Among the most difficult thing for individual workers to understand is the overall pace of productivity changes within their own industries or occupations. We all tend to be close to our own work, and don’t readily observe what happens around us to employment and production, the two factors that dictate productivity growth.

Even if our industry is dominated by technology, we often fail to see how this affects production within our firms.

This failure is due to two truths. One is that we just cannot see or observe much beyond our own rather limited span of control. Second, most workers today are using technology as a complement to their labor skills, not as a substitute for work. The simple reason for this is that if technology were a labor substitute, those workers wouldn’t be working at that firm any longer.

For these reasons, we don’t observe the jobs that aren’t created due to productivity growth, and so miss the breathtaking growth of productivity in U.S. firms. Let me use a quick example.

During the past 20 years (1998 through 2017) Indiana’s factory employment declined significantly, from roughly 651,000 to 530,000 workers. During the same time, the gross domestic product of Hoosier factories rose from $73 billion to $93 billion in inflation-adjusted terms. This GDP measurement doesn’t include imports from outside the state or nation, so it is very close to the actual value of manufactured goods in the state.

To put it another way, in 1998, each Hoosier factory worker produced $124,000 in goods each year. When accounting for inflation, this jumped to $182,000 per year in 2017. So, by 2017 it took only 680 workers to produce what 1,000 factory workers could make in 1998. This is a strong period of productivity growth, so it is worth thinking about the cause of productivity changes over this time.

From 1998 to about 2010, productivity growth was very strong. Since 2010, it has been relatively slow, resulting in a small growth in the number of factory jobs to meet the growing demand for goods during the recovery. Productivity can be measured many ways, but I’ll focus just on the value of goods produced by each worker in the industry.

Growth in the value of goods produced by each worker can have many causes. Historically, it has been due to workers having available more machinery, transportation equipment and the like. It might be that workers are getting better, either through better training and education or through the removal of less-qualified or less-skilled workers. Factory processes can also get better with new techniques or processes enabled by technology.

During the past two decades, the relationship between productivity and the purchase of capital investment in machinery or equipment seems to have broken down. While this might be merely a statistical or data collection problem, it may be that the price of machinery and technology doesn’t well reflect the business value of capital investments. This is especially true for modern technologies. A used dump truck can cost more than a 25-computer network, with advanced website, with inventory and point of sale systems. These likely have very different productivity effects, but look the same to someone doing a statistical model of capital investment and growth.

A far more likely explanation for productivity growth is human capital deepening. Here in Indiana, the relationship between factory productivity growth and the changing share of factory workers with a college degree is astonishingly high. From 1998 through 2010, the share of workers with a college degree rose from 48 percent to 51 percent, while the value of goods produced by the average Hoosier factory worker grew by 5 percent per year.

Since 2010, the share of college-educated factory workers dropped back below 50 percent and productivity declined an average of 0.6 percent per year.

There is more to the story. Union membership in Indiana factories plummeted 63.9 percent from 1998 through 2009, as productivity soared. Since 2009, union membership increased about 15 percent, making up a small share of the 1998 membership, but productivity growth slumped. The research evidence suggests unions are big drags on productivity, despite what many business leaders say. This clearly remains a contemporary issue.

Finally, factory processes may be very sensitive to worker skills and education. It is likely mundane and relatively low-cost production techniques, such as computerized inventory management and statistical process control, have dramatic effects on productivity. These processes require much differently trained employees than the pre-computer-age factory floor.

There is also growing evidence that technological change of this type requires more adaptive learning by employees. Some of the details of how this will emerge remain unclear, but what is sure as the sunrise tomorrow is that better trained men and women remain the key to productivity growth in factories and across the broader economy.

Michael J. Hicks is the director of the Center for Business and Economic Research and an associate professor of economics in the Miller College of Business at Ball State University. Send comments to [email protected].