JOURNAL
OF COMPARATIVE
Incentives,
Income Sharing, and Institutional Innovation in the Yugoslav Self-Managed Firm’ LAURA
Department
3, 285-301 (1979)
ECONOMICS
of Economics,
D’ANDREA
University
TYSON
of California,
Berkeley,
Tyson, L. D.- Incentives, Income Sharing, and Institutional Yugoslav Self-Managed Firm
California
94720
Innovation
in the
The incentive properties of income-sharing arrangements are analyzed for situations in which workers are allowed to form group or coalition decisions on effort. Both the feasibility and enforceability of such decisions are shown to depend on a variety of organizational characteristics, including the degree of worker participation and the size and divisionalization of the firm. Finally, the recent decentralization of Yugoslav firms is shown to enhance the incentive effects of income sharing. J. Comp. Econ., Sept. 1979, 3(3), pp. 285-301. University of California, Berkeley, California 94720. Journal
of Economic
Literature
Classification Numbers: 052, 5 11.
1. INTRODUCTION In a labor-managed enterprise workers do not receive a fixed wage in payment for a specific labor contract. Instead, they earn a share in the net income remaining to the firm after payment of all nonlabor operating costs. There is a presumption in the literature on incentives that incomesharing (or profit-sharing) arrangements of this variety have superior incentive effects in the sense that they encourage greater worker effort for a given amount of labor time than do fixed-wage contracts. Indeed, this is considered by the proponents of income-sharing schemes to be one of the main arguments in their favor. Existing theory, however, also suggests that the incentive properties of income-sharing schemes are weakened considerably by the free-rider problem. In the absence of supervision of individual workers, income sharing fails to distinguish variations in worker effort, and each worker must share the income generated by his 1 The author is appreciative of research support provided by the American Council of Learned Societies and by the Hoover Institution. The current version of this paper was completed while the author was a National Fellow at the Hoover Institution. The author is grateful to Hajime Miyazaki and David Conn for helpful comments. 285
0147-5967/79/030285-17$02.00/O Copyright Q 1979 by Academic Press, Inc. All rights of reproduction in any form reserved.
286
LAURA
D’ANDREA
TYSON
additional effort with his co-workers. Under these circumstances, as the number of workers covered by an income-sharing plan increases, the incentives for additional effort by an individual worker correspondingly decline. An implicit assumption underlying this conclusion is that under an income-sharing scheme each worker is certain only of his own effort and of the fact that he must share the proceeds of his effort equally with his co-workers. Furthermore, each worker is assumed to choose his own level of effort without any information about the effort supplied by other workers, and the choice of each worker is assumed to be independent of the choices of others. Under these individualistic circumstances, the possibility that workers might jointly determine coalition or collective effort is ruled out by assumption. Once the possibility of coalition decisions is introduced, however, the traditional propositions about the incentive properties of income sharing must be reconsidered. In particular, as this paper demonstrates, if workers are able to form a coalition decision about the effort to be supplied by each of them and if they are able to enforce this decision in some formal or informal manner, then the freerider problem of income sharing is mitigated, and its incentive properties are correspondingly enhanced. Naturally, the feasibility of coalition formation among groups of workers depends on certain characteristics of the larger enterprise organization to which the workers belong. One such characteristic is the extent of worker participation in enterprise decision making. Following Steinherr (1977)) worker participation is understood here as an overall organizational characteristic rather than as a choice variable for individual workers. In this view, enterprises differ in the extent of worker participation. At one extreme, there is no participation, the firm is hierarchically organized, and managers hire workers who are subject to an authority relation. At the other extreme, there is full participation, and workers hire managers who act as consultants. In between the two extremes, labor and management share decision-making rights in some predetermined manner. The main objective of this paper is to formalize and underscore the intuitive notion that labor management, defined by income sharing and full worker participation, provides a unique enterprise structure that is conducive both to the formation and enforcement of coalition decisions among workers and to the strengthening of the incentive effects of income sharing. A second objective is to demonstrate how the divisionalization of a labor-managed firm into separate groups of workers, each with its own imputable income, can itself enhance both the feasibility and enforceability of coalition decisions, thereby further strengthening these effects. A third objective is to examine recent organizational reforms in the Yugoslav labor-managed firm in light of the formal analysis of the inter-
INCENTIVES
IN THE
SELF-MANAGED
FIRM
281
relationships between enterprise organization and incentives. The purpose here is to demonstrate that these reforms, like many others introduced during the postwar period in Yugoslavia, can be understood as an attempt to design enterprise structures that enhance the incentive effects of income sharing. 2. INCOME
SHARING AND WORKER A SIMPLE MODEL
INCENTIVES-
In the simplest textbook case of an employment contract between a hierarchical firm with no participation and an individual worker, a fixed quantity of labor, measured in units of time, is hired in return for a fixed wage paid per unit of time. In this simple contractual setting there is no uncertainty from the point of view of either the employer or the worker. The worker knows he will receive the contractual payment if he supplies the stipulated labor time, and the firm is able to supervise the labor input to guarantee that the contract is honored. In this simple model, the only dimension of labor input is a quantity dimension that is easily specified, measured, and monitored, and that is traded between the buyer of labor and the seller at an explicit price. In this model the relationship between realized output and labor input can be expressed as 4 = q(L), where q is output, L is labor input measured in units of time, and q1 > 0. The employment contract becomes more complicated when the buyer of labor cannot easily monitor all dimensions of labor performance. Suppose, for example, that besides its quantity dimension, labor has a quality dimension, called effort, that is costly to measure and monitor. In this case, the production relationship is q = q(L,e), where q1 > 0, q2 > 0, and e is labor effort2 Under these circumstances, the labor contract may fail to specify explicitly the terms of exchange for labor effort, although effort is indeed one of the commodities traded in the contractual relationship.3 Labor effort is then an unmeasured and unpriced commodity in exchange and, as a consequence, there is nothing to guarantee that its exchange will be optimal from the point of view of either the buyer or the seller. An example illustrates this point. Suppose because of the difficulties of monitoring worker effort, the firm offers only one kind of contract: a fixed-wage contract based on some estimate of the minimum or average effort con2 There are problems involved in measuring e or labor effort in the real world. We abstract from these problems here and assume for simplicity that e can be interpreted as either a cardinal or ordinal index of labor effort. 3 Stiglitz (1974, p. 243) argues that because of the complexity of specifying and enforcing labor inputs by contract, the assumption that the employer cannot supervise the contract performance may be a better approximation of reality than the assumption that the employer can costlessly supervise the contract. Ironically, however, it is the latter assumption that underlies the traditional literature on employment contracts.
288
LAURA
D’ANDREA
TYSON
sistent with acceptable costs of supervision. This contract prevents the worker from choosing the optimal effort-income configuration that would maximize his utility for a given quantity of labor. Under the usual assumption that increases in effort are sources of disutility to the worker, the constrained equilibrium position will be found at the minimum effort level consistent with the contract. Because of the inefficiencies inherent in a contractual arrangement that fails to specify a price for an important traded commodity, namely, labor effort, alternative contractual arrangements that explicitly allow for the exchange of labor effort between employer and employee are preferable. One such arrangement is an income-sharing contract where workers receive a share in enterprise income either in addition to or in lieu of a fixed wage.4 This arrangement allows each worker to choose his own utility-maximizing combination of effort and income.5 The problem with such an arrangement, however, is that in the absence of supervision of individual-worker effort, the sharing of income fails to distinguish variations in the effort levels of different workers. Because returns to the additional effort of an individual worker must be shared with his co-workers, the incentive effects of income sharing, as perceived by the individual worker, are correspondingly reduced. To demonstrate this proposition, we begin with a simple employment situation in which an individual worker is paid a fixed wage in return for a constant amount of labor time and some minimum (zero) effort leve1.6 We define for the worker a utility function reflecting his preferences for ease, which is the negative of effort, and earnings, holding labor time constant. That is, U = U( -e, z; t), where -e is ease, z is labor earnings, and i is the fixed amount of labor time supplied on the job. Given the employment contract, the utility level of the worker is just u = U( 4, W), where -2 is the ease level consistent with the minimum effort requirement of the contract and W is the fixed wage. We assume that this represents 4 The optimal composition of total worker income between a fixed payment and a share in income depends not only on the incentive effects discussed here but also on the relative degrees of risk aversion of the employer and the employee. See Stiglitz (1974, 1975) for a full discussion of the tradeoffs between incentives and risk sharing that influence the optimal compensation structure. 5 When the only available contracts offer an effort-income tradeoff that diverges from the one workers would choose on their own accord, the employment relationship is one of continual disequilibrium requiring employer supervision to prevent employees from adjusting to their preferred position. Only when individual workers are at this position will the employment relationship be in overall equilibrium. This conclusion mirrors a more general conclusion about complex organizations-namely, that the organization is in equilibrium if and only if each member of the organization is in equilibrium. @The analysis that follows is a formalization of arguments presented in Vanek (1971, Chap. 12).
INCENTIVES
IN THE SELF-MANAGED
FIRM
289
a constrained equilibrium position for the worker in the sense that he would choose a higher level of both effort and earnings if the employment contract allowed him to do so. Now suppose that an incentive scheme is introduced that allows the individual worker to earn all of the returns to additional effort beyond the required minimum. To find the worker’s optimal effort-earnings choice under this new arrangement we first redefine the origin of the utility function so that @O,O) = 0 = U( -2, W). The worker’s utility-maximization problem under the new incentive scheme can be expressed formally as Max o( -e,y)
(1)
subject to Y =f(e),
(2)
wheref is the transformation function relating effort to income and y is worker income (in excess of the fixed wage). Substituting (2) into (1) and solving for the first-order conditions yields 01/02
= f ‘(e),
(3)
which implies that at the new optimal position the marginal rate of substitution of income for ease will just equal the marginal product of effort7 The equilibrium effort-income configuration implied by problem (1) and characterized by condition (3) is pictured in Fig. 1 at point E, which has an associated utility level I?( -4, 9). For simplicity, the effort-income transformation function is assumed to be linear, reflecting constant returns to effort and ruling out possible production complementarities and indivisibilities. Now consider a firm with n workers each of whom has an identical utility function. Assume that the firm operates with a simple incomesharing scheme under which the income earned from the additional effort of each worker is shared equally among all workers. In this case, from the point of view of the individual worker, the effort-income transformation function must be rewritten as Y =f(eYn,
(2’)
which indicates that for the expenditure-of-effort level C, the worker now receives only f (2)/n rather than f (2) as in the original problem.8 Substituting this relationship into Eq. (1) and solving for the new first-order ’ This condition is analogous to the usual condition in the labor-supply literature that the marginal rate of substitution of income for leisure just equals the marginal product of labor. a For simplicity we assume that the effort-income transformation function for the individual worker is unaffected by variations in the effort levels chosen by other workers. We are thus ruling out production complementarities among the effort inputs of individual workers.
290
LAURA
D’ANDREA
TYSON
cEffort\Ease
FIGURE
conditions
-
I
yields O,lO,
= f ‘(e)/n,
(3’)
which implies that at the new utility-maximizing position, the marginal rate of substitution of income for ease will just equal (l/n)th of the marginal product of the individual worker’s effort. The new equilibrium position for a fixed n is shown in Fig. 1 as point Ez which has an associated utility level I!?( -$, f(&)/n) that is less than the utility achieved at E,. Comparing the E2 equilibrium with the El position and condition (3’) with condition (3), we see that an increase in the number of workers reduces the willingness of an individual worker to give up ease in order to produce additional income, and reduces the attainable utility level as perceived by the individual worker. In the limit, as n becomes very large, each individual worker will prefer to exert no effort beyond the minimum required for receipt of the fixed wage and will generate no additional income. In this limiting position each worker will return to his original constrained equilibrium position 0 = U( -6, @). This simple analysis thus supports the intuitive notion that as the number of workers in an income-sharing scheme increases, the incentive effects of that scheme, as perceived by an individual worker, diminish. An implication of this conclusion is that income sharing is likely to be more effective as an incentive mechanism, the smaller the number of workers participating in the sharing arrangement. 3. INCOME
SHARING
AND WORKER
COALITIONS
The arguments presented in the previous section rest on the assumption that under an income-sharing scheme, each worker is certain only of his
INCENTIVES
IN
THE
SELF-MANAGED
291
FIRM
own effort level and of the fact that he must share the returns to his effort with his co-workers. Furthermore, worker communication is ruled out by assumption. Each worker is assumed to choose his own effort level independently of the choices made by his fellow workers. These assumptions appear to be excessively restrictive and unrealistic in the context of a labor-managed firm, where full labor participation requires that individual workers communicate with one another to establish a collective or coalition decision on all aspects of enterprise performance, including the effort to be supplied by each of them. Indeed, recent theoretical work by Ichiishi (1977) and Greenberg (1975) suggests that the most appropriate model for a labor-managed firm is the model of a production coalition. Such a coalition consists of a group of agents bound together for production purposes by an ex ante implicit or explicit contract or pact governing individual duties and responsibilities and specifying a payoff or sharing rule whereby the income of the coalition is distributed among coalition members.s For the purposes of this paper an n-member coalition C = { 1, . . . , by its transformation function or proJ, . . . , n} can be characterized duction-possibility set F( e, C) = Y, where e = (e,, . . . , e,) is a vector of the effort levels supplied by coalition members, ej is the jth member’s effort input, Y is coalition income, and where we assume that labor time is fixed for each coalition member. In order to specify a payoff function mj for the jth member we use a convenient notation dj = (e,, . . . , ej-1, ej+l, - - . , e,), where tj is an (n - I)-dimensional vector describing the effort levels of all coalition members other than the jth. Using this notation, the payoff to thejth member can be expressed as a function rj(ej,Zj,C) that is the indirect utility function of ej conditional on (Zj,C). Thus, mj(ej,dj,C) E max-,, Uj(-ej, yj), subject to yj = Sj(ej,dj,C), where Sj is the jth member’s share of income in the coalition, conditional on 6,; yj is the jth member’s income from the coalition; and Uj is the jth member’s utility function. In the simple labor-managed coalition where equal sharing is the rule Sj(ej,Zj,C) = F(ej,t?j,C)/n = Y/n. Using this simple model, it is easy to demonstrate that once coalition decision making is allowed, the participation of large numbers of workers in an income-sharing scheme need not reduce the incentive effects for individual workers. In the simplest case, where all workers joining the coalition have identical utility functions and where the transformation function for the coalition is just the sum of the transformation functions for the individual members, F( ej, 2j, C) = nf( ej), the demonstration of this proposition using Fig. 1 is straightforward. We begin with a group of workers each of whom, in the absence of a coalition decision, will choose the effort-income configuration at E2, 9 The Miyazaki
aspects of implicit and Neary (1979).
contracting
among
the
coalition
members
are
analyzed
by
292
LAURA
D’ANDREA
TYSON
which implies an effort level sj. Now suppose that the workers form an ex ante coalition in which each of them agrees to work at effort level Pj provided that others do likewise. As long as each individual worker can be certain that his choice of 2, will be matched by identical choices for all other workers-that is, as long as each individual worker can be certain that the coalition decision will be enforced-he can be certain that his total earnings-both from his share in the income generated by his own effort and from his share in the income generated by the effort of the remainingn - 1 workers-willjust equalF(Pj,Zj,C)/n = nf(8j)ln = f(cj), which is exactly equal to gj, the income associated with the El equilibrium in the single-worker case. The coalition thus allows each individual worker to achieve that utility level realizable in the single-worker case and which is greater than that attainable in the n-worker case in the absence of the coalition. Furthermore, in more realistic circumstances when interdependence among effort levels in production exists, coalition formation can make individual workers in a sharing agreement better off than they would be in the single-worker case. For example, an effort level Pj by each individual worker is likely to produce a coalition income F(Oj,$j,C) greater than nf(gj) and a per-worker share Sj(ci,Pj,C) greater than jtj. From the perspective of the individual worker such a coalition can support a utility level, such as the one pictured at E3 in Fig. 1, that is in excess of the utility level at E, associated with the single-worker case. Overall, the results derived here support the general proposition that enterprise organizational structures, such as labor management or full participation, that allow communication and coalition formation among workers will, ceteris paribus, enhance the incentive effects of incomesharing arrangements among those workers and will allow each individual worker to achieve a utility-maximizing position superior to that attainable in the absence of coalition decisions. 4. EX ANTE AND EX POST PROBLEMS IN THE FORMATION ENFORCEMENT OF WORKER COALITIONS
AND
Although coalition decisions can make each individual worker better off, such decisions may prove difficult to negotiate or enforce in practice. At the negotiation stage, problems may arise because workers fail to reveal their true abilities and preferences. The theoretical literature on the labor-managed firm as a production coalition does not confront these problems but instead simply assumes the negotiation and existence of a reliable ex ante contract specifying the responsibilities of firm participants and the payoff rules. It is clear, however, that a potential coalition member might find it advantageous to misrepresent information at the contract-negotiation stage. As a typical example, one can imagine a “bad” or “lazy” worker who negotiates his way into a profitable coali-
INCENTIVES
IN THE SELF-MANAGED
FIRM
293
tion by pretending to be a “good” worker. Recent contributions in the literature on screening-signalling and optimal taxation (see, for example, Stiglitz, 1975, and Groves-Ledyard, 1977) indicate that such ex ante problems in contract negotiation stemming from the uneven distribution of information among negotiating agents are theoretically solvable. On a more practical level Williamson (1975) has argued that ex ante “adverse selection” problems in contract negotiation tend to encourage the development of an elaborate internal organization to serve as a screening device. In a labor-managed firm, for example, ex ante difficulties in coalition formation might be alleviated by dividing the larger group of workers into smaller subgroups each of which earns its own income and each of which negotiates its own coalition decision among members. A firm-wide coalition agreement could then be worked out to specify the responsibilities and payoff rules for these subgroups within the larger organization. Decentralization along these lines might enhance the coalition-formation process since smaller groups of workers tend to be better able to communicate with one another because of the common or related production tasks and skills that they share, and because they are better equipped to know the requisite attributes to look for in new members. In addition, the information and communication costs of negotiating a coalition decision through a process of consensus formation are smaller, the smaller the number of agents involved in the process.‘O The development of an internal labor-market structure within a labormanaged production coalition is another organizational device that might mitigate the problems confronting truthful ex ante coalition formation. Suppose, for example, that an existing labor-managed coalition is considering a new member but cannot be certain that he is truthfully representing his intentions and/or abilities. If a well-defined job ladder relating payoff rules to jobs rather than to individuals exists, then the prospective coalition member can be admitted to the firm at some point of entry on the job ladder and can move up as he reveals his true productivity and attitudinal and reliability characteristics. l1 Indeed, an internal lo Williamson (1975) notes that the feasible size of a “peer group” or a group of agents who negotiate consensus or coalition decisions is inherently limited by information-processing constraints. In his words, in such a group “everything cannot be communicated to everyone and joint decisions reached without pre-empting valuable time that could be productively used for other purposes.” Since the number of information-flow linkages in a peer group goes up as the square of the number of members (the all-channel network phenomenon), the information and communication costs of peer-group decision making increase dramatically as the number of members included in the group increases. I1 The Yugoslav experience with job-evaluation procedures indicates how an internal labor market of this variety might be structured in a labor-managed firm. Within each labormanaged enterprise there is a formal job evaluation procedure. Every job gets assigned a number of points (shares) according to such criteria as the level of skill required, the level of education required, the level ofjob experience required, and working conditions on the job.
294
LAURA
D’ANDREA
TYSON
labor market may be especially important in a labor-managed firm because new participants admitted into the firm become more than just employees: they become coalition members with associated decision-making rights. Consequently, existing members will want an opportunity to observe new members at entry-level positions to determine where in the job, payoff, and decision-making ladders they belong.12 Suppose for the sake of argument that ex ante problems in coalition formation are solved by some of the organizational techniques suggested here. Even then, an ex post production coalition may produce an outcome other than that prescribed in the ex ante coalition decision. Such a divergence between ex ante and ex post performance can result from the Nashselfish motives of individual coalition members. Even when truthful ex ante negotiation supports a coalition decision that promises to make each participant better off, ex post Nash-selfish behavior on the part of some of these participants can destroy the coalition and make everyone worse off. The ex post problem is similar to the “moral hazard” problem that arises because of difficulties in the enforceability of contracts. The point to be emphasized here is that any viable coalition in the labormanaged firm must embody some internal mechanism to audit and curb ex post Nash-selfishness. The implications of Nash-selfish behavior for coalition viability can be formalized using the model of the labor-managed firm introduced in the previous section. Let e* be an ex ante coalition decision about the effort to be supplied by each coalition member. The jth member’s Nashselfishness is then expressed by the fact that there may exist an ej such that 7rj[ej,2S,C]
> 7rj[eT,i?g,C].
(4)
Now, let ej denote the solution to thejth member’s Nash maximization problem maxq Tj[ej,gf ,C] conditional on 2;. By assumption, ej(gj*) < ej*, or in words, the optimal effort supplied by thejth member’s Nash-selfish behavior is less than the effort promised by him in the coalition decision. An ex post Nash-selfish equilibrium within the coalition is given by E* = (~3) that solves the Nash maximization problem for all agents. The notion that such an intracoalitional Nash solution is inferior to the ex ante coalitional solution is formally expressed by the following inequality 7Tj[eT,Zr,C]
> ?Tj[ET,Z$,C],
In the case of extreme shirking and attempted
eT >E:.
(5)
free riding by coalition
A worker who is assigned a job is then automatically assigned the number of points or shares associated with that job. Under this system, incentives for worker effort and cooperation take the form of promotion possibilities up the job ladder. I2 Miyazaki and Neary (1979) discuss the internal labor market as a theoretical “dual” to the labor-managed coalition.
INCENTIVES
IN THE SELF-MANAGED
FIRM
295
members, E* will be a vector of very small effort levels and the realized payoff for each individual worker will be very small. A comparison of (4) and (5) indicates that an individual member’s anticipated gain from ex post Nash behavior is significantly reduced when other members fail to adhere to the ex ante coalitional rule e*. In other words, each member has a strong incentive to make sure that other members do their share in the coalition, so that he can gain from cheating. In addition, if thejth member observes that a nonnegligible set of coalition members are deviating from their ex ante promise, then in the absence of any formal sanction mechanism, thejth member himself will find it in his selfish gain to deviate from e* as well. In the extreme case when everyone adopts an ex post Nash behavior, the result is an outcome in which everyone is strictly worse off. If an individual’s horizon were extended slightly beyond his immediate myopic gain, then his understanding of the mutual dependence of selfish incentives for all members would deter him from behaving selfishly. The foregoing discussion suggests that even in the absence of formal sanctions against an individual’s ex post selfish behavior, the possibility of everyone’s ex post selfishness in toto implicitly constitutes a mutual sanction to an individual’s selfish act. In the event that everyone is constrained by this implicit mutual threat, there will materialize no ex post selfish motives.13 This point can be further emphasized using the framework of a supergame. Imagine a situation in which the members of a labor-managed firm interact with each other repeatedly over some period of time. In such an environment an individual worker is likely to perceive that selfish behavior vis-a-vis his fellow workers will invite retaliation in subsequent interactions. This follows because member behavior is tacitly correlated, and one’s own “bad” behavior invites “bad” responses. A technical prerequisite for such a correlated behavior strategy is that the environment in which interactions occur be repetitive, ideally an infinite number of times. Each round of interactions can be regarded as a game, and a series of repeated rounds as a whole constitutes a “supergame” (see Aumann, 1959; Kurz, 1977). In the present context, a “correlated” strategy for the jth member would be to act according to the coalitional norm e* in the first round, and to continue this norm until he discovers that some agents are deviating from their norms to their own selfish gain. Subsequent to the discovery of deviations, thejth member himself adopts his own selfish behavior. When all members individually adopt such a two-tier strategy, the strategies as a whole are correlated throughout the coalition and constitute an equilibrium in the intracoalitional supergame. A member, I3 For the sake of argument, it is assumed that effective formal sanctions to compel ex post compliance with an ex ante contract do not exist. Within a self-managed firm such sanctions might include demotion, firing, and/or reduction in profit share.
296
LAURA
D’ANDREA
TYSON
who is supposed to act selfishly upon discovery of someone else’s selfish deviation, will himself be sanctioned. No one will benefit by defecting from this strategy as long as he remains in the coalition. In short, in a situation approximated by a supergame a member will be punished whenever he deviates from the (gentlemen’s) silent agreement on the correlated strategy. The supergame framework suggests that the effectiveness of implicit sanctions against selfish behavior in worker coalitions depends on two conditions-the repetitive nature of coalitional association and the ability of coalition members to monitor one another’s performance. The first condition is likely to be satisfied in practice because both hiring and firing rigidities are likely to restrict labor turnover and mobility among self-managed firms. For example, Meade (1972) has argued that two conditions should be satisfied before a worker leaves one coalition and joins another. First, the worker must want to leave the coalition; and second, the coalition must want the worker to leave or must agree on a set of arrangements under which the remaining workers will be willing to allow the worker to leave. Furthermore, the worker’s desire or willingness to leave is likely to depend on the willingness of some other coalition to admit him. Because of the difficulties inherent in satisfying all of these conditions, only the first of which is relevant to labor mobility under usual circumstances, workers are likely to remain with labor-managed coalitions for long periods of time, and coalitional decisions are likely to have the long-term repetitive quality conducive to the development of an effective implicit sanction mechanism.14 The second condition for such a mechanism-namely, that workers in a coalition be able to monitor one another’s behavior-is likely to depend on the size of the coalition. Ceteris paribus, the smaller the number of workers in the coalition, the easier it will be for them to observe and evaluate one another’s performance. In small groups, workers who perform similar or integrated tasks are able to evaluate automatically the characteristics of one another’s performance. Information about the characteristics of co-worker performance in such groups is frequently simply a joint output of productive activity. These observations suggest that the prospects for effective monitoring and hence the development of an effective implicit sanction mechanism are likely to be enhanced by the subdivision of the coalition into smaller groups of workers who enjoy informational economies in the task of supervising one another’s perI4 As is widely recognized, collective-ownership rights such as those that prevail in Yugoslavia are also likely to discourage labor mobility among self-managed firms. (See, for example, Pejovich, 1973.) Seen in the context of the present analysis, reduced labor mobility, which is frequently thought to have negative effects on economic performance, is seen to have some positive effects by strengthening implicit sanctions and thereby enhancing the incentive effects of income sharing.
INCENTIVES
IN
THE
SELF-MANAGED
291
FIRM
formance. In the framework of the supergame discussed here, the divisionalization of the coalition is important solely because it provides a monitoring mechanism for preventing coalition-breaking ex post Nash selfishness. In a more complete model, however, where differences in worker abilities and in the characteristics of jobs are introduced, divisionalization as a monitoring device can play a critical role in the provision of information needed for the establishment of an internal payoff schedule. In summary, the arguments presented here indicate that worker communication and coalition decision making can enhance the incentive properties of an income-sharing scheme and can make all workers better off than they would be in the absence of such communication. Furthermore, both the ex ante feasibility and ex post viability of coalition decisions are likely to depend on the number of workers formulating or executing these decisions. In general, the smaller the firm, the greater the probability that workers will be able to reach and enforce a collective agreement. Furthermore, for a given firm size, coalitions will be encouraged by an organizational structure that divides workers into smaller subgroups. 5. INCENTIVES
AND INSTITUTIONAL INNOVATION YUGOSLAV SELF-MANAGED FIRM
IN THE
Recent experiments with new organizational forms of the labor-managed firm in Yugoslavia can be understood as attempts to provide intrafirm institutions that will strengthen the incentive effects of income sharing. According to the 1974 Constitution and the more detailed 1976 Law on Associated Labor, if the performance of a subunit of workers in an enterprise “can be expressed in terms of value within the work organization or on the market [that is, if the subunit has an imputable income] . . . the workers shall have the right and duty to organize . . . [that unit] as a basic organization of associated 1aboP” (or BOAL, as the separate divisions have come to be called in the Western literature). Legally, BOALs hold all final authority in enterprise decision making. This authority is exercised by the workers’ council elected by each BOAL to function in a managerial and decision-making capacity. Relationships among BOALs are governed in two ways: first, by an inter-BOAL agreement, called a self-management agreement, which sets out the basic rights and responsibilities of individual BOALs and spells out general principles to guarantee enterprise consistency in certain inter-BOAL decisions;16 and second, by the coordinating activities of the enterprise workers’ council, I5 Constitution of the SFR Yugoslavia, Art. 36, 1974. Repeated verbatim in Basic Law on Associated Labor, Art. 14, 1976. I6 Such general principles frequently include enterprise-wide criteria for the sharing of income among different categories of jobs and labor skills and criteria for labor relations, including hiring and firing policies.
298
LAURA
D’ANDREA
TYSON
which is composed of delegates elected by each BOAL17 and of the enterprise director and his technical supportive staff. The director, who is the chief coordinating agent of the enterprise, is elected for a four-year term by the enterprise workers’ council from a slate of candidates meeting the conditions set forth in the self-management agreement.18 In the spirit of the new legislation, the enterprise workers’ council and the director along with his technical staff are to serve as the general coordinating agents for the multidivisional enterprise, exchanging information among the BOALs and between the BOALs and “third parties” outside the firm, and monitoring adherence to the self-management agreement by individual BOALs. In addition, the legislation allows BOALs to delegate their decision-making authority to either the enterprise workers’ council or the director. In certain decisions, such as those involving the distribution of divisional income between payments to workers and enterprise savings, however, delegated decisions by the coordinating agents must be ratified by the divisions. Although the actual extent of the delegation of authority to central organizations is left open for specification in the enterprise self-management agreement, it is clear that the director and technical staff are likely to be accorded significant decision-making rights in many areas because of their access to superior information. On the other hand, the legislation clearly reserves authority in decisions governing the distribution of BOAL income and internal working conditions to the BOALs themselves. Within the framework of the models discussed in this paper, decision-making rights in these areas are analogous to rights of subgroups of workers to form their own coalition decisions about the effort-payoff positions for each of their members. According to the analysis presented in this paper, the division of the Yugoslav firm into subgroups of workers with imputable incomes is likely to enhance the incentive effects of income sharing in a variety of ways. First, even if all members of the firm negotiate some ex ante coalition decision, the formation of subgroups permits more effective monitoring of ex post selfish behavior. In addition, the possible development of trust or “associational benefits” (Williamson, 1975) among smaller groups of workers with a common interest in maximizing group income should itself act as a hindrance to such behavior. Second, in most real-world situations characterized by uncertainty about the actual relationship I7 Delegates are elected to the workers’ councils for a period of 2 years and no person can be elected for more than two consecutive terms. I8 The slate of candidates for the director is made up after a public advertisement of the position has been made. As Conner and Vukmir (1976, p. 116) argue, the law does not prescribe any qualifications other than a record free of such offenses as “crimes against the people and the state.” Nevertheless, it is generally agreed, as a practical matter, that the managing director be a Yugoslav citizen and probably (although not necessarily) a member of the League of Communists.
INCENTIVES
IN THE SELF-MANAGED
FIRM
299
between effort and income, it may be virtually impossible for workers to negotiate an ex ante firm-wide coalition decision simply because they do not know enough about one another’s jobs, abilities, and preferences. Lack of knowledge in these areas will preclude the kind of detailed communication required for the formulation of an ex ante pact. Under these circumstances, only subgroup coalitions may be feasible, and the divisionalization of the firm into groups of workers whose interests and jobs are directly interdependent may foster the development of such ex ante coalitions. Third, even when smaller groups of workers cannot agree on or enforce a coalition decision among themselves, they still provide a more effective vehicle for individual incentives under income sharing, because each worker shares the income attributable to his effort only with members of the subgroup rather than with all enterprise members. Finally, the creation of small groups may enhance incentives by providing a more attractive work environment. The discussion throughout this paper has rested on the implicit assumption that each worker’s utility function is unaffected by the structure of the organization in which he works. Suppose, instead, as Williamson (1975) has suggested, that the work environment is itself an item of value that either is an argument in the worker’s utility function or affects his perceived tradeoff between income and effort. If workers prefer small work groups and the interpersonal relationships that they tend to foster, then the divisionalization of the firm may encourage greater effort on the part of workers simply through the creation of a more desirable work environment. The general point to be stressed here is that regardless of the incentive scheme in operation, organizational and environmental circumstances directly affect the amount of effort supplied by individual workers, and any organizational change that improves these circumstances from the workers’ perspective is likely to increase worker effort.lg 6. CONCLUSIONS An analysis of collective decision making suggests that the incentive properties of income sharing can be enhanced by forms of enterprise organization that are conducive to worker communication and the development of a “team” spirit among workers. One dimension of enterprise organization, namely, worker participation in decision making, is of special interest in this paper. The analysis suggests that worker participation encourages both the ex ante negotiation of an implicit or explicit coalition decision among workers and serves as an effective sanction mechanism for its ex post enforcement. Furthermore, the arguments inI9 This point has been forcefully argued by Leibenstein X-efficiency and enterprise organization.
(1974) in a recent discussion of
300
LAURA
D’ANDREA
TYSON
dicate that the divisionalization of the labor-managed firm into smaller groups of workers can facilitate both the ex ante formation and ex post enforcement of coalition decisions, thereby enhancing the incentive effects of income sharing. This observation provides a rationale for recent reforms in Yugoslavia calling for the compulsory divisionalization of existing enterprises into BOALs, each of which has decision-making rights over the distribution of enterprise income imputable to it. Of course, a full evaluation of the economic impact of these reforms requires consideration of their probable effects not only on worker incentives but also on the costs of interdivisional coordination and communication. In addition, the new organizational reforms actually introduce an incentive problem of their own-namely, how to motivate the individual BOALs to operate as if they were a team, or, in the terminology of this paper, how to get the BOALs to negotiate and enforce an interdivisional coalition decision on the objectives of the entire enterprise. Moreover, the reforms leave another perplexing incentive problem unresolved-namely, how to motivate the enterprise director and his technical staff to exercise the decision-making rights delegated to them in the best interests of the workers whom they ostensibly represent. The implications of both of these incentive problems must be determined before a complete assessment of the Yugoslav reforms can be made. Although this paper has focused on the labor-managed firm, the arguments suggest the general conclusion that income sharing is likely to be a more effective incentive mechanism when combined with organizational devices that encourage the formation of a team perspective among workers. One might expect, for example, that profit sharing will be more effective when combined with team-production models, like those in the Swedish automobile industry, than when introduced into a traditional hierarchical firm where worker communication is difficult. Similarly, one might expect profit sharing to be particularly effective in the Japanese enterprise where workers perceive themselves to be part of a company team and where the team outlook and the relative stability of team membership foster the kind of supergame atmosphere conducive to the ex ante negotiation and ex post enforcement of coalition decisions. On a general level, these examples and the analysis presented in this paper serve to reaffirm the general principle that aspects of enterprise organization have a powerful impact on the effectiveness of worker incentive schemes. REFERENCES Aumann, Robert J., “Acceptable Points in General Cooperative n-Person Games.” In A. W. Tucker and R. D. Lute, eds., Confriburions to the Theory of Games, Vol. IV, Annals of Mathematical Studies, No. 40, pp. 287-324. Princeton, N. J.: Princeton Univ. Press, 1959.
INCENTIVES
IN THE SELF-MANAGED
FIRM
301
Conner, John, and Vukmir, Branko, “The Legal Anatomy of a Yugoslav Enterprise.” The Business Lawyer 32, 4:99- 117, Nov. 1976. Greenberg, J., “Pure and Local Public Goods: A Game-Theoretic Approach.” Research Report No. 65, Department of Economics, Hebrew University of Jerusalem, 1975. Groves, Theodore, and Ledyard, John, “Optimal Allocation of Public Goods: A Solution to the “Free Rider” Problem.” Econometrica 45, 4:783-809, May 1977. Ichiishi, Tatsuro, “Coalition Structure in a Labor-Managed Economy.” Econometrica 45, 2:341-360, Mar. 1977. Kurz, Mordecai, “Altruistic Equilibrium.” In B. Balassa and R. Nelson, eds., Economic Progress,
Private
Values
and
Public
Policy:
Essays
in Honor
of William
Fellner,
pp. 177-200. Amsterdam/New York: Elsevier, 1977. Leibenstein, Harvey, “Aspects of the X-efficiency Theory of the Firm.” Bell. /. Econ. 5, 3: 125-137, Autumn 1974. Meade, James, “The Theory of Labour-Managed Firms and of Profit-Sharing.” Econ. J. 82, 1:402-428, March 1972. Miyazaki, Hajime, and Neary, Hugh M., “The Illyrian Firm Revisited.” Draft Mimeo., Department of Economics, Stanford University, 1979. Pejovich, Svetozar, “The Banking System and the Investment Behavior of the Yugoslav Firm.” In Morris Bornstein, ed., Plan and Market. New Haven, Conn.: Yale Univ. Press, 1973. Steinherr, Alfred, “On the Efficiency of Profit Sharing and Labor Participation in Management.” Bell J. Econ. 8, 3:545-555, Autumn 1977. Stiglitz, Joseph, “Incentives and Risk Sharing in Sharecropping.” Rev. Econ. Stud. 41, 2:219-256, April 1974. Stiglitz, Joseph, “Incentives, Risk and Information-Notes Towards a Theory of Hierarchy.” Bell J. Econ. 6, 3~552-579, Autumn 1975. Vanek, Jaroslav, The General Theory of Labor-Managed Market Economies. Ithaca, N. Y.: Cornell Univ. Press, 1971. Williamson, Oliver E., Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press, 1975.
Comments MICHAEL Department
of Economics,
University
of New
A. CONTE Hampshire,
Durham,
New
Hampshire
03824
Laura Tyson’s paper focuses our attention on several issues of importance in the area of the labor-managed firm, but primarily on the relevance of profit sharing as a mechanism for distribution. Tyson assumes that the labor-managed firm actually distributes “profits” (defined as revenues less operating costs-labor is not one of these costs) equally among workers in the enterprise. While the analysis is interesting and provocative, one must first check its relevance to the Yugoslav economy. Wachtel has informed us, in his 0147-5967/79/030301-03$02.00/O Copyright 0 1979 by Academic Ress, Inc. All rights of reproduction in any form reserved.