Is Amazon Prime Killing the Planet?

by Larry Snyder

With Amazon Prime, we’re always just a click away from a UPS truck making a special trip to our homes to deliver a package of lightbulbs or a bag of dog food. Over the past few years, I’ve occasionally heard people worry that Amazon Prime will increase carbon emissions and clog the roadways.

Enter: Amazon’s Echo device, which lets us order dog food without even clicking. This seems to only exacerbate the concern.

It is, of course, true that more Prime ordering means more UPS deliveries, and therefore more emissions and traffic from those deliveries. But worrying about higher emissions and worse traffic overall from Amazon Prime means forgetting an important fact:

Increasing the load on UPS trucks will reduce the traffic from passenger cars that are making solo trips to the hardware or pet store.

That’s because you might drive a few miles to the hardware store, whereas UPS might be able to stop at your house on the way to your neighbor’s, adding negligible miles.

This is essentially the difference between truckload (TL) and less-than-truckload (LTL) logistics. When you drive straight to the hardware store and back, you are making a TL-type pickup. When UPS delivers to your house, it uses LTL logistics, which allows it to consolidate many deliveries onto the same vehicle.

As a result of this consolidation, UPS may burn far less fuel, drive far fewer miles, and cause far less road congestion than we would experience if Amazon Prime customers instead drove to local brick-and-mortar stores, even accounting for the poorer gas mileage of UPS trucks compared to standard passenger automobiles.

To demonstrate this possibility, I coded a simple simulation in which there are 100 customers randomly located in a 100-mile-by-100-mile region, and 5 stores — let’s call them hardware stores — randomly located in the same region. Each customer drives to her nearest hardware store in a car that gets the average car fuel economy of 23.4 mpg.

Those trips represent a total of 2137.5 miles, and 91.3 gallons of fuel.

Now suppose that these customers instead order their products from Amazon. My simulation assumes there are two UPS distribution centers (DCs) located in the region, that customers’ demands are each between 0 and 50 cu ft, and that delivery trucks have a capacity of 1,000 cu ft with a fuel efficiency of 10.2 mpg. It also assumes that in addition to these 100 customers, there are another 50 existing customers that UPS would be delivering to anyway, whether or not our new customers order from Amazon.

In this scenario, UPS trucks drive a total of 492.0 miles, burning 48.2 gallons of fuel. That’s a 77% reduction in total miles and a 47% reduction in fuel.

My simulation leaves out a number of important factors when asking whether Amazon Prime is, on balance, environmentally beneficial or harmful. For example:

• Inbound logistics. My simulation models the outbound logistics only. My hunch is that the inbound logistics would also favor Amazon Prime, because Amazon’s supply chain (even including the UPS component) is probably much more consolidated and less fractured than brick-and-mortar retail supply chains. More consolidation means more efficiency.
• Packaging. Amazon Prime shipments typically involve much more packaging, both cardboard and plastic, than products sold in stores. This takes Amazon Prime’s environmental score down a big notch.
• Running errands. My simulation assumes that customers make a round trip to the hardware store and back home. It’s entirely possible that customers make other stops on the same trip — to the pet store, the grocery store, the bank — in which case the brick-and-mortar trips look more like LTL routes and are more efficient (for the same reason that consolidation makes the UPS routes more efficient). On the other hand, if the items purchased at those extra stops could also be purchased from Amazon, then that would hurt the brick-and-mortar strategy since Amazon becomes a one-stop-shop that replaces multiple local stops.
• Increased consumption. It’s possible that we consume more simply because it’s more convenient to do so under Amazon Prime. This would, of course, hurt the environmental impact of Amazon Prime versus local stores.
• Creating a behemoth. This entire analysis ignores the social, financial, labor, and other impacts — good and bad — of a mega-retailer like Amazon. In the extreme case where everyone buys everything from a single retailer, my analysis suggests that the environmental impact may be positive, but I am by no means arguing in favor of such a direction — the other implications would be disastrous.

Other people have looked at this question of UPS deliveries vs. brick-and-mortar shopping, with generally similar conclusions as mine.

1. A study by Carnegie Mellon researchers found that e-commerce uses less energy and fewer emissions than brick-and-mortar retail, even taking packaging into account.
2. A master’s thesis conducted at the MIT Center for Transportation & Logistics had a similar conclusion.
3. On the other hand, a study conducted by the Simon Property Group (a major owner of shopping malls) concluded that online shopping can result in 7% more emissions, though at least one supply chain expert disagrees with the assumptions of that study.

By the way, I tried to make conservative assumptions in my simulation. For example:

• An average customer demand of 25 cu ft is far too large. In reality, customer orders are much smaller, and trucks make over 100 stops per day rather than the 40 implied here. More realistic numbers would mean more efficient deliveries, and a bigger delta between the Amazon and brick-and-mortar scenarios.
• The ratio of new to existing customers is probably much smaller than the 100:50 ratio here. For every customer that switches from local stores to Amazon, there are probably already several existing UPS customers. That, too, would make UPS deliveries more efficient (since we could fit those customers onto existing routes) and make the Amazon scenario look better.
• In a 100x100 region, there may be more or fewer than five hardware stores. Having more hardware stores would help the local-store scenario; having fewer would hurt it.
• In a 100x100 region, there may be more or fewer than five UPS DCs. The directional trend is less clear here: Having fewer DCs would allow more consolidation of loads, but would also entail longer hauls for the delivery trucks.

Still, there is a lot of wiggle room in these numbers. You can change them significantly without changing my overall conclusion, which is that:

Amazon Prime deliveries are much more efficient, from a mileage and fuel perspective, than brick-and-mortar shopping.

If you liked this blog post, check out more of our work, follow us on social media (Twitter, LinkedIn, and Facebook), or join us for our free monthly Academy webinars.