2 Exploring Change in Operating Routines 33
3.4 Sample and Method 68
3.4.3 Measures 70
In developing the items and scales for the measurement of the focal concept, we used existing measures, scales, and items wherever possible. However, only a limited number of prior empirical studies on dynamic capabilities proved suitable (e.g., Newey and Zahra, 2009; Prieto, Revilla and Rodríguez-Prado, 2009). Therefore, we decided to integrate existing items from related research contexts and adapt them to the concept of dynamic capabilities. The present inquiry also has to account for possible time lags between the introduction of dynamic capa- bilities (independent variables) and their fitness effects (dependent variables) (Ambrosini and Bowman, 2009). In line with recommendations of Drnevich and Kriauciunas (2011), measures of the independent variables were set retrospec- tively to the year 2005; that is, five years before the survey was conducted. The
dependent variables were measured as an average annual rate of change over the last five years (2005 to 2009).
Independent Variables. By focusing on the three observable dynamic capabili-
ties manifestations of higher-order sensing, learning, and reconfiguring routines,
our study is able to describe and measure dynamic capabilities straightforwardly. In order to capture their procedural and recurring character, we relied on items proposed by Hambrick (1981) as well as Pfeffer and Leblebici (1973) in their empirical studies on organizational routines and activity patterns. However, since these items do not originate from the dynamic capabilities literature, they were reformulated in order to fit each of the three dynamic capabilities components. The reformulations and adaptations were guided by the current dynamic capabili- ties literature (e.g., Ettlie and Pavlou, 2006; Newey and Zahra, 2009). In order to provide survey respondents with a clear understanding of this study’s conceptual position, we described each of the three components in managerial language. In total, this survey measured dynamic capabilities with six items: The first three items relate to the frequency with which each of the three activity components (i.e., sensing, learning, and reconfiguring) was executed by the purchasing de- partment five years ago (daily, weekly, monthly, once a quarter, once a year). The remaining three items focused on the percentage of average monthly work- ing hours the department spent on the execution of sensing, learning, and recon- figuring activities, and thus enabled us to calculate the average daily working hours. Our extensive pre-test showed that procurement department managers are able to look into and regularly draw on objective archival data, such as procure- ment controlling reports, when answering these questions. Based on these measures, a multiplicative index (working days per year × working hours per day = working hours per year) for each of the three components that manifest dynam- ic capabilities is calculated. This calculation was done in order to generate a valid estimate of the prevalence of dynamic capabilities components in the organiza- tion based on the actual time spent per year on these activities in the specific pur- chasing department. The values of these indices were skewed to the right—above acceptable limits. In order to use structural-equation modeling appropriately, the skewed distribution had to be corrected by applying a logarithmic transformation (West, Finch and Curran, 1995).
Dependent Variables. Evolutionary fitness refers to goal achievement (Helfat et
al., 2007). Therefore, we operationalize evolutionary fitness of operating routines as effectiveness; that is, the extent to which the operating routine achieves prede- fined operational goals (Pavlou and El Sawy, 2006). Drawing on the current pur- chasing literature, the survey measures effectiveness using three percentage-
scaled items: to what extent orders arrived on time, to what extent they were the expected quality, and to what extent they contained the correct quantity (Shin, Collier and Wilson, 2000; Chen, Paulraj and Lado, 2004; Vaidyanathan and De- varaj, 2008).
In general terms, technical fitness refers to how well a routine performs its func- tion (Helfat et al., 2007). Therefore, we operationalize technical fitness as operat- ing-routine efficiency; that is, the ratio of effective output to input (Drucker, 1967). These variables were measured indirectly by asking whether the surveyed firms changed their cost structure. To capture efficiency through routine optimi- zation (functional technical fitness), we collected information on the average an- nual percentage change in functional costs, including purchasing department costs such as labor, proportional IT, proportional accounting, and equipment costs. To capture the company’s efficiency in handling unexpected changes (adaptive technical fitness), the survey collected information on changes in costs for ad-hoc deliveries (e.g., short-notice changes in supplier’s delivery dates, air- freight charges resulting from missing spare parts). The respondents were asked to answer these questions with respect to their most-important purchased articles as well as to discount for industry-wide cost changes. Given that we collected data on both operating-routine effectiveness and costs (functional and ad-hoc) using percentage scales, each respondent company may judge our scale using different baseline values. Therefore, calculating efficiency measures based on a ratio would have yielded an invalid efficiency metric. To avoid this validity issue,
we calculated the functional technical fitness by using the difference between the values for effectiveness and functional costs, and the adaptive technical fitness by the difference between the values for effectiveness and ad-hoc costs. This trans- formation provides a metric for efficiency with a symmetric distribution around zero and a foundation to derive statements about technical effectiveness—that is, the output in relation to the input (costs).
Control variables. To control for other possible influences on the evolutionary
and technical fitness of purchasing routine, five organization-related control vari- ables were included in the model. Firm size (measured by the natural logarithm of the overall number of employees in relation to the natural logarithm of the purchasing departments’ number of employees) may have an impact on operating routine performance, since larger firms are more likely to realize economies of scale and scope and firm size might impact the evolutionary and technical fitness of the operating routine. Firm age (measured by the number of years from found- ing to 2009) may influence operating routine performance because older firms are more experienced and therefore are expected to have more-elaborate operat-
ing routines. Sales (measured by average annual percentage change in sales dur- ing the firm’s previous five years) might have an effect on this study’s dependent variables because a high volatility in sales volume influences the workload, which in turn could have an impact on the evolutionary and technical fitness of purchasing routine. The membership and position within a group was measured as a dummy variable indicating on the one hand whether the organization is part of a conglomerate and on the other whether the company owns a subsidiary company. Both variables might impact routine performance because organiza- tions belonging to a conglomerate or managing other companies can access more slack resources to enhance routine fitness than can single firms.
In addition, the model includes six purchasing-related control variables. Purchas-
ing volume (measured by average annual percentage change during the previous
five years) may influence the purchasing routine performance because firms with high purchasing volumes are more likely to realize economies of scale and scope.
Procurement mode (measured by one survey item asking whether important parts
are bought as single components or modules) might have an effect on the de- pendent variables because procuring modules is likely to lower information and coordination effort, resulting in both more effective and efficient operating rou- tine. Routinization of the purchasing routine (measured by two survey items ask- ing about task variety and two items asking about task analyzability suggested by Withey et al. [1983]) might influence routine performance because routinization implies learning effects, which in turn leads to higher degrees of goal achieve- ment (at a lower cost). Perceived relevance of purchasing in the organization (measured by a statement indicating whether the purchasing function makes an important contribution to organizational performance, which we then dummy- coded around the mean value) might influence the dependent variables because a high level of perceived relevance implies that more resources are allocated to the purchasing department, enhancing its evolutionary or technical fitness. Job ten-
ure (measured by years on the job) controls for the respondent’s experience level,
which may affect the validity of the information provided, since respondent expe- rience relates to the ability to judge the effectiveness and efficiency of firms’ purchasing routine. Finally, vocational training (measured by changes in the pro- curement department’s expenses for vocational training over the last five years, which we then dummy-coded around the mean value) increases the job-related skills of the employees. Therefore, vocational training might impact the effec- tiveness of the procurement routine while also influencing the department’s cost structure.