Deutsche Bank “Breaks Through the Noise” on Amazon and the Post Office

SteveContent type, Slideshow, Story

Back in April 2018, the trade publication FreightWaves published an article about a new research report done by Deutsche Bank entitled “Breaking through the Noise on AMZN, USPS & FDX/UPS” (April 24, 2018).  The report responds to Trump’s tweets that the U.S. Postal Service is undercharging Amazon by $1.50 per package, and it evaluates the potential effects of various possible price hikes.

As explained by the FreightWaves reporter, John Paul Hampstead, “The questions Deutsche Bank’s analysts were trying answer were essentially two: is it true that the USPS loses $1.50 on every Amazon package it delivers, and what would be the effects of a USPS price hike?  The answer to the first question is ‘no,’ and the answer to the second is ‘Amazon would probably be fine, and FedEx and UPS would benefit.’”

Aside from FreightWaves, media outlets did not take note of the Deutsche Bank report.  Deutsche Bank may have wanted to break through the noise on Trump’s tweets, but it chose to do so rather quietly.

The DB report was not available online, so we weren’t able to see it when we did our previous post, “An Amazon puzzle: How many parcels does it ship, how much does it cost, and who delivers what share?”

But a few days ago the DB report appeared among the source materials on the new website of the Package Coalition, which was created this week by Amazon and other stakeholders in the parcel business to “to preserve an affordable, reliable and profitable postal package delivery system.”

It turns out the DB report has an excellent analysis of Amazon’s volumes and shipping costs.  And while it differs in some interesting ways from our analysis in “An Amazon Puzzle,” the two solutions to the puzzle are actually quite close.

Before we get to the DB numbers in more detail, though, let’s return to the question that prompted the DB report: Does the Postal Service lose $1.50 on every Amazon package it delivers, as Trump has tweeted?  


The “miscalculation” behind Trump’s tweets

Back in early April, we addressed that question in several articles, including “Talk about Fake News: How a flawed Citigroup analysis led to Trump’s bogus tweets about Amazon and the Postal Service.”  As explained in these posts, Trump seemed to be basing his claims on an op-ed in the Wall Street Journal by Josh Sandbulte that cited a Citigroup report saying that parcels were underpriced because they weren’t covering their share of the Postal Service’s institutional (aka “fixed”) costs.

According to Citigroup, competitive products (aka shipping services) were covering 5.5 percent of institutional costs, rather than a figure that reflected how much revenue these products generate, at the time around 24 percent of total USPS revenues. If the Postal Service had to increase parcel prices to cover this larger share of institutional costs, said Citigroup, it would need to raise parcel prices by about 50 percent.

The overall average parcel costs about $3.50.  (That’s for all parcels, including Ground, Priority, etc.  In FY 2017, competitive products generated $19.5 billion in revenues on 5.7 billion pieces).  Ciitgroup therefore figured the Postal Service would need to raise prices by over $1.50 per package.

The Citigroup analysts apparently didn’t understand that this 5.5 percent was only a floor set in 2006 by the Postal Regulatory Commission.  In actuality, competitive products have been contributing much more than that — nearly 24 percent of institutional costs at the time. In other words, there really was no foundation to the claim that Amazon parcels were underpriced by $1.50.

The DB report makes exactly this point in explaining why Trump’s tweets were wrong.  Without citing the Citigroup report, DB writes:

The $1.50 Loss per Package Estimate Seems to be a Miscalculation. Critically, USPS package-related revenue net of ‘attributable cost’ covered 23.6% of total institutional costs in FY2017, well above the above-mentioned 5.5% threshold – making the losses associated with package deliveries likely significantly lower than the $1.50 loss per package mentioned in a recent Presidential tweet – it appears this number is being calculated assuming the 5.5% allocated cost metric, when in fact, it’s significantly higher.”

It’s too bad the DB report hasn’t been more widely distributed.  It might have helped to quiet the noise about Trump’s bogus claims.  Unfortunately, the tweets led to a Task Force on the Postal Service, with a price hike on its agenda.  The Task Force’s report comes out on Friday of this week.  If it mandates a price hike for parcels, the Deutsche Bank report may turn out to be a useful guide.


Possible effects of a price hike

The main purpose of the DB report is to offer advice to investors, so the report focuses on several scenarios involving potential price increases.  Here’s what DB concluded:

  1. A price increase of 9.5 percent on a Ground package, about $0.21, would increase the competitive services contribution rate to 30 percent (from the current 23.6 percent), which is more in line with the current revenue contribution.  This price hike would increase Amazon’s fulfillment costs by $378 million.
  2. A price increase of $0.26 results in the USPS reporting breakeven controlled income — competitive products cover all the shortfall in institutional costs.  This rate hike would increase Amazon’s fulfillment costs by $484 million.
  3. A price increase of $1.50 per package, per Trump’s tweets, would increase Amazon’s fulfillment costs by $2.76 billion.
  4. A price increase of about $1 per package — just to split the differences in the other scenarios — would increase Amazon’s costs by $1.8 billion.

One of the main effects of a price hike by the Postal Service would be that UPS and FedEx could raise their rates too, by a similar amount, without fear of losing business.  Deutsche Bank estimates that a 10-percent increase in USPS Ground prices would “conservatively” add about $400 million to FedEx revenues, most of it as profit, and over $700 million to UPS revenues.

The DB report doesn’t examine how such an increase would impact the Postal Service, but it would obviously increase the profit margin on parcels (i.e., increase their contribution to institutional costs).  But higher prices could also drive business to the private carriers, as well as motivating Amazon to do more deliveries itself through programs like Flex and Delivery Service Partners.

For Amazon, DB says that such an increase would translate into about 40 cents per package overall and add about $380 million in incremental costs, “which while significant, is well below fears implied by headlines.”  In any case, these increased costs on Amazon deliveries would just be passed on to consumers in higher prices.


The Deutsche Bank and STPO analyses compared

In order to get to its estimates about the effects of a price hike, the DB analysis first needed to do some figuring about Amazon’s volumes and shipping costs.  Here’s the key table from the DB report (click on the table for a larger version):

The DB analysis is worth going through line by line.  While we’re at it, we’ll compare it to our analysis in the Amazon Puzzle piece.  In order to do that, though, we need to make an adjustment for the DB numbers.

Amazon and the Postal Service are on different fiscal calendars.  The Postal Service’s fiscal year runs from Oct. 1 to Sept. 30, while Amazon is on a regular Jan. 1 to Dec. 31 calendar.  DB therefore used Amazon’s quarterly reports to produce numbers that would correspond to USPS FY 2017; DB calls this 2017A.  (Amazon’s shipping costs are going up about 35 percent a year, so going back in time by even one quarter has a big impact on the calculations.)  In order to compare DB’s numbers with ours, we have to adjust the DB’s numbers so they represent data for Amazon’s actual fiscal year.

The following table shows the DB data in column B (2017A), the adjusted DB numbers in column C (FY 2017), and our Save the Post Office numbers in column D.   Notes follow.

2017 worldwide shipping costs (row 2): DB estimates total shipping costs for 2017A of $19.98 billion.  The actual costs for FY 2017 were $21.7 billion, about 9 percent more than DB’s number.  Most of the other numbers in column C reflect this 9 percent increase.

U.S. revenue as percent of total revenues (row 3): DB estimates that U.S. costs were 63.8 percent of total shipping costs.  Our analysis used 66 percent.  Both estimates are based on the proportion of domestic revenues for N. America compared to international revenues (numbers that appear in the Amazon 10-K), but DB figured in that 98 percent of N. America are U.S.  Our analysis didn’t get quite so granular. 

U.S. shipping costs (row 4): DB estimates total shipping costs for the U.S. of $12.775 billion for 2017A, and hence $13.9 billion for the actual FY.  Our analysis didn’t make an estimate on this, but based on our earlier numbers, it would be about $14.3 billion.  These totals include all shipping costs — both the cost for outbound delivery with the Postal Service and private carriers, plus (as the Amazon 10-K explains) costs associated with sortation, delivery centers, and transportation.

U.S. volume (row 5): DB estimates a total of 2.554 billion Amazon packages for the U.S. in 2017A, which comes to about 2.78 billion for FY 2017.  Our estimate was 2.4 billion.

The numbers are in the same ballpark, but they were arrived at in much different ways.   We came up with the estimate by looking at media reports and guessing.  Here’s how DB came up with its number for volumes.

As we noted in our analysis, many packages contain more than one item.  DB actually uses this fact to estimate total packages.  Its estimate for the U.S. is based on 4.26 billion units, with 1.67 units per package (1.34 units/package for Prime, and 2 units/ 2,554 packages for Non-Prime shipments and 50:50 split between the two).  It’s not clear where DB got its numbers for total units and units per package, but the report contains a detailed table explaining the calculations.

Average cost per package (row 6): These averages are for all shipping costs (sortation, transportation, and outbound) in the U.S.  DB figures about $5 per piece; our analysis suggests about $6 per piece.

Volume shipped in the U.S. by other carriers (row 10): DB didn’t disaggregate by carrier, but our totals are similar: DB, about one billion; STPO, about 1.3 billion.  Most of the discrepancy is due to the fact that DB gives USPS a much larger market share (see below on row 17).

Average Shipping Cost per package for other carriers (row 11): For the other carriers, DB estimates an average total shipping cost of $6.55. To come up with that figure, DB takes the $8.20 average cost per piece for UPS Ground shipping (as reported in the 10-K), and then figures a 20% volume discount.

Given how it was derived, the estimate apparently includes only the carrier costs and not any other shipping costs incurred before handing off to the carrier.  DB includes both costs when it looks at USPS, but for some reason, it doesn’t do so for the other carriers.

It’s not clear why DB figures things this way.  It doesn’t seem possible that there wouldn’t be significant  costs for sortation, transportation, etc., before the package gets handed off to UPS and other private carriers. Amazon’s package processing network wasn’t developed just for USPS Parcel Select.  It’s huge, it’s international in scope, and it is also designed to take advantage of short-zone rates with the private carriers.

Our original estimate had not included a number for average total cost per package for other carriers, but we figured an average carrier cost of $4.50 — due to much steeper discounts than 20 percent — so let’s allocate about $2 for these other costs, for a total average shipping cost of $6.50 for the other carriers.  That’s about the same number that DB uses, but ours includes both pre-outbound and outbound costs.

Volume Shipped through USPS (row 17): DB estimates that 1.53 billion pieces went through the Postal Service in 2017A, which comes to about 1.67 billion for FY 2017.  That’s a lot higher than our estimate of 1.1 billion because DB assigns a much larger market share to USPS.

DB estimates that the Postal Service delivered 60 percent of domestic packages, whereas our estimate was 45 percent. DB doesn’t say much about how it came up with this number: the table just says, “DB estimate, based on our internal survey, and confirmed by industry experts.”  Since the estimate seems pretty high, it’s worth a closer look.

Elsewhere in its report, DB says this: “Since 2012, the overall volume of USPS’s package business doubled, and according to UPS, the USPS delivers over 40% of all Amazon packages.”

So, in one part of the report, DB uses the 40 percent figure and in another, 60 percent. Why the apparent discrepancy?  It’s possible that DB is thinking that USPS delivered 40 percent of Amazon packages worldwide, and 60 percent of US volumes.  The math works out that way, so it makes sense.

But where did DB come up with the idea that USPS delivered 40 percent of Amazon worldwide?  The report says “according to UPS,” but there’s no footnote to check further.  Most likely the number came from a UPS filing with the PRC, submitted on behalf of UPS by J. Gregory Sidak in March 2017.

In his comments, Sidak writes: “Sanford C. Bernstein & Co. has estimated that the Postal Service delivers 40 percent of Amazon’s packages.”

A footnote takes us to a 2015 article by Devin Leonard in Bloomberg, “It’s Amazon’s World. The USPS Just Delivers in It.”  The article is available only to subscribers, but most likely the Bernstein analyst (probably David Vernon) was referring to domestic volumes.  In a 2014 Wall Street Journal article, a Bernstein analyst said that Amazon accounted for 35 percent of U.S. volumes in 2013.  By 2015, that number probably grew to 40 percent, and by 2017, 45 percent — the number we used.  There’s no evidence here that the Postal Service delivered 60 percent of Amazon’s domestic shipments.

One possible explanation is that DB is including parcels delivered by the Postal Service that came through FedEx’s SmartPost or UPS’s SurePost.  There are a lot of them — DB says 30 percent of the packages Amazon ships via third party providers also use USPS for the last mile.  But this is apparently not the explanation because DB explicitly says, “In this analysis, we have not made any assumptions regarding Amazon’s unit volumes that flow through the USPS indirectly” via SmartPost or SurePost.

In any case, it just doesn’t seem likely that the Postal Service is delivering 60 percent of Amazon’s domestic shipments.  If that is true, though, it’s a significant revelation.

(Update, August 8, 2018: According to a report in the Seattle Times, shipping consultant Satish Jindel, president of ShipMatrix, says the Postal Service carries about 62 percent of Amazon’s parcels.  Perhaps that is the source of DB’s figure, rather than the previous explanation.  It should be noted that Jindel does not specify that his estimate excludes Amazon parcels delivered by USPS via SurePost and SmartPost, as the DB analysis does.)

Average Shipping Cost per package for USPS (row 18): DB estimates that shipping through the Postal Service costs Amazon an average of $3.96 for total shipping costs, with $2.15 going to the USPS and the remainder, about $1.80, going for sortation, transportation, etc.

For our estimates here (cell D18), we’re figuring even higher pre-outbound costs for the Postal Service — about $3.30 per package, for a total average cost per piece of $5.30.  That’s more than the $2 we used for pre-outbound for the other carriers to reflect the fact that Amazon must do more to prepare and deliver its packages to the Postal Service than it does with the private carriers.

Whether it’s $3.96 or $5.30 or somewhere in between, the “true cost” for shipping with Postal Service is thus much more than the $2 per package paid to the Postal Service.  Amazon incurs a lot of costs optimizing its packages for what it calls “postal injection” — all sorted, labelled, and delivered to local post offices — which is why it’s getting the worksharing discount in the first place.

Cost per Package to USPS (row 21): DB uses $2.15, which is the average for USPS Ground in the Cost & Revenue Analysis; our analysis used $2, the average for Parcel Select (the main but not only type of Ground) from the Revenue, Piece, Weight Report.

Amazon’s total postal cost (row 23): DB’s total for what Amazon pays the Postal Service would be about $3.6 billion in FY 2017, compared to about $2.2 billion in our analysis.  Again, the difference is due to DB giving a much higher market share to the USPS (60 percent instead of 45 percent).

Cost of sortation network (row 28): DB’s total for all pre-outbound costs in 2017A is $2.77 billion, which would be about $3 billion for FY 2017; our estimate is $3.6 billion because we’re including pre-outbound costs for the private carriers, which DB did not do.


Here’s a table showing the key totals and averages.

Deutsche Bank didn’t draw out its analysis to figure the numbers on international shipping and totals worldwide, but one can do so using the numbers in its analysis.  With total shipping costs of $21.7 billion in FY 2017, and domestic accounting for $13.9 billion, the remaining $7.8 billion would be international.  At $6.55 per piece for an average total cost per piece, that would be about 1.2 billion pieces for international for FY2017.

That brings DB’s estimate for total volumes worldwide to almost 4 billion for FY 2017.  Our estimate was about 3.3 billion for volumes worldwide — about 2.4 billion for U.S. and 850 million for international.  (We noted in our original analysis that we might be underestimating international volumes.)

DB doesn’t provide a number for what portion of costs went to outbound versus sortation-fulfillment, but again, we can extrapolate.  Worldwide, for DB, total outbound costs for 2017A were about $17.2 billion, while costs for sortation, etc., were about $2.8 billion (all associated with pre-outbound costs for the USPS).  Total outbound costs represent about 86 percent of total shipping costs.

In our analysis, outbound account for 55 percent of total shipping costs, with 45 percent going to pre-outbound fulfillment costs.  (The Citigroup analysis figured 75 percent for outbound costs, 25 percent for pre-outbound.)

As explained above, it seems unlikely that such a large portion of total shipping costs would go to outbound and such a small amount for sortation and transportation, but who knows?

The bottom line, anyway, is that it looks like Amazon probably sent out somewhere between 3.3 and 4 billion packages in 2017, for an average all-in cost per package of $5.50 to $6.50.

With numbers like these in play, if Trump’s Task Force can engineer a price hike of any significance, the effects could be huge.

(Photo: Postal trucks with Amazon deliveries, from GAO report on Parcel Select)