703
THE ACCOUNTING REVIEW
Vol. 77, No. 4
October 2002
pp. 703–729
Using Electronic Data Interchange
(EDI) to Improve the Efficiency of
Accounting Transactions
Shannon W. Anderson
Rice University
William N. Lanen
University of Michigan
ABSTRACT: Electronic data interchange (EDI) is an information technology
that standardizes the exchange of information between transacting parties.
Using data from a major U.S. office furniture manufacturer that adopted EDI
primarily to improve the efficiency of accounting transactions, we evaluate
whether EDI reduces order-processing time (the time from sales order receipt
to sales order scheduling) and whether this improvement is greater for more
complex orders. Our measure of complexity reflects both the mix of different
products the dealer orders as well as features and options the dealer selects
for each product in the order. We find that EDI is associated with faster order
processing, independent of complexity, and that EDI mitigates most of the
negative effects of complexity on processing time. We also find that dealers
learn to submit error-free orders to the manufacturer, and that previous errors
provide feedback that helps dealers submit more accurate orders. However,
we find only mixed evidence that order complexity impedes learning.
Keywords: accounting information systems; complexity; learning; transaction
costs.
Data Availability: All data are the property of the company. The authors are
not permitted to redistribute the data.
We are grateful to the managers of the company for providing access to data and to Doug Murphy for helping
us understand the order-entry process. We are grateful for computing assistance from Paul Michaud and for helpful
comments from Bob Kaplan, Chris Ittner, Joan Luft, David Platt, Steve Rock, Phil Shane, Frank Selto, K.
Sivaramakrishnan, Naomi Soderstrom, and workshop participants at The University of Texas, University of Colorado,
and Harvard Business School. Professor Anderson acknowledges research support from Arthur Andersen and
the Tauber Manufacturing Institute at the University of Michigan.
Submitted June 2000
Accepted April 2002
704 The Accounting Review, October 2002
I. INTRODUCTION
Electronic data interchange (EDI) is an information technology that standardizes the
exchange of information between two parties to a transaction. Cushing and Romney
(1994) classify EDI as a ‘‘transactional’’ accounting information system that reduces
the cost of transactions between firms and provides timely information for decision making.
Respondents to an AICPA survey judged EDI to be the most important information technology
for the accounting profession, in terms of both organizational productivity and
projected impact on the work of the professional accountant (Anonymous 1995).1 However,
there is little evidence that EDI improves the performance of administrative work (e.g.,
Borthick and Roth 1993).
This study uses data from a large, office furniture manufacturing firm to evaluate the
performance effects of EDI. The firm adopted EDI primarily to improve the efficiency of
accounting transactions with many of the licensed dealers who order its products (e.g., to
expedite acknowledgment of customer orders and to reduce order and billing errors).2 We
evaluate whether EDI reduces order-processing time (the time from order receipt to order
scheduling), and whether this improvement is greater for more complex orders. Our measure
of complexity reflects both the mix of different products the dealer orders as well as features
and options the dealer selects for each product in the order. Data limitations prevent us
from determining whether EDI reduced the level of order errors as compared to conventional
order submission methods. However, we are able to investigate whether dealers who
adopt EDI submit fewer orders with errors over time and with experience (i.e., learning),
whether feedback from the EDI system on the dealer’s prior errors stimulates their learning,
and whether order complexity impedes learning.
Research in accounting and operations management has documented that the production
of complex products reduces manufacturing performance, measured as productivity, cost,
and quality (Banker et al. 1990; Banker et al. 1995; Foster and Gupta 1990; Banker and
Johnston 1993; Datar et al. 1993; Noreen and Soderstrom 1994; Anderson 1995, 2001;
Ittner and MacDuffie 1995; MacDuffie et al. 1996; Ittner et al. 1997). Several studies also
consider how technology and management practices affect the relation between product
complexity and manufacturing performance. In their study of 70 automotive assembly
plants, MacDuffie et al. (1996) find that differences in the sophistication of information
technology and the extent to which the plants use advanced manufacturing practices account
for differences in the effect of complexity on plant performance. Specifically, plants that
employ ‘‘lean’’ manufacturing methods ‘‘seem to be more capable of minimizing the complexity
penalty arising from higher product variety than traditional mass production plants’’
(MacDuffie et al. 1996, 364). This is consistent with Sakurai’s (1990) description of Japanese
manufacturers who pioneered advanced manufacturing methods to ‘‘overcome the
high cost of producing lower volumes and a wider variety of products.’’ In a study of three
textile-manufacturing plants, Anderson (1995) finds that the plant where complexity is the
norm has a lower marginal cost of complexity than the two plants with less complex product
mixes. She posits that managers may ‘‘learn’’ to accommodate complexity, with the associated
costs of complexity declining as learning takes place.
1 Tsay (1993, 20) describes the accounting applications of EDI: ‘‘An important task of accounting information
systems is to support business operations. Source data automation allows an EDI-integrated accounting system
to provide timely information that supports management decision making.’’ 2 We use the term ‘‘dealer’’ for the firm that orders products directly from the furniture manufacturer. The term
‘‘customer’’ refers to the end-user of the product. Typically, the customer places an order with the dealer.
Anderson and Lanen—Using EDI to Improve Accounting Transactions 705
In this study, we use data on more than 1.5 million orders submitted through either
EDI or conventional methods to investigate whether information technology—in this case
EDI—reduces the negative effects of complexity on administrative processes. We find
strong support for the hypotheses that order complexity slows order processing and that,
independent of order complexity, EDI is associated with faster order processing. We also
find that the benefits of EDI, in terms of faster order-processing times, increase in order
complexity. The benefit of EDI that increases with order complexity offsets most of the
negative effect of order complexity, supporting the proposition that information technologies
may render the notion of ‘‘complexity’’ obsolete. Unfortunately, when EDI orders contain
errors, the time to correct the errors offsets most of the improvement associated with EDI
submission. We find that dealers who adopt EDI learn to submit error-free orders over time
and with experience, and that errors in prior periods provide feedback that improves current
period performance. We find only mixed support for the hypothesis that order complexity
impedes learning.
Our findings have several implications for accounting research and practice. First, research
in the operations and the information technology literatures documents benefits of
EDI in manufacturing (e.g., reduced inventory holding and spoilage, faster delivery). However,
there is no evidence on how EDI affects its primary users—back office, accounting
functions of the firm and its transaction partners. We provide evidence that EDI improves
performance of administrative work in the adopting firm; however, evidence on how EDI
affects the back office operations of transacting partners is still needed.
Second, evidence that order complexity impairs performance in an administrative setting
(e.g., slower order completion by the furniture manufacturer and increased errors by
dealers) is consistent with earlier evidence that product and process complexity impair
manufacturing performance. Our findings suggest that research that focuses on the effect
of these characteristics only on manufacturing efficiency underestimates the effects of product
and process complexity on firm performance.
Third, evidence that performance improvements associated with EDI increase with order
complexity has implications for investment justification. Our results indicate that in
addition to making the traditional order-entry process more efficient, EDI also mitigates the
negative effect of order complexity on order-entry time. The implication for accountants is
that, as with advanced manufacturing technologies, information technologies that enhance
flexibility and reduce economies of scale in administrative work may require different
evaluation approaches than other capital investments.3
Finally, evidence that dealers learn, with time and experience, to submit error-free
orders through EDI and that feedback about prior errors helps dealers learn, has implications
for the design and implementation of new accounting information systems. Recent experimental
research in accounting has explored the role of feedback on learning (e.g., Hirst et
al. 1999). Our finding that experience and feedback on prior errors stimulates learning in
the EDI task environment is consistent with theory.4 An implication for accounting system
designers is that feedback is an important feature that may reduce the payback period for
3 Kaplan (1986) and Sakurai (1990) discuss how traditional capital investment analysis may cause manufacturing
firms to underinvest in technologies that enhance flexibility and reduce economies of