6_ Abernathy - Utterback -Innovation Over Time and in Historical Context

5/14/2018 6_ Abernathy - Utterback -Innovation Over Time and in Historical Context - sl...

http:///reader/full/6-abernathy-utterback-innovation-over-time-and-in-histor

~.- 1I N N ! O V A T I O ' N OVER TIME

AND IN H ~ S T O R ,I C A L :C O 'N T E X T

Patterns of Industrial InnovationWilliam]. AbernathyJames M. Utterback

.i no;.!. 'model suggests how t he c ha ra c te r o f it s innovat ion ,h.(!n g~ s a s a su cc ess fu !eruerprise matures; and how a/A ir compan ie s may c ha nge themselues to fo ster 'in no va tio na s ( h e y g row an d p ro sper.

How does a company's innovation-e-and its re-sponse to innovative ideas----change as (he com-pany grows and matures?

Are there circumstances in which a pat-tern generaUy associated with successful inno-vation is in fact more likely to be associated withfailure?

Under what circurnstance s : will ne~"Jyavailable technology, rather than the market.be the critical stimulus for change?

When is concentration on incrementalinnovation and prcductivity ,gains likely to be of

The research reported in this ani de was supportedby !.he Nationai Science foundation Division of Policy Re-search and Analysis.

maximum value (00 a firm? In what situationsdoes this straregy' instead cause instability andpotential for crisis in an organization?

Intrigued by questions such as these, wehave examined how w e kinds of innovationsattempted by _ productive units apparentlychange as these units evolve. Our goal was amodel relating patterns of innovation within a .unit co that unit's cornperirive s[ra~eg;y, produc-don. capabilities, and organizarional character-rsucs,

This article summarizes our work andpresents (he basic characteristics of the model(0 which it has led us. We conclude that a pro-ductive unit's capacity for and methods of inn a-varian depend critically on its smge of evolution

25 .



SECTION I tnnovetion Over Time and in Histories! Coruext

from a small technology -b ased enterprise to am ajor h igh -volum e p roduc er. ~!an:character is-tics of innovation and the innovative proc esscorrelate with such an historical anal pis; and onthe basis of our model he can now attemptanswers to questions such as those above.T.1.BL t.

A SPECTRU~t OF E'\,'l','OYATORSPast studies of innovation imply that an~ 10novating unit sees. most of its inuovauons a:new products, B~t that observation masks aress.eruial difference: hat is a product inriova

Fluid Pattern Transitional Pattern Specific PatternCompeti tive emphasison

COS! reductionunct ional productperformance

P ro du ct v ari ati on

Innovation stimulatedby

Information on users'needs and users'technical Input;

Opportunities createdbv expanding internaltechnical capability

Pressure to reduce CGJstand improve qualit ,

Predominant type ofmnovauori

Frequent major changesin products

\!ajor process changesrequired by risingvolume

Incremental for productand process, withcumula t iveImprovement inproducriviry and qualitv

Product line :-"'!ostiy undifferentiatedstandard pr oduct s

Diverse, often includingcustom designs

Includes O le least oneprod uct design stab leenough to havesignificant productionvolume

Production processes Flexible and in.efficient:major changes easilyaccommodated

Becoming more rigid.with changes occurringin major steps

E m cien t,capual-iruensive. andrigid; cost o r change ishigh

Equipment General-purpose,requiring high!)' skilledlabor

Some subprocessesauromared. creating"islands of automation"

Special-purpose. mostlyaut ornaric with labortasks mainly monitoringand control

Materials Inputs are Limited toge neral Iy-a vailablematerials

Specialized materialsmay be demanded fromsome suppliers

Specialized materials. . . . i L l be demanded; ifnot available. verticalim egrarion will beextensive

Plant Small-scale" locatednear user or source oftechnology

General-purpose ....ir hspecialized sections

Large-scale, highlyspecific to particularproducts

Organizational controlIS

Informal andentrepreneurial

Through liaisonrelationships, projectand task groups

.Through emphasis onstructure, goals" andrules

26



William J. Abernetnv and James M. Unerbeck

cion by a small. technology-based unit is often. the process equipment adopted by a large unitco improve its high-volume production of astandard product. We argue {hat these twounirs=-rhe small. erurepreneuna I organiza uo nand [he larger unit producing standard pro-ducts In high volume-are at opposite ends ofa spectrum, in a sense formIng boundary condi-tions in the e ....luti on of a urut and in the char-acrer of its innovat ion of produc t and processtechnologies.

One distinctive pattern of technologicalinnovation is evident in the case of established.high-volume products such as incandescentlight bulbs. paper. steel, standard chemicals.and internal-combustion engines, Eor exam-ples,

The .rnarkets for such goods are well de-fined: the product characteristics are well un-derstood and ofcen standardized: unit profitmargins are typically low: production rechnol-ogy is efficient, equipment-intensive. and spe-cialized [Q a particular product: and competi-tion is primarily on the basis of price. Changeis costly in such highly integrated systems be-cause an alteration in anyone attribute or pro-cess has ramifications for man}' others.

In this environment innovation is typt-Rare ofM.ajor

Innovation

ProductInnovation

cally incremental In nature, and it has a gradual .cumulative effect on productivity, For example.Samuel Hollander has shown chat more {han,half of the reduction in the COSt of producingrayon \0 planes ofE. I.du Pont de ;\emours andCo, has been th e result of gradual process irn-provernenrs w hich could not b e identified asformal projects or changes, A sirnilarsrudy b yJohn Enos shows chat accumulating. icrerneri-{. d developments in pecroleum relining pro-cesses resulted in productivity gains \...hichoften eclipsed th e gain from rh e original inno-vation, Incremencal innovations. such as theuse of larger railroad cars and unit trains. haveresulted in dramatic reductions in the cost ofmoving large quantities of materials by rail.

In all these examples. major systems 10-novations have been followed b y countlessminor product and systems improvements. andthe latter account for more than half of the totalultimate economic gain due to their muchgreater number'. \Vhile cost reduction seems tohave been the major incentive for most of theseinnovations, major advances in performancehave also resulted from such small engineeringand production adj ustrnencs.

Such incrememal innovation typicallyresults in an increasingly specialized system in

PrOC~5S

FfGl'RE I.27



SECTION f lnnovs tion OVer Time' and in Historicet Context

wh i c h economies of scale in produc tion and th edevelopment of mass markets are extremelyjrnportant. The productive unit loses its flexi-bilitv, becoming increasingly dependent onhigh-volume production to cover its fixed costsand increasingly vulnerable to changed de-rnand and technical obsolescence.

Major new products do not seem to beconsistent with this pattern of incrementalchan gf.". i\ew produc ts w h ich req u ire reorien ta-[ion of corporate goals or production facilities{end co originate outside organizations devoted(0 a "specific" production system; or, if origi-nated within, to be rejected by' them. '

A more Aui d p att ern of product changeIS associated with [he identification of anemerging need or a new w ay to meet an existingneed; it is anerurepr eneurial act. Man}' studiessuggest that such new product innovationsshare common traits. The>' occur in dispropor-tionate numbers in companies and unitslocated in or near affluent markets with strongscience-based universities or other research in-sritutions and entrepreneurially oriented fi -nancial institutions. Their competitive advan-tage over predecessor products is based onsuperior funct ional performance rather thanlower ini tial cost, and so these radical innova-tions lend [0offer higher unit profit margins,

When a major product innovation firstappears, performance, criteria are typicallyvague and little understood. Because they havea more intimate understanding ofperfonnancerequirements, users may p r a y a major role in -,suggesting the ultimate form of the innovationa s well 3S the need. For example. 'KennethKnight shows that three-quarters of the com-puter models which emerged between 1944and 1950, usually those produced as one or tWOof a kind, ....ere developed b y users.

It is reasonable that the diversity and un-certainty of performance requirements for newproducts give an advantage in their innovationto small, adaptable organizations w-ith flexibletechnical approaches and good external com-28

municacions, and historical evidence suppOrtSthat hypothesis. For example, John .Tiltonargues that new enterprises led in the applica-t ion of semiconductor technology, often trans-ferring into practice technology from more es-tablished firms and laboratories. He argues thateconomies of scale have not been of prime im-porrance because products have changed so ra-pidly that production technology designed fora particular product is rapidly made obsolete,And R. 0, Schlaifer and S. D. Heron have ar-gu-ed (hat a diverse and responsive group ofenterprises struggling against established units(Q enter the industry contributed greatly to [heearly advances in jet aircraft engines.

A TRA..NSITION FROM RADICALTO EVOLUTIONARYINNOVATIONThese two patterns of in nov arion may be takenro represent extreme types-in one case involv-ing incremental change to a rigid, efficient pro-duction system specif ical ly-designed to producea standardized product, and in the other caseinvolving radical innovation with product cha-racteristics in flux . They are not in 'faC[ rigid,independent categories. Several examples w illmake it clear that organizations currently con-side-red in the "specific" caregory-i-where in-cremental innovation is now motivated by C o O S treduction-were at their origin small, "fluid"units intent on new product innovation.

John Tilton's' study of developments inthe semiconductor industry from 1950 through1968 indicates that [he rate of major innovationhas decreased and that the type of innovationshifted, Eight of the 1.3 product innovations heconsiders to have been most important duringthat period occurred within the first s.evenyears, while the industry was making less than.5 per cent of its total 1B-year sales. Two typesof enterprise can be identified in this early pe-



Will/am J. Abernathy and James M. Utterbeck

nod of the new industrv-c-cstahlisbe d units (hatcame into semiconductors from vested posi-tions in vacuum tube markets, and new entriessuch as Fairchild Semiconductor. 'LB.:'.!., andTexas Instruments. Inc. The established unitsresponded rocornperir ion from th e n ew c om ersb y emphasizing process innovations .. i\Jean-while , the latter sought entry and strengththrough product innovation. The three verysuccessful new enrrarus jusr listed were respon-sible for half of the major product innovationsand only one of the nine process innovationswhich Dr. Tiltonidemified in that IS-year pe-riod, while three principal established units(divisions of General Electric, Philco. andR.C.A.) made only one-quarter of the productinnovations but three of rhe nine major processinnovations in the same period. In this caseprocess innovation did' not prove to be an ef.fe ct iv e c omp et it iv e stance; b y 1966 the threeestablished units together held only 18 per cent'of the market while the three new urucs held 42 'per cent, Since 1968. however, (he basis ofcompetition in the industry has changed; ascosts and productivity have become more im-porcant, the rate of major product innovationhas decreased. and effective process innovationhas become an important factor in competitivesuccess. For example, by 1973 Texas Instru-rnents which had been a flexible. new entrant in[he industry two decades earlier and had co n-tributed no major process innovations prior to1968, was planning a single machine mae wouldproduce 4 per cent of world requirements forits integrated-circuit unjt,

Like the transistor in the electronics in-dustry. the DC-3 , stands out as a major changein the aircraft and airlines industries .. AlrnarinPhillips has shown that the DC-3 was in fact acumulation of prior innovations. It ..as not thelargest, or fastest. or longest-range aircraft; itwas the most economical large. fast plane ableto fly long distances. All the features whichmade this design so completely successful hadbeen introduced and proven in prior aircraft.

Arid the DC-3 has essentially the first cornrner-cial product of an. erit e ririg firm (rhe C -I an dDC-2 were produced by Douglas only in s rnal]numbers).

JUSt 3.5 the trans is tor PUt the el ectron icsindustry on a new plateau, so the DC-3 changedthe character of innovation 10 the aircraft in-dustry for the next 1.5 vears. : - - ; 0 major innova-tions ",'ere introduced into commercial aircraftdesign from 1936 unril new jet-powered air-cr:J.fl appeared in the 19505. Instead. there "eresimp I;' many refinements to the DC-:) coo-cepe=-stretching the design and adding ap-pointments; and during th e period of these in-'erememal changes airline operating COSt perpassenger-mile dropped an additional 50 percent.

The electric light bulb also has a historvof a loner series of evolutionary irn orovernerusC! which starred ",'I(h a fe\~' major innovations andended in a highly standardized cornrnodicv-likeproduct.iBy 1909 the initial tungsten filamentand vacuum bulb innovations were in place;from then until 1955 there came a series ofincr ernencal changes= beuer metal alloys forthe filament, the use of "geu ers" to assist inexhausting the bulb. coiling the filaments."frosting" (he glass. and many more. In thesame period (he price of a 50-watt bulb de-creased (even \..irh no inflation adjustment)from S 1.60 to '20 cents each. [he lumens outputincreased b y 17 5 per cent, the direct labor con-Lent was reduced more than an order of magni-tude, from 3 to 0.18 minutes per bulb. and the-production process evolved from a flexible job-shop configuration. involving more than 1 1separate operations and a heavy reliance on theskills of manual labor, to a single machine at-tended by a few workers.

Product and process evolved in a similarfashion in (he automobile industry. During afour-year period before Henry Ford producedthe renowned Mode! T. his company devel-oped. produced. and sold five different en-gines. ranging from two (0six cylinders. These



SECTION I tnnovetion Over Time and in Histories! Context'" .

THE UNIT OF A. . l~ALYS]SAs we show in t his article. innovation within an estabfished industryis ofte n limited ro incremental improvements of both products andprocesses ..iViajor product change is of len introduced from outside anestablished industry and is viewed as disruptive; its s ource is typically[he start-up of a ne\\'; small firm. invasion of markets by leading firmsin other industries.or government sponsorsh ip of chan t!c ei ther as aninitial purchaser or th rough direc t regul< l.lion.

These c i rcumstances m C : : 1 1 1 (hat (h e standard unit s of analysisof industry-i-firm and product type-.are of little use in understandinginnovarion. Technological change causes these t erms i((I change theirmeaning. and {he "very shape of {he production pro:~~ is altered.

Thus the questions raised in this article require (hal a productline and its associated production process b e taken LOg-eth er as th eunit of analy sis. Th is w e term a "productive unit." For :1 simple firmor a firm devoted to a single product, the productive unit and the finnwould be one and the same. In the case ora diversified finn. a produc-tive unit would usually report to a single operating manager andnormally be a. separate operating division The extreme of a highl}'fragmented production process mighll mean that sc\'('ral separatefirms taken together would be a productive unit.

For example. analysis of change in the rexrile industry requiresthat product ive units in the chemical . plastics. paper,



William J Abe'rnq(hy and Jernes M. Um:rbiJc:k

In vet another case. Robert Buzzell ....ndRobert 1';'ourse, tracing irmovations in proc-essed foods, show - (hat new products such assoluble coffees, frozen vegetables, dry pelfoods, cold breakfast cereals, canned foods,and precooked rice came hrst from individualsand small organizations where research was inprogress or which relied heavily upon informa-cion from users. As each product won ~ccept-arice. its. productive unit increased in size andconcentrated its innovation on improvingmanufacturing, marketing, and distributionmethods which extended rather (han replacedthe basic technologies. The major source of chelatter ideas is now each firm's own research anddeve 1 0 prneru organ iza tion.

The sh ift frorrr radical to evolutionaryproduct innovation is a common thread inthese examples. It.is related to "he develop-ment of a dominant product design, and i t isaccompanied b " y heightened price competitionand increased emphasis on process innovationSmall-scale units that are Aexible and highlyreliant on manual labor and craft skills utilizinggeneral-purpose equipment develop into units(hat rely on automated, equipment-intensive,high-volume processes. We conclude thatchanges in innovative pattern, production pro-cess, and scale and kind of productioncapacityall occur together in a consistent. predictable

Though many observers emphasize new-product innovation, process and incrementalinnovations may have equal or even greatercommercial importance, A high rate of produc-tivity improvement is associated with processimprovement in every case we have studied.The CQH of incandescent light bulbs, for exam-ple, has fallen more thanStl per cent si, ce theirinrroduction.iAirline operatingcosts were cutby half through the development and improve-merit of the DC-S. Semiconductor prices havebeen f:alling by 20 (0 30 per cent with eachdoubling. of cumulative production. The intro-

duct ion of the ;"'Iodel T Ford resulted iria pricereduction from S3,OOO to less than 51,000 (inL958 dollars). Similar dramatic reductions havebeen achieved ill the costs of computer corememory and television picture tubes.

MANAGING TECHNOLOGIC\LINNOVATIONIfit is true (hat [he nature and goals of an indus-tr~~l unit's innovations change as [hal uni t rna-rur es from pioneering {Q large-scale producer",what does this imply for the management oftechnology?

We believe that some significantmanagerial concepts emerge from our anaiv-sis-or model', if you will--ofthe characteristicsof innovation as production processes and pri-mary competitive issues differ. As a unit movestoward large-scale production. (he goals of itsinnovations change from ill-defined and uncer-tain targets to well-articulated design objec-tives. In the early stages there is a proliferationof product performance requirements and de-sign criteria which frequently cannot be statedquantitatively, and their relative importance orranking may be quite unstable. It is preciselyunder such conditions, where performance re-quirements are ambiguous, chat users are mosrlikely to produce an innovation and wheremanufacturers are least likely to do so. One wayofviewing regula~ory constraints such as thosegoverning auto emissions or safety is W ac m e radd new performance dimensions co be re-solved b y [he engineer-and .50 may lead tomore inn ova rive design irnprovernents. Theyare also likely to open market opportunities forinnovative charige of thekind characteristic offluid enterprises in areas such as instrurnenta-ilion, components. process equipment; and soon.

The stimulus for innovation changes. as

31



SECTION! Innovation Ov'er Time and in Historical Context

a unit matures. In the initial fluid ;5tage, marketneeds are ill-defined and can be stared only.....th broad uncertainty: and the rele-vant tech-nologies are as yet liule explored. So there aretwo sources of ambiguity about the relevance ofany particular program of research and devel-opmem-e-targ et uncertainty and technical un-certainty. Confronted with both types of uncer-tain ty , the decision-maker h as Iirrle incentivefor major investments in formal research anddevelopment.

As the enterprise develops, however, un-certainty about markets and appropriate tar-gets is reduced, and larger research and devel-opment investments are justified. At some. point before [he increasing specialization of theunit makes the cost of implementing technolog-ical innovations prohibitively high and beforeincreasing- cos t competition erodes profits wichwhich 1O fund large indirect expenses, thebenefits of research and development effortswould.reach a maximum. Technological oppor-tunities for improvements and additions 1O ex-isting .product lines will then be dear, and astrong commitment to research and develop-men t will be characteristic of productive unitsin the middle stages of development. Suchfirms ....i l l be seen as "science based" becausethey invest heavily in formal research and engi-neering departments, with emphasis Onprocessinnovation and product differentiation throughfunctional improvements.Although data on research and develop-ment expenditures are not readily available onthe basis of productive units, divisions, or linesof business, an informal review of the activitiesof corporations with large investments in re-search and development shows that chey tendto support business lines that fall neither nearthe fluid nor the specific conditions but are inthe technologically-active middle range. Suchproductive units tend (0 be large, to be irue-grated, and to have a large share of their mar-kets. A small, fluid entrepreneurial unit re-32

quires general-purpose process equipmentwhich is i t ) 'p icaUy purchased. As it develops,such a unit is' expected to originate some pro-cess-equipment innovations for its own use;and when i ( is fully matured its entire processesare likely to be design~d as integrated systems"specific to particular products. Since the ma-ture firm is new fully specialized, all its majorprocess innovationsare likely to originate OUt-side the unit.

, But note char the supplier companieswill now see themselves as making product->-not process-innovations. From a differentperspective, George Stigler finds stages of de-velopment-similar to those we describe-inFinns that supply production-process equip-ment. They differ in the marker structure t heyface, in the specialization of [heir productionprocesses, and in [h e responsib ili t ies rh ey mustaccept in innovating co satisfy their own needsfor process technology and materials.

The organizariori's methods of coordi-nation and control change with [he increasingstandardization of its products and productionprocesses. As task uncertainty confronts a pro-ductive unit early in its development, the unitmust emphasize its capacity to process informa-tion by investing in vertical and lateral informa-tion system5 and in liaison and project groups.. Later, these may be extended to the creation offormal planning groups:, organizational manife-stations of movement from a product-orientedto a transitional state: controls for regulatingprocess functions and management. controlssuch a s job procedures, job descriptions, andsystems analyses are also extended to become

- < I . more pervasive feature of the production net-work.

As a productive unit achieves standard-ized products 'and confronts only incrementalchange, one would expect it to deal with com-plexity by reducing the need for informationprocessing. The level at which technologicalchange takes place helps to determine the ex-tent to .....ich organizational dislocations take



,William J. Abemarhr and James M. Utterback",

DESIGN AS A MILESTONE OF CHANGEThe rnilestorie in all the examples of transition in (he accompanyingarticle is a dominant new produce synrhesized from in divid ual tec hno -.logical innovations introduced independently in prior products. Thisdominant design has the effect of enforcing standardization so thatproduction economies can be sought. Then effective cornpetitionbegins to take place on [he basis of COS! as well as of product perform-ance.

Similar product design milestones can be identified in otherproduct lines: sealed refrigeration units for home refrigerators andfreezers, effective can-sealing technology in the food canning industryand the standardized diesel locomotive in the locomotive and railroadindustry ..In each case the milestcne'signals a significant transforma-tiori, affecting the lype of innovation which followsit; (he source ofinformation. and the size, scope, .and use of formal research anddevelopment.

In an earlier articleIn this series, George . R : 'vVhile (see h is"Management Criteriafor Effective Innovation, .. February, p 'p . 14-23) con-tends that dominant designs canbe recognized in the ear ly stages oftheir de' velopmenr. His analysis .suggests (hat dominant designs willmoreIikely display one or more of the following qualities:

Technologies which lift fundamental technical constraints limit-ing the prior an while not imposing str inge nt new constraints.Designs which enhance the value of potential innovations Inother elements of a product or process.Products which assure expansion into new markers,'r1lj.A, , j .1VJ. U

FOSTERING INNOVATIONBYUNDERST \NSmON .

. place, Each of these hypotheses helps to ex-plain the firm's impetus to divide into homoge-neous productive units as its products and pro-cess technology evolve.

The hypothesized changes in centro!and coordination imply that the structure of theorganization will also change as it matures,becoming more formal and having a greaternumber oflevels of authority. The evidence isstrong tha i t such structural change is a charac-teristic of many enterprises and of units withinthem.

Assuming the validity of this model for the de-velopment of the innovative capacities of a pro:ductive unit, how can it be applied to furtherour capacity for new products and to improveour productivity?

We predict thou units in different stagesof evolution w in respond eo differing stimuliand unde-rtake different types of innovation.This idea c a n readily be extended to the ques-

33



SECTION' Innovation' OViH Time end in Hisrorical Context. .,WHEN TRANSITION IS iNVISIBLE-OR EVEN ABSENT

Idenrifving th e evolutionary transition from product co, process inno-vat ion is sometimes troublesome. In some cases the transition mayhave occurred so rapidly as (0be unrecognized; this appears to be the'case ~;'ith some continuous-flow processes where advanced, elaborate,and large-scale equipment is necessary co make a new product virtu-ally from ir s initial, iruroducrion. Rapid rransition i s a lso ch arac ter isr icof certain products \\fl(llI0,o/ unit values, such as cigarettes and simple'plastic and metal pares, w h ere rh e availab ili ty ora process technologymay have made th ee produce feasible i.n (he first place.

More in(eres( ing cases are th ose where (he rransition fromproduct to process innovation arid from unit production to massproduction, though predicted, has not come about. Examples includehome construction, nuclear power, and sorne other energ)' alterna-r i ves, In each of these examples, experimental programs to stimulatecost reductions, greater standardization, or other aspects oftransitionhave b een undertaken under government and private sponsorship;but none h as had long-run impact. T hese eas,e.s are of special. interestbecause the model may help in identifying barriersand pinpointingappropriate responses.IV./A..j.M. U.

tion of barriers LO innovation; and probably topatterns of success and failure in innovation forunits in different situations. The unrnet condi-lions for rrans ition can be viewed as specificbarriers ..... ich must be overcome if transition is(0 take place.

We would expect new, fiuid units to " , i ' e l . ' as barriers any factors that impede productstandardization and marker aggrega.ion, w hilennns in the opposite cat.egory lend to rank un-cerrairuy over government regula[ion or vui-nerab iliry of existing investments as more irn-portaru disruptive factors. Those .....ho wouldpromote innovation and productivity in U.S.industry may find this suggestive. (See "lVh)'ln-nova l i on s F ail . " b y S um ner lvf)'~ s a nd E ldQ 1! S wec .y ,. IV/arch/Apri l , p p , 40-46.)'

We believe the most useful insights pro-vided by . th e model apply (0 production pro-cesses in which features of the products can be34

varied. The most interesting applications are (0situations where product inn ova. ion is com"perirively important and difficult co manage: (h emodel helps to identify [he full range of otherissues with which (he firm. is simultaneouslyconfronted :in a period of growth and change.

CONSISTENCY OF MANAGEMENTACTIONMany examples of unsuccessful innovationspoint to :a common explanation of fallure: cer-rain conditions necess.ary to support el i sought-after technical advance were not present. Insuch cases our model may be helpful because i tdescribes conditions that normally support ad-vances ar each stage of developm ent; ac cord-ingly. if we can, compare eXlstmg conditions



WifJiam J. Abetflsrhy tJnd James M Utterbeck

with those prescribed by (he model \,'e may dis-cover how [0 increase innovative success. Forexample, we may ask of the model such ques-tions as these about different. apparently iride-pendent, manage ria! actions:

Can a nrl11 increase the variety and diversityof its produce line while simultaneouslyrealizing (he highest possible level of effi-ciency?Is a high rare of product innovation COnSI5['em \,'ith an effort C O substaruially reduceCOStS through extensive backward integra-lion?Is gOI'ernment policy to maintain diversi-fied markets for tech.nologically active in-dustries consistent. with a policy. chat seeks!i high rate of effective product Innovation?W O L ) l d a firm's action to restructure itswork environment for empioyees iso thattasks are mere challenging and less reperi-me be compatible with a policy of rnechani-zation designed to reduce the need forlab o r?Can the government stimulate productivityb y forcing a young industry (0 standardizeits produces before a dominant design hasbeen realized?The model prompts an answer of "no"

CO each of these questions: each question sug-g,es(s actions which the model tells us are rnutu-aUy inconsistent: We believe that as these ideasare further developed they can be equally ef-fective in helping to answer many far more sub-tie questions about [he environment for inno-vation, productivity, and growth.

FURTHER READINGSfor readers, who wish to' explore this subject ingreater derail, the authors recommend,Abernathy. W.J. and P, L. Townsend, "Technology,

Productivity and Process Change," Technological

F or ec as tin g a nd S oc ia ! C ha ng , Vol. 7, : - - ; 0 . -l. Au-gusc. 197,5, pp . 379-396.

Abernathy. W, J. and K. W

Documents

6_ Abernathy - Utterback -Innovation Over Time and in Historical Context