Introduction
Retail employment is in trouble: in February and March 2017, employment declined by 56,000. An Axios article greeted the news with dismay, noting that "retail workers are being displaced in droves." The Economist remarked that retailers are "a giant established industry ... descending into crisis." Edward Luce of the Financial Times was more blunt: "America faces a retail crunch as big as the manufacturing crisis of the last generation." Is that so? What exactly is happening to retail jobs?
Employment over time
In the grand scheme of things, there is nothing unusual about the job losses seen so far in 2017. Employment in the retail sector fluctuates a lot—sometimes it grows and sometimes it shrinks, even when the economy as a whole is growing. In fact, if we look all the way back to 1939, and we exclude all months when the economy was in a recession, employment fell about 19% of the time (151 months of job losses out of a total of 797 months). So in any given year, we should expect about two months of job losses.
Retail Employment (in millions, seasonally adjusted)
Source: Bureau of Labor Statistics, Current Employment Statistics
Nor are the job losses unprecedented. Employment fell by over 20,000 five different times between 2010 and 2016. And it was only recently, in early 2014, that employment last fell two months in a row—with 44,000 jobs lost. Of course, this does not mean that there is no underlying trend; or that the job losses in 2017 are not worrisome; or that a broader structural shift is not happening. Not at all. It is merely a caution against reading too much into two months of data.
Breaking down retail
Retail employs almost 16 million Americans—roughly 11% of total non-farm employment. It is also a broad category that includes department stores, pharmacies, gasoline stations, liquor stores, motor vehicle dealerships, supermarkets, furniture stores, and many others. It encompasses Walmart but also Amazon, CVS and Office Depot. To understand retail employment, it is essential that we disaggregate the totals into smaller chunks.
The most convenient disaggregation is into twelve sub-sectors (industries with a three-letter code under the North American Industry Classification System, or NAICS). If we visualize employment since 1990 across these twelve main sub-sectors, we observe several trends.
Retail Employment by Main Subsectors (in millions, seasonally adjusted)
Source: Bureau of Labor Statistics, Current Employment Statistics
First, three sectors account for about 50% of total employment: general merchandise stores, which includes department stores; food and beverage stores, which includes grocery and liquor stores; and motor vehicle and parts dealers. Other sectors with employment above 1 million are: building material and garden supply stores; clothing and clothing accessories stores; and health and personal care stores. Together, these six sub-sectors make up three-quarters of all employment in retail.
Second, these sectors do not, typically, follow the same cycle. Employment in some sectors peaked in the 1990s; in others, it is still growing. Some sectors rebounded sharply after recessions, others struggled to bounce back (recessions: July 1990—March 1991, March 2001—November 2001, December 2007—June 2009). Most sectors have grown in recent years, although some have not (clothing and clothing accessories stores, electronics and appliance stores; sporting goods, hobby, book, and music stores). Total retail employment is, thus, an aggregation of many different sub-sectors with different cycles.
Third, and this is not immediately obvious from the chart, the recent losses have come almost exclusively from general merchandise stores. About 79,000 jobs have been lost between November 2016 and March 2017. Of course, here too, we need to place numbers in context: if one looks at the last twelve months, job losses are more modest—roughly 31,000, or about 1% of total employment in general merchandise stores (3.1 million in April 2017). Either way, it is clear that this sector is shedding jobs faster than any other.
General merchandise stores
General merchandise stores consist of four categories: (a) department stores, except discount; (b) discount department stores; (c) warehouse clubs and supercenters; and (d) all other general merchandise stores. Since 1990, these four sub-sectors have experienced vastly different trajectories.
Employment in non-discount department stores started to decline in 2000: since then, jobs have almost halved, as more than 400 thousand jobs have been shed. Discount stores fared better and kept growing through the 2000s; but starting in 2011, they experienced a sharp drop from which they have yet to recover—employment is about 18% lower (minus 176 thousand jobs). Meanwhile, the real growth has come in warehouse clubs and supercenters—which encompasses outlets such as Walmart. Since the early 2000s, employment has almost doubled from about 750 thousand jobs to over 1.4 million. Employment in all other stores has shown a more modest increase.
Retail Employment in General Merchandise Stores (in millions, seasonally adjusted)
Source: Bureau of Labor Statistics, Current Employment Statistics
If we look merely at the last twelve months, the jobs losses have come mostly from discount stores (-28.5 thousand) and secondarily from warehouse clubs and supercenters (-17.8 thousand) and non-discount department stores (-13 thousand). In other words, the recent weakness reflects: (a) the long-term structural decline in non-discount stores; (b) the continued inability of discount stores to rebound from their 2011 slump; and (c) the recent weakness in warehouse clubs and supercenters, which was, until recently, the main job creator under general merchandise stores.
The big structural shift
What comes next? To answer this question, it is helpful to revert back to the macro view: the role of retail employment in the American economy. For about two decades after World War II, retail made about 10% of total, non-farm employment. Then, around the late 1960s, its share started to rise. This coincided with the rise of shopping malls and the gradual transition from mail-order to in-person shopping.
Retail Employment (percent of total non-farm employment)
Source: Bureau of Labor Statistics, Current Employment Statistics
This change is visible if one looks at employment by the U.S. Postal Service. Through the late 1960s, there is a spike in employment around Christmas, when the post office would add about 300,000 jobs. By the early 1970s, that seasonality has dissipated, and by the mid 1970s, it has disappeared. Shopping has moved to the store—the mall, to be precise. This shift lasts through the mid to late 1980s, when retail employment reaches its peak—making up over 12.2% of total employment in August 1988 (at its height).
Then comes a steady drop, and by 2017, retail employment comprises about 11% of total employment. In part, this is just a matter of relative growth: most retail sub-sectors are still adding jobs in the 1990s, just not as quickly as other sectors of the economy. In the 2000s, we observe a more obvious disruption, which, in part reflects the decline of the department store. At first, the disruption comes from warehouse clubs and supercenters, so the net employment effect is minimal.
But more recently, even warehouse clubs and supercenters are in trouble as shopping shifts to the web. The jobs in non-store retailers, which includes Amazon, are growing but the growth is modest—just about 25,000 jobs added over the past twelve months. Yet, once again, we see that the delivery of goods is producing a transportation boom. Not in the US Postal Service this time but in warehousing and storage and in couriers and messengers. Relative to the pre-recession peak, there are an additional ~300,000 jobs in warehousing and storage, and another 50,000 in couriers and messengers.
Employment by U.S. Postal Service and in Warehousing and Storage (in millions, seasonally adjusted)
Source: Bureau of Labor Statistics, Current Employment Statistics
In other words, the American economy is shifting back to the mail-order world: sales happen through catalogs rather than salespeople at the store. In today's world, of course, the catalog needs not just writers, designers and photographers, but also programmers, developers and engineers. But this shift is pulling transportation along as people staff warehouses and drive trucks to deliver goods to consumers. If retail were to fall back to its post-World War II average—roughly 10% of total employment, we could expect another ~1.3 million jobs lost. Some of these will be forever gone due to technology or automation; but others will show up in different sectors, as American consumers increasingly rely on the modern-era catalog for their purchases.
Notes
For more information on the data sources and approach, please visit my GitHub repository.