The Standardization Problem - An Economic ... - Semantic Scholar

33 downloads 9151 Views 37KB Size Report
have evolved in companies throughout the past years. This has led to higher costs for exchanging information between actors, e.g. when media dis- continuities ...
The Standardization Problem - An Economic Analysis of Standards in Information Systems Peter Buxmann, Tim Weitzel, Falk v. Westarp, Wolfgang König Research Program 403 „Competitive Advantage by Networking", Project B3 „Economics of Standards“ J. W. Goethe-University, Frankfurt, Institute of Information Systems

{pbuxmann|tweitzel|westarp|koenig}@wiwi.uni-frankfurt.de

Abstract: Choosing the right set of communication standards is fundamental in any kind of interaction, especially in the coordination of economic activities. This decision problem, generally referred to as "the standardization problem", mainly results from interdependencies due to positive network effects. This paper provides an economic analysis of using standards in information systems. We introduce a decision model for solving the standardization problem, which can calculate the optimal assignment of standards to actors in any given communications network. The relevant model parameters representing the costs and benefits of standardization are derived. Furthermore, we discuss possibilities and limitations of using the decision model. I

INTRODUCTION

The common use of standards generally simplifies transactions carried out between actors or eases the exchange of information between them. A current example of this thesis is the European Monetary Union. The advantages of a uniform currency are evident. For example, companies and tourists can avoid the numerous and costly conversions of various European currencies, and currency risks in the export business can be reduced. The same considerations can be applied to information systems: many different computer networks, operating systems, user surfaces and application systems have evolved in companies throughout the past years. This has led to higher costs for exchanging information between actors, e.g. when media discontinuities occur. One method to improve the flow of information is by using standards, e.g. EDI for electronically exchanging commercial documents, or TCP/IP for platform-independent communication. Using standards also has some disadvantages, such as the possible loss of uniqueness. For instance, a widespread opinion is, that by introducing a uniform European currency, individual countries lose a part of their uniqueness as well as a monetary policy instrument. Braunstein and White generally

refer to a „loss of technologies that may have some unique properties“ [1]. Loss of uniqueness is not the inevitable result of using standards. The Internet offers a good example of this. The variety offered is only made possible by the common use of standards. This paper deals with the selection of standards from an economic viewpoint. The analysis presented is a first building block towards a framework that is being developed as part of a general network theory within the interdisciplinary research program "Competitive Advantage by Networking" at Frankfurt University (http://www.vernetzung.de/eng). In this paper, we analyze the decision between existing standards from a user's perspective. The influence on the standardization decisions of others through market power or marketing strategies is not explored yet. The model parameters relevant for making decisions for selecting standards will be derived and integrated into a decision model. In the second chapter, we discuss general considerations concerning the use of standards in information systems. The third chapter aims at defining the standardization problem as the foundation of the economic selection of standards. Some experiences connected to applications of the model are described in chapter four. The fifth chapter deals with the influence of competition between standards on model parameters. The paper concludes with a short summary of the results as well as an outlook on future research activities. II

STANDARDS IN INFORMATION SYSTEMS

Standardization activities extend over a wide variety of areas. The activities of standardization organizations such as ISO, ANSI or DIN can confirm this [2]. Standards set by these types of organizations are often referred to as norms [3]. „Ironically, standards have not been completely standardized“ [4]. However, there is a general consensus that the use of standards leads to the uniformity of objects [5] and therefore enables and facilitates interaction between at least two objects. Standards provide compatibility by specifying extensional structural forms as well as the behavior

2

of the subjects involved. For example, a standard defines the track width of a railroad network or the rules for exchanging information. In the following we examine the use of standards in information systems. An information system consists of a set of system elements that carry out the necessary tasks. They can represent humans or machines. The relation between the system elements consists of the exchange of information. The tasks of the system elements include processing, storing and transferring information to other elements [6]. Standards are available for fulfilling such tasks, e.g. SQL as a data base language, standard software solutions as well as protocols such as TCP/IP or ATM or EDI standards for transferring commercial documents. The examples demonstrate how standards support interactions on different levels. Communication protocols or natural languages set general rules for exchanging information, regardless of the content to be transferred. In contrast to this, a standard software solution also defines contents. In other words, similar to an EDI standard, the information to be transferred on an intensional level, such as between modules, is determined. Furthermore, the use of a standard software solution determines rules for processing information, e.g. the use of specific cost accounting systems. These brief considerations also demonstrate that the problem of a loss of uniqueness or certain characteristics does not apply to all standards equally. While it seems to pose less of a problem for communication standards, it plays a significant role in evaluating the application of standard software. As the basis of an economic evaluation of using

actors. In our model the standards are implemented in the nodes of the information system, and the information is transmitted over the edges. The decision problem, generally called the standardization problem, is therefore to determine when, with what standard, which system element of an information system should be standardized. Figure 2 shows how the standardization problem can be differentiated into the various dimensions. The single period standardization problem assumes all occurring advantages and disadvantages of the standardization are unrelated to time. In turn, the multiperiod and single level standardization problem implies that the advantages and disadvantages of standardization can occur at different points of time, for which the resulting payments are to be discounted to the time the decision was made. The decision to use standards, however, is only made once at the beginning of the planning horizon - similar to a simple investment theoretical approach. The multiperiod/multilevel standardization problem also allows decisions to be made in different periods. Each of these dimensions can be further differentiated into a simple and an extended standardization problem. The simple problem assumes only one standard is available, while the extended problem also includes the decision to choose one standard from a selection of available alternative standards. In this paper we mainly focus on the simple, single period standardization problem. classification of the standardization problem

single period

2

simple

1

extended

multiperiod/ single level

simple

extended

multiperiod/ multilevel

simple

extended

3

Figure 2. Dimensions of the standardization problem 5

4

Figure 1. Information system with 5 nodes

standards in information systems, let N represent the set of system elements, i.e. N={1, 2, ... , n}. We model these system elements as nodes of a graph. Between these nodes a maximum of n ⋅(n − 1) undirected (or n⋅(n-1) directed) edges 2

exists. Each of the edges describes the information relation between two system elements Si (i ∈ N) and Sj (j ∈ N, i≠j). Figure 1 depicts an information system with five nodes. In the following, we assume the analyzed standards are communication standards which define rules for the exchange of information between

III

MODELING THE STANDARDIZATION PROBLEM

We model the standardization problem in such a way, that the nodes i represent the respective costs of standardization (Ki) while the (directed) edges show the costs of information transfer (cij). Cost reductions as the benefits due to standardization can be realized only when both communicating nodes have implemented the same standard. Thus, the decision problem arises which nodes should be equipped with which IT standard. There is a tradeoff between the node-related costs of implementing a standard and the edge-related savings of information costs. Let the nodes in the simple network in figure 3 represent two firms. The edges describe the com-

3

munications relationships between them. The exchange of information without standards yields the costs of information cij, paid by the respective sending node (i). These costs include both direct costs of information exchange and opportunity costs of sub-optimal decisions, which could be avoided through an increased value of information as a result of standardization. c 12

1 K1

c21

n

OF = ∑ K i xi + i =1

s.t.:

n

n

∑ ∑c

ij

i =1

yij

Min!

(1)

∀ i, j; i≠j

(2)



j =1 j≠ i

x i + x j ≥ 2 − M y ij

(3) x i , x j , y ij ∈ {0,1} ∀ i, j; i≠j Equation (1) describes the costs of a standardization decision. For xi = 1 and xj = 1, the restrictions (2) in combination with the equation in (1) requires that yij = 0.1

2 K2

IV

APPLYING THE DECISION MODEL

Figure 3. Costs of nodes and edges

A If node i standardizes, it pays the standardization costs Ki. According to the assumption however, if node j also implements the same standard, the information costs cij can then be saved. This does not mean that no costs whatsoever occur to transfer information when both nodes are standardized. Rather, cij can be interpreted as the difference between the information costs before ( cijb ) and after ( c ija ) standardization along the respective edge, thus cijb − cija = cij . Explicitly regarding changes in the information value before ( wijb ) and after ( wija ) standardization, the information costs savings potential resulting from standardization can be derived from the equation: cij = cijb − cija + wija − wijb . The benefits of implementing a communications standard must be determined for each node i. In order to do so, the costs of standardization or the node costs Ki, must be compared with the savings cij to be realized along the edges. If the savings are greater than the costs, then the standard will be implemented. The savings from the edge cost cij can only then be realized however, if the partner node j also implements this same standard, while the node costs Ki occur independently from the decision of the partner node. In the algebraic formularization we introduce a binary indicative variable xi, which takes on a value of 1 when node i is standardized and 0 when not (no investment). If i is standardized (that is, xi=1), then standardization costs Ki occur. The standardization costs for the entire network are n described by K x . In the case of the informa-



i =1

i

i

tion costs along the edges, the binary variable yij is equal to 0 if both nodes i and j indicated on the edge are standardized (xi=1 ∧ xj=1). Then, and only then, can the information costs along that edge be saved. This leads to the following formulation of the objective function OF:

Determining the Data for the Decision Model

Apart from the combinatorial complexity of the standardization problem, the main problem of applying the model is determining the model parameters. Following are a few experiences gathered from examining communication standards. The example is based on a case study that we conducted in 1998 with 3Com. It will illustrate empirical figures of costs and savings (see for a more detailed version of the case study [7]). 3Com uses EDI with about 15% of its approximately 200 business partners (suppliers and distributors). Determining the standardization costs in the case of 3Com was generally rather simple. The start up costs for the EDI solution were less than $25,000. With growing data traffic, new EDI project implementation took place. Establishing the EDI Operations department, new personnel was hired. Also, the technical infrastructure was upgraded, installing a new Unix translator for $100,000. The setup of a new trading partner for EDI at 3Com takes about 2-3 days of a programmer’s work. The annual costs of running the solution are estimated at $350,000 for personnel, $36,000 for the data transmission (VAN services), and about $17,000 for additional external services, such as software license agreements and outside contracting consultants. Compared to the benefits, the costs of the EDI solution seem to be rather reasonable. At 3Com the costs of manually processing an order process are calculated at $38 compared to $1.35 using EDI, summing up to estimated savings of $750,000 in sales order and invoice processing. Taking also the reduction of data entry errors, efficiency increases due to better warehouse management, and reduction of processing delays into account, the EDI Operations department estimates overall savings of $1.3 million. The friction costs which arise on account of a „non-use“ of standards, and therefore a lack of available information, are certainly more difficult to determine - but not impossible as the

4

case with 3Com shows - than direct cost savings. Theoretical models have been developed that deal with the problem of determining friction costs or the value of information [8; 9; 10]. However, an operationalization of these models, aiming at a practical application, has not yet been accomplished. B

Simplifying the Standardization Problem

On the basis of the economic application of the decision model, available alternatives to simplify standardization problems must be discussed. These can be attained in two ways. The first alternative includes only a subset of possible alternative procedures. This approach suggests that for specific practical use experiences have shown that employees already have alternative procedures in mind that can be examined by using a type of binary „yes-or-no-decision“. Simulation solutions [11] indicate that in many cases it is optimal to equip either all or none of the elements of an information system with a given standard. The second option aims at simplifying the data determination. This can be achieved by determining the standardization or information costs of only a subset of edges. The arithmetic mean of the collected values for the edges could be included into the model. A similar procedure determines the standardization costs. We refer to a homogenous standardization problem when the standardization costs for all nodes and the information costs for all edges are identical. In this case, the optimal solution can easily be derived analytically [11]. V

COMPETITION BETWEEN STANDARDS

The introduced decision model is based on the assumption that standards already exist, i.e. the development of standards on the market or the definition of standards by committees is not discussed. The dispersion and prevalence of standards are the result of a complex process primarily emerging from competition between standards. The particularity of this type of competition is mainly due to the fact that standards are network effect goods. That is, the benefit derived from using a standard depends on the degree of its dispersion [12; 13; 2; 5]. For instance, the benefit of E-Mail obviously depends on how many others also use it. The degree of dispersion determines the amount of information costs, which can be reduced by using the standard. The benefit of the first connection clearly equals zero, since there is no communication partner and therefore no information costs can be saved. This indicates that the benefit derived from using standards also depends on the

decisions made by others. The induced network effects can be differentiated in direct and indirect effects. Direct network effects refer to the possibility of building a network between the involved actors, Katz and Shapiro refer to „direct physical effects“ [12]. These direct network effects can occur on an intra- as well as an intercompany level. In turn, indirect network effects arise from interdependencies in the consumption of complementary goods. Thus, widespread use of a standard can be expected to lead to an increased supply of complementary products. Examples are the many applications for the Windows platform or the extensive supply of consultants for a standard software such as SAP R/3. These network effects, or demand-side economies of scale in economic terms, are the main reason inefficient standards or technologies could become established or can cause a lock-out concerning superior products. If future decisions concerning the use of standards made by other economic actors were known with certainty, the decision problem of selecting standards becomes trivial. In other words, the problem is asymmetrical information between the involved actors. A complex decision problem arises when a new standard, higher in technical quality, emerges on the market. Farrell and Saloner deal with the question of how in this case information and coordination problems influence the dispersion of standards [13]. According to Farrell and Saloner, the total benefit for all market participants can be raised if they all decide to use the new (higher technical quality) standard. Due to asymmetrical information, it is impossible to determine with certainty if such a transition will occur or not [13, 75ff.]. There is a start-up problem because imperfect information causes transitions to be risky since other market participants may not follow. In our system of model parameters, this would mean that standardization costs incurred would not yield a return in the form of information cost savings. This uncertainty concerning the reaction of other market participants can cause an economy to remain in its present state, due to coordination problems [14]. Farrell and Saloner refer to this constellation as the penguin effect.2 VI

CONCLUDING REMARKS

While the theory of positive network effects primarily examines the use of standards from a descriptive viewpoint, this paper introduces a normative decision model for selecting standards. In addition, the potentials and problems of the model applications are discussed, in connection with the combinatorial complexity as well as the data collection. The decision model provides the basis for demonstrating the advantages of using standards

5

in a model theoretical approach, without having to recourse to general requisitions such as „avoid partial solutions“ or the often vague statements of „securing strategic competitive advantages”. However, due to the extreme simplifications, the model obviously needs to be extended in order to be applicable to real world problems, and it mainly serves to show how the structure of the standardization problem can be modeled, i.e. as a trade-off between node-related standardization costs and edge-related information cost savings. We plan to extend the framework to also model dynamic standardization decisions in different periods of time, between different, possibly hierarchically interdependent sets of standards. This will, among others, raise the question of when to adopt a standard. Additionally, the presented approach implicitly applies a collective or centralized utility function as a measure of the quality of decisions. In those cases in which autonomous actors (nodes) make standardization decisions and are credited individually with the results, however, this collective measure at the aggregate level of the entire network is unsuitable. The optimization of the individual objectives of each actor with respect to the implementation of communication standards in the absence of a central, controlling unit is described by the decentralized standardization problem [14]. Together with results from an empirical study [15], this extended framework will be the basis for an evaluation of various hybrid forms of coordinating standardization decisions in communications networks. Additionally, we will use relational diffusion models to examine the decentralized coordination of consumers in software markets. Acknowledgments This work is part of the research project ”Economics of Standards in Information Networks” of the interdisciplinary research program “Competitive Advantage by Networking - the Development of the Frankfurt and Rhine-Main Region”. We thankfully acknowledge the financial support from the German National Science Foundation. We are very grateful to Tom Trunda, 3Com Global EDI Project Manager, whom we conducted the 3Com case study with.

VII

REFERENCES

[1]Braunstein, Y.M.; White, L.J.: Setting technical compatibility standards: an economic analysis, in: Antitrust Bulletin Summer 1985, 337-355. [2]Farrell, J.; Saloner, G.: Competition, Compatibility and Standards: The Economics of Horses, Penguins and Lemmings. In: H. Landis Gabel

(eds.): Product Standardization and Competitive Strategy, Amsterdam 1987, 1-21. [3]DIN, EN 45020: Allgemeine Fachausdrücke und deren Definitionen betreffend Normung und damit zusammenhängende Tätigkeiten, August 1991, 15. [4]Hemenway, D.: Industrywide voluntary product standards, Massachusetts 1975. [5]Wiese, H.: Netzeffekte und Kompatibilität - Ein theoretischer simulationsgeleiteter Beitrag zur Absatzpolitik für Netzeffekt-Güter, Stuttgart 1990. [6]Ferstl, O.; Sinz, E.: Grundlagen der Wirtschaftsinformatik. 2nd ed., München 1993. [7]Westarp, F.v.; Weitzel, T.; Buxmann, P.; König, W.: The Status Quo and the Future of EDI, in: Proceedings of ECIS'99, http://caladan.wiwi.uni-frankurt.de/IWI/project b3/deu/publikat/edi/ index.htm. [8]Demski, J.S.: Information Analysis, 2nd edition, Reading, Massachusetts 1980. [9]Drukarczyk, J.: Zum Problem der Bestimmung des Wertes von Informationen, in: ZfB 44, 1974, 1-18. [10]Laux, H.: Entscheidungstheorie 1, Grundlagen, (3rd ed.), Berlin. [11]Buxmann, P.: Standardisierung betrieblicher Informationssysteme, Wiesbaden 1996. [12]Katz, M.L.; Shapiro, C.: Network Externalities, Competition, and Compatibility. In: AER 75, 1985, 424-440. [13]Farrell, J.; Saloner, G.: Standardization, Compatibility, and Innovation. In: Rand Journal of Economics 16, 1985, 70-83. [13]Thum, M.: Netzwerkeffekte, Standardisierung und staatlicher Regulierungsbedarf, Tübingen, 1995. [14]Buxmann, P.; Weitzel, T.; König, W: Auswirkung alternativer Koordinationsmechanismen auf die Auswahl von Kommunikationsstandards, in: ZfB, Ergänzungsheft 02/99, 133-151. [15]Westarp, F.v.; Buxmann, P.; Weitzel, T.; König, W.: The Management of Software Standards in Enterprises, SFB 403 Research Report (99-7) 1

Precisely, equation (2) requires the indicative variable yij to take on a value of 1 for xi+xj < 2 . If xi+xj = 2, meaning bilateral standardization, yij = 0 because of (1). 2 The expression is based on the following analogy: hungry penguins stand at the edge of an ice floe. Fearing fish of prey, they hope other penguins will jump into the water first to reduce the risk involved. As soon as a few penguins dare to dive into the water, the danger for the other penguins decreases and the „free-rider penguins“ follow [2].