3—
The Problem of Interdependence

Sets of simultaneous equations as a way to "model" or define a system are, if linear, tiresome to solve even in the case of a few variables; if nonlinear, they are unsolvable except in special cases.
—Ludwig von Bertalanffy, General Systems Theory

At 5:42 A.M. on Monday, the sky is overcast. Philip Wilson boards an AC Transit bus in the Berkeley hills and pays his fare. In downtown Berkeley Phil exits the bus and walks into the BART station. He boards a San Francisco-bound train. He exits at the BART Embarcadero station and walks the four blocks to his office.

An hour later Steve MacFarlane leaves his driveway and parks at the Tiburon ferry terminal, where he catches a Golden Gate ferry to San Francisco. On exiting the San Francisco Ferry Terminal he crosses the street and boards a Muni trolley bound down Market Street. In a few minutes he exits the trolley and walks a short distance to his office.

At 7:31 A.M. Lila Langtree boards a Muni light-rail vehicle headed toward downtown San Francisco from her apartment. She exits the bus to enter a BART station where she catches a train bound for Daly City. At the BART Daly City station she takes a Samtrans bus, which drops her at the departures level of the main airline ter-

minal at San Francisco Airport. She catches the daily Pan Am flight to Paris.

About 11:00 A.M. the cloud cover has begun to break. Bill Jones walks from his house in Oakland to the BART station. He boards a train for Berkeley. At the Ashby station he catches a special AC bus bound for Golden Gate Fields Race Track. By 2:00 P.M. he has bet and lost $150 on the first three races.

At 1:06 P.M. the sun is high overhead, the sky is a brilliant blue, and Thad Slimwit has four minutes to get to his organization theory seminar at Cal. Instead, he catches an AC bus to San Francisco. At the Transbay Terminal he boards a Golden Gate bus to Marin County. In San Rafael he transfers to another Golden Gate bus, which will take him to Muir Beach. By 4:17 P.M. he has had one pint of bitter and finished his first dart game at the Pelican Inn.

About the same time, Jim Fratz leaves his Oakland office and descends into the BART 19th Street station where he boards a train for Fremont. At the terminus he walks outside to catch a Santa Clara bus on its way to San Jose. During the trip he disembarks, walks several blocks and meets his girlfriend of many years at their favorite restaurant, where they order drinks and watch the sun recede in a blaze of orange in the western sky.

Taken together these scenarios depict the connectedness of the Bay Area transit operators at the level at which it matters most—that of the passenger. Each trip, because it involves more than one transit operator, creates interdependence, which then raises issues of coordinating bus and train schedules and routes, physical plant layouts, joint fares, "fastpasses," and transfer discounts, along with issues of long-term capital acquisitions and transit planning. The characters and plot lines of these scenarios are, of course, pure fiction, but as John Steinbeck once said of a story he told in Sweet Thursday, "There are people who will say that this whole account is a lie, but a thing isn't necessarily a lie even if it didn't necessarily happen."

Basic Premises

My views on interdependence start from two basic premises. Organizations are not autonomous entities; not only are their

parts interdependent, but the best laid plans of managers have unintended consequences and are conditioned or are upset by other social units—other complex organizations or publics—on whom the organization is dependent.^[1] Following this premise, the organization cannot be analyzed as a closed system: it is an open system, interdependent with its environment.

Although at times we seem to forget this simple point, to speak a language is to follow the system of organization and classification of data that the language itself decrees.^[2] We focus on some things as our central concerns and ignore others as irrelevant, using the cues of language as guidelines. By the nature of the language involved, an open-system model of organization points the researcher in directions the use of a closed-system model would not. Drawn from biology, the open-system model directs one to discover the linkages between the organization and its environment: one searches for those factors external to the organization that affect its behavior, those factors that describe its interdependence.^[3]

By interdependence, I refer to a condition where two (or more) organizations require each other, are dependent each upon the other. This state is characterized more or less by symmetry among the parties to the relationship. Connections between the organizations involved are nonrandom. However, there is no implication of subordinate status for any of them. It is further recognized that interdependence causes uncertainty for the focal organization. I take uncertainty to mean a "lack of information about future events so that alternatives and their outcomes are unpredictable,"^[4] where "events" are understood as actions by other organizations in the focal organization's environment. What distinguishes relevant uncertainty from the larger universe is threat to the organization, where "threat" is defined as something likely to injure—to be a source of danger, to endanger actively.^[5]

Thus, the behavior of a particular organization in a state of interdependence cannot be understood in isolation: its behavior is affected by and in turn affects the behaviors of those involved in the relationship. Therefore, to say that a set of organizations is described by interdependence is to claim that the set possesses systemic properties such that it is not immediately decomposable into the individual units of which it is comprised.

More specifically, following Lindblom, I define interdependence this way:

Within the set, each decision-maker is in such a relation to each other decision-maker that unless he deliberately avoids doing so (which may or may not be possible), he interferes with or contributes to the goal achievement of each other decision-maker, either by direct impact or through a chain of effects that reach any given decision-maker only through effects on others.^[6]

An organization's interdependence with other organizations in its environment causes uncertainty for it in two ways. Cause-and-effect relations between these units and itself may not be well understood because of insufficient knowledge. Even if the organization achieves an understanding of such relations, it may not then be able to control them to its satisfaction. It will have to discover through research those factors in the environment that cause it uncertainty, and then seek ways to contend with those that threaten it.

Following Thompson, the central problem for complex organizations is one of coping with uncertainty.^[7] Although there are obviously many other potential sources of uncertainty, such as mechanical failure or technological innovation, I focus here on the uncertainty that results from an organization's interdependence with other organizations in its environment.

When another premise is added—that organizations are strongly motivated to achieve what Thompson calls "norms of rationality"—"it becomes abundantly clear why organizations seek to understand and then manage or eliminate the factors in the environment which pose uncertainty for it."^[8] Interdependence with other organizations gives rise to uncertainty, which the organization seeks to reduce or eliminate while operating under norms of rationality. The central problem arising from a state of interdependence is to reduce uncertainty (particularly that which threatens the organization) to an acceptable level by ordering the behaviors of all relevant organizations to lessen the extent to which they impinge on each other's behavior and increase the benefits from their behaviors. Organizations attempt to reduce uncertainty in various ways—through expansion and merger, consolidation and centralization, government regulation, lateral coordination, and warfare.

Policy makers and academics who demand better coordination between and among a given set of organizations implicitly assume the existence of some form of nontrivial interdependence. And they assume that the best if not the only way to deal with it is through centralization. However, given the costs of reorganization and formal hierarchical approaches to coordination, and given Simon's argument about the typical decomposability of most systems,^[9] simple assumptions about interdependence and integration are risky, and the cost of error may be high.

Any analysis that seeks to discover the need for coordination within a given set of organizations and then to make an evaluation of how well that need is being met by existing institutional arrangements must assess first whether any interdependence exists at all, and must then ascertain the extent and character of that interdependence. Only then can we consider what constitutes an appropriate organizational design for handling the particular problems of coordination raised by the actual interdependence (which may include recommending remedies for the deficiencies).

A prime danger of designs not founded on such knowledge is that a formal coordinating structure so developed may itself intensify existing problems of coordination: new problems of coordination may themselves be artifacts of organizational reform. The organizational structure may also be so poorly adapted to the specific problems of coordination it is intended to address that it is incapable of solving them while simultaneously creating additional problems. Such an approach is obviously inconsistent with the view that organizations are rational instruments that must be carefully designed for achieving specific ends under particular circumstances. And yet it occurs all too often. Consolidation plans for the Los Angeles and San Francisco Bay Area public transit systems are merely two of the most recent significant examples.

Operations, Services, and Planning in Bay Area Public Transit

We must first establish the contours of the Bay Area transit system's interdependence^[10] before rendering a judgment on the suitability of existing institutional arrangements. Where does one look

for relationships of interdependence? One practical approach is to begin wherever two organizations have any sort of exchange. For Bay Area public transit several different areas must be examined: where formal jurisdictions of the different organizations are contiguous. where they overlap, or where the organizations in the set interact in common arenas (such as the Metropolitan Transportation Commission or MTC).

Map 1 shows the jurisdictions of the major transit agencies of the Bay Area. including the six operators constituting (with the MTC) the set of organizations under consideration: Alameda-Contra Costa County Transit District (AC); Bay Area Rapid Transit District (BART); the Golden Gate Bridge, Highway, and Transportation District (Golden Gate), San Francisco Municipal Railway (Muni); San Mateo County Transit District (Samtrans); and Santa Clara County Transportation Agency (Santa Clara).^[11]

BART's jurisdiction includes Alameda, Contra Costa, and San Francisco counties, thus containing completely within its bounds areas also served by AC and Muni. AC and Muni touch boundaries in San Francisco. Golden Gate's jurisdiction includes, among others, Sonoma, Marin, and San Francisco counties, thus overlapping with the Muni, with BART, and touching the boundaries of AC. Santa Clara touches BART and Samtrans. Samtrans overlaps geographically with BART and touches the Muni. Where service territories of organizations touch or overlap with others we would expect to find the potential for interdependence.

Interdependence is found in three areas: operations, services, and planning . Because problems of coordination require resolution of what may seem to some rather small details, I hope the reader will bear with the ensuing discussion of the finer technical aspects of interdependence in public transit. The picture drawn here is reasonably accurate and complete for the period of this study (1978–80) but undoubtedly looked somewhat different before and looks different today.

In San Francisco, three operators use the same streets simultaneously: Muni in the conduct of its entire operations, and Golden Gate and Samtrans in the course of delivering and picking up passengers in the downtown area to and from outlying counties. Using the same streets creates several problems. There is the potential for traffic congestion, a problem Muni considered sufficiently serious in 1958 to

^[

Map 1
Jurisdictions of the Six Major Operators and Caltrain

have a section of the Police Code amended to place certain areas of San Francisco off limits to all operators but Muni. There is the matter of limited space and the siting of bus stops and buspads. The type of service to be offered by each operator has also come into question: Can Golden Gate and Samtrans offer local service similar to that of Muni? Muni has vigorously resisted what it considers encroachment on its territory and has come into considerable conflict with Samtrans on the issue, because it believes that its passenger load (and therefore its revenues) is affected by services provided by these two operators. Samtrans and Santa Clara operate on the same streets (although in a less extensive way than in the San Francisco case) in the area in and around Palo Alto.

Joint use of facilities has resulted in interdependence similar to the joint use of streets, the Transbay Terminal in San Francisco being the most important instance. AC, Golden Gate, Samtrans, and Muni all use the terminal, administered by the California Department of Transportation, to load and unload passengers. Problems of access, traffic flow, buspad numbers and siting, parking space, sales, and advertising have arisen. With some private operators also using the terminal, in competition for space with the public operators, there have been additional facilities-related conflicts.

Facilities-related interdependence also results from the use of particular BART stations by more than one bus operator for passenger loading and unloading. Both Samtrans and Muni have bus lines with stops at the Daly City BART station. AC and Santa Clara use the Fremont BART station. AC shares the Hayward BART station with Samtrans. The issues typically involve the number and physical location of buspads at the stations. Conflicts on these issues have occurred at the Daly City station between Muni and Samtrans. Issues of access, signing, location of buspads, and the like occur even where only one bus operator uses a BART station.

Another facilities-related interdependency results from the joint use of a number of rail stations in San Francisco by BART and Muni. BART's trains run in tunnels at a lower level while the Muni Metro (a light-rail system that supplements the old surface streetcar system) uses tracks in an upper set of tunnels. BART constructed the stations and owns them. Important issues have included security, maintenance, allocation of space, and dispensing passenger information.

The second category concerns services-related interdepenence.

Figure 2
Interoperator Transfers, All Trips, Weekdays
Source: Wolfgang S. Homburger and James A. Desveaux,  Joint Transit
Fares in a Multi-Operator Region: A Conceptual Plan for the San
Francisco Bay  Area (Berkeley: University of California, Institute of
Transportation Studies, 1980), p. 18.

Because no single operator's jurisdiction encompasses all possible travel patterns originating or terminating within it, patrons daily make trips that involve more than one operator. Figure 2 shows the extent of interoperator trips on weekdays among the six major operators. Interoperator transfers create interdependence between pairs of operators: problems of passenger load, the routing of the bus lines, and the physical location of transfers become issues, as well as financial questions involving joint fares and transfers and schedules for buses and trains. How many should run and at what times? Will the operators involved have signaling devices to indicate to each other when loads of passengers are about to arrive?

Decisions by any one operator on any or all of these issues affect the ridership of other operators with which it shares patronage. Issues concerning transfers send financial reverberations through the operators involved. Figure 2 shows some transfer traffic between AC and BART, AC and Muni, BART and Muni, Samtrans and BART,

and Golden Gate and Muni. Figure 2 does not indicate transfer traffic between Southern Pacific (now Caltrain) commuter trains on the peninsula and Samtrans, Santa Clara, and Muni; but that traffic is also significant.

Overlap in transit services creates a second set of interdependencies in the service area. If we assume a basically fixed market size or at least a slowly growing market, decisions on fare structures and other services to be offered by one operator directly affect other operators by changing their shares of the market. The transbay travel corridor is the most important instance of this type of interdependence. BART and AC both provide service from several areas in the East Bay to San Francisco. To a lesser extent, BART and AC provide competing services along a north-south corridor in the East Bay from Richmond to Fremont. In San Francisco, BART and Muni offer similar services in some areas, while BART and Samtrans duplicate each other's services on part of the northern peninsula travel routes.

In each area of overlap the interdependent character of operator relationships has become readily apparent when one has altered fares (the AC-BART transbay fare issue has been especially noteworthy), altered headways, changed hours of operation, rerouted its service, or in other ways changed the attractiveness of its services relative to those of other operators. The depth of interdependence on the transbay corridor is never more evident than when through catastrophe or labor-management dispute one operator is unable to provide its regular service.

Another area of interdependence became evident when the imminent termination of a set of transit services threatened to place difficult burdens on several operators and, in general, harm the quality of public transit. For many years Southern Pacific provided commuter service to and from San Francisco along the peninsula corridor running south toward San Jose. If this service was terminated, Muni, Samtrans, and Santa Clara would all be affected in terms of transfer traffic and primary passenger loads. In the late 1970s Southern Pacific sought to withdraw commuter service because of its increasing unprofitability, making clear to these operators its importance to their own operations. There are also several cases in which interdependence in the services area has been created by contractual arrangement for provision of services by one operator for another. Once established, contractual arrangements acquire lives of their own: where

no interdependence existed, it now does. AC runs express buses for BART that serve as extensions of rail service in bart's jurisdiction where rail service is not yet available. This contract service creates many questions about routing, schedules, labor agreements, insurance, maintenance, and bus specifications. In a similar case, BART was engaged by Muni to maintain the tracks and tunnels of Muni Metro (the light-rail system), which raised issues of labor agreements and recruiting, specific responsibilities, and payment schedules. And Golden Gate rented shortened buses from AC for use in weekend service on Mount Tamalpais. Issues included timing and method of bus pickup and return, emptying the fareboxes, and methods of payment.

Another instance of interdependence in the services area also has to do with contractual arrangements. Under the dual rationales that when one operator loses passengers to a service breakdown all public transit suffers and that one operator is obligated to assist another in times of emergency,^[12] "mutual-assistance" agreements were forged between AC and BART and between Muni and BART. Should either party to the agreement suffer a temporary service breakdown, the other provides service to fill in until the situation has been remedied.

The operators have also come together to exploit "efficiencies" resulting from economies of scale. Economies of scale and sharing of expenses cover a wide range of activities. Through the Regional Transit Association (RTA), various items have been obtained at lower cost than any single operator could secure on its own. Other joint activities have included bus "roadeos," marketing research projects, and a Bay Area transit information center. Another program jointly administered by the six operators is an elderly and handicapped discount identification card.

A final set of interdependencies falls into the general category of planning. As with the operations and services areas, planning interdependence is multifaceted. Interdependence in this area stems from several independent factors that have interactive effects beyond their simple additive properties. AC, BART, Golden Gate, Muni, Samtrans, and Santa Clara all fall within the jurisdiction of the MTC, which was established in 1970. Among its major responsibilities, MTC is to provide a long-range plan for all transportation modes for the nine counties of the San Francisco Bay Area and is responsible for seeing that the transit operators participate in that plan as they

develop plans of their own; the operators' decisions must also adhere to the overall plan as it is completed. Additionally MTC administers all federal and state grants and some local transit tax monies. By 1986, this amounted to over $500 million. Its role includes ensuring that recipients—the local operators—adhere to federal, state, and local guidelines, which have often included requirements for standardized procedures and increased interaction among the operators.

Placement under the jurisdiction of a single transportation planning organization affects interdependence among the transit operators in several ways. Although federal and state monies have always amounted to less than the total of the desires of all operators, the fact of MTC placed them in direct competition with each other for funds by virtue of a common arena for application and evaluation. Several factors combined to heighten interdependence from the competition.

First, the statewide ballot initiative, known as Proposition 13, which passed in June 1978, greatly eroded the tax revenues available to local governments (including transit agencies) by placing strict limitations on real property tax rates. Such taxes were the principal tax base for many transit operators. AC, for example, depended on property taxes to make up the bulk of its operating losses. In general, the passage of Proposition 13 meant increased dependence of transit operators on federal and state subsidies.

Second, in the face of increasing demand, intensified by dramatic increases in petroleum prices beginning in 1973–74, the transit operators have attempted to provide expanded services, requiring larger subsidies. Applications by one operator for grant monies affect the abilities of other operators to attain their goals. If each operator had an independent source of revenue sufficient to support its operations, this concern would be irrelevant; but such self-sufficiency is a thing of the past. For example, in Marin County, if certain funds administered by MTC were allocated to Marin County Transit District (MCTD), they are invested in local service. If those funds went to Golden Gate, they were put into long-haul service. Outcomes differed, depending on the actions of each operator. No operator can afford to ignore the actions of others.

Two related factors serve to increase interdependence in the planning area. The fact of a common arena (MTC) invites direct comparisons among the operators across a broad range of subjects, from planning departments to the quality of maintenance schedules for

buses. Such comparisons, made only indirectly and much less frequently prior to MTC, are now made directly and regularly. The operators can no longer act in isolation. The mere presence of a single transportation planning organization has provoked demands for consistency of approach in all kinds of areas. For example, when one operator moves to change its fares, that move is felt by other operators and monitored by the MTC. There have been and continue to be pressures for a unified regional fare structure.

Similarly, the operators must comply with the components of the regional transit plan, as well as with state and federal guidelines, as a condition for receiving federal and state monies. For example, Muni did not have a planning unit, nor was it interested in establishing one on its own. But because of federal and state guidelines, MTC refused to release certain funds to Muni until Muni created such a planning unit. Muni reluctantly complied. These factors have combined to increase interdependence among the transit operators over the decade 1972–82.

Bilateral and Multilateral Interdependence

How can we summarize interdependence in these three areas in a way that will help decide the appropriate machinery for coordination? Given that the complexity of problems heightens with increases in the number of actors involved, it is sensible to start with numbers. How many organizations are involved in any given instance of interdependence? Is it bilateral or multilateral?

The relevance of these questions lies in the attempt to discover exactly how loosely or tightly coupled a given system of organizations might be. Table 3 shows the interdependence of the operators distributed into bilateral (simpler cases) and multilateral (more complex cases) categories.

The rationale for making this distinction hinges on an idea sparked by Thompson's brief discussion of "localizing interdependence,"^[13] Bertalanffy's arguments about the complexity of systems, and Simon's contention that most systems are decomposable.^[14] Thompson suggests that interdependence within single organizations can be localized by the creation of departments along functional lines, thus reducing costs of coordination by locating highly interdependent functions of an organization close together.

As opposed to localizing highly interdependent parts of an organization, can different areas of interdependence be isolated from each other in the multiorganizational context? If one interdependency exists between two organizations and another interdependency exists between one of those organizations and a third organization, are these two different interdependencies connected? Must the first instance be treated as a single occurrence of multilateral interdependence or can it be treated as two separate cases of bilateral interdependence?

Following Simon's assertions about decomposability, the set of relationships among the three organizations is probably less complex than appearances might indicate. Can we specify the conditions under which they will be connected and the conditions under which they will be separate? More concretely, are the interdependencies so structured that decision makers can isolate one from another in their decision-making processes, thereby simplifying the problems? If so, we should be able to employ simpler coordination mechanisms than a first glance might suggest.

Localizing interdependence improves the likelihood of finding a solution. Treatment as an instance of two bilateral interdependencies lessens cognitive complexity. As Hardin notes in his discussion of contract by convention,

the more people whose behavior one must know well enough to consider it predictable, the less one will be able to know about each of them on average. Hence, the prospects for successful contract by convention decline as group size increases.^[15]

Simultaneously such treatment reduces the minimum number of parties necessary to its successful resolution, potentially diminishing conflict over values as well. Another way to approach this question is to think of solving a set of problems sequentially or serially instead of attempting simultaneous solutions or setting the solution of one problem as a precondition for the solution to another. Localizing interdependence permits fuller enjoyment of the fruits of coordination by reducing the complexity of any single problem, thereby permitting the use of simpler mechanisms for coordination.

The capacity of this approach for reducing the difficulty of coordination is clearly seen in an example from the Bay Area. Several thousand passengers per day use both Samtrans and BART to make a single trip. Close to fifteen thousand trips are made daily using both AC and BART. Yet there is virtually no linkage between AC and Samtrans—perhaps two hundred and fifty trips per day. Is this a single instance of multilateral interdependence or are there two instances of bilateral interdependence that involve one organization in common? Do actions of AC affect the goal achievement of Samtrans and vice versa? Do decisions made by BART concerning its links with AC affect goal achievement at Samtrans? The response to each question is no.

The ratio of total daily trips on BART to those trips involving transfers with either AC or Samtrans is very high. Such transfer trips, although significant, remain a small portion of bart's overall business (about 10 to 15 percent), and AC and Samtrans have such a low direct transfer rate between them that their interdependence in this area is negligible. Because BART runs constant headways instead of a fixed time schedule for its trains (although it does publish a schedule), there is little sense in trying to arrange specific train-bus "meets," thus eliminating one complication. The stations at which

AC transit and Samtrans meet BART are mutually exclusive (with the exception of the Hayward station). Problems arising from the shared use of facilities by AC and Samtrans are not significant.

Because of these facts, on the surface what appears to be a case of multilateral interdependence is in fact two instances of bilateral interdependence. BART can (and does) treat its relations with AC and its relations with Samtrans as two distinct problems. These conditions also permit AC and Samtrans to deal with BART without either having to worry about the actions of the other. Furthermore, there is no incentive for any of the actors involved to treat the situation as a multilateral interdependency. If, however, the percentage of BART's total passengers transferring to and from AC and Samtrans increased dramatically, or if the number of stations at which AC and Samtrans both transfer passengers to BART increased, or if BART changed to a fixed schedule, or if some combination of these three changes occurred, then it might become a situation of multilateral interdependence, with correspondingly more complex issues and more difficult solutions.

Similarly, can one interdependency between two organizations be separated from another interdependency between the same two organizations? Isolating interdependencies between the same two organizations lessens the spillover effect from one to the other. The most "dangerous" problem posed by spillover from one area of interdependence to another, described by Schelling, involves two types of bargaining, each distinguished by the nature of the interdependence involved. There is "what might be called the 'distributional' aspect of bargaining: the situations in which more for one means less for the other. . . . When two dynamite trucks meet on a road wide enough for one, who backs up? The other aspect of bargaining consists of exploring for mutually profitable adjustments, and that might be called the 'efficiency' aspect of bargaining."^[16]

AC and BART both provide transbay service. The most lucrative lines for AC have historically proved to be its transbay routes. When BART was in the planning stage and during the early days of operations (which included transbay service), it pushed AC to drop its transbay lines and become solely a feeder service for BART. AC resisted successfully, and the two operators remain competitors for that market. Fierce at times though it may be, this competition has not precluded coordination in a second area of interdependence. One

way of making AC and BART more effective transit operators is to simplify transfers from one mode to the other. AC and BART have worked out transfer arrangements comparable with and surpassing in certain respects the arrangements for rail-bus transfers at the Washington Metro, a unified rail and bus agency. BART and AC have been able to put the conflict over transbay service on hold, agreeing to disagree on that point, thus isolating it from the interdependency that has the promise of mutual gain (intermodal transfers). And there have been no quid-pro-quo discussions of concessions in one area in return for concessions in the other.

Although the confounding of any one area of interdependence with any other makes the situation more complex intellectually and politically, a special potential for difficulty exists when a "distributional" problem is linked with a problem of "efficiency." Competition for transbay patronage is an example of the "distributional" element, whereas interoperator transfers represent the "efficiency" aspect. Keeping the two areas independent of each other, as in the AC-BART example, illustrates the positive benefits of localizing interdependence. Under these conditions, we would certainly prefer to employ coordination mechanisms that permit the separation of these different types of interdependence.^[17]

What explains the AC-BART situation? It is characterized by a high degree of conflict in an area of interdependence where the probability of a clear winner and a clear loser is high. BART seeks to change the status quo and eliminate AC from competiton for the transbay market. At the same time, BART needs AC's feeder service to and from BART. There is no reason why the two interdependencies need be treated as one, and there is no political incentive for BART to accept such an arrangement. There is an incentive for AC to make the link, however. It has not done so, largely because BART has lessened its demands.

Scott notes that greater interdependence is associated with more elaborate coordination structures.^[18] He suggests that higher levels of interdependence necessitate more extensive and complicated coordination mechanisms, an argument similar to the one made here. However, if this statement were turned on its head, it would be interpreted to mean that more elaborate coordination structures precipitate higher and more complex levels of interdependence. By increasing the requirements for successful coordination, the level of interdependence

among the components of the system may be artificially increased as an unintended consequence. Under these conditions, coordination requires greater agreement across a broader range of values and the solution of cognitively more complex problems. This appears to be so in the Washington Metro transit system, where issues whose analogues are but loosely connected in the Bay Area are tightly linked together. In general, the relatively lower capacity of the Metro to localize interdependence, as compared with the Bay Area agencies, appears to result from the fact of a unitary, centralized decision-making body—a single arena for resolving all issues.

Metro is a unified operating authority whose jurisdiction includes parts of the states of Maryland and Virginia and the whole of the District of Columbia. Real differences in interest and ideological perspective mark the seven local jurisdictions that are members of Metro. Issues of rail construction finance, rail construction schedules, fare levels, bus and rail subsidy costs, and transfer policies all arise on a regular basis. However, rather than treating them as separate issues whenever possible and solving them independently, at Metro these issues tend to become intertwined. Consequently, separating different areas of interdependence between the same two members at Metro is exceptionally difficult. Similarly, areas of interdependence that in this study would be characterized as bilateral tend to be treated as multilateral, with the dual results of increasing conflict within Metro and rendering coordination solutions correspondingly more difficult and costly. Not only are distributional aspects such as bus subsidy allocations tied to efficiency aspects such as transfers, but issues that need involve only two members typically find their way into the common arena for resolution, where more actors become involved.

Furthermore, if problems of interdependence can be dealt with separately, the quest for consistency becomes less fervent. Discretely arranging coordination solutions for problems of interdependence is more readily accomplished when a push for consistency among those solutions is absent. Consistency is a far more mischievous hobgoblin in Washington Metro than in Bay Area transit, resulting, I contend, from the institutional arrangements.^[19]

The need for coordination has so far been considered as deriving directly from a state of interdependence. With interdependence, positive benefits may be gained from some sort of coordination, or

negative consequences may be avoided—uncertainty and conflict may be reduced. However, when mechanisms for coordination are overly complex relative to the extent and type of interdependence, the causal flow may be reversed, with the mechanisms themselves creating higher levels of interdependence, linking organizations more tightly than before, and necessitating more complex and less readily achieved solutions.

Origins of Bilateral and Multilateral Interdependence

I began the search for patterns in interdependence by distinguishing multilateral from bilateral cases. I continue that search by looking for differences in its origins. Three general origins can be distinguished. Interdependence occurs naturally when a variety of forces beyond the control of the organizations immediately involved come together to cause them to become connected. This is the type most often considered (explicitly or implicitly) in discussions of interdependence. For example, whenever a significant number of patrons elect to take trips involving more than one transit operator, interdependence is naturally created. The scenarios presented at the beginning of this chapter are all cases of natural interdependence. In response the operators may do nothing or may actively encourage or discourage such travel. But whatever they do, they remain linked through these "naturally" occurring external factors. The behavior of one affects the goal achievement of the other.

Instead of being caused by a natural series of events or a compendium of individual actions, interdependence may result from deliberate efforts of an outside party to link two or more organizations for some purposes of its own (which may have little to do with the goals or interests of those two organizations). Call this "artificial" interdependence (in the sense of artificial organization versus natural organization). The actions of MTC in Bay Area transit illustrate this type of interdependence. Operators applying for capital or operating subsidies are forced into interdependence through demands by MTC for conformity to various standards and plans of its own construction.

Interdependence also results when organizations voluntarily enter

into arrangements to realize some array of mutual benefits, as when one operator seeks to have another perform services on a contractual basis. Contractual arrangements of this sort are extensive in the Bay Area and are of two types. In the first, one operator performs services for another for compensation. The second type involves coordinated joint action to secure the benefits of scale economies. Over time these arrangements acquire an institutionalized presence, so that interdependence develops between the organizations that does not hinge on external circumstances. This interdependence goes beyond simple interlocking effects of decisions concerning the contracted service. Simply observe the effects of cancellation of such service on both organizations. Call this "voluntary" interdependence. In natural interdependence, coordination follows interdependence. In voluntary interdependence, the converse is true. However, this is to be distinguished from a situation where use of an inappropriate device for coordination leads to increased complexity of the interdependence. Table 4 shows the interdependencies of Bay Area transit categorized by their origins.

Bay Area transit is characterized by both bilateral and multilateral interdependence, with most interdependence taking the bilateral form. There are several important differences in the origins of bilateral and multilateral interdependence, as shown in Table 5. Naturally occurring interdependence is, with one exception, bilateral, whereas no bilateral interdependence is of the artificial type. Without external intervention, naturally occurring patterns of interdependence take the bilateral form: they are relatively simple in character. Conversely, without exception, the artificially developed interdependence in the Bay Area is of the multilateral type. Voluntary interdependence falls into both bilateral and multilateral categories.

Figure 3A shows the character of the Bay Area transit system, when natural, artificial, and voluntary origins of interdependence are considered together. The number of lines connecting any pair of operators indicates roughly their degree of interdependence. The six major operators are linked together in a system characterized by multilateral interdependence, such that decisions by one operator affect all the others, both directly and indirectly. But a very different picture emerges when only natural interdependence is taken into account, as shown in Figure 3B. Instead of a tightly coupled system of

multilateral interdependencies. Bay Area transit appears to be a more readily decomposable system. It is more loosely coupled and characterized by a series of bilateral interdependencies that include overlapping sets of operators. Even when voluntary interdependence is considered with the natural type, the picture is not greatly complicated; the predominant character of the system is bilateral. One concludes that, in good measure, the current complexity of Bay Area transit results from the coordinative arrangements employed there, especially from actions by the MTC.

Figure3
Interdependence in Bay Area Public Transit

Conclusion

In general most systems are decomposable. Empirically, it has been shown that for the most part, naturally occurring interdependence in the Bay Area is relatively simple, taking a bilateral form, permitting the transit system to be considered as a series of smaller problems of coordination—that is, it can be decomposed.^[20] More complex multilateral interdependence has resulted from voluntary action of the operators themselves or has been forced on them by external parties such as MTC.

Because the ability to comprehend a system decreases as the number of components and the complexity of their interactions increase—coordination structures may themselves increase complexity and decrease comprehension—and because the distributional and efficiency aspects of bargaining and the danger of spillover remain distinct, great caution should be exercised when altering institutional arrangements of any multiorganizational system such as Bay Area transit. It will likely serve to increase interdependence in the system without guaranteeing significantly better coordination. Furthermore it is foolish to impose coordinative arrangements predicated on high levels of multilateral interdependence on a system actually described by a series of bilateral interdependencies, which themselves do not affect the individual transit agencies greatly. In this case, following the rule of the lowest common denominator is probably the most effective strategy: use the minimum mechanisms necessary to achieve a satisfactory level of coordination.