How productivity drives wages – The theory and evidence

Wages and productivity – The theory

I’ve written before about what I call the ‘real iron law of wages’. Simply put, this says that no hire will take place at a wage* above the employer’s estimate of what the employee will add to revenue. So, if the employer thinks the employee will add $10ph to revenue, they will pay up to that. If they pay $9ph, they are adding $10 to income and $9 to costs. And profit is just revenue – costs. The employer is making a profit of $1ph on the hire.

How much value an employee adds, in this case the $10ph they add to their employer’s revenues, is their ‘productivity’. An employer might like to pay this worker $6ph and make $4ph profit on the hire. But if another employee also thinks that worker will add $10ph to their revenues, they can offer them $7ph and still make $3ph profit. So, in a competitive labor market, wages would get bid up to something close to the $10ph figure for labor productivity. That is why economists often say that wages are linked to productivity. In this sense, higher pay is a result of higher productivity.

This is a model, to be sure, abstracting from real life. As Michael R. Strain of the American Enterprise Institute noted recently:

This theory leaves out a lot, of course. Pay and productivity can diverge for any number of reasons not included in the standard economic model. Workers may not know how much revenue they create, or what other employment options are available to them. And changing jobs has its own costs, which in the real world gives employers some power over wages.

Wages and productivity – The data

Strain goes on to note:

For critics of the current system, “some power” is a drastic understatement. In their telling, the decline of labor unions; erosion of the minimum wage; rise of non-compete and no-poaching agreements; inadequate enforcement of workplace standards and the like have dramatically reduced the bargaining power of workers. This has allowed businesses to drive down wages to the bare minimum job applicants and current workers will accept, pushing their pay below what their productivity suggests it should be.

This argument is based on data such as that presented in the chart below, a variation of which you are likely to have seen:

This chart shows average real hourly wages and productivity growth. It appears to show that, while productivity has doubled since 1973, average real hourly wages have fallen by 7%.

But, as I’ve written before, that isn’t the whole story.

The economist James Sherk answered this question in an excellent paper back in 2013.

Compensation is what matters, not wages

Using wages as a variable excludes all kinds of worker compensation such as health insurance, retirement benefits, and paid leave. These have become more important over time. In 1973, non-wage benefits accounted for 13% of employee compensation. By 2012 that figure had risen to 20%.

Furthermore, the wage figures usually come from the Bureau of Labor Statistics (BLS) payroll survey. This covers only the pay of “production and non-supervisory” employees and excludes managers and many salaried employees. These figures also exclude bonuses and other irregular cash payments, thus missing many forms of performance-based cash pay which have become much more common since the 1970s. As a result, the payroll survey misses these pay increases.

There is data on total compensation which, in addition, covers all workers, including managers and salaried employees. If we use the more accurate, comprehensive data for compensation over that for wages, we get the chart below. As Sherk writes:

“While hourly cash wages measured by the payroll survey have fallen 7 percent since 1973, total compensation as measured by LPC has risen 30 percent. Part of the apparent gap between pay and productivity stems from not including all elements of employee earning and using different data sources.”

More accurate measures of inflation

The purchasing power of money changes over time. How do we account for this?

The BLS adjusts productivity for inflation using the Implicit Price Deflator (IPD). But most analysts adjust wages and compensation for inflation using the Consumer Price Index (CPI). Sherk explains that:

“These two inflation measures are not directly comparable. They use different methodologies and cover different goods and services. Comparing CPI-adjusted compensation growth to IPD-adjusted productivity growth produces inaccurate conclusions.”

For a variety of reasons, the CPI is inferior to both the IPD and the personal consumption expenditures index (PCE). If we use the former to adjust for inflation, we get the following chart. This shows that while CPI-adjusted real compensation grew 30% over the past four decades, IPD-adjusted real compensation grew by 77%.

In summary,

Sherk notes that “Much of the apparent divergence between pay and productivity stems from using different surveys and formulas to calculate inflation.” Selecting more representative variables, using more accurate statistics, and making use of a better measure of inflation largely closes the supposed gap between earnings and productivity that appears to have opened up since the early 1970s. The argument that productivity drives earnings holds up. If we want higher wages, we must have higher productivity.

There is academic support for Sherk’s argument. In a 2017 paper titled ‘Productivity and Pay: Is the link broken?‘, economists Anna M. Stansbury and Lawrence H. Summers note that “Since 1973 median compensation has diverged starkly from average labor productivity”, as seen in the first chart above, and also that “Since 2000, average compensation has also begun to diverge from labor productivity.” They ask “to what extent does productivity growth translate into compensation growth for typical American workers?” They find:

…substantial evidence of linkage between productivity and compensation: over 1973-2016, one percentage point higher productivity growth has been associated with 0.7 to 1 percentage points higher median and average compensation growth and with 0.4 to 0.7 percentage points higher production/nonsupervisory compensation growth. 

A more recent paper by economist Edward P. Lazear titled ‘Productivity and Wages: Common Factors and Idiosyncrasies Across Countries and Industries‘ finds that “Average wage growth is closely related to aggregate productivity growth across countries and within countries over time”. This is the case for low-, middle- and high-wage workers, who all benefit from growth in average productivity. This suggests that improvements in overall economic efficiency help all workers, not just the rich.

Of course, talking about averages can obscure as much as it reveals. If the average productivity in an economy grows by 10% but that of an individual workers doesn’t, you wouldn’t expect to see that worker’s wage rise by 10%. After all, employees hire based on their employer’s estimate of what they will add to revenue, not an estimate of what the economy’s productivity gains will be.

To address this, Lazear looks at productivity at the industry level, comparing industries that employ highly skilled workers with those that employ lesser-skilled ones. This allows him to investigate whether changes in the productivity of, say, low-wage workers affect the pay of that specific group.

Lazear finds that finds that productivity in industries dominated by higher-skilled workers increased by about 34% between 1989 and 2017 with the wages of those workers rising by 26%. For industries requiring lesser skills, productivity increased by 20%, while wages rose by 24%. As Michael Strain explains:

In other words, pay increased faster than productivity in industries with lesser-skilled workers, and slower than productivity in industries with higher-skilled workers. Another striking implication of this finding is that “productivity inequality” — the gap in productivity between workers — may have grown faster than wage inequality over this period. While wage differences have increased over time, differences in productivity between groups of workers have increased even more.

The upshot: Slower wage growth for lesser-skilled workers is not prima facie evidence that employers have significant power over wages or that productivity doesn’t determine wages. Instead, Lazear concludes that productivity growth for high-skilled workers has increased rapidly enough (actually, more than enough) to account for growing inequality.

How do we explain this?

Technological change disproportionately benefits the highly skilled, increasing their wages and productivity. And the globalization-led shift to a services economy has reduced the productivity of goods-producing, lesser-skilled workers.

Lazear also suggests that colleges may have improved more than high schools in their ability to impart skills to graduates. If so, industries dominated by college graduates would be expected to have had faster productivity growth over the last three decades. This would have caused both a wider dispersion in productivity across industries and in wages across groups of workers.

What does it all mean?

Two important conclusions flow from this research.

First, the idea that the best way to increase wages is to increase productivity holds. Government attempts to hike wages by diktat, ignorant of the underlying economic forces, are, at best, pointless, and, at worst, actually harmful.

Go back to our initial example of the worker who the employer estimates will add $10ph to revenues. If a minimum wage is set at, say, $15ph, then hiring that worker will now add more to the employer’s costs (the $15) than to revenues (the $10). As profit is just revenue – costs, in this case the hire would result in a loss of $5ph. No employer would hire a worker in such a basis. The actual result of the diktat would be a wage of $0ph.

Second, it reinforces the argument that wages and earnings are driven, ultimately, by economic forces; they are not, in the main, set nor manipulated by nefarious individuals.

It has become depressingly common to hear, even from respected economists like Joseph Stiglitz, that the divergence between wages and productivity – which doesn’t exist in any case, as we’ve seen – is the result of conscious decisions taken by a handful of mean oligarchs. What this research shows is that, instead, such trends in earnings and productivity as actually have been seen have been driven by impersonal forces such as globalization and technological change. The former makes for a simplistically memorable ‘goodies vs baddies story’, but the latter is actual, fact based analysis.

*In this post I used wage/wages interchangeably with total earnings or remuneration, which includes things like health insurance. I probably shouldn’t for the sake of exactness, but for the sake of ease of reading, I’ve gone with it.