Computerized Loan Origination Systems: An Industry Case ... · PDF fileAn Industry Case Study of the Electronic Markets Hypothesis By: ... erty meet the firm’s approval criteria

Electronic Markets Hypothes:ts

Computerized LoanOrigination Systems:An Industry CaseStudy of theElectronic MarketsHypothesis

By: Christopher M. HessValueQuest, Ltd.Roundy’s HillMarblehead, Massachusetts 01945U.S.A.

Chris F. KemererMassachusetts Institute of

TechnologyE53-315 Sloan School of

ManagementCambridge, Massachusetts 02139U.S.A.

AbstractMuch has been written in recent years about thechanges in corporate strategies and industrystructures associated with electronic coordinationof market activities. This paper considers the ad-vent of electronic market coordination in thehome mortgage industry, focusing on Com-puterized Lean Origination (CL O) systems. Casestudies of five CLOs (First Boston’s Shelternet,PRC’s LoanExpress, American Financial Net-work’s Rennie Mae, Prudential’s CLOS, andCiticorp’s Mortgage Power Plus) reveal a rangeof system functionalities. Predictions from theElectronic Markets Hypothesis (EMH) are testedagainst the empirical results of the five casestudies. As suggested by the EMH, financial in-termediaries have been threatened by the in-"troduction of CLOS, and in some casesopposition has been mounted against thesystems. On the other hand, despite theavailability of the technology and mortgages’seemingly favorable characteristics as an elec-

tronically mediated market product, the industryhas not been fundamentally changed by the in-troduction of these systems, despite more thana decade of experience with them. Of the twocase studies that could be characterized as elec-tronic markets, neither continues to exist in thatform today. And the system with the largest dollarvolume of mortgages of the five is bestcharacterized as an electronic hierarchy. Theseresults suggest that either the full resultspredicted by the EMH require a longer gestationperiod or that the underlying hypothesis will re-quire augmentation in order to fully explain theresults in the home mortgage market.

Keywords: Computerized loan originationsystems, electronic markets, elec-tronic hierarchies, incomplete con-tracts, Shelternet, mortgages.

ISRL categories: AM02, BA0215, EL0202GA01, HA0702, HB32

IntroductionWilliam Bolt of Cinnaminson, N.J., knows wellthe anxieties of buying a new house. In threesuch transactions, he waited an average of sixagonizing weeks for loans to be approved. It waswith a certain foreboding, therefore, that Mr. Boltfaced buying another new house last August--this time before the lease on a rented house ranout. "By the time we found the home we wanted,we had to settle in two weeks, " he recalls."Everyone said it was impossible." But everyonewas wrong. Mr. Bolt’s mortgage banker turnedto a nationwide, computerized mortgage-searchservice to process the Bolt loan application. Themortgage was approved just eight working dayslater (Lipman, 1984).

Although he may not have been aware of it at thetime, Mr. Bolt had happened upon an innovationthat many in the early to mid-1980s felt wasdestined to revolutionize the market for homemortgages. Computerized Loan Originationsystems, or CLOs, were heralded as mortgagebanking’s savior by some and as a dire threat byothers. For years, mortgage lenders hadbenefited from computerizing the "back end" oftheir operations, where records were archived,

MIS Quarterly~September 1994 251

Electronic Markets Hypothesis

balances were transferred, and bills were sentout. The "front end," where loans were con-tracted for by a home buyer and a lending officer,was dominated by face-to-face interaction, pilesof paperwork, and, as Mr. Bolt experienced in histhree previous outings, a lot of waiting.

CLOs promised to change that. Automation firstcrept into the front end of the mortgage businessthrough the local lending officer’s door. Largelenders offered their agents terminals they coulduse to tap into centralized databases in order tohave the latest information about volatile interestrates and to transmit application information backto the corporate underwriters as an aid in theorigination process, the creation and delivery ofmortgage loans. These early CLOs generally of-fered loans from only one lender and left the taskof picking that lender and traveling to his or heroffice to the borrower. The critical change thatmade CLOs the source of hope and controversyoccurred when the systems went out the lender’sdoor and set up shop at the local real estateoffice.

Suddenly, CLOs were much more interesting.The new systems linked the entire loan produc-tion value chain from its beginning at the pointof home sale to the lenders’ back offices whereloans were underwritten and approved. In a 1984Wall Street Journal article, the pronouncementsand predictions of industry leaders about CLO-induced changes to the fundamentalcharacteristics of the industry came fast andfurious:

Financial-services networks will mushroom un-til real-estate agent and the banker melt into one,providing buyers with one-stop shopping ....The transformation has already begun ....These national networks will become the rulerather than the exception .... Once computer-ized mortgage searches become morewidespread.., anyone can, and will, originatemortgages (Lipman, 1984).

Against this backdrop of media excitement, theinformation technology (IT) literature offers theory upon which to base predictions about thefuture scope and direction of this market. Thetheory, developed by Malone, Yates and Ben-jamin, and referred to as the Electronic MarketsHypothesis, suggests that the introduction of ITwill, all other things being equal, generally lead

to greater use of markets rather than hierarchiesfor economic transactions (Malone, et al., 1987).

Predictions from the Electroni(; MarketsHypothesis (EMH) are tested in this paper againstthe empirical results of case sudies of five CLOsfrom the United States (First Boston’s Shelternet,PRC’s LoanExpress, American Financial Net-work’s Rennie Mae, Prudential’s CLOS, andCiticorp’s Mortgage Power Plus). As suggestedby the EMH, IT has reduced the time and effortrequired to select and secure a mortgage, andfinancial intermediaries have been threatened bythe introduction of CLOs, with active oppositionmounted in some cases against the systems. Onthe other hand, despite the availability of thetechnology and mortgages’ favorable character-istics as an electronically mediated market prod-uct, the industry has not yet been fundamentallychanged by the introduction of these systems.01~ the two case studies that could be character-ized as true electronic markets, neither continueto exist in that form today. And the most suc-cessful system, that is, the one with the currentlargest dollar volume of mortgages, is bestcharacterized as an electronic hierarchy.

These results suggest that either the resultspredicted by the EMH require a longer gestationperiod or that the underlying hypothesis will re-quire augmentation in order to fully explain theresults in the home mortgage market. Somepossible barriers to the advent of full electronicmarkets in the home mortgage industry are sug-gested as possible directions for future researchto explore in continuing the validation of the EMH.

This paper is organized as follows. The next sec-tion presents a very brief introduction to the mort-g-’Lge banking industry, intended as backgroundfor subsequent sections. This is followed by asummary of recent research on the advent andattributes of electronic markets and hierarchies,wl~ich focuses on a list of predictions stemmingfrom the EMH. The analysis section develops athree-level categorization of these systems anddescribes the experience of five CLO systems inlight of the predictions of the EMH. These resultsare discussed in the penultimate section, whichalso offers suggestions for future research. Con-cluding remarks are presented in the finalsection.

252 MIS Quarterly/September 1994


Mortgage Banking IndustryBackground’During the late 1980s more than $400 billion inhome mortgages were originated each year(Miller, 1992, p. 91). Every stage in the processby which loans are made depends on the transferof volumes of information from one party toanother. While lenders have made great progressin now offering more than 200 different loan-typechoices, the means by which borrowers selecttheir loans has been slower to evolve (Lipman,1984). Home buyers still scour local newspaperslooking for the right rate and terms, and spendhours on the phone with lending officers. Figure1 shows the relationships, activities, and par-ticipants that characterize the traditional homemortgage process.

Lender and loan selection is the first of five stepsin the process of home mortgage origination.

Subsequent steps are application, prequalifica-tion, underwriting, and closing. In submitting anapplication, the potential mortgagor provides themortgagee with information about the propertyin question and his or her present financial situa-tion, including income, current housing costs, job,assets, and debts. During underwriting, thelender verifies the claims made on the applica-tion and determines if the applicant and the prop-erty meet the firm’s approval criteria for the loanin question. Closing a home mortgage involvesthe actual transfer of the funds in question andsigning of various loan documents. After a loanis closed, the chief task remaining for lenders ortheir agents is servicing the loan, which entailsprocessing the periodic loan payments.

Savings and loan associations and savings bankshave traditionally provided the largest portion ofmortgage financing. In addition, while demanddeposit accounts are the primary focus of com-mercial bankers’ attention, mortgages are an im-

Home Buyer ~. . / "-----.~_~/</ ~ Closed Loan\

r~nancla! /~ j~ ~¯ I . ~

Info / I \ IInvestment IHousing Amount to ~ . ~ . . I ~ I Bankers INieds

.borr~~e~c~a~__l

_ _ ~~/ ~ / Lending Practices (,~1_~ FNMA-

~~ ~ ~:n ~ge ~ " [ Other In, titution~

~ /Rates, terms

Property Loan Infolistings

I Realtor ]

MortgageBroker

Primary Market _1_ Secondary Market

Figure 1. Traditional Mortgage Market

MIS Quarterly/September 1994 253


portant secondary activity. Mortgage bankers usetheir own funds or capital from other sources(banks, pension funds, etc.) to originate homemortgages. They often act as middle-men, seek-ing out borrowers who meet the desired risk pro-files of the investors who provide loan capital. Inthese cases, it is not uncommon for the mortgagebanker to retain the servicing of the loan for afee paid by the investor (Barrett, 1992).

Realtors2 are important to mortgage lending fora number of reasons. They act as intermediariesbetween home buyer and seller. They are at the"point of sale" throughout the home-buying pro-cess. Realtors’ success often depends on theirability to help their clients find attractive financ-ing quickly. One study showed that 60-70 percentof home mortgages result either directly or in-directly from realtor referrals (Lewis, 1991). In poll conducted by the Mortgage Bankers Associa-tion, 90 percent of responding realtors said theymade recommendations on financing (Anderson,1987). Realtors and others may act as mortgagebrokers, matching borrowers with lenders for afee. Mortgage brokers can be a significant marketparticipant, and it has been estimated that 45 per-cent of all mortgages were originated by mort-gage brokers (Stark, 1992).

As shown in Figure 1, there are two markets forhome mortgages. Most home buyers are familiarwith what has been described above, called theprimary market, where lenders and home buyersmeet to finance the purchase of a property. Thereis also a secondary market for mortgage money,where similar loans are bundled and sold to largeinstitutions, either to be held in a loan portfolioor perhaps to be used as the basis for mortgage-backed securities. While the primary market isgenerally a local one, this secondary market isnationwide. These three players, plus the smallerprivate firms who compete in the secondarymarket, provide much of the investment capitalfor home financing. The secondary market alsosets loan standards that originators use to pro-duce mortgages that will be readily accepted in-to the secondary market. Mortgages meetingsuch criteria are known as "conforming" loans.The secondary markets make mortgage moneymuch more liquid: loans can be turned into cashwith relative ease, the proceeds going to newloans or other applications.

Research QuestionsMortgage banking is an information intensivebusiness, characterized by local markets andfinancial intermediaries. Considering the im-mense amounts of money moving through thismarket, it is not surprising that some firms havetried to capture more of it using informationtechnology. This section presents prevailingtheories of electronic market coordination andtransformation on which consideration of CLOsin, mortgage banking will be based.

The electronic markets hypothesisBeginning with Malone, et al., some authors havesuggested that there are higher-order benefitsavailable to market participants who rethink andrecast the nature of their organizations’ activitiesbased on the emerging capabilities of IT (Gurbax-ani and Whang, 1991; Malone, et al., 1987). Thebasic argument is that by decreasing many of thecoordination costs associated with doingbusiness both within the firm and outside, IT willprovide opportunities for cost reduction andrevenue expansion that entail either changing thestructure of markets or the boundaries separatingthe firms in those markets. Coordination costsrefer to the resources expended processing in-formation in order to select suppliers, enter intocontracts, schedule deliveries, and other ac-tivities associated with doing business outside theorganization. Specifically, two varieties of markettransformation have been identified that aremade possible by relatively recent improvementsin coordination technology: electronic marketsand electronic hierarchies both coordinate "theflow of materials through adjacent steps in thevalue-added chain" (Malone, et al., 1987, p.485).In markets, the basic forces of supply and de-rnand determine how products and services aretransferred between multiple firms and customersand in what quantities. Customers compare of-ferings from many vendors in order to find a goodmatch for their specific needs with regard to prod-uct attributes, service, price, and other factors.Comparing more vendor offerings is likely to im-prove the match, but also adds to the consumer’ssearch costs. In electronic markets, IT facilitatescustomers’ comparison of purchase alternatives.Generally speaking, electronic markets can im-prove both the amount of information available

254 MIS Quarterly~September 1994


to customers and lower their search costs, enabl-ing volumes and speeds that human middle-mencould not feasibly accomplish.

In traditional hierarchies, there is ownership andcommon, centralized managerial control, andthis, rather than market forces, dictates howgoods and services are transferred between apurchaser and one supplier. Malone, et al. (1987)coin the term "electronic hierarchy" to includethe situation where buyers are linked by com-puters and telecommunications technology to apredetermined source for the product or servicein question. Although this arrangement forfeitsthe access to multiple providers that marketsfeature, it eliminates all the costs associated withidentifying and doing business with more thanone firm. For some firms and some industries,this is a worthwhile tradeoff. As customers movefrom one to many suppliers, the cost of coor-dinating business rises, and conversely, as thenumber of competing suppliers falls, productioncosts tend to rise because of the absence of pricecompetition to encourage firms to carefullymanage production costs. From this analysis, itfollows that if coordination costs are decreased(by IT or some other factor) customers willnaturally tend to favor markets because they willpay lower prices without paying as much in coor-dination costs. Malone, et al. argue that pro-ducers are, in general, motivated to supporthierarchies as the coordinating scheme of choice.Producers do not like price competition with otherfirms, which drives down margins and requiresproducers to focus more of their resources onadvertising and controlling production costs.Most importantly, though, Malone et al. suggestthat the benefits to buyers will increasinglyoutweigh the benefits to suppliers, as IT drivesdown the costs of coordination through electronicmarkets. They propose an evolutionary path thatindustries will follow to electronic markets. Begin-ning with a single source electronic channel("electronic hierarchy," e.g., the early versionsof the American Hospital Supply/Baxter ASAPsystem) the first step is the biased electronicmarket, where suppliers, often the providers ofthe coordinating technology, use the technologyto push customers toward their product or ser-vice, while providing access to other firms’ offer-ings as well. The next step is the unbiasedelectronic market, where all vendors are givenequal chance to win customers based on the

merits of their goods and services. A third andfinal evolutionary step is the personalized elec-tronic market, which provides decision supportfor customers who may find that they now havemore product and vendor information than theycan efficiently search.

EMH predictions for mortgagebankingPrior to CLOs, the primary lending activity ofmortgage banking could be described as beingcoordinated by a local market dominated by in-termediaries. Generally, prospective home buy-ers chose among the loan programs of nearbylenders, perhaps with guidance from newspapePadvertisements, realtors, or possibly mortgagebrokers. In essence, even mortgage bankerscould be considered to be intermediaries, chan-neling investment capital from the secondarymarket to borrowers. It is not surprising that elec-tronic markets and hierarchies have become anissue in mortgage finance, as the market hasbeen characterized by two critical imperfections.First was the geographic fragmentation of themarket. Price competition was intense primarilyamong neighboring lenders. A particularly lowrate offered by a lender in Los Angeles wouldpose little competitive threat to a lender inBoston. Second, the aforementioned intermedi-aries were able to be quite profitable, acting asmiddle men between the huge pools of invest-ment capital in the secondary markets and homebuyers. In some cases, two layers of intermedi-aries, primary lenders and brokers, took a shareof the profits generated by loan originations,presumably leading to higher origination costs toborrowers than would be the case if borrowershad more direct access to the capital markets.

Are home mortgages amenable to being handledin electronic markets? In addition to coordinationcosts, Malone, et al. (1987) argue that lowerlevels of two factors, asset specificity~ and com-plexity of description, will favor markets in a givenindustry. Complexity of description is defined asthe amount of information that must be trans-ferred between buyers and sellers of a givengood or service in order to describe its attributesin sufficient detail that buyers can make an in-formed choice between competing suppliers(Malone, et al, 1987). Malone, et al. specificallysuggest stocks and bonds as examples of pro-



ducts with low complexity of description and largebusiness computer systems as examples of highcomplexity of description (p. 486). Applying assetspecificity to the market for home mortgages, intheory, a loan for $100,000 can buy $100,000worth of home equally well in any real estatemarket, and this is the principle behind borrowers"prequalifying" for mortgages.4 In terms of thecomplexity of description many loans, chiefly thetraditional fixed-rate, 30-year variety, are easilydescribed such that the average home buyer us-ing an amortization table or financial calculatorcan readily determine monthy payments, interestcharged, and principal paid at any stage in thelife of the loan. While other, more complex loans,such as ARMs indexed to volatile standards, re-quire more information to be passed betweenmortgagee and mortgagor prior to closing theloan, even the descriptions of these more com-plex products can generally be accommodatedin the commonly accepted industry terminologyand therefore should not pose a barrier to elec-tronic mediation.

In general, from the beginning of the value-added chain to the end, mortgage banking is con-cerned with managing only one resource: infor-mation. Fund transfers, applications, under-writing, monthly payments--all are, in essence,just transmitting or verifying information.Therefore, based on the information provided bythe EMH proposal, it would appear that CLOswould provide an opportunity for an electronicmarket to arise. Given this, what changes doesthe EMH predict for the home mortgage market?As information technology reduces the unit costsof coordination, markets will be substituted forhierarchies. This happens because relativelylower coordination costs are a traditional sourceof advantage for hierarchies in environments withno or limited information technology. The benefitsof electronic tools in matching buyers and sup-pliers is referred to as the electronic brokerageeffect, and this also contributes to the trendtoward electronic markets.

It suggests that coordination technology providesopportunities to lessen or alleviate the market im-perfections. This could occur, for example, iflenders from across the country included theirloans on CLOs. They could then compete withlenders in any locality where the CLO wasavailable. A related effect of electronic coordina-

tion would be a significant reduction in the searchetlort required on the part of loan shoppers tocompare a larger number of available loans.

It also follows from the general hypothesis thatthere will be fundamental changes in themarket’s structure catalyzed by electronic coor-dination. The hypothesis predicts that financialintermediaries, in this case mortgage brokers andmortgage bankers, are threatened by electroniccoordination and should expect to be hurt or evenrnade obsolete by electronic markets. Further-rnore, the hypothesis suggests that agents whot-’lke advantage of electronic coordination willhave some initial competitive edge over theircounterparts who do not.

The hypothesis predicts that customers, in thiscase borrowers, will be driven by their desire forlower interest rates and closing costs to favorelectronic markets over electronic hierarchies asforms of industry coordination. If this predictionholds true, there should be a migration away fromthe traditional loan selection methods mentionedabove, presumably through electronic hierarchiesand biased electronic markets, to CLOs providingunbiased, efficient national markets for homemortgage financing. This should occur despiteproducers’ (lenders’) wishes to establish hierar-chies that secure a non-competitive distributionchannel for loans. Given consumers’ desires, thisfurther suggests that the predicted evolutionarypath toward electronic markets is from electronichierarchies to biased electronic markets to un-biased electronic markets to personalized elec-tronic markets.

Alternative views of electronicmarkets.Flecently, a number of authors have suggestedalternative views as to how informationtechnology will affect market structures. In par-ticular, they independently argue that the in-troduction of IT may result in fewer’, rather thanmore, suppliers, despite the reduction in transac-tion costs.

"rhe first such objection to the EMH is also rootedin the transaction cost economics literature.~Nhile IT may reduce coordination costs, in-creased coordination can create transactionrisks, specifically, increasing one’s exposure to



opportunistic behavior on the part of the otherparty in the cooperative arrangement. IT canreduce these transaction risks, which, combinedwith the reduced costs of coordination, suggestsa move toward tightly coupled, cooperative rela-tionships. These relationships are neither trans-actions in the spot market, nor are they verticalintegration within the firm, and therefore, they.represent a "middle" position between these twoextremes. Clemons and his colleagues argue foran amended version of the EMH, which they term"the move to the middle" (Clemons and Row,1992; Clemons, et al., 1993). Clemons, et al. fur-ther.decompose transactions costs into coordina-tion costs, opportunism risk, and operations risks(e.g., uncertainty regarding the quality of the pro-duct being supplied), and argue that the use ofIT is favorable toward all three dimensions. Theirconclusion is that there will be increased coor-dination (outsourcing), but via a limited numberof long-term suppliers. Therefore, in this view atrue electronic market, as proposed by theoriginal EMH, would be unlikely.

A second objection to the EMH comes from ananalytic model to investigate the effects onmarket structure of the adoption of electronic datainterchange (EDI) in the case of a single buyerwith multiple competing heterogeneous suppliers(Seidmann and Wang, 1993). A simple analysisof EDI might argue that since EDI reduces trans-action costs, all suppliers would agree to par-ticipate in the arrangement, assuming that thecosts to do so are not prohibitive. However, thisis shown to be unlikely to be the case becausemost of the benefits tend to be captured by thebuyer (with some flow-through to the endcustomer). More importantly, there are decliningmarginal returns to the suppliers with eachsubsequent supplier added. This creates a situa-tion whereby the buyer will encourage early sup-priers’ adoption (perhaps even providingsubsidies for adoption), but where equilibriummay be reached before complete adoption by allcurrent suppliers. In general, the price offered tonon-adopting suppliers will be lower than theprice prior to the introduction of EDI, and theresulting cost differentials may result in fewertotal suppliers than was originally the case,despite the fact that EDI has reduced transac-tion costs. Therefore, the introduction of informa-tion technology may not unambiguously moveindustry structures toward a greater amount of

spot market transactions because some firmsmay choose not to participate.

A third objection to the EMH comes from theeconomics literature on incomplete contracts, butwas motivated by empirical observations in theU.S. and Japan that note a general movementtoward fewer, rather than more, suppliers (Bakosand Brynjolfsson, 1993). In particular, Bakos andBrynjolfsson note that in buyer-supplier relation-ships there are likely to be a set of attributes suchas quality, responsiveness, and innovation, thatare "non-contractible investments" due to the dif-ficulty in specifying their levels in advance in acontract. The authors argue that the introductionof IT will increase the importance of these non-contractibles. Supplying firms will only make suchinvestments based on their ability to capture thebenefits ex post, which depends on their relativeex post bargaining power. Therefore, to thedegree to which such non-contractibles are im-portant in the supply relationship, buyers will findit in their best interests to limit the number of sup-pliers so as to provide the remaining supplierswith sufficient bargaining power and the resultingincentives to make non-contractible investments.The net result is a smaller number of suppliers,all of whom have sufficient incentives to investin those non-contractible investments that willultimately benefit the buyer. While drawing fromdifferent economic theories, the Bakos/Bryn-jolfsson and SeidmannANang models providecomplementary explanations for why theresulting equilibrium number of suppliers may besmaller after the introduction of IT.

These alternative views of electronic mediationfocus on the repeated nature of the transaction.They note that information technology, whilerapidly declining on a per unit basis, stillrepresents a significant investment whenemployed on the scale required to transform anindustry. In applying these models to the marketfor home mortgages, the buyer is the "retailer"providing the mortgage to the consumer/homebuyer, while the suppliers are those firms withthe capital to lend.s Therefore, the conclusion ofthe alternative views is that a limited number ofcapital lenders (suppliers) are likely to be offeredvia the CLO because the incentives for both thelenders, and the retailer providing the system willbe to constrain membership in the system.



Computerized LoanOrigination Systems

Models of electronic mortgagemarketsThere are a range of design options for electroniccoordination of this market. In an "electronicmarket" for home mortgages, prospective bor-rowers would be able to use computer technologyto search through and compare various loan pro-grams from awide variety of lenders. In an "elec-tronic hierarchy," these same borrowers couldcompare all of the loan programs of one lender.Under both schemes, subsequent steps (applica-tion, prequalification, and underwriting) in theorigination process could be automated as well,saving time for the borrower and lender alike. Theuser hardware in both cases could be found ina realtor’s office, a mortgage broker’s office, orin the office of a lender.

Figure 1 presented a view of the traditional mort-gage origination process. In order to provide aframework in which to describe the existingCLOs, three new market diagrams that are alter-natives to the traditional model are presented inthis section. Each represents a step along a con-tinuum, away from hierarchies, middlemen, andlocalized markets, and toward efficient competi-tion in an electronic market. It is important tounderstand that these market prototypes areneither discrete nor exclusive possibilities, butrather signposts along a possible evolution. It isclearly possible that some CLOs will exhibitcharacteristics of more than one model.

CL01: The Loan Listing Service

In the CL01 model, only the fender/loan selec-tion process is automated by the CLO system.Application, prequalification, underwriting, andclosing all occur just as they did under the tradi-tional scheme: through the time-consuming,personnel-intensive transfer of many paperdocuments. Figure 2 diagrams the CL01 processand shows a CLO-equipped realtor brokering theloan. While a traditional mortgage broker couldsubscribe to a CLO and broker loans via thatchannel, there are several reasons why Figure2 shows a realtor performing this function. Thefirst reason is the simplest: most CLOs have thus

far been targeted at realtors. System operatorsand lenders favor this approach because it movesthe CLO as a marketing mechanism to the pointof sale of the home purchase that initiates theneed to borrow in the first place. This, in theory,gives the CLO an advantage over traditionalmarketing channels like brokers and newspaperadvertisements, because it gets to the prospec-tive buyer first.

A more important reason for diagramming theCLO1 process this way is that it shows that evenat this initial step toward an electronic market anintermediary entity’s business is hurt or made ob-solete. If home buyers can compare many loansfrom many lenders in their realtor’s office, it isless likely that they will expend the money andeffort to visit a mortgage broker’s office to do ex-actly the same thing, unless the broker providessome other incentive. A CL01 speeds up onlylender selection and therefore does not offer asignificant time savings over the traditionalscheme from beginning the mortgage search un-til closing a loan. The major benefit to consumersis the increased amount of more readily availableinformation, which presumably leads to an abili-ty to choose a less expensive loan.

There are two key benefits to participatinglenders. As mentioned above, the marketing oftheir loan programs now occurs at the point ofsale, conveying a competitive, advantage overnon-participants. Also, CLO1 offers lenders thechance to market their programs more widelywithout incurring bricks-and-mortar expense fornew loan offices. Apart from a shift in theirmarketing focus, CLO1 does not significantlychange the activities of the participating lenders.Once a prospective borrower selects a lender, theprocess is basically the same as the traditionalmodel.

CLO2: The Application Processor

In the CLO2 model, as shown in Figure 3, muchmore of the origination process is automated, andinformation flows both ways between borrowerand lender across the CLO, After the prospec-tive borrower selects a lender and loan, he or shethen uses the CLO to transmit an applicationback to the lender. Document requests,necessary to verify the information provided onthe application, are also carried over the network.



Property~t~

listings

~ome Buyer1 Financial/# ~ \ Closed LoanInf° /-- l ~’~

HN°::c~sng ALm°untt° Selected Loan / ~ ~Bankers |

I ~ / Lending Practices ~Sel I~

~~~~

’Mortgage ~~ ~ ~ Banker ~ Ins~itutio~]

Available

koan~N~es, terms

CLO-equipped Realtor

¯ Mortgage ’brokered’ by Realtor¯ Traditional brokers excluded¯ Home buyers potentially gain

easier access to moreloan info

Primary Market Secondary Market

Figure 2. CL01: Loan Listing Service

Under the CLO2 model, some lenders may pre-qualify borrowers, issuing a legally binding com-mitment to loan money at a particular rate andunder certain terms, provided that the informa-tion in the application and the property in ques-tion are verified. This prequalification can be apowerful bidding tool for home buyers, servingas proof that their offer on a home is backed byproven borrowing power. Another benefit to mort-gagors using CLO2 is that processing time isdecreased significantly in comparison to the tradi-tional process.

CLO2 includes the features and implications ofCLO1, plus several others. Much of the workassociated with originating loans, including tak-ing applications and processing document re-quests, moves from the lender to the realtor. Inexchange for taking on this added responsibili-ty, the realtor receives some portion of theorigination fees assessed by the lender to the bor-

rower. An additional bonus to the borrower is thathe or she can now make status requests over theCLO. Because the system tracks loan status, thereply can be generated automatically, decreas-ing labor effort for the lender and providing aquicker response to the borrower. It is also moreconvenient for the borrower to have the realtoracting as a clearing house for information aboutthe status of both purchase bids and loan re-quests. Lenders benefit because they canoriginate mortgages faster and more economical-ly. Much of the approval process is automatedby including decision logic in the system. Creditchecks are conducted via electronic ties to creditservice bureaus. Although the final decision toapprove or reject a loan application remains inthe hands of a human underwriter, the CLO2system allows the human participants in the pro-cess to spend less time coordinating information,freeing them to spend more time on individualloans or processing a higher volume ofapplications.



Paper Documents ~~Home Buyer ~ ~ \ Closed Loan

Financial Info, / "~ Paper~. . I

Housing Amount to ~/" ~ ~)ocuments~ I Inves,men,Needs borrow, /~ ~__~.~ "~ ...... ~ .J Bankers

Other Criteria / o,,up \ ~ ~,I Status ~.. / for Loan \ \~ /~

.L Requests’~’_.. I ( ,o_~_,_ .~ ERA, FNMA, / - .. \ Mu,,gagel. v \LUa,,/

f .. -- /ransml[ Banker I~(~hop for---’~ Application, " V

~Other Docs

i -- [Other Institutions J

.~

Loan Info,

\ Monitor Doc Requests

\ Status Status Info

.P.r.o.perty ~ \ / .Attributes of CLO1, plusiis,lngs

"~ ~ .. I -Diminished role for lender,~ Selection & expanded for Realtor

~ Processing Info\ I Proces ":Realtor receives origination fee

CLO-equipped Realtor .~_~ Loan processed faster

!< Primary Market ~ I < Secondary Market-

Figure 3. CLO2: Applicat~on Processor

CL03: The Transformed Market Case studies of CLOs

CLO3 represents the ultimate realization of thegeneral electronic market model as applied tohome mortgages. The CLO links the realtor’s of-fice directly to investment capital available in thesecondary market. The realtor performs the en-tire origination, closing the loan using a credit linemade available through the CLO operator. TheCLO system, in turn, bundles loans for sale intothe secondary market. As shown in Figure 4, bothintermediary entities from the traditional model--mortgage brokers and mortgage bankers--areexcluded.

As in CLO2, the realtor, now the realtor/lender,is the conduit for information about both the loanand the home purchase. This model offers fasterand cheaper loan processing because more ofthe process is automated, and there are fewermiddle-men handling information.

Tllis section presents an account of the ex-periences of five leading CLO systems, basedprimarily on secondary sources.6 The fivesystems are presented in the order in which theywere introduced:

¯ First Boston’s Shelternet¯ PRC Advanced Systems Inc.’s LoanExpress¯ Rennie Mae, developed by the realtors’ trade

association¯ Prudential’s CLOS¯ Citicorp’s Mortgage Power Plus

While there may have been other such systemsover the past decade (most of them quite localin scope~) these five are believed to representthe most significant initiatives in the decade priorto the amendment of the Real Estate SettlementProcedures Act of 1974 (R.ESPA) effective inDecember of 1992.6 These "first generation"systems form the basis for the analysis of the



Paper Documents./*~~ ~Closed LoanPaper

~ App~d L~O.~~ ~en~

F nancia ’nfo, ~:ar::ma~ l LMe°nr~gearge ~’~~1

Amount to borrow Application,Other Criteria Other DocsStatus Requests

Monitor .

I ~. Status /

~ ~

Propertylistings

Housing

~

!4 Primary Market -~ I ~

InvestmentBankers

FHA, FNMA

Other Institutions

¯ Realtor replaces lenderoCLO links borrower,

Realtor/lender,secondary market

¯ Very fast processing

Secondary Market

Figure 4. CLO3: Transformed Market

EMH. Table 1 summarizes the basic data aboutthese five systems.

Shelternet (First Boston Capital Group)

Shelternet was developed to be a nationwideautomated network offering home buyers accessto mortgage financing from a wide variety oforiginators from across the country. At its peakin 1985, Shelternet processed approximately $1billion worth of home mortgages, from 125originators to borrowers in 44 states, using 140realtors’ offices as the points of contact(Economist, 1986; Runde, 1986). In addition having the greatest geographical scope,Shelternet was one of the few CLO systems thattook borrowers through all five steps of the mort-gage process: lender/loan selection, application,prequalification, underwriting, and origination.Shelternet received considerable positive publici-ty in the literature on strategic informationsystems (Wiseman, 1985). Only more recently

has there been any follow-up study (Kemerer andSosa, 1991).

Shelternet used as input a borrower’s financialdata, the cost of his or her intended home pur-chase, estimations of homeowner’s insurancepremiums and property taxes, and the amountof money the borrower had available for a downpayment and closing costs. The system appliedthis information to various loan programs andcould immediately generate monthly payments,amortization schedules, closing costs, privatemortgage insurance premiums, and even the taximplications to the borrower. Within" an hour oftaking an application for a selected loan, thesystem could prequalify the borrower, issuing aloan guarantee good for 60 days. The systemthen automatically generated and mailed pro-cessing documents, such as appraisal requestsand salary verification forms, to the underwriter.Loan processing took 15 to 20 days, during whichtime Verex Corporation, a private mortgage in-



Table 1. Basic Data of Cases

Citicorp’sFirst Boston’s PRC’s AFN’s Rennie Prudential’s Mortgage Power

Shelternet LoanExprsss Mae CLOS Plus

YearStarted 1981 1983 1985 1988 1989

Approx. $1 Billion/year $250 Million/ $1 Billion/year $120 Million/year $9 Billion/yearPeak yearLoanVolume

Parent Investment Bank System Provider System Provider Insurer/Real CommercialFirm Estate Bank

Peak # of 125 40 100 Approx. 5 perLenders geog. market. 1

surer under contract with Shelternet, wouldassess the loan and either approve or reject itfor insurance. If the loan closed and was ap-proved for insurance, Shelternet would then buyit from the originator and sell it to an investor.Shelternet typically owned each loan for approx-imately 15 days.

Shelternet was developed by Andreas Kissal inthe late 1970s, while he was working for the mort-gage finance subsidiary of a construction com-pany (LaGesse, 1984a). First Boston, theinvestment banking firm, purchased the systemfrom Kissal in 1981 and subsequently investedroughly $10 million. First Boston developed theproject in secret, presumably to gain competitiveadvantage and to defer the negative reactionfrom mortgage bankers who were customers ofother First Boston products or services. As lateas a year prior to roll-out, First Boston represen-tatives denied the project’s existence (Guenther,1983). The system was rolled out nationally inApril of 1983 (Brownstein and Lore, 1984).

Most sources agree that Shelternet was originallymarketed to realtors as a conduit to connect them(and their clients) directly to the capital markets,bypassing mortgage lenders in the process(LaGesse, 1984b). Kissal, hired by First Bostonto run the network, publicly disputed that thesystem was designed to exclude traditional mort-gage lenders from the loop. What is certain is thatFirst Boston met with enormous resistance fromthe mortgage banking community, which initiateda boycott of First Boston and Verex’s services(Basch, 1985). The roll-out also raised regulatory

issues, regarding the legality of realtors accept-ing fees for loan originations.

Either to ameliorate the mortgage bankers’ con-cerns or simply to provide more primary mort-gage funds for the network, First Boston quicklytempered its revolutionary stance, focusing onproviding service to the traditional mortgagelenders and selling the software to other networkmanagers. By 1986, private-label versions of thesystem were in operation in the realty offices ofColdwell Banker, Century 21, Better Homes andGardens, Finance Partners and Realty World,who together accounted for 30 percent of theresidential market (Nelson, 1986). Shelternet pro-per originated $800 million in home mortgagesin 1984, $1 billion in 1985, but only $650 millionin 1986. By 1987, the system’s origination ratehad dropped 50 percent since the peak, and FirstBoston cut the Shelternet work force by 20 per-cent (NMN, 1987). By the end of 1988, the systemhad been repositioned to deal solely with thesecondary market, acting as a conduit betweenoriginators and investment banks wishing to sellmortgages to their clients (Miller, 1989;Petramala, 1988).

Shelternet was the first CLO to receive much na-tional attention and was therefore the impetus formany other systems and much of the controver-sy surrounding CLOs in the 1980s. As originallyconceived, Shelternet came as close to a CLO3system as any has since. First Boston, an invest-ment bank, seemingly intended to build a directpipeline between its vast pools of investmentcapital and home buyers, bypassing mortgage



lenders and mortgage brokers in the process(Guenther, 1983). This may account for the highlevel of secrecy surrounding the project in its ear-ly stages, as First Boston may have hoped todefer the negative reaction of traditional mort-gage lenders, many of whom were customers ofother First Boston services. The mortgagelenders’ eventual boycott is just one indicationof how serious a threat Shetternet was perceivedto be. First Boston had threatened the homemortgage industry with drastic change, and theindustry resisted: the middle-men wanted topreserve their positions, the realtors were reluc-tant to take on additional work to learn and runthe system, and the home-buying public did notshow a sufficient preference for shopping formortgages electronically. Shelternet’s first raposi-tioning brought the system closer to the CLO2model: facilitating lender and loan selection andapplication processing, but keeping traditionalmortgage lenders in the loop.

In terms of the EMH, there were aspects of a per-sonalized electronic market evident in Shelternetfrom the beginning: given a prospective bor-rower’s financial data, the system could generatea list of loans that the user might be interestedin and qualify for. However, because Shelternetoriginally carried only First Boston mortgages,"personalized electronic hierarchy" might be amore accurate descriptor. The system’s firstrepositioning, which expanded the loan programsoffered to include other lenders, is very much inkeeping with the tenets of the EMH: there wasindeed a powerful impetus to move from hierar-chy to market. But, it seems to have primarily oc-curred as a result of pressure applied bycompeting suppliers who saw their businessthreatened.

The EMH makes no provisions for looking at thefailure or devolution of electronic markets, andas such offers little guidance for assessing thesecond and final repositioning of Shelternet to awholesale system dealing solely with customersin the secondary market. However, three case-specific factors seem worth noting. First,Shelternet was expensive to join and use relativeto the low or non-existent costs realtors paid torefer clients to local lenders, and as such maynot have commanded as much interest on thepart of realtors as it might have at a lower pricepoint.9 Second, it is important to remember that

mortgage banking is not First Boston’s primarybusiness, so that when investment banking prof-its fell in the late 1980s, the primary mortgagelending operation was one of the first activitiesto be cut back. Third, First Boston was acquiredduring this period by Credit Suisse. Shelternet’sfinal repositioning may have been to a large ex-tent the result of a new corporate-level concep-tion of the business unit’s function. For these orperhaps other reasons Shelternet represents afailed attempt to establish an electric market forhome mortgages.

LoanExpress (PRC Advanced Systems, Inc.)

PRC Advanced Systems, Inc. (PRC) was a suc-cessful large vendor of Multiple Listing Services(MLS) to local realty boards. LoanExpress, PRC’smortgage network, was originally developed asa support service for the first four steps of thehome financing process: lender/loan selection,application, p~’equalification, and underwriting.The system was introduced in September of 1983and by the end of 1984 carried information about350 loan programs from 40 lenders into morethan 2,000 realtors’ offices (Mariano, 1984).There, home buyers used the system to selecta loan and then had to visit the local LoanExpressoffice to fill out an application and other process-ing documents. The system expanded from theWashington D.C./Northern Virginia market intoTacoma, Phoenix, and Memphis.

LoanExpress was developed in-house and ran onthe same terminals that PRC had already in-stalled in 12,000 realty offices to run their MLSsystem. The system was well-received initiallypartly because for MLS users, there was no costassociated with hardware. Realtors paid $20/month and borrowers paid $150/application. Par-ticipating lenders paid $450/month to be listedplus 1 percent of the value (i.e., 1 point) of anyloans originated over the system. Borrowers likedthe system primarily for the wide range of loansit could access quickly. An interesting, thoughminor, change in the system came when PRCdecided to list loans by lender name rather thananonymously as it had originally. The rationalebehind this change was that it would encouragelocal lenders to participate if they felt that namerecognition would give them a marketing advan-tage over other lenders without a local presence(Naylor, 1985). In September of 1985, PRC an-



nounced a major strategic shift for LoanExpress.Mortgage origination activities were completelyabandoned, leaving the network as an automatedrate listing service. PRC spokespeople describedorigination as being outside of their firm’s corecompetence. The refocused network continuedto expand into new markets, but positioned asa customer-service enhancement for realtors, asopposed to an origination tool.

In terms of the EMH, PRC’s LoanExpress beganlife as a CLO2 system, supporting loan selection,prequalification, application, and underwriting.1°

LoanExpress carried primarily local loans, so thatit effectively automated the existing local marketsin which it operated, rather than provided a com-munication backbone for a national market.PRC’s decision to change from listing loansanonymously to listing them by name appears in-consistent with the EMH prediction that marketswill evolve from biased to unbiased forums, sincethe anonymous listings exhibited somewhat less"tilt" than the lender-named listings. On the otherhand, despite the widespread perception of mort-gages as a commodity, perhaps consumers doplace a value on certain "brand name" lendinginstitutions. In this view the lender name may beseen as part of the product description.

After just two years of operation, PRC announceda drastic change in LoanExpress’ positioning. Allorigination activities were dropped, moving thesystem from being a CLO2 to a new orientationas a CLO1. Under its new definition, Loan-Express became basically just a multiple listingservice, which was PRC’s business prior to thesystem’s debut. Like Shelternet, this againreflects a failure of an attempt to establish anelectronic market.

Realtors National Mortgage Access System(American Financial Network)

The Realtors National Mortgage Access System,nicknamed :’Rennie Mae," was developed by theNational Association of Realtors (NAR), theU.S.’s largest trade organization, for use bymember brokers. As originally conceived, Ren-nie Mae was to be a non-profit service that wouldautomate the first two stages of the home mort-gage process--lender/loan selection and applica-tion. The system also allowed loan applicants totrack the progress of their applications. Unlike


Shelternet and systems like it, Rennie Mae in-itially left the processing, underwriting, origina-tion, and securitization tasks to traditionalmortgage lenders. Although intended for use byNAR members across the U.S., after three yearsof operation only a few states had many subscrib-ing realtors. However, despite this limitedgeographical presence, Rennie Mae processed$182 million in home mortgages in 1987(Roosevelt, 1988b). By 1990, the system had ex-panded to more than 100 lenders, 2,000 loan pro-grams, and over $1 billion in loan volume(Finkelstein, 1991).

In its initial configuration, Rennie Mae was essen-tially analogous to the Multiple Listing Servicethat many local realty boards use. It allowed bor-rowers to compare rate and fee information, sort-ing loan programs by type, rate, loan-to-valueratio, or other criteria. Once a loan was selected,an application could be transmitted using a stan-dard form. The system sent Federal Express anelectronic order to pick up the original signeddocuments and deliver them to the lender. Fromthen on, the loan was a matter between thelender and borrower.

NAR developed the system in cooperation withSolomon Brothers at a cost of approximately $3million (Guenther, 1986). Rennie Mae was firsttested in San Diego, with 40 realtors and 25lenders participating. The response wasfavorable, and a limited national roll-out into fiveor six key markets was planned. The feesassociated with Rennie Mae were initially quitemodest, in keeping with the system’s position-ing as a "public utility" focused primarily on sup-porting NAR members. Shortly after the SanDiego pilot, in late 1985, NAR signed an exclusivelicensing agreement with American FinancialNetwork (AFN), a Dallas-based firm, to marketand manage the system. The licensing arrange-ment represented a major shift of focus for Ren-nie Mae, from non-profit to for-profit and fromlisting service to origination system. A complex,unorthodox fee structure was adopted to insureprofitability for AFN and circumnavigate regula-tions regarding origination fees collected byagents not approved by federal licensing boards.Lenders agreed to take 70 basis points (.7%) offtheir origination fees, which would instead bepaid to AFN, which in turn would distribute upto 150 basis points to the originating realtor. So,


on a loan of $100,000 originated through Ren-nie Mae, AFN would get $700, $500 of which itwould pay the realtor.

A unique aspect of Rennie Mae was its reposi-tioning from non-profit to for-profit and from listingservice to full-blown origination system at a timewhen other CLO systems were retreating fromorigination activities. Rennie Mae’s strategic shiftcame at a time when other networks, such asPRC’s LoanExpress and Shelternet, were back-ing away from direct participation in the origina-tion process. AFN felt that Rennie Mae would beless susceptible to resistance from the traditionalmortgage lending community because the parentcompany, unlike, for example, Shelternet’s spon-sor, First Boston, was not in the business oforiginating loans. Initially, Rennie Mae was slight-ly less than a full CLO2. It facilitated loan selec-tion and transmitted applications, but went nofarther into the origination process. Perhaps see-ing Shelternet’s chilly reception deterred NARfrom initially choosing a more aggressive posture.

As the system that appears closest to offeringsomething akin to an electronic market, it is in-teresting to note that Rennie Mae was developedby the realtors association, a financial in-termediary. It seems clear from the pricing of thesystem that it was not initially intended to makemoney solely through loan originations (Bender,1985).11 The NAR’s adoption also makes sensein the context of their goal of selling properties.Providing ready access to financing is a key waythat realtors can encourage the sale of moreproperties, thereby generating more sales com-mission revenue. Via Rennie Mae, realtors canbe seen as trying to usurp the role of the mort-gage intermediary in order to promote their corebusiness.

Computerized Loan Origination System(Prudential Real Estate Affiliates)

In 1987, the Prudential Insurance Company (Pru)announced it was going into the real estate fran-chise business and that by 1993 it would sell3,000 franchises, placing it between Century 21and Coldwell Banker in number of member of-fices (Sichelman, 1987). One component of Pru’sfranchise strategy was a comprehensive com-puterized broker support system that franchiseeswould be required to install and use. One of itssubsystems was CLOS, an electronic mortgage

network that gave the prospective home buyeraccess to loan programs from different lenders.CLOS supported the first three steps of the homefinancing process: lender/loan selection, applica-tion, and prequalification. In January of 1988,Prudential Real Estate Affiliates (PREA) had 180members, and CLOS went into operation, carry-ing six national lenders (Roosevelt, 1988a). 1992, PREA had installed CLOS in 700 local real-ty offices and had expanded the lenders carriedto include regional players, so that in any givenoffice a home buyer might have access to loanprograms offered by three to five lenders (Thomp-son, 1992). It currently offers just two nationallenders, Prudential and Countrywide Funding(Quinn, 1993).

CLOS’s loan selection function allowed bor-rowers to enter selection criteria such as apreference for an adjustable rate mortgage, forthe lowest closing costs, or for the lowest month-ly payments. The system would then displayanonymously the loans that best matched thespecific criteria. Once a loan program wasselected, the borrower could then transmit an ap-plication electronically. Approval came in two tofour days and closing in 20. Many loan docu-ments were not transmitted electronically. Under-writing and closing documents were handled bythe selected lender. PREA spent $3.5 milliondeveloping the system. CLOS is perhaps a pro-totypical example of the CLO2 model, im-plemented with a small number of participatinglenders by a non-lender firm. CLOS is uniqueamong electronic mortgage networks in that itwas developed as a tool to help its parent get in-to the real estate business, rather than the mort-gage lending or brokering business. Pru was alatecomer to the franchise market, and as such,can be seen as needing a product or service likeCLOS to lure top franchisees into its stable. Thenetwork was included in the franchise purchaseprice, which was "in the low $20Ks." (Sichelman,1987). There were no transaction or monthly feesto the realtors. Lenders paid PREA $450 for eachapplication transmitted, and PREA forwarded$100 of that to the originating franchisee, therebyattracting negative attention from the same mort-gage bankers, regulators, and citizen’s groupsthat had been criticizing CLOs since 1981(especially Citicorp, see below). PREA obtaineda letter from HUD stating that their fee structurewas not in violation of RESPA, but the system



remained controversial. The Pru system con-trasts with the earlier systems in that it seems tohave changed little, except for scale, since itsinception.

In terms of the EMH, like some of the others, theCLOS system offered some features associatedwith a personalized electronic market. Borrowerscould use the system to screen loans based ontheir personal selection criteria. The system’sdesigners were committed to providing an un-biased market and to that end built CLOS as avendor-blind system, where loans were displayedanonymously until one was selected. However,the small number of loan suppliers offered is mostconsistent with alternative theories that argue thatsuccessful electronic markets will inherently of-fer a limited number of suppliers, in order to main-tain the appropriate levels of incentives.

Mortgage Power Plus (Citicorp Mortgage)

Among the five CLOs examined, the newest,Mortgage Power Plus, is also the most biasedtoward one particular lender: Citicorp Mortgage,the only lender carried on the system. By 1990,it was also the fastest, returning a legally bindingloan agreement to the realtors’ offices in 15minutes and closing many loans in three days(Lewis, 1991). Citicorp Mortgage’s MortgagePower program, minus the "Plus," was startedin 1981 as an automated system to originate lowdocumentation jumbo loans12 at an acceleratedpace to back the Citicorp parent’s privatemortgage-backed securities. By 1989, MortgagePower had 4,000 member realtors in 37 stateswho processed approximately 75 percent ofCiticorp’s $11.9 billion home mortgage business.Loan closing took from 12 to 15 days (Miller,1989). In 1989, Citicorp initiated a pilot programcalled Mortgage Power Plus, designed to shortenthe time required to close loans even further. Theidea behind the enhanced functionality was anelectronic linkage between the Mort Power main-frame and credit verification firms. Immediatelyupon receiving an electronically transmitted ap-plication, the system automatically pulled the ap-plicant’s credit history from several soumes. If thecredit and financial information were verified andmet the loan requirements, the system sent abinding loan guarantee, conditional on furtherverification of all information provided. Threedays later, the loan was closed through the mail.

The 30-40 percent of applications that were notimmediately guaranteed by the system’s ap-proval routine were passed to human analysts forfurther scrutiny. After the pilot, Mortgage PowerPlus became available in Pennsylvania,Massachusetts, and Florida.

Mortgage Power and the enhanced Plus systemmay be the CLO that has provoked the most opoposition from industry players and regulators.Perhaps because the network was (a) very suc-cessful and (b) limited to loans offered by theparent, citizen’s groups and other lenders labeledit "anti-competitive," and accused Citicorp ofprice-gouging, especially on loans for lower in-come buyers (BNA, 1990). Mortgage Power alsobecame the focal point for the regulatory disputeover brokers receiving fees for selling propertiesand directing buyers to loans, a problem of dualagency, meaning that they represented two par-ties with conflicting goals. In the press, onegroup, Citizen’s Action, called Mortgage Power"an elaborate kickback scheme." Citicorpwithdrew from the Mortgage Bankers Associa-tion, apparently over the CLO dispute. Despitethe external resistance, Mortgage Power wasvery successful, originating more loans than anyother network. In 1991, Citicorp Underwent amassive restructuring, and Citicorp Mortgagewas dissolved. Home financing came under theaegis of the regional Citicorp consumer bankingoperations. Citicorp publicly announced that thenetwork would not be abandoned in thechangeover (Sichelman, 1991).

Mortgage Power Plus (MPP) is unique among theCLO systems examined in this paper because itwas cor~ceived as and remained an electronichierarchy rather than an electronic market, car-rying only Citicorp loans. Because it was createdto produce loans primarily to provide raw materialfor Citicorp’s mortgage-backed securities, MPPis essentially a hierarchical version of the CLO3model, providing almost direct access to thecapital markets and excluding other lenders fromthe loop. Because it was perceived to be so com-pletely anti-competitive, MPP met with strong op-position from government regulators, citizensgroups, and competing lenders. However, onegroup that did not seem to offer much resistancewas the borrowing public, who made MPP themost successful CLO in existence. This is in-teresting since from the EMH it would be ex-



pected that borrowers would avoid a biasedsystem. Under the current scenario it is hard toimagine that Citicorp would ever seriously con-sider moving to an electronic market by includingother lenders on the system, following the predic-tion of the general hypothesis, unless drasticchanges in the system’s performance or theregulatory climate took place.

DiscussionAnalysis of resultsGiven these case studies in computerized loanorigination systems, what can be said about theapplicability of the EMH? Table 2 summarizes thehistory of five case studies.

In all five cases information technology reducedthe time and effort required on the part of pro-spective borrowers to select and secure a loan,as required by the EMH. In terms of reducingmarket imperfections, there is some evidence forthis to the degree that consumers who use suchsystems compare more alternatives than theywould have without such systems. However,despite the estimated 30 percent of the market’ssales force having access to CLOs, as recently

as 1991 computerized orignations represented"only a tiny portion" of all originations, andtherefore no strong effect can be said to havebeen observed (Lewis, 1991). There is, howeverstrong positive evidence for the prediction thatthe electronic market will be a potent threat tointermediaries, most clearly in the cases ofShelternet and Mortgage Power Plus. Interesting-ly, none of the five systems was developed bya mortgage broker. This is somewhat surprisingsince (a) they are most obviously threatened the CLO development, and (b) this group’s cur-rent primary expertise is searching through theexisting market of mortgage offerings and it mighttherefore be expected that at least one group ofbrokers would have elected to offer such asystem as a way to avoid being squeezed out ofthe process. Instead, what has happened is eitherforward integration (realtors have offered a CLOto their home-buying clients) or backward integra-tion (the mortgage lenders themselves have of-fered systems).

The EMH’s prediction that customer needs wouldbe a key factor driving the evolution toward elec-tronic markets appears inconsistent with the factthat the system with the current largest dollarvolume is an electronic hierarchy, because it

Table 2. History of Cases

Cltlcorp’sFirst Boston’s PRC’s AFN’s Rennle Prudentlal’s Mortgage Power

Shelternet LoanExpress Mae CLOS Plus

System Began as near- Began as CLO2, CLO2 CLO2 ElectronicType CLO3, devolved devolved to hierarchy

to CLO2, then CLO1secondarymarket only

StrategicHistory

Originally PRC tried to Developed by Developed to at- Conceived ofdesigned to cir- leverage ex- National Associa- tract franchisees and operated ascumnavigate perience with tion of Realtors to the newly a one lenderother lenders, Multiple Listing as a non-profit formed Pru Real system, facilitat-who boycotted Service ter- loan listing ser- Estate Sub- ing completeparent firm in minals. Began vice to members,sidiary. Supports origination pro-response, as comprehen- Sold to AFN first three stages cess for CiticorpRepositioned to sive origination after pilot proof origination loans. Met withinclude other tool. After two gram, shifted to process, much regulato.rylenders. Aban- years, refocused comprehensive and citizen’sdoned orgina- as simple infor-profit orgina- group resistance,tions in 1986, terest rate listing tion network, but successfulfocusing on service, nonetheless.wholesaling insecondarymarket.



might be argued that this hierarchy actually pro-motes and enables higher prices whereas anelectronic market enables consumers to haveready access to the prices of many vendors(Bakos, 1991). And neither of the most ag-gressive systems that ventured into the realm ofCL03s--the form closest to a pure electronicmarket--continues to exist in that form. However,it should be pointed out that in none of the fivecases did a market evolve into a hieramhy. Whenchange occurred, it was in the other direction,as per the EMH.

In contrast to the mixed support for the EMH, thealternative versions of the effects of electronicmediation on market structure do seem largelyconsistent with the history of the CLO market.These approaches suggest that a large-scale (interms of number of suppliers) electronic marketfor mortgages would not emerge, as has beenthe case despite more than 10 years of ex-perience with this concept. The alternativetheories’ emphasis on the need for a limitednumber of suppliers to maintain incentives isquite consistent with the retreat of Shelternet andLoanExpress and the recent experience of CLOSand Mortgage Power Plus. However, it is lessclear how these alternative views explain thelarge size of Rennie Mae.

The Bakos and Byrnjolfsson (1993) descriptionof the role of non-contractible investments is apossible explanation for the failure of a full elec-tronic market to develop in the home mortgagemarket, especially if this market is compared withthe much more successful advent of an electronicmarket in the secondary market. Mortgage sup-pliers need to make considerable investments ingathering and analyzing borrower data, such assources of income, assets and liabilities, previouscredit history, preferences in loan terms, etc. Inaddition, they need to make similar types of in-vestments in gathering information about theproperty, i.e., conducting an assessment.13 Tothe extent that there are many suppliers com-pet!ng to provide th~ mortgage, the incentives forany one supplier to make these investments isreduced. The EMH is much more consistent withevents in the secondary market (witnessShelternet’s transformation) where these in-vestments need not be made, because loans aresimply categorized as "conforming" and are thenbought and sold as commodities. Therefore, the

original EMH may prove to be a much morepowerful explanatory model in the secondarymarket than in the primary market. "[’able 3 pro-vides a summary of these results.

Why doesn’t the EMH predict wellfor CLOs?As Table 3 summarizes, the EMH has limited utili-ty in explaining the CLO phenomenon. In orderto better understand the forces at work in theCLO electronic market, it seems useful to con-trast it with the situation where the EMH has beenof greater descriptive accuracy. Malone, et al.make clear reference to the well-known historyof the airline reservation systems, and thesesystems provide the prime example of the evolu-tion from biased to unbiased to personalizedmarkets. How might the market for home mort-gages be contrasted with this more successfulexample?

At the level of the market, airline tickets are soldthrough a relatively simple transaction, oftendirectly to the end consumer from the supplier(the airline), or, at most, through one additionallayer, the travel agent, who is compensated bythe supplier for his or her services. The trans-action therefore appears costless to the end con-sumer. The dominant systems (e.g,, SABRE,APOLLO) were developed by the supplier andrepresented an additional source of revenue. Themarket for aidine tickets was already national inscope, by definition of the travel business, evenprior to the electronic systems. The product isrelatively simple, is purchased with relative fre-quency, and typically does not represent a signifi-cant fraction of a consumer’s disposable income.

In contrast, purchase of a home mortgagerepresents a relatively complex transactionbecause, in part, the purchase of a home mort-gage is often one half of a composite transaction,and the other half is the purchase of a home. Thisseparate but related transaction is typically con-ducted through the aid of another third party, therealtor. Other intermediaries also play importantroles, as diagrammed in Figure 1, and the mort-gage transaction is only the "front end" (theprimary market) for the entire economic process,which includes the sale of the mortgage in thesecondary market. Some systems were devel-oped by parties other than the mortgage supplier,



Table 3. Summary of EMH Results

Major Proposal Results to Date (1981-1993)

Use of information technology will result inthe development of electronic markets forhome mortgages.

Use of information technology will result infundamental changes in market structure;"financial intermediaries will be threatened.

Customers will drive the movement towardelectronic markets over electronichierarchies.

Markets will evolve from electronichierarchies to biased electronic markets tounbiased electronic markets to personalizedelectronic markets.

Limited, with even some contradictoryevidence of the two case studies that couldreasonably be characterized as trueelectronic markets, neither continues toexist in that form today.

While no fundamental changes in marketstructure are apparent, financialintermediaries, in the form of mortgagebrokers and mortgage bankers, have feltthreatened.

No support. The most successful system(the one with the current largest dollarvolume of mortgages) is best characterizedas an electronic hierarchy.

No support.

such as realtors, who were initially prohibitedfrom charging a fee for the system. The primarymarket for mortgages was local in scope prior tothe introduction of electronic systems. Finally, theproduct is complex, is purchased infrequently,TM

and represents a significant financialcommitment.

How might these factors affect the advent of elec-tronic markets? The mortgage product itself maybe relatively hard for suppliers to differentiate.Given the structure of the two markets for mort-gages, the presence of the large and well-organized secondary market may end up exert-ing a considerable amount of discipline on theprimary market. While at one level this might ap-pear to work in favor of the creation of electronicmarkets, further analysis may suggest otherwise.If it is difficult to differentiate the product on adimension like "unique features," then competi-tion will tend to settle around price. Naturally,sellers find it in their best interest to resist suchtrends. And, in the mortgage industry, the ex-isting financial intermediaries may have beenrelatively successful in resisting a change to amore open market structure, owing to their pro-

portionately large and well-organized marketpower relative to that of the consumer. Bakos(1991) argues that electronic marketplaces usual-ly favor the buyers by lowering buyers’ searchcosts, thus reducing the market power and prof-it of the sellers. Sellers will be reluctant to developsuch systems, and in markets such as consumermarkets, where the sellers are highly concen-trated relative to the buyers, they may be ableto slow or even halt their development.

The aspect of market power seems to fit well withthe home mortgage cases. Much of the evolutiontoward electronic markets appears to dependupon the end consumer, who ultimately will bethe primary beneficiary, exerting sufficientpressure to provoke the evolution. In home mort-gages, the home buyers are fairly unorganizedan.d powerless compared to airline passengers,where it is quite common to witness, for exam-ple, large corporations negotiating favorable rateswith air carriers for ticket prices,is Not only arebuyers fragmented, but mortgage transactionshappen infrequently, and therefore there may beinsufficent motivation or opportunity for buyersto organize. Another interesting contrast is that



in the airline example, it was the supplier of thegoods who provided the electronic market. Of thefive systems studied in the mortgage industry, themost successful in terms of dollar volume wasalso provided by a supplier of mortgages--Citicorp. It seems to have been designed to cap-ture additional product sales rather than generatefee income. It is, of course, possible that thissystem will evolve into offering other suppliers’products, presumably for a fee. However, it iseasy to imagine that Citicorp could easilyduplicate another vendor’s offering, since theultimate product is simply money offered undera particular contract. In contrast, different airlinesown the rights to specific routes and airport land-ing slots, and therefore a certain amount ofcooperation is required. All of these differencessuggest obstacles to the creation of electronicmarkets that were not present in the initial airlinereservation system example.

As one might expect, the most adamant detrac-tor of CLO systems continued to be smallerplayers in the mortgage banking community,whose middle-man niche is squeezed byautomated vertical integration. With a few excep-tions, the CLO technology is being explored mostaggressively by firms interested in building elec-tronic hierarchies rather than markets. Thesefirms fall into two groups: realtors hoping to speedloan approvals and capture additional revenuefrom the financing process and lenders hopingto position their loan products closer to the pointof home sales. An additional explanation for theearly slow adoption of an electronic market formortgages may be the consumers’ inability or un-willingness to engage in it given the product’srelative complexity. They may prefer to have anintermediary lead them through the transaction.This might suggest that more sophisticatedsystems that allow greater ease of interaction bythe consumer are needed. This possibility mayprovide a market niche for a new competitor inthe development of an electronic market,is

More recent developments--the.role of government in incentives

However, one recent development that may en-courage further development of CLOs is a

regulatory change by HUD to RESPA.17 Prior toan amendment of RESPA effective in Decemberof 1992, receipt of fees by realtors for matchingbuyers and mortgage providers was prohibited.Now a fee can be charged so long as a serviceis provided, although "kickbacks" for pure refer-als are still prohibited. This is believed by someto lower a barrier to CLOs in real estate offices,because it is now quite clear that the realtor cancharge the buyer a fee to use a CLO (NMN,1992). The Mortgage Bankers Association (MBA)vigorously fought this change and failed in theirattempts to get HUD to either (a) cap the fees thatcould be charaged, or (b) require that the systemsprovide access to multiple lenders. The MBA hasfiled a lawsuit to reverse HUD’s decision(Lehman, 1992). The MBA, although publiclyarguing that it is attempting to protect consumers,is assumed to be motivated by loss of servicefees by its members if mortgage loans areoriginated in realtors’ offices (Lubinger, 1993).Another outcome of the amendment was to lowerbarriers to vertical integration, for example, themerging of real estate brokerages and title in-surance companies,

By December of 1993 these new RESPA rulessupporting increased realtor participation in thernortgage origination process had been in effectalmost a year, and the debate among variousrnembers of the lending and realty communitiesover the new rules showed no signs of quietingdown. Some new CLO systems were being in-itiated by a variety of players: on the real estateside, interest in CLOs seems to be growing. TheNational Association of Realtors was consider-ing a system for inclusion in the Multiple ListingService package available to member realtorsthat would eventually include "hundreds" oflenders and feature on-the-spot qualification(Sichelman, 1993). Master Mortgage LP, a con-sortium of 23 large real estate firms, had a singlelender system in the design stage (Salt, 1993).Large lenders showed interest in CLOs as well.GE Capital (Cornwall, 1993) and Sears Mortgage(Harney, 1993) both initiated systems; Searsforged ahead with a new CLO despite a $700Ksettlement of a lawsuit alleging that Sears realtyoffices unfairly steered borrowers to Sears Mort-gage loans using a discontinued CLO system(Downey Grimsley, 1993).



Suggestions for future practiceand researchWhat, then, are some of the most generally ap-plicable lessons for practice inherent in this casestudy of the electronic market for home mort-gages? Some of the aspects of the EMH appearto be validated by these examples, in particularthe danger to current market intermediaries ofinformation technology. Therefore, anyone in therole of a market intermediary should carefully ex-amine the likelihood of their continued economicexistence under the intervention of informationtechnology. More proactively, they might ex-amine whether to be the first to provide the struc-ture for the electronic market, and thereby securetheir place in the transaction.

On the other hand, the failure to achieve a fullelectronic market, despite a significant amountof effort, suggests that significant barriers can ex-ist to this change, barriers that must beacknowledged and accommodated by any partywho wishes to create an electronic market. Oneclear lesson from the alternative views of theEMH, and one that appears consistent with thecase of home mortgages, is the need for ap-propriately aligning the incentives in the market.Suppliers only have incentives to participate inelectronic markets where they can either differen-tiate their products or directly compete with arelatively small number of other suppliers. Choos-ing the participants to be included in the marketmay be the most critical decision in establishingsuch an electronic market.

This is clearly related to the role of market power,a notion that is under-emphasized in both theoriginal EMH and the alternative views. Few realworld markets exhibit the perfect characteristicsof markets described in neoclassical economics.If such theory is to be directly applied to a specificmarket, then that market’s local imperfectionsmust be taken into account. In the home mort-gage market the mortgage lenders and therealtors exhibit a tremendous degree of marketpower. In addition, this mad~et is strongly affectedby government regulation, as witnessed by theimpact of the HUD regulatory changes and thehigh degree of political lobbying that precededsuch changes (Markus, 1983). The writings the effect of information technology on marketstructure to date have largely ignored the role ofgovernment regulation. Table 4 lists some of

these possible factors to consider in predictingthe likelihood of the development of an electronicmarket.

Table 4. Possible Electronic Market Factors

¯ Complexity of the transaction.*¯ Frequency of transaction¯ Relation of goods/services supplier and

system supplier¯ Current market structure¯ Relative power of buyers and suppliers¯ Possibility of transaction fees¯ Ability of suppliers to match competitors’

product offerings

This is in addition to the EMH’s proposed com-plexity of product description.

Finally, this analysis of the electronic market forhome mortgages raises some additional ques-tions for future research. One is the identity ofthe provider of the market. In the case of thedominant airline reservation systems, the pro-vider was also the supplier of the goods. In thelargest dollar volume CLO, this was also the case.Is there an important relationship here, or is thisa simple coincidence? After all, the fact that theairline systems were developed by the airlinescould be argued to be an historical accident,stemming from the airlines’ early familiarity withthe then exotic technology of computers as afunction of their operational needs to managetheir highly perishable inventory of seats.Hypothetically, if airline reservations systems didnot exist today, could they be successfullycreated by a travel agency, serving as an elec-tronic market provider, or would such a systembe successfully resisted by the airline supplierswho would be necessary participants in such asystem? If airline reservations systems were tofail to arise, then the identity of the market pro-vider might be construed to be an important fac-tor. Further research in other markets may shedlight on this issue.

Concluding RemarksFrom the analysis in the previous section it ap-pears that at the present time CLOs providelimited, support for the Electronic Markets



Hypothesis. Is this an appropriate test? Malone,et al. (1987) suggest the need for such tests:"... our forecasts are based on a simple concep-tual analysis rather than on systematic empiricalstudies. A conclusive test of our model and ourpredictions will, therefore, require further em-pirical and analytical work" (p. 484). Home mort-gages are a financial product like stocks andbonds that have previously been suggested asgood examples of electronic markets at work. Thetechnology necessary for CLOs has been in placefor over a decade and, therefore, arguably hashad time for the forces to work. At least five ma-jor economic agents have offered CLOs, yet noneexists today as a pure electronic market as sug-gested by the hypothesis. Of course, this areacontinues to evolve, and such a market may yetemerge. The EMH does not give guidelines onhow long it will take for an electronic market toemerge in an area like home mortgages. Whatthe current analysis points out is that such anevolution is by no means quick nor is it asstraightforward as might be imagined from acasual interpretion of the EMH.

However, it should not pass unnoticed thatMalone, et al. caution the reader: "In addition tothe changes in information technology that wediscuss here, there are, of course, other impor-tant fomes--such as changes in stock prices, an-titrust regulations, and interest rates--that mightaffect firm and market structures" (p. 484). Clear-ly, no single variable-based theory, even if in-tuitively appealing, will successfully predict allevents in a phenomenon as complex as the struc-ture of markets. And the EMH has been suc-cessful in informing people about the significantpotential impacts of information technology. Whatis needed is further empirical study leading toaugmentation of our understanding of the impactof information technology on market structure inorder that future events may be forecast with in-creasingly greater accuracy.

AcknowledgementsResearch support from the Center for Informa-tion Systems Research (CISR) at MIT is gratefullyacknowledged. Helpful comments were receivedfrom Y. Bakos, E. Brynjolfsson, E. Clemons, G.Hurst, T. Malone, A. Seidmann, and participantsat the Twenty-Sixth Hawaii International Con-ference on System Sciences.

Endnotes

1This section may be skipped by readers familiar with thehome mortgage industry. Alternatively, readers interested ingreater background detail than is provided here are referredto Hess (1992).

=~The term "realtor" is a designation for a set of real estatebrokers who have met the requirements for this professionalcertification. While not all real estate brokem are realtors, theterm realtor will be used in this paper so as to avoid any con-fusion with the term "broker" between real estate brokersand mortgage brokers.

:~ Asset specificity refers to the degree to which an asset canbe redeployed to alternative uses and by alternative userswithout sacrifice of productive value (Williamson, 1989).Williamson’s examples of asset specificity include sitespecificity, e.g., the physical location of an asset chosen toeconomize transportation costs for a particular customer, andphysical asset specificity, e.g., specialized capital equipmentdesigned to optimally produce a very specific product or lineof products. Other types of asset specificity include humanasset specificity (specialized knowledge arising from learn-ing by doing), dedicated assets (discrete investments general purpose assets at the behest of a specific customer),and brand name capital, which may not be easily transferredto another product.

4 However, it should be noted that once the actual mortgageis finally approved, the general funds are transformed intoa contract with a single buyer for a single property and arenon-transferable.

5 Note that this generic description is purposeful; while in thetraditional market the "retailer" is most often the mortgagebanker (see Figure 1), with electronic mediation the retailercould be a different party, such as a realtor.

~ Data collection spproaches are described in the Appendix.7E.g., ColdwelI-Banker offers a service called "Borrower’sChoice" in some offices in parts of New Jersey (Quinn, 1993).

~See the discussion section below for examples of post-RESPA amendment systems.

~ The fee structure to member realtors was as follows: $16,000initiation, which included the hardware, software, andmarketing materials, $100 per hour usage fee, $500 stafftraining fee, $100 per month maintenance. Offsetting thesefees were the origination fees from the borrowers, which inthe traditional lending process would go to a hometown lenderor mortgage banker (Guenther, 1983).

1o The system did not, however, alleviate as much borrower ef-fort as other CLO2s, because it required borrowers to visita PRC office to apply for a loan after selecting one througha terminal in a realtor’s office.The fee to connect to the network was $50, and there wasa $10 per month use fee. Status check requests cost $5 each,as a small disincentive to apply for multiple loans (Bender,1985).

12Jumbo describes mortgages that do not conform with Fred-die Mac or Fannie Mae guidelines because their balancesexceed the limits established by the quasi-public secondarymarket makers.In addition, it might be expected that local firms will have ac-cess to better information, and therefore face less risk, thannon-local firms.



Although the recent low levels of interest rates have prompteda flurry of refinancing that has made this transaction a morefrequent event for many consumers.Although it should be pointed out that one exception to thisis the case of corporate re-location services, which can, dueto their large volume, sometimes negotiate relatively morefavorable rates.

lePersonal communication, G. Hurst, January 1994.17For a useful history of RESPA, see Bernstein (1993).

ReferencesAnderson, K. "CLO Update: Bring Consumer

Realtor and Lender a Step Closer," MortgageBanking, 1987, p. 154.

Bakos, J.Y. "A Strategic Analysis of ElectronicMarketplaces," MIS Quarterly (15:3),September 1991, pp. 295-312.

Bakos, J.Y. and Brynjolfsson, E. "From Vendorsto Partners: Information Technology and In-complete Contracts in Buyer-Supplier Rela-tionships," Journal of OrganizationalComputing (3:3), 1993, pp. 301-328..

Barrett, A. "Countrywide’s Home Sweet Loans,"Business Week, September 14, 1992,pp. 86-87.

Basch, M. "Living with a Loan Network Lion,"American Banker, April 29, 1985, p. 22.

Bender, R. "The Home Guide of the FutureSpeaks Computerese," American Banker,April 29, 1985, p. 26.

Bernstein, L.A. "RESPA Changes Wreak Havocon Safe Harbors," New Jersey Law Journal,January 25, 1993, p. $10.

BNA. "Survey of Mortgage Power ProgramHighly Inaccurate Citicorp Charges," BNA’sBanking Report, November 12, 1990, p. 780.

Brownstein, P, and Lore, K. "Explosion Con-tinues in Secondary Mortgage Market," LegalTimes, February 6, 1984, p. 15.

Clemons, E.K., Reddi, S.P., and Row, M.C. "TheImpact of Information Technology on the Organ-ization of Production: The ’Move to the Middle’Hypothesis," Journal of Management Informa-tion Systems (10:2), Fall 1993, pp. 9-35.

Clemons, E.K. and Row, M.C. "InformationTechnology and Industrial Cooperation: TheChanging Economics of Coordination andOwnership," Journal of Management Informa-tion Systems (9:2), Fall 1992, pp. 9-28.

Cornwell, T. "CLO Supporters Say Rural Areaswill Benefit Greatly," National MortgageNews, July 12, 1993, p. 25.

Downey Grimsley, K. "Coldwell Banker SettlesHome-Sale Closing Case; Sears AffiliateAgrees to Pay $700,000," The WashingtonPost, May 22, 1993, p. El o

Economist. "Who’s Who of Networks," TheEconomist (U.S. Edition), June 14, 1986, p.

Finkelstein, B. "CLO Buys Mortgage Banker."National Mortgage News, July 8, 1991, p. 9.

Guenther, R. "First Boston’s Mortgage Plan StirsUp Real Estate Industry," Wall Street Jour-nal, April 6, 1983, p. 1.

Guenther, R. "Computerized Mortgages HitSnags with Lenders Brokers," Wall StreetJournal, March 19, 1986, p. 33.

Gurbaxani, V. and Whang, S. "The Impact of In-formation Systems on Organizations andMarkets," Communications of the ACM (34:1),January 1991, pp. 59-73.

Harney, K.R. "Loan Overages Come UnderFire," Los Angeles Times, August 15, 1993,p. K4.

Hess, C. Electronic Markets For Home Mort-gages, unpublished MIT Sloan School ofManagement S.M. thesis, 1992.

Kemerer, C.F. and Sosa, G.L. "SystemsDevelopment Risks in Stratregic InformationSystems," Information and SoftwareTechnology (33:3), April 1991, pp. 212-223.

LaGesse, D. "In the Exploding World of Elec-" tronic Mortgage Marketing RKI Databank’sLee is the Sort Who Lights the Fuses,"American Banker, January 23, 1984a, p. 22.

LaGesse, D. "Realty World Plans a Computer-ized Mortgage Network," American Banker,October 31, 1984b, p. 1.

Lehman, H.J. "New Rules on Settlement Pro-cedures Could Alter Buying, Borrowing Pro-cess," Washington Post, December 12, 1992,p. El.

Lewis, J. "Long Fought RESPA Debate May beResolved by Fall; Compromise HUD Rule Ex-pected in October," Mortgage Commentary,August 23, 1991, p. 1.

Lipman, J. "Home-Buying Process is ChangingRapidly Because of Technology," Wall StreetJournal, January 25, 1984, p. 1.

Lubinger, B. "HUD Revises Its Rule on MortgageReferrals," Plain Dealer, January 18, 1993,p. lF.

Malone, T., Yates, J., and Benjamin, R. "Elec-tronic Markets and Electronic Hierarchies,"Communications of the ACM, June 1987,



pp. 484-497.Mariano, A. "Revolution Under Way in Real

Estate Sales; Computers Deregulation Shak-ing Up Industry," Washington Post, February4, 1984, p. El.

Markus, L.M. "Power, Politics, and MIS Im-plementation," Communications of the ACM(26), 1983, pp. 430-~.~.~..

Miller, P. "Citicorp Mortgage Battle ProposedHUD Rules Would Hurt Firm," St. LouisBusiness Journal, March 13, 1989, p. 13B.

Miller, P. The Common-Sense Mortgage, HarperPerennial, New York, NY, 1992.

Naylor, B. "Ground Control to Major Match,"American Banker, April 29, 1985, p. 23.

Nelson, J.F. "Friend or Foe," U.S. Banker,February 1986, p. 34.

NMN. "Drop in Originations Causes LayoffsRise," National Mortgage News, November23, 1987, p. 10.

NMN. "Don’t Buy RE-Referred Loans," NationalMortgage News, November 16, 1992, p. 1.

Petramala, T.L. "One-Stop Shopping for a Homeis Near," New York Times, April 10, 1988, p. 9Section 10.

Quinn, J.B. "One-Stop Shopping For a NewHouse," Washington Post, February 21, 1993,p. H3.

Roosevelt, P. "Computerized Loan SystemReadied; Prudential Unit Shakes Up SomeMortgage Bankers," American Banker,December 9, 1988a, p. 6.

Roosevelt, P. "High-Tech Loan System Grows;But Lenders Question Fairness of Rennie MaeNetwork," American Banker, September 2,1988b, p, 6.

Runde, Ro "Mortgage, Where is Thy Sting?"Money, April 1986, p. 87.

Saft, J.H. "23 Realty Firms to Originate Loansfor a Single National Lender," The AmericanBanker, September 17, 1993, p. 11.

Seidmann, A. and Wang, E. "Electronic Data In-terchange: Competitive Externalities andStrategic Implementation Policies," SimonSchool, University of Rochester Computersand Information Systems working paper92-03, February 16, 1993.

Sichelman, L. "RE Franchise Effort to IncludeCLO Plans," National Mortgage News,November 23, 1987, p. 12.

Sichelman, L. "Mortgage Power Effort Not En-tirely Disbanded," National Mortgage News,

November 18, 1991, p. 1.Sichelman, L. "NAR’s International MLS Could

Expand CLO Efforts," National MortgageNews, November 8, 1993, p. 42.

Stark, J. "Real Estate Rules Draw Criticism," St.Petersburg Times, November 25, 1992,p. 14A.

Thompson, S. National Accounts Manager forPrudential Real Estate Affiliates, phone inter-view, May 1992.

Williamson, 0.E. "Transaction Cost Economics,"in Handbook of Industrial Organization, R.Schmalensee, and R.D. Willig (eds.), North-Holland, Amsterdam, 1989, pp. 135-182.

Wiseman, C. Strategy and Computers: Informa-tion Systems as Competitive Weapons, Dow-Jones-Irwin, Homewood, IL, 1985.

About the AuthorsChristopher M. Hess is information systemsmanager for ValueQuest, Ltd., an investmentmanagement firm in Marblehead, Massachu-setts. He received his B.A. degree in English fromHaverford College and S.M. degree in manage-ment from the Massachusetts Institute ofTechnology. His current focus is on using infor-mation technology to improve investmentanalysis and trade execution in global equitymarkets.

Chris F. Kemerer is the Douglas Drane CareerDevelopment Associate Professor of InformationTechnology and Management at the MIT SloanSchool of Management. He received a B.S.degree in decision sciences and economics fromthe Wharton School of the University of Penn-sylvania and a Ph.D. degree from the GraduateSchool of Industrial Administration at CarnegieMellon University. His primary research interestsare in the measurement and modeling of softwaredevelopment for improved performance. He haspreviously published articles on these topics inCommunications of the ACM, IEEE Computer,IEEE Software, IEEE Transactions on SoftwareEngineering, Information and SoftwareTechnology, Information Systems Research,Management Science, Sloan ManagementReview, and others. Professor Kemerer serveson the editorial board of MIS Quarterly as wellas four other journals.



AppendixEmpirical data for this research were collected primarily through investigation of secondary sources,particularly industry trade journals such as Mortgage Banking, American Banker, BNA’s Banking Report,National Mortgage News, Mortgage Commentary, and U.S. Banker. The primary tools in this investiga-tion were the online search services ABI/Inform and Mead’s Lexis/Nexis. These databases were initiallysearched in December 1991 under the main topic headings "computerized loan origination systems"and "automation of mortgage banking operations," resulting in 51 sources that were then examinedfor relevance to the research topic.

After this initial evaluation, a small number of unstructured interviews with primary sources in the mor-tgage industry were conducted in order to either investigate issues that were incompletely describedin the secondary sources, or to corroborate particularly significant insights gleaned from single secon-dary sources. Additional online searches were conducted in April and November of 1993 in order toupdate the data where necessary. During the course of this research initial drafts of this paper werereviewed by knowledgeable sources to avoid egregious errors.

However, despite the careful nature of the investigation, this approach to data collection remains non-traditional. Therefore, the data have been treated conservatively. In particular, attributions of the motiva-tions of specific parties involved in the systems are presented in the text only circumspectly, given thepublic nature of the data. In all cases direct citations to the source of specific relevant background materialhave been provided, resulting in a somewhat longer than normal Application section bibliography.
