[Economic Report of the President (1997)]
[Administration of William J. Clinton]
[Online through the Government Printing Office, www.gpo.gov]

[DOCID: f:erp_c4._]
Economic Report of the President - - - - - - - - - - - - H. Doc. 105-002
[From the online service of the U.S. Government Printing Office]
[wais.access.gpo.gov]


CHAPTER 4--The Labor Market

THE RECENT STRENGTH OF THE ECONOMY has created a large number of new
jobs, and the unemployment rate is low by historical standards. Between
January 1993 and December 1996, economic growth produced 11 million new
jobs. At the end of 1996 the unemployment rate was 5.3 percent.
Jobholding increased dramatically even among groups whose members
traditionally have difficulty finding employment.
Economic progress has greatly benefited many American workers, but
it poses important challenges as well. New technologies have led to
explosive growth in some industries, but to the decline of others. With
deregulation and expanding international trade, firms that once enjoyed
market power and could share the resulting economic rents with their
workers are now forced to compete more aggressively in the marketplace.
Technological change and greater competition have eliminated the jobs of
some workers, but many others have found new jobs in industries that
these same powerful forces are causing to expand. Yet some workers may
discover a mismatch between the skills they needed for their old jobs
and those required in the newly expanding sectors. These workers are at
risk of significant unemployment and may have to accept lower wages when
they finally do find work. The benefits that come from an economy that
has been strengthened by technological progress and more intense
competition should be tempered by the recognition that these same
changes may have hurt some working Americans.
To what extent have structural changes in the labor market reduced
the well-being of American workers? Some analysts claim that a
fundamental change in the nature of employment has taken place. While
acknowledging the robust growth in the number of jobs, they maintain
that this growth is concentrated in low-paying jobs, that wages overall
are falling, that layoffs are increasing despite a growing economy, and
that the promise of long-term employment on which many American workers
rely can no longer be kept.
Recent studies suggest that these claims are exaggerated. Although
it is true, as some critics point out, that the number of low-paying
jobs has increased, that of high-paying jobs has increased even more
rapidly. It is the jobs in the middle, the ones offering wages close to
the median, that have become somewhat scarcer. Layoffs, meanwhile, are
not rising: the rate of job loss has actually declined somewhat,
although it does appear that certain categories of workers previously
less affected by job loss are now more at risk. Real, inflation-adjusted
wages have generally been stagnant over the longer term, but standard
methods of adjusting wages for inflation may have masked a real rise,
and total compensation, including fringe benefits, has increased.
Finally, some evidence indicates that the high level of average job
tenure first identified in the early 1970s has changed little since
then, although other recent research disputes this claim. This chapter
examines these and other labor market trends in some detail, describes
how workers have responded to these changes, and discusses policy
alternatives to address some of the real problems that exist.

TRADITIONAL LABOR MARKET INDICATORS

Traditional indicators of labor market performance point to
substantial improvement in the last few years. Perhaps the single most
important indicator, the unemployment rate, is as low today as it has
been at virtually any time in the last 20 years--and lower than it was
through most of the economic expansions of the late 1970s and the mid-
to late 1980s (Chart 4-1). The unemployment rate for the whole of 1996,
at 5.4 percent, was below the rate for any full year since 1973, except
for 1989 when the rate was 5.3 percent. And not only is the overall
unemployment rate low, but groups that traditionally have experienced
greater difficulty in finding jobs are doing better as well. For
example, the unemployment rate for blacks in both 1995 and 1996 was
almost a full percentage point lower than in any of the last 20 years.
The unemployment rate measures unemployment as a percentage of the
labor force, not of the entire working-age population. It would be
little cause for celebration if the unemployment rate has fallen merely
because some jobless workers have become discouraged and have stopped
seeking work, thus removing themselves from the labor force altogether.
Recent data, however, strongly reject this explanation of today's low
unemployment picture. Employment gains have been strong over the last 4
years: the employment-to-population ratio indicates that almost as large
a share of the population is working now as at any time since annual
statistics began to be collected (Chart 4-2).

THE QUALITY OF NEW JOBS

A large number of new jobs have been created over the past 4 years,
but concerns have been expressed about the quality of these


jobs. Recent research finds that most of the new positions created in
the 1990s are ``good'' jobs. The number of lower paying jobs also
increased, however, as employment in the middle of the earnings
distribution fell.

JOB GROWTH WITHIN SERVICE-PRODUCING INDUSTRIES

A disproportionate share of employment growth in the current
expansion has occurred in service-producing industries. But contrary to
the popular notion that service jobs are primarily low-paid positions,
jobs in these industries are actually quite diverse, including many
high-wage positions in such industries as financial services, health
care, and computer and accounting services. For this reason it is
important to determine at which end of the wage spectrum the employment
growth within services has mainly occurred.
The evidence indicates that managerial and professional occupations
have been the main contributors to recent job growth within service-
producing industries, accounting for most of the net increase in
employment in this sector occurring between February 1994 and February
1996 (1994 is chosen as the base year because the Current Population
Survey, or CPS, underwent a major redesign that makes comparisons before
and after 1994 difficult; see Box 4-1). Managerial and professional
occupations within service-producing industries have been large
contributors to employment growth in each of the past three major
expansions. But gains in these occupations have been even more important
in the current expansion. Employment increases in these generally higher
paying occupations may not necessarily translate into high pay for
workers immediately, but the greater opportunities for advancement in
these jobs promise higher wages in the future.

ECONOMY-WIDE JOB GROWTH

A more detailed picture of recent job creation emerges from an
examination of changes in employment within specific industry and
occupational categories. A study conducted by the Council of Economic
Advisers and the Department of Labor compared full-time employment in
February 1994 with that in February 1996 in 45 specific occupations in
each of 22 major industries, for a total of almost 1,000 industry/
occupation ``cells.'' For each of the 287 cells in which employment was
large enough to provide reasonably reliable within-cell wage levels,
median weekly earnings as of February 1994 were determined, along with
the median wage across all cells in that month. Employment growth
between February 1994 and February 1996 in high-wage job cells, defined
as those in which median earnings were above the overall median, was
then compared with overall employment growth. The study found that 68

Box 4-1.--Effects of the Redesign of the Current Population Survey

The Current Population Survey (CPS), conducted monthly by the
Bureau of the Census for the Bureau of Labor Statistics (BLS), is a
major source of data regarding the U.S. labor market, including the
monthly unemployment rate. In January 1994 the BLS revised the
questionnaire to adjust for changes in work patterns and implemented
computer-assisted interviewing to improve the quality of data collected.
The BLS estimates that the overall unemployment rate was not
significantly affected by the redesign. This finding is contrary to
early reports that the new survey produced a slight rise in measured
unemployment. The new survey did change the measured composition of
unemployment, however. For example, measured unemployment among 55- to
64-year-old workers and workers 65 and older increased by about 12 and
50 percent, respectively.
The breakdown of reported reasons for unemployment was also
affected by the redesign. Whereas the old survey asked directly whether
the unemployment spell began because of a quit or a lost job, now
respondents must first report that they were working just prior to their
unemployment spell before that question is asked. Evidently, asking the
question directly induced some workers to report that their unemployment
spell began for one of these reasons, because the number of workers
classified as job losers or job leavers declined using the new survey.
In addition, expanding the definition of previous work experience to
include part-time work led to more workers being classified as
reentrants rather than as new entrants.
The redesign also affected reported unemployment durations,
because computer-assisted interviewing allows the interviewer to check
whether a respondent's answers are consistent from month to month.
Respondents used to overestimate short-term and underestimate long-term
unemployment spells. These changes increased the proportion of spells
longer than 14 weeks and decreased that of spells shorter than 5 weeks.
Other labor market indicators were also affected. Both the labor
force participation rate and the employment-to-population ratio are
about half a percentage point higher when measured using the new
techniques.
percent of the net growth in full-time employment over this period
occurred in these higher paying job categories.
The results of this research were similar to those reported in a BLS
study that divided employment into 90 industry/occupation categories and
then identified jobs in these categories as either high-, middle-, or
low-paying. Between 1989 and 1995, employment in the high-paying and
low-paying categories increased by 13 percent and 7 percent,
respectively, while employment in the in-between category fell by about
3 percent.
An alternative disaggregation of jobs into extremely detailed
occupational categories (also by the BLS) supports these findings. The
BLS compiles responses from a full year of CPS data to examine wages and
employment growth for almost 500 occupational categories. Between 1994
and 1995, some of the categories with the largest employment gains
included sales supervisors and proprietors, electricians, marketing and
advertising managers, and electrical and electronic engineers.
Consistent with the Council's calculations, occupations in the top half
of the wage distribution accounted for 70 percent of net employment
growth.

FULL-TIME VERSUS PART-TIME JOBS

Even if today's new jobs are more likely than before to be in the
higher paying sectors of the labor market, not all of these jobs provide
workers with full-time employment. Data from the CPS provide an
opportunity to explore trends in part-time employment. Chart 4-3 depicts
the proportions of employed persons reporting that they work part-time
for ``economic'' reasons (i.e., who would prefer a full-time job but
cannot find one). Most of those who work part-time seem to do so by
choice; moreover, the proportion of part-time workers who do so for
``economic'' reasons has been declining.

THE LEVEL OF WAGES

The economic growth of the 1980s produced only small real wage gains
for workers. Moreover, real wages, when adjusted for inflation by
consumer prices, have failed to keep pace with worker productivity since
about 1983--a clear departure from the pattern of preceding years. (See
Box 4-2 for a discussion of potential biases introduced in measuring
consumer price inflation.) Although productivity growth has slowed, from
around 2.8 percent per year before 1973 to 1.1 percent per year since,
it has not stagnated; it therefore cannot explain these wage trends.
After documenting the trends, the following discussion explores two
possible explanations for them: changes in the relationship between the
consumer price indexes, used for measuring real wages from the worker's
perspective, and overall price indexes used for measuring real wages
from


the producer's perspective; and changes in the relationship between
wages and total compensation, which includes fringe benefits as well as
wages.

Box 4-2.--The Influence of Inflation Adjustments on Measured Real Wages
and Incomes

Standard analyses of wage and income trends use the CPI to adjust
for inflation. But the CPI is a biased measure of the cost of living
because it does not adjust for substitution between goods and may not
fully account for changes in their quality--problems that are described
more fully in Chapter 2. To the extent that the CPI overstates
inflation, adjustments to wages and income using the CPI will understate
actual growth in purchasing power.
Chart 4-5 displays trends in wages from the ECI, adjusted by the
official CPI and adjusted by CPI inflation less 0.5 percentage point. If
the CPI overstates inflation by \1/2\ percentage point, real ECI wages
have actually risen by almost 10 percent since the early 1980s. Trends
in real income, described in the following chapter, show the same
sensitivity to bias in inflation adjustment.

TRENDS IN WAGES

Chart 4-4 shows annualized real changes in wages and earnings over
the past decade and a half using five different data sources (Box 4-3)
adjusted for inflation by the consumer price index (CPI). For three of
the sources, wages were virtually unchanged over the period. Changes in
median real weekly and annual earnings for full-time workers, estimated
from the monthly CPS and the annual March CPS, were similar to those
from the employment cost index (ECI) despite several differences in
methodology: the ECI data measure mean rather than median wage changes,
compute hourly wages rather than weekly or annual earnings, and include
part-time as well as full-time workers. Of the five series, only average
hourly earnings, as measured in the BLS's Current Employment Statistics
(CES) program, fell noticeably over this period. Unlike the other
series, the CES covers only production and nonsupervisory workers, who
suffered relative wage declines in the 1980s.


An interesting feature of wage trends is that they display no
apparent pattern over the business cycle. Economic theory does not offer
a clear prediction of how real wages should move over the


cycle. On the one hand, we might expect the greater demand for labor
during an expansion to lead to real wage increases. On the other hand,
as the economy expands, it puts into production its less efficient
capital stock. To induce firms to do this, prices of the goods they sell
must rise relative to wages, which means that real wages must fall.
Empirically, the fact that aggregate measures of the real wage show
little cyclicality may indicate that these two effects are offsetting.
A difficulty in identifying changes in wages over the business
cycle, however, is that the pool of employed workers changes. During
recessions, lower skilled and less experienced workers are more likely
than others to lose their jobs. When the economy recovers, these same
workers become reemployed. Therefore, during an expansion the labor
force is likely to include more low-paid workers; this depresses the
average wage. Research shows that once the composition of the pool of
employed workers is controlled for, the wages of male workers are
considerably more procyclical than the aggregate wage statistics
indicate.

WAGES VERSUS TOTAL COMPENSATION

The discussion so far has mainly focused on wages. However, for many
purposes total compensation, which includes fringe benefits, may be a
more useful measure. Although real wages have changed

Box 4-3.--Sources of Wage Data

Several data sources can be used to track trends in wages. Five
commonly used sources are the following:
 The March CPS, conducted by the Bureau of the Census,
reports median annual earnings for full-time, year-round workers
for the preceding calendar year.
 The monthly CPS, conducted by the Census Bureau for the
BLS, asks one-quarter of all respondents about their ``usual''
weekly earnings and hours worked on their main job, in order to
estimate the median wage for all full-time workers. Earnings
data from this source are reported quarterly.
 The employment cost index, produced by the BLS, is based on
a survey of wages, salaries, and benefits in approximately 4,700
establishments in the private sector. Firms surveyed are chosen
so as to maintain a constant industry and occupational mix of
workers, to eliminate the effects of employment shifts between
industries and occupations.
 The Current Employment Statistics survey, conducted by the
BLS, obtains data from nearly 400,000 establishments in private
nonagricultural industries regarding earnings and hours worked
for all production and nonsupervisory employees. The data can be
used to construct a measure of average hourly earnings.
 The wage data in the national income and product accounts,
produced by the Bureau of Economic Analysis, are based on
quarterly earnings records for workers covered by State
unemployment insurance. Data on the number of paid hours from
the Current Employment Statistics survey are used to translate
these quarterly data into mean hourly wage measures, and these
data are supplemented by imputation for those workers not
represented in that survey.
little in the last decade and a half, total compensation has risen
modestly since the mid-1980s. Meanwhile fringe benefits, which comprise
roughly 30 percent of total compensation, have risen sharply. This rise
is driven primarily by rapid increases in the cost of employer-provided
health benefits, which increased over 20 percent in real terms between
1982 and 1994. However, employer health costs have stabilized since
1994, reflecting some combination of slower increases in the prices of
medical care services, a shift toward managed care, increased premium
cost sharing with employees, and a reduction in the share of the
workforce with employer-paid health insurance (Box 4-4). In competitive
labor markets, a rise in one component of compensation might be expected
to lead firms to reduce another component, so as to keep total
compensation in line with worker productivity. This may have happened
during the 1980s and early 1990s, as wages remained relatively stagnant
to compensate for sharply rising health benefit costs.
Even so, total compensation has risen more slowly than have
increases in productivity, when nominal compensation is adjusted for
changes in the prices of consumer goods. A possible explanation is that
producer prices have fallen relative to consumer prices, largely as a
result of the decline in the prices of many industrial goods, such as
computers. From the perspective of firms, prices for all output,
including investment goods, offer a better method of adjusting trends in
compensation. Because firms hire an additional employee only if the cost
of doing so is less than or equal to the value of that employee's
output, a more appropriate measure to compare with productivity may be
compensation adjusted for all output prices. As can be seen in Chart 4-
6, changes in real compensation, when deflated by output prices, have
tracked changes in total productivity more closely since the mid-1980s
than when consumer prices are used for the adjustment.

JOB LOSS

The threat of losing one's job engenders justifiable anxiety,
because job loss can result in a lengthy spell of unemployment and a
long-lasting reduction in earnings even after a new job is found.
Economic expansion creates dynamism in the labor market, with
reallocation of workers across sectors, and in such periods growth in
new jobs typically is sufficient not only to lower the aggregate
unemployment rate and to create jobs for new entrants into the labor
force, but also to accommodate those workers displaced from their old
ones. Historically, the highest rates of job loss tend to occur during
recessions. Some have claimed that it is high today for an expanding
economy.

TRENDS IN THE RATE OF JOB LOSS

The Displaced Workers Survey, published by the BLS since 1984 as a
biennial supplement to the Current Population Survey, has become an
important source of data on job loss. This survey identifies workers who
have lost jobs within the 3 to 5 years before the survey date, either
because their plant closed or moved, because their position or shift was
abolished, or because of insufficient work. The

Box 4-4.--Trends in Employer Health Care Costs

The cost to employers of providing health insurance to their
employees rose more rapidly than inflation throughout the 1980s and
early 1990s. Since then, however, this trend has reversed: in the past
few years firms' health insurance costs have actually fallen in real
terms. This turnaround is the result of a combination of factors
including slower growth in medical expenditures, employers switching to
lower cost managed care plans, declining health coverage of retirees,
and, possibly, modest cost shifting to employees.
Slower Growth in Medical Spending. Overall private medical
expenditures are increasing much more slowly than in the past. Premiums
(employer and employee) at medium-size and large firms rose by about 11
percent in 1991 and 1992, but only 2.1 percent in 1995 and 0.5 percent
in 1996. The move to managed care may help explain why growth in health
costs has moderated so sharply. Not only are managed care plans cheaper,
but their expansion may also be forcing the competing traditional plans
to become more efficient.
Coverage Trends. Over the past 15 years, employers have reduced
the number of workers for whom they provide health insurance coverage.
But most of that reduction occurred before the recent slowing in health
benefit costs. However, employers have continued to decrease the share
of their retirees eligible for health benefits.
Cost Shifting. Employers have tried to hold down rising health
benefit costs by shifting more responsibility for premiums and other
expenses onto employees. But this trend has moderated recently. Since
1992, the percentage of workers whose employers fully finance their
health insurance has changed little. Nor has the average premium
contribution that firms require their employees to make been modified
much in recent years. Deductibles and out-of-pocket spending have
increased little in the same period. One reason is that coverage has
shifted dramatically into managed care plans, which typically have low
copayments and deductibles.
survey, which is conducted in January or February of every even-numbered
year, can be used to examine trends in displacement rates, the
characteristics of dislocated workers, and the costs associated with
permanent job loss. Most of the results reported in the survey, and all
those reported here, reflect job displacement for so-called long-tenure
workers: those who were employed in their previous job for 3 or more
years. The rationale for this focus is that


individuals with lengthy job tenure are likely to have the most severe
adjustment problems when displaced.
Chart 4-7 shows trends in the rate of job displacement among long-
tenure workers since the early 1980s. As one might have expected from
the deep recession of 1981-82, job dislocation rates were high during
this period. As the economy recovered in the mid-1980s, displacement
rates fell. The recession of the early 1990s again saw increasing rates
of displacement: job loss in 1991-92 was as prevalent as it had been in
1981-82, even though the earlier recession was much more severe.
Although displacement statistics from the 1993-94 period are calculated
from unpublished data and may not be directly comparable to earlier
years, displacement rates appear to have subsided to the level that
prevailed for most of the late 1980s. Displacement rates were quite a
bit lower, however, in 1987-88, even though the unemployment rate in
those years was close to that in 1993-94. One may infer from these data
that some of the problems of job loss are persisting even in the face of
a healthy economic expansion.
Other measures, such as the monthly CPS, indicate that the rate of
job loss has fallen significantly in recent years. The monthly CPS
obtains information not only on labor market status, but also on the
reasons why an unemployed worker began looking for work and the length
of time spent looking. Job losers who are not on layoff


may be thought of as ``permanent'' job losers, even though they may have
been fired for cause or have some chance of eventually being recalled.
The number of these job losers unemployed for less than 5 weeks is an
indicator of the number experiencing permanent job loss. These data are
valuable because the CPS is the standard survey of labor market behavior
and because the data are available on a regular basis. However,
displaced workers who find a new job without an intervening spell of
unemployment are not captured by this measure. Chart 4-8 shows that job
loss by this measure has declined over the last few years and is
currently comparable to the rates observed throughout much of the late
1980s.
Initial unemployment insurance (UI) claims provide another measure
of job loss. Initial UI claims have declined throughout the current
expansion: weekly claims have fallen by about one-third since the 1990-
91 recession. Although the share of unemployment spells that are
compensated has declined over time, recent trends fairly accurately
reflect changes in the number of workers who have lost jobs or been laid
off. These data are obtained from the administrative records of the UI
system and represent a complete count of layoff activity that leads to a
UI claim, rather than a sample. The weaknesses of these data are that
they include temporary as well as permanent job loss and that they do
not capture job losses that do not lead to a UI claim.


The distribution of job displacements has apparently changed over
time. Workers in service-producing industries and white-collar
occupations have become more vulnerable to job displacement, whereas
blue-collar and manufacturing workers have become relatively less prone
to lose their jobs. Thus, whereas service-producing industries accounted
for about a third of all long-tenure displaced workers in the 1979-84
period, this sector's share has recently climbed to over one-half.
Similarly, white-collar workers represented about 40 percent of those
displaced in the early 1980s but now constitute more than half of job
losers. Older and more educated workers also are exposed to greater risk
of displacement than in the past. The bottom line is that the risk of
job loss is now spread over a wider cross section of employees.

THE COSTS OF JOB LOSS

The costs of losing one's job include lost wages during any
subsequent unemployment and any wage reduction or loss of fringe
benefits that results when a new job is obtained. Displaced workers are
now finding new jobs more quickly than in the past, thus reducing the
first of these costs. Among workers displaced in the 1979-83 and 1981-85
periods, 60 percent and 67 percent were reemployed by 1984 and 1986,
respectively. In contrast, 68 percent and 74 percent of workers
displaced in the 1991-93 and 1993-95 periods were reemployed by 1994 and
1996, respectively, even though the shorter time period should have
produced lower reemployment rates. The shift in the composition of
displacement, from less educated to more educated workers, may explain
some of the increase in reemployment probabilities, as more schooling
generally helps ease workers' adjustment into alternative career paths.
Dislocated workers who find new full-time jobs often suffer a
lingering decline in real earnings. Some evidence indicates that 6 or
more years after displacement, the median displaced worker's earnings
remain roughly 10 percent below what that worker might otherwise have
expected to earn. That figure does not appear to have changed much over
time. More educated workers appear to face smaller displacement costs,
as their earnings losses are smaller than those of less educated
workers. Furthermore, currently almost 15 percent of reemployed workers
who had health insurance at their old jobs receive no such coverage from
their new employers. However, this represents a considerable improvement
from the early 1980s, when over one-quarter of previously insured
displaced workers did not receive health insurance at their new jobs.
Nevertheless, the costs of displacement are substantial for a large
number of workers.
Taken as a whole, these results suggest that any sense of greater
vulnerability to job loss is likely to be the result of a broadening of
the risk of job displacement to groups of workers who had been
relatively immune. Among those who do lose their jobs, the adjustment
difficulties that follow job displacement are actually modestly less
than in previous years.

JOB STABILITY

A number of prominent U.S. firms that used to maintain policies of
``lifetime employment'' for their workers have recently abandoned those
policies. These well-publicized reversals may have led to the widespread
perception that jobs in general are less stable than they used to be.
However, jobs at these firms probably never comprised more than a very
small share of national employment. To arrive at a more accurate picture
of job stability in the United States, one needs to examine the evidence
for the labor market as a whole.
One well-known study that explored job duration in the 1970s found
that many workers could reasonably look forward to ``lifetime jobs.'' A
significant proportion of workers held their jobs for 20 years or more.
A more recent investigation shows that lifetime jobs are just as
prevalent in the 1990s as they were during the 1970s. For instance, in
1993 the median 45- to 54-year-old male worker and the median 55- to 64-
year-old male worker had been employed at their current jobs for about
12 and 14 years, respectively, and over one-quarter of both groups had
held their jobs for 20 years or more. These statistics are virtually
identical to those obtained in several different surveys throughout the
1970s and 1980s (Chart 4-9), refuting the notion of a widespread
reduction in employment stability.


Job durations have changed for certain demographic groups, however.
In particular, the trend toward greater female labor force participation
is likely to have contributed to greater job tenure among currently
employed women. Conversely, employment stability appears to have
declined for high school dropouts.
A study conducted by the National Commission on Employment Policy,
however, reported potentially contradictory evidence regarding job
stability for the workforce as a whole. Using a method similar to the
study just discussed, this study also found no change in employment
stability. It also implemented an alternative method, using longitudinal
data for each year of the 1970s and 1980s and examining respondents'
answers to the question, ``Did you have another main employer during the
previous 12 months?'' It found that the share of workers reporting
having had another main employer two or more times in that period had
increased between the two decades. Because these data pertain to other
main jobs, they do not necessarily provide direct evidence for job
stability on the first job when workers hold multiple jobs.

WORKER ANXIETY

How have workers responded to the changes facing them in the labor
market? Press reports suggest that a prevailing general sense of
economic uncertainty has led workers to worry about their own prospects
in the labor market. Researchers can get a reading of workers' anxiety
over their economic circumstances in at least two ways. Public opinion
polling directed at workers' sense of job security is one approach.
Another is to examine aspects of worker behavior that are linked to
feelings of security.

PUBLIC OPINION POLLS

For more than two decades, a leading nationwide opinion research
organization has been asking workers, ``How likely is it that you will
lose your job over the next 12 months?'' The proportion of respondents
who believed that they were ``not at all likely'' to lose their jobs was
lower in 1996 (51 percent) than in 1991 (about 60 percent), even though
the economy was suffering through a recession in the earlier year. In
fact, the low proportion of workers with this strong sense of job
security in 1996 is similar to the unusually low level reached in 1983,
shortly after the unemployment rate peaked at nearly 11 percent during
the worst recession since World War II. However, the decline in the
share of respondents who considered job loss ``not at all likely'' has
been mirrored by an increase in the share saying that it was ``not too
likely.'' The share saying that it was ``very or fairly likely'' that
they would lose their jobs has changed little. Accordingly, these polls
suggest that more people than before are feeling a moderate, but not a
high, risk of job loss.
At the same time, workers also express a perception that jobs are
readily available. For many years a national business association has
surveyed individuals about their views on the availability of jobs. The
pattern of their responses has closely matched trends in unemployment.
Another survey of consumer sentiment, conducted by the University of
Michigan, also shows that consumer perceptions about the job market are
consistent with economic conditions prevailing at the time.
Appropriately, the current low level of unemployment is reflected in
recent results from both these surveys, which indicate that workers are
not overly concerned about job availability.

QUIT BEHAVIOR

Workers do appear to have changed their behavior in ways that are
consistent with feelings of increased anxiety about their jobs. In
particular, workers have become more reluctant to quit their jobs.
Typically, during periods of prosperity, workers employed in jobs they
feel are a bad match for them often quit to look for new work for which
their skills would be more appropriate. Quits generally fall during
recessions, when new jobs are harder to find. For any two comparable
points in the business cycle, a lower overall quit rate may indicate
greater worker anxiety, because it suggests that workers fear they will
not be able to find or keep a new job if they quit their current one.
One measure of how many workers are quitting their jobs to look for
new work is provided by the CPS, which reports the percentage of the
labor force that has become unemployed within the previous 5 weeks
because of having quit. Chart 4-10, which plots this measure, shows the
expected strong cyclical pattern to quit behavior. The current expansion
is no exception, although the rise is less sharp than that in the
previous expansion, and quits fell slightly last year. Five years into
the current expansion, quits are still considerably less prevalent than
in the 1970s or 1980s--a finding that is consistent with lingering
worker anxiety.


POLICIES TO MITIGATE THE COSTS OF ECONOMIC CHANGE

The Federal Government has many policies and programs at its
disposal to reduce the costs that economic growth and change impose on
some workers. The main policy instrument that addresses some of the
immediate needs of workers who lose their jobs is the unemployment
insurance system. Other policies, such as mandatory advance notice of
layoffs, may provide short-term benefits as well. Still other policies,
including education and training programs, are vital for improving the
longer term fortunes of those hurt by economic change. These are
discussed briefly here and in more detail in Chapter 5.

UNEMPLOYMENT INSURANCE

Created in 1935 as part of the Social Security Act, the UI system
has two main goals: to work as an economic stabilizer, expanding
consumer spending during periods of heavy job loss, and to provide
economic security for workers through income maintenance. The Federal
Government maintains control over the broad design of the UI system, but
States have considerable autonomy in tailoring the program's features
within their jurisdictions. UI provides weekly benefits to workers who
have been laid off or who have lost their jobs for reasons other than
misconduct or a labor dispute. Only workers with a sufficiently long
employment history (usually two calendar quarters of significant
employment) are eligible. Benefits are a fraction of average weekly
earnings on the job that was lost, up to a maximum dollar amount, and
paid up to 26 weeks in most States. This fraction, called the
replacement rate, is typically between 50 and 70 percent. Benefits are
financed, in most States, by a payroll tax levied on firms.
UI benefits help workers weather periods of unemployment, since the
benefits allow workers to maintain consumption patterns closer to those
observed prior to the job loss. Another potential benefit of the UI
system is that it may improve the match between workers and firms upon
reemployment: UI may provide individuals the financial resources to
prolong their job search until they receive an offer appropriate to
their skills. However, little empirical evidence supports the
proposition that longer search periods translate into better job
matches, as measured by higher future earnings.
Although the UI system has benefited millions of workers over the
years, these benefits do not come without costs. In particular, a
significant body of evidence supports the contention that higher UI
benefits lead to longer unemployment spells. Providing benefits to
unemployed workers reduces their incentive to search intensively for a
new job. Research suggests that a 10-percentage-point increase in the
replacement rate of UI benefits leads to an additional 1 to 1.5 weeks of
unemployment, when an insured unemployment spell typically lasts roughly
15 weeks. Job-finding rates also increase somewhat as the exhaustion of
benefits approaches.
Some States and the U.S. Department of Labor have investigated
whether changes in the UI program can reduce unemployment durations and
improve subsequent employment outcomes in a cost-effective manner. The
research was undertaken in the form of controlled experiments. Workers
were randomly assigned to treatment and control groups; those in the
control groups received benefits under the rules of the existing
program, while treatment-group participants were subject to an
alternative, experimental set of rules. With random assignment, members
of the different groups can be assumed to have similar characteristics,
so that any differences in outcomes can be attributed specifically to
the difference in policy.
The first set of experimental policies included job search
assistance. Treatment-group members were eligible for services such as
instruction in how to find a job, and for periodic meetings with
employment counselors. These programs were generally found effective
both in reducing unemployment durations and in increasing earnings
during the first year or two following reemployment. One difficulty in
interpreting the results, however, is that one cannot be sure whether
the favorable effect was caused by the job search services themselves or
by the more rigorous monitoring of worker search activities that
accompanied them. Nevertheless, the apparent success of these
experiments led the Congress to pass a law in 1993 requiring States to
institute job search assistance for workers identified as likely to be
hard to place.
States have also experimented with paying reemployment bonuses to
workers who find jobs within a certain period after filing a UI claim;
self-employment assistance programs with UI payments as support; and
training programs targeted at dislocated workers. Of these, only the
self-employment assistance programs yielded generally positive results.
The proportion of unemployed workers starting their own businesses
roughly doubled, although it remains quite low. Over an 18-month follow-
up period, failure rates for these businesses were no different from
those observed for businesses started by control-group members.
The reemployment bonus experiments yielded mixed results: in some
but not all cases the savings in reduced UI benefits exceeded the costs
of bonus payments and additional administrative expenses. It is also
possible, however, that a more widespread use of bonuses would increase
the share of workers filing UI claims.
Short-term training programs generally have not been as successful
as other policies in improving the labor market outcomes of dislocated
workers--a result that contrasts with the findings of similar programs
targeted at low-income, low-skilled workers. Programs to support longer
term training--for example, those that provide funding for higher
education--may yield significant benefits, but no formal, controlled
experiment has so far examined such programs.
Changes in the economy have also had profound effects on the UI
system. Most notably, the share of unemployed workers who received UI
benefits has fallen dramatically since the early 1980s. This reduction
has been attributed to demographic shifts in the workforce, a reduction
in union membership, regional shifts in employment, and tightened State
eligibility requirements. Payment of extended benefits during recessions
(beyond the regular maximum duration) has been less likely, because the
trigger that starts these payments is tied to an insured unemployment
rate that now is a less reliable indicator of economic conditions. As a
consequence, during the last two recessions the Congress authorized
temporary emergency programs that did not depend on the extended
benefits triggers. Such ad hoc adjustments may not be well timed to the
beginnings and ends of recessions. The question of which is the correct
trigger to use for this program has resulted in changes in the law,
which now authorize States to adopt a total unemployment rate trigger
for the extended benefits program if they so desire.
In addition, inflation has significantly eroded the value of the
taxable wage base, upon which UI taxes are imposed. The Federal wage
base, currently set at $7,000, is not indexed for inflation and has
fallen dramatically in real terms. (Although many States have a higher
base, it is less than $10,000 in most larger States.) Early in the life
of the UI system, in the late 1930s, the taxable wage base was set at
$3,000 (over $35,000 in 1996 dollars), and only relatively small,
infrequent adjustments have been made since then. Such a low base makes
the UI tax similar to a head tax that is disproportionally levied on
firms that employ low-wage workers. The nominally rigid taxable wage
base, combined with the fact that UI benefits are indexed in many States
and increased regularly in others, requires periodic adjustments in
State UI tax rates.

ADVANCE NOTICE

Another way to reduce the costs of job loss is to require firms to
give advance notice to workers about to be displaced. Prenotification
has a variety of potential benefits. It gives individuals time to search
for a new position while still working, which may shorten unemployment
spells or prevent them altogether. Other types of adjustment assistance
(e.g., job counseling, skills retraining, or outplacement assistance)
may also be more effective and easier to administer if individuals are
still reporting for work. Finally, if the notice is given sufficiently
far in advance, workers may be able to switch their human capital toward
skills that are likely to be useful to their future employers. Although
legislation requiring advance notice has been enacted, a variety of
exemptions limit the number of firms required to provide notice. It is
unclear whether the legislation has increased the share of workers who
are actually notified.
For those displaced workers who receive it, advance notice does
appear to reduce adjustment problems. Recent studies suggest that
individuals receiving at least 2 months of advance notice are out of
work up to 1 week less and earn around 10 percent more in their new jobs
than do those receiving no notice. Despite only modest reductions in
joblessness, pay might increase through at least two mechanisms. First,
employers who provide advance notice may also tend to provide other
forms of readjustment assistance that might lead to wage gains upon
reemployment. Second, notified workers remain jobless almost as long as
other workers, but may find new jobs that better match their skills and
qualifications. The available evidence lends support to both of these
possibilities.

REFORMING TRAINING AND REEMPLOYMENT SERVICES

Both the Administration and the Congress have proposed consolidating
many of the roughly 100 separate education and training programs now
administered by the Departments of Labor and Education and reforming the
overall system. Some of the proposed reforms are intended to help
dislocated workers. A crucial element is the establishment of one-stop
career centers where workers can find out about employment opportunities
and training programs and apply for unemployment benefits. These centers
are already being established in many States.

PORTABILITY OF PENSION AND HEALTH CARE BENEFITS

The costs of job transition are higher than they need to be in part
because of the frictions involved in transferring pension and health
care benefits. This is a significant cause of ``job lock,'' in which
workers are reluctant to leave their current jobs because they fear they
will not be able to transfer their benefits. Many of these frictions can
be eliminated, and recently some important strides down this path have
been made. The minimum wage legislation passed in August 1996 contained
a pension simplification initiative aimed at making portable pensions
more available. New Internal Revenue Service regulations seek to do the
same. Another recent success is the enactment of the Health Insurance
Portability and Accountability Act of 1996 (the Kassebaum-Kennedy bill),
which ensures continued health care coverage for workers with
preexisting conditions who lose or change their jobs. The Administration
has also proposed continuing health insurance for unemployed workers.
Such a policy would further the goal of reducing the frictions
associated with changing jobs.

CONCLUSIONS

Over the long run, sound economic policies that lead to low levels
of unemployment and high rates of economic growth are likely to produce
gains for most workers. Technological change and an increasingly
competitive marketplace also help promote the conditions necessary for
such growth. Most of the available evidence suggests that the U.S. labor
market is quite robust, with significant job growth in the higher paying
sectors, some evidence of reduced job loss, and a level of job stability
that probably is no different today from what it was 20 years ago.
Nevertheless, some costs have been incurred. Government has a role
in lessening the burden that economic growth causes for some workers.
Some policies have been put in place, and others have been proposed,
that should help reduce these costs without sacrificing growth in the
economy.
One important potential cost of economic growth that this chapter
has not addressed is increased inequality: the danger that those at the
bottom of the earnings distribution will find themselves falling ever
further behind the rest. Chapter 5 explores issues of inequality in far
greater detail.