Store Wars: Imperfect views of imperfect markets

Economics has something of an edge when it comes to respectability among the social sciences, but it is often squandered by talking heads acting like everyone else is an idiot on TV. The benefit of taking an introductory economics class is not prestige or a set of talking points but rather a set of tools for looking at the world that will hopefully equip you to understand a problem better. There’s been a lot of talk about monopoly with regards to tech, and there will be a lot more now that Fortnite has been kicked off the App Store. This article is going to describe a few ways (none of them perfect) to apply an economic lens to the dispute between Epic and other digital storefronts, and the implications they carry.

Monopoly or monopsony?

In addition to sharing a name with a popular board game, monopoly is a very old term, coined by Aristotle (this does not contradict the article below which dates the first English usage). Its definition is broadly understood: a single seller controlling a market for a good or service, and the arguments for why monopolies are socially undesirable are almost as well known. Epic’s filing describes Apple as a monopoly. However, depending on how you view digital storefronts, you may prefer to use a lesser known (your spell checker might even try to correct you for using it) term for a single purchaser of goods and services from many sellers: monopsony. It was first employed by a Cambridge economist in the 1930s (though according to the Journal of Economic Perspectives, the term was coined in collaboration with a classicist). For the purposes of the models below, the difference between monopoly and monopsony won’t change the results, but since we are trying to gain some intuitions about the problem (or identify if there even is a problem), assigning a label forces us to be clear about what we think a digital storefront actually does. If you think that services like the App Store, Google Play, and Steam are selling access to an audience and server space for downloads to developers, then they are a monopoly (though careful readers may question the ‘mono’ of this monopoly. This is the subject of the next paragraph). If, instead, you think that the App Store is more like a conventional store like Walmart, buying goods from different suppliers and selling it to customers, albeit with a very efficient inventory management, then they are a monopsony.

Whether you consider digital storefronts to be buyers (monopsonies) or sellers (monopolies), the use of the plural immediately calls into question the choice of terms. Surely the existence of Google Play implies some kind of duopoly (or, more generally oligopoly). Those fearing more etymology can rest easy because there is a straightforward concept that applies to both that we can employ: monopolistic competition. Monopolistic competition describes a market in which more than one producer sells similar products that are not perfect substitutes. Chances are you did not think about what farm you bought your carrots from. If Warren Buffett and Old Macdonald both had farms, you would likely just look at the price and pick up whichever one cost less, meaning Warren Buffett’s and Old Macdonald’s carrots are perfect substitutes. Contrast this to the market for running shoes. Most people are not indifferent to what brand of running shoe they buy, and will tolerate price increases for a favourite brand up to a certain point before they consider switching. Monopsonistic competition is similar except that there are now a few buyers purchasing from many sellers of differentiated products. In both cases the description seems more intuitive since it aligns with the simple observation that publishers can and do choose which platforms to support and storefronts to publish on (including switching) .

An unfortunate reality in writing an article like this is that the inclusion of any mathematics will immediately turn off a number of readers. This is a shame because mathematical arguments should provide clarity, not turn people away. However, this is not the article to divest readers of this belief and so I have opted to choose models that are available in most economics texts (and Wikipedia) and will forego formal arguments in favour of describing the assumptions and the results.

Differentiated products

A model of monopolistic competition rests on the assumption of a relatively large number of firms selling products that are similar but not identical. Firms are assumed to be profit maximizing and that there are relatively few barriers to entry and exit. The monopsonistic counterpart has the same assumptions but for buyers. A result of this model is that firms are capable of making short term economic profits, but that there are no economic profits in the long run. Before moving forward I would like to distinguish economic profits from a more traditional understanding of the term. Economic profits can be thought of as excess profits. That is, economic profits are profits above and beyond the opportunity cost of whatever the individual is doing. The conclusion that there should be no profits does not say that Apple, Google, Valve, or any other owner of a digital storefront should have financial statements with a 0 in the income box, only that these profits should not be in excess of a comparable expenditure of resources. Why is it that there are no long run profits? This is a function of the ability for firms to enter and exit.

Beyond the confusions that arise from introducing the idea of economic profit, there are some problems with this model. Our conclusion that there are no long run profits rests on low barriers to entry and exit, and the whole idea of there being a large number of storefronts seems counterfactual. Are these reasonable assumptions? On one hand, Blackberry (then Research in Motion) used to dominate the smartphone category and even after the introduction of the iPhone, BlackBerry App World had the highest revenue per app in 2010 (though this is due to a small denominator. Apple had over 80% of the market at that time). Android was able to enter despite the dominance of iOS and has been competing at the high end of the smartphone market for years. Google Play is not the only place to get apps on Android with Amazon Appstore already operational and a promise from Epic that the Epic Games Store will be coming to Android too. Depending on your definition of app, we can also add the fact that Steam is a dominant player but Origin, Uplay, Humble, GOG, Kongregate, Discord, Twitch, Paradox, Amazon, and, of course, Epic (among others), have entered this market with varying degrees of success (remember, the assumption is low barriers to entry and exit).

If we think monopolistic/monopsonistic competition is a good model for app stores then it is hard to argue that what Epic is trying to accomplish is going to produce some kind of net benefit for everyone. Entry and exit into the market is already possible and the best that a firm can hope for is short term profit. Making iOS more like Android is certainly beneficial to Epic, but nothing in the model says that the short term profits will go away and the Android experience hasn’t stopped Google from charging the exact same fee that Apple does.

Suppose we don’t think this is a good model for app stores. The biggest objection stems from the number of firms and their ability to enter, and this ultimately comes down to how we define the market for apps. Epic’s filing has chosen the narrowest possible definition by saying Apple has a monopoly on apps on iOS. This article has entertained the possibility of a broad definition that includes PC (and this is not just a rhetorical touch. Fortnite is available on PC, console, and mobile, even if it does not participate on all storefronts). Where we fall on this issue is likely going to determine what model we think is most appropriate for analysis. If I am skeptical about Epic’s claim, I might point out that with some capital I can become a monopolist with the world’s only Hungarian-Mexican fusion restaurant with live polka music and belly dancing, but that an injunction is the least of my worries. Arguments against the broader definition might point out a correlation between Apple’s tendency to charge fees in sectors it has a product in, and that its market share in the US is much higher than the total market share that is often reported (even higher when considering dollars spent on apps).

Few firms

In order to rid ourselves of the undesirable assumptions above, we can consider a model of oligopoly or oligopsony (again, the results don’t change). There are multiple models, but in each case wind up trading one set of questionable assumptions for another. Broadly, oligopoly concerns itself with a fixed number of firms producing (buying in the case of oligopsony) an undifferentiated product. Because they have market power, firms’ decisions will affect the price of the good.

Cournot’s model was first and it had firms simultaneously competing on quantity. Working through the model shows that the industry will produce a larger quantity than a monopoly case and with a lower price, but that it will be a higher price and lower quantity than the case of perfect competition. To put it in plainer terms, the firms are able to claim long run economic profits, but less than in the case of a monopoly. Cournot equilibrium produces no economic profits as the number of firms gets large (an assumption we are trying to avoid).

In reviewing Cournot’s book, Bertrand argued that the model would produce a different result if firms competed on price. Here the assumption about an undifferentiated product is quite strong since it means that consumers will only care about price, which is what the firms are now competing on. The result from this model is that even with as few as two firms the price will be equal to the competitive equilibrium and there will be no profits. This is because for any price higher than this level it is profitable for the competing firm to go fractionally lower and capture the entire market. Adding additional firms does not change this result.

We might find the model of Cournot competition appealing because it produces a result that seems to be in line with our intuitions about the market for apps, though conforming to our priors should not be the only criteria by which we judge a model. The Cournot model does have some unappealing features. Not everyone will agree on the importance of product differentiation in this market, but its absence makes it an unappealing choice for anyone who considers it an important feature. Even if you think undifferentiated products is a fair price to pay, it is very strange to consider app stores competing on quantity (that is, Apple and Google sit at head office each year and say “this is the number of apps we’re going to sell to keep at a profit maximizing price”). Epic’s goal with the lawsuit appears to be geared towards increasing the number of firms, but can we really expect a large enough number of firms to enter to compete away the profits? If not, the lawsuit gets Epic their piece of the pie leaves us with the undesirable outcome of long term profits (albeit with diminished effect). This might recommend some other intervention other than what Epic is proposing.

Bertrand’s model is more appealing for its focus on price but it throws into relief just how much of a concession the shift to an undifferentiated product is. This may be a perfectly reasonable model for the market for carrots we mentioned above, but competition on price seems to be the only feature that aligns with our experience of using app stores. One unstated assumption of the two models is the impossibility of collusion. In the Bertrand case this seems to be a particularly strong assumption because the gains from sharing monopoly profits through collusion are so large compared to the equilibrium. If you find the Bertrand model the most credible representation of this market though then we know there isn’t a problem and the Epic lawsuit is just two rich companies fighting over how to cut the pie.

The two previous models of oligopoly assumed everyone made their decision on price or quantity at once, but there are also models for market leaders. In Stackleberg competition a market leader chooses a quantity first, this decision is observed by the other firms and they then set their own quantities. The Stackleberg leader is at an advantage since their worst case is just to fall back on a Cournot equilibrium, but they can consider the best responses of the followers and set a quantity that is more advantageous for them. There is an analogous model for price in which the leader uses their advantage to consider the responses of the followers and set a profit maximizing price that they know the other firms will comply with (again, showing the teeth of the assumption of undifferentiated products). The prices and quantities of these two equlibria are generally not the same.

The Stackleberg model is appealing because it’s not hard to see a company like Apple being the first mover in an oligopoly/oligopsony. Yes, we still have the weirdness of working with competition on quantity, but the first mover feature feels like it’s something that really fits this particular discussion and aligns with our thoughts that maybe they do have a particular advantage. What is the outcome? While the leader does quite well for themselves, the overall outcome is a quantity that is higher and a price that is lower than the Cournot equilibrium. This is an interesting result because it is a model that doesn’t align with our traditional ideas of fairness (the leader does better than the followers), but produces a consumer surplus that is higher than in the case where all the firms are on a level playing field. This raises an important question we haven’t even considered yet: how concerned are we about the interests of an individual company relative to overall social welfare?

What do we get out of all of this?

We have reached the limits of where only words can take us with these models. None of these models give us the full picture. There is an entire field dedicated to studying these kinds of problems called Industrial Organization and its work cannot be reduced to five models you can find in a good undergraduate microeconomics textbook. What conclusion are we supposed to draw? How are we supposed to think about this?

The point of this article is not to argue for a particular point of view on the lawsuit or on the store wars in general. Instead, the point is to share some competing abstractions of the problem to promote more careful thinking about the issues. George Box is credited with the aphorism “All models are wrong, but some are useful” (though, as is common with such sayings, the history of the quote could fill a small essay). These models should stimulate your own thoughts, not become dogmas. For the mathematically inclined, these models are not very difficult to work through (some textbooks reduce them to algebra, though the usual treatment involves calculus for optimization), which means assumptions can be adjusted to gain deeper insights. Sometimes the best way to see learn is to turn the knob all the way to 11 and see what happens.

For those who don’t want to go deeper into these models, the previous discussion should still provoke some reflection. The opening pages of Epic’s filing really seem to be the sort of thing written for the public rather than the court (does the legal argument change at all if you remove the first paragraph? Do you see this kind of writing on page 39 or does the #FreeFortnite crowd not read that far?). There are already plenty of people willing to jump on “Team Steam”, “Team Epic”, “Team Apple”, “Team Amazon”. But as the Stackleberg model showed, companies getting screwed does not necessarily mean consumers are getting screwed. As a society we are willing to endure certain costs so long as we achieve an even greater benefit, which is why we talk a lot about why monopolies are bad and yet have an institution dedicated to creating them in the form of patents.

Even the Monopoly or Monopsony question provokes some interesting thoughts. I’ll admit, I’m a very happy Apple user. I specifically like Apple because of the things that developers hate about Apple. Apple has created a product that I’m not only happy to use, but reduces the app purchase decision to a simple “do I think this thing is worth the asking price?” with all my payment information accessible with a fingerprint. I am so happy with this state of affairs that I genuinely don’t know how Android users get anything done without constantly checking the access and privacy settings of their devices. This makes me think that Apple is ultimately selling me and people like me to developers, because we’re willing to spend more and happy to get on with our day knowing that someone else has taken care of the hard stuff. If Apple reliably provides access to people willing to pay a premium for apps, then they’re worth every penny they’re asking for and the question is what the other stores are doing to justify their 30%. The fact that the status quo makes all the hard stuff someone else’s problem naturally makes me resistant to moves that make Apple’s enforcement of those rules weaker.

All of this makes me sound like I’m on Team Apple doesn’t it? That’s where the models come in. I know that while I might like the status quo, it’s entirely possible that the cost of the current policies are more and better apps that never made it to the App Store. More importantly, my personal preference for Apple’s policies do not necessarily generalize to all users (and they certainly don’t give developers any standing). This is why the abstracted view is useful. We look at the whole market and ask “what are we giving up and what are we getting in return?” Hopefully the models have given you some new dimensions to consider and, with luck, some examples about how work through our assumptions about what’s going on.


Note on affiliate links: I am a member of Epic’s Support-a-Creator Program for which I am given a portion of sales for traffic I drive to the site or which contain the creator tag SYSTEMCHALK. I am also a member of the Humble Affiliate Program. The inclusion of a given title is for illustrative purposes first, but when it is available on either link I will provide such a link. As always, I encourage you to do your own price comparison or buy on your preferred platform.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s