BACKGROUND
This project involves a comparative empirical analysis of the development, adoption, implementation, and diffusion of vertical (or industry-wide) information system standards in different industries. We aim to contribute to knowledge about IT-enabled interorganizational collaboration, which has consequences for national productivity improvement. This work is motivated by the potential opportunities of Internet-based XML standards, which promise to lower the costs of interorganizational collaboration significantly. Achieving these gains, however, will depend on 1) the successful development of higher-level Internet-based interorganizational coordination standards, 2) widespread adoption of these standards by both large and small organizations in specific vertical industries, and 3) appropriate technical choices by these industry participants during implementation of the standards.
Standards development, adoption, and implementation are challenging processes and collective (often at the industry-level) activities. Because an interorganizational interconnection standard must harmonize the activities of many different organizations, no one organization can develop a successful standard for all others: Representatives of all interdependent organizational segments must join in. The extent to which a vertical information system standard diffuses in an industry is also a collective phenomenon, because individual organizations base their adoption choices on what other organizations do.
The actual adoption and implementation of standards, however, are organization-level activities, leading to complex interactions between individual organizational decisions and collective processes and outcomes. Certain technical choices during standards development can make it easier to gain consensus and successful development of an industry-wide standard, while simultaneously making organization-level adoption and effective implementation of the standard less likely to occur—for example, by failing to account for the incentives of commercial software developers. Our aim is to better understand how these processes interact in shaping vertical information system standards creation and use.
This project began with funding from the National Science Foundation's Digital Society and Technology Program (Award Numbers: 0231584, 0233634, and 0323961). Early work focused on the home mortgage industry, including a detailed case study of the Mortgage Industry Standards Maintenance Organization (MISMO). Research activities involved case studies and interviews with industry professionals, a review of literature, analysis of company literature and Web use, and review of available statistical data on the mortgage sector. Collectively, these efforts were oriented towards identifying how emerging uses of the Internet for all facets of the home mortgage industry value chain have influenced the structure and activities of industry participants. Topics include the impact of ecommerce on market structure at different mortgage industry segments such as loan origination, servicing and sales to the secondary market, the evolution of IT standards in the mortgage industry, the use and impact of the Internet in various customer-facing activities, and the impact of industry-level IT standards on mortgage industry practices, performance and structure.
New work extending early mortgage industry findings to other industries is now underway with new funding from NSF's Human Centered Computing program awarded in the fall of 2007 (Collaborative Research: Interorganizational Information Systems Integration Through Industry-wide IS Standardization: Technical Design Choices and Collective Action Dilemmas (NSF award numbers: 0704629, 0704978, 0705186). Over the past two years, in-depth interviews and analyses have been conducted in the mortgage industry, the automotive industry, and the apparel retail sector to enable comparisons of vertical information system standards development and diffusion across industries.
A number of papers are now available from the VISTA Project.
Our research questions map to the three major stages of our research model.
Developing the Standard
- How does the collective approach taken by a VIS consortium influence the technical design choices embodied in the standard?
Does the composition of the consortium influence which data elements are defined, which transactions are defined, and which implementation platforms are considered acceptable? - How does the collective approach taken by a VIS consortium influence the industry's ability to complete the development of a standard?
Issues here include membership policies, aspects of governance including how committees are structured, how voting is managed, how intellectual property rights are determined, how costs of contribution are determined, etc.
Implementing the Standard
- How do specific design choices made by VIS standards consortia influence the future likelihood that they will be implemented in software and adopted by end users?
How difficult will the standards be for software vendors to build into new systems for the industry? To what extent are the standards that are developed perceived to be innovative and flexible enough for long term use? What choices are made by technology vendors in interpreting VIS standards and embedding standards into software? Are vendors encouraged to implement VIS standards in such a way that their software is interoperable with other vendors in the industry? How are end users integrating standards-based software into their interorganizational transactions? What types of user organizations are more likely to pursue a "straight-through" processing approach vs. a translation approach? - How does the collective approach taken by each industry influence the technical implementation choices that vendors and users make when implementing the standards? Does earlier involvement with standards development encourage adoption? Are those who have participated more likely to integrate the standard more heavily into their existing systems? Do the policies related to intellectual property influence the ability of vendors and users to implement the standards?
Diffusing the Standard
- How do technical implementation choices by vendors and users influence the extent to which standards-based systems diffuse throughout an industry?
To what extent do the standards represent the data types and transaction types of all segments in the industry, encouraging all to adopt? Will they have differential cost saving benefits for larger vs. smaller organizations? Do they threaten the position of certain segments of the industry value chain?
Comparing across Industries
- How (if at all) and why do the answers to questions 1-5 differ across industries? In particular, do institutional factors at the industry level vary significantly in a way consistent with differences in the findings?
The Influence of Technical Design and Implementation Choices on Electronic Interchange Outcomes
To pursue our intuition that the most significant potential effects of IT use in the home mortgage industry involve the relationships among organizations, we compiled a historical portrait of IT uses in the industry and an overview of industry structure from a variety of sources, including prior academic research (such as (Jacobides, 2000)), articles in the trade press, interviews with executives of the leading industry association (Mortgage Bankers Association of America, MBA), and various statistical reporting agencies. Guidance for this work came from a seminal study of information technology and mortgage industry market structure by Hess and Kemerer (Hess & Kemerer, 1994). Those authors sought to test the theory of electronic markets Malone, Yates and Benjamin (1987), but found little evidence for theoretically-derived predictions. They concluded that either the theory needed adjustment or that not enough experience with electronic markets had accumulated at that point in time.
Since a dozen years have elapsed from the publication of Hess and Kemerer's study, we believed it timely to revisit the role of electronic marketplaces in the mortgage industry. We learned that a new wave of electronic marketplace formations had occurred in the late 1990s, but that most of them had perished in the dot.com meltdown. The net result was that no more progress toward a hypothesized shift from hierarchies to markets (Malone et al., 1987) than Hess and Kemerer had observed.
We did, however, observe a significant development that those authors had not envisioned. Since Hess and Kemerer's publication, automated underwriting was introduced to the mortgage industry by the Government-Sponsored Enterprises (GSEs) Fannie Mae and Freddie Mac, which purchase and securitize mortgage loans, thereby creating a secondary market. The effects of automated underwriting have been considerable, but these effects are value chain integration effects, not brokerage effects, as Hess and Kemerer had anticipated. Prior to automated underwriting, the process of obtaining a home loan was largely manual, lengthy and costly, had a long processing time and was not as accurate as it could be, because lending decisions were based on simple heuristics like "loan to value". Within about four years of the 1994 introduction of automated underwriting, nearly 100% of US mortgage lenders had adopted it, leading to noticeable efficiency improvements. Today, for example, obtaining GSE approval on most residential mortgages takes a matter of minutes instead of the 4-6 weeks it took a decade ago. Lending decisions are more accurate: fewer defaults have occurred in recent years than formerly, even during the recent recession-a testament to the ability of the technology to accurately assess borrowers' credit risk. According to executives at the Mortgage Bankers Association of America, which conducts periodic cost studies, the cost of originating a mortgage loan has declined about 50% over the last ten years-owing in large part to IT applications and more efficient use of expensive underwriting labor. Second order effects of automated underwriting have shown up with startling clarity in mortgage industry structure. In 1994, the top ten lenders had about 25% of the loan origination market; the top 25 lenders had 40%. By 2004, the top ten lenders (one industry expert asserts that it's really the top five lenders) held 50% of the origination market; the top 25 lenders held 79%. There are still tens of thousands of mortgage lenders in America. However, the vast majority of them are very small, while a few of them are huge and growing huger. More people than ever have been able to take out mortgage loans.
Evidence and expert opinion suggest that opportunities for still more integration effects in the mortgage industry are great. For example, although loan approval takes minutes, it still takes a month or more to "close" on a loan, owing to time-consuming processes of loan "fulfillment" (appraisal, title, hazard insurance, etc.). In addition, the number of errors made during origination is high, leading to expensive post-closing reconciliations. On average, basic mortgage data may be manually re-entered into computer systems seven times due to the use of different forms and processes. The process of recording loans with county recorders is also problematic, as is the process of transferring loans to investors. The MBA estimates that another 50% decrease in the cost of lending activity could be realized over the next decade if there is widespread adoption of the industry-wide Internet-based standards for interorganizational coordination now under development in the mortgage industry. (Again, these IT-enabled effects are believed to result from cross-value chain integration effects, rather than from brokerage effects, as earlier theorists had predicted.) An extensive body of theory and research has documented the economic and social/organizational consequences of standards for communication, transportation, and organizational control. One frequently stated claim is that standards can contribute to change in industry structure, for example by promoting consolidation, vertical dis-integration, disintermediation, new entrants, etc. The potential ndustry level of analysis issues related to vertical information standards are explored in Steinfield et al., (2005). In Wigand et al., (2005), we showed how Internet-based standards can significantly reduce the costs of adoption relative to earlier EDI standards. Therefore, whereas EDI provided benefits to larger organizations at the expense of smaller ones, thus promoting industry consolidation, Internet-based standards have the possibility to benefit smaller organizations as well as larger ones. Therefore, Internet standards could alter the course of existing trends, for example, by leavening consolidation at the large end of the organizational size spectrum with opportunities for smaller organizations, including new entrants and new types of intermediaries.
We also show, however, that the potential benefits of standards for small organizations are highly dependent on whether and how the standards are implemented in both large and small organizations. We review the available implementation options in some detail and conclude that the specific technical design features of standards are likely to have important influences on the nature of the effects observed in various settings. For the current proposal, understanding how technical features and physical implementations of interorganizational coordination standards can affect the outcomes observed is highly relevant.
Collective Action Dilemmas in the Emergence of Vertical Industry Standards
While exploring the integration effects of automated underwriting in the home mortgage industry, we learned about the efforts by MISMO to develop XML-based electronic interconnection standards. The issue of standards had come up in writings about automated underwriting, because that technology involves standardization of credit scores and mortgage risk assessments. More recently, standardization efforts in the industry has focused on electronic interconnections among the GSEs, mortgage bankers, and all other players in the sector: credit companies, title and escrow companies, property assessors, etc.
In addition to the cost advantages of the XML standards relative to EDI, the MISMO effort created a data dictionary, which has two key advantages. First, by ensuring that all member organizations use business terms in the same way, the data dictionary helps prevent the "splitting" of the standard into separate dialects-as happened with EDI. When splitting occurs, the burden often falls on small suppliers to maintain separate versions of the "standard" to communicate with large, powerful customers. Second, by eliminating redundancy in the set of business terms in use, the data dictionary reduces software vendors' costs to develop applications that support the standards, thus lowering barriers to adoption by mortgage sector organizations without the resources to develop their own software.
In Markus et al., (2006), we studied the emergence of MISMO standards as a collective action involving three groups of actors: IT vendors, the GSEs, and other industry participants (e.g., mortgage bankers). We reviewed prior literature on standardization, industry collaborations, and collective actions. We observed that organizational collaborations to develop industry-specific standards for interorganizational coordination differ significantly in characteristics such as objectives, member types, and relationships among members from other collaboration types such as R&D collaboratives, industry associations, open source software development projects, and information/communication technology standards consortia. We concluded therefore that new theoretical development was required to explain the requirements for successful development of these industry-specific coordination standards. Markus et al., (2006) presented a theoretical framework for explaining this phenomenon as two separate but related collective action processes-development of the standards and diffusion of the standards-each with different dynamics such that technical choices in the design of the standards can promote consensus about the design of the standards but also discourage adoption and depress the level of benefits achieved by the standards.
It is too early to analyze the diffusion of Internet-based mortgage industry standards. Therefore, in a second investigation Markus et al., (2005), we examined the diffusion of an earlier innovation for interorganizational coordination in the mortgage industry-automated underwriting. In a retrospective study over a twenty-year time span, we traced the interplays among several theoretically important elements: the technical development of the automated underwriting innovation, social constructions of the innovation, the behaviors of key institutional actors, and the individual and collective actions of other participants. We concluded that all of these elements contributed to the successful diffusion of automated underwriting in the mortgage industry. These theoretical elements are all reflected in the current proposal.