Trends in Men's Wages, 1981–2014

Introduction

Every year, employers report their employees' wage and salary earnings to the Internal Revenue Service (IRS) and the Social Security Administration (SSA) on IRS Form W-2.¹ SSA stores those earnings records in its Master Earnings File (MEF), which it uses to administer the Old-Age, Survivors, and Disability Insurance (OASDI) programs.² For research and statistical purposes, SSA extracts data from the MEF and other administrative files each year to create the Continuous Work History Sample (CWHS). The CWHS contains earnings records for more than 3.7 million individuals, representing 1 percent of all Social Security numbers ever issued. For researchers, the large number of earnings records in the CWHS, its longitudinal structure, and its accuracy have advantages over household surveys, which consist of smaller samples, typically collect data for relatively short periods, and are subject to reporting and recording errors.

This article describes the trends in real wages and salaries recorded in the CWHS among men aged 25–59 from 1981 through 2014. It briefly describes the change in real wages and salaries for all men aged 25–59 during this period, then examines trends for individual birth cohorts in each of seven age groups: 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, and 55–59. Using a series of charts, I show how men's real wages changed across age groups and birth cohorts within each age group.

Data and Methods

The CWHS is an analytical master file created from 1 percent samples of the Master Beneficiary Record (MBR) and the MEF, both of which SSA uses to administer the OASDI programs. To maintain the CWHS's 1 percent sample size, each year, SSA adds the earnings records associated with a random selection of newly issued Social Security numbers. The records of deceased workers remain in the CWHS, allowing researchers to study the wages of entire birth cohorts over time. When needed, SSA updates the CWHS earnings records for adjustments and corrections to the MEF.

The CWHS includes data on Social Security taxable wages in covered employment since 1951.³ Covered employment refers to jobs for which employers submit payroll-tax deductions to the IRS and report wages to SSA to determine a worker's eligibility for Social Security benefits and the amount of those benefits. Taxable wages are earnings in covered employment equal to or less than an annually adjusted threshold amount called the taxable maximum.⁴ Since 1978, the CWHS has included records on wages in noncovered employment and earnings exceeding the annual maximum taxable amount.

This article describes results derived from the 2014 CWHS file, the most recent available when the analysis was conducted. Following the methods of Leonesio and Del Bene (2011), the earnings analyzed in this article consist of wages and salaries since 1981 in both covered and noncovered employment, including wages and salaries exceeding the annual taxable maximum. Earnings from self-employment are not included.⁵ The analysis includes only men's earnings because the changes that have occurred in employment and earnings among women warrant separate analysis.⁶ It focuses on ages 25 to 59 because those are the ages with the highest employment rates.⁷ For brevity, I refer to “wages and salaries” hereafter simply as “wages.”

To focus on workers who had substantial wages, the analysis includes only individuals with annual wages equal to or greater than the amount needed to earn four quarters of coverage under Social Security.⁸ This amount ranged from $1,240 in 1981 ($2,827 in 2014 dollars) to $4,800 in 2014. All wages have been indexed to 2014 values by the personal consumption expenditure (PCE) index of the National Income and Product Accounts.⁹

In addition to excluding individuals with wages lower than the amount needed to earn four quarters of coverage, this analysis excludes the top 0.1 percent of earners each year. I exclude those records because in some cases, very high wages recorded in the CWHS indicate data-reporting errors, coding errors, or fraudulent use of a Social Security number, and there is no way to distinguish between the accurate and inaccurate records. This exclusion also reduces the effect of extreme outliers at the high end of the wage distribution on the measured mean and variance of wages. The 1981 sample was bounded at the high end at $432,197, the amount of wages (in 2014 dollars) above which a man would have been in the top 0.1 percent of male earners that year. The 2014 sample was bounded at the high end at $1,522,006, the amount of wages above which a man would have been in the top 0.1 percent of male earners that year.

The 2014 CWHS file consists of 3,727,665 individual person-records.¹⁰ Of these records, 53.1 percent are for men and 46.9 percent are for women. For this analysis, the sample was restricted to men aged 25–59 in the year observed. Thus, for 1981, the sample includes men born from 1922 through 1956. For 2014, the sample includes men born from 1955 through 1989. Overall, the sample consists of 18,228,530 person-year observations from 1981 through 2014, with an average of 536,133 unique individuals observed each year. The number of observations ranges from a low of 434,328 for 1982 to a high of 586,865 for 2007. There are an average of 15,318 records for each year observed for each single year of age.¹¹ The fewest records for any year observed for a single year of age is 6,839, for men aged 59 in 1992 (born in 1933). The most records for any year observed for a single year of age is 20,467, for men aged 38 in 1998 (born in 1960).

In the next section, I summarize previous research based on the CWHS. I then describe broad trends in wages from 1981 through 2014 for men aged 25 through 59. A discussion of the main findings follows, in a section that describes the changes in median real wages from 1981 through 2014 for men in seven age groups: 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, and 55–59. These age-earnings profiles show how men's real wages changed from 1981 through 2014 across age groups and birth cohorts within each age group.

Previous Research

Several analysts have used the CWHS to study the growth and variance of earnings over time. Kopczuk, Saez, and Song (2010) investigated trends in the variance of annual earnings from 1970 to 2004. They found that almost all of the increase in variance was “due to [an] increase in the variance of permanent earnings, as opposed to transitory earnings.” Sabelhaus and Song (2010) found that between 1980 and the early 1990s, the variability of earnings growth rates across the working population declined significantly, and that the lower volatility continued through the early 2000s. They suggested that over that period, both permanent and transitory components of earnings shocks had become more moderate.

Leonesio and Del Bene (2011) used the CWHS to study the distribution of men's and women's wages from 1981 through 2004. They observed that “among prime-aged men, real earnings have declined or stagnated for low-wage earners, have increased modestly in the middle of the distribution, and have risen substantially for high earners.” They also found among men “an increase in long-run earnings inequality of roughly the same magnitude as the trend seen in annual earnings dispersion.” They observed relatively little increase in the dispersion of long-run earnings among women. They concluded that the trends they observed were “consistent with the view that more highly skilled and educated workers have been paid higher premiums for their labor over time, while the productivity and earnings of lower-skilled workers have not similarly benefited from improvements in technology.”

Guvenen, Kaplan, and Song (2014) used the CWHS to measure the progress that women have made toward achieving earnings parity with men. They found that although the share of women in the top 1 percent of earners increased by a factor of more than three from the early 1980s to 2012, women's earnings constituted only 11 percent of the earnings of the top 1 percent of earners in 2012. Guvenen and others (2015) examined changes in annual earnings and found that in any given year, most workers experience very small changes in earnings, but a small percentage experience very large shocks. They found that positive shocks to high-income individuals are transitory, but negative shocks are persistent. For low-income individuals, however, large earnings shocks are more common but less persistent. The authors concluded that in general, high-income individuals experience earnings shocks that are persistent but that their income shows lower volatility than that of lower-earning workers. Song and others (2015) matched CWHS records to employer data to compare the dispersion of earnings within firms to earnings dispersion across firms. They found an increase in earnings inequality among workers of different firms between 1978 and 2012, while differences in earnings within firms remained almost unchanged.

This study differs in focus from those described above. It exploits the large CWHS sample and its longitudinal structure to compare the real wages of men in seven age intervals over a period spanning 33 years. Charts show real median wages each year for each age group, allowing us to observe trends in men's real median wages across age groups and birth cohorts within each age group. First, however, I summarize the broader trends in men's real wages in the study period.

Men's Wages 1981–2014

Chart 1 shows the median and mean wages along with the standard deviation of wages for men aged 25–59 from 1981 through 2014. Men's real wages during that period had a flat median, a rising mean, and increasing variance. Real median wages were $42,973 in 1981 and $45,000 in 2014, an increase of $2,027 (4.7 percent) overall and an average annual increase of 0.1 percent. Much of the growth in men's wages occurred over a relatively short period in the late 1990s. Real median wages fell from $42,973 in 1981 to $39,968 in 1993. From there, median wages rose to $45,620 in 2001 and then remained almost level over the next 6 years, rising by $489 (1.1 percent) to $46,109 in 2007. Real median wages fell during the Great Recession, declining to $44,170 in 2011 before recovering slightly to $45,000 in 2014. Nevertheless, the real median wages of men aged 25–59 in 2014 were $1,109 (2.4 percent) lower than they had been in 2007.

Real mean wages rose from $47,720 in 1981 to $64,181 in 2014, an increase of $16,461 (34.5 percent), or an average annual increase of 0.9 percent. Much of the growth occurred between 1993 and 2001. Early in the period, real mean wages rose from $47,720 in 1981 to $51,128 in 1992, an increase of $3,408, or 7.1 percent. By 2000, they had risen to $61,587, an increase since 1992 of 20.5 percent. Mean wage growth slowed after 2000. Wages rose to $64,282 in 2007, then fell to $62,027 in 2009 (during the Great Recession), before rebounding to $64,181 in 2014. Between 2000 and 2014, men's mean wages rose by 4.2 percent.

As men's real mean wages increased from 1981 to 2014, so did the standard deviation, a measure of how widely the values are distributed around the mean. In 1981, men's real mean annual wages were $47,720 and the standard deviation was $34,797. By 2014, the mean value of men's wages had risen by 34.5 percent to $64,181, yet the standard deviation had more than doubled, from $34,797 to $80,635, indicating a substantial increase in the dispersion of wages around the mean. In both 1981 and in 2014, the distribution of wages was skewed to the right: The highest values were much farther from the mean than the lowest values; recall that the latter are equivalent to the annual earnings needed to earn four quarters of coverage under Social Security.

Chart 2 shows the real wages of men aged 25–59 each year from 1981 through 2014 at the 10^th, 25^th, 50^th, 75^th, 90^th, and 99^th percentiles of the wage distribution. In 1981, a worker with wages at the 10^th percentile earned $12,894 in 2014 dollars. In the study period, the wages of men at the 10^th percentile peaked at $14,085 in 2000. By 2014, real wages at the 10^th percentile were $13,387, only 3.8 percent higher than in 1981. Real wages at the 25^th percentile were $25,848 in 1981, $26,809 at their peak in 2000, and down to $25,339 in 2014, 2.0 percent lower than in 1981. As noted earlier, median real wages among men aged 25 to 59 rose from $42,973 in 1981 to $45,000 in 2014, an increase of 4.7 percent. Men's real median wages peaked in 2007 at $46,109.

In contrast with the low rates of growth at the median and lower percentiles, real wages rose substantially more rapidly in the upper half of the earnings distribution. Real wages at the 75^th percentile were $61,747 in 1981, changed relatively little through 1995, then began a steady rise to $71,601 in 2000 and finally to $75,413 in 2014—a level that was 22.1 percent higher than in 1981. Wages at the 90^th percentile rose nearly continuously through the period, declining only slightly during 1989 and the recession years of 1982, 2002, and 2008–2009. Real annual wages at the 90^th percentile rose from $80,791 in 1981 to $121,763 in 2014, an increase of 50.7 percent, representing an average annual growth rate of 1.25 percent. The most striking feature of Chart 2 is the steep increase in wages at the 99^th percentile. From 1981 to 2014, real wages at the 99^th percentile more than doubled, rising from $180,214 to $392,250—an increase of 117.7 percent, or 2.38 percent per year on average.

An individual's lifetime path of wages depends on a number of factors, including education, occupation, industry of employment, economic conditions, and the worker's personal traits. For many workers, annual wages are relatively low when they are in their 20s, rise rapidly in their 30s as they develop skills and gain experience, and then increase more slowly as they enter their 40s. Annual wages for many workers peak between ages 45 and 55. By the time many workers reach their late 50s, annual wages begin to decline. Some workers choose to work fewer hours as they get older, while some move to lower-paying jobs, either voluntarily or involuntarily, depending on their circumstances (Sonnega, McFall, and Willis 2016). For example, some are unable to continue in their career occupation because of chronic illness or work-limiting disabilities. Many workers experience declining wages in their late 50s; yet since the mid-1980s, the median wages of men aged 55–59 have been higher than those of men younger than 40.

Chart 3 shows median wages for 1981–2014 among men aged 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, and 55–59. All amounts are in 2014 dollars. In both 1981 and 2014, the two age groups with the lowest median wages were 25–29 and 30–34. Also in both years, men aged 35–39 had lower median wages than those aged 40–54. Men aged 55–59 had higher median wages than did those in the two youngest age groups in both 1981 and 2014, and they experienced a greater rise in median wages over that period than did those in any other age group. Real median wages among men aged 55–59 rose from $46,334 in 1981 to $51,871 in 2014, an increase of 12.0 percent. Some of this increase was due to a trend of more hours worked in the latter years of the period while some of it may reflect increases in hourly wages for older workers. As will be seen later, the real median annual wages of men aged 55–59 rose from 1981 to 2014, yet they declined for each successive year of age from 55 to 59 throughout the period, reflecting both fewer hours of work and lower wages among workers approaching retirement.

Changes in the age distribution of workers can affect the wage growth rate. For example, if the proportion of workers who are in their peak earnings years (ages 40–54) rises, the median annual wages of all men aged 25–59 may rise even if median wages within each age group remain flat. Chart 4 shows that from 1981 to 2014, the proportion of working men aged 25–59 who were 25 to 39 years old fell from 56.4 percent to 43.8 percent and the proportion who were 40 to 54 years old rose from 33.9 percent to 43.2 percent. All else being equal, the increase in the proportion of working men who were aged 40–54 would have illustrated the example noted above by causing the median annual wages of men aged 25–59 to increase even if median wages within each age group had not risen.¹²

We can estimate the effect of the change in the age distribution of working men on their median wages by reweighting the records from the CWHS so that the age distribution is constant each year, and then recalculating the annual median wage. Of course, if the distribution of workers by age had not changed over time, a static distribution of workers by age itself would likely have had some effect on wages. Nevertheless, estimating a “fixed population weight” median wage gives us an idea how much of the increase in men's real median wages during 1981–2014 was due to the growth in the proportion of workers who were in their prime earning years. After reweighting by age, men's estimated real median wages for 1981 are $44,199 (2.9 percent higher than the actual median of $42,973), and for 2014, they are $44,256 (1.7 percent lower than the actual median of $45,000). In other words, if the proportion of working men in their prime earning years had not risen, the real median wages of men aged 25–59 likely would not have risen at all from 1981 to 2014, all else being equal. The observed increase in men's real median annual wages was therefore due almost entirely to the increase in the proportion of men aged 40–59 and the corresponding decrease in the proportion aged 25–39.

Age-Earnings Profiles

The CWHS' large number of records and its longitudinal structure allow the construction of age-earnings profiles that show the median wages of workers from many birth cohorts over long periods. This section contains charts showing real median wages of men in seven age intervals over a 33-year period, allowing us to compare real wages across age groups and birth cohorts within each age group. The period 1981–2014 included four recessions and four expansions, and the charts illustrate the effects of the business cycle on the age-earnings profiles.¹³ Specifically, the charts show men's real median wages from 1981 through 2014 for each of the following seven age intervals:

• 25–29, comprising the 1956–1985 birth cohorts;
• 30–34, comprising the 1951–1980 birth cohorts;
• 35–39, comprising the 1946–1975 birth cohorts;
• 40–44, comprising the 1941–1970 birth cohorts;
• 45–49, comprising the 1936–1965 birth cohorts;
• 50–54, comprising the 1931–1960 birth cohorts; and
• 55–59, comprising the 1926–1955 birth cohorts.

The oldest men in the sample, the members of the 1926 birth cohort, attained age 55 in 1981. Because they (as well as men born 1927–1930) were older than 59 for all but the first few years of the observation period, I track their wages only in the 55–59 age interval. The youngest men in the sample were born in 1985; they attained age 25 in 2010. Because they (as well as men born 1980–1984) were younger than 25 in all but the final few years of the observation period, I track their wages only in the 25–29 age interval. Although no birth cohort can be fully tracked through each of the seven age intervals in the 1981–2014 span, men born 1951–1960 are fully tracked in six of the seven charts below. In total, I track the wages in 1981–2014 of men representing 60 birth cohorts (1926 through 1985).¹⁴

Each chart includes a note highlighting the average change in real median wages over the entire observation period for all members of the subject age group. Additional notes identify the single birth cohort whose members experienced the smallest wage growth (or greatest decline) and the cohort whose members experienced the greatest wage growth (or smallest decline) over that age interval and show the corresponding percentage changes.

Appendix A contains tables that correspond with Charts 5–11.¹⁵ The tables show the specific real median wages for each year and cohort covered in each chart.

Chart 5 tracks the real median wages of men born 1956–1985 in the years when they were aged 25–29. For men in the 1956 birth cohort, wages rose from $27,921 at age 25 to $34,730 at age 29, or by 24.4 percent. For men in the 1985 birth cohort, wages rose from $25,720 at age 25 to $35,302 at age 29, or by 37.3 percent. Thus, wages at age 25 were $2,201 (7.9 percent) lower for men born in 1985 than those of men born in 1956, but at age 29 the wages of men born in 1985 were $572 (1.6 percent) higher than those of men born in 1956. On average, real median wages for all members of the 1956–1985 cohorts increased by 32.6 percent from age 25 to age 29.

Chart 6 tracks the real median wages of men born 1951–1980 in the years when they were aged 30–34. For men in the 1951 birth cohort, wages rose from $37,600 at age 30 to $42,273 at age 34, or by 12.4 percent. For men in the 1980 birth cohort, wages rose from $36,903 at age 30 to $43,343 at age 34, or by 17.5 percent. The wages of men born in 1980 were 1.9 percent lower than those of men born in 1951 at age 30 but were 2.5 percent higher at age 34. On average, across all cohorts, real median wages increased by 15.6 percent from age 30 to age 34.

Chart 7 tracks the real median wages of men born 1946–1975 in the years when they were aged 35–39. For men in the 1946 birth cohort, wages rose from $46,073 at age 35 to $50,130 at age 39, or by 8.8 percent. For men in the 1975 birth cohort, wages rose from $44,075 at age 35 to $48,642 at age 39, or by 10.4 percent. The wages of men born in 1975 were 4.3 percent lower than those of men born in 1946 at age 35 and were 3.0 percent lower at age 39. On average, across all birth cohorts from 1946 through 1975, real median wages increased by 9.1 percent from age 35 to age 39.

Chart 8 tracks the real median wages of men born 1941–1970 in the years when they were aged 40–44. For men in the 1941 birth cohort, wages rose from $49,563 at age 40 to $51,979 at age 44, or by 4.9 percent. For men in the 1970 birth cohort, wages rose from $49,290 at age 40 to $52,625 at age 44, or by 6.8 percent. The wages of men born in 1970 were 0.6 percent lower than those of men born in 1941 at age 40 and were 1.2 percent higher at age 44. On average, across all cohorts from 1941 through 1970, real median wages increased by 5.6 percent from age 40 to age 44.

Chart 9 tracks the real median wages of men born 1936–1965 in the years when they were aged 45–49. For men in the 1936 birth cohort, wages rose from $49,613 at age 45 to $50,895 at age 49, or by 2.6 percent. For men in the 1965 birth cohort, they rose from $49,911 at age 45 to $51,897 at age 49, or by 4.0 percent. The wages of men born in 1965 were 0.6 percent higher than those of men born in 1936 at age 45 and were 2.0 percent higher at age 49. On average, across all cohorts from 1936 through 1965, real median wages increased by 2.8 percent from age 45 to age 49.

Chart 10 tracks the real median wages of men born 1931–1960 in the years when they were aged 50–54. For men in the 1931 birth cohort, wages fell from $50,830 at age 50 to $50,267 at age 54, a decline of 1.1 percent. For men in the 1960 birth cohort, wages rose from $51,627 at age 50 to $52,879 at age 54, or by 2.4 percent. The wages of men born in 1960 were 1.6 percent higher than those of men born in 1931 at age 50 and were 5.2 percent higher at age 54. Men in several birth cohorts experienced declines in real median wages from age 50 to age 54. On average, across all cohorts from 1931 through 1960, real median wages fell by 0.2 percent from age 50 to age 54.

Chart 11 tracks the real median wages of men born 1926–1955 in the years when they were aged 55–59. For men in the 1926 birth cohort, wages fell from $47,960 at age 55 to $44,379 at age 59, or by 7.5 percent. For men in the 1955 birth cohort, wages fell from $51,748 at age 55 to $50,710 at age 59, or by 2.0 percent. The wages of men born in 1955 were 7.9 percent higher than those of men born in 1926 at age 55 and were 14.3 percent higher at age 59. Members of all of the birth cohorts from 1926 through 1955 experienced declines in real median wages between the ages of 55 and 59; however, although the average rate of decline for the 1926 through 1939 cohorts was 9.2 percent, it was only 6.6 percent for the 1940 through 1955 cohorts. This could have resulted from members of the later cohorts working relatively more hours, earning higher hourly wages, or both. On average, real median wages across all cohorts fell by 7.4 percent from age 55 to age 59.

Discussion

Several patterns emerge in Charts 5–11. First, the relationship between age and median wages is evident. Median annual wages grow rapidly when workers are young, as they gain skills and experience. It is not age itself that influences wage growth, but rather the increase in a worker's “human capital”—his skills and experience—that leads to the rise in earnings with age, especially in the first 10 to 20 years of a worker's career. From 1981 through 2014, men's real median wages at age 29 were, on average, 32.6 percent higher than their median wages at age 25. The rate of growth of wages slows in middle age. Men's real median wages at age 49 were, on average, just 2.8 percent higher than their median wages at age 45. Finally, real median wages fall in the later years of workers' careers. From age 55 to 59, men's real median wages fell by an average of 7.4 percent, likely through a combination of reduced hours of work and movement to lower-paying jobs before retirement.

A second pattern illustrated in the charts is the relationship between real median wages and economic expansions and contractions. Every birth cohort from 1981 to 2014 experienced its lowest rate of growth in median annual wages during one of three overlapping periods: 1987–1991, 1988–1992, or 1989–1993. Economic growth was weak in those years; real median household income in the United States fell each year from 1990 through 1993. The fastest rate of wage growth for six of the seven age intervals—all but the 55–59 age group—occurred either from 1995 through 1999 or from 1996 through 2000. Economic growth was robust in that period. From 1995 through 2000, real median household income grew at an average annual rate of 2.1 percent (Federal Reserve Bank of St. Louis 2017). The slow growth of men's median annual wages from 1987 through 1993 and their rapid growth from 1995 through 2000 illustrate the strong effect of the business cycle on annual earnings.

Finally, although both age—as a proxy for experience—and the business cycle affect the growth of wages over time, behavioral changes can also lead to patterns of wage growth that are unique to particular birth cohorts. Chart 11, for example, appears to indicate a behavioral change among workers aged 55–59 beginning in the mid-1990s. On average, from 1981–2014, median wages at age 59 were 7.4 percent lower than median wages at 55. Among men born from 1940 through 1954, however, the reduction in real median annual wages that occurred after age 55 was smaller than it was among earlier birth cohorts. For men born from 1926 through 1939, median wages at age 59 were 9.2 percent lower than those at age 55. For men born from 1940 through 1955, however, median wages at age 59 were only 6.6 percent lower than those at age 55. This may indicate that workers had begun to work more hours per year after reaching age 55.

Conclusion

This article has summarized trends in men's real wages from 1981 through 2014 using CWHS data. Over that period, men's median wages rose relatively slowly and mean wages rose more quickly. The wage distribution became more unequal as wage growth in the top 10 percent of earners substantially outpaced the rate of growth for earners below the 90^th percentile. Men's real wages from 1981 through 2014 exhibit a relatively flat median, a rising mean, and increasing variance. The real median annual wages of men aged 25 to 59 were $42,973 in 1981 and $45,000 in 2014, an increase of just 4.7 percent in 33 years. Mean wages grew much faster, rising from $47,720 to $64,181 in that span, an increase of 34.5 percent. As the mean of men's wages increased, so did the standard deviation—from $34,797 in 1981 to $80,635 in 2014—indicating a substantial increase in the dispersion of wages around the mean.

From 1981 to 2014, men's real annual wages increased more quickly in the upper half of the income distribution than in the lower half. Wages at the 99^th percentile rose from $180,214 in 1981 to $392,250 in 2014, an increase of 117.7 percent. Wages at the 90^th percentile rose from $80,791 to $121,763, an increase of 50.7 percent. By contrast, men's real median annual wages increased by 4.7 percent and real wages at the 25^th percentile declined by 2.0 percent. Wages at the 10^th percentile rose from $12,894 in 1981 to $13,387 in 2014, an increase of 3.8 percent.

Other things being equal, the increase during 1981–2014 in the proportion of men who were in the peak earnings age range of 40 to 54 would have caused the real median wages of all men aged 25–59 to rise even if median wages within each 5-year age interval had not risen. In other words: If the average age distribution of men over the entire period had been maintained for each year within the period, the real median annual wages of men aged 25 to 59 would have been essentially the same in 2014 as they were in 1981; again, with all else being equal.

Members of every birth cohort experienced their lowest rate of growth in median annual wages during one of three overlapping periods: 1987–1991, 1988–1992, or 1989–1993. The fastest rate of growth for six of the seven age intervals occurred in one of two overlapping 4-year periods: 1995–1999 or 1996–2000. Among men born in 1940 or later, the reduction in median annual wages after age 55 was smaller than that for men in earlier birth cohorts. Although median wages of men at age 59 were 7.4 percent lower on average than their median wages at age 55, the average rate of decline was lower for men born during 1940–1955 (6.6 percent) than it was for men born during 1926–1939 (9.2 percent).

One limitation of this study is that the CWHS accounts for cash compensation only. Many workers receive additional compensation in the form of employer payments for health insurance and contributions to retirement accounts. During the period from 1981 to 2014, health insurance premiums for many workers rose more rapidly than wages; consequently, employers' payments toward health insurance coverage constituted an increasingly greater share of employees' total compensation over time.¹⁶ On the other hand, workers in the lower half of the wage distribution are less likely to have employer-sponsored health insurance than are those in the upper half, and it was in the lower half of the distribution that wage growth was slowest over this period.

Government officials at the federal, state, and local level recognize the importance of identifying and pursuing economic policies that promote employment and wage growth. To evaluate the effectiveness of economic policies, officials need detailed, accurate, representative long-term data on workers' wages. The wage data recorded in the CWHS are ideal for this type of research and can contribute much to our knowledge of trends in the growth and distribution of wages.

Appendix A