Entrepreneurs and American Economic Growth
Entrepreneurs and Path Dependence
Sources for the Lecture:
Carroll, Paul (1993). "The Day Bill Gates Overthrew Big Blue." Wall Street Journal, 16 August 1993.
Churella, Albert (1995). "Corporate Culture and Marketing in the Railway Locomotive Industry." Business History Review, 69, pp.191-229.
David, Paul (1986). "Understanding the Economics of QWERTY: The Necessity of History." In William Parker, ed. Economic History and the Modern Economist. New York: Basil Blackwell.
Diamond, Jared (1997). "The Curse of QWERTY." Discover Magazine, April 1997.
Folsom, Burton W. Jr. (1991). The Myth of the Robber Barons: A New Look at the Rise of Big Business in America. Herndon, Virginia: Young America’s Foundation.
Gomes, Lee (1998). "QWERTY Spells a Saga Of Market Economics." Wall Street Journal, 25 February 1998, p.B1.
Hughes, Jonathan (1986). The Vital Few: The Entrepreneur and American Economic Progress. New York: Oxford University Press.
Klein, Maury (1994). Unfinished Business: The Railroad in American Life. Hanover, NH: University Press of New England.
Liebowitz, Stan J. and Stephen E. Margolis (1990). "The Fable of the Keys." Journal of Law and Economics, 32.
Manes, Stephen and Paul Andrews (1993). Gates: How Microsoft’s Mogul Reinvented an Industry – and Made Himself the Richest Man in America. New York: Doubleday.
McDonnell, Greg (1997). "American Standard." Trains, Sept. 1997.
In a free market economy with potential Schumpeteran entrepreneurs all around, how do standards evolve? More precisely, is market failure possible with respect to the choice of a standard? That is, in a free market with several potential standards, is it possible for a technologically inferior standard to win? If so, why?
Numerous examples of "inferior" technologies winning in the market are produced as evidence by many authorities – DOS/WINDOWS versus the Macintosh; VHS versus BETA; and the most famous is the QWERTY versus the Dvorak typewriter keyboard.
The concept of path dependence is used to explain why some seemingly inferior standards (like DOS/WINDOWS 3.1 vs. the MAC) have evolved. Paul David defines it as follows: "A path-dependent sequence of economic changes is one in which important influences upon the eventual outcome can be exerted by temporally remote events, including happenings dominated by chance elements rather than systematic forces." The essence of the argument is that people get "locked-in" to the inferior standard. A critical mass of people adopts the standard thereby making the related goods cheaper (or more valuable) and thereby attracting more people to adopt the standard. In other words, a feedback effect. For example, as more people acquire DOS/WINDOWS computers, more software will be written for computers running this standard thereby making DOS/WINDOWS computers cheaper and more attractive for potential users. And so on.
This "lock-in" on an inferior standard is understandable in the case where switching to the superior standard would be very costly. For example, the U.S. uses 550 scanning lines on its television sets while Europe uses 800 scanning lines. European televisions are superior but the changeover would be enormously expensive. Another example is electricity. The U.S. uses 110 volt 60 cycle AC while Europe uses 220 volt 50 cycle AC. Switching to the superior European standard is totally impractical because of the costs.
But here is the rub. Consider two popular examples of "inferior" winning standards: DOS/WINDOWS vs. Macintosh; and VHS vs. BETA. Are these examples the same as the choice between 110 volt 60 cycle versus 220 volt 50 cycle AC? Clearly not.
Aside from the obvious switching cost differences, the U.S. AC standard is stable! In contrast, the so-called WINTEL standard is still evolving! The DOS vs. Macintosh example is something of a red herring because DOS supplanted CP/M and then later evolved into WINDOWS 3.1 which in turn has evolved into WINDOWS 95 and WINDOWS NT. The WINTEL standard is very rapidly absorbing and surpassing those features of the Macintosh that made it so popular.
In the VHS vs. Beta case, the feedback/lock-in theory does not work well because Beta was introduced first. In addition, although the picture quality of Beta is considered to be "superior", not everyone agrees and VHS tapes had longer recording times.
- The QWERTY Versus Dvorak Keyboard Debate
The "cleanest" example of market failure appears to be the QWERTY typewriter keyboard. It took its present form in 1874 when E. Remington & Sons first shipped a commercial "Type Writer". That keyboard layout is still the standard and appears on millions and millions of computer keyboards today. The Dvorak keyboard, patented by August Dvorak in 1936, appears to be a superior keyboard. It equalizes the loads on the left and right hands, it places the most commonly used letters on the middle row of the keyboard, and it is at least as good as the QWERTY keyboard in terms of maximizing the frequency of alternating hand sequences. The first two criteria produce less hand and finger strain while the third criteria increases the speed of the typing.
The arrangement of the letters on the QWERTY keyboard was the result of exhausting experiments by its inventor Christopher Sholes. The arrangement of the letters was due to his efforts to prevent jamming of the type bars. The standard story of why the QWERTY keyboard won in the marketplace during the latter part of the 19th Century is essentially a feedback/lock-in story well told by Jared Diamond and in Paul David’s articles.
The QWERTY keyboard thus appears to be a stable standard in the same sense that the 110 volt 60 cycle AC standard is – namely, it is not evolving in the sense that DOS/WINDOWS standard is (indeed, the DOS/WINDOWS standard is really the WINTEL standard now). Furthermore, switching from QWERTY to a better standard such as the Dvorak would not be cost prohibitive in the sense that a switch in the AC standard or the television-scanning standard would be. Indeed, mechanical typewriters are fading from the market and computers running word processing programs such as Microsoft Word or WordPerfect have largely taken their place. Remapping computer keyboards is relatively cheap and easy. Consequently, if anything, the costs of switching from the QWERTY standard to an alternative (especially for people who have yet to learn how to type) are falling.
It is these facts that make the debate over the superiority of the Dvorak keyboard over the QWERTY keyboard the focus of an important intellectual debate.
On the one hand, the QWERTY keyboard is used by many to argue that market share and technical superiority are not necessarily related. Indeed, some argue that it is a frequent occurrence (Liebowitz and Margolis cite numerous examples of authors who make these sort of claims).
The intellectual stakes here are quite high. If it true that "first-movers" have an advantage in that they can lock-in a standard through feedback and thereby prevent superior rival standards from emerging later, then this implies a role for government! After all, if the market fails in choosing the best standard, then government should step in and do so! Panels of experts (of course they should be appointed by the government) can do a better job!
Thus, as Liebowitz and Margolis point out, although not exactly in these words, the QWERTY example is yet another attempt by many authors to justify an industrial policy. (Actually, technology policy is more accurate.) No matter how it is phrased, however, substituting the government for the market is socialism plain and simple.
If market failure is as frequent as doomsayers suggest, then this means the role of the market entrepreneur would be limited indeed. What point would there be to producing a superior product unless you were the first mover! This is political entrepreneurship not market entrepreneurship.
In addition, if the government starts picking the technological winners, how will that effect future inventions within the purview of the standard? Jacob Schmookler’s research strongly suggests that inventive activity in terms of patents closely tracks investment. Creative minds go where the money is! The government destroyed the railroad industry as a vital part of the American economy through oppressive regulation early in the 20th Century. Inventive minds went elsewhere.
Funding basic scientific research is one thing (after all, that is where the internet came from!), but picking the winners in a dynamic economy is quite another. The government’s track record of picking winners is not encouraging.
Given the importance of the QWERTY example, Liebowitz and Margolis expend considerable effort at debunking what they dub as "The Fable of the Keys." Although I do not fault their scholarship, there is no "smoking gun". What appears to be missing from this literature is a set of convincing tests that one keyboard is clearly superior to the other. The Dvorak advocates appear to have a compelling case and Jared Diamond lays it out well, but provides no sources for his statements.
The common sense conclusion here is that the QWERTY keyboard is not that inferior for most of the vast number of people using it! Most people do not need to type 100 words a minute! What is missing from this whole debate is a convincing cost analysis. In other words, if sheer speed is the issue, the QWERTY keyboard appears to be fast enough for almost all uses of it. If you are just driving around town you do not need a 500 horsepower V8. My point is simple, the set of keyboard layouts that are fast enough for almost all useful purposes may be large and it may include QWERTY, Dvorak, and a lot of other keyboards.
- Bottom Line: I agree/disagree with Liebowitz and Margolis. I believe that QWERTY is clearly an inferior standard but it is not a compelling example of market failure. I think it is closer to the AC example in that the gains from switching to another keyboard are very small and they are far outweighed by the costs.
- The Case of Diesel Locomotives
The concept of path dependence is not about just standards like QWERTY and WINTEL. Recall David’s definition: "A path-dependent sequence of economic changes is one in which important influences upon the eventual outcome can be exerted by temporally remote events, including happenings dominated by chance elements rather than systematic forces." This definition also encompasses the effects of corporate culture on the development of new products. The failure of the Steam Locomotive industry is an excellent example of a "path-dependent sequence" in which the "remote" influences produced the prevailing corporate culture that in turn destroyed the industry.
The steam locomotive industry was undone by a classic Schumpeteran entrepreneur, Harold Hamilton, who founded the Electro-Motive Engineering Corporation. Like Carnegie, Hamilton’s basic innovation was to transfer business methods from one industry to another. In Hamilton’s case, he transferred marketing techniques he learned in the auto industry to the locomotive construction industry.
Hamilton founded EMC to enter a niche market – gasoline powered railcars. These were, in effect, small busses mounted on railroad wheels and powered by a conventional internal combustion engine. They were intended to replace passenger trains on little used branch lines. Hamilton used marketing techniques he learned in the auto industry to market his railcars. EMC made performance guarantees, provided field instructors who went on site to teach railroad crews how to operate and maintain the railcars, provided a quick spare parts service using standardized parts, and, most importantly, he bypassed the railroad operating officials and dealt directly with the financial departments.
His technique was to sell his standardized railcars based on cost. In contrast to the way the steam locomotive makers sold to the railroads, Hamilton started at the top and worked down rather than dealing with the operating executives who then went to their bosses for purchasing approvals.
By 1930 the railcar market was saturated and Hamilton correctly decided that the diesel locomotive market would be a natural for EMC. By joining with General Motors in 1930 he obtained the necessary financial backing to tackle the diesel locomotive market.
Diesel locomotives had many advantages over steam that were already apparent by the 1920s but the leading manufactures of steam locomotives failed to fully appreciate the implications of these advantages. Diesel locomotives were initially more costly per horsepower but were far more energy efficient. The advantages were:
Thermal efficiency was three times greater than steam.
Water not required (a big advantage for desert railroads).
Diesels were on line 90% of the time, steam only 50%. The Pennsylvania
Railroad's experience was typical of the major railroads. In 1945 it purchased two
diesels and quickly discovered how efficient they were. The two ran 69,000 miles
in their first six months without a single failure. In contrast, even the best
steam locomotive only ran about 3000 miles before needing some service. By the
early 1950s the PRR owned about 1900 diesels and was 100 percent diesel by 1957.
Diesels were lighter per horsepower and less likely to damage track and bridges.
Although both Diesels and Steam locomotives were too large to be manufactured on an assembly line, because the construction of Diesels required special-alloy steels and closer manufacturing tolerances, Diesels were manufactured according to standardized designs producing interchangeability of parts. This was not true of Steam locomotives.
Indeed, by the 1960s, the quality of the manufacturing of Diesel locomotives
was such that 95% of the SD40 family of Diesels built by EMD (see below) between 1966
and 1986 are still in service!
The one major advantage of Steam was in sheer pulling power. By 1940 EMC technicians were able to demonstrate that multiple-unit Diesels could be operated by a single engineer thereby producing the horsepower necessary to pull large trains.
Because of its marketing skills, the Electro-Motive Company (in 1940 it was made the Electro-Motive Division of GM) dominated the Diesel market by 1940. The same techniques that Hamilton used to sell railcars were used to sell Diesel locomotives. It provided demonstration units for free, it could help arrange financing through the General Motors Acceptance Corporation (GMAC), and it trained railroad employees how to run and maintain the units.
The American Locomotive Company was unable to compete. Its corporate culture resulted in the promotion of executives who were wedded to Steam through a lifetime’s worth of experience. The surge of orders in WWII fooled them into believing that Steam still had a long life. This plus their failure to adopt the marketing techniques of EMC proved to be their downfall. By the time that ALCo changed direction, it was too late. The entry of General Electric into the Diesel locomotive business sealed their fate.
- Entrepreneurial Standard Setting: DOS and the Evolution of WINDOWS
The success of Microsoft Corporation is, in part, also a story about
how a corporate culture – in this case IBM’s – can handicap an established firm
and allow a Schumpeteran entrepreneur –
William H. Gates – to win a market.
Microsoft began when Paul Allen, Bill Gates, and Monte Davidoff
wrote a BASIC interpreter for the first minicomputer in kit form, the Altair 8800,
in the spring of 1975. Allen devised an 8088 emulator that ran on a PDP10 so the
BASIC code could be developed. Gates wrote the BASIC code and Davidoff worked
on the math routines.
The company that made the Altair (MITS – Micro Instrumentation and
Telemetry Systems) – located in Albuquerque, New Mexico – signed an agreement
with Gates and Allen on 22 July 1975 (but made effective as of 1 March 1975 when
Allen first got the BASIC running on the Altair in Albuquerque). Micro-Soft
began as a simple partnership of Gates and Allen and was headquartered in
Albuquerque from 1975 to 1978.
By December of 1978 Microsoft had 11 employees and an income of
over $1,000,000 in that calendar year. Its major moneymaker was BASIC which
was in big demand on a variety of microcomputers. Microsoft had agreements
with Altair, Apple Computer, Radio Shack, Commodore, and several other computer
Microsoft moved to Bellevue, Washington in December 1978. By 1980
Microsoft had 40 employees and had branched out into FORTRAN and COBOL. On
22 July 1980 IBM met with Gates and Steve Ballmer at the Microsoft offices and
sounded Microsoft out about some vague experimental computer project (after
having Gates and Ballmer sign nondisclosure agreements). On 21 August 1980
the IBM people came back and told Microsoft that it was their first choice to
provide interpreted BASIC, COBOL, FORTRAN, Pascal, and a BASIC compiler for
their new personal computer.
IBM needed an operating system so Gates sent the IBM people to see
Gary Kildall of Data Resources who was developing CP/M-86 – a follow-on to
his highly popular CP/M operating system. The IBM people flew down to San
Francisco to see Kildall but came away empty handed. Kildall was out flying to
another business appointment and his wife – Dorothy McEwen who handled all
the business deals – met with the IBM people. She refused to sign the
non-disclosure agreement and IBM left. However, even had she signed, CP/M-86
was not ready and IBM could not have used it in any event.
The rest, of course, is history. In September Paul Allen negotiated
a deal with Seattle Computer Company to license 86-DOS (originally, QDOS –
Quick and Dirty Operating System – written by Tim Paterson) for $10,000 for a
single company ($15,000 if the source code was included). Microsoft could
distribute DOS to an unlimited number of end users using that company’s computer
but would have to pay an additional $10,000 for each additional company.
On 27 July 1981 Microsoft bought DOS outright from Seattle Computer for an
On 6 November 1980 Microsoft signed its deal with IBM to supply DOS,
Pascal, BASIC, FORTRAN, and COBOL. IBM could not license Microsoft’s programs
to third parties but Microsoft retained the rights to do so.
In hindsight, the deal of the Century. On 12 August 1981
IBM announced the IBM Personal Computer. Gates and Allen were not invited to
the roll out.
WINDOWS 1.0 was introduced in November of 1983, WINDOWS 2.0 in
October of 1987, and WINDOWS 3 in May of 1990, WINDOWS 3.1 in April of 1992,
and WINDOWS NT 3.1 in July of 1993. In 1987 Bill Gates became the youngest
- William H. Gates: Summary of Business Style: 1) hard working;
2) very competitive; 3) a skilled business strategist – persistent if he
thinks he is correct but willing to reverse course overnight if need be;
4) the ability to pick gifted associates.
NOMINATE Data, Roll Call Data, and Software
Course Web Pages: UC San Diego (2004 - )
University of San Diego Law School (2005)
Course Web Pages: University of Houston (2000 - 2005)
Course Web Pages: Carnegie-Mellon University (1997 - 2000)
Spatial Models of Parliamentary Voting
Recent Working Papers
Analyses of Recent Politics
About This Website
K7MOA Log Books: 1960 - 2006
Bio of Keith T. Poole