The Ecommerce Counterfactual

By / 3.12.2018

I’ve been arguing that the shift to ecommerce has improved the position of workers, by increasing the number of jobs and boosting wage payments. In a nice twitter discussion last week, Jose Azar pointed out that I had not specified a counterfactual, and he was right.

So here I will make up for that omission, at least a bit. Let’s start by laying the groundwork. The direct employment impact of ecommerce primarily shows up in three industries: Retail, couriers and messengers, and warehousing and storage. Most ecommerce fulfillment centers are reported in the warehousing and storage industry, though some are found in the electronic shopping, which is part of retail.  The companies that deliver the packages seem to be mostly reported in the couriers and messenger industry. And of course the brick-and-mortar stores hurt by ecommerce are in the retail industry.

Let’s look at the recent employment history of these three industries (see table below). We will focus on hours worked by production and nonsupervisory workers in what we call the “consumer distribution sector,” the  combination of retail, couriers and messengers, and warehousing and storage.

The Jobs Impact of Ecommerce
Percentage change, aggregate hours worked, production and nonsupervisory workers
Retail 3.8%
Couriers and messengers 19.9%
Warehousing and storage 33.5%
Total (consumer distribution sector) 6.3%
All private sector 5.8%
The consumer distribution sector is defined as including retail, couriers and messengers, and warehousing and storage


According to BLS CES data, aggregate weekly hours of production and nonsupervisory workers in the warehousing and storage industry are up 34% over the past 3 years, reflecting the rapid expansion of ecommerce fulfillment centers.  Aggregate weekly hours of production and nonsupervisory workers in the courier and messenger industry are up 20%, and hours worked in retail are up 4%

Taken together, hours worked in the consumer distribution sector are up 6.3% over the past 3 years. That’s faster than the 5.8% gain in hours worked in the overall private sector of the economy.

How does this gap compare to the historical pattern? We’ll look at the previous two business cycles, 1990-2000 and 2000-2007.

Let’s start with the 1990-2000 business cycle. Even though Amazon was founded in 1994, ecommerce was fairly insignificant for the labor market in this decade. In 2000 Amazon had only 6 fulfillment centers in the United States, and employed a grand total of 9,000 full-time and part-time workers globally. In total, ecommerce amounted to less than 1% of retail sales in 2000.

During this decade, hours worked in the private sector grew at a 2.1% annual pace. By comparison, hours worked in the consumer distribution sector grew at only a 1.6% annual rate.

The next business cycle was terrible for employment. Between 2000 and 2007, private sector hours growth slows to a 0.5% rate,  while consumer distribution sector hours rose at only a 0.2% pace.

However, it’s worth noting that the gap between  hours growth in the private sector and consumer distribution sector narrowed as ecommerce became more important. By 2007, ecommerce amounted to about 3.5% of retail sales.

From 2007 to 2017, the share of ecommerce rose to roughly 9%.  At the same time, hours growth in the consumer distribution sector accelerated to an 0.6% annual pace. The gap with the private sector narrowed even further, to less than 0.1 percentage points.

And when we focus on the last three years–the period of the supposed retail apocalypse–we see that hours growth in the consumer distribution sector is now outpacing private sector hours growth, even as Amazon and other online retailers have opened up ecommerce fulfillment centers all over the country.

How can this be? The short answer is that ecommerce is sucking unpaid hours out of the household sector. In effect, consumers are paying workers do their picking, packing, and driving for them. Ecommerce is hours-creating, rather than hours-destroying.

This analysis is indicative rather than conclusive, of course. But it suggests that ecommerce has had a positive effect on hours growth in the consumer distribution sector, relative to private sector hours overall.