Strategic Consensus And Manufacturing Performance [*].

The procedure used to assess the business unit's manufacturing performance involved each SBU's general manager. The general manager was asked to assess the level of focal product manufacturing performance for each performance criteria specified. Response items were then weighted by the degree of importance (1-5) attached to that performance dimension, as separately and independently designated by the general manager. Weighted responses were summed producing a total measure of the business unit's manufacturing performance (MP).

Descriptive Statistics

The variables used in this research are summarized in Table 2. Descriptive statistics and correlation coefficients are provided in Table 3. Recall strategic consensus and manufacturing task consensus were measured by computing the standard deviation of the six product group team member responses for each questionnaire item, and then summing the standard deviations for all items (25 items on the SC scale and 12 items on the MTC scale). Since standard deviation measures the dispersion or differences in perception by the product group team members, a lower mean value indicates a higher level of consensus. The sample statistics indicate that product group team members achieved a much higher degree of consensus with respect to manufacturing tasks ([micro] = 7.71) than they did for the firm's overall business-level strategy ([micro] = 19.25).

The range of values for product process alignment (0,1) reflects the fact that the SBU's in our sample operated, at most, only one unit off the diagonal position. A mean value of 0.44 suggests our sample is nearly evenly distributed among firms operating on and one unit off the diagonal position. The use of flexible manufacturing systems and robotics was measured on a 5-point Likert scale (0 = does not use; 1 = little use; 5 = extensive use). A mean value of 2.39 and a standard deviation equal to 2.13 indicates our sample has a variety of usage levels from essentially absent to quite extensive. Manufacturing performance was measured using a twelve-item, 7-point Likert scale (1 = poor relative to the industry; 7 = exceptional relative to the industry). Responses were weighted by the degree of importance assigned to that dimension by the firm's general manager and then summed to produce a total measure of operational performance.

A correlation matrix is provided in Table 3. Of the ten potential variable combinations, four produced significant correlation coefficients (all positive). The relationship between FMS/RB and MP appears to be the strongest (.5116). It is interesting to note that the correlation coefficient between SC and MP is insignificant, implying that strategic consensus and manufacturing performance are not related. Of course, this simple comparison does not control for the confounding effects of other variables.

Path Analytic Model

Path analysis was employed to empirically ascertain the magnitude of the causal relationships among the operations strategy variables hypothesized to be related. In Figure I both strategic consensus and flexible manufacturing systems/robotics (SC, FMS/RB) are defined as exogenous variables, presumed to cause variation in the endogenous or dependent variables (PPA, MTC, MP). Assumptions of causal order, deduced from the literature, are represented by the arrows. Any variations in the exogenous variable are not to be explained by the model. [D.sub.1], [D.sub.2], and [D.sub.3], are disturbance terms associated with the three endogenous variables which account for variations not explicitly included in the model. The model does not deny the existence of other variables that may be relevant, but not included. Their impact is captured by the disturbance terms.

Figure I can be converted into a system of equations that reflects the linkages drawn. One structural equation can be written for each endogenous variable. The structural equations are linear in the path coefficients and do not have a constant term. The constant term can be omitted if the variables are standardized and if the unmeasured residuals are also assumed to be standardized. Included in each equation are those variables that directly affect the endogenous variable in question, weighted by the appropriate coefficients. Path coefficients, interpreted as structural parameters that represent the true causal structure linking the variables in the model, are most easily obtained by employing ordinary regression techniques (Asher, 1983). The general form of the system of structural equations for Figure I is:

Equation 1: MP = f (SC, PPA, FMS/RB, MTC, [D.sub.1])

Equation 2: PPA = f (SC, FMS/RB, [D.sub.2])

Equation 3: MTC = f (SC, [D.sub.3])

Variable Distribution Tests

Prior to estimation, the Kolmogorov-Smirnov test for distribution normality was performed to examine whether or not the sampled values of each variable approximated a normal distribution. The results indicated that the sampled distributions of four variables (SC, MTC, EMS/RB, MP) were approximately normal, while the distribution of one variable (PPA) appeared to be non-normal. This, however, was expected due to the bi-variate nature of the PPA values in the sample, as previously described.

Due to the smaller size of the research sample (27 SBUs) and the non-normality of at least one of the measured variables (PPA), a nonparametric multiple regression approach was employed. Nonparametric statistical procedures are appropriate when the assumption of a normal distribution of variable measurements is not warranted, as they are distribution-free tests that do not require restrictive assumptions about the shape of the population and/or sample distributions.

We use the following for the extension of nonparametric regression methodology to multiple regression analysis; X- and Y-values are separately ranked from 1 to n and any usual multiple regression method is then employed on the ranked data. Thus, all causal modeling in this research was carried out using a combination of path analytic and nonparametric multiple regression procedures. Specifically, all path coefficients were estimated by means of a multiple regression approach on standardized ranked data.

RESULTS

Path coefficients (standardized regression coefficients) obtained from the original model's regression analysis are given in Table 4. Table 5 summarizes the outcomes of the hypothesized relationships ([H.sub.1]-[H.sub.7]). In accordance with a theory-trimming approach to path analysis, we excluded all coefficients not significant at the 0.10 level from the final estimation of path coefficients (James et al., 1982; Wiersema and Bantel, 1993). The final model and estimated path coefficients are reported in Figure II. Three general results were obtained.

First, contrary to Hypothesis 1, strategic consensus did not directly influence manufacturing performance, rather its impact was indirect through its influence on manufacturing task consensus. Specifically, strategic consensus had a strong direct effect on managerial task consensus, supporting Hypothesis 2, and managerial task consensus had a strong direct effect on manufacturing performance, supporting Hypothesis 3.

Second, in support of Hypothesis 4, strategic consensus was found to be directly related to on-diagonal product-process alignment. Product-process alignment, however, did not have a significant influence on manufacturing performance, rejecting Hypothesis 5. Taken together the results suggest that, even though a higher level of strategic consensus facilitated the correct degree of product-process alignment, this influence did not ultimately result in superior manufacturing performance.

Finally, the use of flexible automation was found to impact both product-process alignment and manufacturing performance. The relationship between FMS/RB and PPA, while significant, was opposite in sign from that predicted by Hypothesis 6. Specifically, the use of flexible automation resulted in a more correct degree of product-process alignment. The use of flexible automation was found to be positively related to manufacturing performance, providing support for Hypothesis 7.

One advantage of path analysis is that it enables one to measure the direct and indirect effects that one variable has on another. We found that manufacturing performance is affected either directly or indirectly by all of the variables in the model. As illustrated by the decomposition analysis in Table 6, MTG and FMS have the greatest effects on manufacturing performance. Both of these are entirely composed of direct effects and are of similar magnitude (.36 and .38, respectively). The indirect impacts of SC and PPA on manufacturing performance, although smaller in magnitude relative to the direct effects, are similar in size to one another (.17 and .21, respectively). The strongest effect in our model was not related to manufacturing performance, rather it was related to the influence of SC on MTC.

DISCUSSION

Managerial Consensus and Manufacturing Performance

Our results provide important evidence toward the specification and clarification of the linkage between strategic consensus and manufacturing performance. Specifically, we found that consensus among business unit managers and manufacturing level managers regarding the SBU's overall competitive strategy (SC) had no direct positive influence on manufacturing performance (MP). Rather, the effect of strategic consensus on manufacturing performance was indirect through an important intermediate variable, manufacturing task consensus (MTC).

The notion that strategic consensus, or agreement on the business unit's overall competitive strategy, does not directly influence manufacturing performance is intuitively appealing. Product group managers may agree on the method chosen to compete in the product group's industry; however, unless consensus is also obtained with respect to the manufacturing-specific tasks that are necessary to support that competitive method, high levels of manufacturing performance may not be obtained.

Print Get the Mag Weekly Updates