research and Quality, show that
23.5 million people were uninsured throughout a recent 2-year period,
and that 80.2 million were without insurance at some time during that
period. For those who lose coverage, the median spell without insurance
is 5 months.

In the extreme, the combination of a high rate of workers changing jobs,
short durations of many employment relationships, and short average
durations of unemployment could reflect either of two possible
scenarios. One is that a large fraction of workers are experiencing
frequent but temporary layoffs and recalls, such that a re-sorting of
workers is taking place among an unchanging set of existing jobs. The
other is that workers are fluidly pursuing job opportunities that are continually being created to replace other jobs that are continually
being destroyed. Both scenarios are likely at work, but studies show
that a substantial amount--35 to 45 percent--of worker turnover is
driven by the destruction and creation of jobs. Each year roughly 10
percent of all existing jobs are destroyed, and a roughly equal number
of new jobs take their place.

Data from the Bureau of Labor Statistics' Job Openings and Labor
Turnover Survey (JOLTS) document that the common notion of a static
labor market does not fit the facts even during periods of slow
employment growth. The JOLTS gathers data on job openings and job
turnovers from a nationally representative sample of roughly 16,000
business establishments. Those data reveal that, in October 2002, there
were 3.2 million job openings--that is, available but unfilled
positions--the equivalent of 2.5 percent of total employment of roughly
131 million. Moreover, in that same month 4.1 million workers--3.1
percent of total employment--were hired into new positions (from other positions or from nonemployment), and a nearly equal number quit or
lost their jobs. The majority of these separations were not layoffs,
however; 2.2 million of those 4.1 million workers left their jobs
voluntarily. Thus, although nonfarm payrolls increased by only 69,000
between September and October, and unemployment increased slightly
(from 5.6 percent to 5.7 percent), there was a large amount of movement
both into and out of jobs.

What kinds of policies work best to support workers in need of
assistance while maintaining the dynamism of a constantly changing
labor market? The Earned Income Tax Credit (EITC) is an example of a
policy that works

----------------------------------------------------------------------
Box 3-2. Two Ways to Look at Income Mobility

Some studies find substantial income mobility among Americans, whereas
others find much less. The differences between these studies depend in
large part on whether the income mobility that comes with increased
labor market experience is included in the analysis. Studies that
include all sources of income mobility are sometimes referred to as ``absolute`` measures of mobility, whereas those that compare incomes
over time of cohorts of individuals of the same age and approximately
the same level of experience are sometimes called ``relative`` measures
of mobility.

Studies of absolute mobility find that 80 percent of individuals in the
bottom quintile of the income distribution were in a different quintile
10 years later. This finding suggests that most people at the bottom of
the income distribution move up as they gain labor market experience.
Even studies that examine the absolute mobility of men in their prime
working years (ages 25 to 44), after many job changes and after much
wage growth has already occurred, find a substantial amount of mobility.

Studies of relative mobility find less movement out of the bottom
quintile: only about half of workers in that group are no longer there
after 10 years. These studies show that much of an individual's upward mobility is shared among all members of the cohort. Changes in the
relative ranking of incomes among members of a cohort are a good
measure of social mobility, whereas changes in the absolute level of
incomes are a good measure of mobility in economic well-being. Taken
together, these studies show a substantial amount of both concepts of mobility.

----------------------------------------------------------------------

because it encourages rather than discourages mobility in the labor
market (Box 3-3). It does so because its implicit subsidy to earnings,
which can be as large as 40 percent, increases the rewards associated
with work for the low-income individuals to whom it is targeted. The
credit thus provides an incentive for those without jobs (including
those on public assistance) to enter or reenter the labor force. Indeed, several studies have found that the EITC increases labor force
participation among those eligible. The effect is particularly strong
for single parents. One study found that, between 1984 and 1996, the
EITC accounted for roughly two-thirds of the 4.7-percentage-point rise
in labor force participation among single mothers with children. By
2001 the labor force participation rate of these women had risen an
additional 8.6 percentage points (Chart 3-3). In addition, studies
have found the EITC to be more than twice as effective as the minimum
wage at lifting families with children out of poverty, partly because
of the program's positive employment incentives.

----------------------------------------------------------------------

Box 3-3. The Earned Income Tax Credit

The EITC is a tax credit for the working poor. Benefits are paid only
to those who work, and these benefits rise as earnings increase.
Because the tax credit is refundable (that is, it can exceed the amount
of income tax otherwise due), families who pay little or no income tax
can benefit fully from the program.

The program works as follows. Families are eligible for the credit if a
member of the family works. The benefit amount depends on the family's
labor market earnings, the number of children in the family, and the
marital status of the tax filer. In 2003 a family with two or more
children receives a subsidy of 40 cents for each dollar of earned
income up to $10,510. From that level the credit remains stable at
$4,204 until earnings reach $13,730 ($14,730 for a married couple).
Single individuals and families without children are also eligible but typically receive less. The credit phases out over a range of income
from $13,730 through $33,692 ($14,730 to $34,692 for a married couple).
Over this range there is a relatively high implicit marginal tax rate
on earnings. For example, for each dollar earned between $13,730 and
$33,692, a family with two or more children sees its EITC benefit
reduced by roughly 21 cents (Chart 3-4). According to the latest
estimate from the Bureau of the Census, the EITC lifted 3.7 million
people out of poverty in 2001.
----------------------------------------------------------------------




Unfortunately, the EITC can also provide an earnings disincentive for
some low-income families who are already working. This disincentive
comes about because, over the income range in which the EITC is phased
out, recipients face a relatively high implicit marginal tax rate on
earnings as the subsidy is withdrawn. For example, for families with
two or more children, each additional dollar earned between $13,730 and $33,692 of income reduces the credit by roughly 21 cents. In effect,
this places an additional 21 percent tax on these families' work
efforts over that range of income. Of course, if the phaseout were
steeper and the implicit marginal tax rate higher, fewer families would
be affected by the disincentive. A further concern is that a
substantial amount of noncompliance or error occurs within the program.
The Internal Revenue Service has estimated that, of the roughly $31.3
billion in EITC claims filed in 2000 for tax year 1999, between $8.5
billion and $9.9 billion (27.0 to 31.7 percent) was improperly claimed
and should have been disallowed. This raises questions as to whether
the resources devoted to the EITC are being targeted in the most
effective and efficient way possible.

In stark contrast to the EITC, which recognizes the dynamics of labor
market mobility and fosters labor force participation, the
quintessential static labor market policy is the minimum wage, or the
closely related variant known as the ``living wage.`` Policies such as
these, which mandate that employers pay their workers higher wages than
they might pay voluntarily, could be justified by the view that most
labor market entrants will be stuck in low-wage jobs and will not
experience substantial wage growth over their careers. Both the minimum
wage and the EITC increase the earnings of those low-income individuals
who work. But whereas the EITC increases employment, the minimum wage
likely reduces it: the most recent studies have found that significant employment losses are associated with minimum wage policies.

What accounts for this difference in effects on employment? The EITC effectively lowers the wage at which potential low-income workers are
willing to work but does not affect the demand of employers for their
labor services. A minimum wage, on the other hand, increases the cost
to an employer of hiring a low-wage worker and consequently reduces
that employer's demand for labor services. Even when the minimum wage
does not lead firms to reduce employment, it has been found to reduce
the amount of employer-based training young workers receive. Another
reason why the EITC is a more effective policy is that it is targeted
to those workers who need it most: workers, especially workers with
children, from low-income families. The minimum wage, on the other
hand, applies to all workers whose wages would otherwise be below the
minimum; this includes low-wage workers from families whose other
working members earn high wages.


Unemployment Assistance Policy

As noted at the outset, 6.0 percent of the labor force were unemployed
in December 2002; many more Americans face the risk of becoming
unemployed. On December 14, 2002, the President called on the Congress
to extend unemployment benefits for the 750,000 unemployed workers whose benefits would have otherwise expired. He further asked that this
benefit extension be retroactive, so that no one who is unemployed
would fail to receive any portion of benefits to which he or she is
entitled. The Congress responded to the President's call, and on
January 8, 2003, the President signed this extension into law.

Unemployment and the risk of unemployment are a reality in a flexible
labor market like that of the United States. But this same flexibility
also results in higher overall employment than would prevail in an
inflexible, static labor market. Recognizing the job uncertainty inherent
in a dynamic, flexible labor market, government has long undertaken to
provide social insurance against the risk of lower income resulting
from job loss. However, the government's unemployment policies should
always take into account the substantial and continual movement of
workers into and between jobs and into and out of unemployment.

The Federal-State Unemployment Insurance (UI) program provides unem-
ployment benefits to eligible workers who are unemployed through no
fault of their own (with fault being determined under each State's law)
and who meet other eligibility requirements set by each State
individually. Workers who are unemployed because they are new labor
market entrants, have recently reentered the labor market, have quit a
job, or were fired for cause are not eligible for UI benefits. Although
the formula used to determine benefits varies from State to State, the
dollar amount always depends on the worker's previous earnings up to a specified maximum. There is also a minimum UI benefit for workers with especially low earnings. Because of this truncated benefit structure,
the UI replacement rate (the ratio of the benefit to the recipient's
previous earnings) is higher for low-paid than for high-paid workers,
making UI relatively more attractive to those who earned low wages
while working. In most States workers can receive up to 26 weeks of UI benefits; States with unusually high unemployment may offer an
additional 13 weeks of extended UI benefits.

Statistics on the duration of unemployment show that although most unemployment spells are short, their average duration is longer in the
period immediately following a recession. (These statistics cover all unemployed workers, not just those receiving UI benefits.) On average
over all recessions and expansions since 1970, the median duration of unemployment has been 8.2 weeks in the year following a recession and
6.6 weeks at other times. Similarly, 38.2 percent of unemployment
spells are of 5 weeks or less immediately after a recession, compared
with 44.0 percent at other times. However, the surveys used to generate
most labor market statistics may overstate the duration of the typical unemployment spell. In one study that examined completed spells of UI recipients after the unemployed worker had found another job, it was
estimated that 35 percent had returned to work within 4 weeks of their
job loss.

The most recent recession has followed the pattern of previous
recessions: the median duration of unemployment spells rose from 6.4
weeks in March 2001 to 9.6 weeks in December 2002. In March 2002 the
President responded to this need by signing the Job Creation and
Worker Assistance Act (JCWAA), which provided an additional 13 weeks
of temporary extended unemployment benefits to all eligible unemployed workers, and in January 2003, as noted above, the President again
extended unemployment benefits.

Any time that policymakers consider offering or extending UI benefits,
they face a difficult tradeoff. UI can provide valuable assistance to unemployed workers, but it may also create a disincentive for benefit recipients to return to work. Unemployed workers who rationally
evaluate their options may postpone accepting new work until their UI
benefits are exhausted or nearly exhausted. The result is higher
unemployment and longer average spells of unemployment. In the study
cited above, for example, 40 percent of those who had not received UI benefits, but only 35 percent of those who had, returned to employment
within 4 weeks of their job loss. This 5-percentage-point difference
hints at the disincentives built into UI, since fewer of those receiving
it returned to employment quickly. Another study found more direct
evidence: each additional week of UI benefits was estimated to increase
the duration of the average unemployment spell by about a day. Many
other studies have also found an association between the level of
weekly UI benefits and the duration of unemployment. Still more
evidence comes from Europe, where most countries have more expansive UI policies than the United States and have higher rates of unemployment
and longer average unemployment spells. Although these differences in unemployment outcomes may not be due to differences in UI policies
alone, the totality of the evidence suggests that they contribute.

Chart 3-5 illustrates another aspect of the relationship between the availability of UI benefits and incentives to find a new job.
Unemployed workers who receive UI benefits are more than twice as
likely to find a job in the week before their regular benefits expire
than in the several weeks immediately preceding. As noted above, UI
benefits expire after 26 weeks unless extended, in which case they
expire at 39 weeks (for workers receiving either extended UI benefits
or temporary extended UI benefits). Perhaps not coincidentally, peaks
in the fraction of unemployed workers finding work also




occur around these expiration dates. Among unemployed workers who do
not receive benefits, in contrast, there is no substantial difference
in the likelihood of finding a job at these points in their
unemployment spell.

Moreover, although in theory workers should benefit from the longer
time that UI allows them to search for a new job, evidence of such a
benefit is hard to come by. Some States have experimented with giving
UI recipients a cash bonus if they start a new job before exhausting
their benefits. These reemployment bonuses have been found to reduce
the number of weeks of UI receipt, as was hoped. But researchers also
found that those unemployed workers who received bonuses--and
consequently returned to work sooner--did not, on average, end up
taking lower paying jobs upon reemployment than those who did not
receive bonuses. These findings suggest that the longer period of time
that traditional UI recipients remain unemployed does not necessarily
lead them to find jobs better matched to their skills. One possible
downside to the reemployment bonuses is that the prospect of the bonus
may induce more unemployed workers to claim UI in the first place,
especially if they believe they will find work quickly and therefore
might not bother to claim UI were it not for the bonus.

The President's Personal Reemployment Accounts proposal, announced on
January 7, 2003, builds on the demonstrated potential of reemployment
bonuses to speed unemployed workers' reentry into the work force. The
proposed accounts would also add flexibility to the provision of
training for unemployed workers while avoiding penalizing those who
quickly return to work. Under the proposal, qualifying unemployed
workers would each be given an account with a value of $3,000, which
the recipient could use for reemployment services, training, or
supportive services such as transportation or child care. Recipients
who become reemployed within 13 weeks of receiving their first UI
payment would be able to retain any balance remaining in the account
as a cash reemployment bonus. Those who do not find work within that
period would not be able to cash out their account but could continue
to use it for services while receiving UI benefits.

The President has proposed that States be granted a total of $3.6
billion to create the new accounts, enough to provide immediate
assistance for up to 1.2 million unemployed workers. The accounts
would be targeted at those unemployed workers who are very likely to
exhaust unemployment benefits before finding a new job. In some
circumstances, States would be able to provide the accounts to those unemployed workers who have already exhausted their UI benefits within
the last 3 months.

The flexibility that the new accounts would provide in accessing
unemployment services and benefits is important, because research has
shown that the economic impact of unemployment differs greatly from
worker to worker, reflecting differences in their underlying skills and
in their circumstances. For those unemployed workers whose skills are
no longer valued in the marketplace, extensive retraining may be
appropriate. Other unemployed workers may need help relocating or
weathering a spell of unemployment but have marketable skills and
require little or no retraining. The President's proposal recognizes
that the people best suited to evaluate their current skills and match
them with market opportunities are the displaced workers themselves.

Personal Reemployment Accounts are not intended as a replacement for
UI but rather would be structured as a new component of the UI system.
They would be offered as an additional option to those UI recipients
who, under current UI rules, are referred to reemployment services. Eligibility for an account would be a one-time event.

Who would be eligible to receive Personal Reemployment Accounts? In
October 2002 there were 8.2 million unemployed workers, and in that
same month roughly 700,000 workers received first payments from the UI
system. Current law requires that States identify those UI applicants
who are likely to exhaust their benefits and refer these individuals to reemployment services. Although each State applies different criteria,
the factors used to identify these workers include local unemployment
rates, level of education, recent job tenure, and prior employment in
an industry or occupation in decline or particularly hard hit by
economic downturn. From July 2001 through June 2002, 10.4 million
individuals began to receive UI benefits, and 1.2 million, or about 12 percent, were judged to be very likely to exhaust 26 weeks' worth of
regular UI benefits and were referred to reemployment services.
Personal Reemployment Accounts are targeted to those workers.

In more specific terms, Personal Reemployment Accounts would work in
the following way. UI recipients identified by their State as being
very likely to exhaust UI benefits under current law already must
register with the State's Workforce Investment Act program to become
clients of the already-established network of one-stop career centers.
These recipients would be given the option of receiving in addition a
Personal Reemployment Account as part of the intensive services they
receive. The career centers would administer the accounts on the
recipients' behalf. The worker would continue to be eligible for and
receive UI benefits and would be free to use the core services provided
by the one-stop career center. Personal Reemployment Accounts thus
represent additional dollars available to the unemployed recipient.
Funds from the accounts could be used for other training and support
services (such as transportation and child care) at the recipient's discretion. The career center would use an ``advanceable`` process such
as smart cards or an allowable billing process to permit recipients to
make payouts from the account.

If the recipient is reemployed within 13 weeks of starting UI benefits,
the career center would pay him or her, in cash, any balance remaining
in the account. The account would then be closed. States would have the
option of providing the cash balance as a single lump sum or in two installments of 60 percent and 40 percent, the latter after the
recipient has been on the new job for 6 months. The one-stop career
center would distribute these bonus payouts according to the policy of
the State in which it is located. After the cash payout is completed,
the recipient could continue to use all of the no-cost automated and
staff-assisted basic reemployment services available at the career
centers. He or she would not, however, be eligible for intensive
services such as counseling, case management, or training under the
Workforce Investment Act for a period of 1 year after the cash payout. Recipients who do not find employment within 13 weeks of starting UI
benefits would be able to continue to use the resources in the account
for intensive, training, or supportive services.

The potential to receive a reemployment bonus would provide eligible
workers a greater incentive to find new employment. At various times
from 1984 to 1989, four States--Illinois, New Jersey, Pennsylvania, and Washington--conducted controlled social experiments to determine the effectiveness of providing reemployment bonuses to unemployed workers.
In these experiments, a random sample of new UI claimants were told
they would receive a cash bonus if they became reemployed quickly. The advantage of these experiments is that the effect of offering a
reemployment bonus on the duration of unemployment and on earnings upon reemployment can be directly evaluated by comparing the experiences of
UI claimants randomly chosen to be offered a reemployment bonus with
those of UI claimants not chosen for the bonus (who received the
regular State UI benefit).

An evaluation by the Department of Labor of the reemployment bonus
experiments conducted in the States of Washington, New Jersey, and Pennsylvania showed that a bonus of $300 to $1,000 motivated the
recipients to become reemployed, reduced the duration of UI by almost a
week, and resulted in new jobs that were comparable in earnings to
those obtained by workers who were not eligible for the bonus and
remained unemployed longer. Similarly, a study of the experiment
conducted in Illinois found that a reemployment bonus of $500 reduced
the duration of unemployment by more than a week and did not lead to
lower earnings at the worker's next job. Therefore it is likely that
giving unemployed workers the option of receiving the unspent balance
in their Personal Reemployment Accounts will provide them an incentive
to find a new job quickly, reducing the time spent unemployed, but will
not result in workers taking lower paying jobs than they would get if
they searched longer.

A potential problem with Personal Reemployment Accounts is that, like
other reemployment bonuses, they may make UI benefits more attractive
for unemployed workers who expect to find new employment quickly and
thus would be unlikely to apply for traditional benefits. However, the
fact that Personal Reemployment Accounts would be targeted to those
workers whose characteristics are highly correlated with long-term unemployment makes it much less likely that the accounts would induce
entry into the UI system.

Workers adversely affected by international trade are eligible for
support from another Federal program separate from the UI program: the
Trade Adjustment Assistance program. To further assist these dislocated workers, the President and the Congress extended benefits under the
program as part of the Trade Adjustment Assistance Reform Act of 2002.
The main features of this part of the legislation include an extension
of eligibility and an expansion of benefits. To be eligible for these benefits, laid-off workers must have been working in an industry in
which either sales or output has declined, and increased imports must
have contributed importantly to their being laid off. (Workers
subjected to partial rather than full layoff are also eligible.)
Benefits include both cash and training benefits, a tax credit for
health care expenses, and eligibility to participate in State-run
high-risk insurance pools and other State-based efforts to extend
health care coverage. A pilot program for wage insurance has also been launched for these workers. The program offers a wage subsidy for
eligible workers over 50 who take a new job at a lower salary. The
subsidy pays half of the difference in wages between the old and the
new job, up to $10,000. This program is particularly noteworthy because
it provides a direct incentive for seeking reemployment quickly.


Dynamics of Program Participation
and Social Policy

Government social support is, of course, not limited to the unemployed. Disability and spells of low income resulting from any cause are
additional risks against which the government may have a role in
providing social insurance. In 2002 approximately 2 million families
received TANF cash assistance in any given month; another 5.2 million individuals received Supplemental Security Income (SSI) payments, 6.9
million received Social Security Disability Insurance (SSDI) payments,
and some received both.

Most spells of welfare benefit receipt are of short duration: studies
of AFDC typically show that half of such spells ended within 1 or 2
years. However, a significant fraction of welfare spells last a long
time. In a study conducted before the passage of welfare reform in
1996, 18 percent of spellswere found to last 5 years or longer, and
one-quarter of recipients had spent 10 years or more on welfare,
although not necessarily all in one spell. Since 1996, substantial
progress has been made: welfare caseloads have fallen by 54 percent,
and is it likely, although no studies are yet available, that the
duration of welfare spells has shortened as well.

SSDI provides benefits to disabled and blind individuals who are
insured through workers' payroll tax contributions. The worker must
have worked and paid Social Security taxes for a sufficient number of
years and must have worked recently to qualify for benefits. SSI, in
contrast, is a means-tested program for persons who are 65 or older,
or of any age if the recipient is blind or disabled. (A means-tested
program is one in which eligibility is determined by income or some
other measure of the applicant's means of self-support, as opposed, for example, to a record of past contributions to an insurance fund.) SSI
is a program of last resort; its benefit formula takes into account
income received from other sources (including other Federal, State, and
local programs as well as private efforts). It does not duplicate these sources but rather fills the gap between them and a specified minimum
level of income. Both SSDI and SSI define adult disability as the
inability to engage in any substantial gainful activity because of a
mental or physical impairment that is expected to result in death or
that lasts for a continuous period of at least 12 months. As of 2002 ``substantial gainful activity`` was defined as work paying over $780 a
month, when the impairment is other than blindness, and over $1,300 a
month for blindness. The average monthly SSDI benefit in 2002 was $817,
and the maximum monthly SSI benefit was $545.

In contrast to TANF, participation in the SSDI and SSI programs has not decreased in recent years (Box 3-4). Of course, unlike with TANF,
individuals
----------------------------------------------------------------------
Box 3-4. The Growth in SSDI and SSI Disability Caseloads

The number of people receiving disability payments through either the
SSDI or the SSI program has increased dramatically. From 1990 to 2002,
the number of SSDI recipients rose by 3.0 million, and from 1990 to
2001 the number of SSI recipients rose by 2.1 million. The President
supports a program that would address this rise in disability caseloads
by helping people with disabilities reenter the work force.

In 1999 Congress passed the Ticket-to-Work and Work Incentives
Improvement Act, which addresses the disincentives to return to work
that many individuals with disabilities face. The act allows recipients
of SSDI and SSI to choose their own vocational rehabilitation and
support systems, and it extends the Medicare benefits of SSDI recipients
so that they do not lose health benefits on returning to work. The act
also expands Medicaid eligibility for persons with severe disabilities.
The President has promised swift implementation of this initiative, to
be completed by the end of 2003.
----------------------------------------------------------------------

apply to receive disability payments both because they require income
support and because health impairment limits their ability to work and
perhaps increases their demand for medical services. Thus one would
expect a lower rate of exit from SSDI and SSI than from TANF, even if
their incentive structures were identical. Indeed, a low rate of exit
has been the norm for these programs: each year only about 1 percent of
those who receive SSDI or SSI leave the rolls to go to work.

How should welfare programs be designed for a dynamic labor market? If
labor markets were static, the design of social insurance to provide
welfare, like the design of UI, would be straightforward: the
government would simply provide cash assistance to needy families. In a dynamic labor market, however, needy families typically require welfare benefits only for brief spells. This very dynamism makes the design of
welfare programs more difficult, because policymakers again face a
tradeoff. Welfare programs that provide cash benefits without work requirements or time limits, such as the former AFDC program, provide
eligible families with needed assistance, but they also create a
disincentive for the adult members of those families to acquire skills,
to enter or reenter the work force, and to escape poverty. (They also
create a modest incentive to remain unmarried and to have children out
of wedlock, because the presence of a working husband reduces the
benefit whereas that of an additional child increases it.) The work disincentives that were part of AFDC (which in part remain under TANF)
arose because benefits were phased out as family income increased,
imposing a high implicit marginal tax rate on income earned by
families receiving AFDC. Although any well-designed means-tested public assistance program would include an income phaseout and thus face this
problem of high marginal ``tax`` rates, the AFDC program
unintentionally promoted dependency on welfare and induced some
families to have longer spells of welfare receipt.

The Personal Responsibility and Work Opportunity Reconciliation Act of
1996 was motivated by the recognition that a better policy for families requiring welfare assistance was needed. The reform granted greater
program authority to State governments and replaced the AFDC program,
which was based on Federal provision of matching funds to the States,
with TANF, which is a block grant program. These reforms essentially
abolished Federal eligibility and payment rules, giving States much
greater discretion in designing their own cash public assistance
programs, and eliminated the Federal entitlement to cash assistance.
TANF not only gave States the freedom to set their own eligibility
criteria and benefit levels but also created work requirements for
recipients, set a lifetime limit of 60 months of TANF assistance, and
rewarded States for strong performance in terms of reduced caseloads.

As noted previously, caseloads have fallen by 54 percent since PRWORA's enactment. However, because the unemployment rate was falling during
much of this period, an important question is whether the decline in
caseloads was due to welfare reform itself or to the strength of the
labor market. A number of studies based on experiences during the
period of extensive State experimentation with welfare program waivers
have found that economic growth and the consequent decline in
unemployment rates likely had a secondary role in the decline in
caseloads. As Chart 3-6 shows, the correlation between unemployment and
the number of AFDC/TANF recipients that is evident in the 1990s, and particularly after welfare reform, was not evident in earlier periods of declining unemployment rates.

The era of innovation in welfare policy began with the granting of AFDC waivers in the late 1980s: certain restrictions under the AFDC program
were waived for States wishing to experiment with alternative welfare
program designs. PRWORA continued this process, with the result that
the particulars of State programs now vary widely. One important
dimension on which they differ is the rate at which welfare benefits
are reduced as the recipient's income rises. This benefit reduction
rate had been set by Federal law under AFDC. Under PRWORA, many States
have chosen a lower implicit rate in an effort to increase the
incentive to work and to provide more assistance to low-wage workers.
Lower benefit reduction rates, in conjunction with increased work
mandates, time limits, and work support programs, do appear to have
increased work incentives. By 2000, States reported that, in




the aggregate, 34.0 percent of the welfare caseload were engaged in
work or job-related activities, up from 20.4 percent in 1994. Although
States remain subject to a Federal 60-month maximum time limit for
individuals receiving TANF funds, they can set shorter time limits or
use State funds to extend benefits. PRWORA has also increased
expenditure on work support programs such as subsidies for child care;
between 1993 and 2000 annual Federal child care subsidies almost
doubled, from $9.5 billion to $18 billion.

With the recent economic slowdown and continued weakness in job growth,
a critical question is whether the number of TANF recipients will
increase to their pre-1996 levels. Analysis based on the relationship
between unemployment rates and recipients, combined with current
forecasts for unemployment in 2003 and beyond, suggests that the number
of recipients will increase only slightly and will not approach
prereform levels (Chart 3-6).

Although it is still too soon to reach a final conclusion, welfare
reform--both TANF and the innovative policies implemented by States
before and since the enactment of PRWORA--seems to have had a
remarkable impact on public assistance caseloads. The reductions in
caseloads, moreover, have not been concentrated geographically but have
been seen across the Nation. PRWORA has also shown positive, although
still preliminary, effects on employment, earnings, marriage rates, and
the prevalence of single female-headed families.

At the same time that caseloads began to fall, employment increased dramatically among the population most affected by the caseload
declines. Among those who reported receiving public assistance income
in the previous year, the share reporting being employed in March of
the following year rose from 19.8 percent in 1990 to 44.3 percent in
2000. Even among those who remained on welfare, work effort greatly
increased, possibly reflecting both the work requirements and the rates
at which benefits were reduced with income. Among women on welfare,
those who reported labor earnings rose from 6.7 percent in 1990 to 28.1 percent in 1999. The research literature suggests that approximately
two-thirds of welfare leavers are employed at any future point in time.
In addition, employers have often rated welfare recipients as performing
as well as or better than other employees. One study finds that former
welfare recipients have higher rates of wage growth than do other
workers.

Poverty and income levels are directly tied to employment and wages.
Studies suggest that, just as it has raised employment, welfare reform
has also reduced poverty and increased income. For example, poverty
among all families headed by a single mother declined from 35.4 percent
in 1992 to 26.4 percent in 2001. This finding is consistent with
research showing that States that adopted innovative welfare programs
under AFDC waivers before 1996 recorded an average 2.4-percentage-point decline in the poverty rate of the entire population of less skilled
women. PRWORA itself was associated with a 2.0- to 2.2-percentage-point decline in the poverty rate.

Research shows that, although many lost government assistance, single
mothers saw their incomes increase on average during the 1990s. Studies
of those who have left welfare reveal that around half remain in
poverty, but evidence also points to increases in family income over
time. Data on consumer expenditure meanwhile reveal increases in
spending by low-income single mothers in the 1990s.

Why has welfare reform been so successful? Studies of the States'
experiments with work requirements under AFDC waivers suggest that
these requirements led to increases in employment and reductions in
welfare payments. The effect of time limits is less well established,
because few recipients have yet exhausted their eligibility under
PRWORA, and evidence from those States that implemented time limits as
part of their AFDC waiver programs is mixed.


Fostering Skill Development

Labor market experience fosters the development of valuable skills. As
noted earlier, in addition to formal educational institutions, family
members and firms play a central role in skill development in the
modern economy. Job mobility, workplace education, and on-the-job
learning by doing account for as much as 50 percent of all skill
formation. Training on the job, together with simply the experience of
being in the labor market, has been found to be more effective at
increasing the earnings of young workers than are government training programs. Indeed, evidence from evaluations of formal, publicly
provided job training programs for youth demonstrates that they have
little or no impact on earnings.

When younger workers change jobs, the switch is usually accompanied
by an increase in wages, possibly because they have both increased
their skills and moved to jobs that use those accumulated skills more effectively. In contrast, job changes for more experienced workers
often result from job loss and may result in lower earnings.
Experienced workers who lose their jobs at a given time are more than
three times as likely to experience one or more additional spells of unemployment in the following 2 years than similar workers who did not
lose their jobs at that time. In addition, more than one-quarter of experienced workers who lose their jobs suffer substantial wage
reductions when they do return to work. The reductions in employment
are short-lived: within 4 years of the job loss, workers who lost their
jobs have a virtually identical likelihood of being employed as workers
who did not lose their jobs. But the wage losses are long lasting: 4
years after a job loss, the average weekly earnings of job losers are
10 to 13 percent below those of workers who did not lose their jobs.
These permanent declines in wages likely reflect a deterioration in the
value of the skills these older workers had previously acquired.  This
makes fostering the reacquisition of skills among experienced workers
who have lost their jobs a policy priority.

Some types of worker retraining have been effective at increasing the
earnings of displaced workers. One evaluation of a training program
that subsidized community college attendance by displaced workers found
that 1 academic year of community college raised these workers' earnings
by about 5 percent.

Technically oriented vocational skills and science and math skills are particularly important for displaced workers, because investments in
these skills result in much higher returns in the labor market than
does non-technically oriented training. One study found that the
expected return on earnings from a curriculum that provides an academic
year of more technical and applied coursework ranges from 10 to 15
percent.

The labor market rewards skill accumulation and investment in human
capital. In particular, it rewards with higher wages those who obtain
more schooling. Studies estimate that each additional year of education increases a worker's wages by 6 to 10 percent on average. The Bureau of
Labor Statistics reports that, in the fourth quarter of 2002,
bachelor's degree holders over the age of 25 had an unemployment rate
of 3.0 percent, and those working full-time earned a median weekly
income of $944, whereas workers with only a high school degree earned a
median weekly income of $545 and had an unemployment rate of 5.1
percent. Americans have responded to the benefits of human capital
investment: in 1959 only 2 in 10 jobholders had some college education;
today roughly 6 in 10 are college educated.

The benefits of education not only are large but have increased. The difference between the average earnings of college-educated workers and
those of high school-educated workers has increased by almost 70
percent since the early 1980s. Education may also generate gains for
society at large: it is correlated with better public health, better parenting, lower crime, a better environment, wider political and
community participation, and greater social cohesion, all of which may contribute to economic growth.

Earnings increase with age, with increased tenure on a job, and with
the accumulation of both general and job-specific human capital.
Between 1963 and 1989, men with 30 years of job experience earned 75 to
85 percent more, on average, than men in their first 5 years out of
school. Furthermore, one study finds that the past three decades have witnessed an increase in this premium: whereas in 1969 high
school-educated men with 30 years of work experience earned 62 percent
more than new entrants with the same education, by 1989 they were
earning 110 percent more. In addition, workers who have been at the
same job a long time tend to stay there: accumulated tenure is negatively related to turnover rates. The rising importance of experience points
to the value of employer-provided training. One study finds that
on-the-job training accounts for at least two-thirds of the growth in
wages in any given year.

The President, recognizing the individual and economy-wide benefits of
an educated society, has vowed to make educating every child in America
a top priority. On January 8, 2002, he signed into law the No Child Left Behind Act, designed to improve elementary and secondary education. The
act requires stronger accountability and high standards of achievement,
to be measured through annual testing of third through eighth graders
and publicly released report cards of school performance. It gives
students who attend low-performing schools, and their parents, greater
scope to seek better options. The act gives State and local governments greater control over Federal education funding, which was increased by
49 percent from its 2000 level, to $22.1 billion in 2002. It creates a
highly qualified teacher initiative, supported by investment, research,
and training, and it increases Federal money devoted to the teaching of reading. The Administration's commitment to education highlights the importance of investing in the Nation's human capital, benefiting both individuals and the economy as a whole.


Conclusion

Policymakers can help labor markets work better, but they need to
remember that labor markets are dynamic, and that the policies that
work best for a dynamic labor market are very different from those
that work best for a static labor market. Static labor market policies
may unintentionally induce workers to accept longer spells of poverty
and unemployment and to remain in lower paying jobs. The policies
described in this chapter should encourage mobility and help workers
smooth over the difficulties they encounter during labor market
transitions.