is owned and operated by the Economic History Association
with the support of other sponsoring organizations.

Learning by Doing in Markets, Firms, and Countries

Author(s):Lamoreaux, Naomi R.
Raff, Daniel M. G.
Temin, Peter
Reviewer(s):Gemery, Henry A.

Published by EH.NET (February 2000)

Naomi R. Lamoreaux, Daniel M.G. Raff, and Peter Temin, editors, Learning by

Doing in Markets, Firms, and Countries. Chicago: University of Chicago

Press (National Bureau of

Economic Research Conference Report), 1999. vii +

347 pp. $65.00 (cloth), ISBN: 0-226-46832-1; $22.50 (paper), ISBN:


Reviewed for EH.NET by Henry A. Gemery, Department of Economics, Colby College.

The “learning by doing” in the title of this NBER volume has little to do with

the classic case of the Horndal effect or with the productivity effects of

learning associated with the long production runs of aircraft or ship

construction. Instead, this volume deals with “scaled-up” learning by doing

concepts with an analytical reach that extends to a number of forms of

organizational learning and, in Gavin Wright’s concluding chapter, to learning

as a “national network phenomenon” (p. 296). The volume, drawn from an

interdisciplinary conference of business and economic historians,

is premised on the notion that information – its acquisition and use –

“effectively determines whether firms, industry groups, and even nations will

succeed or fail” (p. 15). Thus the learning examined in these studies falls

within that broad compass.

The first two essays examine how firms learn of the technological frontier and,

as well, learn how to appropriate best-practice technologies for competitive

advantage. In “Inventors, Firms, and the Market for Technology in the late

Nineteenth and Early Twentieth Centuries,” Naomi R. Lamoreaux and Kenneth L.

Sokoloff utilize patent data to develop a quantitative picture of the market

for technology. That market, they conclude, was well developed and thus

allowed firms to keep track of technological advances via intermediaries in the

market (patent agents and solicitors) or by direct contact with inventors. By

the beginning of the twentieth century,

however, firms increasingly attempted to move inventive activity within the

firm and that in turn required further learning on the part of the firm,

e.g., in minimizing employee turnover and insuring that patents received by

employees were assigned to the firm. The following essay by Steven W.

Usselman examines exactly this form of learning by American railroads and

their “internalization of discovery” (p. 63). Railroad managers from early in

the nineteenth century saw technical innovation in their industry as stemming

largely from “the efforts of ordinary

mechanics and engineers, not through discrete acts of patentable invention” (p.

63). Since railroads saw firm-specific knowledge as critical to the innovation

process, they not only attempted to internalize inventions but also attempted

to buffer the impact of external technical developments by forming railroad

associations and patent pools that insured patents would be cross-licensed to

the member firms. In a detailed and perceptive comment on the Usselman paper,

Jeremy Atack notes that such “… collusion stilled the winds of ‘creative

destruction’ that jeopardized the value of existing investment” (p. 101).

Forms of collusion or, more neutrally, institutionalized forms of information

interchange, were not confined to railroads. Avoiding the cartel label and yet

still providing interfirm coordination on pricing represents another form of

organization learning. In “The Sugar Institute Learns to Organize Information

Exchange,” David Genesove and Wallace P.

Mullin study a “technologically stagnant industry” (p. 106), U.S. sugar

refining from 1928 to 1936, where the learning question shifts away from

production technology to organizational innovation in interfirm information

sharing. The Institute did learn to organize and collect data while insuring

members’ confidentiality, thus allowing for “increases in the correlation of

firm decisions” (p. 133) as price and sugar stock data became available to all

members of the Institute. Not incidentally, the availability of common

information also precluded secret

price concessions.

A Supreme Court decision ended this particular form of organizational learning.

Kazuhiro Mishina’s paper on “Learning by New Experiences: Revisiting the Flying

Fortress Learning Curve” is the only paper in the volume that approaches

learning in its familiar learning curve form and the only one to draw on

econometrics in its analysis. The magnitude of the productivity increase in

Boeing’s B17 production from 1941 to 1944 was huge: the direct labor hours per

airframe dropped from 142,83 7 to 15,316, falling to nearly a tenth of the time

required at the beginning of the production run. What accounted for a

productivity increase of that size? Mishina rejects “the learning-by-doing

hypothesis that holds direct workers or engineers as the learning agent” (p.

175). Instead he finds the answer in the reduction in through-put time and “the

operating know-how that enabled it” (p. 175). No direct econometric test of

that conclusion is possible and the absence of learning taking place by direct

labor and engineers appears improbable. Not surprisingly then, Ross Thomson,

in his comment raises the question of whether the learning involved might have

been a cumulative process in which output growth, productivity growth, and

prior learning interacted.

The next two essays are intensive examinations of organizational

decision-making/learning. David Hounshell focuses on one critical meeting of

the Ford Motor Company Executive Committee on December 2, 1949. This is the

“Whiz Kid” era at Ford and Hounshell sees the meeting as defining a turning

point in Ford’s strategic course since the meeting reversed Ford’s strategy of

a decentralization of production. Hounshell asks how such a reversal came

about, explores several hypotheses, but concludes he can do no more than

speculate on the mechanisms that might have accounted for the Executive

Committee’s about-face on strategy.

Daniel M.G. Raff and Peter Temin’s essay also examines strategy decisions

within a firm, in this case two marketing decisions made by

Sears, one in the 1920s and a second in the 1980s. At the earlier date,

retailing channels were expanded from mail order operations to own retail

stores; in the latter case, financial services were added to the product array

in its retail stores. Again, as with Ford, the question is how these decisions

were made and whether they relied on the firm’s learning of its corporate

strengths and accurate perceptions of its competitive advantages in evolving

markets. Differences in leadership capability in the two eras were, in Raff

and Temin’s view, the critical variable at work. Leadership in the 1920s

focused on an attractive market that could be tapped by

“exploiting [the] firm’s existing competitive strength” (p 246). The 1980s

leadership failed in both learning the market and in recognizing Sears’

competitive strengths.

Perhaps of most interest methodologically is Leslie Hannah’s test of whether

the “lump of corporate capability” (p. 257) presumably possessed by the giant

corporations of 1912 grew or declined by 1995. Survivability to the 1995 date

is the first test, but Hannah also poses a second: among the survivors, how did

a given firm’s growth in market equity capitalization compare with a

price-deflated market index? Using those tests, Hannah notes that

“disappearance or decline was nearly three times more likely among the giants

than growth” (p. 271). Observing that high incidence of corporate decline and

failure, he turns to a consideration of what types of

“corporate architectures” and strategies

allowed large firms to “retain their position, continue to add value, and

expand their capabilities” (p.


In the final essay, Gavin Wright questions whether learning should be equated

solely with changes in total factor productivity. Rather, when looking at the

learning associated American economic growth in the nineteenth century, the

learning “was substantially a national network phenomena” (p. 296). As such,

“collective national learning may reside just as much in the discovery,

expansion, and accumulation of the factors of production as in their

productivity” (p. 296). To develop his point, he examines the U.S. mineral

industry, “one of the earliest and largest American technological networks,”

(p. 307) and the development of chemical engineering as it changed the way in

which chemical knowledge was acquired.

A collection of learning-by-doing studies as diverse as these serve to expand

definitions of the forms of learning. Can one measure the learning taking place

or generalize from the case studies, as Leslie Hannah and the editors attempt

to do? The answer would appear to be: with considerable difficulty. The problem

lies not only with the diverse definitions of learning employed, but also with

the difficulty of devising any empirical measures of the learning taking

place. Once one moves beyond the patent data of Lamoreaux and Sokoloff or the

production data of Mishina, measurement is elusive. One test would appear to

be success in the marketplace,

perhaps indicated by firm size and survivors hip – a measure that Hannah

attempts to make explicit. Market success may be an appropriate measure, if the

firm’s organizational capability, its use of patented technologies, or its

“ability to collect and use information effectively”

(p. 15) represent the major forms of learning occurring and can be linked to

market outcomes. However, as Bruce Kogut points out in his comment on Hannah’s

paper, there are more variables involved. “… a firm’s duration is contingent

on the evolution of its broader competitive and institutional landscape. This

broader landscape consists of firms, workers (sometimes organized in unions),

governments, political interests, research centers,

suppliers and buyers, idea merchants, and, of course, mechanisms of financial

intermediation and corporate governance” (p. 289). With that array of

variables at work, it may be that business and economic historians will not be

able to move significantly beyond case studies in examining these larger forms

of organizational and national learning. Or, as Leslie Hannah resignedly puts

it for the large corporation case: “To date, … we have made great strides in

storytelling, but a clearer, surer recipe for sustained success for large

corporations has remained elusive” (p. 270).

Henry A. Gemery is the author of “The Microeconomic Bases of Short-Run

Learning Curves: Destroyer Production in World War II,” (with Jan Hogendorn) in

Geoffrey Mills and Hugh Rockoff, editors, The Sinews of War:

Essays on the Economic History of World War II, Ames:

Iowa State University Press, 1993.

Subject(s):History of Technology, including Technological Change
Geographic Area(s):General, International, or Comparative
Time Period(s):General or Comparative