Retail apocalypse? More like metamorphosis.

It’s game on in the debate over whether the internet and other technologies are killing retail jobs, according to economist Michael Mandel. In his analysis, Mandel examines growth in occupations that are closely related to e-commerce, such as warehouse jobs, and finds that it outweighs losses in traditional retail occupations. Mandel’s findings are certainly controversial, but the assumption underlying his second look at this data is sound.

Occupations and jobs are not naturally occurring roles with fixed sets of characteristics, as we tend to think of them. Instead, each category is an evolving concept that is created by people within markets, industries and government.

The tasks that make up a particular job shift over time as technological capabilities change. It is easy to cast the stories of occupational decline as black and white, either jobs are destroyed or they are created. But the situation is actually much more complex, with tasks being shifted into and out of different job categories.

As more of the tasks that make up one job shift elsewhere, that job category will look like it is disappearing, but that neither means that the job is being stolen by technology, nor does it mean that this shift will result in fewer jobs overall.

Take typing for example: the Bureau of Labor Statistics (BLS) includes in its occupations list a category called Typist — those who “use word processor(s), computer(s) or typewriter(s) to type letters, reports, forms, or other material from rough draft, corrected copy, or voice recording.” In 1983, there were 906,000 typists in the US; last year there were 67,000.

This decline in the number of typist jobs is not due to technology taking over the primary task of typists; people still type (as I am doing right now). Rather, the decline stems from technology changing how this task is executed and therefore changing who does this task. Typing is no longer a specialized task for only some workers, but a general one that most workers engage in.

Like typing, pumping gas or operating an elevator, new technologies have made certain retail tasks — such as checking out at a store and collecting information to compare products across a variety of dimensions — easy enough for a non-specialist to do them. Many tasks that we once hired someone (such as a clerk or a salesperson) to do for us, we now do ourselves.

This movement of tasks from and between different job categories also means that certain occupations that we once thought fit firmly within one industry, turn out later to be industry-spanning occupations. That is what Mandel argues: warehouse jobs used to just be warehouse jobs, but with the rise of Amazon, some warehouse jobs are now also retail jobs (the same could probably be said for parcel delivery jobs).

Unfortunately, the categories — of industries, jobs and tasks — that we come up with do not adapt as fast as technology progresses. Just like when you wear old glasses to read, if the lens you are using to look at the data no longer fits, you won’t get a clear picture of what you are trying to see. We should be cognizant of how old categories are distorting our read of current data, so we can determine whether we are really seeing what we think we are seeing.

To get more labor market insights sent directly to your inbox, sign up for Cathy’s weekly newsletter here: