Standards, Platforms, and Public Policy

Shane GREENSTEIN

Today, I want to talk about the frontier of discussion in standardization and address the question of what the literature on standards suggests about government policy.

As an early example of standards, the Chicago Federal Reserve Bank helped to push through standards for check clearing, resulting in improved efficiency for processing checks, and creating large gains for the U.S. economy as a whole.

There has been a shift in the way of thinking about standards, however, from focusing on compatibility toward competition between platforms which embed standards. I want to look at examples of how standards play a role as a catalyst for the growth of mass markets and how governance of standards influences—and is influenced by—competition.

Taking the Wi-Fi standard as an example, there were many experiments with wireless LAN designs during the 1990s which generated little sales. However the Wi-Fi interoperability standard was released by the IEEE 802 committee in 1997 and updated in 1999 as 802.11a and 802.11b. The standard helped firms commit to a design that was interoperable, and the committee achieved its intention of putting the standard on an unlicensed spectrum to ensure that no firm would own it. Apple commissioned the first product from Lucent, which was followed by Dell, and a competitive market subsequently took off. Firms formed an alliance for conformance testing and branding, and the "Wi-Fi" name was chosen. Hotspots became popular, with many different business models including free access and subscription services. Intel took the step of adding Wi-Fi into motherboard designs with the launch of the Centrino reference model, which fostered its ubiquitous use. The IEEE 802 committee thereafter continued to upgrade Wi-Fi speeds, with redesigns becoming increasingly contentious because of the commercial value of upgrades. The interesting observations here are that experimentation continued for many years without a profitable market emerging; once the standard was launched, the growth into a mass market came from an unexpected use case of hotspots; and profits came from selling both equipment and access.

A second example is Google's Android operating system for mobile devices. Again, there were many experiments in smartphone operating systems in the early part of this century, and the U.S. was not in a position of leadership. This changed with the emergence of the iPhone in 2007, which was a leapfrog design, bringing the iPod installed base to Apple's phone product with a touch interface and online store. An ecosystem developed around the iPhone product as developers found the access to a large user base attractive, but there was resentment about Apple's strict rules and control that dictated standards. Google initially set out to deter a proprietary standard rather than to make huge revenue. It made the Android OS available with all information about its APIs accessible to the public, giving the ability to developers and handset makers to make changes if they wished. The Android operating system has seen growth due to fewer restrictions and lower prices. The competitive landscape has therefore changed, with Apple still very profitable, Android very popular but less profitable, and most of the other players not doing well. The observations here are very similar to those for Wi-Fi, with much experimentation continuing without a profitable market emerging and a breakthrough with an unexpected use case at a low cost price point, but also differing in that Android was initially a strategic competitive response although it is also profitable now.

The lessons are that experimentation can continue for quite some time, breakthroughs are often unexpected, and sponsors attempt to profit, but are often quite late.

Many firms and consultancies are asking how we think about this in broader terms. One key concept is that you must be clear about what roles the standards play. The first role is facilitation of inter-networking, which is now very old and well-known and allows data in one location to be shared with many others. A second role is facilitating a network effect, in which the value of the standard rises with participation and creates a virtuous cycle. A third role is platforms which reduce transaction costs, in which a standard facilitates communities to come together at low cost and do business with one another. The situation of people talking about multisided platforms has its own language. What is interesting is that people in the industry who think a lot about platforms will not use the language of standards, but it is actually all about standards. It is so well understood that it is about standards that nobody uses the term anymore—it is taken for granted. A platform is defined as a reconfigurable base of compatible components on which participants build applications. A standard is inherent in this. It is multisided in the sense that there are different distinct groups of participants with very contrasting interests, and the firms that make platforms have standards that vary among groups, so it is always a collection of standards. A key point is that platforms serve an economic function as an intermediary between the different groups. Another key point is platform leadership, which is affiliated with designing hardware or software that mediates economic activity, and entails designing and maintaining standards, setting out roadmaps and targets, keeping users informed, and providing assistance so that business partners can use the standards to serve their interests. There is a timing mismatch of costs to revenues, in which platform leadership inherently assumes significant costs early and revenues later, with much exploration early on and profitability later. This means that there is very high risk, with only some firms eventually profiting.

There are also not-for-profit platforms and hybrid platforms, and Wi-Fi is one example of the former. The difference between for-profit and not-for-profit platforms is mostly in the restrictions and limitations on participation and the release of information. Open platforms usually mean no limits on information and on use. There will occasionally be zero pricing but not always. This also means that competition between open and proprietary platforms tends to be based on multiple margins and is typically quite complicated.

Platforms should thus be thought of as an organizational form for deploying standards that encourage participation by multiple communities with intermediation between them and adoption of technologies that embed the standards. This raises the important policy question of how to think about intervention in this kind of setting. There are many questions there. However, it is not my job to tell you what to do. I want to focus the remainder of the time on the most challenging part of this, which is the evolution of platforms. In most cases in which policy matters, the standards are not static, which is the hardest problem.

How should we think about how platforms shape competition, and particularly the evolution of competition? One important concept is that platforms serve as a focal point for coordination, and all of the participants gain from partaking relative to a situation in which there is no platform, and yet often do not profit early. An early example is retail barcodes. The remarkable thing is that for the first 10 years of its existence, it was not profitable. There was a great deal of experimentation, diffusion, and adoption of the scanners, and the barcodes were put on packaging. But it was not profitable, and the coalition was in fact very close to collapsing. After 10 years, however, when the cost of the equipment came down and learning curves brought the cost far down, barcodes became essential for competition for retailers. It then became a competitive advantage for packagers to be a part of the barcode community. Although early on it was not very profitable, platforms have become focal for learning and experimenting, which is a benefit to society. This is also a central problem from a policy standpoint that, quite often, for the purpose of learning, fewer is better in order to focus the learning in one direction. But for the purposes of competition and variety, more is better. There is no obvious answer in any given situation to the tradeoff between variety and competition on the one hand and learning on the other.

Another key issue to think about is the emergence of monopolies. Most platforms tend to lend themselves toward a small number of platforms, and the main reason again is a tradeoff between the tendency to continue to invest in an established platform—where if a platform can add functionality without limit, then it will tend to become better than everything else, which will allow room for only a small number of platforms—while on the other hand, differentiation of platforms to approach different communities is a strong market force working against monopolization. Again, it is not clear what the right answer is in any given situation. You will generally get competition on the margin for differentiated users or communities, and in a multi-sided platform, it is quite hard to think about this because you may get differentiation on one side of the market and not on the other.

Another helpful point to think about but is also quite difficult to do when watching platform competition is the pathways toward development. An observation that is now emerging is that there is more than one way to grow a platform. This is a challenge for competition policy because it is much easier if there is one right way. There are a couple of reasons why you tend to get a variety of paths toward growth of dominant platforms. One is multihoming, which is the tendency for users or firms to use multiple platforms. It sustains more platform competition and tends to be seen early in platform development. Another is that converters tend to emerge to overcome incompatibility across platforms. In some situations, that can have a very dramatic effect on the outcome, such as what occurred in the 56K modem market.

A very common question is how to think about experimentation from a policy standpoint, and as a policymaker, how to know if experiments are going well or badly. One insight is to look for symptoms of health, such as finding lots of experiments, many entrepreneurs, vigorous competition between standards, and few unilateral bargaining. Another healthy sign early on is a virtuous cycle, in which developers react positively to platforms and users react positively to developers. If you do not see virtuous cycles, it typically means things are not going well.

In practice, when looking at platform competition, a complete analysis requires thinking about changes over time rather than a static analysis. Multiple pathways for platform evolution is challenging, and although firms usually want their own work to go well, policy will generally want at least one platform and potentially more. It is in society's interest to have competition and at least some platforms so that there are coordination benefits. There are also tradeoffs which are quite challenging to think through.

To conclude, I would like to consider again a series of questions. First, private orderings can and do resolve issues and are beneficial in the right circumstances but not always. Large firms can sometimes do this well, although it is quite hard because of bargaining and negotiations. Another thing that can be learned is that standards are more than just designs, especially when embedded in platforms, and there is typically the need for ongoing organizational monitoring, conformance testing, support, and planning and testing of investment by private firms, even with breakthroughs.

Another question is how to think about intervention. One of the big issues for intervention is disclosure rules because that is the biggest difference between proprietary and open. Firms play very close attention to disclosure rules, which shape whether firms participate in standards committees and are sometimes the reason why firms are willing to participate. This also played a very important role in the Apple-Android competition. There are many interesting questions there, and the key issue for policy is to think about who is favored by disclosure rules.

This literature about platforms takes us toward some general views about when it makes sense to intervene. One case when a government would think about intervening is bargaining failure, in which firms get together but there is no platform, and there could be endorsements to bring a platform as it is in the public interest. Another case in which you could have platform intervention is if the government has statutory authority over some key input, such as frequency spectrum for radio communications devices, for example, and the statutory authority necessitates government intervention. A third case, which is interesting for those of us doing antitrust, is government intervention that can facilitate entry or experimentation, which benefits society. The cases in which the government generally should not intervene are when private orderings are working well, with virtuous cycles and significant competition, or when there are other signs of vigor, such as symptoms of good experimentation. Also, when the use case remains uncertain but substantial private incentives to experiment with new use cases exist. It is worth noting that government mandates for designs will rarely work very well, and literature heavily leans toward private experimentation, except in situations in which there is leadership failure. One example of a government-mandated design that worked very well was the FCC's intervention to make designs for the interconnection of telephone equipment.

NAGAOKA Sadao

Economic literature suggests three important sources of market failures. One is the critical mass issue. In the Wi-Fi case, your talk suggests that early experiments were very important but not well rewarded, and the IEEE establishing the standard helped to create a critical mass. There can be a tradeoff, in that if you postpone standardization, there will be more experimentation. Considering that early experimenters are not really rewarded, there is an issue of how to encourage early experimentation while also not comprising the chance to create a critical mass. Thus, policy for overcoming the critical mass issue is one question.

The second source of a market failure is the possibility of choosing the wrong standard. There can be multiple equilibriums, and the market may have chosen the wrong standard. Your view seems to be that the government cannot dictate the choice of the standard. On the other hand, there can be a risk that the incumbent monopoly platform may prevent the emergence of a competing platform, so there is a role for antitrust. Other than antitrust, can there be any role for the government with regard to preventing the choosing of the wrong standard?

The third is fragmentation issues. As you know, in Japan, there is no national electricity grid because the western and eastern parts of Japan use electric grids with different frequencies. It is very difficult to integrate the market now because of the huge sunk costs involved in both parts of Japan. Even with a standard, if there is fragmentation, it can be very costly, and there is the question of how this can be avoided.

Shane GREENSTEIN

Thinking through critical mass is very important. When I talk about wanting at least one platform, rather than none, that is a part of the policies for critical mass. That interacts with questions about the timing for intervention to generate critical mass. It is hard to make a general statement other than early on it is very hard to make a good answer. A government policy is often like a blind giant in that it has a huge effect on the people around it when it moves, but it is blind in the sense that it does not know better than anyone else. Regarding the Wi-Fi case, there were three initiatives in wireless local-area networking, and two of them succeeded. Bluetooth is a private consortium that chose to differentiate by focusing on short-range wireless networking. There was a third community called HomeRF, which had taken an approach very similar to the 802.11 community, that failed. Its problem was that it did not have a design early, and when private firms decided to make a product, they went to 802.11 rather than HomeRF because it was ready. As such, timing can be a very interesting question.

Your other two questions on wrong standards and fragmentation are good ones, but I am not sure that I have a good answer to those.

Q: When you talk about policies, you spoke of intervention or mandatory standards, but there are other kinds of policies. One is government procurement. The U.S. has a huge defense budget, and the equipment it chooses might have a great impact on a platform. Another area is consumer policy. You said that the government is like a blind giant, but the technology is now such that it is difficult for normal consumers to make the right choice, which makes it difficult for market choice to arrive at the right standard, and the government might know better than an ordinary consumer. In some products, there may be safety problems or concerns over information leakage.

Shane GREENSTEIN
Governments can play a very large role in procurement. The history of computing has a number of examples in which large procurement orders from government buyers can change standards. The most famous is TCP/IP. The Department of Defense established TCP/IP as a standard for all Unix-based systems in 1982, and as a consequence, every Unix system provider in the world consequently made their operating system TCP/IP compatible, which in turn facilitated the growth of the Internet. The Department of Defense also has had standards that have not taken off because they have no commercial value. So if government procurement needs overlap with commercial needs, it can move markets.

Government policy through consumer protection can play a very important role, particularly in young markets because consumers are generally less informed than suppliers about issues. I am particularly sympathetic to this issue in the privacy rules, in which the firms understand much better than users about how the information will be used.

Q: Standards issues for the IT or telecommunications industry have similarities to other industries, like reducing costs or increasing demand, but are there additional considerations for platforms or standards for the IT or telecommunications industry? Also, when thinking about the world landscape, there are European countries that like to make dual standards for many industries, and China also likes to set up its own standards, so there may be room for government to intervene to make a standard to be internationalized. I would like to hear your view on this.

Shane GREENSTEIN
Your intuition that some of the arguments for increasing returns to scale and demand in communications are very similar is right. Looking at text messaging, for example, you could have increasing returns to scale from a network effect if everyone is using the same protocol, and in principle, government intervention could facilitate that. Similarly, this applies to intervention to generate interconnection in communications to facilitate increasing returns. What is interesting is that when you go across international borders, from a supplier standpoint, the principal problem is high transaction costs, with the question of whether hardware needs to be redesigned with different safety standards, different communication protocols, different licensing, and so on. Suppliers talk all the time about how different the legal regimes are in every country in which they do business, which reduces their ability to realize economies of scale. When you have uniform standards across boundaries and borders, it facilitates international trade. Export-oriented Japanese firms must also be very cognizant of that.

One of the biggest concerns for people working on Internet policy is that the Internet may fragment soon, and China is often brought up as the example of where this could happen first. The real concern is over intervention and fragmentation at the infrastructure layer, such as in the transport layer and domain name system, in which there are international agreements which lower transaction costs.

Q: I understand that there are no standards for batteries for electric cars at present. For the combustion engine, there is a worldwide platform called gasoline. Do you have any thoughts about what the platform for electric cars will be?

Shane GREENSTEIN
Investment in better battery technology is viewed as a key learning curve. Sufficient technical advance in batteries that would bring costs down and/or functionality up within a weight constraint would facilitate an electric car. One venture capital strategy is to fund many startups working on battery technology and take an ownership stake in the intellectual property so that even if the car does not sell well and the enterprise does not succeed, they still have a financial interest in the intellectual property and can facilitate inter-temporal investment. Generating learning of battery technology is the key issue. Government subsidy would probably have a large benefit in accelerating technical development, but what is interesting is that the benefit is worldwide, rather than country specific. The intuition is that we are in a period of experimentation, and there are potential benefits to society that are very diffused.

Q: My understanding is that converters can solve the problem of tradeoff between economies of scale and variety or competition. There might then be a role for governments to subsidize the development of converters.

Shane GREENSTEIN
Yes, that is a justification for government subsidy, but the economic consequence of a converter can be quite different from what was anticipated. E-mail and 56K modems were positive examples, but sometimes converters can change the nature of competition in a way that was not intended.

Q: Are the concepts of de facto and de jure standards still relevant?

Shane GREENSTEIN
Today, I was mostly trying to talk about the frontier, but when I talked about private ordering that is really de facto, and the intuition about how platforms lead to standards comes from the intuition of increasing returns to scale in de facto standards, which are part of the older literature. When you start thinking about platforms and understanding the organizational form, the division between de facto, de jure, and mandated standards is not quite so clear anymore because platform organizational forms have hybrids. The FCC is now considering releasing more unlicensed spectrum in light of the value created from its use in Wi-Fi, but this is really neither de facto nor de jure. For some policies, you want to think slightly differently.

Q: If you take a long-term view, you often see examples of one standard overtaking another, such as in mass information storage, but in the long-run, things tend to come about unexpectedly. Is it right to say that the government should not think too much about the long-term consequences of intervention?

Shane GREENSTEIN
Governments should not think too much about design because they typically will not be able to design the product. On the other hand, government intervention should be focused on the long term, in the sense that its role in the competitive process is quite essential. The role for antitrust and government intervention is to commit to facilitating entry in the next period. Nobody else can do that. Thus, governments absolutely should think about the long term because the commitment to a legal regime that facilitates entry down the road also brings about the process that everyone anticipates, which is a Schumpeterian type of setting in which you get successive generations replacing one another and is quite essential to thinking about healthy competitive replacement in multiple generations. Even if governments don't commit to particular designs, committing to a legal regime that facilitates replacement is quite healthy. However, the specifics of what governments commit to change and not change are quite hard to do.