Manufacturing strategy–technology relationship among auto suppliers

MANUFACTURING STRATEGY-TECHNOLOGY RELATIONSHIP AMONG AUTO SUPPLIERS 1 César H. Ortega Jiménez (a) & (b) 2 Pedro Garrido-Vega (b) José Luis Pérez Díez de los Ríos (c) Santiago García González (d) (a) Universidad Nacional Autónoma de Honduras Instituto de Investigaciones Económicas y Sociales (IIES) Edificio 5, Planta Baja, Ciudad Universitaria, Blvd. Suyapa Tegucigalpa, MDC., Honduras (Central América). (b) Universidad de Sevilla Facultad de Ciencias Económicas y Empresariales Departamento de Economía Financiera y Dirección de Operaciones Grupo de Investigación en Dirección de Operaciones en la Industria y los Servicios (GIDEAO) Avenida Ramón y Cajal, 1. 41018 – Sevilla (Spain). (c) Universidad de Sevilla Facultad de Ciencias Económicas y Empresariales Departamento de Economía Aplicada I Avenida Ramón y Cajal, 1 41018 – Sevilla (Spain) (d) Universidad de Huelva Departamento de Economía Financiera, Contabilidad y Dirección de Operaciones Administración y Modelización de Organizaciones (G.I.A.M.O.) Facultad de Ciencias Empresariales. Plaza de la Merced, 11. 21002-Huelva (Spain). 1 Acknowledgement: This study is part of the Spanish Ministry of Education and Science National Programme of Industrial Design (DPI-2006-05531 and DPI 2009 -11148) and the Junta de Andalucía (Spain) PAIDI (Plan Andaluz de Investigación, Desarrollo e Innovación) Excellence Projects (P08-SEJ-03841). 2 Corresponding author: telephone: +504-22391849, fax number: +504-22391849, e-mail: cor[email protected] 2 ABSTRACT Each manufacturing plant has to develop its own path to success based on contingencies and on manufacturing practices links. On the basis of the latter, this paper tests the link between two of the most important manufacturing practices areas, manufacturing strategy (MS) and technology, without addressing causality or their combined effect on performance. This is done by selection fit, i.e. congruency adjustment. However, this paper goes beyond grouping both sets of practices in pairs, by using a more general selection view version, with practices from both sets related multidimensionally and subordinated by regression analysis to test for any congruent pattern. Regression results from a wide-ranging survey of auto supplier plants show that, in general, MS seems to have some kind of impact on technology, and that technology has some kind of influence on MS. In addition, a strong congruency between both practices areas is observed when using correlation. This suggests that when implementing or adjusting MS or technology, the other should also be considered; otherwise they may not operate effectively. Keywords: Congruency, Selection, Fit, Manufacturing Strategy, Technology 1. INTRODUCTION Each manufacturing plant must find its own unique path to success, based on contingent factors and the links between manufacturing practices. Previous studies on this topic still shed little light on the reasons why the application of the same manufacturing practices works well in some plants, but worse in others (Primrose, 1992; Olhager, 1993; Nassimbeni, 1996). Thus, before the selection, adaptation (when required), implementation and interconnection of manufacturing practices, there should also be a strategic, well-conceived plan based on the particular situation of the company. Without it, the designed strategy will not have the desired 3 effect: the achievement of success. All of the above should be linked to a planned path of continuous improvement. Hence plants should be dynamic, constantly drawing upon the best manufacturing practices for their possible inclusion as part of the manufacturing process. Such inclusion depends on both the context of the plant (contingency) and on the effect that the introduction of new practices will have by linking them to what the plant is already doing or is planning to do. This results in a synergy of processes designed to achieve a sustainable world-class competitive advantage by means of the continuous improvement of the manufacturing capacity (Schroeder and Flynn, 2001). However, achieving a sustainable competitive advantage, by means of using manufacturing practices, is itself an evasive goal: world class plants may sometimes have relatively poor implementation levels of practices. In such cases, it may well be that the success of the plant will quickly diminish when the conditions change, as the solid foundation of a correctly connected network of practices is not supporting the whole. Likewise, there may be cases where plants have implemented a high level of practices and still be unsuccessful. In the latter case, the plants need to consider whether they have chosen the correct practices for their own circumstances and whether the practices are appropriately linked to the overall strategy and with one another (Schroeder and Flynn, 2001). On the other hand, the effective use of technological resources—amongst other things—is essential for achieving a sustainable competitive advantage and for increasing the effectiveness of the company. Therefore, taking into account the importance of MS and technology, as well as the proposition that the lack of success in some plants may be partially due to a faulty link between practices (Schroeder and Flynn, 2001), the present study examines the link between practices from manufacturing strategy (MS) and from technology from an international auto supplier sector survey. The 4 need to investigate the interconnection between strategy and technology has also been stressed by Porter (1983, 1985). Accordingly, the present paper is primarily centred on the following research question: Are there any links between practices from manufacturing strategy and practices from technology? This is answered by way of exploratory and confirmatory research. A review of the literature is made in section 2. Research propositions are described in section 3 along with their respective hypotheses. The research methodology of this work is explained in section 4, describing the constructs and concepts used. Subsequently (section 5), the results are discussed. Finally, in section 6, some conclusions and final considerations are outlined, highlighting the implications and limitations of this study. 2. LITERATURE REVIEW In relation to the MS-technology relationships, some authors (Hofer and Schendel, 1978; Porter, 1983; Hayes, 1985; Maidique and Patch, 1988; Parthasarthy and Sethi, 1993; Parker, 2000) present a mainly static and unidirectional perspective. In this perspective, the causal relationship goes from technology to strategy and not vice versa (since the existing technical capabilities should guide the formulation of strategy). According to this perspective, competitiveness in a company’s manufacturing technology is a springboard for the development of strategy (Parthasarthy and Sethi, 1993). Therefore, manufacturing strategy should reflect manufacturing capacities, including technological initiatives. This argument of complementarities implies that plants which try to achieve high effectiveness from technological practices should implement these in conjunction with the appropriate manufacturing strategy (e.g. Corbett and Van Wassenhove, 1993; Parthasarthy and Sethi, 1993). Technology is therefore a factor that limits strategy in two ways: 1) the existing 5 technology determines the strategy that an organisation can pursue (Itami and Numagami, 1992), and 2) the company, wanting to pursue a different strategy, should expand or change its technological base (Hofer and Schendel, 1978; Maidique and Patch, 1988; Parker, 2000; Porter, 1983). Taking the opposite view, other researchers (Skinner, 1969; Stobaugh and Telesio, 1983; Dean and Snell, 1996) uphold that strategy should determine the selection of technology. According to this perspective, for an organisation to be competitive, strategy must drive technological development (Porter, 1983). In this way, technological development can bring both a group of competitive weapons and a deeper technological base applicable to other products/markets to the plant (Itami and Numagami, 1992; Zahra and Covin, 1993). The accumulated resources of past products/markets may change into the driving forces behind the diversification strategy of the plant. The true sources of competitive advantage may be derived more from consolidating technologies with manufacturing skills in the core areas of competition than from generating products that the competition does not anticipate (see Chandler, 1962; Prahalad and Hamel, 1990). Thus, the most important plant decisions in manufacturing should be made to improve the chosen base of competitive advantage (Hayes et al., 1988; Garvin, 1993). Manufacturing technology can clearly be one of these, since it is a significant element in manufacturing (Leong et al., 1990; Marucheck et al., 1990). Hence, in order to use strategy effectively, technology should be considered through its lens. However, the present study will go beyond the limitations of any single approach regarding the directions of the relationships between manufacturing strategy and technology that can be explored. 6 Thus, the research question of this paper could be nuanced as to how to identify the MS practices that affect technology practices and vice versa, and to explore the nature of these relationships. Among the possible models to analyse these relationships, selection fit 3 has been chosen since it has proven to be the best way to examine how variables interact to explain each other’s designs/implementations (Gerdin and Greve, 2004). Additionally, selection is the most common and simplest form of fit in the literature (Burns and Stalker, 1961; Morse, 1977; Drazin and Van de Ven, 1985; Galunic and Eisenhardt, 1994; Meilich, 2006). For this, exploratory and confirmatory research based on three relationships, namely a bidirectional and two unidirectional views of selection (also termed congruency) will be used. The adjustment premise that is assumed in selection is a congruency between both practice sets mutually influencing each other while operating in a plant (see Hannan and Freeman, 1977; Aldrich, 1979; McKelvey, 1982; Van de Ven and Drazin, 1985; Drazin and Van de Ven, 1985). A closer look at the way the MS-technology relationships have been researched reveals that only nine studies from over 110 papers compiled in a book edited by Schroeder and Flynn (2001), whose two main High Performance Manufacturing (HPM) research foundations were contingency and links between manufacturing practices, directly dealt with linkages between practices (Flynn et al., 1992; Flynn, 1994; Morita and Sakakibara, 1994 a, 1994 b; Flynn et al., 1995; Morita and Flynn, 1997; Ahmad, 1998; Morita et al., 1999; Cua, 2000). Furthermore, Morita and Flynn’s paper (1997) is the only study of these nine that is directly concerned with the relationship between MS and technology. However, it does not deal with 3 Fit could be defined as the correlation between two or more factors that leads to a better result. 7 this relationship in an exclusive or exhaustive way, since, on the one hand, it approaches the relationship of MS (considering only strategic adaptation) with other practices and, on the other hand, it only takes on board the concept of technological adaptation with its scales. The authors do conclude, however, that there is an important link between this technological concept and strategic adaptation. Since said book, only three works in this same line of HPM research have directly examined this important subject. In these papers there are findings that tend to confirm the importance of this relationship. Matsui (2002) studies the contribution of different practices (including MS) in the development of technology in three practices of process and product technology (effective implementation of processes, interfunctional design effort, simplicity of product design). Parts of his results constitute clear evidence that the participation of manufacturing practices (MS included) in the development of technology has a strong impact on the competitiveness of the production plant. McKone and Schroeder (2002) seek to determine the type of companies making use of process and product technology by taking the relationship within the context of the plant (they include strategic aspects) but without considering performance. Finally, a part of Ketokivi and Schroeder’s (2004) study considers the strategic eventualities involved in the adoption and implementation of several manufacturing practices to achieve high performance. However, they include "design for manufacturability" as the only technological variable. Regarding the general Production and Operations Management (POM) literature, most of the previous studies have explored the relationship between business strategy (not MS) and technology, either in a one-dimensional or a multidimensional way. Some researchers have classified the essential dimensions/practices of technology that are inherent in a specific strategy (e.g. Ford, 1980). On the other hand, Parker (2000) tries to test for current and future 8 dynamic interaction between business strategy and technology and its effect on the plant’s performance, but without using a time series (a longitudinal study). This literature shows some empirical interconnections between specific dimensions/practices of technology and business strategy. Some of the discoveries indicate the need to determine the fit/adjustment between these practices (e.g. Parthasarthy and Sethi, 1993; Croteau and Bergeron, 2001). Thus, some of these studies have indeed proposed integrated models that describe fits between several dimensions/practices of technology and business strategy (Maidique and Patch, 1988; Zahra and Covin, 1993). However, they have not empirically shown if there is a relationship of mutual adaptation in the design and implementation of MS and technology practices, which ensures that only world class organisations will survive thanks to the existence of a supposed isomorphic process between the two practice areas (selection fit). In conclusion, although the above studies have increased the general understanding of strategy-technology relationships, they have not examined the possible congruency/selection aspects of this rapport. Moreover, although they have had an influence on the generation of ideas concerning the relationships between strategy and technology, to date the corresponding empirical validations have been minimal and there have been even fewer regarding MS, since most of these past papers analyse relationships from a business strategy perspective. With this in mind, it is possible to conclude that: 1) previous research has fundamentally been oriented towards theory and 2) the possible impact of a selective relationship between MS and technology has not been well documented. Due to the above, it is not clear whether the relationship between MS and technology is inherently selective in its nature. Therefore, the present work tries to shed more light on this 9 subject by verifying a possible congruency between MS and technology (T) practices, taking data from an auto supplier survey conducted in ten countries. 3. ANALYTICAL FRAMEWORK AND HYPOTHESES One important focus of POM research in recent years has been linkages between manufacturing practices. Drawing on this, this paper tries to find whether the variables (in our case, a set of 3 MS practices and another of 3 T practices) show a certain degree of congruency. Thus, this is different from addressing the relationship of how these same variables influence performance (i.e. universal perspective) or from finding whether both practice sets interacting with each other affect performance (i.e. interaction perspective, where one of the sets interacts with the other) (Hartmann and Moers, 1999; Luft and Shields, 2003). The fundamental difference compared to the congruency/selection view is that in the two latter (universal and interaction) the researcher is not primarily interested in examining how variables interact to explain each other’s designs, but in showing that some combinations are more related to higher performance than others. In the contingency literature, the selection form of fit envisions primarily a linear correspondence between the structural and contingency variables. Thus, as a starting-point for the adjustment between both practice sets, this concept of fit could be described as the correlation between two or more factors that leads to a better result (see Venkatraman and Prescott, 1990; Milgrom and Roberts, 1995; Cua et al., 2001). In keeping with this, propositions for the relationship studied here are first described and then their respective hypotheses are presented. On the basis of the fit and misfit concepts, this paper will therefore address the concepts involving the direct relationship between the two sets of practices in question using a bivariate selection model. 16 4.2. Measurement All of the measurements used in this study were performed using perceptual scales, each consisting of several questions (items). Each question was answered using a seven-point Likert scale. Reverse-worded items were reverse scored. Content validity was ensured through both a representative collection of items, as well as a method of test construction (Nunnally, 1967). A comprehensive review of the extant literature was used for the representative list of items. The test construction method followed questionnaire preparation, pilot testing, structured interviews, translation, and back translation when the questionnaires were administered in countries whose mother tongue was not English. For construct validity, the items of each factor were checked to see if they loaded onto one factor. For this, within-scale factor analysis was performed to test whether each scale from both manufacturing practice sets formed corresponding unidimensional measures, as follows: three scales were used to measure MS practices according to the definition of MP practices described earlier. An item was deleted if it loaded onto a second factor. All factor loadings of the scales were above 0.60, much higher than the cut-off value of ± 0.40 (Hair et al., 1998). A similar procedure was used to construct the technology practices set with its three scales (all of the factor loadings were above 0.70 except for one (0.476, but still higher than cut-off)). Both the MS and technology practice sets are conceptualised and defined as unidimensional constructs. Meanwhile, a reliability analysis was conducted at the plant level for each scale to evaluate internal consistency. The reliability of the scales was measured by Cronbach’s alpha according to Nunnally (1978) and all were greater than 0.7 (the corresponding analysis with an acceptable degree of reliability and validity will be provided upon request). 17 4.3. Methods The functional form of selection fit is linear correspondence between MS and technology. Some of the advantages of this model are its simple procedure and the fact that it does not require the measure of a third variable as an outcome. In addition, operationalising the selection method is very straightforward using correlation, regression, analysis of variance (ANOVA), and so on. This study uses both correlation and regression. 4.3.1. Correlation The typical testing scheme associated with the selection approach is the assessment of simple correlation between each pair of MP variables (e.g. Aiken and Hage, 1968; Cohn and Turyn, 1980; Damanpour, 1991). Thus, the first method is the most common in selection and in this paper consists of grouping both sets of variables in pairs, where a series of canonical correlation analyses could demonstrate whether the set of technology practices used here is congruent with the MS set. Hypothesis H1 requires the strength of the relationship between two sets of variables to be tested. Canonical correlation analysis is used to test this relationship. It constructs a weighted linear combination of the variables in each of the two sets being correlated, with weights selected to maximise the correlation between the two weighted vectors, or canonical variates. One of the advantages of canonical correlation analysis is that it requires only multivariate normality of the variables in the data sets. In addition, canonical correlation permits the use of multiple dependent variables. Three criteria were considered to assess the strength of the overall relationship described by canonical correlation analysis (Hair et al., 1998): 1) level of statistical significance; 2) magnitude of the canonical correlation coefficient; and 3) redundancy measure for the percentage of variance explained by the two sets of variables. The first canonical pair comprises the two canonical variates that have the strongest relationship with each other, and is sufficient evidence to reject the null hypothesis. For the significant canonical pairs, canonical cross-loadings are calculated as the correlation between 18 each of the original variables in one set and the weighted canonical variate from the other set of variables. This set of cross-loadings is used to interpret the strength of each of the variables in explaining the relationship with the other set as a whole. Naturally, canonical correlation analysis is feasible if you do not want to consider one set of variables as the outcome and the other set as predictor variables. This paper therefore presents the following method. 4.3.2. Regression There might be some limitations to the use of canonical correlation analysis for testing the proposed hypotheses. The main basis for this is that the consideration of variables from the two domains in isolated pairs and the extrapolation of the findings to inferences associating the root domains are problematic. This problem, however, is overcome to an extent by using regression as a more general version of the selection approach, consistent with the definition of congruency. Fit has been widely measured through regression coefficients in the selection perspective (see Simons, 1987; Kaplan and Mackey, 1992; Hair et al., 1998). This analysis not only shows the general direction of the association, but also determines the degree to which the independent variables affect the dependent variables. Here, as opposed to treating the variables as independent pairs, sets of variables from the two domains are related, essentially depicting a congruent pattern in a multi-attribute configuration, where the practices of the two MP sets are related multidimensionally and subordinated by multivariate multiple regression analysis (MMRA), in order to observe whether they follow a congruent pattern. This type of regression is used when you have two or more variables that are to be predicted from two or more predictor variables. From the research variables, this method will predict firstly technology (T) from MS (H2), and secondly MS from T (H3). In both cases, this paper uses their specific practices. This regression is "multivariate" because there are three outcome variables (scales of practices) from one of the MP sets. It is a "multiple" regression because there are three predictor variables (scales of practices) for the corresponding MP set. This paper does not recommend this regression method for simultaneous equations, because it may cause the regression coefficients to be biased. Therefore each of the tests (i.e. MS to T and T to MS) is mutually exclusive. 19 MMRA is a logical extension of the multiple regression concept to allow for multiple response (dependent) variables. Multivariate regression estimates the same coefficients and standard errors as would be obtained using separate OLS regressions for each outcome variable. However, the OLS regressions will not produce multivariate results, nor will they allow for testing of coefficients across equations. On the other hand, multivariate regression, being a joint estimator, also estimates the between-equation covariances. This means that it is possible to test coefficients across different outcome variables. Hence, MMRA allows for multivariate tests for a collection of two or more responses, each in two or more practices. In other words, it allows testing for two or more responses of Ys predicted by two or more practices of Xs. Finally, there are at least two issues to consider when applying MMRA in this paper: 1. The residuals from multivariate regression models are assumed to be multivariate normal. This is analogous to the assumption of normally distributed errors in univariate linear regression (i.e. OLS regression). 2. The outcome variables should be at least moderately correlated for the multivariate regression analysis to make sense. 5. RESULTS A two-step procedure was used when performing the data analysis. First, canonical correlation analysis was performed to test the multivariate relationship across the variables representing T and MS practices (H1). The significance of this test provided the basis for two series of individual and mutually exclusive multivariate multiple regression analyses—one for each of the next two hypotheses (H2 and H3). This is a regression procedure that enabled assessment of the effect of all three practices of an MP domain on all three practices from the other MP domain. 20 The canonical correlation analysis indicated a significant multivariate relationship across MS and T variable sets, thus lending support to the relationship hypothesis H1. The statistical analysis regarding the selective fit between MS and technology practice sets through the association of canonical correlation between these variables allows for the deduction of the combinations that described the following results in Table 1. Take in table 1 Table 1 shows the results of a canonical correlation analysis between three technology practices and another three manufacturing strategy-related practices representing the main operations management areas. Only the first canonical pair was statistically significant. The canonical correlation (0.77) is high. Although there are no guidelines about the minimum acceptable value for the redundancy index, generally the higher the value of the index the better. Thus, there is evidence of the impact between the MS and T practice sets, since the redundancy index shows that close to half of the variance in the T practices set is explained by the first canonical variables of MS-related practices and that around one third of the variance in the MS practices set is explained by the first canonical variables of T-related practices. These results indicate that there is a very strong relationship between MS practices and T practices. Traditionally, canonical pairs have been interpreted by examining the sign and the magnitude of the canonical weights. However, these weights are subject to considerable instability due to slight changes in sample size, particularly where the variables are highly correlated. Canonical cross-loadings have been suggested as a preferable alternative to the canonical weights (Hair et al., 1998). The canonical cross-loadings show the correlations of each of the dependent variables with the independent canonical variate, and vice versa. Table 1 shows the 21 canonical cross-loadings for the first canonical pair. A loading of at least 0.31 is considered significantly different from zero at the 5% significance level (Graybill, 1961). According to this criterion, each of the MS variables is significantly related to the T canonical variate (canonical variate representing practices). On the other hand, all T variables (practices) are significantly related to the MS canonical variate (canonical variate representing practices). It is important to stress that the manufacturing-business strategy linkage is the most important factor in accounting for the first canonical variable of T-related practices, but the other two MS practices are not far behind. On the other hand, effective process implementation shows the highest correlation with the first canonical variable of MS-related practices, far in advance of the other two T practices. These results for the international auto supplier plants support hypothesis 1 since there is a congruency displayed through a relationship between Manufacturing Strategy and Technology. Thus, the success of manufacturing industries may often be attributed to the links between their own particular practices: technology-related practices must be accompanied by MS-related practices, which is one of the most important reasons why some manufacturing companies achieve a desirable effectiveness level in the global marketplace. Therefore, canonical correlation analysis provides a good test of the overall relationship specified by the hypothesis, as well as a basis for further regression analysis of the effects of the individual variables. Next, two separate multivariate multiple regression analyses (one per each MP set as a predictor) were performed to test hypotheses H2 and H3. Thus, two stages for both independent regressions will be shown, the first stage focusing on the multivariate tests and the second on the tests of between-subjects effects. The second stage of MMRA may be treated in a similar way to multiple linear regression. Thus, in line with Umanath and Kim’s 22 (1992) and Umanath’s (2003) conclusions on congruency and from the first part of the MMRA, equations [1] and[2] were used, where MS represents Manufacturing Strategy and T, Technology. The MS and T indexes 1, 2, and 3 represent the three corresponding practices for each set of manufacturing practices (3 MS practices and 3 T practices) explained in section 4.1 (page 15), the βs are the fit coefficients associated with their respective variables, i=1-3 represents the same three practices above for each set of MS and T manufacturing practices, and ε is the error. MSi = βmsi + βmsiT1T1 + βmsiT2T2+ βmsit3T3 + εmsi [1] Ti = βti + βtims1MS1 + βtims2MS2 + βtims3MS3 + εti [2] The selection perspective is supported by the statistical significance of β associated with the interest independent variable (MS1, MS2 and MS3 for equation 1 and T1, T2 and T3 for equation 2). Thus, for the first MMRA with MS as a predictor, multivariate tests give the following for the MS practices: Pillai’s Trace, Wilks' Lambda, Hotelling's Trace and Roy's Largest Root are all significant. All practices from MS collectively may predict the practices of T as an output. Hence, for the first stage of the first regression, all MS practices are significant for potentially predicting the T set, or in other words, all 3 MS practices may predict all 3 T practices (Table 2a). Following up, Table 2b shows all results of the second stage of the first MMRA (equation 1) using arrows to indicate significant relationship directions from tests of between-subjects effects. In view of the foregoing results (regardless of the fact that there does not seem to be complete congruency), this paper could conclude with reservations (MS does not influence 23 T3) that hypothesis 2 has been partially proven: manufacturing strategy influences technology. Take in table 2 Table 2b shows the results for the first model (equation 1) in more detail. The columns represent MS practices, which were tested to see whether each practice predicted the rows as technology practices. The consequent estimated parameters from this test show technology practices that are influenced by the manufacturing strategy practices. Thus, only formal strategic planning (MS2) does not significantly predict interfunctional design effort (T1), probably due to some type of restriction caused by planning. MMRA showed that both MS1 and MS3 have positive impacts on T1 (βs are 0.300 and 0.328). In the next row, the effective process implementation (T2) row shows that this is significantly predicted by all the manufacturing strategy practices (at the 1%, 10% and 5% significance levels, respectively). MMRA calculations showed that all MSs have positive impacts (βs are 0.309, 0.169 and 0.281) on T2. Finally, technology supplier involvement (T3) does not seem to be significantly predicted by any of the manufacturing strategy practices, possibly due to the fact that it is something that the company cannot completely control (contextual factors related to suppliers). This can all be summarised as follows. 5 out of 9 configurations are significant: • MS (all but MS2) predicts T1 • MS (all its 3 MS’s) predicts T2MS does not predict T3 The following possible unidirectional congruency relationships are therefore obtained: 1. Manufacturing strategy (except MS2) interfunctional design effort. Manufacturing strategy  effective process implementation. 24 Thus, these results show that all of the MS variables (except MS2, which is partial) in the model have statistically significant relationships with the joint distribution of interfunctional design effort and effective process implementation. Therefore, it can be said that manufacturing strategy influences technology to a certain degree, as reflected in most practices, and that as a result, hypothesis H2 has been partially fulfilled. Furthermore, whilst stressing that bidirectional relationships are not within the scope of this part of the study, hypothesis 3 was independently proven to a certain degree in the MMRA second stage: technology influences manufacturing strategy (Table 3b). This will be explained in detail below. Thus, in the first stage of the second independent regression, only T1 is not significant for potentially predicting the MS set (possibly due to coordination problems), or in other words both T2 and T3 may predict all 3 MS practices (Table 3a). Take in table 3 As in Table 2b, Table 3b sets out the results of the second model (Equation 2). In this case, rows are the technology practices, where T2 and T3 were each 5 tested to check whether they might influence the manufacturing strategy practices (columns). Thus, only effective process implementation (T2) predicts anticipation of new technologies (MS1) at the 1% significance level (MMRA showed a β of 0.657), probably due to technology processes that affect this anticipation strategy. All technology practices but T1 significantly predict formal strategic planning, MS2 (at the 1% significance level), since interfunctional actions may require more room to work. MMRA showed T2 and T3 both had positive impacts (βs are 0.794 and 0.157). Finally, it can be seen that all the technology practices except for interfunctional design effort 5 The first stage of MMRA, multivariate test, showed T1 was not significant. 25 (T1) (see Table 3b) significantly predict the manufacturing strategy and business strategy link, MS3 (at the 1% and 5% significance levels), possibly due to strategies being somewhat rigid. MMRA showed that both T2 and T3 have positive impacts (βs of 0.660 and 0.103). As with the other regression, on this basis it can be stated that 5 out of 9 configurations are significant: • T (all but T1) predicts MS2. • T2 predicts MS1 • T (all but T1) predicts MS3 Therefore, the unidirectional relationships can be summarised as follows: 1. Technology (except T1)  formal strategic planning 2. Effective process implementation  anticipation of new technologies. 3. Technology (except T1)  manufacturing strategy-business strategy link Thus, these results show that the T variables (except T1) in the model have a statistically significant relationship with joint distribution of MS practices (effective process implementation is the only technology practice with a statistically significant relationship with the MS practice, anticipation of new technologies). Therefore, it can be stated that technology influences manufacturing strategy to a certain degree, thus partially confirming hypothesis H3. 6. CONCLUSIONS AND FINAL CONSIDERATIONS 32 Hartmann, F., Moers F., 1999. Testing contingency hypotheses in budgetary research: an evaluation of the use of moderated regression analysis. Accounting, Organizations and Society, 24, 291-315 Hayes R.H., 1985. Strategic planning-forward in reverse?. Harvard Business Review, 63, 67-77. Hayes R.H., Wheelwright S.C., Clark K.B., 1988. Dynamic manufacturing: Creating the learning organisation. New York: Free Press. Hofer, C. W., Schendel, D., 1978. Strategy Formulation: Analytical Concepts. St. Paul, MN: West Publishing. Itami, H., Numagami, T., 1992. Dynamic Interaction Between Strategy and Technology. Strategic Management Journal, l, 13, 119–36. Kaplan, S.E., Mackey, J.T., 1992. An examination of the association between organisational design factors and the use of accounting information for managerial performance evaluation. Journal of Management Accounting Research, 4, 116–30. Ketokivi, M., Schroeder, R.G., 2004. Manufacturing practices, strategic fit and performance. A routine-based view. International Journal of Operations & Production Management, 24 (2), 171-91. Leong, G. K., Snyder, D., Ward, P., 1990. Research in the process and content of manufacturing strategy. OMEGA, 18(2), 109—122. Luft, J., Shields M.D., 2003. Mapping management accounting: graphics and guidelines for theory-consistent empirical research. Accounting, Organisations and Society, 28, 169–249 Maidique, M.A., Patch, B.J., 1988. Corporate Strategy and Technology Policy. In: Tushman, M.L., Moore, W.L., (Eds.). Readings in Management of Innovation (2nd Ed.). Cambridge, MA: Ballinger, 236–248. Maier, F., Schroeder, R., 2001. Competitive Product and Process Technology. In: Schroeder, R., Flynn, B., (Eds.). High Performance Manufacturing-Global Perspectives. New York: John Wiley & Sons, Inc., 74-114. Marucheck, A., Pannesi, R., Anderson, C., 1990. An exploratory study of the manufacturing strategy process in practice. Journal of Operations Management, 9 (1), 101–23. Matsui, Y., 2002. Contribution of manufacturing departments to technology development: An empirical analysis for machinery, electrical and electronics, and automobile plants in Japan. International Journal of Production Economics, 80, 185–97. McKelvey, B., 1982. Organisational Systematics: Taxonomy, Evolution, Classification. Berkeley, CA: University of California Press. 33 McKone, K. E., Schroeder, R.G., 2002. A plant’s technology emphasis and approach. A contextual view. International Journal of Operations & Production Management, 22 (7), 772-92. Meilich O., 2006. Bivariate Models of Fit in Contingency Theory. Critique and a Polynomial Regression Alternative. Organisational Research Methods, 9 (2), 161-93 Meyer, M. W., Zucker, L. G., 1989. Permanently failing organizations, Sage: Newbury Park, CA. Milgrom, P., Roberts, J., 1995. Complementarities and fit: strategy, structure and organisational change in manufacturing. Journal of Accounting and Economics, 19 (2-3), 179–208. Morita, M., Sakakibara S., 1994a. Linkage as a Key for Excellence of Management, Part 1. Management 21 (Japan Management Association), 4 (8), 48-52. Morita, M., Sakakibara S., 1994b. Linkage as a Key for Excellence of Management, Part 2. Management 21 (Japan Management Association), 4 (9), 44-48. Morita, M., Flynn, E.J., 1997. The Linking among Management Systems, Practices and Behavior in Successful Manufacturing Strategy. International Journal of Operations and Management, 17 (10), 967-93. Morse, E.V., 1977. The effects of formalization organizational adoption behavior. Organization and Administrative Sciences, 8, 107-122. Nassimbeni, G, 1996. Factors Underlying Operational JIT Purchasing Practices. International Journal of Production Economics 42 (3), 275-88. Nunnally, J.C., 1967. Psychometric Theory. New York: McGraw-Hill. Nunnally, J.C., 1978. Psychometric Theory. 2nd ed., New York: McGraw-Hill. Olhager, J., 1993. Manufacturing flexibility and profitability. International Journal of Production Economics, 3031, 67-78. Parker, A., 2000. Impact on the Organisational Performance of the Strategy-Technology Policy Interaction. Journal of Business Research, 47, 55–64. Parthasarthy, R., Sethi, S., 1993. Relating strategy and structure to flexible automation: a test of fit and performance implications. Strategic Management Journal, 14, 529–49. Pennings, J.M., 1992. Structural contingency theory: a reappraisal. In: Staw, B., Cummings, I.L., (Eds.). Research in Organizational Behavior. Greenwich, CT: JAI Press, 267-309. 34 Porter, M.E., 1983. The Technological Dimension of Competitive Strategy. In: Rosenbloom, R.S., (Ed). Research on Technological Innovation, Management, and Policy. Greenwich, CT: Jai., 1–33. Porter, M.E., 1985. Competitive Advantage—Creating and Sustaining Superior Performance. New York: Free Press. Prahalad, C.K., Hamel, G., 1990. The Core Competence of the Corporation. Harvard Business Review, 79–91. Primrose, P.L.., 1992. Evaluating the introduction of JIT. International Journal of Production Economics, 27, 9– 22. Robbins, S. P. 1990. Organization theory: structure, design, and applications (3rd ed.). Englewood Cliffs, NJ: Prentice-Hall Inc. Sakakibara, S., Flynn, B.B., Schroeder, R.G., Morris, W.T., 1997. The Impact of Just-In-Time Manufacturing and its Infrastructure on Manufacturing Performance. Management Science, 43 (9), 1246-57. Schroeder, R.G., Flynn, B., (Eds.), 2001. High Performance Manufacturing-Global Perspectives. New York: John Wiley & Sons, Inc. Simons, R., 1987. Accounting Control Systems and Business Strategy: an Empirical Analysis. Accounting, Organisations and Society, 12, 357-74. Skinner, W., 1969. Manufacturing - Missing link in corporate strategy. Harvard Business Review, May-June, 136-45. Stobaugh, R., Telesio, P., 1983. Match manufacturing policies and product strategy. Harvard Business Review, 61(2), 113—20. Umanath, N., 2003. The concept of contingency beyond ‘‘It depends’’: illustrations from IS research stream. Information & Management, 40, 551–62 Umanath, N., Kyu. K., 1992. Task-Structure Relationship of Information Systems Development Subunit: A Congruence perspective. Decision Sciences, 23 (4), 819-38. Van de Ven, A.H., Drazin, R., 1985. The concept of fit in contingency theory. In: Staw, B.M., Cummings, L.L., (Eds.). Research in Organisational Behavior. Greenwich, CT: JAI Press. Venkatraman, N., Prescott, J.E., 1990. Environment-strategy coalignment: an empirical test of its performance implications. Strategic Management Journal, 11 (1), 1–23. Zahra, S., Covin, J., 1993. Business Strategy, Technology Policy, and Firm Performance. Strategic Management Journal, 14, 451–78. 35 a. Selection: no performance variation b. Interaction: performance variation Figure 1. Fit vs. misfit 2a. All plants 2b. Low level MP2 plants 2c. High level MP2 plants Figure 2. Relationship in a Selection Fit (Adapted from Gerdin and Greve, 2004) a. Bidirectional b. Unidirectional MS-T c. Unidirectional T-MS Figure 3. MS-Technology Relationship in a Selection Fit MS T MS T MS T 36 Table 1. MS and T correlations First canonical pair Canonical Correlation 0.7711 Likelihood ratio 0.3803 Significance 0.0000 Redundancy index: MS 0.4361 Redundancy index: T 0.3117 Correlations between manufacturing strategy practices and canonical variable of technology related practices (canonical cross-loadings) Anticipation of New Technologies (MS1) 0.665 Formal Strategic Planning (MS2) 0.627 Manufacturing - Business Strategy Linkage (MS3) 0.669 Correlations between technology practices and canonical variable of MS related practices (canonical cross-loadings) Interfunctional Design Efforts (T1) 0.490 Effective Process Implementation (T2) 0.743 Technology supplier Involvement (T3) 0.329 Table 2. MS set as predictor a. Predictor regression: significance on multivariate tests b. MS to T: tests of between-subjects effects MS1*** GL: 0.210, 0.790, 0.266, 0.266 F 6.478 MS2* GL: 0.096, 0.904, 0.106, 0.106 F 2.573 MS3* GL: 0.086, 0.914, 0.094, 0.094 F 2.297 MS1 MS2 MS3 T1 *** F 7.922 F 0.444 ** F 3.818 T2 *** F 16.863 * F 2.640 ** F 5.633 T3 F 1.301 F 1.027 F 0.305 GL: Pillai’s Trace, Wilks' Lambda, Hotelling's Trace and Roy's Largest Root respectively; * P ≤ 0.1, **P ≤ 0.05, *** P ≤ 0.01 37 Table 3. Technology set as predictor a. Predictor regression: significance on multivariate tests b. T to MS: tests of between-subjects effects T1 GL: 0.053, 0.947, 0.057, 0. 057 F 1.375 T2*** GL: 0.459 , 0.541, 0.847 , 0.847 F 20.609 T3** GL: 0.107, 0.893, 0.120, 0.120 F 2.919 MS1 MS2 MS3 T1 F 0.388 F 2.770 F 0.074 T2 F 25.372 *** F 39.187 *** F 34.857 *** T3 F 1.962 F 7.658 *** F 4.249 ** GL:Pillai’s Trace, Wilks' Lambda, Hotelling's Trace and Roy's Largest Root respectively; **P ≤ 0.05, *** P ≤ 0.01