The U.S. labor market in 2003: signs of improvement by year's end

In 2003, total nonfarm employment continued to decline until late in the year. The unemployment rate rose during

Several factors affected employment trends last year. Lingering effects from both the late1990s technology bubble and the 2001 recession, as well as related State and local government budget crises, led to continued job losses or to rates of employment growth slower than those seen in recent years. Strongly competitive markets drove ongoing structural change in several industries, while lackluster tourism translated into little or no growth in some tourismrelated industries. In contrast to an otherwise weak labor market, historically low interest rates were a catalyst to job growth in interest-rate-sensitive industries. The data in this article are primarily from the Current Employment Statistics (CES) survey, the Current Population Survey (CPS), and the local Area Unemployment Statistics (LAUS) program. (See page 4 for an explanation of differences between the CES survey and the CPS.)

Total nonfarm employment declined through the first three quarters of the year and then registered small gains. Over the year, net payroll employment was down by 243,000. Manufacturing lost 642,000 jobs, the largest drop of any sec tor. Information, transportation and warehousing, and wholesale trade also posted employment declines. Job losses were not unique to private industry: government employment contracted for the first time since 1982. Private education and health services added 329,000 workers in 2003, more than any other sector, although the rate of growth slowed relative to the rate in recent years. Employment in professional and business services, financial activities, and construction grew faster in 2003 than 2002. Largely on the basis of strength in food services, the leisure and hospitality sector also added workers, while retail trade and the natural resources and mining sector showed little change in employment.

Although the 2001 recession had ended, many of the payroll survey's most sensitive cyclical series did not show signs of improvement until late 2003. While real gross domestic product, consumer and business spending, and the stock markets each grew at an accelerated pace after the 2001 recession, total nonfarm employment continued to contract through the third quarter of 2003. The lack of job growth contributed to a sense of uncertainty regarding the strength of the recovery from the recession by raising concerns about consumers' ability to continue spending.1 Earnings growth slowed over the year, while average weekly hours for all private industry dipped to an all-time low in mid-2003 and ended the year flat relative to 2002.

Worker hours in manufacturing often lead the business cycle. Average weekly hours in manufacturing hovered near a recession low for much of 2003 before lengthening in the last quarter. Weekly overtime hours in manufacturing had bottomed out at the end of 2001, but growth in overtime stalled in mid-2002, and the workweek did not lengthen again until the end of 2003. Over the year, the manufacturing workweek expanded by 0.2 hour and overtime by 0.3 hour.

The temporary help services industry supplies labor to a wide variety of industries, and its employment trend is often considered a leading indicator of payroll employment. Employers who face uncertain demand frequently hire temporary workers before hiring permanent workers. The temporary help industry showed clear signs of recovery in 2003 as its employment began to expand in the second quarter. Over the year, the industry added 133,000 workers.

Trucking activity is often considered a cyclical indicator, because it fluctuates with the demand for goods. Truck tonnage is an important measure of the quantity of goods shipped via truck, while employment in truck transportation reflects the industry's labor needs. Although tonnage remained below its long-run growth trend in 2003, it strengthened relative to its recent recession levels.2 Despite some recovery in employment during the second half, employment in truck transportation was flat over the year.

Activity in machinery manufacturing also is considered a cyclical indicator, because the industry sells its products to other manufacturers and its expansion generally coincides with increased capital investment. Industrial production of machinery grew in the last quarter of 2003.3 Over the year, activity was up slightly. Despite the upturn in output, employment in machinery declined through the fourth quarter of 2003. The industry lost 59,000 jobs in 2003, although the pace of job loss slowed as the year progressed.

The job situation in the highly cyclical securities industry also showed relative improvement. After contracting by 44,000 jobs in 2002, securities, commodity contracts, and investmerits employment leveled out in 2003. (See table 1.) Demand for financial services corresponded with strengthening in the U.S. stock markets; the Standard and Poor's 500 bottomed out early in the year.

It was not until the final quarter of 2003 that payroll indicators aligned more clearly and hinted at a labor-market recovery. After falling in each of the first three quarters, total nonfarm employment edged up in the fourth. (See chart 1.) Promisingly, the job gains were spread among a number of industries.4 Aggregate weekly hours for total private industry also appeared to have bottomed out in the third quarter. Another positive factor was relative improvement in manufacturing employment: although job losses in the industry continued through the fourth quarter, the pace and scope diminished in the second half of the year. (See chart 2.) In particular, job losses in durable goods manufacturing slowed dramatically in the fourth quarter; on average, industries in this sec tor lost 133,000 jobs each of the first three quarters, compared with 23,000 jobs lost in the fourth quarter.

Lingering effects from both the 1990s technology bubble and the 2001 recession continued to influence the employment situation in 2003. Telecommunications and information-technology-related industries continued to deal with overcapacity, while industries associated with commercial construction suffered from lackluster spending. all of these industries shed workers in 2003.

Driven by high expectations of growth in demand, telecommunications companies quickly expanded their transmission capacity during the 1990s. When it later became clear that these expectations would not be met, companies speedily cut their prices to capture as much demand-and revenueas possible. Facing stiff competition and hefty debts, telecommunications companies also shed workers as they struggled to profit.5 Employment in telecommunications reached a peak in the first quarter of 2001 and shrank 19.9 percent through 2003. The industry lost 78,000 jobs in 2003, the largest portion of job losses in the information sector. While the job situation in telecommunications continued to weaken in 2003, the pace of job loss slowed over the year. Employment declines were concentrated in wired telecommunications carriers-an industry which includes companies that facilitate communications via landlines-and in telecommunications resellers.

Spurred by the popularity of Internet and e-mail applications, as well as by concerns of a year-2000 computer glitch, investment and spending on information processing equipment and software grew an average of 20 percent per year between 1995 and 2000. The information-technology-related components of manufacturing, namely, computer and peripheral equipment, communications equipment, and semiconductors and electronic components, together averaged 2.4percent annual employment growth over the 5-year span. Spending on information processing equipment reached a high in the last quarter of 2000 and contracted each quarter of the next year.6 The peak in information-technology-related manufacturing employment lagged spending by one quarter, although the industry continued to shed workers through 2003, 2 years after spending had begun to recover. Employment did show relative improvement in 2003, as the pace of job loss slowed by one-third compared with 2002.

The job situation in computer systems design and related services also mirrored the technology bubble's expansion and contraction. After reaching a high in the first quarter of 2001, employment in computer systems design fell 17.5 percent from the second quarter of 2001 to the end of 2003. The pace of job loss moderated over the year, although the industry still lost 22,000 jobs. The practice of outsourcing computer-related jobs offshore is yet another factor underlying the industry's weak employment situation.7

The technology bust and the 2001 recession were characterized by cutbacks in business spending. Besides slashing spending on information processing equipment and software, businesses reduced their investment in structures. The rate of expansion of telecommunications infrastructure dropped once the industry revised its expectations of future demand. Employment in utility system construction, which includes construction of communication lines and related structures, followed this boom and bust: from the second quarter of 2001 through mid-2003, the industry shed 30,000 jobs; after that, construction improved somewhat. The result was that, over the year, employment was flat.

Architectural and structural metals manufacturing, which produces materials such as prefabricated metal buildings and concrete-reinforcing bars for commercial construction projects, shed 13.4 percent of its workforce from 2000 through mid-2003. Job losses moderated in 2003, although the industry still shed 10,000 jobs over the year. Employment in commercial building construction contracted 7.1 percent from mid-2001 through 2003 and declined by 16,000 over the year. Employment in industrial building construction dropped by 8,000 in 2003, marking the industry's third consecutive year of decline.

Long-term structural change driven by foreign and domestic competition, as well as technological advances, led to continued job losses. Manufacturing employment declined by 4.3 percent over the year. Until the fourth quarter, the losses were more severe in durable goods manufacturing; for the year, the pace of job loss exceeded 3 percent in both durable and nondurable manufacturing. In recent years, cyclical factors exacerbated manufacturing's long-run employment decline, and the pace of job loss accelerated, with the industry having lost 2.9 million jobs since 2000. Foreign competition has played an important role, both in driving the adoption of laborsaving technologies in the United States and in moving manufacturing jobs offshore. This relationship is evidenced by the long-run growth of both manufactured imports and industrial production, coupled with manufacturing's downward trend in employment.8

Primary metals manufacturing lost 35,000 jobs in 2003, continuing an employment slide that accelerated markedly in 2001. Intense competition among U.S. and foreign competitors has forced the steel industry to consolidate in recent years by shedding excess capacity and workers. In March 2002. steel tariffs were implemented in order to protect the industry from foreign competition while it underwent major restructuring. After the tariffs took effect, the industry's pace of job loss slowed from 12.6 percent in 2001 to 7.2 percent, on average, in 2002 and 2003.9 In December 2003, the tariffs were lifted, reducing the likelihood of retaliatory trade actions from foreign steel-producing countries.

Although conditions in the steel industry improved after the tariffs were put into place, motor vehicle parts manufacturers were hurt by higher steel input prices. With profits squeezed between higher steel costs and automobile producers' demands for low-priced parts, the industry shed 30,000 jobs in 2003. Data on producer prices illustrate this constriction well: the cost of steel mill products rose 11.3 percent from 2001 to 2003, while prices received by producers of motor vehicle parts and accessories shrank 1.3 percent over the same period.10

In contrast to motor vehicle parts manufacturers, motor vehicles manufacturing added 7,000 workers over the year, after shedding employees the previous 2 years. Before the labor negotiations that began in july 2003, the "Big 3" auto producers were locked into contracts that restricted them from closing plants. Suffering from overcapacity, high retiree health and pension costs, and a shrinking market share, the Big 3 convinced the United Auto Workers labor union of the need for closings. New contracts were ratified in September 2003, and the first of several plant closings occurred the following month." While the traditional automobile industry struggled in 2003, foreign automobile producers expanded their market share and production in North America.12

Apparel, textile, and leather products manufacturing industries have each been shedding workers for several decades. Apparel shed 48,000 workers, and textile mills, which manufacture basic fibers into products such as yarn and fabric, lost 38,000 jobs over the year. Textile product mills, which purchase primary textiles and manufacture nonapparel textile products such as sheets and towels, also shrank in 2003, as the industry cut 9.4 percent of its workforce. Employment in leather and allied products decreased by 5,000 in 2003, a decline that was part of an ongoing contraction that has separated about two-thirds of the industry's workers since 1990.

Competition from foreign producers is one factor that has led to the employment declines. Indeed, U.S. industrial production of textiles contracted each year from 2000 to 2003, and the production of apparel and leather goods declined each year from 1997 to 2003. '3 Also in recent years, imports of textiles and of leather and allied products have grown.14 Automation is yet another factor affecting employment trends in these industries.15

Paper and paper products manufacturing lost 25,000 jobs over the year, and employment in the industry has fallen by 129,000 since 1994. Over the long term, the industry has struggled with overcapacity and increased global competition and has shed jobs through consolidation. The pace of job loss accelerated during the 2001 recession.16

The U.S. Postal Service, too, has closed and consolidated several facilities in recent years. Since mid-1999, the Postal Service has cut 10.9 percent of its workforce. In 2003, employment declined by 22,000 workers. In recent years, the Postal Service has reduced its labor needs by adopting technologies that facilitate automated mail sorting. Decreased mail volume also has reduced the need for workers. Mail volume has shrunk due to the popularity of on-line bill paying, e-mail, increased competition, and economic "softness."17 Couriers, an industry that includes private parcel delivery, has also become more automated in recent years. This industry has shed 58,000 workers since 2000, although, in 2003, employment was flat.

Automation, such as self-scanning equipment at checkout counters, has become more prevalent in food and beverage stores in recent years and is one factor behind the industry's job decline.18 Fierce competition is another factor. Employment in food and beverage stores has contracted by 180,000 since reaching a peak in mid-2000; employment fell by 43,000 in 2003. Shrinking employment in grocery stores has driven the decline, although specialty food stores also have shed workers in recent years. Warehouse clubs and supercenters, which sell groceries as well as general merchandise, have captured a large portion of grocery stores' traditional markets.19

Employment in warehouse clubs and supercenters expanded by 63,000 over the year. Bargain-hunting and convenience-seeking consumers have supported growth in this industry by increasingly shopping at warehouse stores and supercenters, rather than at more traditional stores. Indeed, warehouse clubs and supercenters have registered doubledigit sales growth in recent years, while traditional-style department stores have struggled to compete.20 The job situation in department stores, except discount stores reflects the woes of traditional-style department stores, as 55,000 jobs were lost over the year. Since reaching a high in mid-2001, employment in the industry has contracted by 17.3 percent. Over the same time span, employment in discount department stores has been relatively flat. Unlike traditional-style department stores, discounters generally have cashiers located at stores' exits; they differ from warehouse clubs and supercenters because they do not sell groceries. Despite dynamic job trends within the retail trade sector, overall retail employment was essentially flat in 2003.

The public sector was not immune to the weakening job situation in 2003, and government employment contracted for the first time in more than two decades. (See chart 3.) Although cutbacks by the U.S. Postal Service account for a portion of the job decline, employment in the Federal Government, except U.S. Postal Service also fell over the year. Job growth stalled in State government and local government.

Strong tax revenues during the 1990s facilitated escalated government spending. However, when the 2001 recession hit, tax revenues shrank considerably and pushed government budgets out of balance. State governments also were saddled with growing unemployment compensation and Medicaid costs, while both State and local governments were burdened by increased security costs.21 In addition, nearly all State and local governments are required, by law, to balance their annual budgets.22

One-time accounting fixes that shifted spending from one fiscal year to the next, increased fees, and tobacco settlement funds helped States avoid job cutbacks overall in 2002.23 With continued budget woes in 2003, however, States were forced to make difficult spending decisions. In State government, excluding education, the rate of job loss accelerated over the year and employment declined by 38,000. Appreciating property values drove local tax revenues and helped local governments balance cutbacks in State aid. Still, employment in local government, excluding education grew only 16,000 over the year, compared with an average growth of 120,000 workers per year from 1997 to 2001.

Job growth in education, both public and private, slowed in 2003. Employment in local government education was flat over the year, after adding an average of 183,000 jobs per year from 1997 to 2001. Employment in State government education, which includes mostly public universities and colleges, grew by 38,000 over the year, compared with an average growth of 68,000 per year from 1997 to 2001. Employment in private educational services grew by 53,000 in 2003, about one-half of the category's growth in 2002; both primary and secondary schools, as well as private colleges, experienced slower growth over the year. Private colleges suffered from shrunken endowments and cutbacks in donation levels, while public universities and colleges faced reductions in government funding.24

Health care and social assistance services also rely on government support, and in recent years health-care-related industries have faced cuts in Medicare and Medicaid reimbursement rates. The reduced benefits to these industries coincided with increased spending on liability insurance and disaster readiness and also with slower employment growth.25 Social assistance maintained its 2002 rate of growth in 2003, adding 53,000 jobs over the year. Overall, the pace of job growth in health care and social assistance slackened in 2003, but the industry still experienced considerable growth and added more jobs than any other industry.

Tourism remained muted in 2003, as did employment in related industries. Transportation and warehousing shed 57,000 workers over the year, with air transportation accounting for the majority of the decline. Airlines continued to struggle with overcapacity in 2003. Geopolitical and terrorist concerns played a role in the lackluster demand for air travel, especially in the first half of the year. Beginning in 2001, reduced business travel crimped the airlines' revenues. A sharp decline in air travel after September 11, 2001, further intensified the airlines' difficulties. With its customers financially distressed, aerospace product and parts manufacturers shared the airlines' pain,26 shedding 26,000 jobs in 2003. Overall, the industry has lost 83,000 jobs since the third quarter of 2001.

Leisure and hospitality added 87,000 workers over the year, with strength in food services bolstering weakness in accommodations and entertainment. Employment growth in food services and drinking places corresponded with strong sales in the industry.27 However, with business and leisure travel in a slump, the job situation in hotels and motels, except casino hotels remained lackluster. Since reaching a peak in 2000, employment in this industry has decreased by 137,000-55,000 alone in 2003. Arts, entertainment, and recreation shed workers, too, and reduced its employment by 14,000 over the year.

Low long-term interest rates fueled employment growth in mortgage- and housing-related industries, generating a pocket of contrast to an otherwise weak employment situation in 2003. Housing became more affordable as mortgage rates dipped to record lows that spurred both sales and refinancing activity.28

In 2003, housing starts increased from already high levels and drove job growth in construction. An employment increase of 61,000 in specialty trade contractors reflected greater demand for residential construction. The pace of newand existing-home sales accelerated over the year and drove employment growth in real estate.29 The job situation in building material and garden supply stores also reflected the housing market's strength. The industry's employment expanded by 31,000 over the year, marking the 12th consecutive year of job growth. Sales at building material and garden supply stores accelerated slightly in each of the last 3 years, reaching 6.6 percent in 2003.30

Refinancing activity reached historically high levels as homeowners took advantage of low mortgage rates. Credit intermediation and related activities added workers to meet the increased demand for refinancing, and employment swelled by 60,000 over the year. All of the industry's growth occurred in the first three quarters of the year; by the final quarter of 2003, refinancing activity had slowed somewhat, and the credit intermediation industry cut back its workforce.

Unemployment rose in mid-2003, but began to decline at year's end. Estimates from the CPS showed that the unemployment rate in the fourth quarter of 2003, 5.9 percent, was unchanged from the jobless rate in the fourth quarter of 2002. After having ticked up to 6.1 percent during the second quarter of the year, the unemployment rate began a slight downward movement during the last half of the year. (See chart 4.) Unemployment rates for all the major demographic groups-adult men, adult women, teens, whites, blacks, and Hispanics-peaked at midyear and then declined to their 2002 levels, or below in the case of Hispanics. (See table 2.) The unemployment level also was little changed, on net, over the year, and the number of unemployed persons totaled 8.6 million by year's end.

The number of employed persons rose fairly steadily by 1 million in 2003. Employment growth, however, was slower than population growth, and the employment-population ratio fell by 0.2 percentage point from the fourth quarter of 2002. The employment-population ratio declined throughout 2003, to a rate of 62.3 percent in the fourth quarter, the lowest level in 10 years. Since the end of the recession in the fourth quarter of 2001, the employment-population ratio has declined by 0.7 percentage point.

The labor force grew by 1 million from the fourth quarter of 2002 to the fourth quarter of 2003, due mainly to the increase in employment over the year. The labor force participation rate fell during the year from 66.5 percent in the fourth quarter of 2002 to 66.1 percent in the fourth quarter of 2003. The drop was caused by population growth that outpaced the growth in the labor force.

Adult women and men (those aged 20 years and older) fared somewhat better than teenagers during 2003. The unemployment levels of both men and women did not increase significantly over the year, and the unemployment rates for both sexes were unchanged as well, at 5.5 percent and 5.1 percent, respectively. Nor did labor force participation rates or employment-population ratios change much for both men and women over the year. The participation rate for women was 60.5 percent in the fourth quarter of 2003, and the rate for men was 76.0 percent. After a half century of relatively steady growth, women's labor force participation has moved little since 1998. Employment for adult women increased by 505,000 over the year, while that for men increased by 858,000, largely on the strength of the fourth quarter.

Continuing a declining trend, the number of teenagers in the labor force fell by 418,000 over the year, to a level of 7.0 million, and the labor force participation rate of teenagers declined by 3.2 percentage points, to 43.6 percent, close to the lowest ever recorded for this group. Teenagers are less likely to participate in the labor force during economic contractions, and the teen labor force participation rate had been declining since the mid-1990's, due at least in part to increased rates of school enrollment.31 The employment level for teenagers declined by 349,000 over the year, to 5.9 million, and the group's employment-population ratio declined by 2.6 percentage points, to 36.5 percent. At 1.1 million, the unemployment level for those aged 16 to 19 was about unchanged over the year. The unemployment rate for teenagers peaked at 18.3 percent in the second quarter of 2003. This was the highest unemployment rate for the group since the second quarter of 1993. The rate began falling in the third quarter of 2003 and ended the year at 16.3 percent, the same as in 2002.

Among the major racial and ethnic groups, Hispanics saw improvements in their unemployment level and rate, while the levels and rates for whites and blacks were little changed over the year. By year's end, the number of unemployed blacks was 1.8 million, and their unemployment rate was 10.7 percent, both measures little changed from 2002. The black unemployment rate was more than double the unemployment rate for whites. The number of employed blacks edged down by 126,000 over the year, to 14.7 million, and the employment-population ratio for the group fell from 57.7 percent to 57.0 percent. While the employment level for whites was unchanged over the year, Hispanics saw an increase of 523,000. Still, at 63.1 percent, the Hispanic employment-population ratio was little changed over the year. Unemployment levels for whites were about unchanged over the year, while the unemployment level for Hispanics dropped by 326,000. The unemployment rate for Hispanics fell by 0.8 percentage point, to 7.1 percent, between the third and fourth quarters of 2003, after having stayed flat throughout the first three quarters of the year. The unemployment rate for whites was about unchanged over the year and stood at 5.1 percent in the fourth quarter.

The labor force participation rates for most of the major age groups dropped markedly during the latest recession and during the initial recovery, relative to the previous recession in the early 1990s.32 (See chart 5.) The sharpest decline in the participation rate during the most recent period occurred among youths 16 to 24 years of age. Women in the prime working-age group, 25 to 54 years, also experienced a large decline in their participation rate relative to their change in participation during the early 1990s. The labor force participation rate for women in that age group increased during the previous downturn of 1990-92 and decreased during the most recent recession and recovery period.33

These data are supported by Social security Administration data which show that the number of women under age 55 who received Social security disability benefits more than doubled over the same period.

For further discussion of these women's declining participation rates, see Steven Hippie, "Labor Force Participation during Recent Labor Market Downturns, "Issues in Labor Statistics, Summary 03-03, September 2003.

The lone exception to the trend of declining labor force participation has been among persons aged 55 and older. This group's participation rate increased by 3.2 percentage points over the most recent period, compared with a decrease of 0.7 percentage point during the previous recession.

Workers at most education levels experienced little change in their unemployment rates over the year, except for high school graduates without college-level training.34 At 8.5 percent, the unemployment rate of persons with less than a high school diploma was higher than that of persons with more education. (See chart 6.) In the fourth quarter of 2003, persons with a bachelor 's degree or higher had an unemployment rate of 3.0 percent, a figure that was unchanged over the year after having doubled between 2000 and 2002. The 4.7-percent unemployment rate of those with some college training, but without a degree, also was little changed over the year. The only group for whom the unemployment rate rose over the year-by 0.3 percentage point, to 5.5 percent-was high school graduates with no college. The rate for this group reached a peak of 5.6 percent during the second quarter of 2003. (See table 3.)

Large employment gains in professional occupations and in installation, maintenance, and repair occupations accounted for much of the employment growth over the year. Employment in professional and related occupations grew by 840,000 during 2003.35 (See table 4.) The occupations within this category that experienced the largest employment increases were education, training, and library occupations and health-care practitioner and technical occupations. The gains in these occupations were shared by both women and men. Life, physical, and social science occupations also saw small increases in employment. The other occupations within the professional category were little changed over the year.

Natural resources, construction, and maintenance occupations posted an over-trie-year employment increase of 693,000. While all the components of this group rose over the year, about three-fifths of the gain occurred in installation, maintenance, and repair occupations.

The gains described for the preceding occupation groups were partly offset by relatively small employment declines in production, transportation, and material moving occupations. The number of persons employed in production occupations fell over the year, accounting for more than two-thirds of the overall decline in the broad group. The majority of persons employed in production occupations work in the manufacturing industry, which continued to experience substantial job losses in 2003. Employment in transportation and material moving occupations also declined over the year. Most of these workers are employed in the transportation industry, a cyclical industry that is still recovering from the 2001 recession. Employment in service occupations and in sales and office occupations was flat over the year.

The number of self-employed persons in nonagricultural industries rose by 312,000 over the year, to 9.5 million. Since the official end of the recession in the fourth quarter of 2001, self-employment has risen by 567,000, or 6 percent. In the recession of 1990-91, by contrast, self-employment was unchanged 2 years after the official end of the recession in the first quarter of 1991. (See chart 7.)

The number of persons working part time involuntarily increased in the second half of 2003, pointing to labor underutilization other than unemployment. One measure of underutilized labor is the number of persons who are working part time involuntarily, also referred to as those working part time for economic reasons. These individuals are persons who prefer a full-time job, but who cannot work full time due to slack work or business conditions or because they can find only part-time work. Over the year, the number of persons who were working part time for economic reasons increased by 480,000, to a level of 4.8 million. (See chart 8.) This number has increased by 1.5 million since the fourth quarter of 2000. The measure of persons working part time involuntarily is sensitive to business cycles, and the level increased more, proportionately, during the recession of 2001 than in the previous recession.

Broad measures of underutilization changed little over the year; the numbers of marginally attached and discouraged persons, rose in early 2003, but declined by year's end. In addition to estimating the number of persons who are employed and the number who are unemployed, the CPS provides data on persons who are not in the labor force, but who have some attachment to the job market. The labor force is defined as the sum of those who are employed and those who are unemployed, but who are actively seeking work. all other persons are considered to be outside of the labor force. Among these persons are some who indicate that they want a job, but who were not actively searching for work or were unavailable to take work. The number of persons outside the labor force who said they wanted a job was little changed over the year, at about 4.4 million. Another measure of underutilized labor is the number of persons who are marginally attached to the labor force. Marginally attached persons, a subset of those who say they want a job, are those who looked for a job sometime during the year prior to the survey and who would be available to take a job if one were offered to them. At 1.5 million, the number of marginally attached persons was up by 98,000 over the year.36 Discouraged workers, a subset of the marginally attached, are persons who have given up searching for work specifically because they believe that no jobs are available for them. The number of discouraged workers rose sharply in the first quarter of 2003, but then remained at about 450,000 throughout the year, about 69,000 higher than at the end of 2002. (See table 5.)

The majority of the unemployed had lost their jobs, although the number of job losers did not increase over the year. Those who were unemployed found it increasingly difficult to find work in 2003, as the average duration of unemployment was up over the year. The number of newly unemployed persons (those unemployed for 5 or fewer weeks) continued a downward trend that began in 2002. (See chart 9.) By the fourth quarter, the number of newly unemployed was 2.7 million, down 200,000 from a year earlier. This group made up just under one-third of total unemployment at the end of the year.

The number of persons who are newly unemployed typically rises at the beginning of a recession, when job losses start accumulating. After the recession is underway, unemployed persons who are unable to find work swell the ranks of the long-term unemployed.

The number of persons unemployed for 15 or more weeks increased by 292,000 over the year, to 3.5 million; the group made up about 40 percent of the unemployed. A subset of this group, the long-term unemployed (those unemployed for 27 or more weeks), increased by 197,000 over the year, to 2 million, and made up 23 percent of total unemployment. The 23-percent figure is the highest proportion of total unemployment that the long-term unemployed have been since the third quarter of 1983. The average number of weeks unemployed for all unemployed groups combined rose by 1.7 weeks over the year, to 19.7 weeks. (See table 6.)

The reasons persons are unemployed afford interesting insights into the labor market. Historically, most of the unemployed have been persons who have lost their jobs. Since this series began in 1967, job losers have made up roughly half of the unemployed. The series is sensitive to business cycles and has tended to peak during recessions. The proportion of the unemployed who had lost a job began trending upward in the second quarter of 2000 and continued this upward trend until the fourth quarter of 2001, when the percentage peaked at 55 percent. The proportion has remained at that level since then. (See chart 10.) By contrast, the number of persons who were unemployed because they left their jobs is negatively correlated with business cycles. Workers are less likely to leave a job when the prospect of finding a new one is more uncertain. Reentrants follow the same cyclical pattern as job leavers do over recessions, and new entrants have a weaker relationship to business cycles. At 2.4 million, the number of reentrants to the labor force was about unchanged, and the number of new entrants increased by 111,000 over the year, to 655,000.37 (See table 6.) Reentrants made up 28 percent of total unemployment, and new entrants constituted 7.6 percent. The number of persons unemployed because they had left a job was 835,000, and these job leavers made up about 10 percent of the unemployed.

The median usual weekly earnings of full-time wage and salary workers rose over the year, but the increase was offset by rises in consumer prices.38 From 2002 to 2003, median weekly earnings for all workers, as measured by the CPS, rose to $620, an increase of 2.0 percent. However, over the same period, the Consumer Price Index for All Urban Consumers (CPI-U) increased by 2.3 percent. Women's median earnings rose 4.3 percent over the year, faster than the men's 2.4-percent increase; still, women's earnings were $552, compared with $695 for men. Over the year, the ratio of women's earnings to men's earnings rose to 79.4 percent, up from 77.9 percent in 2002. Since 1979, the year the Bureau began collecting data on usual weekly earnings, the ratio of women's to men's earnings has increased by 16 percentage points. (Note that the comparisons of earnings set forth in this section are on a broad level and do not control for many factors that can be significant in explaining earnings differences.39)

All the major demographic groups saw earnings growth over the year, and all groups' earnings growth outpaced the rise in consumer prices, except for white men's. Among women, blacks had the largest earnings growth, 3.8 percent, followed by whites, at 3.7 percent. Hispanic women experienced a slightly lower earnings growth of 3.3 percent. Black men's earnings grew by 5.9 percent over the year, the largest increase in earnings among all the demographic groups. Hispanic men's earnings grew by 2.9 percent, higher than the white men's rate of 1.9 percent.

Workers in management, business, and financial operations occupations continued to have the highest median earnings of any intermediate occupation group, $961 per week in 2003. The group also had the second-highest earnings growth rate over the year, 4.2 percent. Professional and related occupations continued to be the second-highest paid group, with weekly earnings of $845 in 2003; the group's earnings increased at a rate of 2.7 percent. all the other occupation groups had earnings growth rates lower than the increase in consumer prices. The smallest growth rate in earnings, 1 percent, was among sales and related occupations. (See table 7.)

Among the four major educational attainment groups, high school graduates with no college and workers with a bachelor's degree or higher had earnings growth of 3.4 percent and 2.4 percent, respectively, both figures outpacing the increase in consumer prices.

Labor market conditions were little changed throughout much of the country in 2003. all four census regions and about two-thirds of the States and the District of Columbia recorded unemployment rate shifts of one-half percentage point or less between the fourth quarters of 2002 and 2003.40 (See box on page 21 for information on the methodologies used to develop unemployment data for local areas.) Among the census regions, jobless rates in the Northeast, South, and West were little changed or down slightly, following 2 years of increases. (See table 8.) In the Midwest, though, the unemployment rate ticked up again in 2003. The Midwest has been the region hardest hit by manufacturing job losses in recent years, and 2003 marked the fourth consecutive year during which its unemployment rate rose. In the fourth quarter of 2003, the South registered the lowest regional jobless rate. At the same time, the West continued to post the highest rate, for the 12th straight year.

States. Over the year, unemployment rates were up in 26 States and the District of Columbia, down in 20 States, and unchanged in 4 States. (See chart 11.) Michigan and Tennessee experienced the most marked increases in joblessness, 1.2 percentage points and 1 point, respectively. For 10 States, 2003 was the fourth consecutive year during which the unemployment rate increased relative to the previous year on a fourth-quarter-to-fourth-quarter basis. Half of these 10 States were located in the Midwest, including Michigan, where the jobless rate cumulatively more than doubled during the 4-year period. The most sizeable unemployment rate decreases in 2003 were reported by Mississippi and Arizona, 1.1 percentage points and 1 percentage point, respectively.

In the fourth quarter of 2003, Alaska recorded the highest unemployment rate, 8.1 percent. Michigan, Oregon, and Washington also posted rates of more than 7 percent. Unemployment rates were little changed or unchanged over the year in the three Pacific division States (Alaska, Oregon, and Washington), in contrast to the notable increase in Michigan. The lowest jobless rates at the close of 2003 were reported by North Dakota and South Dakota, 3.8 percent each, followed closely by Virginia and Nebraska, 3.9 and 4 percent, respectively.

Total nonfarm employment increased in 28 States and decreased in 22 States and the District of Columbia in 2003.41 (See chart 12.) Florida added the most jobs over the year (131,900). Nevada and Arizona recorded the next-largest job increases (40,600 and 32,500, respectively). In percentage terms, the most robust employment expansion occurred in Nevada (3.8 percent), followed at some distance by Hawaii (1.9 percent) and Florida and Wyoming (1.8 percent each). Alaska, Montana, New Mexico, North Dakota, and Wyoming are notable for being the only States that posted at least some job growth during each of the last 3 years on a fourth-quarter-to-fourth-quarter basis.

Two East North Central division States experienced the largest total nonfarm employment declines in 2003: Michigan (70,000) and Ohio (63,600). California, Illinois, Massachusetts, New York, Oklahoma, Pennsylvania, and Texas also shed more than 30,000 jobs each. On a relative basis, Oklahoma recorded the sharpest employment contraction (2.2 percent), followed by Massachusetts (1.8 percent) and Michigan (1.6 percent). For 17 States, 2003 was the third consecutive year of over-the-year total nonfarm employment losses. In Michigan, New York, and Ohio, the cumulative declines since the fourth quarter of 2000 exceed a quarter of a million jobs each.

Metropolitan areas.42 Thirty of the 42 most populous metropolitan areas saw their unemployment rates fall in 2003.43 (See table 9.) Three areas-Atlanta, Georgia; Phoenix-Mesa, Arizona; and San Jose, California-registered declines of more than a full percentage point each. Of the seven areas where jobless rates rose, only Charlotte-Gastonia-Rock Hill, North Carolina-South Carolina; and Detroit, Michigan, experienced increases of one-half percentage point or more.

In the fourth quarter of 2003, unemployment rates were above 7 percent in New York, New York; Portland-Vancouver, Oregon-Washington; and San Jose, California. Unlike San Jose, New York and Portland-Vancouver had rates that were little changed from a year earlier. At the same time, jobless rates were below 4 percent in Norfolk-Virginia Beach-Newport News, Virginia-North Carolina; Orange County, California; and Washington, DC-Maryland-Virginia-West Virginia. While Orange County posted a decline of one-half percentage point from the fourth quarter of 2002, neither of the other two low-rate areas recorded a notable change in rate over the year.

Twenty-two of the most populous metropolitan areas experienced declines in their total nonfarm employment levels in 2003. (See table 10.) The heaviest job losses were registered in New York, New York (52,400); Detroit, Michigan (51,200); and Boston, Massachusetts-New Hampshire (48,700). For the third straight year, San Jose, California, reported the steepest percent decline in employment (3.8 percent), and neighboring San Francisco had the second-largest contraction (3.6 percent). Boston, Detroit, and Oakland, California, also posted declines of more than 2 percent in 2003. Eighteen of the 42 most populous metropolitan areas have lost jobs during each of the past 3 years.

Twenty large metropolitan areas experienced increases in total nonfarm employment in 2003. The most jobs were added in Las Vegas, Nevada-Arizona (34,400); Phoenix-Mesa, Arizona (27,800); Washington, DC-Maryland-Virginia-West Virginia (24,100); and Orlando, Florida (20,600). In percentage terms, Las Vegas posted the largest employment expansion (4.3 percent), followed by Orlando (2.2 percent). Growth rates of more than a full percentage point were recorded by six other large areas in 2003. Seven areas-Fort Lauderdale, Florida; Las Vegas; Norfolk-Virginia Beach-Newport News, Virginia-North Carolina; Orange County, California; Riverside-San Bernardino, California; Sacramento, California; and San Diego, California-are notable for having reported employment gains during each of the past 3 years.