Real Wage Growth, America's Aging Workforce and the Impact on the Economy

A recent two-part article posted on the Liberty Street Economics website shows what lies ahead for the American economy when one considers the relationship between the aging of the labour force and the behaviour of real (after inflation) wage growth.  For those of you who are not aware of the Liberty Street Economics website, it is a blog that is sponsored by the Federal Reserve Bank of New York, publishing analysis from economists that may or may not reflect the actual position of the FRBNY.

The authors, Robert Rich, Joseph Tracy and Ellen Fu, started by looking at real hourly wage growth rates for all employed individuals aged 16 and older over the period from January 1982 to May 2016 as well as the May data from 1969 to 1981 using the CPI to convert nominal hourly wages to real hourly wages in Q1 2014 dollars.  The data is then divided into 140 different demographic cohorts based on decade of birth, gender, race and educational level (i.e. one cohort would be white male high school graduates born in the 1960s).

Here is an example of their findings which looks at the real wage profiles plotted on a logarithmic scale of white males born in the 1950s divided into groups based on educational attainment:

   

While this is only a single sample of the 140 cohorts, the five cohorts within the white male, born in the 1950s cohort, the authors found that the same pattern developed across all of their demographic cohorts.  They observed that real wages tended to shift up as educational attainment rose and that real wages tended to rise early in a worker's career, flattening out in mid-career and declining as the worker approaches retirement.  

While all of this is not terribly unexpected, the focus of the authors was to find out the implied growth rate of real wages, not simply the level of real wages, a task that is made easier because the authors used a logarithmic scale to plot the real wage profiles.  The authors computed implied real wage growth rates as the slope of the estimated cohort-specific real wage profiles at a given age.  As the slope of the line steepens, real wage growth increases.  As the slope of the line flattens, real wage growth decreases.  When the slope turns downward as age increases to the end of one's working career, the real level of wages is actually falling.

Here is a graphic showing the percentage of real wage growth for white males born in the 1950s by educational level and age:

All five educational levels show rapid real wage growth in the early part of their career with positive real wage growth ending in the mid-forties.  At this point, real wage growth is either flat for less than high school graduates and declining for white males with some college.  By the age of 55, all categories are experiencing negative wage growth; in other words, wage growth is not keeping up with inflation.  

From my own personal experience in the workforce, early wage growth was at very high rates, partly because of the high inflation levels in the late 1970s and early 1980s, however, a great deal of the wage growth was related to on-the-job-learning and on-the-job-training.  This generally makes a worker a more valuable commodity to their employer.  After a few years, workers, myself included, tended to look for more suitable employment that better matches their acquired skills with a new employer.  At this point, wage growth is also substantial.  After that point, as workers age, there is a diminished incentive for employers to invest in on-the-job training since the worker's working like is shorter.  This results in significant slowing of wage growth rates.  As such, the authors found that real wage growth profiles can be divided into three segments:

1.) fast real wage growth up to age 40

2.) flat real wage growth between the ages of 41 and 54

3.) negative real wage growth over the age of 55

Now, how is this going to impact America's workforce.  Here is a graphic showing how the age distribution of the adult population of the United States has changed between 1980 and 2015:

The authors' analysis shows that the fraction of the U.S. population that is in the fast real wage growth phase of their careers has dropped from nearly 60 percent in 1980 to just under 45 percent in 2015.  As shown in this graphic, this aging of the American workforce has had a substantial impact on aggregate real wage growth rates since 1980:

Aggregate real wage growth for all workers in America has declined from 1.8 percent in the mid-1980s to 1.2 percent in 2015, a decline of 33 percent.  

In conclusion, the authors note that U.S. real wage growth has slowed down over the past three and a half decades as the American workforce aged.  With real wage growth becoming negative for workers over the age of 55, this trend will continue for the coming decades as the proportion of older workers rise.  Since real wage growth generally reflects the rate of labour productivity growth, it certainly appears that the ongoing aging of the American workforce will negatively impact future labour productivity and wage growth, realities that will ripple through the economy, affecting future economic growth rates.

    

The Real Wage - Productivity Conundrum

Updated February 2015

In the early days of my career in the late 1970s and early 1980s, one thing that employees could count on was getting a raise that would include a cost of living allowance or COLA.  That way, we had some assurance that our compensation was keeping up with increases in what we were spending to live.  A paper by Lawrence Mishel and Kar-Fai Gee gives us some insight into the conundrum that faces workers in America.  While labour productivity increased over the four decades from 1973 to 2011, the same cannot be said for gains in real wages, that is, wages that are corrected for increases in the cost of living.  

The median wage represents the wage earned by a worker at the midpoint between the highest and lowest paid worker.  Using median values rather than average values avoids the problem that can occur when wages change only at the top and bottom of the distribution.  Over the period from 1973 to 2011, the long-term growth in the median real compensation, which includes all benefits plus salary and wages, has been very low, averaging only 4 percent over the four decades as shown on this graph:

Over the period, the median hourly compensation has risen from $18.08 (in 2011 dollars) in 1973 to $20.01 in 2011, an average annual increase of 0.27 percent per year.  The graph shows that the real median compensation level was pretty stagnant from 1973 to 1992 when it increased rather sharply until 2003 and then remained stagnant thereafter.  In fact, in 2011, the real median compensation fell by 2.5 percent; this in a year when the economy was in full recovery mode!

Now, let's compare what happened to the average and median real hourly compensation, again, keeping in mind that average compensation is affected by changes in compensation at the top and bottom of the entire distribution (i.e. if wages increase at the top, the average will increase but most workers may still have seen little increase in compensation):

Real average compensation has risen from $25.54 (in 2011 dollars) in 1973 to $35.05 in 2011 which results in an annual growth rate of 0.87 percent, just over three times the growth rate of the median.  The faster growth in the average means that earning inequality is growing and that the highest paid workers are seeing their compensation rise at rates much higher than among their lesser paid peers as you can see on this graph:

The share of all wages paid to the top one percent of earners has nearly doubled from 1973 to 2010; in 1973, the top earners received 6.8 percent of all wages paid.  By 2010, this had increased to 12.9 percent of all wages paid.  In fact, the top 0.1 percent of earners received 4.7 percent of all wages paid in 2010, up from 1.5 percent in 1979.  This is in sharp contrast to the earnings of the bottom 90 percent of earners who saw their wages in 2010 being only 115 percent of their wages way back in 1979!

Now, let's switch gears for a minute.  Let's look at how productive America's workers have been.  Labour productivity is defined as the output produced on an hourly basis by an average worker and is measured as real GDP per hour worked as shown on this graph:   

Between the first quarter of 1973 and the first quarter of 2011, real output per hour rose from 51.26 to 104.03, an increase of 102.9 percent.  In real dollar terms, labour productivity grew from $33.68 (in 2011 dollars) in 1973 to $60.77 in 2011, an average increase of 1.56 percent per year.  The real growth rate in worker productivity is 5.8 times the growth rate in median real hourly compensation!

Here is a graph showing the labour share which is defined as the share of total nominal worker compensation in nominal GDP:

Between 1974 and 2011, labour's share of GDP fell from 67.7 percent to 62.2 percent, the lowest level on record looking back to 1950.  Note that during the 1980s and early 1990s, labour's share was pretty consistent, rising between 1997 and 2001 when it hit 66.4 percent.  From there, it began its downhill slide to 62.2 percent as noted above.

What does all of this mean?  It means that, on average, workers in America have benefitted very, very little from growth in labour productivity (for which they are largely responsible) in terms of real wage growth and that their share of the economy has declined to levels not seen going back six decades.  Sadly, this is a situation that has become worse over time.  The gap between median wage and productivity growth was 0.84 percentage points per year between 1995 and has increased to 1.84 percentage points per year between 2000 and 2011.   There has been a dramatic increase of redistribution of income from labour to capital as a share of America's total income, largely related to these factors:

1.) de-unionization.

2.) globalization.

3.) high trade deficits.

4.) high unemployment.

5.) higher levels of investment in short-lived capital assets including information and communication technologies that require frequent updating and replacement.

The sluggish increase in the level of real median worker compensation and the contrasting substantial increase in worker productivity speaks to the "new way of doing business in America".  As the Great Recession showed, workers are deemed necessary only because they are critical to increasing output and are quickly ushered to the door when times get tough.

The Impact of Sorting and Effort on Productivity During Recessions

As an employee, have you ever wondered whether your employer understands the effectiveness of positive and negative reinforcement, particularly the threat of losing your job, on your performance and output?   Does that raise result in higher productivity or is the threat of a potential layoff more effective at milking that extra bit of effort? Does a recession have an impact on productivity levels because employers get rid of non-performers or because workers are willing to put in that extra effort to avoid the stigma of unemployment?

An examination of the issue by Edward Lazear, Kathryn Shaw and Christopher Stanton in a paper entitled "Making Do With Less: Working Harder During Recessions" looks at what happened to productivity during the period from December 2007 to July 2009, the official beginning and end of the Great Recession.

Let's open with this graph from FRED showing output per hour of all persons in the nonfarm business sector for the aforementioned period of time:

At the beginning of the Great Recession, output per hour was 96.187, rising to 99.499 by the end of the contraction, not exactly what one would expect.  Over the latest recession, aggregate output dropped by 4.35 percent but the aggregate number of hours worked dropped by even more, falling by 10.54 percent.  This means that labor productivity rose, in fact, it rose by 3.16 percent in nonfarm businesses.  This is not generally the case historically; in recessions before the 1990s, productivity dropped and the decrease in labour is generally less than the decrease in output.  In Europe, this was the case even in the latest recession as shown here, noting that the dotted red line (United States) is heading in the opposite direction to the major European economies including Germany, Italy, France and the United Kingdom:

The difference between Europe and the U.S. could be related to Europe's policies on restricting firings and the areas intentional employment retention policies, a policy that clearly does not exist in the increasingly non-union workplace in America.  

Why did productivity rise even as employment fell in the U.S. during the Great Recession?  There are two possibilities as follows:

1.) Sorting: The quality of the workers that retained their jobs during the Great Recession were of better quality than in the period before the recession.  This is referred to as "sorting" whereby employers layoff their less productive workers (laggards) and keep their more productive workers (stars).

2.) Effort: The workers that were employed during the great recession were of the same average quality as before the Great Recession but were able to produce more because they made more of an effort.  This is known as the "making do with less" since employers are able to get more production out of the same number of workers as they had prior to the recession.  Workers, knowing that there is a greater chance that they might be laid off, may be willing to trade a higher degree of effort for a given level of compensation given that the option is termination for poor performance.

By studying data for one very large, multi-faceted nationwide technology services firm that employed 20,386 moderately skilled labourers and 5.1 million data points on daily performance measured by computer monitoring of activities from June 2006 to May 2010, the authors were able to measure the impact of "sorting" versus the impact due to "effort".

Productivity at the company in question rose from an average of 9.87 units per hour during the period prior to and after the recession and rose to an average of 10.76 units per hour during the period during the Great Recession as shown on this graph plotting both unemployment (red crosses) and the log of output per hour (OPH) (blue dots):

Output per worker during the recession rose by 5.33 percent.  During the recession, the company added employees, however, the growth rate of employment fell.  Note the rapid drop off of output after the Great Recession even though unemployment remained high.

Through the use of regression analysis, the authors were able to determine that:

1.) Laggards (poorer performers) were able to increase their productivity by 5.1 percent during the recession compared to stars (better performers) who increased their productivity by only 3.4 percent.

2.) A very tiny percentage of the productivity gains during the recession can be explained by the movement of lower quality workers out of the workforce and the resulting concentration of higher quality workers.

The main finding of the research is that this particular company, a reasonable representative of the economy as a whole, made do with less.  Increased effort by employees led to higher output rather than the sorting out of poorer performing employees.  What is even more interesting is that in areas of the country where unemployment was higher due to a more pronounced recessionary economic contraction, the productivity gains were the highest and the increase in effort was the most pronounced.

To summarize, the fear factor associated with unemployment, particularly in the last, very deep recession, caused people to work harder.  The spectre of having to find a job in a dreadful economy put the "fear of God" into Main Street America, causing workers to work harder and productivity to rise even as employment fell.  The implicit use of the "you're lucky to have a job" mantra can work wonders on a fear-ridden workforce, pushing us to work beyond our normal comfort zone in a desperate effort to remain employed.

Apparently, Corporate America is learning that they really can make do with less.