پنج یافته متضاد با منطق در خرید فناوری اطلاعات

Since 1995 we have been collecting quantitative data about the purchasing of IT-products and relations between buyers and IT-suppliers in The Netherlands, together with a team of colleagues. The data include the way in which buyers search and select their supplier, the way in which they negotiate with their chosen supplier, the kind and content of the contracting that is used, the kind and number of management staff involved, the importance of the IT-product or service to the buyer and supplier, the performance of the supplier, and the problems that were eventually encountered. Using our database of transactions in IT-purchasing, we present five empirical findings that we believe to be counterintuitive: (1) though the ability to deal with IT-purchases has increased over the years, the amount of problems experienced has not diminished, (2) the types of problems with IT-transactions that are encountered most, are not the ones managers expect to occur most often, (3) large investments in planning and contracting to prevent problems are not useful, (4) current rules and procedures concerning purchasing management within firms lead to larger management investments, while they do not lead to fewer problems, and (5) although large firms are more bureaucratic and deal with more complex transactions, they are not so different from SMEs as one might think.

Since 1995 we have been collecting quantitative data on a large scale about the purchasing of IT-products and relations between buyers and IT-suppliers in The Netherlands, together with a team of colleagues. Our unit of analysis is the transaction: a single purchase of IT-products and/or services of a buyer (mostly a small or medium sized firm) from a seller, in The Netherlands. For each transaction we collected data about the way in which buyers search and select their supplier, the way in which they negotiate with their chosen supplier, the kind and content of the contracting that is used, the kind and number of management staff involved, the importance of the IT-product or service to the buyer and supplier, the performance of the supplier, and the problems that were eventually encountered. Taken together, these data provide a set of more than 2000 “quantitative snapshots” of the actual purchasing management that takes place, collected in three separate surveys in 1995, 1998, and 2003. Our research is part of a larger research program, initiated at Utrecht University (the “Management of Matches” program, Raub and Tazelaar, 2000; Raub and Weesie, 1992/3). The program considers decentralized mechanisms for cooperative relations on the basis of different kinds of theoretical models: game theoretical models with repeated interaction (Taylor, 1987; Axelrod, 1984), models based on trust and social exchange (Dasgupta, 1988; Coleman, 1990; Kreps, 1990), transaction cost theory (Coase, 1937; Williamson, 1975 and Williamson, 1985), and theories regarding the social embeddedness of economic behavior (Granovetter, 1985). The emphasis of the program is on the integration and application of these theories to different kinds of cooperative undertakings: the governance of households, employment relations, and cooperative relations between organizations (such as joint ventures, R&D alliances, and cooperative relations between buyers and suppliers). Within this larger project, several papers with systematic and rigorous tests of theoretically derived hypotheses on buyer–supplier relations have been published (see: Raub and Weesie, 1990; Raub and Snijders, 1997; Rooks et al., 2000; Blumberg, 2000 and Blumberg, 2001; Buskens and Raub, 2002; Batenburg et al., 2003; Rooks et al., 2005). We feel that, especially in the field of management and organization studies, one of the main purposes of quantitative data collection and data analysis is that they can be used to expose ideas about buyer–supplier interaction that have some intuitive or even theoretical appeal, but are in fact wrong. Especially, given the fact that a substantial part of the research in purchasing is based on case studies (Morlacchi et al., 2002), interviews with key informants, participant observation, or other forms of more qualitative data collection, it can be useful and refreshing to see which of the ideas that seemed so plausible and compelling can stand a quantitative comparison with reality. Of course, one can also work the other way around, and this is what we do in this paper. We present five empirical findings related to problems with transactions in IT-purchasing that we believe to be interesting and counterintuitive. The first two findings concern the amount, degree and type of problems that are encountered; the other three findings concern the relationship between the buyer's management, both at the transaction and the organizational level, and the problems encountered. Whenever possible, we try to find reasons for our findings; in any case, we feel that all of them are worthy of further inquiry. In order to get a feel for what the beliefs and perceptions of managers are with respect to the likelihood and causes of problems in IT-purchasing we used several sources: (a) popular IT-magazines, purchasing magazines, and newspapers, (b) studies with respect to purchasing and with respect to experienced successes and failures in IT-projects (e.g. Riesewijk and Warmerdam, 1988), and (c) discussions with experts in IT (automation managers, consultants). To get a more systematic and less biased insight to managers’ perception, we also collected: (d) survey data from our audience when presenting the results of our analysis in non-academic settings (such as at a meeting of a local branch of the Dutch Association for Purchasing Management [NEVI]), (e) experimental data (conjoint analysis) from managers and laymen who were tested on their ability to assess the likelihood of problems, and from these same individuals we also collected data on the perceived relative importance of factors influencing the likelihood of problems occurring in IT-transactions (see Snijders et al., 2003; Tazelaar and Snijders, 2004). Based on these empirical sources we get a clear picture of the beliefs and perceptions managers have concerning not only the probability of problems with purchasing transactions under specific conditions, but also with respect to the relationship between various management mechanisms and this probability. For instance, purchasing managers have strong beliefs about their ability to predict the kind and type of problem that will occur, and have strong beliefs that large investments in planning and contracting help to prevent problems to occur. They also believe that on the organizational level strict internal rules and procedures with respect to the purchasing process do help to prevent problems to occur. Whether these beliefs and perceptions are sound in the field of IT-purchasing will be put to the test here.

We conclude by first summarizing our main findings. First, though the ability to deal with IT-purchases has increased over the years, the amount of problems experienced has not diminished. IT-managers have become better at judging IT-purchases, but IT-projects have in the meantime also grown more complex, and the net effect on the ability to preclude problems is zero. Second, the types of problems with IT-transactions that are encountered most are not the ones managers expect to occur most often. The ones that occur most often are generally related to after sales issues, and much less related to exceeding the budget, compatibility issues, and time delays. Third, we find no evidence that large investments in planning and contracting are useful in preventing problems, although they might have other benefits (such as improving the relationship with the supplier). A certain minimum investment in planning and contracting is necessary, but after that the extra investments do not—on average—help prevent problems. Fourth, the current rules and procedures within our firms generally lead to larger management investments, but not to fewer problems. Finally, large firms are not so different from small- and medium-sized enterprises (SMEs): with respect to investment in purchasing management and the problems that go with it, our data indicate that the underlying mechanisms are actually similar for small and large firms. All of these findings warrant further research as to their general validity and implications, but we do want to emphasize that being able to come up with findings that are counterintuitive and have a factual basis is only possible given that we have been collecting quantitative data over a larger period of time on a relatively large scale. Without denying the importance of qualitative research, we feel it is curious and unfortunate that so little of the purchasing literature is developed in quantitative ways (note: the MAT95 and MAT98 data are freely available from the Steinmetz archives, www.niwi.knaw.nl/en). An issue that complicates the analysis of the supplier's performance is that, even if quantitative data are available, one needs relatively sophisticated ways of analyzing the data. Our analyses show that if one—inappropriately—applies standard regression models, the conclusions can turn out to be simply wrong. Of course, large-scale quantitative data collection suffers from its own disadvantages. For instance, because we collected data only from the buyer, this gives us by definition a one-sided view of the transactions under study. Moreover, although we tried hard to find respondents who were involved in the transaction under study and sufficiently aware of its details, the data are collected retrospectively (some more than others), so that recollection biases are a general risk. Nevertheless, we feel that our approach is promising also in a different respect. As our last finding shows, even when one considers purchasing transactions in a relatively small domain (IT-products and services by SMEs), careful and elaborate measurement can add to the general validity of the results. In part, this is because we control for many intervening variables. This implies that if one expects that, for instance, differences exist between SMEs and larger firms with respect to the factors that drive the probability of problems, they must be related to variables other than the ones we already control for. However, as we saw using MAT03, the differences between SMEs and larger firms are in fact negligible with respect to the underlying mechanisms. That brings us to the scope of our findings. Basically, we see no reason why IT-purchasing in The Netherlands would be different from IT-purchasing in any other developed western country, so we do expect our results to hold for IT-purchasing in general. Our general suspicion is that after controlling for characteristics of transactions, buyers, sellers, and investment in management, not many differences with purchasing of other products in other contexts will remain. This implies that we are quite confident about the validity of our results for purchasing in, say, facility management or consulting. What we can imagine is that differences will exist for purchasing of other products in different contexts when the relative efficiency of management mechanisms differs from that in purchasing. The purchasing of IT is in many ways ‘uncertain’, especially when the IT is non-standard, and it is hard to make contractual agreements about it. Compared to other products or services, it is difficult to specify concrete targets for software or even hardware in contractual agreements. Moreover, the buyers of IT—especially the SMEs—are not very well connected socially, so that more informal mechanisms of governance are not likely to work. One way to progress that we are currently undertaking (Kamann et al., 2005) is to run a similar study in a context where contracting is easier because products are more standard, and the social network between buyers is more dense, such as contractor–subcontractor relations in the (Dutch) construction industry. In such a context both formal management through contracts and informal management through networks that spread information and reputation should be more efficient. The latter issue is another interesting question for future research. Given that we find relatively little support for the use of formal investment mechanisms such as contracts, one could try to understand better how informal ways of safeguarding a transaction might work (cf Rooks et al., 2005). Finally, our findings about rules and procedures suggest an obvious follow-up question: how can rules and procedures be designed so that they do reduce problems in buyer–supplier relations, or are there other benefits from standardized rules and procedures that compensate for the increased level of management they tend to invoke?