Organizational Routines as a Source of Continuous Change

In this paper I claim that organizational routines have a great potential for change even though they are often perceived, even defined, as unchanging. I present descriptions of routines that change as participants respond to outcomes of previous iterations of a routine. Based on the changes in these routines I propose a performative model of organizational routines. This model suggests that there is an internal dynamic to routines that can promote continuous change. The internal dynamic is based on the inclusion of routine participants as agents. When we do not separate the people who are doing the routines from the routine, we can see routines as a richer phenomenon. Change occurs as a result of participants’ reflections on and reactions to various outcomes of previous iterations of the routine. This perspective introduces agency into the notion of routine. Agency is important for understanding the role of organizational routines in learning and in processes of institutionalization. (Routines; Learning; Change; Performative; Structure; Agency) Introduction Routines are temporal structures that are often used as a way of accomplishing organizational work. Students of organization have long recognized routines as an important element of organizational behavior. Routines are important in organizations, in part because a lot of the work in organizations is performed through routines (Cyert and March 1963, March and Simon 1958). Researchers have considered routines as they relate to organizational structure (Jennergren 1981, Blau and Schoenherr 1971), technology (Galbraith 1973, Gerwin 1981, Nelson and Winter 1982, Stinchcombe 1960, Thompson 1967), innovation (Beyer and Trice 1978, Hedberg et al. 1976), socialization (Beyer 1981, Kanter 1977, Kaufman 1967, Sproull 1981), and decision making (Allison and Zelikow 1999, Cyert and March 1963, Lindblom 1959, March and Simon 1958, Selznick 1957, Steinbruner 1974). Despite this considerable attention, I claim that organizational routines are still underappreciated because their potential for change has not been sufficiently explored. In this paper I present observations of routines that altered my understanding of their potential for change. I began my fieldwork in a student housing department of a large state university with the idea that organizational routines are repeated patterns of behavior that are bound by rules and customs and that do not change very much from one iteration to another. Because stability is often used as a defining characteristic of routines, I intended to study what factors contribute to this stability. The definition I used is consistent with the work of Cyert and March (1963) on standard operating procedures and Nelson and Winter (1982) on routines. While this definition helped me to identify several routines that I could follow, it did not help me pursue my original objective. Indeed, I found that most of the routines I was studying were undergoing substantial change. This discrepancy between the concept and the observations raises questions. People have asked me how there can be such a thing as a routine that changes. Isn’t that, by definition, not a routine? They certainly do not fit Webster’s first definition of ‘‘a regular, more or less unvarying procedure,’’ though they come much closer to fitting the fourth definition: ‘‘a series of steps for a dance’’ (1984, p. 1241, see also Feldman and Rafaeli 2000 on routines as dance). The routines I studied were hiring, training, budgeting, moving students into residence halls at the beginning of the year, and closing up residence halls at the end of the year. These routines fall into the category of task performance standard operating procedures identified by Cyert and March (1963). They also fit well Nelson’s and Winter’s definition ‘‘that range from well-specified technical routines for producing things through procedures for hiring and firing. . .’’ (1982, p. 14) and with the definition proposed by a group of scholars meeting at the Santa Fe MARTHA S. FELDMAN Organizational Routines 612 ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 Institute who defined routine as ‘‘an executable capability for repeated performance in some context that has been learned by an organization in response to selective pressures’’ (Cohen et al. 1996, p. 684). A hiring routine provides a good example of how a routine can change and still be the same routine. The hiring routine I observed has standard features that most of us would expect. People submit applications, they are screened and interviewed, they are given letters of rejection or job offers. These standard sequential elements of the hiring routine continued to be included in the routine. These are the aspects of the routine that did not change. However, at the beginning of my observations, an applicant for a job in this organization would have to submit applications to every residence hall he or she wanted to work in, would go through a separate screening and interviewing process in each hall, and may receive multiple rejections and/or offers. During the observation period, the routine was changed so that applicants submit only one application, are screened in a centralized process, then interviewed in each of the halls they are interested in working for. They receive only one offer of a job at the end of the process. In this case, the elements of the routine have not changed, but how they are accomplished has. I show later in the paper that how these elements are accomplished has implications for what elements are in the routine as well as for the outcome of the task of hiring. The preponderance of attention to organizational routines has focused on them as stable and unchanging (Gersick and Hackman 1990, Ashforth and Fried 1988, Weiss and Ilgen 1985). Though Cyert and March specifically acknowledge change in standard operating procedures (which they refer to as adaptation), they also state that ‘‘because many of the rules change slowly, it is possible to construct models of organizational behavior that postulate only modest changes in decision rules’’ (1963, p. 101). Nelson and Winter also acknowledge the possibility of change, which they refer to as mutation (1982, p. 18), but their definition of routine focuses on the lack of change: ‘‘Our general term for all regular and predictable behavioral patterns of firms is ‘routine’ ’’ (1982, p. 14). Recent experimental research has suggested that the stability of organizational routines is attributable, at least in part, to their being stored as distributed procedural memory that is not readily available for discursive processing (Cohen and Bacdayan 1994). Though change is not generally seen as a dominant aspect of organizational routines, scholars from a number of different perspectives have acknowledged it. One way of thinking about change in routines is change that is provoked by a crisis or an external shock. Gersick and Hackman list five reasons for change in habitual routines of groups: ‘‘(a) encountering a novel state of affairs, (b) experiencing a failure, (c) reaching a milestone in the life or work of the group, (d) receiving an intervention that calls members’ attention to their group norms, and (e) having to cope with a change in the structure of the group itself’’ (1990, p. 83). Financial crises or new ideas in the industry, for instance, cause routines to change. Technology is one explicit impetus that has been shown to bring about changes in the way an organization structures the accomplishment of work (Barley 1986, 1990; Orlikowski 1992). This is similar to the view that change in routines is associated with their origins, and that after a period of flux, an equilibrium is established that does not entail change (Cohen et al. 1996). There is no doubt that new beginnings and major transitions are powerful incentives to change the way work is accomplished, but our understanding is limited if we think of this as the only way that organizational routines change. We specifically omit the possibility that routines are continuously changing. An evolutionary or ecological perspective on routines also suggests a role for change in routines (Nelson and Winter 1982). Baum and Singh, for instance, categorize routines as genealogical entities that ‘‘pass on their information largely intact in successive replications’’ and that are ‘‘concerned with the conservation and transfer of production and organizing skills and knowledge’’ (1994, p. 4). Nonetheless, these entities are influenced by a variety of factors in the organizational context. Baum and Singh, for instance, suggest that routines are influenced not only by changes in jobs, but also by changes in the incumbents of these jobs and by the ideas and mistakes of these incumbents (1994). Miner and Estler (1985) show how responsibility accrual by individuals can be a vehicle for or a reflection of organizational change through the redefinition of jobs. They also show that such changes are influenced by factors at the individual, organizational, and environmental levels. Miner (1991) argues that these evolved jobs are routines and that their survival depends on features of the organizational context, features of the jobs and features of the individuals who made the initial change in the jobs. Burgelman (1994) presents a different source of change in organizational routines when he shows how the product mix for Intel evolved as a result of mid-level managers following internal rules rather than as a result of decisions by top management. The perspective presented in this paper adds to this picture of routines by focusing on the role of agency in the process of routine change. Pentland (1995) and Pentland and Rueter (1994) come closest to the perspective that I develop here. They have pointed out that routines have qualities of both stability MARTHA S. FELDMAN Organizational Routines ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 613 and change. Their use of a grammatical analogy for understanding routines produced the following insight: ‘‘An organizational routine is not a single pattern but, rather, a set of possible patterns—enabled and constrained by a variety of organizational, social, physical and cognitive structures—from which organizational members enact particular performances’’ (1994, p. 491). They suggest that organizational participants have a repertoire of actions they can take. The choice from among the repertoire varies according to preceding actions and is guided, though not determined, by the grammar or ‘‘rules’’ about what actions go together. This work claims that the unvarying qualities of routines are the rules about how to put parts of the repertoire together and the repertoire itself. These two elements constitute the structure that enables and constrains the actions that take place. My observation of organizational routines extends this understanding of change in organizational routines. Like Pentland and Rueter, I find that variation is a common part of organizational routines in large part because they are not mindless but ‘‘effortful accomplishments’’ (1994, p. 488). I also find, however, that change is more than choosing from among a repertoire of responses, and that the repertoire itself, and the rules that govern choice within a repertoire can also change. In addition, the changes in the repertoire and the rules have implications for what it means to accomplish a particular task. My observations suggest that work practices such as organizational routines are not only effortful but also emergent accomplishments. They are often works in progress rather than finished products. Naming routines emergent accomplishments, however, does not help us understand where new repertoires and rules come from. Previous observations that exogenous change in the form of a change in the context of the organization or of the introduction of new technology are surely important motivations to changing rules and repertoire. My research, however, points to the internal dynamic of a routine as another source of change. This perspective moves away from viewing routines as either behavioral or cognitive and toward thinking about routines as something that includes both of these aspects. One can think of routines as flows of connected ideas, actions, and outcomes. Ideas produce actions, actions produce outcomes, and outcomes produce new ideas. It is the relationship between these elements that generates change. The fit between the ideas, actions, and outcomes is not always tight. Ideas can generate actions that do not, in fact, execute the ideas. Actions can generate outcomes that make new and different actions possible or necessary. The outcome could, for instance, be a disaster that encourages one to try something different next time. Outcomes, in turn, can generate new ideas. Some of these changes have the potential to be continuous. I have identified two kinds of outcomes that are implicated in continuous change: outcomes that fall short of ideals and outcomes that present new opportunities. Outcomes that fall short of ideals can, in the right circumstances, motivate continued striving. Outcomes that present new opportunities suggest an expanding notion of what is possible and worth trying. Outcomes can open up new possibilities by, for instance, creating new resources (Feldman 2000). New ideas may be required to deal with these outcomes. The adjustment process that results from either of these sources has the potential to be continuous. People who engage in routines adjust their actions as they develop new understandings of what they can do and of the consequences of their actions. This adjustment does not necessarily constitute movement to a new equilibrium. This perspective on routines fits with an understanding of organization (or organizing) as an ongoing accomplishment. This perspective has been an established part of organization theory at least since Weick transformed the famous title of the Katz and Kahn (1966) book, The Social Psychology of Organizations (Katz and Kahn 1966) into The Social Psychology of Organizing (Weick 1979). It shifts focus from what an organization or structure is to how it is accomplished. There has been much interest among both social and organizational theorists in structuring as a process rather than structure as a thing. Theorists have suggested that structure consists of patterned actions (Manning 1977, 1982; Weick 1979) and of recursive relations between actions and the residue of past actions (Giddens 1979, 1984; Sewell 1992; Barley 1986, 1990; Orlikowski 1992). As we move toward a notion of organization (or organizing) as an ongoing accomplishment we need a notion of routine to match. The performative model of routines that I propose in this paper provides an image of routine as an ongoing accomplishment. The change process described here is similar to the teleological change model described by Van de Ven and Poole (1995). They describe teleological change as incorporating a ‘‘constructive mode of development’’ in which ‘‘the process is emergent as new goals are enacted’’ (1995:523). One difference between the change in routines that I have observed and Van de Ven’s and Poole’s notion of teleological change is the idea that change is based on consensus. As illustrated by one of the routines I describe, conflict as well as consensus can be an important part of the process of routine change. Agency is an important aspect of this perspective on routines. When we do not separate the people who are doing the routines from the routine, we can see routines MARTHA S. FELDMAN Organizational Routines 614 ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 as a richer phenomenon. Routines are performed by people who think and feel and care. Their reactions are situated in institutional, organizational and personal contexts. Their actions are motivated by will and intention. They create, resist, engage in conflict, acquiesce to domination. All of these forces influence the enactment of organizational routines and create in them a tremendous potential for change. In this paper I show only a small part of the effects of agency on routines. I focus on the effects of agency on what I call the internal dynamic of routines, which involves participants’ reactions to the outcome of previous iterations of the routine. I take into account certain, very general characteristics of the participants mostly having to do with role perceptions, but myriad specific characteristics of the situated individuals who brought about the changes are omitted. In the following, I describe how routines changed in an organization I observed for four years. First I discuss the methods of study and the organizational context of the routines. Then, I present descriptions of these routines and how they changed over the course of the study. Following the descriptions, I discuss how the routines changed and propose that the internal dynamics of change in routines suggest a source of continuous change. In the conclusion I claim that this way of viewing change in organizational routines has some important implications for our understanding of learning and our understanding of processes of institutionalization in organizations. Observations of Routines In my fieldwork I focused on routines that are repeated annually and that involve many organizational participants. I focused on this kind of routine because I was at the time studying the barriers to change in organizational routines, and I thought that these would be the most likely not to change. I did not choose what Stinchcombe (1990) refers to as batch routines because I am more interested in operations in which the agents have discretion. Organizational members identified five routines for me: (1) budgeting for maintenance and renovation of the buildings and operations within the buildings; (2) hiring; (3) training the student resident staff; (4) moving students into the residence halls in the beginning of the school year; and (5) closing the residence halls at the end of the school year. Within each of these routines there are multiple routines, and there is some variance in what is included in each of routines depending on who is describing them. Nonetheless, organizational participants would have a good understanding of the rules and actions implied if one were to say, ‘‘now we are doing budgeting’’ or ‘‘now we are doing hiring.’’ In the following sections I describe how I gathered and analyzed information about these routines. Methods Data Gathering. I gathered data in stages. The first stage involved 20 formal but unstructured interviews with members of units throughout the Housing organization. In these interviews I simply asked people what their jobs were and how they performed them. I asked for examples and used much of the interview, which usually lasted about an hour, to gain more specific information for each example. These interviews gave me a feel for the work and the culture of the organization, for how units were organized and how they coordinated with other units. Based on these interviews I focused on the five routines mentioned above. Each routine was broadly recognized within the organization. I viewed these routines primarily from the perspective of one unit. This unit was the only one involved in all of the routines. In addition, much of the coordination in this unit took place in regularly scheduled meetings, which facilitated observation. This unit also welcomed my examination of their operations. The details of data gathering over the next four years take a very long time to describe and involve much knowledge of the specific routines. Suffice it to say that I attended every meeting relevant to these five routines (that I knew of and that my schedule allowed me to attend). I attended meetings of the upper-level supervisors as well as of their subordinates. I shadowed both supervisors (four people) and subordinates (three people) during times when they were particularly engaged in the routines. I often had lunch with members of the organization and had conversations as we walked to and from meetings. I also attended such things as birthday and employee of the month celebrations to get a sense of what it was like to be a member of this organization. In the last year I also engaged in participation. I consider this to be an important stage of the research that deepens the understanding of what organizational members know and feel. I participated in budget discussions, I taught a class associated with the hiring routine, and I participated in a committee that wrote a report on one of the positions that was involved in all of the routines. I spent approximately 1,750 hours in observation, participation and conversations of various sorts. Over the four years, this averages to between five and ten hours per week. There were some weeks when I spent much more time in the organization, and others when I spent much less time. During all this time I kept field notes. I also kept artifacts such as agendas for meetings, budgets, newsletters, commemorative pins, and articles used for discussion in MARTHA S. FELDMAN Organizational Routines ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 615 meetings. I audiotaped discussions only during the last year and only when I thought that my taping would not be disruptive. During this last year, I also gathered 10,000 electronic messages from both supervisors and subordinates. Electronic mail had become a common form of communication over the period of observation, and increasingly coordination was taking place through this medium. After the formal observation period, my interactions continued at a decreased rate. I did two consulting sessions with members of Residential Life that were probably more instructive for me than for them. I coauthored a chapter in a book with one of the central administrators. I continued to receive materials from mailing lists. I continued to meet with members of the organization from time to time. As a result, data gathering continued, but in a less formal and less systematic way. I tend not to put as much emphasis on information gained after the formal observation period ended, but I cannot completely discount what I learned during this time. Data Analysis. It is always hard to say where data gathering stops and data analysis begins. Whether explicitly as proposed by Glaser and Strauss (1967) or implicitly, one is always trying to make sense of one’s data and thinking about what more one can find out. My approach at this stage was to find out as much as I could about the organization, its members, and the routines they were engaged in. This is partly because my initial question of why routines do not change was replaced before long with a more general question about how to make sense of routines that do change. Conscious analysis of these data waited until the formal observation period ended. Formal analysis involved three steps that took place roughly concurrently and over a period of several years. The first step was to write a manuscript that pulled together the information I had gained about both the organization in general, and the specific routines. This manuscript included detailed descriptions of organizational units and positions, organizational culture and attitudes, and dispositions of individuals as they pertained to the organizational routines I studied. It also contained detailed descriptions of each of the routines, who had participated in them, what they had done, and how the routines had changed over the years of observations. The second step involved using several different metatheories to think about this information. The metatheories I used were ethnomethodology, semiotics, dramaturgy, and deconstruction (Feldman 1995). These metatheories all have assumptions that were consistent with the setting I was concerned with. They each allowed me to develop new understandings of the data I had gathered without going beyond what I had actually observed or been told. The reason for this part of the analysis was consciously to break the order of information as it had been presented to me. I did not discount the original order, but sought to develop alternatives. The third step took place once I had found some theories that helped me to think about both change and stability in organizational routines. The theories I found most useful were structuration theory (Giddens 1979, 1984, 1993) and the theory of practice as developed by Bourdieu (1977, 1990), Lave (1988) and Ortner (1984, 1989). As I read these theories, I used the concepts to organize my observations of the routines. This exercise led me to an appreciation of the relationship between action and structure through the medium of practice. This appreciation underlies much of what I understand about why and how organizational routines change. A final step is ongoing as I write articles in which I try to explain what I have come to understand and why I believe it is important. The effort involves shaping the data in a way that will help people to understand the point I wish to make without violating the sense of the observations. The reason for this effort is that ethnographic research yields observations that are relevant to many points of theoretical interest, and these observations are tangled and interwoven in the fabric of everyday life. As I attempt to pull out and follow one strand, I must make decisions about what constitutes a ‘‘strand’’ and about what surrounding fabric needs to be explained in order to make sense of the ‘‘strand.’’ During this process I find that questions arise that did not arise from any of the previous analytical efforts. I take this to be a function of the richness of the data rather than a failing of any of the earlier analytical efforts. Organizational Setting The organization I observed, among other things, provides housing for approximately 10,000 single students and 4,000 family members. One could say that this organization operates dormitories, and in some respects this is true. But it gives the wrong impression. The people in this organization are extremely committed to providing a living environment that helps students deal in a variety of ways with the stresses of university life. Saying that they run dormitories does not capture this commitment, which is part of why they refer to the halls as residence halls instead of dormitories. Their commitment is not only expressed in such surface manifestations, but also in the level of attention paid to the quality of the living environment and to the programs offered. The Housing organization is quite complex, consisting of a broad range of professional and nonprofessional members. The three major functional units that are responsible for the day-to-day operations of the residence MARTHA S. FELDMAN Organizational Routines 616 ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 Figure 1 Organization Chart of Residential Life halls are Facilities (responsible for maintenance and renovation of the halls), Food Service (responsible for providing meals for residents) and Residential Life (responsible for maintaining order and developing the potential of the halls as communities in which students could make a transition from life with parents to life on their own). Most of my observations took place in Residential Life. Residential Life consisted primarily of central administrators (an associate director who reported to the director of housing and three assistant directors) who neither lived nor worked in the residence halls, building directors who worked in the residence halls but did not live there, and student resident staff who both lived and worked in the residence halls. There are also systemwide professionals such as the head librarian who work in but do not live in the residence halls. These professionals are different from the building directors in that they have responsibilities in all of the residence halls, whereas the building directors were usually responsible for one hall. The systemwide professionals had student staffs who usually, but not always, were residents. Figure 1 is an organizational chart for Residential Life. My descriptions of routines involve the central administrators, building directors and resident staff members. Descriptions of Routines In the following I describe the changes that I observed in four of the routines, and some of the implications of these changes. The first vignette describes the routine of damage assessment that was part of the closing routine. The second vignette deals with the move-in routine. The third vignette talks about changes in the hiring and training routines. I deal with the two routines in one vignette because the changes I discuss implicated both routines. I do not describe the budget routine here because it is very complex, takes a great deal of space to describe and even more space to describe the ways that it did and did not change. In all cases the changes that took place are much more complex than I could ever describe in an article. I have portrayed the essence of the most significant changes for organizational members. The vignettes vary in length because the complexity of the changes in the routines and of the implications of the routine changes varies. I have placed the vignettes in order of increasing complexity and, therefore, length. Who Broke this Mirror? There are many aspects to finishing the school year and closing the residence halls. People who have been together for the year need to have opportunities to say good-bye to each other. Students need to pack up their belongings and move out. Facilities needs to clean the rooms and make them ready for summer conferences and other uses. I focus on one aspect of this process, which involved the inspection of rooms and the assessment of fines. The logistics of damage assessments are complicated by several factors. The assessment needs to be made after most possibility of incurring damage has passed. Student schedules at the end of the term are both extremely constrained and unpredictable, influenced by finals schedules, parents’ demands, and the need to spend time with good friends and lovers they will not be seeing for several months. Unresolved roommate disputes often further complicate the process of assessing damages. Add to all this the fact that the assessments are done by resident staff who are often experiencing many of the same pressures. One consequence of these complications is that room inventories for damage assessment often took place after the students left. Building directors were unhappy with this situation. When they talked to me about their concerns about room damages, the problems they cited surprised me. They were not that Housing had a hard time getting parents to pay the bills or that there were disputes about which roommate was responsible for the damage. Instead, building directors related that they were paid too easily, that they often dealt only with the parent’s (usually the father’s) secretary, and that they did not have the sense that the student was being held accountable in any way for behavior that resulted in the damage. They felt that learning how to take care of one’s room was part of the education about the transition from living with one’s parents to living on one’s own that Housing was supposed to provide. One of the building directors developed a system for checking people out of their rooms that resolved this MARTHA S. FELDMAN Organizational Routines ORGANIZATION SCIENCE/Vol. 11, No. 6, November–December 2000 617 problem. It involved developing a room inventory that was filled out at the beginning of the year by the room residents and resident staff. This inventory listed the furniture in the room and its condition as well as the condition of the room itself. At the end of the year, resident staff members met with the room residents before they moved out to go over the inventory. Ideally, both room residents were there for the inventory. This allowed for a discussion of what deterioration had taken place in the room and for assignment of financial responsibility for the deterioration. This system provided good information about room damage, but more than that it confronted students with the damage that they had incurred and the fact that someone would need to take financial responsibility for the damage. Thus, it increased the likelihood that students would take personal responsibility for damages they make to the room, and enabled building directors to fulfill more of what they saw as their educational mission. This new system started to be adopted by building directors in the second year of my observations. Every building director used it by the end of the four years of observation. Moving In and Moving On. ‘‘Smooth Move’’—the local newspaper announced the first year of a new process for moving 10,000 students into a dozen residence halls over a three-day weekend. ‘‘Last year was hellish,’’ said one student. ‘‘People were running every which way. People were stuck in traffic jams for