INDUSTRY CHANGE THROUGH VERTICAL DISINTEGRATION: …...INDUSTRY CHANGE THROUGH VERTICAL DISINTEGRATION: HOW AND WHY MARKETS EMERGED IN MORTGAGE BANKING MICHAEL G. JACOBIDES London

INDUSTRY CHANGE THROUGHVERTICAL DISINTEGRATION:

HOW AND WHY MARKETS EMERGEDIN MORTGAGE BANKING

MICHAEL G. JACOBIDESLondon Business School

This paper provides an inductive theoretical framework that explains how and whyvertical disintegration happens, showing that transaction costs are an incidentalfeature of industry evolution. I find that gains from intrafirm specialization set off aprocess of intraorganizational partitioning, which simplifies coordination along partsof the value chain. Likewise, latent gains from trade foster interfirm cospecialization,which leads to information standardization. Given standardized information andsimplified coordination, new intermediate markets emerge, breaking up the valuechain, allowing new types of vertically specialized firms to participate in an industry,and changing the industry’s competitive landscape.

Industry evolution has long been a central con-cern to management scholars and managers alike.Scholars have studied such facets as the life cycleof an industry’s product and process technology(Abernathy & Utterback, 1978; Klepper, 1997),changes in technology and their impacts on firms’success (Henderson & Clark, 1990; Tushman &Anderson, 1987), and adaptation to changing de-mands (Burgelman, 1991; Winter, 1984). Yet oneaspect of industry evolution that has been rela-tively neglected by management scholars, despiteits great importance, is vertical disintegration: theemergence of new intermediate markets that dividea previously integrated production process be-tween two sets of specialized firms in the sameindustry.

What is particularly interesting about verticaldisintegration is that it often happens even whenthe underlying products, services, and core tech-nologies remain the same. Thus, while it may be arelatively invisible part of industry evolution, itcan radically transform the sectors in which it oc-curs. The automobile sector, once the textbookexample of integration, has been on a steady trajec-tory of vertical disintegration, with automotivecompanies giving up parts of the value chain1 tonew specialists and new intermediate marketsemerging (Fine, 1998; Fine & Whitney, 1996).Similar trends are evident in semiconductor man-ufacturing, with the emergence of “fab-less” (non-manufacturing) chip design companies and the cor-

The paper (and the author) benefited greatly from theadvice of a late friend, colleague, and mentor, SumantraGhoshal: He is sorely missed. Also, I would like to thankSid Winter for his extensive feedback; Freek Vermeulenfor constructive suggestions; Mark Zbaracki, Nick Ar-gyres, Carliss Baldwin, Yiorgos Mylonadis, Julian Birkin-shaw, Phanish Puranam, Costas Markides, Peter Moran,Nikolaos Pisanias, Ranjay Gulati, and Madan Pillutla foruseful remarks; and Peter van Overstraeten for editorialassistance. I would also like to thank the AMJ editor, TomLee, and three anonymous AMJ reviewers for the thor-ough and insightful comments and guidance that helpedshape the paper. Doug Duncan provided invaluable sup-port in the industry of interest. Generous funding fromthe Mortgage Bankers Association of America, the Centerfor the Network Economy, and the Leverhulme TrustDigital Transformations Programme at the London Busi-ness School is gratefully acknowledged. The usual dis-claimers apply.

1 A value chain, as defined in this paper, consists ofthe set of value-adding activities that need to be under-taken for a product to be made or a service rendered. Theconcept of the value chain was originally used to denotethe set of activities that take place within the boundariesof any given firm or business unit, including inboundlogistics, production, outbound logistics, marketing, andservice and support functions (see Porter, 1985); itsmeaning has, however, been enlarged and altered re-cently. My definition is consistent with the general use(e.g., Evans & Wurster, 1999; Grant, 2004: Ch. 13) of theterm to mean the structured set of activities that takeplace in an industry, regardless of whether they takeplace within the boundaries of one integrated or manycospecialized firms. Focusing on activities—the tasksthat need to be taken care of—provides an efficient wayof examining how firm and industry boundaries changeand how these changes create different types of “institu-tional packages” along some or all of the activities (or“steps”) of the value chain that are undertaken in a sector.

� Academy of Management Journal2005, Vol. 48, No. 3, 465–498.

465

responding growth of firms that only make, but donot design, their chips (Macher, 1998); the biotechsector provides yet more evidence, with the emer-gence of specialized research firms, specializeddrug-testing firms, and so on (Arora, Fosfuri, &Gambardella, 2000). The story of the minicomputerand then the personal computer (PC) is a furtherexample of an industry’s witnessing not only tech-nological change but also vertical disintegration.While the minicomputer business started off inte-grated, dominated by companies such as DEC andIBM, it soon opened up, and this process becameeven more pronounced with the advent of PCs.Veering far from the relatively integrated structuresof pioneers like Apple, the PC industry today has amultitude of segments, including component man-ufacturers, software and operating system provid-ers, specialized assemblers, resellers, consultantswho integrate products into client companies, andmore (Baldwin & Clark, 2000; Gawer & Cusamano,2002). Once disintegration occurred, the nature ofthe industry, its very definition, and its competitivedynamics were radically transformed, even for theplayers who chose to remain integrated by keepingall the production processes inside.

Such vertical disintegration has profound impli-cations. First, it changes the nature of the firms thatcan participate in an industry. In semiconductors,for example, the newly available vertically disinte-grated structure enabled engineering firms with noproduction capabilities or facilities to become sig-nificant players. In banking, the fact that data pro-cessing became a separate part of the productionprocess—that, through vertical disintegration,could be procured through the market rather thandone inside—led nonfinancial firms such as EDS,IBM, and Accenture to become key participants inthe financial institutions sector. The nature of ca-pabilities, the types of entrants into an industry,and the structure of competition can all changewhen disintegration happens, as the PC industryvividly illustrates (Baldwin & Clark, 2000; Gawer &Cusamano, 2002). To use a biological analogy, ver-tical disintegration allows a new, vertically cospe-cialized ecosystem to compete and partly cooperatewith the old, integrated structures, thus altering thelandscape for all involved (Jacobides & Winter,2005).

Although vertical disintegration can have pro-found effects, scholars know surprisingly littleabout it beyond what can be garnered from anec-dotal evidence or from the popular business press(e.g., Evans & Wurster, 1999). The contribution ofthis paper is its consideration of how as well as whydisintegration happens, focusing on the underlyingdrivers that bring it about.

THEORETICAL BACKGROUND

Existing Research on Vertical Disintegration

Several research strands have indirectly ad-dressed vertical disintegration, although none hasreally directly focused on it. An exception may bethe work of Stigler (1951), who suggested that thesize of a market limits the extent of specialization(which disintegration represents). But althoughmost industrialized countries and most sectorshave the requisite scale to support vertical special-ization outside firms, it doesn’t always occur. Thus,scale is generally not considered to be a good ex-planation for disintegration (see Langlois & Robert-son, 1995). There are almost no systematic studiesof the emergence of vertical disintegration, despitesubstantial research on the social and institutionaldynamics of product-category creation (Carruthers& Babb, 2000; Porac, Thomas, Wilson, Paton, &Kanfer, 1995; White, 1981); substantial research onthe social institutions of market exchange in gen-eral (Carruthers, 1996; Fligstein, 2001); and newresearch on vertical scope—that is, the amount ofvalue-adding steps in the production process that afirm is active in (Jacobides & Winter, 2005).

The main reason for the relative dearth of knowl-edge is that the literature, particularly transactioncost economics (Williamson, 1985, 1999), haslargely focused on a firms’ decisions to “make ver-sus buy” in given transactions. Such analysis isconducted at a subfirm level, in that the units ofanalysis are particular choices made “on the mar-gin” by individual firms.2 It does not look at entireindustries by examining, for example, how marketsemerge to create vertical disintegration. Nevertheless,transaction cost economics may be a useful startingpoint for understanding vertical disintegration.

Transaction cost economics focuses on thechoices an existing firm makes in deciding whetherto make or buy a particular input (Williamson,2000). If transaction costs—that is, the costs of us-ing markets rather than producing in-house—arehigh, then firms choose to make a particular input,thus bypassing the markets by integrating. If trans-action costs are low, firms rely on markets and buy.Transaction costs largely depend on the risk ofopportunistic renegotiation (Williamson, 1975,

2 Note that the literature also does not address howparticular firms’ choices in terms of making or buyingdifferent products and services are intertwined; that is, itdoes not directly address the question of how the bound-aries of one particular firm are set, and how they evolveover time. For an analysis of the logic underpinning thedesign of a particular firm’s vertical structure, see Jaco-bides and Billinger (2005).

466 JuneAcademy of Management Journal

1985, 1999). The risk of opportunism emergeswhen a potential party to a transaction, such as asupplier, needs to make dedicated, long-lasting in-vestments to make the relationship work, invest-ments that will be valuable in that relationship butnot in other contexts. The party undertaking such“asset-specific” (i.e., relationship-specific) invest-ment may be “held up” by the other party (here, thebuyer), who might have promised to compensatefor the investment but will be tempted to break thatpromise. The supplier, knowing this, is not likelyto make such a valuable but risky investment (Hart,1995). As a result, it is impossible for the buyer toget the benefits of such a specialized investment,unless the buyer gives an ironclad guarantee of noopportunism, which, in a dynamic and unpredict-able environment, is not economically feasible. Sothe main way to achieve specialized investments isvertical integration (Williamson, 1985). Thus, ac-cording to this theory, when asset specificity ishigh, and hence transaction costs owing to oppor-tunism are significant, firms will be integrated.When asset specificity is low, firms will not beintegrated.

The question of integration on the industry level,however, is not examined in transaction cost eco-nomics, nor is the option of using markets. Instead,the theory is informed “by the presumption that ‘inthe beginning, there were markets’ ” (Williamson,1985: 87). It begins with the presumption that thechoice of relying on a market exists and explainswhy firms integrate, bypassing the market. It doesnot extend to the development of new markets.Transaction costs theory does not address how in-dustries or value chain structures evolve and doesnot ask the question of whether firms can choosewhether to make or buy. To use a schematic illus-tration from the computer industry, transactioncost economics can explain why a computer man-ufacturer decides to procure a component from themarket, or make it in-house, but it cannot explainthe fundamental industry change from a structureof integrated entities, active throughout the valuechain, to a structure with a distinct market forcomponents that firms can either make or buy (cf.Baldwin & Clark, 2000, 2003). So, given that re-searchers know that many industries start off inte-grated before breaking up into cospecialized pieceslinked through a market (cf. Christensen, Verlin-den, and Westerman, 2002; Jacobides & Winter,2005; Langlois & Robertson, 1995; Stigler, 1951),the lack of a consistent theory and a correspondingbody of empirical evidence on vertical disintegra-tion leaves both empirical and theoretical gaps.

Some related research pertinent to the analysis of

changes in industry scope exists. In their work onthe historical dimension of transaction costs, forinstance, Argyres and Liebeskind (1999) and Mad-hok (2002) suggested that scholars should considerthe specificities of firms’ histories to understandwhat drives their choices of scope. These authorstoo, however, focused on the individual choices ofspecific firms that can either make or buy. Silver(1984) and then Langlois (1992, 2003; Langlois &Robertson, 1989,1995) have argued that transactioncosts are dynamic or even transient and hence thatthey decline over time as potential suppliers andbuyers learn to work together. Christensen and co-authors (2002) argued that disintegrated structuresare “worse” than integrated ones, but that theybecome “good enough” after a while, without, how-ever, explaining why this is so, and, more to thepoint, without explaining how or why vertical dis-integration emerges in the first place. So, while allthis research has correctly pointed out the need fora historically informed analysis of vertical scope, ithas not offered a framework that explains how anindustry’s scope evolves—in particular, how verti-cal disintegration happens—nor has it explainedwhy rapid disintegration happens in some settingsbut not in others.

Theoretical Gaps and the Contribution of thisPaper

The discussion above raises an interesting set ofopportunities. On the empirical side, it is desirableto better document and explain vertical disintegra-tion and market creation, as this process has signif-icant implications for industries and the firmswithin them. On the theoretical side, the need to doso becomes particularly important inasmuch ascurrent theory is ill suited to explaining disintegra-tion. This is the case for three reasons:

First, existing literature has focused on the indi-vidual transaction, as opposed to the changes in anindustry that enable a market to appear, but factorsthat operate on the industry level of analysis areunlikely to be fully reducible to those operating onthe level of the individual transaction.

Second, most of the arguments in transactioncost economics are designed to explain why firmsabandon markets and opt to integrate instead. Onemight simply argue that to understand why marketsemerge, all one has to do is construct the inverse ofthe story, using the same factors to explain disin-tegration. Yet organizational processes and, in par-ticular, processes of institutional change are rarelysymmetric. To wit, a theory of organizationalgrowth cannot be the mirror inverse of a theory of

2005 467Jacobides

organizational decline (cf. McKinley, 1993), be-cause the factors that explain why and how firmsgrow are qualitatively different from the factors thatexplain why and how firms decline. The causallogic relevant to explaining the processes and mo-tivations of growth and decline may differ in itself,so that one cannot use theories of growth to fullyexplain decline, and vice versa. Likewise, the pro-cesses that bring about vertical disintegration arelikely to differ from those that bring about integra-tion: To explain disintegration, a theory must ac-count for what allows a market to be set up in thefirst place. Existing theory has not done this, nor is itsafe to assume that existing theory could be extendedto account for the initial appearance of markets.

Third, most of the existing literature cannot ex-plain how and why transaction costs can be re-duced, as it does not focus on the dynamics of theevolution of an industry. Moreover, even in thelimited research on “dynamic” transaction costs(Langlois, 1992, 2003), there is not much that canhelp explain why transaction costs decrease insome settings and not others. For all these reasons,it seems that scholars may well obtain new theoret-ical insights by studying the process of verticaldisintegration without being entirely bound by ex-isting theory.

It follows from these observations that the bestway to understand the emergence of vertical disin-tegration is to undertake an inductive analysis that,while building on existing literature, allows newanalytical insights on how the intermediate mar-kets resulting from disintegration emerge. To gleansuch insights, I examined an industry that under-went significant vertical disintegration: mortgagebanking in the United States. Whereas until the late1970s mortgage loans were originated, funded, andserviced by integrated financial institutions (banksor savings and loan associations), by the end of thecentury, the industry had become a collection ofvertically cospecialized entities. To understandwhat drove this vertical disintegration, I examinedthe history of the entire industry as it evolved andfocused on the emergence of three new markets—three episodes of vertical disintegration—over a20-year period. Finally, in addition to studyinghow these new markets emerged, I also examinedwhat drove and motivated this process of verticaldisintegration.

The contribution of this study, then, is twofold.First, it provides an analysis of the market emer-gence process that goes beyond the traditional pur-view of transaction cost economics and examinesthe factors that enable the initial vertical break-upin an industry. Second, it explores the drivers of

this process: it analyzes when and why such disin-tegration and new market creation come about. Byshifting the level of analysis from the transaction tothe industry and the firms within it, this studyprovides a template that can explain when and whyvertical disintegration happens.

As this is an inductive study, I first describe themortgage banking industry and introduce the meth-ods. I then present the induced theoretical frame-work and discuss how it relates to existing theory.I conclude by considering the insights offered bythis new framing and theoretical approach on theorganizational drivers of vertical scope as they af-fect research and practice alike.

METHODS

Analysis and Procedures

This paper presents an inductive theory based onan in-depth, qualitative study of the evolution ofthe mortgage banking industry. The level of analy-sis in the study was the entire value chain of oneindustry. I chose this approach in order to portrayboth the process and the context of change alongwith their interconnections over time (Mohr, 1982;Pettigrew, 1990). As process research, this studyfocused on understanding the causal dynamics of aparticular setting (Mohr, 1982). The central as-sumption was that “causation is neither linear norsingular . . . [hence] explanations of change arebound to be holistic and multifaceted” (Pettigrew,1990: 269). The causal relations derived here areexplained in detail in the main body of the paperand are closely linked to the evidence, so as toallow their internal validity to be assessed (Yin,1994: 33).

This setting was chosen on conceptual grounds,rather than for representativeness (Miles & Huber-man, 1994: 27). The aim was to select a setting thatwould allow isolation of the central concepts per-taining to the research questions. Mortgage bankingwas thus selected because it underwent a processof disintegration and market creation, and in par-ticular because it witnessed several episodes of ver-tical specialization. Furthermore, the decision wasinfluenced by the fact that the product/service of-fered, mortgages, had not changed substantially inthe period of this study. In other settings that Ievaluated (e.g., semiconductors, pharmaceuticals,auto manufacturing), the process of value chainand industry change was often entangled with sig-nificant changes in the product provided. Avoidingthese complications enabled me to identify the ma-


jor underlying drivers of market creation for a givenproduction sequence.3

This study is based on multiple sources of evi-dence: archival data, industry publications andmanuals, interviews, and direct observation. Al-though it covers a 20-year period of vertical disin-tegration, my direct observation in the field lastedabout 38 months. My primary objective was to cre-ate an accurate depiction of the evolution of theindustry’s value chain, something that did not existwhen I went to the field.4 To do so, I used manysources of evidence: archival data, qualitative evi-dence, and corroborating quantitative evidence; thelatter was used to create a synthetic map of theindustry structure as it evolved.

In terms of data reliability, the study conformswith Yin’s (1994: 32) focus on developing constructvalidity and with Guba and Lincoln’s (1982) em-phasis on confirmability. The theory is built uponrepresentations of phenomena. It is essential forsuch representations to be accurate and meaning-ful; that is, they must reflect the shared views ofindustry participants and observers. I used multi-ple sources of evidence to ensure an accurate rep-resentation of the industry’s evolution. The archi-val data were particularly useful in reconciling thediverse views I obtained through my interviews,which are described below. Thus, whenever I en-countered a new idea or view, I would test itagainst the archival data and reports of the time. Ifdata did not exist, and I had discrepant informa-tion, I tried to determine if different constituenciesin the mortgage sector accepted a particular view orargument. If there was disagreement, I would mod-

ify the argument until a reasonable amount ofagreement could be reached and, failing that,would consider the argument as unsupported.Methodologically, my objective was to ensure I hadan accurate depiction of the industry—of what hadhappened—and to provide an interpretation,through inductive theorizing, of what broughtabout disintegration.

In keeping with Pettigrew’s (1990) directives forinductive, case-based research, while I did ap-proach the organizational field of interest with the-oretical constructs in mind, I did not impose them.Instead, I considered how the detailed evidencegathered in the field might inform existing theoryor constructs, such as transaction costs. To do so, Iexamined how the data informed my understand-ing of (1) the vertical disintegration and marketcreation process and (2) what motivates and drivesit. Then, working within the emerging theoreticalframework, I reconsidered the data and clarifiedparticular issues, which led me to refine the devel-oping theory, and so on. As my theory developed, Iwould communicate it to key industry participants,and in particular the chief economist of the Mort-gage Bankers Association of America, with whom Iworked closely in the project. Thus, I used an iter-ative process of theory development and data anal-ysis (cf. Eisenhardt, 1989, 1991) that led to theframework proposed in this paper. (The next mainsection of this article, “The Setting: Mortgage Bank-ing and the Genesis of Three Sets of IntermediateMarkets,” describes the process of discovery andthen validation of my framework.) As the evidencewas historical, archival evidence was given dueattention. The archival data I used included indus-try manuals, each issue of the monthly publicationMortgage Banking from 1968 onwards, the manualsfor the School of Mortgage Banking (see Lederman,1985, 1995), all mortgage-related bulletins of theFederal Reserve, earlier studies of the industry(e.g., Tuchman, 1986), annual reports of major in-stitutions, regulations and congressional acts re-lated to the sector, all academic research on thesector that I could locate, analysts’ reports (espe-cially from Morgan Stanley, which covers the sec-tor extensively), and other material described below.I also gathered industry statistics and demographics,especially from the Housing Mortgage Disclosure Act(HMDA) database of the Department of Housing andUrban Development, the industry reports of theMortgage Bankers Association (including its an-nual cost and servicing studies), and data fromprivate consultancies, such as Inside Mortgage Fi-nance and Wholesale Access, as well as confiden-tial data assembled by two consultancies.

3 This kind of purposive choice may raise questionsabout the external validity of the theory (Yin, 1994), orthe extent to which the insights from this study aretransferable to other settings (Guba & Lincoln, 1982).However, the usefulness of generalization from this typeof data is analytical rather than statistical (Firestone,1993). While claiming no statistical significance for thesefindings, I suggest that choosing a setting that allowedisolation of key theoretical constructs makes the theoret-ical findings applicable to other industry settings andprovides a grounded theoretical framework that can betested in other settings in subsequent research.

4 As there was no existing theory, either in the indus-try or in academe, on the drivers or even the nature ofdisintegration, my investigation required an intensiveinvolvement so that the basic value chain layout could bederived. I thus had to become familiar with the technicaldetail of the industry, so as to “speak the language” of theparticipants in the various bits of the value chain. Thequestions that I would then ask would not be couched inmy “native” terminology, but rather in that of industryparticipants.

2005 469Jacobides

TA

BL

E1

Sou

rces

and

Act

ivit

ies

for

Pro

ject

Sta

gesa

Ch

arac

teri

stic

sof

Sta

geS

tage

1:A

ugu

st19

98–D

ecem

ber

1998

Sta

ge2:

Jan

uar

y19

99–A

pri

l20

00S

tage

3:M

ay20

00–N

ovem

ber

2001

Obj

ecti

ves

Un

der

stan

din

gin

du

stry

stru

ctu

res

and

pro

cess

esE

xplo

rin

gev

olu

tion

ofin

du

stry

stru

ctu

rean

dm

arke

tcr

eati

onB

ecom

ing

fam

ilia

rw

ith

ind

ust

ryp

arti

cip

ants

Iden

tify

ing

key

par

tici

pan

tsfo

rsu

bseq

uen

tin

terv

iew

s

Exp

lori

ng

inm

ore

det

ail

the

spec

ific

sof

ind

ust

rych

ange

s:p

rod

uct

s,te

chn

olog

y,p

laye

rs,

and

stru

ctu

reF

inal

izin

ga

map

ofth

ein

du

stry

and

its

evol

uti

onC

reat

ing

am

apw

ith

dri

vers

ofve

rtic

ald

isin

tegr

atio

n

Fin

aliz

ing

map

pin

gof

the

mec

han

ics

and

cau

ses

ofin

du

stry

evol

uti

onan

dm

arke

tcr

eati

onC

omp

lem

enti

ng

the

fiel

dev

iden

ceR

esol

vin

gre

mai

nin

gd

iscr

epan

cies

Inte

rvie

ws

18in

terv

iew

sw

ith

exec

uti

ves

from

ind

ust

ryre

gula

tors

and

over

sigh

tag

enci

es,

incl

ud

ing

the

MB

AA

,N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Ban

kers

,H

UD

,O

FH

EO

,F

ann

ieM

ae,

and

Fre

dd

ieM

ac

84in

terv

iew

sw

ith

ind

ust

ryp

arti

cip

ants

,in

clu

din

gba

nke

rs,

brok

ers,

serv

ice

and

tech

nol

ogy

pro

vid

ers,

tech

nol

ogy

pro

vid

ers,

ind

ust

ryan

alys

ts,

fin

ance

con

sult

ants

,an

dre

gula

tors

27in

terv

iew

sw

ith

MB

AA

sen

ior

exec

uti

ves

and

mor

tgag

eba

nke

rsF

ollo

w-u

pp

hon

eca

lls

and

shor

td

iscu

ssio

ns

wit

hm

any

info

rman

ts

32in

terv

iew

sw

ith

exec

uti

ves

from

mor

tgag

eba

nks

,in

du

stry

anal

ysts

,an

dco

nsu

ltan

tsS

econ

dar

yso

urc

esH

isto

rica

lst

ud

ies

Com

pan

ym

anu

als

His

tori

cal

acco

un

tsS

ecto

rd

escr

ipti

ons

Tra

de

arti

cles

Com

pan

ym

anu

als

Tra

de

pu

blic

atio

ns

Pre

ssre

leas

esT

rad

ep

ubl

icat

ion

sH

isto

rica

lac

cou

nts

ofse

ctor

evol

uti

onE

ven

ts19

98M

BA

AN

atio

nal

Con

ven

tion

Tw

o-w

eek

cou

rse

for

mor

tgag

eex

ecu

tive

s19

99M

BA

AN

atio

nal

Con

ven

tion

Tw

oM

BA

Ain

du

stry

mee

tin

gsT

hre

ein

du

stry

rou

nd

tabl

es

Pre

sen

tati

onof

init

ial

fin

din

gsto

ake

ynot

ep

anel

ofth

e20

00M

BA

AN

atio

nal

Con

ven

tion

Pre

sen

tati

onof

fin

din

gsto

the

2001

MB

AA

CE

Oco

nfe

ren

ce

aT

he

MB

AA

isth

eM

ortg

age

Ban

kers

Ass

ocia

tion

ofA

mer

ica.

HU

Dis

the

U.S

.off

ice

ofH

ousi

ng

and

Urb

anD

evel

opm

ent.

Th

eO

FH

EO

isth

eO

ffic

eof

Fed

eral

Hou

sin

gE

nte

rpri

seO

vers

igh

t.

Data Collection

The study was underwritten by the MortgageBankers Association of America (MBAA) and theWharton Financial Institutions Center, a researchcenter sponsored by the Sloan Foundation. It lasted38 months and proceeded in three stages; Table 1summarizes the evidence examined, analyzed, andused in these three stages.

The first stage of the research took place betweenAugust and December 1998 and focused on identi-fying the key industry participants and getting agood idea of the industry and its evolving structure.I conducted 18 interviews with industry regulatorsand oversight agencies (such as the Department ofHousing and Urban Development and the Office ofFederal Housing Enterprise Oversight), as well asthe major government-sponsored “securitizers”(Fannie Mae, Freddie Mac, and Ginnie Mae.5 Theseregulator interviews were followed by 32 semi-structured interviews with executives who hadbeen in the field for a long time and who were ableto provide an overview of the sector and its evolu-tion. The evidence was complemented with docu-ments from secondary sources, including the fewexisting historical studies and descriptions of thesector, and trade publications. During this stage Ialso participated in the 1998 MBAA NationalConvention.

The second stage of the fieldwork and archivalresearch was conducted between January 1999 andApril 2000. During this period, I delved moredeeply into the history of the sector to explore thespecifics of its evolution in more detail, looking atits products, technology, players, and industrystructure. The ongoing support of the MBAAproved particularly helpful at this stage, as it pro-vided me with critical contacts and in-house data. Ifirst attended a two-week course for mortgage ex-ecutives who were either from outside of theUnited States or new to the sector; this course fur-ther familiarized me with the way the industry isorganized in the U.S. and abroad. I then conducted84 semistructured interviews with industry partic-ipants and consulted a large number of articles andmanuals, as well as all the published historical

accounts of the sector. In this stage, I attended the1999 MBAA National Convention and two moreindustry meetings, where I interacted formally andinformally with industry participants. I was alsoallowed to sit in on three industry roundtables or-ganized by the MBAA. From the evidence collectedduring the first two stages, I created a preliminarymap of the industry and its evolution, which waschecked by the chief economist of the MBAA.

The third stage of the research project, conductedbetween May 2000 and November 2001, focused onsolidifying the view of the industry and its evolu-tion, as well as resolving any remaining discrepan-cies and complementing the field evidence. I con-ducted 27 interviews with MBAA senior executivesand mortgage bankers. I also conducted follow-upinterviews with many informants and had numer-ous phone calls and short discussions to confirminformation and fill in gaps. I also communicatedthe basic findings of the research and received ex-tensive feedback on the validity and accuracy of mydescription. This stage culminated in an industrypublication (Jacobides, 2001), which I circulated toseveral executives ahead of time to solicit theirinput.

Throughout the study, I selected interviewees soas to maximize the variety of profiles and hetero-geneity of perspectives in the industry. I choserespondents from different-sized firms, from firmswith vastly different strategies and scope, and fromfirms with various roles (technology, origination,servicing, and so forth). I selected several partici-pants who were not mortgage bankers but wereservice or technology providers, analysts, regula-tors, or consultants. The level of seniority also var-ied, although few interviewees had less than fiveyears of experience in the industry; Figures 1 and 2provide summary statistics of the interviewees forall three stages. The focus of the discussion de-pended on the profile of the individual, and I con-centrated on the respondents’ fields of expertise

5 To “securitize” mortgage loans means to take a set ofrelatively homogeneous loans, bundle them together, andthen create a “security,” which is an instrument that canbe sold (much like a bond) though the capital market.The “securitizer” is the entity buying loans (from com-panies that have issued them to individuals); then creat-ing the infrastructure for making the security; and thenselling the bundled loans to the capital market. SeeFabozzi and Modigliani (1992) or Nadler et al. (1992).

FIGURE 1Interviewees’ Tenures in the Mortgage Sector

2005 471Jacobides

and activity, except when they volunteered to pro-vide their broader assessment.6 Major issues andthreads that emerged in the interviews were incor-porated in subsequent interviews. I undertook all ofthe interviews. Some, but not all, were taped, as theinterviewees were often uncomfortable with taperecordings. For interviews that were not recordedand transcribed, I kept notes, which I consultedagain after the interviews and as I was preparingthe account of industry evolution. Finally, becauseinterviewees were concerned about maintaininganonymity, I sometimes use published sources forquotations, rather than primary evidence, to sup-port my conclusions.

The Setting: Mortgage Banking and the Genesisof Three Sets of Intermediate Markets

A mortgage is a loan collateralized by real estate.To make such a loan possible, a lender, as a conduitfor those with excess funds, must be able to find aborrower who needs a mortgage. Roughly speaking,to make a mortgage possible, either one integratedfirm or a series of vertically cospecialized firmslinked though the market must ensure that the fol-

lowing happens: (1) lenders with excess funds arefound; (2) borrowers willing to take a loan arefound and steered to the appropriate loan type; (3)borrowers are analyzed for creditworthiness, thevalue of their collateral, etc., and are guidedthrough the paperwork and procedures required toinsure titles and deeds and meet all other legalrequirements; (4) the loan is closed, and the trans-action consummated and recorded; (5) the loan isserviced for the duration of its length, which meansreceiving payments from the borrower and manag-ing the account until it is paid off or, alternatively,foreclosing if payments are not received; and (6)payments are made to the lenders or other provid-ers of capital.

These six different functions were originally per-formed in integrated institutions, in particular, inretail banks, which maintained some mortgageloans, or in savings and loan associations (S&Ls),which were primarily focused on mortgage loans.7

The short-term deposits of retail customers andlending from the capital market provided liquidityfor both of these integrated types of firms, allowingthem to fund their loans (Fabozzi & Modigliani,1992; Lederman, 1985). Banks and S&Ls sought outmortgage loan applicants, prepared and processedapplications, and serviced the loans until they ex-6 The four main question areas were (1) How has your

segment evolved? What have been the main differencesthat you observed? Why did these changes occur at thetime that they did? (2) Was there any change in the typesof activities that your segment performs? (3) Has therebeen any change in the scope of activities in your part ofthe mortgage industry? Why did any such changes occurat the time they did? and (4) (if the participants identifiedthe trend toward disintegration), Can you identify whythere has been such a tendency? What factors led to thisbreakup? Why did it not happen before?

7 This study focuses on the evolution of residentialmortgages. The commercial mortgage industry’s struc-ture and evolution is broadly similar, although verticaldisintegration has been less pronounced. The differenceis largely a result of the comparative difficulty of stan-dardizing information (one of the issues I focus on in myinductive framework).

FIGURE 2Backgrounds of Interviewees


pired. This outline captures the earlier, integratedversion of the industry.

Mortgage banking started vertically disintegrat-ing, creating new markets, in the early 1970s. Spe-cialized institutions, each with a narrow verticalscope, began to perform the functions noted above.While the functions and the basic steps in the pro-duction of a loan did not change, the vertical struc-ture of the industry did. With it, the types of indus-try participants and the nature of competition alsochanged. By the mid-1990s, each function could beperformed by a vertical specialist. Mortgage brokersfound mortgage borrowers and steered them to theappropriate loans. Mortgage banks focused on clos-ing (finalizing) loans, funding them, and then ser-vicing them. They held no deposits, nor did theyseek funding for the loans through the capital mar-kets. Instead, a mortgage bank would “warehouse”a loan until it could sell the loan to new specialists,the securitizers, who would then find the lenders tofund it over the long term. To fund loans tempo-rarily, mortgage banks would use lines of credit,working capital obtained from commercial banks toenable loan warehousing. Then, securitizers, hav-ing purchased individual loans from several differ-ent mortgage banks, would bundle loans togetherand then turn them into securities (unbeknownst tothe borrowers whose loans were being securitized)and sell these mortgage-backed securities to thecapital markets, earning fees on the securities theyproduced (Fabozzi & Modigliani, 1992). Later,some specialized mortgage banks began to focusmore on servicing, while others focused more onoriginating loans. This stunning vertical specializa-tion, depicted in Figure 3, meant that new interme-diate markets had arisen to accommodate interfirmtrade: Mortgage brokers would sell their loan leadsto mortgage banks; mortgage banks would sell theassets (the funded loans) to the securitizers, whileretaining the right to service the loans; and later,some mortgage banks would even trade the rights toservice particular loans.

More specifically, between 1970 and 1990, therewere three major, distinct episodes of vertical dis-integration. The first one was the creation of themortage securitization sector; the second one wasthe creation of the mortgage broker segment; andthe third one was the creation of the market formortage servicing. Each of these episodes consistsof the creation of an entirely different market, in adifferent part of the value chain, each with its ownset of vertically cospecialized buyers and sellers;thus, the analysis of this industry rests on threeinstances of vertical disintegration, rather than one,as Figure 1 indicates. A brief explanation of eachepisode follows.

Episode 1: Loan ownership and loan origina-tion and servicing are separated. The first episodeof disintegration in the industry was securitization,which enabled the growth of mortgage banks (i.e.,nondepository financial institutions.) Althoughmortgage banks existed prior to securitization, theirrole was different, their scope limited, and theirmarket share very small. They were used to gener-ate and service mortgage loans that were held onthe books of insurance companies or the federalgovernment (which wanted to subsidize particularneedy groups). The big change in the housing fi-nance industry happened at the end of the 1960sand the early 1970s. Because of shortages in hous-ing finance, especially in the quickly developingSunbelt (Florida, Texas, and California), the gov-ernment sought additional means of providinghousing finance. Limits in the federal budget madeit attractive to rely on capital markets, as opposedto government funds, to support the availability ofhousing finance (Lederman, 1985; Tuchman, 1986).Thus, by 1970, the United States had created threegovernment-sponsored enterprises, Fannie Mae,Freddie Mac, and Ginnie Mae, to facilitate the pro-vision of mortgage finance. Mortgage bankers is-sued and serviced loans according to particularcriteria on behalf of these enterprises. In 1972, Gin-nie Mae created the “mortgage-backed security”: itcreated a pool of loans, sold shares in these pools,and insured the pool against default (a defaultwould be treated as a prepayment). Henceforth, assecuritizers the three government agencies wouldspecify in broad terms what loan bundles theywanted; they would then buy them from mortgagebankers; and then they would package the loansthey had bought into new bundles and sell these tothe capital markets. Thus, an alternative to the in-tegrated S&Ls and banks came about; this newmode relied on capital markets for funds, on secu-ritizers for access to funds and relationships withlenders, and on mortgage bankers for the origina-tion of loans, their servicing, and the payment tosecuritizers (Baldwin & Esty, 1993; Fabozzi &Modigliani, 1992).

The government, having set up the government-sponsored enterprises, soon reduced its involve-ment, providing them with public company char-ters. (Fannie Mae was listed on the New York StockExchange between 1970 and 1972; Freddie Macwas listed in 1989). Although the initial loans thethree agencies securititized were government-spon-sored loans of the Federal Housing and VeteranAuthorities, from 1972 onwards they focused onloans made by mortgage banks, which were not partof any government initiative (Fannie Mae, 1992).The market had taken off, with the government-

2005 473Jacobides

FIGURE 3The Empirical Context: The Disintegrating Mortgage Banking Sector


sponsored enterprises competing with other secu-ritizers (especially Salomon Brothers, LehmanBrothers, and First Boston). It is important to notethat the government’s intervention was limited andlargely focused on the creation of the template thatenabled the market to emerge (Department of Hous-ing and Urban Development, 1996; Tuchman,1986). Once that template was set, this disinte-grated “ecosystem” took off, as shown in Figure 4,which shows the growth of the mortgage-bank-cum-securitizer model through the growth of secu-ritized loans. In this particular episode, then, theU.S. government was market-augmenting, to useOlson’s (2000) term. Its initiative and resulting reg-ulatory framework enabled a new market betweenprivate firms to take off, and as such prompted theemergence of a new mode of organizing.

Episode 2: Loan origination is separated intobrokerage and warehousing. The next instance ofvertical disintegration happened within mortgagebanks themselves, with no intervention from gov-ernment. Whereas until the 1980s, mortgage banksalways generated their own loans, this soonchanged. A new, vertically specialized segmentemerged—mortgage brokers, who would seek, qual-ify, and educate customers and prepare mortgageloans. They would then sell the mortgage loan ap-plications to the mortgage banks; mortgage bankscould stop worrying about seeking customers andfocus on selecting loans, which they would keeptemporarily (warehouse), sell to the securitizers,and service.

The new mortgage broker segment (and the cor-responding intermediate market) emerged partly asa result of the recession of 1979–81, when bankslaid off loan origination staff but maintained someflexible arrangements with former employees.When the business cycle improved, in 1982, these

arrangements became more permanent (Garrett,1989). A newly independent segment of brokers,generally former loan officers who would identify aloan customer and prepare the “loan lead,” startedformally transacting with banks. As this segmentgrew, a number of mortgage banks began reducingtheir in-house origination staff and using the mar-ket to procure the loans, restricting themselves tothe wholesale side of the business. By 1997, theshare of this market-based arrangement betweenmortgage bankers and the new species of brokershad reached 63 percent of the total volume of loansproduced (LaMalfa, 1990), as Figure 5 shows.

Episode 3: Loan servicing is separated fromorigination. In the final episode of market creation,servicing was separated from origination, throughthe market for mortgage-servicing rights. Until thelate 1980s, a mortgage bank would originate a loan,sell it to securitizers, but retain its servicing, whichbrought in valuable fees. There were only a fewsporadic purchases of mortgage loan portfolios andno real specialization in servicing or origination.The step-change toward the creation of the marketbegan in 1989, when the Resolution Trust Corpora-tion started selling both the loans and the servicingrights of S&Ls that had failed due to adverse eco-nomic conditions (Baldwin & Esty, 1993). To do so,the Resolution Trust had to find a way of pricingservicing rights, which it did. After these emer-gency sales were finished, however, the sales ofmortgage-servicing rights did not stop: as the tem-plate for a market exchange had been set up, firmsthat wanted to specialize vertically and becomemostly originators or mostly servicers used thatmarket to do so.

In the first episode of market creation, mortgagebanks, or nondepository institutions that relied onsecuritizers to generate loans, managed to chal-

FIGURE 4Securitized Loans as Share of Total Loans,

by Valuea

a Source: Federal Reserve Bulletins.

FIGURE 5Loan Originations by Brokers as Share of Total

Loan Volumea

a Sources: Inside Mortgage Finance and LaMalfa (1991, 1998).

2005 475Jacobides

lenge the vertically integrated banks and S&Ls. Inthe second and third episodes of market creation,even mortgage banks themselves disintegrated.These three episodes transformed mortgage bank-ing. To put it in the words of a seasoned industryexecutive, in a feature article in the main industrypublication:

The question begs to be asked: “Are there any realmortgage bankers left?” The term “mortgage banker”has, until recent years, referred to . . . a fully inte-grated loan origination company. . . .This scenariosounds like ancient history today. . . . Loan origina-tion has shifted to the small mortgage company orloan brokerage. . . . [Servicing also became sepa-rated from origination and thus] CEOs look at ser-vicing growth as just a “make or buy” decision.(Jacobs, 1993: 23–24)

Despite disintegration of the value chain in mort-gage banking, however, the product remainedroughly the same, and the sequence of activitiesneeding to be done did not change dramatically.This stablility ensured the possibility of isolatingthe theoretical issue at hand—that is, the changesin the division of labor within a given industry, fora given sequence, between different institutionalarrangements—without the possibly confoundingeffects of product innovation or other structuralchange. Also, the appearance of intermediate mar-kets does not mean that all of the activities thatused to be coordinated through integrated firmswere now coordinated through markets, thoughsome were. In the mortgage banking industry, thebattle between different ecosystems (that is, thecospecialized, market-mediated mortgage-banking-cum-securitizers’ world and the integrated, tradi-tional world of S&Ls and lending banks) has goneon for a long time. The remainder of this paperfocuses on what enabled the new, vertically spe-cialized ecosystem to come about in the first place.

RESULTS: AN INDUCTIVE THEORY OFMARKET EMERGENCE

In analyzing the case evidence during stage 2 ofthe project, I considered whether some of the es-tablished theoretical constructs at hand, such asopportunism and asset specificity, were useful pre-dictors of market creation. The evidence did notpoint to a significant reduction in asset specificity,however, at least not to an extent that could justifythe extensive creation of markets. Consider, for in-stance, the first market creation, securitization. Noreal issues of hold-up or cospecialization were re-solved over time. If anything, the mortgage bankscurrently face huge risks, as their entire operations

depend on only two major players (and a handful ofsmaller ones) who “set the rules” and can in theoryopportunistically renegotiate, leaving the verticallycodependent mortgage banks with few alternatives.What is even more puzzling from a transaction costeconomics perspective is that most of the mortgagebanking firms usually work with only one of thesesecuritizers, thus creating real dependencies. Sothe new, securitized system appears to create more,not less, potential for transaction costs along thesame value chain, and this makes the emergence ofmarkets harder to explain.

A second puzzle is that, as Eccles and White(1986) noted, integration does not always resolvethe problems of opportunism. For instance, whenasked about the dangers of using the market (bro-kers as opposed to in-house retail branches), a se-nior vice president of a mortgage bank said:

I don’t think this is a major issue. Don’t get mewrong—you do get these lemons you talked abouteverywhere. But I don’t think you would choosehow to procure your loans on the basis of defaultrisk or fraud. You could have bad apples in yourown retail branches—loan officers who make a kill-ing and then go—as you could have greedy brokersor untrustworthy correspondents. I guess that if youtrust fly-by-night brokers you can get in trouble, butyou can monitor these things.

This statement might be interpreted as a result ofall problems of hold-up having been mitigated inacross-firm relationships. Yet I did not find muchdirect or indirect evidence for reductions in assetspecificity or opportunistic behavior. This absenceof evidence did not prompt me to discard transac-tion cost theory, but it did prompt a wider reach forother related explanations, especially in view ofpotential asymmetries between the drivers of inte-gration (given the existence of a market) and thedrivers of disintegration, when a market does notexist and needs to emerge.

Through the iteration of data and theory de-scribed in the previous section, I developed mymodel, summarized in Figure 6, which concernshow markets emerge and what motivates this pro-cess. Specifically, I argue that two motivating fac-tors ultimately drive disintegration: The first isgains from specialization, which result when thereare differences in the requisite managerial styles orknowledge bases along the value chain (for in-stance, retail loan production versus loan ware-housing). The second is latent gains from trade,which emerge whenever there are capability differ-ences between specific firms in an industry, orwhen firms can expand only one part of the valuechain, making reliance on the market desirable.


These motivators in turn put in motion two respec-tive enabling processes: Inasmuch as there aregains from specialization, firms engage in intraor-ganizational partitioning; that is, they create clearadministrative separations along the value chain.

Likewise, inasmuch as there are gains from trade,firms engage in a process of vertical cospecializa-tion; that is, they try to find ways to reduce trans-action costs and seek templates for exchange acrossfirm boundaries. As a result of these two processes,

FIGURE 6What Enables Vertical Disintegration and Market Creation to Come About?

2005 477Jacobides

TA

BL

E2

Sou

rces

ofE

vid

ence

for

Iden

tify

ing

and

Val

idat

ing

the

Dri

vers

ofD

isin

tegr

atio

nan

dM

ark

etC

reat

ion

a,

b

Ep

isod

ean

dD

ata

Sou

rce

Gai

ns

from

Tra

de

and

Sp

ecia

liza

tion

Par

titi

onin

gan

dC

osp

ecia

liza

tion

Info

rmat

ion

Sta

nd

ard

izat

ion

Coo

rdin

atio

nS

imp

lifi

cati

on

Ep

isod

e1:

Sec

uri

tiza

tion

Evi

den

ce,s

tage

1;V

alid

atio

n,s

tage

2E

vid

ence

,st

age

2;V

alid

atio

n,

stag

es2

&3

Evi

den

ce,

stag

e1;

Val

idat

ion

,st

ages

1&

2E

vid

ence

,st

age

2;V

alid

atio

n,

stag

es2

&3

Inte

rvie

ws

Fed

eral

Boa

rdof

Res

erve

s;W

all

Str

eet

exec

uti

ves;

fin

ance

con

sult

ants

Wal

lS

tree

tex

ecu

tive

s;M

BA

A;

fin

ance

con

sult

ants

HU

D;

Wal

lS

tree

tex

ecu

tive

s;M

BA

A;

fin

ance

con

sult

ants

Wal

lS

tree

tex

ecu

tive

s;ac

adem

ics;

fin

ance

con

sult

ants

(wea

ksu

pp

ort)

Pu

blic

reco

rds

Aca

dem

icre

sear

chon

secu

riti

zati

on(e

.g.,

All

en&

San

tom

ero,

1997

;F

olla

in&

Zor

n,

1990

)

Des

crip

tion

sof

the

evol

uti

onof

secu

riti

zati

onin

mor

tgag

eba

nki

ng

(e.g

.,T

uch

man

,19

86)

Tec

hn

ical

anal

ysis

onse

curi

tiza

tion

;ac

adem

icre

sear

ch(F

aboz

zi&

Mod

igli

ani,

1992

;N

adle

ret

al.,

1992

);in

du

stry

pu

blic

atio

ns

(Sle

sin

ger,

1994

)

Aca

dem

icre

sear

chon

S&

Lch

ange

s(B

ald

win

&E

sty,

1993

)

Ep

isod

e2:

Bro

ker

mar

ket

crea

tion

Evi

den

ce,s

tage

1;V

alid

atio

n,s

tage

2E

vid

ence

,st

age

1;V

alid

atio

n,

stag

e2

Evi

den

ce,

stag

e1;

Val

idat

ion

,st

ages

1&

2E

vid

ence

,st

age

1;V

alid

atio

n,

stag

es1

&2

Inte

rvie

ws

MB

AA

;N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Bro

kers

;go

vern

men

t-sp

onso

red

ente

rpri

ses;

brok

ers;

ban

kers

;co

nsu

ltan

ts;

anal

ysts

MB

AA

;N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Bro

kers

;br

oker

s;ba

nke

rs;

S&

Lex

ecu

tive

s;co

nsu

ltan

ts

MB

AA

;N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Bro

kers

;br

oker

s;ba

nke

rs;

con

sult

ants

;H

UD

;te

chn

olog

yp

rovi

der

s

MB

AA

;N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Bro

kers

;br

oker

s;ba

nke

rs;

con

sult

ants

;te

chn

olog

yp

rovi

der

sP

ubl

icre

cord

sIn

du

stry

pu

blic

atio

ns

onm

ortg

age

ban

kin

g(L

eder

man

,19

85,

1995

;O

’Don

nel

&D

ivn

ey,

1994

)an

don

brok

erse

gmen

t(e

.g.,

LaM

alfa

,19

90,

1998

;P

rop

per

&F

uri

no,

1994

)

Art

icle

son

chan

ges

inth

ese

ctor

(e.g

.,Ja

cobs

,19

93);

ind

ust

ryp

ubl

icat

ion

son

brok

er/b

anke

rre

lati

ons

(Gar

rett

,19

89,

1993

;L

aMal

fa,

1998

;W

iser

&L

aMal

fa,

1991

)

Ind

ust

ryp

ubl

icat

ion

son

auto

mat

ion

and

brok

erag

e(F

oste

r,19

97;

Leb

owit

z,19

95;

Sch

nei

der

,19

94);

anal

yst

rep

orts

(Mor

gan

Sta

nle

yD

ean

Wit

ter,

2000

)

Ind

ust

ryp

ubl

icat

ion

son

brok

er-

ban

ker

rela

tion

san

dau

tom

atio

n(F

oste

r,19

97,

Gar

rett

,19

89;

Pro

pp

er&

Fu

rin

o,19

94;

Mor

gan

Sta

nle

yD

ean

Wit

ter,

2000

);in

du

stry

man

ual

s(e

.g.,

Led

erm

an,

1985

,19

95)

Ep

isod

e3:

Ser

vici

ng

righ

tsm

ark

etE

vid

ence

,st

age

1;V

alid

atio

n,

stag

es1

&2

Evi

den

ce,

stag

e1;

Val

idat

ion

:N

osi

gnif

ican

tsu

pp

ort

Evi

den

ce,

stag

e1;

Val

idat

ion

,st

age

2E

vid

ence

,st

age

1;V

alid

atio

n,

stag

e2

Inte

rvie

ws

MB

AA

;H

UD

;ba

nke

rs;

con

sult

ants

;an

alys

tsM

BA

A;

ban

kers

;an

alys

tsM

BA

A;

ban

kers

;M

ortg

age

Ele

ctro

nic

Reg

istr

atio

nS

yste

mex

ecu

tive

s;co

nsu

ltan

ts

MB

AA

;N

atio

nal

Ass

ocia

tion

ofM

ortg

age

Bro

kers

;br

oker

s;ba

nke

rs;

con

sult

ants

(wea

ksu

pp

ort)

Pu

blic

reco

rds

An

alys

tre

por

ts;

arti

cles

onse

rvic

ing

(O’D

onn

el&

Div

ney

,19

94;

Wis

er&

LaM

alfa

,19

91)

Art

icle

son

ind

ust

ryu

nbu

nd

lin

g(G

arre

tt,

1993

)A

rtic

les

onse

rvic

ing

mar

ket

evol

uti

on(e

.g.,

Mar

a,19

89;

Wis

er&

LaM

alfa

,19

91)

No

dat

a

a“B

anke

rs”

are

mor

tgag

eba

nke

rs;“

brok

ers”

are

mor

tgag

ebr

oker

s;an

d“a

nal

ysts

”ar

ein

vest

men

tba

nk

anal

ysts

.MB

AA

isth

eM

ortg

age

Ban

kers

Ass

ocia

tion

ofA

mer

ica;

HU

Dis

the

U.S

.Dep

artm

ent

ofH

ousi

ng

and

Urb

anD

evel

opm

ent,

wh

ich

incl

ud

esth

eO

ffic

eof

Fed

eral

Hou

sin

gE

nte

rpri

seO

vers

igh

t(O

FH

EO

).In

terv

iew

sw

ere

wit

hem

plo

yees

ofth

ese

vari

ous

agen

cies

.b

Th

est

age

ofth

ere

sear

chin

wh

ich

init

iale

vid

ence

was

fou

nd

isn

oted

,an

dth

est

age

atw

hic

hva

lid

atio

nw

asob

tain

ed.S

tage

1w

asa

5-m

onth

per

iod

ofex

plo

rin

gm

ortg

age

ban

kin

g,an

dst

age

2w

asa

15-m

onth

per

iod

ofm

app

ing

the

pro

cess

esof

inte

rest

.Fin

ally

,sta

ge3

enta

iled

pre

sen

tin

gth

ere

sult

sto

ind

ust

ryp

arti

cip

ants

and

fin

aliz

ing

the

stu

dy

afer

rece

ivin

gfe

edba

ck.

the market emerges when two necessary conditionsare met: On the one hand, when coordination alongdifferent parts of the value chain is simplified, astask interdependence is reduced; and on the otherhand, when information becomes standardized—that is, when it becomes simple, transmissible, anduniversally understood.

Figure 6 summarizes the argument, and Table 2provides an overview of the process of discovery ofthe different elements of the framework I induced.Table 2 connects the specific factors identified inFigure 6 with the sources used to initially detect orvalidate my analysis, as well as with the researchstage in which I found the initial evidence for, andthen the validation of, each factor. In the remainderof this section, I consider each of the induced fac-tors, starting from the analysis of the necessaryconditions; I then move to the discussion of theenabling processes; and I finally look at what mo-tivates the process of vertical disintegration. Interms of Figure 6, I move from the top to the bot-tom, at each stage going deeper in the chain ofcausation. I thus gradually shift the focus of theanalysis from the question of how vertical disinte-gration happens, to why it happens.

Necessary Conditions: CoordinationSimplification and Information Standardization

Coordination simplification: Evidence. In allthree episodes, the market only emerged after co-ordination was simplified via substantial reductionin task interdependence between the differentstages of the production of mortgage loans (cf.Thompson, 1967). First, securitization became pos-sible when holding a loan did not have to interferewith originating or servicing that loan. Second, thecreation of a separate segment of originating bro-kers became possible when the production andwarehousing of loans became sequentially—ratherthan reciprocally—interdependent. Third, the mar-ket for servicing rights, which enabled servicingand origination to be separated, arose when origi-nation and servicing ceased to have tightconnections.

Securitization, or the separation of the owner-ship of an asset from its origination and servicing,would not have been possible without the arrange-ments that enabled each of the parts to operate inisolation from the others. The key to making secu-ritization work was to create a structure in whichmortgagors would not know who really ownedtheir loans; they would be dealing with the mort-gage banks that originated them. Indeed, many ofthe readers of this paper may not know that theirmortgages are really held by a smattering of finan-

cial institutions around the globe. Likewise, own-ers of the assets do not have any idea of the specificdetails of the loans they own. More important, thesecuritizers do not have to coordinate with themortgage banks on anything more than receivingthe payments from the loans at the prespecifiedintervals; the obligations, in case of foreclosure, arewell defined ex ante; there is no need for any com-munication about the loans among the originator/servicer, securitizer, and owner (Follain & Zorn,1990). Consider loan foreclosure: Securitizationonly became possible when an effective arrange-ment was made that gave mortgage banks an incen-tive to execute unsupervised, effective foreclo-sures, and the holders of the assets (the ultimateloan holders who had the securitized loans) did nothave to worry about when or how foreclosurewould actually occur.8 Such an effective arrange-ment was formalized in the U.S. Universal Com-mercial Code, Article 9. Without a structure to min-imize the requisite coordination among all theseparties, vertical disintegration would have beenvery difficult.

In the creation of the banker-broker market, sim-plification of coordination was also evident. Beforethe mid-1970s, the processes of seeking out cus-tomers and warehousing loans (i.e., preparing themfor sale to the secondary market) were tightly inte-grated, as mortgage banks were keen to originateonly the loans they knew they could sell at a profitto securitizers. From the late 1970s on, however,securitizers introduced optional delivery schemes.These were intended to get business for the securi-tizers offering them, yet they also changed the pro-duction process, as industry handbooks suggest(Lederman, 1985, 1995). As an executive I inter-viewed explained:

What you had to avoid [in the old days] was to bestuck with product the market would not want, be-cause you would have to kill it [incur a loss whenselling or use valuable and scarce capital if holdingit on your books]. On the other hand, if you had anagreed delivery [of closed loans, to an investor], youhad to procure all these loans. . . . The result wasthat you were on the heels of your production guysto fill your quotas. It would distort your pricing, and

8 If a loan was called within its first year and docu-mentation was not foolproof, the securitizer had the rightto force the mortgage bank to buy the loan back. Also,parties further up the value chain evaluated mortgagebanks on the quality of the loans they sold: the pricesthey obtained depended on the overall performance oftheir portfolios over time. Finally, mortgage banks borethe costs of foreclosure, which were fairly high, so theywanted to avoid bad loans.

2005 479Jacobides

you’d have to ensure they were delivering what youwanted or had agreed to. . . . Now with flexibledelivery, you could set your sales guys loose, pro-vided they knew what were the limits they had. Itwas easier to manage, and you could also use leadscoming from outside. It made your life easier as youcould focus separately on each piece of the businesswithout worrying about the other.

Such innovations decreased the coupling be-tween origination and warehousing by reducinginteractions between the two tasks. If a mortgagebanker produced more loans than anticipated, thecompany need not worry about not being able tosell them to an investor; the flexible arrangementsalleviated this concern. Thus, a part of the com-pany could just focus on origination without hav-ing to worry about the next step in the valuechain—in other words, interdependence becamesequential (Thompson, 1967). The availability oflarger and more flexible lines of credit from commer-cial banks had similar effects: it enabled mortgagebankers to construct a buffer between originationand warehousing, reducing task interdependence.

The reduction in coordination difficulties wasfacilitated by changes in information and commu-nication technology (Schneider, 1994). Cheaplyfaxing interest rate sheets en masse every morningto the broker network made real-time loan procure-ment effective; and enterprise information systems(i.e., electronic systems linking banks with outsideagents, brokers, etc.) made it easier for bankers tocoordinate effectively with brokers (Morgan Stan-ley Dean Witter, 2000), especially with the adventof automated underwriting (Foster, 1997), wherebya fixed set of criteria allowed brokers to ensure thatthey could close loans on the behalf of mortgagebankers, and also allowed mortgage bankers toknow that they could sell these loans to securitiz-ers. Managing brokers became only marginallymore complicated than managing the in-house re-tail production network (Lebowitz, 1995). As anexecutive from a mortgage bank put it:

Nowadays you can buy the product as easily fromany source. . . it used to be easier to rely on yourin-house people. . . but nowadays the systems makemanaging the brokers just as easy, so you look to allyour potential channels.

A similar pattern underlay the development ofthe market for servicing rights. In the early days ofthe industry, the activity of servicing a loan waspart and parcel of an integrated process, but in the1980s, the idea of integration started fading away(Garrett, 1993). One reason was that servicing be-came more dependent on computerized systemsthat stored information than on the unarticulated

knowledge of a loan officer in a branch bank thatoriginated a loan (Thinakal, 2001), thus obviatingthe need for integration. As a senior bank vicepresident noted:

Sure, you have these local shops, and they waxlyrical about their being integrated and all. But Ireally don’t see why you should bundle servicingwith origination anymore. I mean, what’s the point?The only value I can see is the emotional link be-tween the customer and the originating bank, interms of cross-selling, and that ain’t high. Perhaps itwas different before, when you didn’t have com-puter systems and the rest of it, but things changedin the 1990s.

Coordination simplification: Theory. The firstcondition that needs to be satisfied for a market toemerge, then, is for the coordination between theadjoining stages of a value chain to be simplified; ifthis does not occur, then it remains impossible forthose completing one stage to turn to an outsideparty to complete the next stage. In the presence ofreciprocal or pooled interdependence (Thompson,1967), it is comparatively hard to enable an activityto be located outside the boundaries of a firm. AsVan de Ven, Delbecq, and Koenig (1976), Nadler,and Tushman (1978), Gulati and Singh (1998), andPuranam and Singh (2000) have demonstrated, ac-tivities that require tight coordination demand theuse of mechanisms that a firm can deploy: commit-tees, meetings, or other authority or coordinationdevices that market-based relations lack, in com-parative terms (Kogut & Zander, 1996; Langlois,2003). Coordination difficulties, then, may be whyRichardson (1972), Teece (1976: 13), and Langloisand Robertson (1989, 1995) argued that integrationis needed to manage a tightly coupled system. Sofor a market to be feasible, interactions along avalue chain must be minimized; production mustbecome modularized (Baldwin & Clark, 2003).

Standardization of information: Evidence. Fordisintegration to occur, however, one more neces-sary condition must be met: Information must be-come standardized—that is, it must become be-come universally understandable and easilyspecified. Information standardization happenedin all three of the market creation episodes.

In the case of securitization, what enabled thesecuritizers to sell loan bundles to the capital mar-kets, and what enabled buyers to raise the capital tobuy them, was that the securitizers standardizedthe information on these loans (Passmore & Sparks,1996). Standardization made it possible for thebuyers to understand what the securitizers wereselling. “Market value was small, that’s whereSalomon Brothers came in. They standardized the


loans and made them fungible” (an executive ofE. F. Hutton, quoted in Tuchman [1986: 34]).

The information standardization that helpedmake securitization possible had two elements:

The first was standardization of loan characteris-tics; types of loans and their documentation andsupport became homogeneous. The second elementwas standardization of underlying customer cred-itworthiness information: the advent of universallyunderstandable and assessable attributes describingloan applicants and securities (i.e., the mortgage-backed debentures sold through the capital market),and the advent of a language to describe the risk thatloans entailed both to the primary market (the marketfor the mortgage loans that mortgage banks buy frommortgage brokers) and to the secondary market (themarket for securitized mortgage loans, sold by themortgage banks to the securitizers and from thesecuritizers to the capital market). Further detailson these two elements of standardization follow.

First, loan types were standardized. A key ingre-dient of the success in the initial development ofthe market was that “residential mortgages werequite homogeneous in terms of their design, terms,and underwriting standards” (Vandell, 2000: 1). Soa driver for the vertical disintegration process inthis industry was the substantial share of loanswith simpler and more tractable characteristicsthan had been seen in the past—initially, relativelystandardized 15- and 30-year fixed-rate mortgages(Lederman, 1985).

Such product standardization also helped verti-cal disintegration both within the origination pro-cess (addressing the broker-banker split) and be-tween origination and servicing, through themarket for servicing rights. Earlier mortgages hadidiosyncrasies and differing term structures, somoving them from one financial institution to an-other as brokered loans to be warehoused, or trans-ferring their servicing rights, was simply too com-plicated. However, given fixed-rate mortgages withset 15- or 30-year terms, with broadly similar char-acteristics, and given the homogenization withinadjustable-rate mortgages, many of these problemsdisappeared, and specialization became more plau-sible (Aldrich, Greenberg, & Payner, 2001).

The role of standardization in disintegration canalso be inferred from the fact that even today, inte-grated institutions specialize in the more compli-cated adjustable-rate mortgages or other loans withunusual clauses or structures. Because these loansare harder for any financial institution to procurefrom outside brokers, they are produced in-house;their servicing rights are harder to sell to anotherparty, so they are serviced in-house; and finally,they are harder to securitize, so integrated firms

such as S&Ls focus comparatively more on theseunusual loans (Garrett, 1993).9

Standardization of the product was also impor-tant because it meant that the information neededfor any potential transaction could be standardized.First, in order for securitizers to be able to sell theirloans, uniform production standards (such as com-mon underwriting procedures) had to be devised.Second, along with product standardization cameprocess standardization, in the form of standardprocedures for the interface between buyer andseller. So the initial involvement of securitizers inthe industry helped by providing a template of loanorigination:

Each thrift [had] evolved its own mortgage contractforms, documentation and record-keeping systems.When the cost of raw computation came down inthe 1960s and 1970s with the advent of computers,lack of standardization remained a barrier to thetransfer and pooling of mortgage assets. By offeringto buy [and then sell to the secondary market] mort-gages that conformed to certain standards, the [se-curities] played an important role in providing in-centives for the fragmented thrift industry tostandardize contracts on single family loans. (Bald-win & Esty, 1993: 40)

Notably, private securitizers such as Lehman andSalomon Brothers used the same criteria and de-scriptions as the government-sponsored enterprisesin buying mortgage loans to sell to the secondarymarket. For the sake of simplicity, they maintainedthe market “language” that had been created forloan specifications (Passmore & Sparks, 1996;Slessinger, 1994) and pushed standardization fur-ther.

The next step in standardizing information wasto standardize the certification of loan applicants’creditworthiness. By the early 1980s, consumercredit assessment agencies such as Equifax had de-veloped reliable data on customers, and specialistcompanies, in particular Fair Isaac, developed pre-dictive scores for individual customers. “Fico,”Fair Isaac’s main predictive score, became the defacto standard for underwriting decisions (Bennett,Peach, & Peristiani, 2001; Mara, 1989). With theinstitution of Fico and other scores, the underlyingcustomer loans became more easily describable.

9 The particularity of the U.S. in terms of the domi-nance of these simple loans is also one of the mainreasons that vertical specialization (and securitization inparticular) was quick to develop (Coles, 1999). In othercountries, such as the U.K., despite efforts to promotesecuritization, it has been slow to catch on. British loanshave a huge variety of product attributes, making a ver-tically specialized chain hard to orchestrate.

2005 481Jacobides

The firm-specific communication shorthands (Ar-row, 1974; Pelikan, 1969) that had been useful inthe past as the means of describing the types of loanapplicants a company wanted to retain were re-duced to a set of scores. As a mortgage brokerobserved:

With the Fico’s and the automated underwritinganyone could do the job. I mean, why should I besitting in First National’s crappy offices originatingthe loans, rather than just do the job myself? All they[bankers] want is a pile of loans which is healthyand meets their criteria. And believe you me, thiscommoditization of information is putting them in arough spot. I can now go around and sell my loan ifI really want to.

With information standardization came the abil-ity to specify, both inside and outside the bound-aries of a firm, the data needed on applications forloans to be warehoused, or on loans to be pur-chased, or on portfolios of servicing rights (Aldrichet al., 2001; Bennett et al., 2001).

Standardization of information: Theory. Interms of theory, it appears that the second neces-sary condition for a market to appear consists ofinformation standardization and the increasingease of information transfer. These features helpcreate and support an intermediate market, as itbecomes possible to certify information. Informa-tion asymmetries diminish, and thus the risk of“lemons” that might go through the market alsodiminishes (Akerlof, 1970; Passmore & Sparks,1996).

Yet over and beyond that, it seems that much ofwhat drove the emergence of markets was the abil-ity to specify what would be transacted across firmboundaries, even regardless of the presence of anylemons or transactional hazards (cf. Argyres, 1999;Monteverde, 1995). As has been pointed out in theengineering literature, specification—the creationof standard templates for product or process defi-nition—is viewed as indispensable for market pro-curement (Baldwin & Clark, 2003; Fine & Whitney,1996) because it replaces each organization’s ownshorthand and particular ways of describing things(Arrow, 1974; Pelikan, 1969; Winter, 1987). A stan-dard informational “syntax and grammar” mustarise for communication to happen (Argyres, 1999).Also, if information on what is desired is unclear ordifficult to articulate, outside procurement arrange-ments cannot be set up, as desired items can’t beidentified ex ante (Barzel, 1982; Boisot & Child,1988; Jacobides & Croson, 2001).

Summing up, the field evidence suggests that thecosts that keep a market from emerging do notprimarily relate to asset specificity. Specifically,

the two necessary conditions that need to be metfor a market to arise are the simplification of coor-dination and the standardization of information.10

The next step, though, is to consider what exactlyallows these two conditions to be met. The data inthis industry revealed that two processes enabled thesimplification of coordination and the standardiza-tion of information: intraorganizational partitioning,whereby parts of the value chain become separated ina single organization, and vertical cospecialization.

Enabling Processes: IntraorganizationalPartitioning and Vertical Cospecialization

Intraorganizational partitioning: Evidence.The first process is intraorganizational autonomyand partitioning, which is separation of the produc-tion process within the boundaries of organiza-tions, through, for example, the creation of differ-ent divisions for each part of a value chain.Separation creates clearly identifiable internalboundaries, which can then lead to market procure-ment inasmuch as one division chooses not totransact with another division, but with anotherfirm instead. So intraorganizational boundariespave the way for the use of a market.

This process happened both in the market be-tween bankers and brokers and in the market forservicing rights. During the late 1970s, private se-curitizers and the government-sponsored enter-prises instituted “optional” and “forward” loancommitments. These were provisions allowingmortgage banks to agree ex ante on the prices ofloans that would be delivered on a particular date,rather than produce loans without knowing exactlywhat price they would fetch when sold. These in-novations meant that the various processes in-

10 A conceptual question emerges as to the relation-ship between the two necessary conditions, which, whilerelated, are analytically distinct: Simplification of coor-dination concerns the extent of interactions between ac-tivities or decisions, exerted from one part of a valuechain to the next, regardless of the information flow,whereas standardization of information relates to thetype of information to be transmitted along the valuechain. If coordination becomes simplified, the requisiteinformation may become simpler or more reduced, yet itwill not necessarily become standardized. Conversely, ifinformation becomes standardized, the interactionsalong the value chain may well remain, as a number ofoutsourcers are painfully finding out in failed efforts tocut across firm boundaries. If information becomes stan-dardized and coordination simplified, and if there is amodicum of gains from trade, vertical disintegrationtakes place.


volved in loan production could become more sep-arable, as banks did not need to worry about theprice a loan they produced could fetch; they werehedged by the optional commitments, and thuscould separate warehousing from retail origination.This separation led to and was, in turn, furtherenhanced by the creation of separate units for thewarehousing and retail production of loans and anincrease in the autonomy of the loan agents whooriginated the loans. Part of what enabled that au-tonomy was, of course, the fact that their perfor-mance could be assessed, and the requirementsspecified, in clearer terms; that measurement im-proved, which led to intraorganizational separa-tion. The intraorganizational separation changedthe way in which mortgage banks were run. Thischange was noted with disdain by an old-time CEO:

It is no secret that the strategies for loan originationhave changed. Any number of companies haveturned the corner from retail to wholesale origina-tions or purchased servicing. [A reason for this isthat] many of today’s major industry participantsdid not grow up in the industry. Most are “johnny-come-lately” entrants [and not part of the tightlyintegrated firms mortgage banks used to be]. ManyCEOs of major mortgage banking firms have neveroriginated, marketed or serviced a loan in theirmortgage banking career [and thus do not under-stand why mortgage banks should be integrated].(Jacobs, 1993: 24)

The structure of the industry had changed, be-cause the structure of the firms had changed. Ac-cording to my interviews, the reason why thenewer generation of CEOs did not believe in inte-gration was that decentralized organizations weremore efficient. The change of internal structure andthe partitioning within organizations had paved theway for disintegration.

The link between organizational partitioning andvertical disintegration was more pronounced in theseparation of origination and servicing. Until thelate 1970s, mortgage banks were small firms thatdid not have separate divisions for servicing andorigination. However, as they grew, the desire tomanage their scope better, and a sense that theremight be some benefits from specialist depart-ments, led to an increasing separation of the origi-nation and servicing components. In the 1980s,several firms adopted separate units, with profit-and-loss reporting responsibilities for each of thesetwo major activities. The diffusion of this organiza-tional innovation made it more obvious to manag-ers in mortgage banks that there were importantdifferences between these stages of the value chain,as they could both measure their own value-addingprocesses better, and compare benchmarks with

other banks. That was made evident to me in aseries of “peer review roundtables” for mortgagebanks that I was allowed to observe.

Furthermore, the separation made it clearer thatthere were managerial differences along these twoparts of the value chain. Servicing is a business inwhich operational efficiencies determine profitmargins, and money is made by earning servicingfees. In origination, however, effectiveness (cus-tomer orientation), warehousing, and short-run in-terest-rate management capabilities matter (Garrett,1993). Furthermore, intraorganizational transpar-ency made firms identify their strength and ledefficient servicers to try expanding their servicingvolume by the occasional purchase of the loan port-folio of a competitor. So the organizational separa-tion facilitated both coordination simplificationand the comparison of performance across firms.

Finally, this process also happened in the firstepisode, securitization. The only difference wasthat, in this context, the process did not relate tothe growth of a set of firms but, rather, to the “socialengineering” of governmental and legislative agen-cies, which tried to facilitate vertical disintegration(Passmore & Sparks, 1996; U.S. Department ofHousing and Urban Development, 1996). The insti-tution of Article 9 of the U.S. Universal Commer-cial Code enabled creation of a blueprint, whichwas further refined by the operation of the govern-ment-sponsored agencies and the investment banksSalomon Brothers and Lehman Brothers (Lewis,1989). This framework was effective because it sup-ported a structure whereby one player could workindependently of another; in that regard, this epi-sode of disintegration in mortgages resembles thedisintegration of the PC sector “designed” by IBM(Baldwin & Clark, 2000).

Intraorganizational partitioning: Theory. Ingeneral, intraorganizational partitioning appears tolead to the creation of autonomous subunits and,ultimately, to market-mediated exchange. Gainsfrom specialization motivate part of this process, asexplained below, but another part of the process issimply time-varying: as firms grow, administrativedivisions become solidified (DiMaggio & Powell,1983). As firms restructure into multiple profit-and-loss (as opposed to cost) centers, decision mak-ers come to see each division’s operation as moreautonomous than it was originally, and they con-template alternative profit opportunities. Internal“shadow prices” and true intraorganizational au-tonomy invariably push units to consider the make-or-buy decision more actively and seek ways tomake it happen. There is also a feedback loop: Ascoordination becomes simpler, administrative par-titions are more likely to occur. This process, while

2005 483Jacobides

not strictly necessary, leads to simplifying coordi-nation as well as to identifying or intensifying pres-sures to outsource, often by active efforts to changethe transactional environment.

Vertical cospecialization: Evidence. As firmsstart considering their strengths and weaknessesmore explicitly, going outside their old boundariesbecomes attractive. The second process, verticalcospecialization, emerges as firms come up withmutually complementary roles in market-basedtransactions. The accompanying learning processenables information standardization to come aboutand creates the institutional background for markettransactions (Fligstein, 2001; North, 1986). The factthat it took about six years for the broker-bankerarrangements to emerge and solidify suggests thatthere needs to be a period of trial-and-error learningacross firm boundaries for a market to fully de-velop, as described in the following published ac-count of the market for mortgage brokers:

In 1981 the first rumblings of change began, . . . theunraveling and unbundling of the housing financedelivery system. . . . With no money [given a finan-cial crisis] with which to make mortgages [banks]started round after massive round of layoffs in theirloan origination staffs. . . . But as loan officerscleaned out their desks and headed for the door,they were often told something like the following:“Listen, Bob, you know we can’t afford to keep you,you know that there’s almost no business and thatwe have almost no money to lend. But we’ve alwaysthought you put together high quality loans andwe’d like to stay in touch. You have good contactswith Realtors, so if you can go out on your own andput together some loans, bring them to us and we’llfund them. And we’ll figure out some way to splitthe loan fee.”

Now things were different. A new kind of mortgageborrower who had access to institutional moneyappeared. The fellow who had been working at FirstFederal Savings was now an independent contractorworking on his own—but he could place a borrow-er’s loan with First Federal Savings. Initially, thoseoriginators who went out on their own would onlybroker to the institution they had previously workedfor. But rates plummeted in August of 1982, andinstitutional lenders were flush with cash andneeded mortgage product. Our broker who had beenplacing loans with First Federal Savings was nowbrokering to other institutions. In need of loans,Second Federal Savings approached him with a con-versation that went something like this: “We knowyou worked for and placed loans with First FederalSavings. But if you bring us some loans, maybe wecan offer better programs, and perhaps we can offeryou a better split on the fees. . . .” Before long, ourbroker was approved to place loans with a great manylenders. Calling themselves wholesalers, lending in-

stitutions set up whole departments not to originateloans at the borrower level, but to take them in fromoriginating brokers. (Garrett, 1989: 30–32)

The effects of this gradual process of verticalcospecialization, which led industry participantsto learn from each other how to work with theindependent brokers, are evident in the industry’sdemographic traits, which reflect the gradualchange from integrated production within firms toproduction of loans through independent brokers,in addition to in-house production. As Figure 4shows, the amount of brokered production in-creased over that period, as S&Ls were reducingtheir own workforces (cf. Mara, 1989; O’Donnel &Divney, 1994). The material from Garrett justquoted suggests that over time, parties that stand tobenefit from a potential transaction find a wayaround managing across the boundaries of a firmthat might have once housed them. The gradualadjustment between brokers and bankers that Gar-rett alludes to in the last paragraph above entailsthe diffusion of institutional innovations that en-able firms to cross their own boundaries and thecreation of mutually accommodating (i.e., cospe-cialized) vertical roles.

In mortgage banking, as in other settings, a grow-ing menu of choices emerged over time in the mar-ketplace. Whereas brokers initially only providedleads or qualified applicants, in the late 1990s agreater variety of arrangements became available(LaMalfa, 1998). Brokers could provide leads onlyor could also close the resulting loans; they couldeven initially fund them, too, through a practicecalled “table funding.” As time went by, finer sep-arations in the market appeared, offering morechoice of vertical scope, with different compensa-tion and quality criteria set for each stage of thevalue chain.

Finally, this cospecialization and learning bothresulted in the standardization of information andwere enabled by it. As a broker remarked:

As soon as people saw they could do business inthis [vertically specialized] way, and as soon as theywere able to iron out the details, they were up andrunning—and driving business away from the inte-grated guys. You know, little details—faxing the rateinformation, getting to a code of conduct—helped alot.

The same pattern happened in the first episode ofmarket creation (Passmore & Sparks, 1996). Securi-tization started with the standard 15- to 30-yearlow-risk loans and then moved into loans withhigher risk, as measured by loan-to-value ratio andtheir underlying customers’ credit (over time, boththe loan-to-value ratio and credit assessment be-


came increasingly more liberal than they had beenformerly) and, from 1984 onward, into adjustable-rate mortgages. As a Wall Street analyst put it:

You just get the market figuring it out and movinginto the more complex stuff. Once you apply thebasic idea in one set of loans, you can then move onand add a few more bells and whistles. You see, ittakes some time for the buyers to understand whatyou’re selling; you got to increase the complexity ofwhat you sell gradually.

This quote suggests that a market emerges onceinformation becomes standardized and that as in-formation on increasingly complex activities andprocesses becomes standardized over time, verticalspecialization can increase (Bennett et al., 2001). Italso suggests that it may well be cumulative expe-rience, learning, and the resulting cospecializationthat create the link between industry growth and spe-cialization—and not scale, as Stigler (1951) proposed.

Vertical cospecialization: Theory. To organizetransactions across the boundaries of firms, poten-tial participants in a market need to find ways oforchestrating the transactions. As Langlois (1992:104) remarked, it takes time for the market to “hitupon institutional arrangements.” Such cospecial-ization is essentially a gradual learning and en-abling process; it sets in once the parties know thatthere is a benefit from transacting across firmboundaries. It is the manifestation of the effort tostandardize information, and it also depends on thecontext of the parties that are engaged in it. Moreimportantly, perhaps, cospecialization connotesthe ability to engage in a new way of organizing,and the interest in doing so. Further, social, legal,and political contexts can enhance or hinder learn-ing new ways of transacting. So, governments (Flig-stein, 2001) and other social forces (Aldrich & Fiol,1994) can and often do play roles by shaping theprocess that will ultimately lead to the creation ofthe functional and institutional background formarket exchange (North, 1986; Olson, 2000).

Drivers of Disintegration: Gains fromSpecialization and Gains from Trade

The analysis so far has identified two necessaryconditions for and two processes that facilitatemarket creation. Still, I have not answered the morefundamental question of what drives vertical disin-tegration and why this sequence of events unfoldsin some settings and not in others. Here, I proposethat disintegration is largely driven by the desire ofindustry participants to take advantage of the gainsfrom specialized production and to realize gainsfrom potential trade.

Gains from specialization: Evidence. The ben-efits of vertically specialized production are a ma-jor driver of disintegration. In my first episode, theevolution of the secondary loan market, disintegra-tion happened because many financial institutionswere itching to hold mortgage loans (that is, holdfixed-income assets that were collateralized withreal estate) but couldn’t do so as they lacked capa-bilities in the origination and servicing that had toprecede and follow the “capital claim” of the mort-gage loan. Hence, once the securitizers found a wayto create marketable securities, finding willing fi-nancial institutions or investors to buy the loanswithout the hassle of producing or servicing themwas easy (Follain & Zorn, 1990). The securitizersalso found mortgage banks and S&Ls that werehappy to sell their loans, as these institutions wereefficient in origination and servicing yet had littledesire, funding, or stomach for managing capitalrisk. Were it not for these gains from specialization,the primary and secondary markets would not havetaken off.

In the development of the mortgage broker mar-ket, the pattern was similar. Firms that were moreadept in warehousing loans than in originationpushed for the development of an intermediatemarket for brokered loans, to increase the pool ofloans they could warehouse, which is where theymade money. As a consultant commented,

It only made sense for production to fall into thehands of brokers. An independent guy is successfulbecause he can focus on what he can do best—keepin touch with the local community, have a goodnetwork, and be an expert salesman who takes cus-tomers, holds their hand, and explains the process.. . . Sure, it [vertical specialization] should havehappened earlier, but I guess that finding a way todeal with brokers took some time.

So while mortgage banks continued in-houseproduction of mortgages, they increasingly reliedon the independent agents because of advantages interms of gains from specialization (Mara, 1989).This quotation from an experienced broker portraysthese advantages well:

Why be a broker and not work in a mortgage bank?Well, first of all, you are in control of your owndestiny. I mean, brokers are better at figuring outwhat the clients need and where the market is going.And letting the brokers loose made this a betterindustry. Second, it’s also about money—while youcan be on commission even in a bank, you still makemore money when you’re on your own or in a bro-kerage. Anyway, it’s a different mentality, differentstyle, different priorities—it just works better.

Gains from specialization: Theory. This evi-dence from mortgage banking, then, is consistent

2005 485Jacobides

with Smith’s (1776/1874) work, which suggeststhat the propensity to trade and barter, in order totake advantage of one’s capabilities, can be a majordriver of vertical specialization (Demsetz, 1988; Ri-chardson, 1972). And there must be some latent,unrealized gains if a market is to appear.

First, gains from specialization exist because uni-fied up- and downstream governance reduces theeffectiveness of production, even while it decreasestransactional risks. For example, in the broker-banker relation, the very same agent, if employed ina bank as part of an in-house production unit, willbe less effective than if he or she is an outsidecontractor (a mortgage broker selling loans throughthe market). As an outside contractor, he or she willhave a much freer hand, without administrativeconstraints; the management style of the upstreambusiness can be quite different from the style of thedownstream one. The heterogeneity of the knowl-edge bases and managerial structures along thevalue chain (Ghemawat & Ricart i Costa, 1993) thusdetermine the gains from specialization. So in areal sense, “managerial diseconomies of scope” re-sult from differences in the requisite capabilitiesand styles of each vertical segment (Richardson,1972). These managerial diseconomies drive thelatent gains from specialization, which, in turn,create the need for an intermediate market toemerge as well as determine the extent of interme-diate trade, given a market (Jacobides & Hill, 2005).

Second, in economic terms specialized produc-tion is incentive-intensive: an outside or free agentis closer to the market of interest and hence may bemore effective (Milgrom & Roberts, 1992; William-son, 1985: Ch. 6; Zenger & Marshall, 2000). Brokerswho moved from banks to form their own firms didso in order to profit more from their abilities. To theextent that integrated structures restrict incentives,perhaps for equity reasons within an integratedfirm, specialization will be preferred.

Third, firms may also specialize to reap the benefitsof demand smoothing and aggregation—that is, anoutside agent sells all possible loans, not only thosein which an employing mortgage bank specializes.11

Latent gains from trade: Evidence. Gains fromspecialization, while related to gains from trade,are analytically distinct. Our evidence suggests thatpure gains from trade played a significant role inthe push towards vertical disintegration.

In our first episode, a factor driving securitiza-tion was that, compared to their competition, S&Ls

had a very poor set of funding activities. To wit,one of the two main government-sponsored enter-prises, Freddie Mac, was initially set up by anassociation of S&Ls, the Federal Home Loan BanksAssociation, which owned Freddie Mac from itsinception in 1970 to its public listing in 1989. Fred-die Mac was put together to enable the S&Ls to getrid of the part of the home loan business they knewthey were bad at—finding the capital to fund theloans. The idea was that Freddie Mac would helpkeep this part out of the value chains of the S&Ls sothat they could focus on their strength, generatingand servicing loans. So the gains that would resultfrom such vertical disintegration (S&Ls originatingand servicing loans, and Freddie Mac outsourcingpart of the capital provision to buyers of mortgage-backed securities) motivated the S&Ls to put in thetime, effort, lobbying, and funds to support FreddieMac (U.S. Department of Housing and Urban De-velopment, 1996).

In the second episode, a major motivation for thedisintegration between bankers and brokers wasthat the banks that were strong in warehousingcould not grow their origination segment (in-housebrokerage) as quickly and efficiently as they couldgrow their warehousing functions. They neededincreasingly more brokered business to cross theirfirm boundaries. The reason was that originationrequires local links with consumers in a specificgeographical area. An industry consultant noted:

You saw some of the warehousing-focused firmsgrowing aggressively. . . and they were able to fuelthe growth using mortgage brokers. You can’t growyour own retail production that quickly—so brokerswere the obvious solution.

Potential gains from trade also propelled the dis-integration in servicing and the creation of the mar-ket for servicing rights (Aldrich et al., 2001). Ac-cording to industry association studies (Duncan,1998), accounts of technology providers (Thinakal,2001), and confidential evidence provided for thisstudy, the difference between what good loan ser-vicers and mediocre ones could charge grew signif-icantly between 1988 and 2001. Firms that weregood at loan servicing sought to focus vertically inorder to reap high potential gains from specializa-tion. As a mortgage banker noted:

Effective servicers decided that, even with the highpurchase prices asked for servicing portfolios, theycould still make money—because they were moreeffective at servicing and they were looking at cross-selling revenues—and so they kept buying [servic-ing rights].

As the efficient servicers managed to scale uptheir operations quickly, they soon sought to create

11 See Jacobides and Billinger (2005) on how the use ofintermediate markets can increase operational effective-ness and efficiency.


and support a market for servicing rights to lever-age their scalable superior capability; as a result,disintegration between origination and servicingensued, which caused concentration in servicing torise faster than concentration in origination.12

In addition to established players attemptinggains from trade, other actors supported the currentcase of disintegration. First among these were po-tential entrants who stood to benefit from a disin-tegrated industry structure. Such entrants couldcome from within the ranks of existing firms (suchas the brokers, who wanted to become self-man-aged entities) or from other segments, such as WallStreet firms that wanted to get securitization vol-ume to generate fees. To do this, they had to findwilling accomplices and also ensure that they le-veraged their own capabilities (for accessing thecapital markets and structuring deals) to becomeactive players in the new sector. Outsiders whostand to gain from vertical disintegration try toshape an institutional environment to fit theircapabilities.

A second group of actors in such cases are tech-nology providers, who often precipitate a change invertical structure inasmuch as they help transferinformation or modularize a production process. Inthis case, technology suppliers such as FiServ andAllTel structured their offerings to accommodate avertical disintegration. As a technology executivenoted:

Look, you have to go to the client with a competitiveedge. So if you tell them they can do somethingnew—better integrate brokers, accommodate the lat-est thing, and then explain how this will save themmoney—you’re more likely to win business. So wetry to be proactive.

Such suppliers are often the first to recognize thepotential gains from a reorganization of production,and they try to take some of the benefits that resultfrom the vertical reorganization of labor; but theselatent benefits must exist for their efforts to beworth the investment, even if the participants arenot always convinced ex ante. Consider the case ofrating agencies, like Moody’s and Standard &Poor’s (for the market for securitized loans), orFitch ICBA or Fair Isaac (for mortgage banking), or

credit record agencies for consumers (for the mar-ket for brokered loans). These firms, in the processof finding new, profitable business for themselves,created the preconditions for the new intermediatemarkets to be created by standardizing information.

The third set of actors includes the governmentand regulators. The government’s key role in themortgage banking industry was its early involve-ment with the government-sponsored enterprisesand with jump-starting the secondary market formortgage loans (U.S. Department of Housing andUrban Development, 1996). Creating the legal back-ground making securitization possible, especiallythrough the institution of Article 9 of the UniversalCommercial Code, was very important as well. Thegovernment created the templates that the marketneeded. However, once the templates were set, itphased out its participation and eventually privat-ized the government-sponsored enterprises (U.S.Treasury Department, 1991). Also, through theGarn-St. Germain Act of 1982, the U.S. governmentlifted the statutory limits that forced thrifts to beintegrated, thus allowing a disintegration it hadpreviously prohibited in part of the industry (U.S.Congress, 1982). The government’s narrowly fo-cused involvement worked because it unlocked thelatent gains from trade, which led to the support forsuch an action. Without this intervention, interme-diate markets might have taken substantially longerto emerge, although they could have manifestedthemselves in some form, as they did in Europe inthe late 1990s (Coles, 1999). But governmental orregulatory intervention is not indispensable. In theother two market episodes—the separations of orig-ination and brokerage and of servicing and origina-tion—the involvement of such a central authoritywas conspicuously absent. Perhaps surprisingly,the mortgage banking industry is the least regulatedpart of the financial services sector in the UnitedStates.

Regulation tends to either institute or legitimizenew rules, such as vertically cospecialized arrange-ments. As players in each part of an industry try tolobby for their interests, they try to promote anindustry structure that is maximally profitable forthemselves. In essence, then, potential gains fromtrade also appear to shape the role of government,in that potential gainers who can be organizedenough and stand to win enough will lobby aggres-sively, by pushing for standardization and the le-gitimization of particular ways of organizing (Abo-lafia, 1996; Zelizer, 1979). As the CFO of amidsized mortgage bank said:

There’s no beating around the bush. As far as regu-lation goes, it’s clear there’s a battle going on. [The

12 Between 1989 and 1999, the top ten servicers’ shareof the market rose from 11 to 41 percent, while the topten originators’ share went from 17 to 36 percent—so theconcentration cumulative growth rate was roughly dou-ble for the former. The most efficient firms have gainedshare more quickly in servicing, and that has happenedthrough the purchase of servicing rights (Inside MortgageFinance Publications, 2000).

2005 487Jacobides

securitizers] want to encroach on our turf, andchange the division of labor so they end up with agreater part of the pie. Which is why we spend allthese lobbying dollars to ensure we keep the struc-ture of the industry more reasonable.

So it appears that the institutional background ofthis industry was driven by industry participants’conscious efforts to shape it to their advantage.This observation offers both support for and quali-fication of Olson’s (2000) analysis of the role ofmarket-augmenting governments. The support isthat, in this setting, government can play a signifi-cant role through providing appropriate infrastruc-ture, or even catalyzing a market through its ownactions, as it did for securitization. Thus, marketsand governments should not be viewed as polaropposites; rather, there is a direct need for govern-mental support to facilitate the emergence of mar-kets (cf. North, 1986). The qualification of Olson(2000) resides in the possibility that governmentsaugment markets not only because they expectgrowth and as such an increase in their own taxrevenues, but also because lobbying pressures andthe desire to placate particular constituencies makethe case novel institutional arrangements (cf. Olson& Kahkonen, 2000; Azfar & Cadwell, 2003). Thewords of a White House official speaking to econ-omists on policy making, as reported by De Long(2000: 139), capture this dynamic well: “What youeconomists don’t see, is that you are pushing for thepublic interest. But there are other interests thatcan be more important.”

Given the current results, I tentatively concur.While the interplay between government, privateinterests, and institutions requires dedicated anal-ysis that far exceeds the scope of this paper, thecurrent evidence points to some interesting possi-bilities: It indicates that the evolution of institu-tional form, and even the role of government inshaping it, might depend on the “interests” at hand,and in turn these depend on the structure of pro-duction that determines the incentives of differentindustry participants, through their expected gainsfrom trade. So these gains need to be understood.

Gains from trade: Theory. Potential gains fromtrade are necessary for a market to emerge—for anexchange to happen, it must be economically mo-tivated and viable (Afuah, 2001; Zenger & Poppo,1998)— but they also facilitate disintegration. First,gains from trade occur when the capabilities in anindustry are not evenly distributed. Whenever afirm is efficient in one part of the production pro-cess (say, loan origination) and not in another (say,servicing), it has the incentive to support the cre-ation of a new market that allows it to eliminate theinefficient part of its organization. So differences in

capabilities between firms along the value chain,with some firms being good upstream and othersdownstream, push disintegration, as a quantitativestudy of the same sector has elaborated (Jacobides &Hitt, 2005).

A subtler but equally important motivation for amarket to emerge is uneven growth along a valuechain. In the presence of bottlenecks resulting fromparts of a chain having different rates of growth andprofitability, efficient firms trying to expand theirvolume want to use outside providers that enablethem to grow more quickly and increase total prof-its (cf. Jacobides, 2004).

Similarly, the presence of economies of scale thatare important in only one part of a value chain(such as those reported in the servicing sector inthis context) make specialization of a few, largercompanies in that scale-intensive sector profitable;so some firms specialize in the segments with higheconomies of scale, whereas the others become spe-cialized, smaller players in the segments with “dis-economies” (Jacobides, 2004; Stigler, 1951).

Such latent gains from trade explain why incum-bents often support rather than fight new special-ized firms, such as mortgage brokers (cf. Carroll,1985). Yet gains from trade do not only concernexisting industry participants, who stand to gainfrom vertical specialization and exchange: Theyalso concern potential entrants, who can only par-take in the industry if and when it becomes verti-cally specialized; and further, they concern tech-nology and infrastructure providers. Potentialentrants and technology providers are keen to seethe structure of an industry become specialized.They thus try to push for disintegration, partneringeither with other outsiders, or with incumbents.

A Model of the Drivers and Mechanics of theProcess of Vertical Disintegration

Figure 7 shows graphically the full model of ver-tical disintegration developed here. This frame-work helps to explain why and when vertical dis-integration happens. To retrace the chain ofcausation, I have explained that the real drivers ofvertical disintegration are (1) benefits from special-ization, which are themselves driven by differencesin the knowledge bases and requisite managerialstyles along the value chain, and (2) potential gainsfrom trade, resulting from imbalances of capabili-ties along the value chain of existing participants orgains that could obtain if new, vertically special-ized entrants were to migrate into an industry. Forinstance, inasmuch as the upstream part of themortgage business relies on financial expertise andmanaging interest rates, and the downstream part


of the business relies on salesmanship, there arepotential gains from specialization within a singlefirm itself. Potential differences in managementstyles or in knowledge bases make administrativeseparations between departments more effectivethan unified management (Lawrence & Lorsch,1967; Gulati, Lawrence, & Puranam, 2001), whichmeans there are some real managerial benefits to behad by specializing. In the presence of such gains,a process of intraorganizational partitioning is putin motion. For instance, firms create clearly dis-tinct departments of interest rate management andsales, perhaps by creating separate divisions, oreven create different profit-and-loss centers foreach part of the value chain. This intraorganiza-tional partitioning allows the coordination betweenthe two adjoining parts of the value chain to be sim-plified through a reduction of task interdependence,which further fuels administrative partitioning.

Also, as a function of history and capabilities,some firms are simply better than others in partic-ular parts of a value chain. Inasmuch as there arelatent gains from specialization—that is, if somefirms are good upstream and weak downstream andvice versa—the prospect of being able to specializebecomes attractive for those who stand to benefitfrom it. Gains from specialization (which meanfirms are more likely to be good at one rather thanboth parts of the value chain) tend to reinforcegains from trade, which may or may not be evidentto potential transactors. Additionally, wheneverfirms cannot quickly expand all the parts of a valuechain, the possibility of relying on an outside party

to help them grow, through vertical cospecializa-tion, becomes a profitable prospect. Technologyvendors and potential entrants who stand to thriveor at least participate in such a disintegrated envi-ronment also try to push for disintegration, oftenencouraging current industry participants to sup-port specialization by making a case for it and byhelping to turn latent gains from trade into identi-fied gains. The attractiveness of a market increaseswith intraorganizational partitioning, which raisescomparisons of efficiency inside and outside afirm. For all these reasons, firms engage in interfirmlearning, with the objective of enabling vertical co-specialization and gains from trade. This processleads to the standardization of information and toironing out problems that market transactionswould induce. In this process, insiders and outsid-ers may engage regulators, the government, or otherindustry bodies that can help information becomestandardized.

With information sufficiently standardized andcoordination simplified, a market can emerge, evenif significant problems of opportunism, depen-dency, and measurement persist. As the marketoperates, the problems are gradually addressed,and institutional arrangements are made to copewith them; information also becomes increasinglystandardized and coordination modularized, en-abling more value to be divided through the mar-ket; the market itself, then, becomes adept at coor-dinating increasingly complex activities through afeedback loop of the cospecialization and learningprocess.

FIGURE 7The Full Model: Drivers and Mechanics of Vertical Disintegration and Market Creation

2005 489Jacobides

DISCUSSION

This industry study provides a framework forexplaining the nature and the drivers of verticaldisintegration. It suggests that coordination simpli-fication and information standardization are thenecessary conditions for market emergence. Fur-ther, this study sets forth the proposition that thesetwo conditions are respectively met through theprocesses of intraorganizational partitioning, andvertical cospecialization with interfirm learning.Going further back in the chain of causation, I pro-posed that gains from specialization and latentgains from trade shape those two enabling pro-cesses. I noted that such gains exist when both theknowledge bases and managerial styles along avalue chain, and firms’ capabilities, differ. Suchdifferences set vertical disintegration into motion.

The analysis of the dynamics of vertical disinte-gration fills a significant gap in the literature re-garding this very important aspect of industry evo-lution. Academic research so far has not examinedvertical disintegration, despite the fact that it radi-cally transforms the industries in which it occurs.Much as the computer industry’s identity radicallychanged with its shift toward a multitude of sub-segments in both the hardware and software sec-tors, the mortgage lending industry was trans-formed through vertical disintegration. The birth ofa vertically cospecialized ecosystem, shown in Fig-ure 3, changed the nature of the industry and rede-fined potential entrants and competitors, whetherfirms chose to be integrated or not (Jacobs, 1993).The dynamics of vertical disintegration that trans-formed the mortgage banking world would havegone undetected with use of the established frame-works of industry evolution, since these do notaddress the evolution of an industry’s verticalstructure; they would also not have been capturedby transaction cost economics, which, focusing onindividual transactions, risks losing sight of theforest for the trees.

This study suggests that by shifting focus from anindividual firm and from the final market for aproduct or service to the value chain structure andthe process by which the good or service is pro-duced, one can capture hidden but important dy-namics. This analysis suggests that exogenous fac-tors did not shape this value chain, but rather, theconscious, even if occasionally myopic, efforts ofindustry participants and potential entrants shapedit. Firms and individuals who break industrynorms instigate disintegration as they, unlike therest of the industry, understand the potential of adisintegrated production structure and are willingto put in the effort to make such disintegration

possible (Knight, 1921; Schumpeter, 1911). Effortsof Salomon and Lehman Brothers exectives (suchas those captured in Lewis’s [1989] best-seller, Li-ar’s Poker, for example), while motivated by achance to profit or profiteer, helped securitizationtake hold and led to disintegration.

Yet, at the same time, changes in vertical scope inthe focal industry were not the result of a well-thought-out plan but, rather, of a quest for near-term profit. Most of the interviewees I spoke tocould not even describe the structure of the mort-gage banking value chain; part of what made myproject interesting to them was the fact that theycould get this strategic overview. What they didknow was that they were constantly changing theirown boundaries to increase their margins, be moreeffective, or gain market share. This quest for prof-its and operational efficiency often came at theexpense of their strategic prospects. While I wasconducting the study, I noticed a significant changein attitudes toward some of the finer degrees ofdisintegration facilitated by information technol-ogy. Increasing specialization, originally hailed asefficiency-enhancing, started hitting firms’ bottomlines through intensified competition, and execu-tives came to grips with the fact that their quest forefficiencies or even short-term growth through spe-cialization could ultimately hurt them. This quota-tion from a 2002 speech by Angelo Mozillo, theCEO of CountryWide, the largest U.S. mortgagebank, captures this dilemma: “For years we havebeen trying to make the system more efficient andget rid of the middleman. . . the problem is, we arethe middleman.”

Purposeful, even if myopic, manipulation of in-stitutional context is way beyond what transactioncost analysis would lead one to posit. Firms, ac-cording to that theory, choose from menus of op-tions, or transactional alternatives, to minimize thesum of production and transaction costs. But thetheory does not address how menus of transactionsevolve over time. By looking at the evolution of anentire value chain over time, I have observed thattransactional environments themselves evolve,shaped by the actions of firms that reduce “mun-dane transaction costs” either consciously (throughtrying to standardize information) or unwittingly(by simplifying coordination along the value chain,initially to improve the division of labor withintheir own scope). So this study suggests that tounderstand vertical scope, scholars have to under-stand, at the industry level, the forces that affect it.In particular, this study suggests a need to examinethe forces driving toward disintegration, ratherthan simply focus on the microanalytics of firms’


individual make-or-buy choices (cf. Langlois &Foss, 1999; Jacobides & Winter, 2005).

My findings also qualify transaction cost eco-nomics by suggesting that the types of transactioncosts operating in the initial disintegration stage arequite different from those that push individualfirms to pursue vertical integration. The costs thatinhibit specialization reflect Coase’s (1937) originalconcept of “friction”—what Williamson (1985),perhaps dismissively, called “mundane transactioncosts” (see Baldwin & Clark, 2003). These mundanecosts are driven by lack of standardization anddifficulties of coordinating across firm boundaries,rather than by the problems of hold-up discussedby Klein, Crawford, and Alchian (1978) and byWilliamson (1985, 1999). My findings suggest thatthese costs are quite important: they determine afirm’s ability to rely on a market in the first place, aquestion that Williamsonian analysts do not evenconsider.

Langlois (1992) and Langlois and Robertson(1995) argued that transaction costs are a dynamicand hence transient phenomenon, but their workdoes not explain when and how these transactioncosts are reduced. This study provides a frameworkthat accounts not only for the mechanisms of trans-action cost reduction and market emergence, butalso looks at the causes and underlying factors thatexplain when and why transaction costs decreaserapidly in some settings, leading to disintegration,whereas in others they do not. It suggests that it isthe nature of a production process, and the result-ing gains from specialization and trade, that drivevertical scope, market emergence, and the potentialfor disintegration.

This research also complements the work of eco-nomic sociologists who have been interested in thenature and role of markets. Sociologists, buildingon White’s (1981) work, have explored the emer-gence of markets, construed as categories of offer-ings that both producers and consumers perceive asclose substitutes, focusing in particular on the so-cial and institutional dynamics of “product cate-gory” creation (Carruthers & Babb, 2000; Porac etal., 1995; White, 1981). They have examined thenorms that sustain marketplaces—the (well-de-fined) junctures where goods and services are ex-changed (Abolafia, 1996), the norms of productionand exchange (Zelizer, 1979), and their social con-struction (Ruef, 1999)—and considered marketsboth as broad societal institutions (Carruthers,1996) and as social fields (Fligstein, 2001). Yet theyhave not directly focused, to date, on explaininghow markets that link two stages of a value chainarise, or on how a productive system evolves andendogenously leads to the emergence of new mar-

kets. This study does so, and in so doing suggeststhat, under some conditions, participants in an in-tegrated status quo may have the incentive to pro-mote institution-changing innovations (cf. Carroll,1985).

This qualitative study, then, suggests that thesociological argument that existing actors (domi-nant firms) try to reproduce their own structures(White, 1992) is not universally applicable. In thissetting, the forces to create new markets came fromor were immediately supported by existing firms.Incumbents, rather than reproducing existing struc-tures, supported disintegration (contrast this viewwith Fligstein [2001: Ch. 3 & 4]). Despite the factthat these representatives of established firms sup-ported the new markets observed here, disintegra-tion ultimately became detrimental to them. Like-wise, the findings are in contrast to the resource-partitioning view (Carroll, 1985), which suggeststhat newly specialized firms have to fight against(integrated) incumbents. Whereas sociological re-search by and large suggests that new markets onlyemerge when they serve the interests of the statusquo, I found that this is not always the case. Inmortgage banking, many established firms helpeddisintegration happen, even if they eventuallyfound it to be strategically detrimental. Furtherstudy could thus help provide a more nuancedview of the social dynamics of new intermediatemarkets.

This study also affords some opportunities forspeculation. While markets may “emerge” (as didthe market for mortgage brokers), they also requireeffort to be set up. In the setting of this research,and possibly in many others, markets are not asself-enforcing as, say, a food market in a rural ba-zaar (Olson, 2000: Ch. 10). My analysis suggeststhat markets are “artifacts” (Simon, 1962), con-sciously designed either by industry participants orby government and regulators. For instance, with-out the creation of an institutional backbone forexchange, including but not limited to informationstandards and norms of collaboration, a marketcould not operate. While law, especially in Anglo-Saxon countries, tends to follow business practice,and as such ratifies and regulates the efficientmodes of organizing that have emerged (cf. Cooter,2000), it is still the case that the legal and institu-tional framework must be such that it can supportmarket creation (Olson & Kahkonen, 2000). Marketsdepend on economic agents who willingly partakewhen they can see and calculate the benefits ofdoing so (Callon, 1998), but markets also criticallydepend on the establishment of infrastructure andnorms for interaction (MacKenzie & Millo, 2003).Infrastructure and norms are the results of both trial

2005 491Jacobides

and error and painstaking design (Coase, 1992;North, 1986), which help partition value chainsand delimit the scopes of activity of all involved(Loasby, 2000). So, for markets to emerge, effortmust be extended; and it may be that, absent regu-latory involvement, no party will find its privatebenefit from a new market worth the set-up costs,and as such a potentially beneficial new structuremay be foregone (Olson, 2000).

Limitations and Extensions

This research has several limitations. This is astudy of a particular industry and hence should notbe hastily generalized to other settings. Like anyother process research (Mohr, 1982), it focuses onunderstanding the causal dynamics of a particularsetting, as opposed to providing information on thegeneralizability of the findings to other settings. Ichose the mortgage banking industry because itunderwent disintegration and market creation, inparticular because it witnessed several episodes ofvertical specialization. I also looked at a stable in-dustry with few changes in product definition inorder to isolate the major conceptual issues. Itwould be interesting to consider how this analysisshould be modified to accommodate substantialprocess and product innovation.

This study does not provide much microanalyti-cal detail on the changes in coordination or infor-mation mechanisms that enabled disintegrationand market creation. It would be useful to study thespecific microprocesses by which latent gains totrade were identified and how entrepreneurs andincumbents tried to change the industry’s structureas a result of such identifications. A look at suchmicroprocesses would help better explain the po-litical and strategic element of the informationstandardization process (Fligstein, 2001; Gawer &Cusamano, 2002) and the agency of actors in shap-ing their institutional and social environment (Cac-ciatori & Jacobides, 2005).

A related topic that has not been fully exploredherein is the mediating role of government in ver-tical disintegration. While not every market re-quires intervention or even support by government,my evidence does suggest that it facilitates or evenprompts the emergence of some markets, as securi-tization demonstrated here. The way in which agovernment or regulator affects vertical specializa-tion and market creation through legislation,through subsuming fixed costs of market infrastruc-ture, or through incentives (such as tax incentives),and, even more, the way in which industry partic-ipants can affect governmental action, all remainfascinating topics for future research.

Finally, in focusing on delineating the conditionsfor and drivers of vertical disintegration, this studyhas underplayed the role of historical accident,which does affect industry evolution. In this case,history partly “paid the cost” of the infrastructureof new markets: For instance, when the ResolutionTrust Corporation was endowed with mortgagesthat someone had to service, it had to come up witha way of selling mortgage servicing rights and thusindirectly subsidized the creation of the marketinfrastructure. Also, historical conditions can pro-vide the external shock that enables a departurefrom previously developed routines inherent in in-stitutional form and vertical scope (Nelson & Win-ter, 1982; North, 1981). For instance, the layoffs ofloan agents in the 1980s brought the latent butpreviously disregarded gains from specializationand trade to the surface, which ultimately led to thecreation of the market for brokered loans: Learningrequires some “unlearning” first, and unlearningoften requires a shock or crisis (Polanyi, 1957).

There are several other possible extensions tothis research. At the micro level, one could explic-itly study the relationship between organizationalidentity and vertical scope (see Cacciatori & Jaco-bides, 2005; Jacobides & Winter, 2005). One couldstudy how framing, identity, and cognitive repre-sentations change as a function of changing indus-try scope, and how changes in identity furthershape the vertical scope of a sector. In the setting ofthis study, one could study the evolving identity ofthe mortgage broker segment, which was bothcaused by and resulted in an increase of verticalspecialization and market creation. Researcherscould further look at how mortgage brokers seethemselves, and how they differ from those origi-nating loans within integrated firms. Last, research-ers could examine how the emergence of mortgagebrokers has changed the identity and nature of loanorigination even within integrated firms.

At the meso level, in addition to looking at howmanagerial differences along the value chain andcapability differences between firms affect verticalscope, researchers could examine how changes invertical scope affect the capability developmentprocess, managerial styles, and knowledge bases.As Jacobides and Winter (2005) also recently sug-gested, disintegration changes the structure of anorganization, and as such changes the process ofcapability development. This theme is developedby Cacciatori and Jacobides (2005), who explainedhow vertical specialization affects and constrainsthe capabilities of an industry as a whole, butclearly more needs to be done in that respect. Ex-plicitly linking capabilities and vertical scope, andin particular focusing on how changes in scope


loop back to capability development, could be apromising research venue.

Another extension to this research would be toreconsider “modularity”—that is, the extent of au-tonomy of different stages of a production process.In the current setting, the modular structure of theindustry was the result of an evolutionary, bot-tom-up process, motivated by gains from special-ization and trade, achieved through organizationalpartitioning and information standardization. Thisevolution does not quite fit recent work on organi-zational theory and modularity (Baldwin & Clark,2000; Schilling, 2000; Simon, 1962), which usuallyexamines design problems from the perspective ofa “system designer.” Thus, an extension of thisstudy would be to consider “modular emergence”and the conditions that allow bottom-up modular-ization in the absence of a designer.

Finally, at the macro level, research could studythe new entrants who arrive with newly disinte-grated sectors (for instance, the entry of EDS [Elec-tronic Data Systems] into banking through informa-tion processing) and explore how such firmsmigrate from one industry to another, oftenprompting disintegration. This analysis could pro-vide some new insights into how vertical disinte-gration affects the dynamics of competition andhelp explain how different sectors converge, oftenthrough vertical disintegration.

More generally, this study is best viewed as onlyone part of a broader investigation into what Coase(1992) called “the institutional structure of produc-tion,” an examination of how labor is divided bothbetween different firms, and across the vertical di-vide (Jacobides & Winter, 2005). Industries do notonly disintegrate. They also reintegrate, or recom-bine on the basis of newly found possibilities fororganizing (Fine, 1998). For instance, in automo-bile manufacturing, where the initial stage of verti-cal disintegration occurred, several firms are nowemerging as “systems providers,” integrating com-ponents once produced by independent firms thatcontracted directly with OEMs (original equipmentmanufacturers). At the same time, these systemsintegrators themselves are relying even more onnew markets, using several subcontractors to helpcreate “systems” (see Brusoni, Prencipe, & Pavitt,2001). Also, as Cacciatori and Jacobides (2005) ar-gued in their study of building construction, verti-cal reintegration in services often requires the cre-ation of new all-in-one markets, where a buyerneeds to find a new way to interact with one inte-grated service provider, abandoning piecemealcontracting. Thus, perhaps paradoxically, scholarsmay need to better understand the process of new

market creation to explain the process of verticalreintegration as well.

Studying the nature and the evolution of produc-tive systems by focusing on their value chains orvalue networks is a promising path for future re-search that can bring to the fore new empiricalregularities and can augment understanding of howfirms and industries coevolve. It can also point tonew explanations for old and new puzzles alike.For instance, as I argue in related work (Jacobides,2005), the “path dependency” in the vertical divi-sion of labor, and country-level operation of thedynamics that shape intermediate markets, meanthat otherwise similar countries may have indus-tries with markedly different vertical structures.Thus, foreign expansion in some sectors may meetlimits inasmuch as two countries have incommen-surate “vertical modules.” So capabilities devel-oped in one context, for one part of a value chain,are not transferable to another context, a trend thatis evident not only in mortgage banking, but also inother sectors, such as construction. Conversely, in-ternational convergence in value chain structurescan drive toward or be motivated by the prospect ofglobal trade (Jacobides, 2005). Studying the verticaldivision of labor thus becomes an important meansto explain globalization patterns—a topical issue,given the recent rise of global trade through “off-shoring” and international outsourcing (see Feen-stra, 1998).

CONCLUSION

In terms of practice, this framework explainswhen and why vertical disintegration occurs andthus provides a tool for managers who want tounderstand whether their industry is or is not liableto this change. Evans and Wurster (1997) and Hageland Singer (1999) suggested that the boundaries oforganizations and industries would be redrawn inthe near future. The framework developed herecontextualizes such claims; it identifies the condi-tions under which disintegration does happen.Evans and Wurster (1999), for instance, argued thatmany sectors were ripe for disintegration and re-configuration; yet many of the sectors they singledout have remained integrated to date. Insurers, forinstance, presumably did not manage to simplifycoordination sufficiently or standardize the requi-site information, and there were not enough gainsfrom trade on the industry level to justify special-ization. In a world of rapidly changing firm andindustry boundaries, analyzing vertical disintegra-tion and its determinants can help guide practiceby providing more robust foundations for strategyand policy alike.

2005 493Jacobides

In terms of research, by changing the level andfocus of analysis—from an individual firm’s staticdetermination of vertical scope to the systemic evo-lution of an industry’s intermediate markets—thisstudy uncovered some dynamics that had not beenstudied to date. By doing so, it qualified and ex-tended existing research (in particular, transactioncost economics) and also went deeply into thechain of causation, showing that transaction costsare an incidental part of an evolutionary process. Ithighlighted the factors that are ultimately respon-sible for vertical disintegration and suggested re-searchers and managers should not take the verticalstructure of an industry for granted. Finally, thisstudy suggests that changes in a value chain’s struc-ture have significant implications for all firms, bethey integrated or not, and may well be the leaststudied, yet a very important, part of industry evo-lution.

REFERENCES

Abernathy, W. J., & Utterback, J. M. 1978. Patterns ofinnovation I technology. Technology Review, 80(7):40–47.

Abolafia, M. Y. 1996. Making markets. Cambridge, MA:Harvard University Press.

Afuah, A. 2001. Dynamic boundaries of the firm: Arefirms better off being vertically integrated in the faceof a technological change? Academy of Manage-ment Journal, 44: 1211–1229.

Akerlof, G. A. 1970. The market for lemons: Qualitativeuncertainty and the market mechanism. QuarterlyJournal of Economics, 84: 488–500.

Aldrich, H. E., & Fiol, C. M. 1994. Fools rush in? Theinstitutional context of new industry creation. Acad-emy of Management Review, 19: 645–670.

Aldrich, S. P. B., Greenberg, W. R., & Payner, B. S. 2001.A capital markets view of mortgage servicing rights.Journal of Fixed Income, 11(1): 37–48.

Allen, F., & Santomero, A. M. 1997. The theory of finan-cial intermediation. Journal of Banking and Fi-nance, 21: 1461–1485.

Argyres, N. S. 1999. The impact of information technol-ogy on coordination: Evidence from the B-2 Stealthbomber. Organization Science, 10(2): 162–181.

Argyres, N. S., & Liebeskind, J. 1999. Contractual com-mitments, bargaining power and governance insep-arability: Incorporating history into transaction costtheory. Academy of Management Review, 24: 49–63.

Arora, A., Fosfuri, A., & Gambardella, A. 200. Markets fortechnology and their implications for corporate strat-egy. Industrial and Corporate Change, 65(1): 31–55.

Arrow, K. J. 1974. The limits of organization. New York:Norton.

Azfar, O., & Cadwell, C. A. (Eds.). 2003. Market-augment-ing government: The institutional foundations forprosperity. Ann Arbor: University of MichiganPress.

Baldwin, C. Y., & Clark, K. B. 2000. Modularization andproduct architecture. Boston: MIT Press.

Baldwin, C. Y., & Clark, K. B. 2003. Where do transac-tions come from? A perspective from engineeringdesign. Working paper, Harvard Business School,Boston.

Baldwin, C. Y., & Esty, B. C. 1993. Financial perspectivesservices and challenges. In S. L. Hayes III (Ed.), Les-sons from the thrift crisis: 35–64. Boston: HarvardBusiness School Press.

Barzel, Y. 1982. Measurement cost and the organizationof markets. Journal of Law and Economics, 25: 27–48.

Bennett, P., Peach, R., & Peristiani, S. 2001. How muchmortgage pool information do investors need? Jour-nal of Fixed Income, 11(1): 8–15.

Boisot, M., & Child, J. 1988. The iron law of fiefs: Bureau-cratic failure and the problem of governance in theChinese economic reforms. Administrative ScienceQuarterly, 4: 507–527.

Brusoni, S., Prencipe, A., & Pavitt, K. 2001. Knowledgespecialization, organizational coupling, and theboundaries of the firm: Why do firms know morethan they make? Administrative Science Quarterly,46: 597–621.

Burgelman, R.A. 1991 Intra-organizational ecology ofstrategy making and organizational adaptation: The-ory and field research. Organization Science 2(3):239–262.

Cacciatori, E., & Jacobides, M. G. 2004. The dynamiclinks of specialization: Vertical re-integration re-considered. Working paper, SIM area, London Busi-ness School; http://ssrn.com/abstract.

Callon, M. 1998. The laws of the markets. Oxford, En-gland: Blackwell.

Carroll, G. 1985. Concentration and specialization: Dy-namics of niche width in populations of organiza-tions. American Journal of Sociology, 90: 1263–1283.

Carruthers, B. G. 1996. The color of money and the natureof value: Greenbacks and gold. American Journal ofSociology, 101: 1556–1592.

Carruthers, B. G., & Babb, S. L. 2000. Economy/society:Markets, meanings, and the social structure. Thou-sand Oaks, CA: Pine Forge Press.

Christensen C., Verlinden, M., & Westerman, G. 2002.Distribution, disintegration and the dissipation ofdifferentiability. Industrial and Corporate Change,11: 955–994.


Coase, R. H. 1937. The nature of the firm. Economics(new series), 4: 386–400.

Coase, R. H. 1992. The institutional structure of produc-tion. American Economic Review, 82: 713–719.

Coles, A. 1999. Securitization: What is different about theUSA? Housing Finance International, March: 9–13.

Cooter, R. D. 2000. Law from order: Economic develop-ment and the jurisprudence of social norms. In M.Olson & S. Kahkonen (Eds.), A not-so-dismal sci-ence: A broader view of economics and societies:228–244. Oxford, England: Oxford University Press.

De Long, J. B. 2000. Overstrong against thyself. In M.Olson & S. Kahkonen (Eds.), A not-so-dismal sci-ence: A broader view of economics and societies:138–167. Oxford, England: Oxford University Press.

Demsetz, H. 1988. The theory of the firm revisited. Jour-nal of Law and Economics and Organization, 1:141–161

DiMaggio, P. J., & Powell, W. W. 1983. The iron cagerevisited: Institutional isomorphism and collectiverationality in organizational fields. American Socio-logical Review, 48: 147–160.

Duncan, D. G. 1998. Relative profitability of the mortgageindustry. Mortgage Banking, 58(9): 95–97.

Eccles, R. G., & White, H. C. 1986. Firm and marketinterfaces of profit center control. In S. Lindberg, J. S.Coleman, & S. Novak (Eds.), Approaches to socialtheory: 203–220. New York: Russell Sage.

Eisenhardt, K. M. 1989. Building theories from casestudy research. Academy of Management Review,14: 532–550.

Eisenhardt, K. M. 1991. Better stories and better con-structs: The case for rigor and comparative logic.Academy of Management Review, 16: 620–627.

Evans, P., & Wurster, T. S. 1997. Strategy and the neweconomics of information. Harvard Business Re-view, 75(5): 70–83.

Evans, P., & Wurster, T. S. 1999. Blown to bits. Boston:Harvard Business School Press.

Fabozzi, F. J., & Modigliani, F. 1992. Mortgage and mort-gage-backed securities markets. Boston: HarvardBusiness School Press.

Fannie Mae. 1992. Fannie Mae Charter Act (as amendedby October 28). http://www.fanniemae.com/global/pdf/aboutfm/understanding/charter.pdf.

Feenstra, R. 1998. Integration of trade and dis-integrationof production in the global economy. Journal ofEconomic Perspectives, 12(4): 31–50.

Fine, C. 1998. Clockspeed. Proteus Books.

Fine, C. H., & Whitney, D. E. 1996. Is the make-buydecision a core competence? Discussion paper, MITCenter for Technology, Policy, and Industrial Devel-opment, Cambridge, MA.

Firestone, W. A. 1993. Alternative arguments for gener-

alizing from data as applied to qualitative research.Educational Researcher, 22(4): 16–23.

Fligstein, N. 2001. The architecture of markets. Prince-ton, NJ: Princeton University Press.

Follain, J. R., & Zorn, P. M. 1990. Unbundling of residen-tial mortgage finance. Journal of Housing Research,1: 117–137.

Foster, D. 1997. The debate over automated underwrit-ing. Mortgage Banking, 57(8): 10–16.

Garrett, J. 1989. New paradigm for lenders. MortgageBanking, 49(8): 30–37.

Garrett, J. 1993. The unbundling of an industry. Mort-gage Banking, 53(5):12–21.

Gawer, A., & Cusumano, M. 2002. Platform leadership:How Intel, Microsoft and Cisco drive industry in-novation. Boston: Harvard Business School Press.

Ghemawat, P., & Ricart i Costa, J. 1993. The organiza-tional tension between static and dynamic effi-ciency. Strategic Management Journal, 14: 59–74.

Guba, E. G., & Lincoln, Y. S. 1982. Epistemological andmethodological bases of naturalistic inquiry. Educa-tional Communications and Technology Journal,30(4): 233–252.

Gulati, R., Lawrence, P. R., & Puranam, P. 2005. Adapta-tion in vertical relationships: Beyond incentive con-flict. Strategic Management Journal, 26: 415–440.

Gulati, R., & Singh, H. 1998. The architecture of cooper-ation: Managing coordination costs and appropria-tion concerns in strategic alliances. AdministrativeScience Quarterly, 43: 781–815.

Hagel, J., & Singer, M. 1999. Unbundling the corporation.Harvard Business Review, 77(2): 133–142.

Hart, O. 1995. Firms, contracts and financial structure.New York: Oxford University Press.

Henderson, R., & Clark, K. B., 1990. Architectural inno-vation: The reconfiguration of existing systems andthe failure of established firms. Administrative Sci-ence Quarterly, 35: 9–30.

Inside Mortgage Finance Publications. 2000. Mortgagemarket statistical annual. Bethesda, MD: InsideMortgage Finance Publications.

Jacobides, M. G. 2001. Mortgage banking unbundling.Mortgage Banking 61(4): 28–39.

Jacobides, M. G. 2004. A model of how capabilities,scale, learning and transaction costs jointly shapevertical scope. Working paper, LBS/LeverhulmeTrust Programme, London.

Jacobides, M. G. 2005. Playing football in a soccer field:Value chain structure, vertical co-specialization, in-stitutional modularity and success in foreign expan-sion. Managerial and Decision Economics: Inpress.

Jacobides, M. G., & Billinger, S. 2005. Designing theboundaries of the firm: From “make, buy or ally”

2005 495Jacobides

to the dynamic benefits of vertical architecture.Working paper, LBS/Leverhulme Trust Digital TransProgramme, April; http://ssrn.com.abstract-670386.

Jacobides, M. G., & Croson, D. C. 2001. Information pol-icy: Shaping the value of agency relations. Academyof Management Review, 26: 202–221.

Jacobides, M. G., & Hitt, L. M. 2005. Losing sight of theforest for the trees? Productive capabilities and gainsfrom trade as drivers of vertical scope. StrategicManagement Journal, 26: In press.

Jacobides, M. G., & Winter, S. G. 2005. The co-evolutionof capabilities and transaction costs: Explaining theinstitutional structure of production. Strategic Man-agement Journal, 26: 395–413.

Jacobs, J. J. 1993. In search of the real mortgage bankers.Mortgage Banking, 53(6): 21.

Klein, B., Crawford, R. A., & Alchian, A. A. 1978. Verticalintegration, appropriable rents, and the competitivecontracting process. Journal of Law and Economics,21: 297–326.

Klepper, S. 1997. Industry life cycles. Industrial andCorporate Change, 6: 145–181.

Knight, F. H. 1921. Risk, uncertainty and profit. NewYork: Houghton Mifflin.

Kogut, B., & Zander, U. 1996. What firms do? Coordina-tion, identity, and learning. Organization Science,7: 502–518.

LaMalfa, T. 1990. The wholesale business. MortgageBanking, 50(1): 53–55.

LaMalfa, T. 1998. Wholesale giants ’97. Mortgage Bank-ing, 58(11): 46–64.

Langlois, R. N. 1992. Transaction cost economics in realtime. Industrial and Corporate Change, 1: 99–127.

Langlois, R. N. 2003. The vanishing hand: The changingdynamics of industrial capitalism. Industrial andCorporate Change, 12: 351–385.

Langlois, R. N., & Foss, N. J. 1999. Capabilities and gov-ernance: The rebirth of production in the theory ofeconomic organization. Kyklos, 52(2): 201–218.

Langlois, R. N., & Robertson, P. L. 1989. Explaining ver-tical integration: Lessons from the American auto-mobile industry. Journal of Economic History, 49:361–375.

Langlois, R. N., & Robertson, P. L. 1995. Firms, marketsand economic change: A dynamic theory of busi-ness institutions. London: Routledge.

Lawrence, P. R., & Lorsch J. W. 1967. Organizations andenvironment: Managing differentiation and inte-gration. Boston: Harvard University Press.

Lebowitz, J. 1995. A new look at technology. MortgageBanking, 55(4): 50–59.

Lederman, J. (Ed.). 1985. Introduction to mortgagebanking. Washington, DC: MBA Editions.

Lederman, J. (Ed.). 1995. Introduction to mortgagebanking. Washington, DC: MBA Editions.

Lewis, M. 1989. Liar’s poker. Norton: New York.

Loasby, B. 2000. Market institutions and economic evo-lution. Journal of Evolutionary Economics, 10: 297–309.

Macher, T. 1998. Reversal of fortune? The recovery of theUS semiconductor industry. California Manage-ment Review, 41(1): 107–137.

MacKenzie, D., & Millo, Y. 2003. Constructing a market,performing theory: The historical sociology of a fi-nancial derivatives exchange. American Journal ofSociology, 109: 107–145.

Madhok, A. 2002. Reassessing the fundamentals and be-yond: Ronald Coase, the transaction cost and re-sources based theories of the firm and the institu-tional structure of production. StrategicManagement Journal, 23: 535–550.

Mara, J. P. 1989. The new economics of mortgaging.Mortgage Banking, 49(3): 89–94.

McKinley, W. 1993. Organizational decline and adapta-tion: Theoretical controversies. Organization Sci-ence, 4(1): 1–10.

Miles, M. B., & Huberman, A. M. 1994. Qualitative dataanalysis: An expanded sourcebook (2nd ed.). Lon-don: Sage.

Milgrom, P., & Roberts, J. 1992. Economics, organiza-tion, and management. Englewood Cliffs, NJ: Pren-tice-Hall.

Mohr, L. B. 1982. Explaining organizational behavior:The limits and possibilities of theory and research.San Francisco: Jossey-Bass.

Monteverde, K. 1995. Technical dialog as an incentivefor vertical integration in the semiconductor indus-try. Management Science, 41: 1624–1638.

Morgan Stanley Dean Witter. 2000. The dream industry:Report on the mortgage banking sector. New York:Morgan Stanley Dean Witter.

Nadler, D. A., & Tushman, M. L. 1978. Information pro-cessing as an integrating concept in organizationaldesign. Academy of Management Review, 3: 613–625.

Nadler, J., Mistretta, S., & Kumar, R. 1992. The newstandards for rating mortgage securities in the sec-ondary mortgage market. Chicago: Probus Publish-ing.

Nelson, R. R., & Winter, S. G. 1982. An evolutionarytheory of economic change. Cambridge, MA:Belknap.

North, D. C. 1981. Structure and change in economichistory. New York: Norton.

North, D. C. 1986. The new institutional economics.Journal of Institutional and Theoretical Econom-ics, 142: 230–327.


O’Donnel, T., & Divney, B. 1994. The mortgage bankingfood chain. Mortgage Banking, 54(11): 29–34.

Olson, M. 2000. Power and prosperity: Outgrowingcommunist and capitalist dictatorships. Oxford,England: Oxford University Press.

Olson, M., & Kahkonen, S. (Eds.). 2000. A not-so-dismalscience: A broader view of economics and societ-ies. Oxford, England: Oxford University Press.

Passmore, W., & Sparks, R.W. 1996. Putting the squeezeon a market for lemons: Government-sponsoredmortgage securitization. Journal of Real Estate andEconomics, 13(1): 27–43.

Pelikan, P. 1969. Language as a limiting factor for cen-tralization. American Economic Review, 59: 625–631.

Pettigrew, A. M. 1990. Longitudinal field research onchange: Theory and practice. Organizational Sci-ence, 1: 267–292.

Polanyi, K. 1957. The great transformation. Boston:Beacon.

Porac, J. F., Thomas, F., Wilson, D., Paton, X., & Kanfer,A. 1995. Rivalry and the industry model of Scottishknitwear producers. Administrative Science Quar-terly, 40: 203–227.

Propper, M. S., & Furino, R. D. 1994. The revolution inmortgage origination. Mortgage Banking, 54(9): 64–71.

Puranam, P., & Singh, H. 2000. Bringing some disciplineto M & A mania (“Mastering management series”).Financial Times, October 30: 85–89.

Richardson, G. B. 1972. The organization of industry.Economic Journal, 82: 883–896.

Ruef, M. 1999. Social ontology and the dynamics of or-ganizational forms: Creating market actors in thehealthcare field. Social Forces, 77: 1403–1432.

Schilling, M. 2000. Towards a general modular systemstheory and its application to interfirm product mod-ularity. Academy of Management Review, 25: 312–335.

Schneider, H. 1994. Great strides in industry automation.Mortgage Banking, 55(2): 10–17.

Schumpeter, J. A. 1911/1934. The theory of economicdevelopment [1st published in German; Leipzig:Duncker & Humblot]. Cambridge, MA: Harvard Uni-versity Press.

Silver, M. 1984. Enterprise and the scope of the firm.London: Martin Robertson.

Simon, H. A. 1962. The sciences of the artificial. Cam-bridge, MA: MIT Press.

Slesinger, P. K. 1994. Data standardization: Streamliningaccess to the secondary market. Mortgage Banking,54(5): 107–109.

Smith, A. 1776/1874. The wealth of nations: An inquiry

into the nature and causes. London & New York:George Routledge & Sons.

Stigler, G. J. 1951. The division of labor is limited by theextent of the market. Journal of Political Economy,59: 185–193.

Teece, D. J. 1976. Vertical integration and vertical dives-titure in the U. S. oil industry. Stanford, CA: Stan-ford University Press.

Thinakal, S. 2001 The new Web-centric servicing busi-ness. Mortgage Banking, 61(2): 41–46.

Thompson, J. D. 1967. Organizations in action. NewYork: McGraw-Hill.

Tuchman, J. N. 1986. Innovation in the US mortgagemarket to 1990. Special report no. 1051, EconomistIntelligence Unit, London.

Tushman, M., & Anderson, P. 1986. Technological dis-continuities and organizational environments. Ad-ministrative Science Quarterly, 31: 439–465.

U.S. Congress. 1982. Garn-St. Germain depository insti-tutions act of 1982. Publ. no. 97–320, 96 Stat. 1469;http://www.fdic.gov/regulations/laws/rules/8000-4100.html.

U.S. Department of Housing and Urban Development.1996. Privatization of Fannie Mae and FreddieMac. Policy paper, Office of Policy Developmentand Research, Washington, DC.

U.S. Department of Housing and Urban Development.2002. Economic analysis and initial regulatoryflexibility analysis for RESPA proposed rule tosimplify and improve the process of obtainingmortgages. Policy paper, Office of Policy Develop-ment and Research, http://www.hud.gov/offices/ogc/ea-chapters.pdf.

U.S. Treasury Department. 1991. Modernizing the finan-cial system. Washington, DC: U.S. GovernmentPrinting Office.

Vandell, K. D. 2000. Securitization of the U.S. mortgagemarket: Progress and pitfalls, with lessons for Ja-pan. Paper presented at the Seventh Annual Inter-national Land Policy Forum, October 31.

Van de Ven, A. H., Delbecq, A. L., & Koenig, R. 1976.Determinants of coordination modes within organi-zations. American Sociological Review 41: 322–338.

White, H. C. 1981. Where do markets come from? Amer-ican Journal of Sociology, 87: 17–547.

White, H. C. 1992. Identity and control: A structuraltheory of social action. Princeton, NJ: PrincetonUniversity Press.

Williamson, O. E. 1975. Markets and hierarchies: Anal-ysis and antitrust implications. New York: FreePress.

Williamson, O. E. 1985. The economic institutions ofcapitalism. New York: Free Press.

2005 497Jacobides

Williamson, O. E. 1999. Strategy research: Governanceand competence perspectives. Strategic Manage-ment Journal, 20: 1087–1108.

Williamson, O. E. 2000. New institutional economics:Taking stock, looking ahead. Journal of EconomicLiterature, 38: 595–614.

Winter, S. G. 1984. Schumperterian competition in alter-native technological regimes. Economic Behaviorand Organization, 5: 287–320.

Winter, S. G. 1987. Knowledge and competence as stra-tegic assets. In D. J. Teece (Ed.), The competitivechallenge: Strategies for industrial innovation andrenewal: 159–184. Cambridge, MA: Ballinger.

Winter, S. G. 1991. On Coase, competence and the corpo-ration. In O. E. Williamson & S. G. Winter (Eds.), Na-ture of the firm, origins, evolution and development:179–195. Oxford, England: Oxford University Press.

Wiser, R. F. & LaMalfa, T. 1991. Mortgage servicingcomes of age. Mortgage Banking, 51(9): 14–18.

Yin, R. K. 1994. Case study research: Design and meth-ods revised (2nd ed.). London: Sage.

Zelizer, V. A. 1979. Morals and markets: The develop-ment of life insurance in the United States. NewYork: Columbia University Press.

Zenger, T. R., & Marshall, C. R. 2000. Determinants of

incentive intensity in group-based rewards. Acad-emy of Management Journal, 43: 149–164.

Zenger, T. R., & Poppo, L. 1998. Testing alternative the-ories of the firm: Transaction cost, knowledge-based,and measurement. Strategic Management Journal,19: 853–878.

Michael G. Jacobides ([email protected]) receivedhis Ph.D. in strategy at the Wharton School of the Uni-versity of Pennsylvania. He is an assistant professor ofstrategic and international management at the LondonBusiness School. His research examines how institutionsand firms coevolve and how this coevolution affectsfirms’ strategic prospects. In particular, he is currentlyexamining how transaction costs and vertical scopeshape the nature of industry participants and the processof capability development—and vice versa. As a princi-pal investigator in the Leverhulme Trust’s Digital Trans-formations Programme at the London Business School,he also examines how technologies transform firmand industry boundaries and reshape the competitivelandscape.


Documents

INDUSTRY CHANGE THROUGH VERTICAL DISINTEGRATION: …...INDUSTRY CHANGE THROUGH VERTICAL DISINTEGRATION: HOW AND WHY MARKETS EMERGED IN MORTGAGE BANKING MICHAEL G. JACOBIDES London