Taking a second look at the idea that antitrust action created the US software industry

James Pethokoukis of AEI.

"Advocates of aggressive antitrust actions against large American technology companies such as Google and Facebook often argue that today’s super-successful tech industry owes a debt to antitrust actions of the past. Under this theory, US government investigations and lawsuits tamed and/or distracted the monopolistic tech giants of the past — even though the companies were not broken up — allowing competitors to bloom and grow. So Google and Facebook should thank the Clinton administration for suing Microsoft in 1998, and the entire US software industry should thank the Johnson administration for beginning a 13-year legal pursuit of IBM in 1969. Likewise, action today is needed to prevent modern tech titans from squashing future competitors.

In a previous blog post, however, I outlined my skepticism about the distraction theory as it applied to Microsoft. And I am also somewhat skeptical of the IBM story. Now as Matthew Stoller of the Open Markets Institute has plainly put it, “IBM implemented a software unbundling policy due to scrutiny from the antitrust suit. This allowed a software industry to emerge.”

This is a strong claim. It refers to the company’s 1968 decision to price and sell software separately from hardware, supposedly the spark that created an independent software market. As Stoller and other activists tell the story, IBM’s decision was in reaction to the antitrust investigation that began in 1967, a failed attempt to head off eventual government legal action.

So two questions here. First, what was IBM’s motivation for the unbundling? IBM argued that the decision was in reaction to changing business conditions, not a ploy to avoid legal action. And there is some reason to think this is not an entirely implausible argument. As Edward Steinmueller writes in his 1995 paper, “The US Software Industry: An Analysis and Interpretive History” (also cited by “The Cambridge Economic History of the United States”):

The motives for IBM’s unbundling decision are disputed. One interpretation is that IBM’s actions were made in response to anticipated litigation. [In “IBM and the US Data Processing Industry: An Economic History,” authors Franklin Fisher, James McKee, Richard Manke] dispute this view, noting that no direct evidence of relation between the announcement and the DOJ antitrust action was discovered during subsequent litigation.  These analysts instead argue that IBM’s costs of software support were increasing rapidly. The costs of developing the System/360 operating system software had proved a major trauma for the company, and seemed to foreshadow still further cost increases.  In addition, the growth of independent software vendors made it possible by the late 1960s for IBM to consider separate pricing for software and to retreat from its commitment to provide all of the software tools that users might need in order to purchase or lease IBM computers.

So if this alternate take is true, market forces rather than government intervention was the action-forcing factor.  Which leads to the second big issue: How important to the emergence of the software industry was IBM’s decision, whatever its reason for taking it?

Clearly there was more happening in the computer industry than just that one decision, no matter how momentous. Indeed, the launch of the aforementioned IBM/System 360 mainframe computer in 1965 seems critical. As Computerworld magazine put it in 1976, citing the US vs. IBM testimony of government witness Larry Welke, president of International Computer Programs: “For the first time . . . software vendors had a large-potential market for their products. Before the 360 and its wide acceptance, the limited transferability of programs for other manufacturers’ machines tended to fracture the market.”

What’s more, the 1960s also saw the development of microcomputers, providing an alternate technology to IBM’s big iron, and an alternative software market to sell into. Again, Steinmueller:

Applications and affordability were responsible for rapid growth in the minicomputer sector during the late 1960s, the period when time sharing operating systems for mainframes were floundering. By 1970 . . . minicomputer unit shipments exceeded those of mainframes and, consistent with their price, were achieving about one seventh of mainframe revenues.

Finally, as for IBM being distracted, guess what: Successful incumbents often fail to keep their eye on the ball, whether or not they have Uncle Sam looking over their shoulder. They find it hard to change what’s been working in response to changing market conditions. This from the MIT Sloan Management Review:

In the 1980s, IBM’s profit margins suffered a steep decline. Because the company’s costs remained level, profits dropped. Critics of the company have widely attributed IBM’s decline to two factors. During this period, IBM became a follower of technological development, more so than in the past. Such a change was marked because, in the 1960s, IBM had led the information technology industry with a grand innovation — the 360 Series of computers. Also, the company displayed a surprising naiveté in its partnering strategies, giving Microsoft and Intel extremely profitable portions of the industry while choosing to retain less profitable portions for itself.
Though these factors are very important, they are not the root causes of IBM’s difficulties. For example, the decline of profit margins was a result of falling customer interest in mainframe computers. That IBM executives failed to foresee this was the result of two more basic factors. First, IBM’s enormous R&D effort of the 1970s should have been directed at the microcomputer, which was about to burst onto the technological scene bringing a future full of personal computers, networks, and computer servers. Instead, the company squandered R&D on building a larger mainframe.  Second, IBM shifted its relationships with customers and lost touch with their interests and concerns. Thus a partial explanation of IBM’s difficulties is that its profit margins on mainframes declined precipitously.

Again, the elegant and simple antitrust explanation may well be correct. And I am certainly open to it. But the issue certainly seems contestable, just as it is for Microsoft. Perhaps more serious research needs to be done before we start breaking up or heavily regulating today’s tech titans in the name of innovation."