The Productivity-Pay gap doesn't exist

Debunking a statistical anomaly

[Post by Dario Marino]

This post is the first of a series of 10 which will cover various issues related to productivity.

In this episode, we will debunk the productivity-pay gap and give an idea of what has historically caused productivity to increase. It has already been debunked in other articles such as this and this but I would like to use this wrong idea as an introduction for a Douglass North contribution that I find way more stimulating.

Several commentators and politicians have argued, especially in recent years, that since 1973, a gap has formed between wages and productivity. The myth comes from the Economic Policy Institute, a progressive think tank. The image showing the gap can still be found on the institute’s website:

In reality, productivity growth has not stopped influencing wage growth. A simple regression between productivity and compensation shows that the relationship shown earlier must have measurement problems, probably intentional.

In the work "Productivity and Pay: Is the link broken?" from Harvard University, Anna Stansbury and Lawrence Summers find that higher productivity growth is associated with higher average and median compensation growth. They use the median wage and the average wage of production and non-supervisory employees, as well as the average wage of all workers. In order to dampen economic cycle fluctuations, they calculate the three-year moving average of the change in inflation-adjusted compensation and labor productivity. Their conclusions:

"We find substantial evidence of a link between productivity and compensation: from 1973 to 2016, one percentage point more of productivity growth was associated with an increase in average and median compensation growth between 0.7 and 1 percent and with an increase between 0.4 and 0.7 percent of production/non-supervisory compensation growth. ... Our findings suggest that productivity growth is still very important for middle-income Americans."

The main reason for the curious structure of the original data was found to be caused by a serious methodological flaw, which occurs by examining only hourly wages and using the Consumer Price Index (CPI) to adjust for inflation.

First of all, by examining only hourly wages, only part of total income is taken into account, which excludes non-monetary benefits such as health insurance, pensions, etc. Benefits have grown over time, especially given the expansion of the American welfare state examined already in "The myth of American Inequality". Regarding income inequality between quartiles, for example, considering the various benefits received and taxes paid, the inequality ratio goes from 1:16 to 1:4. In 2012, legally established benefits accounted for over 20% of employee income. Secondly, the measures used to track inflation and productivity are not compatible. The Bureau of Labor Statistics uses the Implicit Price Deflator (IPD) to adjust productivity for inflation, which uses different measures and goods and services than the Consumer Price Index (CPI).

Conveniently, the CPI excludes quality improvement while the IPD does not. So the union’s think tank EPI compared productivity that takes into account quality improvements with steady prices, like consumer electronic goods from the 90s on, and the CPI instead compared the average income to an ever changing bundle of goods with higher quality goods at similar prices while productivity measured all the improvement that we had at the level of quality.

Furthermore, although the study includes total compensation, it only includes those who hold production positions. The comparison between compensation for some employees and productivity for all employees represents about 45% of the gap.

This further excludes unexpected payment methods such as performance-related bonuses. For example, self-employed workers see their productivity growth included in the data, but their wage growth is ignored. This represents almost 12% of the gap. However, the main flaw is once again adjusting inflation through the CPI, which represents about 39% of the gap. The CPI not only is incompatible with IPD measurements but also tends to overestimate inflation by not taking into account consumer responses to price changes or a "substitution effect".

Once these incorrect comparisons and biases are corrected, the gap is significantly reduced and becomes less relevant. As shown below:

During my (short) academic carreer, I have proposed an additional factor at the production cost level, but that study will be probably available in the future.

This short article was useful to say that a productivity rises does not occur when wages decrease and owners increase their profits. This correlation makes no sense; it has always been observed that wages have increased in sectors that have seen increased productivity. Therefore, the so-called Productivity-Pay gap is nothing more than a statistical anomaly caused by dubious methodological choices.

As Douglass North, an economic historian and Nobel laureate in Economics, explains in "Institutions and Productivity in History," the two main factors for increasing productivity are institutions and technological improvements. Institutional improvements mainly allow companies to operate with less frictions. We have four main categories of institutional improvements which are positive for productivity: those that improve capital mobility, reduce information costs, distribute risks, improve contract compliance. Obviously these improvements do not exclude each other, indeed they go together. Regulatory agencies reduce the costs of obtaining information on certain products, and can therefore encourage greater capital mobility given the respect of contracts that certain states can better understand and follow (thus reducing the risk of arbitrary decision by a state to increase the difficulty for foreign goods to enter the domestic economy). However, if regulatory agencies exaggerate they can also reduce capital mobility and increase the risk of trading with a certain country (or block of countries).

If this first improvement is the environment that allows companies to act, we can expect the other improvements to come directly from companies, and in this set we include all organizational and technological improvements that each company can adopt.

Historically, improvements such as double-entry bookkeeping can be considered a technological adjustment, while strengthening the navy that protected domestic companies' trade can be seen as an institutional improvement.

In the following articles we will mainly see what companies do to increase their productivity. The institutional side largely concerns whether government puts companies in the right conditions to do what is optimal to do. And the market analysis side concerns which economic forces incentivize companies to adopt measures that increase their productivity.

A particular case to be treated before starting to deal with company adjustments are industrial policies, when institutions and productions merge. When the state enters production it usually falls into events known as industrial policy. This will be the theme of the next article, in order to show how what we will deal with, namely the improvement of companies in their processes in a market economy, is mainly something to observe at a micro level, and not at a level of state decisions actively engaged in the production process.

You can read Noah Smith if you want to know already what argument I will challenge.