Earlier this year, we posed the following challenge for our readers:
Why does the following chart, which spans 50 years of data for the United States in the post World War 2 era, look the way it does?
In this chart, we observe that the ratio of the U.S. National Average Wage Index starts off at a level 127.3% of the U.S.' GDP per Capita in 1951, slowly rises to peak at 137.8% of GDP per Capita ten years later in 1961, then falls steadily for the next three decades until 1994 when it flattened out at around 88.3% of the U.S.' GDP per Capita.
Since then, it has been as high as 91.3% of GDP per Capita in 2001, and as low as 86.2% of GDP per Capita in 2006. In 2010, the ratio of the U.S. National Average Wage Index to GDP per Capita is 88.6%.
What we can't explain is why these patterns exist. How can the average wage earned by individuals in the U.S. go from being as much as 37.8% higher than the U.S.' GDP per Capita over forty years ago to being steadily 11.4% below that quantity three decades later. What factors caused this ratio to first rise, then fall, then stabilize?
Several of our readers responded, offering the following possibilities:
- Why the ratio was rising from 1950 to 1961:
- Casualties in World War 2 reducing the pool of able-bodied men available to work, which boosted wages.
- Why the ratio fell so dramatically from 1961 to 1994:
- Technology replacing human capital in the workforce.
- An increasing share of women entering the workforce at lower wages.
- Baby boomers entering the U.S. workforce at lower wages.
- Declining union membership in the private sector.
- A declining portion of GDP going to wages.
- Why the ratio leveled off after 1994:
- Baby boomers no longer entering the U.S. workforce.
It took time to run down all these possibilities, but we now know the basic answer, at least for the period since 1961: the declining portion of GDP going to wages!
We were able to eliminate the other possibilities largely because the timing of factors affecting each of them failed to correlate with the trends we observed in the chart. For example, the number of women entering the U.S. workforce has been rising steadily for all years since 1951, and so have their wages - there have been no surges in female employment that might have sharply reduced wages with respect to GDP after 1961 as we might expect to see if women were responsible for what we've observed.
The entry of Baby Boomers into the U.S. workforce at first seemed like a better fit in explaining the decline, with the oldest Baby Boomers turning the working age of 16 in 1962, but here, only a portion of this population would have done so, with a majority waiting to graduate from high school two years later before entering the workforce, and another very large percentage postponing that event for up to another four years following their graduation from college.
If Baby Boomers were responsible for what we observe, we would see the ratio continue rising but top out in the early-to-mid 1960s. Instead, that clearly happened in the late 1950s and very early 1960s, years before the leading edge of the Baby Boomers reached Age 16.
Similar logic applies for the stabilization of the ratio following 1994. Here, the youngest Baby Boomers, born in 1964, would have reached the working age of 16 in 1980. Adding in two more years of high school and four years of college would have nearly all of the last Baby Boomers entering the U.S. workforce no later than 1986. The stabilization in the ratio didn't begin for another eight years.
Likewise, the role of technology, such as the wide scale introduction of the transistor in the 1950s and 1960s, following by the rapid development of computing technology in the late 1980s and 1990s doesn't well correlate with what we observe. Here, if technology were the culprit, we would expect to see a second decline in the ratio following the widespread adoption of computing technology in the late 1990s and early 2000s. That we don't see such a second decline indicates that technology doesn't explain what we observe.
Like the timing of Baby Boomers entering the U.S. workforce, the timing of the decline of union membership fails to explain what we observe in the chart. Here, union membership in the U.S. peaked in the early 1950s, then entered a period of slow and steady decline until 1974, after which union membership entered into more of a free fall thanks to the loss of automotive industry jobs as a result of the Arab oil embargo, when U.S. consumer demand shifted strongly in favor of more fuel efficient vehicles made by foreign car makers (a situation repeated again in 2008). The slow decline of union membership in the 1950s and 1960s is not consistent with the timing of the decline we observe in the ratio after 1961 and the lack of a similar decline following 2008's event rules out this possibility.
And that leaves just a declining portion of GDP going to wages. Interestingly though, when we went into the Bureau of Economic Analysis' data on the composition of GDP by sources of income, we found that total labor compensation as a percentage of GDP has been rock steady since 1951, averaging 88.6% of GDP and ranging between 86.8% and 90.7% of GDP in the 60 years from 1951 through 2011.
But the composition of that labor compensation has changed! Here, the percentage of labor compensation represented by wages and salaries has fallen from 95% in 1951 to just over 80% in 2011. Meanwhile, the portion of labor compensation represented by employer contributions to Social Security and Medicare (or employer FICA contributions), as well as for regular pensions and health insurance has risen considerably, taking a larger and larger share of total labor compensation for U.S. workers. Our new chart below shows how and when these other forms of compensation cut into U.S. wages and salaries:
Moreover, the timing of when those changes in the composition of labor compensation in the U.S. took place correlate very well with what we observe in the chart. Here, we observe increases in the employer portion of FICA taxes in 1954, 1957, 1959, 1960, 1962, 1963, 1966 (when the Medicare tax was introduced), 1967, 1969, 1971, 1973, 1978, 1979, 1981, 1982, 1984, 1985, 1986, 1988 and 1990. The FICA tax rate has then increased from 1.5% of income in 1951 to 7.65% in 1990, where it has held steady through 2011.
We likewise observe double-digit increases over the previous year for labor compensation in the form of pensions and employer-provided health insurance taking place from 1955 through 1957, 1959, and in each year from 1964 through 1981.
Together, these forms of employee compensation have displaced wage and salary income in the measure of GDP, which largely explains why the ratio of the National Average Wage Index first topped out with respect to GDP per capita in the U.S. in the late 1950s, then dropped after 1961 exactly as we've observed.
That also includes the stabilization in the ratio after 1994, as the year-over-year growth rate for employer-provided pension and health insurance slowed significantly after that year as U.S. employers adopted more effective cost control strategies for their pension and health insurance-related employee compensation expenses. These measures have included shifting away from unsustainable defined benefit pension plans in favor of defined contribution plans, such as 401(k) and 403(b) type retirement saving plans, as well as increasing the amount of the co-payments and annual deductibles paid by employees for their employer-provided health insurance.
And that's it! Our thanks to our readers for their suggestions in helping us finally explain that chart!
The data for the changing composition of labor compensation over time is taken from the BEA's interactive data showing GDP and Personal Income data, specifically Table 1.12, which shows the U.S.' national income by type of income.