Studying purchasing specialization in organizations: a multi-agent simulation approach

https://doi.org/10.1016/j.pursup.2003.09.002Get rights and content

Abstract

This paper reports on a series of multi-agent-based simulation experiments. The purpose of these experiments is to investigate how factors like task-complexity, availability of skills and corporate policies determine the degree to which a firm may benefit from a specific, dedicated purchasing department. Particular attention is paid to the purchasing of non-product-related items and services. Among the factors investigated are: corporate purchasing policies, available information, capacity of organizational members and the nature of the various purchasing activities. Preliminary results suggest that most of these factors impact on the degree of involvement of a purchasing department as well as the quality of the purchasing activities carried out. The results show the limits of Purchasing's added value and the importance of organizational learning in that respect.

Introduction

A number of studies (US Bureau of the Census, 1991; Fearon and Bales, 1995) indicate that the role of the purchasing department is much smaller in the purchase of non-traditional (non-product-related) goods and services than commonly believed. These ‘non-traditional’ (non-product-related) areas, such as insurance, utilities, consultancy, travel and advertizing, are likely to be purchased without the involvement of a dedicated purchasing department. For example, in a study in 1993 the Center for Advanced Purchasing Studies (CAPs, Fearon and Bales, 1993) found that in a surprisingly large percentage of firms, the purchasing department had no input to the purchase of non-traditional goods and services (e.g. in 39% of the firms, there was no involvement in the purchasing of services, in 55% of the firms there was no involvement in the purchasing of utilities). Given the often huge sums of money involved in non-product-related (NPR) purchasing and the supposed saving potential by involving professional purchasing expertise, it seems worthwhile to find out which factors apparently hinder this purchasing involvement.

The literature about this subject is very limited. Most researchers who studied this subject used surveys or questionnaires as a methodology for their studies. De Boer et al. (2002) contemplate on some theoretical explanations for the limited and often problematic involvement of the purchasing department in NPR purchasing. Possible explanations are:

  • the problematic separability of complementary purchasing activities in terms of technologies used and objectives pursued;

  • the tension between the tendency for customer-oriented specialism of a purchasing department and the lack of sufficient market share for such a department;

  • the lack of omniscient agreement and awareness as to who possesses the right skills for the various tasks in a purchasing process;

  • the lack of room for a variety of internal ‘trading’ relationships and fee structures between purchasing specialists and internal customers.

De Boer et al. (2002) conclude that the problem is very dynamic and complex. Therefore, we set out to investigate the usefulness of simulation for studying the problem. More specifically, we use multi-agent simulation technology to obtain insight in the complex mechanisms that determine the modes of organization and coordination of NPR purchasing activities.

Despite its limitations, simulation could be a valuable tool in addition to the existing approaches because of the possibility to model and analyse dynamic and complex systems and control the assumptions and the conditions of the simulation model, such as corporate policies, and investigate the effects of these changes in a controlled environment. A multi-agent system is ideally suited to represent problems that have multiple problem solving entities. Such systems have the traditional advantages of distributed and concurrent problem solving, but the additional advantage of sophisticated patterns of interactions (Jennings et al., 1998).

The aims of our study are:

  • In what way and to which extent do certain factors influence the involvement of the purchasing department in the NPR purchasing activities? For example: various corporate purchasing policies, different types of purchases and purchasing situations and the skills, goals and perceptions of the agents in the organization.

  • How does the involvement of the purchasing department in the NPR purchasing change over time?

  • How does purchasing involvement correlate with the attainment of organizational goals?

Our paper is organized as follows. In the first section we describe the problem setting and starting points of our approach, followed by the simulation model. The next section contains the results of the experiments. Finally, we discuss the results and implications for management regarding possible actions for improving management control over NPR purchases. Also, we reflect on our experiences with using multi-agent simulation as a method for research in Purchasing and Supply Management.

Section snippets

Problem setting

In the simulation model we introduce two types of intelligent agents: multiple internal customer agents (ICA), which represent the departments or business units in the organization, and one purchasing agent (PA), which represents the purchasing department. Every agent is assigned the following properties: (1) a specific level of knowledge about the function and technical properties of a purchased item (content skills) and knowledge about the process-side of purchasing (process skills); (2) a

The simulation model

As explained above, we developed a simulation model based on intelligent agents, programmed in eM-Plant (Tecnomatix, 2000). Franklin and Graesser (1996) state that “an autonomous agent is a system situated within and part of an environment that senses that environment and acts on it, over time, in pursuit of its own agenda and so as to effect what it senses in the future”. The agents in our model can make their own decisions, based on their knowledge and information about the environment, and

Simulation experiments and results

In this section we describe the simulation experiments. We start with a basic scenario, which is the starting point for the rest of our experiments. In each of the following experiments, one of the parameters in the basic scenario is varied. There are five ICA and there is one PA in our model. All job types have equal probability of occurrence (1/3). The required skill levels are drawn from a triangle distribution as shown in Fig. 1 with areal=μ−10% and breal=μ+10%, where μ is the average

Autonomous policies outperform rigid corporate purchasing policies

First of all, given the initial distribution of skills between the ICA and the PA, the overall superiority of cooperation in terms of goal attainment is hardly surprising. Still, it is clear that a more ‘liberal’ corporate policy leads to higher levels of goal attainment than the stricter ‘either/or’ or ‘purchasing only’ policies. In terms of immediate recommendations, it therefore appears that it is not a good idea to impose strict corporate policies, which narrow the involvement of purchasing

Conclusions

The simulation model is a simplified representation of reality. We did not explicitly include transaction costs, which are very important in practice. Cooperation might be the best option from a knowledge perspective; it might not be from a cost perspective. To make the model more realistic we will try to include secondary goals, such as cost, which will lead to more complex decision rules for the ICA. In addition, the PA will be upgraded to the level of sophistication of the ICA in the sense

References (16)

  • Cyert, R.M., March, J.G., 1963. A Behavioural Theory of the Firm. Prentice-Hall, Englewood Cliffs,...
  • De Boer, L., Holmen, E., Pop-Sitar, C., 2002. The bounds to specializing purchasing activities. Working Paper,...
  • Faris, C.W., Robinson, P.J., Wind, Y., 1967. Industrial Buying and Creative Marketing. Allyn & Bacon,...
  • Fearon, H.E., Bales, A.W., 1995. Purchasing of non-traditional goods and services. CAPS Report,...
  • Fearon, H.E., Bales, A.W., 1993. CEOs’/Presidents’ perception and expectations of the purchasing function. CAPS Report,...
  • Freund, J.E., E. Walpole, 1987. Mathematical Statistics. Prentice-Hall, Englewood Cliffs,...
  • Franklin, S., A. Graesser, 1996. Is it an agent, or just a program?: A taxonomy for autonomous agents. In Proceedings...
  • N.R. Jennings et al.

    A roadmap of agent research and development

    Autonomous Agents and Multi-Agent Systems

    (1998)
There are more references available in the full text version of this article.

Cited by (0)

View full text