4—
Informal Coordinative Mechanisms

If individuals act without first establishing arrangements with those whom their actions affect and who affect them, each is likely to be sorry when he sees the price all will pay.
—Philip B. Heymann, "The Problem of Coordination: Bargaining and Rules"

I turn now toward the heart of the matter: understanding informal organizational features—how they work and how they facilitate coordination. My intention is to work backward, using public transit as a vehicle for generating broader generalizations about the problems of coordination inherent in multiorganizational systems and the role of informal mechanisms therein.

Where the modal tendency of contemporary discussions of coordination problems has been to assess the interests of the several actors involved and then to evaluate the attractiveness to each actor of various alternative solutions in light of the costs and benefits involved as a precursor to predicting behavior, here I focus on the mechanisms through which communications take place and solutions are sought and implemented. In this sense, the center of attention in this research has shifted from concerns about motivation and interests to actual observed behavior and processes of coordination.

The argument in this chapter complements rather than competes with theories of rational choice. However, where rational choice theories have largely assumed that near-perfect information is available to all relevant actors and that clear communication is not problematic, the approach adopted here treats these as questionable factors whose presence cannot be assumed;^[1] thus some sort of redress may be required for coordination to occur. Informal channels, in particular, are the conduits through which bargaining and negotiation take place, while informal norms provide foundations for coordination activities, and informal bargains and agreements are frequently their products. Some theorists of rational choice, such as Axelrod and Hardin,^[2] have begun to examine closely the development of informal norms and conventions and their effects on problems of coordinating multiple independent actors. Although the strategy employed here differs significantly from theirs, the conclusions reached appear to be rather similar.

Formal Failures and Informal Channels

The development of informal mechanisms is associated with inadequacies and failures of key aspects of formal organization. One such inadequacy is slowness. Although formal channels provide perfectly expeditious solutions to pressing problems in many cases, sometimes they are too slow to facilitate timely solutions. Or two organizations might benefit by solving a problem more quickly than a formal approach would allow. As one transit manager noted, where fast response is important, formal lines may take so long as to prevent essential action:

When it comes down to the short hairs, you go to the people you have known. You go the shortest distance necessary to get the information. I hate to ask a question and have the person wait to clear through twelve people before he can respond.^[3]

Informal lines of communication work more quickly than their formal counterparts because time-consuming formal constraints and ambiguity about who is responsible for a problem are eliminated.

Informal contacts are used regularly in the grant application process to check on progress of proposals through the formal machinery

as well as to push for greater speed. A planner at BART who has two friends at UMTA notes that "when funding is slow going through UMTA, I call them up to accelerate things a bit."^[4]

Informal channels also work more rapidly than their formal analogues because of a reservoir of mutual trust between the individuals involved. They go out of their way to act expeditiously as a return on prior obligations.^[5] Of course, such favors can exhaust the reservoir. "Hurry-up" favors cannot be asked in unlimited quantity. Nowhere does this show more clearly than in the execution of the express bus contract between BART and AC. Formally, BART sets the broad parameters, and AC works out the details in a two-stage procedure. Informally, however, the process is one of greater interaction:

A senior staff member at AC whom we shall designate as "S" maintains a continuing and close tie with a senior staffer at BART, who we shall refer to as "C." Both operate in "fairly unconventional ways" and bypass official channels easily. They originally got to know each other through their official tasks in connection with the express bus service that AC runs for BART. . . . For AC to establish a new express route by request takes some three months. Requests are routed to the office of scheduling, then referred to AC planning for study, then back to scheduling. In late August 1979, "C" called "S" to inquire about the possibility of a new route in Martinez. "S" immediately invited "C" to join him in a test of the proposed route. Securing a bus, they made a run to check out the availability of stops, potential traffic problems, the ability of an AC bus to negotiate hills and curves, etc. Both decided the route was feasible. On the same day they met with the public works director of Martinez and fixed the route. "S" then presented the request to the director of scheduling along with the necessary technical support. The new service route was established in less than five weeks.^[6]

Insofar as this channel is used regularly and successfully, the formal channel may in time become little more than a pro forma validation of what has been worked out informally. Or the informal channel may retain its role as a backup for the formal in that subset of situations where the formal is too slow.

Informal channels and procedures are also used where formal channels are not simply sluggish but actually blocked, perhaps because of organizational politics or because the person with the formal responsibility is reluctant to make a decision. At AC transit, a num-

ber of standard length (forty feet) buses were sectioned and shortened (to about 30 feet) to use on routes with hills, tight-radius turns, or light passenger loads. The buses were considered successful from the perspectives of both engineering and use. On weekends, Golden Gate runs some buses on narrow two-lane roads with curves difficult for standard-length buses to negotiate safely. After some mishaps, and knowing of the AC success with the shortened buses, a Golden Gate manager "A" wrote to the appropriate AC official "B," proposing that Golden Gate purchase or rent the buses for use on the weekends. Weeks passed with no response. "A" then called "C," an informal contact at AC, and asked him to look into the matter. "C" discovered that "B" had had a difficult time deciding but thought he had responded to Golden Gate. A check of the files indicated in the negative. "C" then brought Golden Gate's proposal to the AC general manager, noting that AC had no use for the shortened buses on the weekends. On the spot, the general manager agreed to rent buses to Golden Gate.^[7] In this case the informal channel served a backup function for the formal channel, ensuring that contact would be made, and some sort of decision reached.

Clearly, organizational actors think of informal contacts in terms of their backup role for formal channels. A Golden Gate planner maintains a good informal relationship (developed through work contact) with a district manager at the California Department of Transportation (Caltrans). He calls and talks with him from time to time, but "since Caltrans has a professional information office, I can call them for my information needs, and since they have proven reliable, I call him less than I might."^[8] The formal channel works well enough that the informal channel stays mostly dormant. It is not usually needed but remains available.

Another difficulty is created when formal channels do not exist in a form appropriate to the problem at hand, a failure that is both inevitable and not specifically predictable. Our knowledge is usually insufficient to permit the design of organizations capable of coping with all the problems they will ultimately have to face. Even if we could, economics would preclude us from doing so. It is also true that formal organizations are designed to operate under a specific set of conditions; but conditions rarely remain constant. A formal structure predicated to function effectively under conditions at Time I will be ineffective at Time 2 when other conditions obtain.

In the short run, however, it is the fact of anomaly that pushes informal channels to the forefront. For example, on the first occasion that BART and AC were to provide service simultaneously for the California versus Stanford football game, no formal procedures had previously been devised. No specific information about the particular problems involved was in hand. The BART and AC planners informally worked out the number of trains and buses necessary and the bus routes AC should emphasize. Because they had no prior experience, they just had to "wing it,"^[9] operating pragmatically and experimentally.

Coordination also requires accurate information. As both Devons and Sapolsky have noted, formal channels often tend to be ineffective when information is sensitive or politically charged: formal counterparts do not trust each other; there are sanctions against the transmission of such information; or there are no formal channels between the person who wants the information and the person who has it. Informal channels are not so constrained.

In one case I observed during the course of my research, the various transit agencies disagreed over the specific form and application of a proposed Bay Area elderly and handicapped fare-discount card. During negotiations, one operator sought to make an end run past another by surreptitiously building an opposition coalition. Through an informal channel (which was confidential) an employee of a third operator apprised a contact at the target agency of the effort. This allowed the latter to prevent the coalescence of a formidable opposition. Whether formal channels could have been used to make such warning is doubtful because of their open, visible nature. The political costs are too high.

In other cases, however, lack of information reliability is due not to any limitation inherent in formal channels but to the politics that typify the relationship of two organizations. Mutual distrust characteristic of the relationships between Bay Area transit agencies and MTC lays a foundation rendering information passed through formal channels nearly always suspect. At MTC, one section manager commented, referring to transit personnel, "they don't trust me completely, and I am never quite sure of how accurate or complete the information I get from them is."^[10] This manager's solution was to develop personal relationships founded on mutual trust and reciprocity with personnel at each operator.

When formal channels fail to provide parties in negotiation with an accurate perception of the reactions of other relevant organizations to their proposals,^[11] informal channels often supply the information. Says one Golden Gate planner:

Muni is a place where formal responsibility for policies cannot be discerned. If I have a plan in mind I go to a fellow I know real well since he used to work at Golden Gate, to see if he can do anything about it directly. I am then able to tailor my proposals to the idiosyncrasies of Muni.^[12]

Organizational actors also use informal channels to let their counterparts know the importance of a particular proposal that has been (or is about to be) transmitted formally. The intention to ensure that some response is made, not necessarily to stimulate a particular response.^[13] However, such informal communication of importance can only be made on a portion of coordination proposals; otherwise its currency is diluted.

Thus, informal channels perform important primary functions in the face of both permanent and episodic failures of formal mechanisms. Other important failures of formal designs with consequences for coordination are redressed by informal conventions and norms. I turn to that topic after first considering the contours of informal channels in the Bay Area public transit system.

Informal Channels and Informal Networks

Informal channels based on personal relationships between pairs of individuals are the most frequent and direct contributors to coordination in Bay Area transit. They have developed on an asneeded basis and have proved surprisingly flexible, adaptive, and effective as devices for coordination. They are predicated on personal relationships and exist independently of formal responsibility or position.

How can an informal channel be recognized when it is present? In this study, I proceeded to interview officials of many kinds and at levels ranging from transportation supervisors to general managers, with the interviews varying in length from about one to more than four hours, the average being about two hours. Although formal

position was a factor in the selection of interviewees, it was not the sole determinant. In fact, in many instances the precise formal title of the individual was not known until the interview. More important, in the selection of those to be interviewed I depended on responses of the interviewees themselves. In seeking to trace out the informal network of relationships among the personnel at the various operators and MTC, I asked the individuals interviewed: "To whom would you go if you needed to resolve an interorganizational problem, or to whom would you speak if you needed absolutely reliable information about some problem?" Those persons so named were then interviewed and asked the same questions. This procedure permitted verification of the statements made by one individual about a relationship with another.

For the purposes of this research, an informal relationship was reckoned to exist when each person named the other, irrespective of formal position, as someone to whom he would go with a problem, when reliable information was needed, or when he wished to transmit information to the other organization. This question was followed with queries about the specific character of the interactions as a check upon the extent of the relationship. Given that the research focused on links between problems of coordination and informal mechanisms, I did not want to include purely social relationships, so the questions focused specifically upon contacts that were directed toward solving work-related problems. However, expressions of friendship were also considered in the evaluations, as were social contacts outside the workplace, as additional evidence of an informal relationship.

On occasion, a pairing was included as an informal relationship where only one person named the other, if it was clear that the two carried on business with each other regularly. If an individual could not name an informal contact at another organization, I did not pressure him to do so. By following these procedures, I believe that I effectively safeguarded against including pairings where no informal relationship existed, preferring to err on the side of underinclusion.

Other approaches to establishing the presence of informal ties between individuals have included asking respondents to name people with whom they personally and frequently communicate, usually being defined operationally as three times per year (in two studies on social networks among academic researchers).^[14] I did not use this

procedure as I was not sure what the meaning of any particular frequency of contact might be and wished to avoid an arbitrary threshold for inclusion as an informal channel. Nor did it seem particularly useful to ask the respondents to make a subjective estimate of the proportion of their work activities that involved such informal contacts.

The interviewing process can thus be best described as a series of branching actions, with succeeding interviews dependent on responses in ongoing interviews. This procedure permitted discovery of the presence of informal channels without imposing a priori theoretical views on the way they might look. Confidence grew that the informal channels of the Bay Area transit system had been encompassed as fewer and fewer new names were obtained in succeeding interviews.

However, this method runs the risk of omitting informal relationships that do exist, because it depends for its effectiveness on tapping into the informal network^[15] initially. Where subsets of an informal network exist, and the subsets are relatively insulated from one another, an entire subset might be missed; thus, more informal channels might exist than were discovered. However, for reasons to be discussed I am confident that no gross errors were made in this regard.

Although multiple subsets of the informal network were discovered in the Bay Area transit system, I also found that they overlap with one another. Thus, if I successfully tapped into one subset, sooner or later I found my way into the others. Furthermore, individuals were asked to name others within their own organizations they thought should be interviewed—for example, planners were asked about people in their respective operations and maintenance sections.

In any case, given the questions of central concern in this study—Can informal mechanisms be instrumental for coordination of interdependent organizations, and are they in fact?—it is acceptable if considerable evidence is found for the existence of informal channels and their importance for coordination. I therefore preferred to err in a conservative direction: I deemed it better to miss some informal channels that were there than to assume the existence of some that were not. I am convinced, furthermore, that the description of the informal network among the Bay Area transit operators, although incomplete, is nonetheless an accurate representation of those that do

exist. Those informal channels that were found are so wide-ranging and so clearly important for coordination among the transit organizations that incompleteness does not weaken the hypothesis.

If informal organizational features arise when formal designs are either nonexistent or inadequate for one reason or another and if interdependence between two or more organizations creates uncertainty, thus giving rise to pressures for coordination, the expectation is that where there is pressure for coordination and no formal mechanisms to deal with that pressure, a more extensive set of informal relationships will develop than where there is no pressure. Thus, I examined informal ties between personnel at pairs of transit operators whose relationships are characterized by varying degrees of interdependence. Informal channels in Bay Area transit meet the expectations rather closely, with some exceptions.

Figure 4 schematically depicts the contours of informal channels in the Bay Area transit system. Existence of an informal relationship between two individuals from different organizations is represented by a dot at the intersection of their respective coordinates. Only direct informal relationships are included. Indirect two-step or three-step relationships are not considered. Figure 5 summarizes the frequencies of informal relationships observed for each pair of organizations.

Although the pattern of informal relationships is relatively stable, it does undergo change. The picture presented in Figures 4 and 5 is therefore not immutable: it is accurate for the 1978–80 period, but undoubtedly looked somewhat different before and looks somewhat different today. The contours change less and more slowly than the personnel involved, because the patterns of channels depend on the patterns of interdependence in the system.^[16] That is, one finds that the same kinds of informal relationships tend to persist despite turnover of personnel, as long as the pattern of interdependence characterizing the relevant organizations remains stable. This suggests that "role" within the system may have an effect independent of other individual traits.^[17]

Figure 6 shows the natural and voluntary interdependence in the Bay Area transit system (as so classified in Chapter 3) by pairs of operators. The two agencies with the greatest interdependence, AC and BART, display a well-developed set of informal relationships. Informal relationships are especially strong between the planning

staffs, and quite durable between management personnel, and between certain operations people. Muni and BART also enjoy a relatively high level of interdependence and reveal a similar set of informal channels. Golden Gate and Muni are well connected informally although their interdependence is very low. AC and Muni touch only at the Transbay Terminal operationally and, as expected, have a less well-developed set of informal channels than pairs of operators with greater interdependence. Golden Gate and BART have virtually no interdependence, and their informal contacts are correspondingly limited. The same holds for Golden Gate and Samtrans, which, like AC and Muni, touch only at the Transbay Terminal. They also share some San Francisco streets but have virtually no overlap in passenger clientele. Golden Gate and Santa Clara have no interdependence at all and show far fewer informal channels than AC and BART. AC and Samtrans show little informal contact when compared with AC and BART or BART and Muni. This finding was anticipated because me two agencies touch only at the Transbay Terminal and share no surface streets. BART and Samtrans are linked through contact at the BART Daly City station and show a commensurate set of informal ties. Santa Clara and Samtrans are connected less than would be anticipated from their level of interdependence. BART and Santa Clara connect at two BART stations, but have little overlap in passenger clientele. Muni and Santa Clara show only slight informal contact, reflecting their low level of interdependence.

Because of the size of its jurisdiction, the extent of its operations, and the fact that it depends on the bus operators to provide feeder service for its rail lines, of all the agencies, BART shows the most extensive set of informal interorganizational ties and channels. Conversely, Santa Clara shows the least informal contact with other transit companies, as would be expected from its geographic isolation and low level of interdependence.

Analysis of these informal relationships must be tempered by the recognition that reliance on sheer numbers as a method of comparison of the extent and importance of informal channels between one pair of operators as opposed to another pair is problematic for two reasons. Bay Area transit organizations vary considerably in the size of their operations, personnel, equipment, budgets, and numbers of passengers carried. Santa Clara and Samtrans are small when compared with the other agencies. Golden Gate is smaller than AC or

Figure 4
Interorganizational Informal Channels in the Bay Area Public Transit System (1978–80)

Figure 5
Frequencies of Informal Relationships by Pairs of Operators

Muni. Sufficient informal contact with smaller operators requires knowing fewer people.

Furthermore, the importance of one informal channel for problems of coordination is not easily calculated in comparison with others. It would be incorrect to assume that all informal channels are created equal, because the importance of business conducted through one channel as opposed to another is not amenable to quantitative measurement (at least at our present state of knowledge). From this perspective, one informal channel may be more valuable than two or more others. In light of these considerations, one ought not assume automatically that informal connections should be equally extensive for two different pairs of organizations subject to similar levels of interdependence. Still, I found more extensive informal channels as interdependence increased.

There are several exceptions to the expected levels of informal contact between pairs of agencies. Muni and Samtrans are opera-

Figure 6
Natural and Voluntary Interdependence in Bay Area Public Transit

tionally linked in a number of ways. However, there is a competitiveness and animosity between their personnel that had, at least at the time of this research, effectively precluded development of informal relationships between them. Despite their level of interdependence, no informal channels existed; the systems appeared to be at war.

Conversely, Golden Gate and AC transit display a more well-developed set of informal ties than their virtually nonexistent interdependence would lead one to expect. This is explained largely by the movement of personnel from AC to Golden Gate at the time the latter commenced bus operations.^[18] Similarly, AC and Santa Clara have more extensive informal ties than would be anticipated from their slight interdependence, principally because AC provided consulting services for Santa Clara when it began bus operations. More generally, contact through the RTA and MTC has increased the overall level of informal ties for the Bay Area beyond what would be anticipated from the level of interdependence.

Informal ties between the transit companies and MTC are extensive, matching the importance of the grants and planning processes to the operators. MTC personnel need and want reliable information about the operators: by 1980, the operators were officially required to submit over twenty reports per year to MTC. The agencies also need and want accurate information about the application of federal and state guidelines, the intentions behind new laws, and information about the availability of grants and other monies. Informal channels meet these needs and also make possible informal consultations and negotiations between the operators and MTC. MTC personnel have most of their informal contact with the planning sections of the transit companies. The most extensive informal connections appear to be between MTC and AC, MTC and BART, and MTC and Muni, with Golden Gate following close behind. Santa Clara and Samtrans are less well informally connected with MTC.

Informal Network and Subsets

Although informal channels are best described as links between pairs of individuals, and have thus far been analyzed in the context of pairs of interdependent organizations, I also found it

useful to group the channels observed on the basis of shared characteristics, in the belief that understanding the informal network in the Bay Area transit system would be enhanced by comprehending the different subsets of informal channels of which it is composed.^[19]

Social networks have been studied through a variety of methods, and a significant body of "network analysis" literature has developed. In recent years significant strides have been made in our ability to describe social networks by formal mathematical models. In particular, the work of Lorrain and White and their colleagues has added considerably to our understanding of patterns and arrangements of social relations.^[20] Although my use of "network" is consistent with theirs, and my approach resembles that of graph theory, both the goals of this research and the methodology employed here are simpler. My interest in network morphology has no intrinsic basis; it results entirely from a concern with the extent to which informal network development varies with observed interorganizational interdependence. Whereas social network researchers are often preoccupied with assessing personal or organizational influence within networks, the development of cliques, the structural equivalence of individuals, or the social structure of a total system, my aims are more modest. I am concerned with informal channels and networks only as they arise and are used to address certain problems of coordination created by interorganizational interdependence.

Furthermore, whereas sociologists have been concerned with various relational types,^[21] the research reported here focuses solely on informal relationships predicated upon and used for the conduct of work-related activities—communications relations and instrumental relations. Communications relations are understood to mean linkages between actors in the form of channels by which messages may be transmitted from one actor to another. Instrumental relations involve contact between actors in efforts to secure valuable goods, services, or information.^[22] Differences in channels stem from variations in the principal kind of business conducted through them.

If we examined the universe of Bay Area transit organization personnel and the ties between pairs of them, we would discover subsets of the larger universe of informal channels. In operational terms, subset is understood to mean a discrete clustering of channels among individuals, who, as a group, communicate informally with each other more often than they communciate informally with those out-

side the group. In the simplest case, members of one subset would have informal relationships only with those within their subset and no such relationships with others outside.^[23] I determined subset boundaries empirically by questioning individuals about those to whom they speak, moving to question that next set of individuals, and so forth, until no more new names are mentioned—the so-called snowballing technique.^[24] However, as I was interested solely in direct first-step informal relationships, the question of where to stop moving outward from the initial sample was not a problem. Furthermore, unlike more broadly cast studies of social networks, the universe of potential actors was clearly limited to members of the relevant organizations. The structure of any subset is composed of the individuals (or nodes) and the channels between them.^[25]

The Bay Area transit system is described by several such informal subsets. The subsets are accurately considered as separate entities, but share some members in common. The kind of business transacted in different subsets varies significantly.^[26] Although exchange of expertise, exchange of factual information, and informal negotiation occur at one time or another in each, one or two activities tend to predominate in any one subset. Thus, there appears to be not only specialization of informal channels, but specialization of entire subsets of channels.

Three informal subsets were observed within the larger informal network of the Bay Area transit system: operations/maintenance, planning, and management. These subsets are portrayed diagrammatically in Figure 7. Although interactions among the management personnel and among planning staffs at the different operators were expected, those among the operations/maintenance subset were not.

Among the operations/maintenance personnel, exchanges of expertise and factual information were the dominant types of informal communication. For example, the bus operations manager at Golden Gate supplied expert advice to Samtrans on the design of a fare-collection system.^[27] One of the maintenance personnel at Golden Gate discovered that four small batteries could replace the single large battery typically in use on buses, resulting in savings when cells in the battery died. He passed the word along to his informal contacts at several other transit agencies.^[28]

Maintenance personnel at AC and Golden Gate worked together informally to solve a problem with the rear doors of the (then) new

articulated buses that both properties had purchased.^[29] Although this type of communication constitutes the bulk of informal interactions among the operations/maintenance personnel, significant coordination has also been achieved through this subset.

Operations people at AC and Golden Gate were responsible for designing and implementing arrangements for weekend loans of shortened AC buses to Golden Gate. More important, all negotiations and arrangements for the AC-BART busbridge agreement were worked out by operations personnel. Even where exchange of expertise does not directly affect the course of coordination, it contributes indirectly by making future interorganizational cooperation easier by promoting similar decision premises at the agencies involved. It also provides opportunity for initial (in some cases) or continued (more often) contact between two individuals, making possible the development of mutual trust, and a chance for the norm of reciprocity to come into play.^[30]

In the planning subset, direct coordination efforts appear to be the modal activity, although exchange of expertise also occurs. AC provided Golden Gate with expert advice on changes in bus-stop lengths and locations along with driver training changes necessary for successful use of the articulated buses. The planning director at Golden Gate called an informal contact at AC planning, who put him in touch with someone in ac's training department. Knowledge AC had acquired through trial and error was thereby transferred informally to Golden Gate.^[31]

Informal communications in the planning cluster often revolve around formal tasks of coordination that have been assigned to the planners. These problems range from fairly high-level discussions to ironing out specific details for implementing coordination. At the higher level, there might be negotiations on problems of a joint fast-pass, involving issues of financial equity and mechanical compatibility. There are also continuing informal contacts over common technical problems of coordination, such as the express bus service that AC operated for BART.

The management subset is composed of fewer individuals than either the operations/maintenance or planning networks. This subset typically involves high-level discussions of policy issues, although exchange of technical expertise is not unknown. Sometimes these informal channels are simply used to communicate the importance of a

Figure 7
Subsets of the Informal Network

particular issue from one agency to another, with its resolution left to subordinates.

However, these subsets are not entirely independent of one another. The management subset intersects with both the operations/ maintenance subset and the planning subset (although the planning and operations/maintenance subsets do not appear to connect significantly). These relationships develop because management concerns overlap with concerns in each of those areas, and because in many cases managers were promoted from within those sections, taking with them contacts from their old positions to their new ones.

Several generalizations can be made about patterns of informal ties between personnel at different agencies. Type of informal tie is closely linked to function. For example, if it is used primarily as an advisory channel for exchange of expertise, the tie tends to be between individuals of equivalent formal status and similar area of activity.^[32] Thus, general managers talk to general managers, controllers talk to controllers, schedulers talk to schedulers, and maintenance supervisors talk to maintenance supervisors.

Conversely, for informal channels used primarily to exchange sensitive information, or for bypassing formal channels, ties are often between people of different formal status within their respective organizations. Apparently, mutual trust is more important than equal rank. These channels tend to retain the elements of a primary, intrinsically valuable relationship, though they are used for essentially instrumental reasons.^[33]

Earlier I expressed surprise that the operations/maintenance subset not only trades expertise but actually coordinates as well. Why? J. D. Thompson has argued persuasively that in most organizations some differentiation of function occurs, referring not to technical specialization in the production of a good or service, but to components of the organization such as the technical core and the buffer. The latter exists to protect the former from changes in the organization's task environment that might disrupt production. While as a whole the organization may be an open system, an adequate buffer permits the technical core to approximate a closed system, thereby improving the effectiveness of production.^[34]

I therefore expected to find that at any particular transit operator a specialized subunit would perform the buffer function for the technical core, in this case rail or bus operations, or both. External com-

munications of all kinds would be conducted by this subunit, allowing bus and rail operations to be run as a closed system, even though interdependence compels the organization as a whole to be treated as an open system. To be sure, the planning sections appear to perform this function.

Against the perspective of Thompson's theory, however, I found extensive contact between the technical cores of the different transit agencies. The cores appear to be far more permeable than Thompson's theory would lead one to expect. It might be countered that his theory does not preclude contact between the technical core of an organization and its environment, that the buffer simply shields the core from unwanted intrusions. But in the Bay Area, the technical cores directly negotiate agreements among themselves to reduce uncertainty and, as I shall show, sometimes intentionally circumvent their own buffers.

Development of extensive interorganizational informal networks means that many more organizational actors become boundary spanners than an analysis of formal responsibilities would suggest. This helps explain the apparent contradiction of Thompson's theory; although he treats organizations as open systems, he considers only formal structure in his analysis of buffer functions. Had he included interorganizational informal mechanisms, he might well have come to a different conclusion.

Informal Conventions and Norms

An important part of coordination is to establish processes for decision making that are essentially continuous, stable, and enduring. Sometimes these processes are spelled out in the formal designs of organizations, often they are not. Even when they are, they are inevitably incomplete, so that effective coordination always depends, at least in part, on the development of informal norms and conventions through group interaction, socialization, and experimentation. Such informal conventions limit the scope of conflict and the range of issues to be considered, establish expectations of behavior on the part of participants, and set out the kinds of factors to be considered in decisions. In short, they provide a ready foundation and a context for coordination. Heymann argues that without

general understandings, there are several reasons why we cannot expect coordination on an ad hoc basis even when a group of parties might see benefits from working together. Some sort of conventions or general rules are essential. Heymann emphasizes rules that are not legally enforceable; he supposes that coordination is at least as dependent upon such rules as it is on formal law itself.^[35] In this sense, motivation for coordination (for example, commonality of interest) is a necessary but not a sufficient condition for coordination to take place. The other necessary condition is the actual capability for coordination, which is provided by informal conventions and norms (as well as informal channels). Public choice theorists have sometimes missed this key point by focusing almost exclusively on differences and commonalities of interest of the parties involved.

Following Lewis, I understand informal norm or convention (I use the terms interchangeably) as follows:

A regularity R in the behavior of members of a population P when they are agents in a recurrent situation S is a convention if and only if it is true that, and it is common knowledge in P that, in almost any instance of S among members of P,

1. almost everyone conforms to R ;

2. almost everyone expects everyone else to conform to R ;

3. almost everyone has approximately the same preferences regarding all combinations of actions;

4. almost everyone would prefer that any one more [person] conform to R, on condition that almost everyone conform to R ;

5. almost everyone would prefer that any one more [person] conform to R', on condition that almost everyone conform to R', where R' is some possible regularity in the behavior of members of P in S, such that almost no one in almost any instance of S among members of P could conform both to R' and to R .^[36]

Informally derived norms are essential even where the formal organization structure is specified in great detail: not all important factors can be taken into account ahead of time, and the character of some vital procedures is such that they are not easily formalized.

When the formal organization is less well specified, informal norms assume greater importance:

The early months of the Economic Cooperation Administration provide a striking example of the ratification of informal relations. Between April 1948, when the agency was established, and about July 15, when it already had some six hundred employees, it operated without any formal plan for its internal structure even so elaborate as an organization chart showing its principal divisions. . . . The real core of this organization, which was in full operation before May 1, lay in a complex set of behaviors and understandings that had grown up almost spontaneously. The formal plans that were finally issued in July and subsequently were in very large pan ratifications of this informal scheme.^[37]

My own reseach on the California Coastal Zone Conservation Commission in 1976 produced similar findings. Few formal structures and procedures were found within the central commission staff, yet there was little confusion as to which persons had which responsibilities and which procedures were to be followed. These functions were all performed by an extensive set of informal conventions that were widely understood by the staff. In response to a query about an organization chart, one staff member replied: "We had one somewhere, but no one pays any attention to it."^[38] Kagan's work on the administration of wage and price controls during the Nixon and Ford presidencies produced similar findings.^[39]

In the case of new organizations, much time must be given to development of informal procedures. Critics expecting immediate substantive results from new organizations often fail to recognize the time and resources required to develop informal norms and procedures. The problem becomes acute when multiple independent organizations come together for the purpose of coordination, because no coercive mechanisms are available to enforce compliance (as there are within single organizations). A complex process of negotiation and accommodation is necessary. The situation is further complicated where no precedent exists on which to base informal procedures.

The RTA illustrates both complications.^[40] Before RTA could produce tangible results on problems of coordination, it had to establish procedures and mechanisms by which it could deal with them.

Much time in early RTA staff-level committee meetings was spent establishing what issues would be considered and how they would be considered. Sometimes these procedures were explicitly specified. The Management Systems Committee tried in its first six months to come up with a viable set of objectives and was finally successful on the third attempt. In other committees, such as Joint Procurement, trial-and-error processes led to an informal procedure for letting out procurement bids. Developing these procedures involved contending with differences in interest and operating procedures among the member agencies. Yet once the procedures were established, they were adhered to closely.^[41]

Washington Metro provides another illustration of the importance of informal norms for establishing decision-making procedures. The construction schedule for Metrorail was formally determined by a long-range plan based on projected patronage figures and cost of construction. In actuality, informal norms have been more important in fixing the order of construction. Given participation in Metro of several local jurisdictions, each with its own constituency and representatives, and economic constraints that preclude simultaneous construction on all planned rail lines, Metro's board had to arrive at some way to "allocate construction." It did so informally, taking into account political factors that the formal procedures did not. According to a former board member, "We give each board member enough construction each year to survive politically."^[42] There are no formal strictures, yet the norm is well recognized, accepted, and defended. A stable pattern of interaction exists, remaining constant even with changes in board membership.

Informal socialization processes and group pressures for conformity produce adherence to the conventions governing acceptable types of decision.^[43] This affects the choice of decision-making style, often excluding conflict-intensifying types, such as unconditional manipulation and prior decision, in favor of bargaining and compensation.^[44] Because at Metro, coalitions are expected to vary from issue to issue—"You never know who you might need next"—persuasion and consensus building are emphasized over confrontation and majority votes over minority objections. These norms are well understood by the board members: "When a new member came on board he didn't know the norms; so we sat on him and he learned."^[45]

Decision makers and others in the organization community know

which approaches are beyond the pale. Manipulation through threats is considered destabilizing and costly to the organizational system. In 1980, Fairfax County began withholding its share of the Metro bus budget, claiming that the agency's irresponsible accounting prevented the county from planning its own budget.^[46] The county's primary concern was the rate of increase in operating costs and the general unreliability of estimates of future bus subsidies.^[47] Condemnation of Fairfax County by other members of Metro and by the press was nearly universal. After almost nine months of withholding by Fairfax, the Washington Star said in an editorial:

Fairfax all along has had democratic access to the Metro Board to make its points on budgeting methods and other issues. And on questions where all of the participating jurisdictions have similar interests (as on the need for accurate subsidy forecasts) it should not be hard to construct a majority of the board to make a desired change.

The Fairfax tactic of withholding money it clearly owes could not be emulated by all of the local jurisdictions simultaneously, without precipitating the financial collapse of the transit system. That would help no one—certainly not Fairfax riders who have used the subsidized service throughout the boycott.^[48]

Besides conformity to group standards, other incentives exist for individuals from various organizations who must work together to adhere to informal conventions: "Where these are not law people may accept them out of the conviction that the stability of the system demands their acceptance."^[49] Certainly this is true at Metro.

Cosmopolitanism and Coordination

Some of these same informal processes also led to the development of cosmopolitan attitudes sympathetic to the idea of coordination among multiple organizations. Cosmopolitan attitudes work as lubricants in situations of interdependence where informal mechanisms are already in place. When representatives of the different transit agencies meet more or less regularly to discuss problems, they may eventually reach accommodations to take back to sell to their respective organizations. Processes of group interaction often result in the development of loyalties to other group members that condition behavior of the individual toward his own organization. For ex-

ample, one San Francisco supervisor who serves on the Metropolitan Transportation Commission is also a member of the governing board of one of the operators. On several occasions he has experienced what might be called "role conflicts" as a result of his overlapping memberships. Two San Francisco mayors have publicly criticized him for misrepresenting or betraying the city's interests in various transit matters before MTC.^[50] Through continued exposure to different points of view, he had developed multiple loyalties and simultaneously a more regional or cosmopolitan attitude.

Furthermore, personal loyalties to personnel from other organizations develop through close working relationships over long periods of time. Such bonds strengthen the social cohesion of a group (composed of individuals from various organizations) that has come together to work on problems of interdependence:

Integrative bonds of social cohesion strengthen the group in pursuit of common goals. Group cohesion promotes the development of consensus on normative standards and the effective enforcement of these shared norms, because integrative ties of fellowship enhance the significance of the informal sanctions of the group, such as disapproval or ostracism, for its individual members. Cohesion, therefore, increases social control and coordination.^[51]

Continued participation in RTA staff-level committees has bred identification with other members of the group and with the group qua group. A Golden Gate planner observed that while the RTA representatives clearly brought the views of their respective organizations to RTA meetings, they were probably more oriented toward regional integration (in whatever form) than others at their organizations.^[52]

A willingness to engage in coordination processes complements an objective need for coordination—that is, participation in coordination activities helps to develop a positive attitude about coordination. For example, the RTA

affords an institutionalized forum for discussion of regional as opposed to operator-specific problems. Such discussions serve as socializing media. They make regional considerations legitimate topics of concern instead of Utopian or unrealistic dreams, and they increase awareness of the advantages associated with cooperative efforts and lay the basis for future cooperation.^[53]

In addition to the development of identification, extraorganizational loyalties, and positive attitudes about coordination, participants in group processes come away with a more thorough understanding of the attitudes of their counterparts and the operating conditions they face. When making proposals for joint action they can better predict the likely responses of their counterparts, thereby improving their chances for success. Thus, informal mechanisms indirectly promote coordination through processes of socialization, knowledge acquisition, and development of extraorganizational loyalties. Members of informal networks and participants in formal coordination efforts develop affiliations with members of other organizations that compete for attention with loyalties to their own organizations. Furthermore, they come to understand coordination problems from a regional perspective rather than from simply the parochial perspective of their own organization. These factors, taken together, work to promote what I call a "cosmopolitan attitude" about coordination.

Following Simon's argument that behavior in organizations is determined by decision premises,^[54] one would conclude that two organizations with similar decision premises are likely to exhibit similar behavior, other things being equal. With respect to coordination, similar decision premises correspond to coordination by standardization as described by J. D. Thompson.^[55] The difference is that in the multiorganizational case similar decision premises are not mandated by some superior authority to the actors in the system.

Similar premises facilitate coordination both directly and indirectly. They narrow the scope of conflict when two organizations come together, providing a foundation for discussion: some issues are simply removed from contention. Through common histories, shared decision premises may also provide such similar approaches to problems that little or no discussion is required for successful coordination:

Common background and training, as well as professional traditions, among public administrators will support some common standards to guide agency decisions.^[56]

Similar premises constitute a potential for what I will call "latent coordination." If two decision makers, X and Y, at two different organizations, base their decisions on similar premises, then their de-

cisions may be well coordinated even if neither X nor Y makes any consideration of the consequences of his decisions for the other. I suspect, but have no conclusive evidence to support the argument, that similar premises cannot overcome serious differences in interest among organizations. But differences in decision premises may inhibit coordination even in the presence of mutual interests. The strained relationship between Muni and Samtrans results in part from very different decision premises for transit operations.

Thus, the public transit system of the San Francisco Bay Area is marked by the absence of a formal mechanism for coordinating the multiple organizations it contains. Simultaneously, the transit system possesses a stable and extensive network of informal relationships between personnel from different agencies, the density of which positively correlates with increased interdependence. Informal channels redress specific failures of formal mechanisms in particular situations and are powerful devices for coordination in their own right and on a regular basis. They provide the conduits through which coordination can take place, while informal agreements and norms of behavior form the foundation for coordination and, under certain conditions, actually coordinate behavior of different organizations without any direct communication.