Amazon fined by the Italian Antitrust Authority. Is this the end of digital ecosystems?

Self-preferencing practices in platforms and digital ecosystems are in the spotlight as the Italian Antitrust Authority fines Amazon Italia €1.13 billion for abuse of dominant position. The decision poses serious questions about what is considered a legitimate design of platform ecosystems and whether firms orchestrating a platform marketplace can actually build an ecosystem (structuring co-specialized relationships). Is this the end of ecosystems?

Facts first. The decision of the Authority can be found here; it's 250+ pages, in Italian. So, for non-Italian speakers, here is a summary of some of the key points. The AGCM, the Italian Authority, accuses Amazon Italia of leveraging its dominant position in the e-commerce marketplace market to induce sellers to use its own logistics services (fulfillment by Amazon - FBA) at the detriment of other competitors in the logistics market for e-commerce.?The AGCM is thus defining two "relevant markets" as separate here, first the market for logistic services in the fast, B2C delivery (yeah, that narrow a logistic services market!) and the market of e-commerce marketplace in Italy.

Sellers using FBA can obtain the "Amazon Prime" badge and with it expect a number of benefits, including participation in Amazon special sales events (eg. Black Friday), “free shipping” to Amazon Prime clients, higher chances to adjudicate the Buy Box (i.e., consumers selecting and buying the seller's product).?Because of it, sellers using FBA obtain greater visibility and higher sales. Also, they are not subject to performance evaluation metrics other (non-Prime) sellers are subject to satisfy Amazon service standards.?

Amazon Italia is accused to use this FBA-Prime tying strategy to exclude non-FBA sellers from Prime benefits, and undermine the ability of third-party logistic providers to compete on the merit with Amazon’s logistic services, and induce sellers to single-homing to the Amazon marketplace - i.e., choose only to transact on the Amazon marketplace. This practice is judged as anti-competitive and detrimental to competition in the e-commerce logistics market, the e-commerce marketplace market, and ultimately producing negative effects for final customers. Is it? ???

So, let's look at some numbers. According to the evidence produced by the AGCM, in 2016?Amazon Italia was head-to-head with eBay Italia. But in 2019 Amazon gained about [70 -75%] market share in terms of the gross merchandise value (GMV) of sellers, while eBay market share dropped to [10_15%], about [20-25%] of Amazon GMV.

Table 17 - Sales value and market share

The number of retailers on Amazon Italia's marketplace has grown over the years while the number of retailers on the major competitor eBay Italia has remained stable. ?

Table 18 - active sellers

Amazon Italia has also seen an increase in monthly visits. More sellers, more products, more buyers, more sellers….the textbook "indirect network effects" in action!?

Figure 22 - user # visits

And now the interesting point: The power of ecosystems! In research commissioned by eBay, they note: “Perceptions of eBay trail those of Amazon, especially in … trust, shipping, and returns”. This shows in the numbers ???

Total online shoppers - eBay internal research

They add: “Preference for eBay is softening, and eBay fails to convert awareness into visitation or preference effectively”… “eBay preference is dropping among its most loyal buyers”. That is not the case for Amazon Italia - very high and consistent conversion rate???

Behavior conversation - eBay internal research

So, is this competitive effect due to market tipping (indirect network effects) or superior customer experience due to Amazon ecosystem around Prime service? Amazon?thinks is the latter. According to Amazon CEO Jeff Bezos in a 2015 letter to shareholders “FBA is the glue that inextricably links Marketplace and Prime. Thanks to FBA, Marketplace and Prime are no longer two things... Their economics and customer experience are now happily and deeply intertwined”.

Why is this about ecosystems? Amazon has turned (a set of) arms-length relationships in the marketplace into co-specialized relationships, creating non-generic complementarities that allowed Amazon to shape a particular customer experience - Amazon’s shopping way!?We discuss specifically the Amazon Prime strategy in our 2018 Strategic Management Journal paper as an example of the firm’s attempt to build an ecosystem of cospecialized sellers on top of the marketplace as a way to shape value creation dynamics in specific ways. ?

SMJ article specific passage on Amazon Prime

Are thus ecosystems type of structures largely anticompetitive? The AGCM?concedes that Amazon Italia has a business model built around a “complete ecosystem” integrating multiple services, whereby Amazon plays multiple roles. This is not a problem per se. However, the AGCM thinks Amazon Italia has “artificially tied two separate services”, Prime Seller benefits and FBA service, to “create an illegitimate incentive” to buy FBA services. This strategy “ falsifies” competition between its logistic service and competitors’ ones. The key issue, thus, is that Prime sellers (ie, those buying FBA service) have benefits over other sellers (those not using FBA). The authority is demanding equal access for ALL (!) sellers to Prime benefits independent of their FBA use on the basis of uniform standards. ?

But, we see these “exclusive” types of relationships with preferential treatment in all kinds of ecosystems, from apps to games to ERP systems. That’s how firms “orchestrate” the ecosystem, promoting some value creation paths, not others. And that's the ley difference between structuring digital ecosystems around specific value propositions vs. running pure platform marketplaces. ?

We are back to the original puzzle: if ecosystems are designed around non fungible investments, creating “meta-organizations” of an aligned set of firms towards specific value propositions for the customer, are ecosystems de facto anticompetitive in excluding not aligned firms on the ground of a generic "self-preferencing" abusive conduct??

My own considerations on the specific (and more general) matter.

First consideration: self-preferencing, in and by itself cannot represent harmful conduct. Coopetition with complementors is a key part of the platform ecosystem model that helps create value and make it work!?Conceptual work by e.g. Andrei Hagiu shows how the dual model of operating the marketplace while also competing with complementors is an efficient if not desirable feature. Check Andrei Hagiu’s work here:?

Should Platforms Be Allowed to Sell on Their Own Marketplaces?

Some empirical work on innovation platform ecosystems shows how platforms use their own products not just to increase efficiency but to steer the ecosystem towards desired tech development trajectories and thus stimulate innovation.?See e.g. work by Thomas Kude et al. in the Android-apps context:

Does Platform Owner's Entry Crowd Out Innovation? Evidence from Google Photos

And my own work (sorry, is this self-referencing?) in the video games context:?

Building the Value of Next-Generation Platforms: The Paradox of Diminishing Returns - Carmelo Cennamo, 2018

And there is also work specific on Amazon’s own prime products by Feng Zhu (HBS) showing how these products, while capturing increasing value, help expand demand for the product category, precisely because of the benefits associated with shipping:?

Competing with complementors: An empirical look at Amazon.com

Second consideration: we thus need a specific and better theory of harm when judging these “self-preferencing” practices. What the decision is after in this specific case might have to do more with bundling services rather than with preferential treatment. In these regards, one would have liked to see evidence that this bundle was not needed to properly run the Amazon marketplace…that is, Amazon can perfectly deliver value by keeping them separate (as the AGCM implicitly holds by treating the two as separate relevant markets). What if instead, it is thanks to it that Amazon can increase value to users??

Third consideration: user benefits! Is new antitrust getting away from any considerations of consumer welfare? Shall we just focus on “fairness” without any considerations on why the practice is put in place and who stands to benefit from it??But how do we then establish an “objective” principle (or at least parameter) of fairness if we just consider the split of the pie without accounting for how big the pie is being made??

Fourth consideration: while arguments and evidence of the abusive conduct by?Amazon Italia might be not grounded on a solid theory of harm and open to several doubts, the remedy being proposed by the AGCM seems in fact very sensible to me.?It does not constrain Amazon’s ability to define standards and operations to run the service the Amazon way or ban them from running their own logistic service. But it puts constraints on Amazon’s *possibly* taking arbitrary excluding choices against third-party logistic services and non-Prime sellers. This, to me, allows Amazon to still keep control over the overall ecosystem, keeping logistic firms and sellers up to its own standards. It allows discretion in the design of the rules but constrains discretion in the application of the conditions, avoiding possible arbitrary prevarication (a limit and a "price" that a successful dominant firm can bear after all). Isn’t this a win-win??