The Wage Stagnation Myth
If you have been remotely in touch with modern day politics, then you have probably heard of the claim that wages in the US (and other developed nations) have been ‘stagnating’ for the past few decades.
And if so, you might have even seen this chart, or one that resembles this:
There are numerous problems with these charts. The issues are centered around:
- Using only the Consumer Price Index (CPI) for measuring inflation
- Does not include overall labor force of the United States
- Does not include pay growth for self-employed
- Does not measure overall compensation
This is where I cite this excellent 2016 study by labor economist James Sherk, titled “Workers’ Compensation: Growing Along with Productivity.”
Sherk goes over the flaws of CPI and uses two other inflation indexes, the Implicit Price Deflator (IPD) and the Personal Consumption Expenditures Index (PCE).
The problem with calculating real wage growth with CPI is due to flawed methodology. The Bureau of Labor Statistics (BLS) use the Consumer Expenditure Survey (CEX) to calculate CPI. Problem with this is that the survey has biases and leads to underreporting.
The IPD also uses more accurate data than the CPI. In calculating the CPI, the BLS uses data from the Consumer Expenditure Survey (CEX) to estimate how much consumers spend on different types of goods and services. This survey has significant biases. Studies show that households recall large and repeated purchases quite well. Consequently, the CEX measures the amounts that Americans spend on rent and utilities reasonably accurately. However, people often forget smaller and less regular purchases during their interviews. This underreporting makes it appear that Americans spend far more of their income on housing, gas, or utilities than they actually do. The costs of these goods have increased faster than other goods and services. This “recall bias” increases CPI-measured inflation — and decreases CPI-adjusted compensation.
Another problem is changing consumer behavior, which Sherk also talks about:
Economists know that consumers respond to shifting prices. As smartphones have become less expensive, consumers have bought more of them, and fewer of those goods and services for which prices have risen. However, the CPI accounts for this “substitution effect” only infrequently. For this reason, most economists believe that the CPI over-estimates inflation.[31] The IPD uses a “chained” methodology that regularly takes account of changing spending patterns.
When IPD and PCE are used to calculate real wage growth, the results are:
As you can see, there still is a pretty decently sized gap between productivity and hourly compensation. But that gap can be explained through other factors as well.
BLS uses it’s payroll survey to find hourly compensation for workers. But the problem with this it does not include all forms of compensation:
The payroll survey also excludes most performance-based compensation, such as commissions, bonuses, and stock options. Performance-based pay has become widespread throughout the economy since the 1970s. It has particularly grown among top earners. Many senior corporate employees receive much of their total compensation in stock options. If the company does well, these stock options become valuable; if it does poorly, they become worthless. For example, Apple paid its former CEO Steve Jobs a salary of $1 a year. Jobs became a billionaire because his stock options became very valuable as Apple’s stock rose under his leadership. The payroll survey excludes such performance-based compensation. Excluding top earners and most performance pay makes average compensation appear to have grown more slowly than it actually did.
When adjust for those various factors, the productivity-pay gap disappears, and you can see significant wage growth since the 1970s:
The chart above illustrates the methodological ways one can create a productivity-pay gap. But once you adjust for various inflation indexes, include the overall work force, and include total compensation, the gap disappears.
Another study titled “Fifty Years Of Growth In American Consumption, Income, And Wages” by Bruce Sacerdote also goes over the flaws of using CPI, and uses Personal Consumption Expenditures index (PCE) to calculate consumption and wages.
He cites several past studies that look into CPI biases, namely The Boskin Commision (1996), Hausman (2003), Broda and Weinstein [2010], Costa (2001), Hamilton (2001), and Bils and Klenow (2001).
Each of those studies try provide estimates on CPI biases, with the range being from 1.1% to 3% in annual bias depending on the decades being evaluated. When calculating long term wage growth, annual biases in the range of 1.1 to 3% compound quickly, leading potential large errors over several decades.
When Sacerdote adjusts for the CPI errors and uses PCE, the results show that American income and consumption has grown significantly since the 1960s and 70s.
My preferred point estimates are based on CEX measures of consumption where the price index has been de-biased following Hamilton and Costa. These estimates suggest that consumption is up 1.7 percent per year or 164 percent over the whole time period. These estimates of growth strike me as consistent with the significant increases in quality and quantity of goods enjoyed by Americans over the last half century. And my conclusions are consistent with the findings of Broda and Weinstein (2008). Estimates of slow and steady growth seem more plausible than media headlines which suggest that median American households face declining living standards. The bias adjusted estimates also provide a more positive outlook on real wage growth in the last 40 years than standard media headlines. PCE adjusted wages appear to have grown at .5% per year during 1975–2015 while the de-biased CPI adjusted wages grew at 1% per year over the same time period.
So adjustments made to CPI and using PCE show positive long term wage growth for workers in the United States.
Lastly, I would like talk about this paper, “Underestimating the Real Growth of GDP, Personal Income, and Productivity” by Harvard Economist Martin Feldstein. As the title states, the paper goes into the various ways where the US government and financial institutions undercount GDP growth, productivity and most importantly personal income.
Feldstein talks about new products and how they increase consumer incomes. He says that the effects of new products is immense but that they are ‘underestimated.’
. . . new products and services are not even reflected in the price indexes used to calculate real incomes and output until they represent a significant level of expenditures. They are then rotated into the sample of products used for price index calculations, and subsequent changes in their price are taken into account in the usual way. It is only at that secondary stage, sometime long after the new product has been introduced, that it affects officially measured changes in real output.
He then provides an example of this happening. Statins, a class of drugs known to reduce cholesterol, was a major healthcare innovation in the second half of the 20th century. The impact of it was substantial:
In 1994, researchers published a five-year study of 4,000-plus patients. They found that taking a statin caused a 35 percent reduction in cholesterol and a 42 percent reduction in the probability of dying of a heart attack.
Statins eventually became the best selling drug in the United States by the mid 2000s, and the death rate from heart disease reduced by a third for people over the age of 65.
Feldstein then cites Grabowski et al. (2012), an empirical study looking into the economic impact of the medical breakthrough. The study finds that the value of lower healthcare costs as well as reduced mortality from heart diseases was $400 billion alone in 2008.
Grabowski et al. (2012) calculated that the combination of reduced mortality and lower hospital costs associated with heart attacks and strokes in the year 2008 alone was some $400 billion, which was almost 3 percent of GDP in that year. None of this value produced by statins is included in the government’s estimate of increased real income or real GDP.
Statin, as well as other new products, take years to potentially being factored into GDP and national income, leading statistics that show in face value that the US economy and incomes have been stagnating. Feldstein goes over other examples where new products were not included in price indexes:
The Boskin Commission (Boskin et al. 1996) noted that at the time of their report in 1996 there were 36 million cellular phones in the United States, but their existence had not yet been included in the Consumer Price Index. The earlier Stigler Commission (Stigler 1961) found that decade-long delays were also noted for things like room air conditioners. Autos were only introduced to the Consumer Price Index in 1940 and refrigerators in 1934. More recently, the Bureau of Labor Statistics has introduced procedures that cause new products to be rotated into the analysis more quickly, but only after they have achieved substantial scale in spending. These delays cause the price index to miss the gains from introducing the product in the first place as well as the declines in prices that often happen early in product cycles.
Since their inclusion is often delayed, official statistics don’t take into account the benefits new products and innovations provide to consumers, leading to the growth of output being underestimated and the overestimate of the rate of increase in the price index.
The paper goes into a detailed history of the underestimation caused by the methodology used by the US government. It is a great read and I suggest everyone to read it
Overall, the claim that wages and compensation have stagnated for decades is not true. Those statistics are based on using CPI for calculating inflation, which produces large biases that compound over decades. Those statistics also do not include all workers and all forms of compensation, such as healthcare benefits, 401Ks, stock options, etc.
