Against Platform Monopolies: Returns

On Sunday I offered one reason that platform monopolies might not stimulate as much innovation as the ideal case would suggest.

Here’s another reason: R&D spending often has diminishing returns. The surplus generated by a successful new platform (DOS/Windows, x86, iTunes/iPod, etc) provides an enormous windfall to the company that creates it–a windfall that may be totally out of proportion to the amount of money the company spent developing the platform. The first version of Microsoft’s DOS operating system, was called QDOS (for quick and dirty OS) and was licensed from another company for a small sum. It’s not at all obvious that there’s any connection between the amount of R&D Microsoft did on early versions of DOS and the large subsequent profits they made.

If we had changed public policy in the 1970s to halve or double the long-run rewards to creating a PC operating system, it’s not obvious it would have changed the outcome for Microsoft one bit. For many technological categories, the winning platform tends to be made by the firm that is in the right place at the right time. Above a certain point, increased R&D spending is likely to show diminishing returns. It only costs so much to develop a microchip, an iPod, or an operating system. If the return on doing so is 10 or 100 times the cost, we don’t want firms to spend money beyond the point of diminishing returns.

Recall that last week I offered some reasons to think that boosting the rewards for platform creators generated some deadweight losses. Given the existence of diminishing returns and zero-sum competition, it’s clear that the goal should be to have the right return on platform creation (i.e. one that’s sufficient to stimulate the creation of new devices), not the largest possible return.

So how large a return is necessary?

It’s important to keep in mind that firms don’t generally attract imitators until after their product has been recognized as a hit. It would be a waste of time to produce a clone of a product that’s not selling very well. This gives the initial firm a substantial head start. Even without a platform monopoly, the inventor of an innovative device is likely to have the field to himself for the first year or more. That gives him time to build brand recognition, refine his product and production process, and begin building supporting products.

Moreover, in the early phase of a new platform, it’s not at all obvious that competitors hurt the incumbent’s bottom line. Especially if a platform has strong network effects, having several firms selling products compatible with your own may stimulate demand for your own product, both because it helps generate buzz and because it makes your product more useful. Although the proliferation of PC clones undeniably hurt IBM in the long run, in the short run it was arguably helpful, as it increased the profitability of producing software and peripherals for the PC platform, thereby making IBM’s PCs more useful.

To put this point another way, device-specific property rights “skew early” in the returns they generate. A device is most profitable in the months after its release, as it’s brand new and can attract a high price from early adopters. In contrast, the returns from a platform monopoly “skew late”–that is, they become important only after that platform has begun to mature.

This is important because technological firms–and especially startups, which produce some of the most innovative new devices–have high discount rates when they’re evaluating the potential returns on a new technology. The fact that creating a new platform could lead to enormous profits 15 years later is unlikely to have much impact on the firm’s decision-making, both because such returns tend to be highly unpredictable, and because the firm likely has much more short-term cash flow needs. If a device loses money in years 1 through 5, it’s likely to be cancelled, no matter how rosy its prospects look in year 15.

I also think there’s plenty of evidence that the originators of new platforms continue to enjoy a healthy market share even after the platform becomes highly competitive. Intel’s the poster child for this theory, holding 80 percent market share despite more than a decade of vigorous competition. IBM is an obvious counterexample, but IBM made some major blunders that largely account for its poor performance in the 1990s. In a nutshell, IBM tried to replace the open PC architecture with new, proprietary platforms of its own making, and the rest of the market wasn’t interested. (Had IBM concentrated in producing best-of-breed ISA/DOS PCs instead of trying to push MIcroChannel/OS2 PCs on its customers, it’s likely IBM would have been much more successful in the 1990s)

If someone forced the iPod/iTunes platform open, it’s likely that the brand recognition they’ve built up would be more than sufficient to allow them to continue selling millions of iPods a quarter for the foreseeable future.

So the incentives created by device rights are likely sufficient to give companies a reasonable return on the new devices they create. On the other hand, platform rights are often too large and too late to have much of a stimulative effect on R&D spending. I’ll wrap this up next time with some additional reasons to think that there will be plenty of innovation without platform monopolies.

SHARE: 8 comments