TY - JOUR
AU - Keil,, Marten
AB - Abstract In this article we examine the scope a principal in a public organisation has for motivating agents for productivity improvements where standard stick and carrot incentives cannot be used. The principal's only incentive device is a reallocation of budgets and tasks across agents depending on the extent of productivity improvements revealed by each agent. We first show that as long as agents do not collude, the principal can use rotation and tournament schemes to eliminate all slack in the organisation. Second, to break collusion between agents, the principal must use discriminatory tournament schemes. In some cases, however, there is no incentive scheme that can overcome collusion. In many organisations the principal typically depends on the help of agents to improve the productivity of an organisation. For instance, agents have accumulated specific knowledge and can reveal unproductive tasks to the principal. However, such activities may reduce the budget and the sphere of control of the agents, and hence may in turn also reduce their utility. In many public organisations, standard stick (e.g. firing) or carrot (e.g. wage increase) incentives cannot be used to motivate agents to reveal unproductive tasks, because of guaranteed life‐time employment and a rigid and predetermined wage structure. Moreover, implicit incentives due to career concern may be limited or even non‐existent, as can be the case in state‐owned universities. In addition, the output of agents may not be verifiable or may only be verifiable with considerable delay, thus limiting the use of incentive devices based on output, as is the case in research and health care. As an example we take the restructuring of German universities in the late 1990s. Departments were forced to cut costs by forfeiting a fixed portion of their budget, irrespective of their past performance, because no performance measures were available. After the fixed budget cuts, it remained unclear whether less productive activities had indeed been eliminated.1 In this article, we examine the scope a principal in a public organisation has for motivating agents for productivity improvements when all the standard stick and carrot incentives cannot be used. Therefore, our central assumption is that the principal's only incentive device is a reallocation of existing budgets and tasks across agents. Our major conclusion is that even if agents collude, the principal can in many cases eliminate all unproductive tasks in the organisation by appropriate mechanisms, such as tournaments or job rotation. We consider a model with a principal and a number of agents characterised by the following information asymmetry. The agents become informed about the extent of unproductive tasks in the organisation. The principal, however, does not observe the productivity of the tasks. We assume that the principal can verify and eliminate a set of unproductive tasks if this set is revealed by an agent and communicated to the principal in the form of hard information. Since the agents derive utility from the size of their unit, and therefore from the unproductive tasks in their unit, the principal must provide the agents with incentives to reveal their knowledge. We consider two scenarios distinguished with respect to the knowledge agents can supply to the principal in the form of hard information. In the first case, agents have knowledge specific to the organisation (organisation specific knowledge), i.e. they are able to reveal the slack in all the units of the organisation to the principal. In the second case, which appears more plausible, agents have knowledge specific to their unit (unit specific knowledge) and are only able to provide hard information on the unproductive tasks in the unit they govern. For both cases, we show that a first‐best solution in which the principal eliminates all unproductive tasks in the organisation can be achieved by appropriate mechanisms involving the revelation of unproductive tasks by agents. If collusion is possible, a first‐best solution can only be implemented if knowledge is organisation specific. In particular, we establish the following results. First, if agents act non‐cooperatively, a tournament guarantees the first‐best solution in the organisation specific knowledge scenario. The prize of the tournament is the award of extra tasks. In the case of unit specific knowledge however, a tournament does not work, since the agent with the largest productivity improvement potential only has to match the agent with the second highest share of unproductive tasks. The principal, however, can realise a first‐best solution through partial job rotation. Under such a scheme, the principal assigns each agent a part of his traditional tasks and a part of the other agent's traditional tasks. The assigned shares depend on the announcements of unproductive tasks. Second, if agents can collude but cannot transfer utility among themselves via side payments, only tournaments in which the principal does not treat agents equally will yield a first‐best solution in the organisation specific knowledge scenario. Such discriminatory tournament schemes reward agents differently, even if agents reveal the same share of unproductive tasks. If agents are treated unequally, it is impossible for less well‐treated agents to offer collusion contracts without side payments. Otherwise, collusion would be feasible and productivity improvements would not occur. The necessity of treating equal agents unequally represents a fundamental trade‐off between fairness and efficiency. In the case of unit specific knowledge, there is no incentive scheme which implements a first‐best solution. Our analysis suggests that in practice task and budget assignments may be used as an incentive device motivating unit managers to improve productivity in public organisations. In the past, uniform cost‐cutting across all departments in public organisations such as in German universities, has probably eliminated both unproductive tasks and productive tasks. In recent years, universities have tried to reward departments by using a tournament‐like scheme with new budgets and possibilities of expanding into other areas in exchange for the revelation of less productive tasks. The paper is organised as follows. Section 1 discusses related literature. We introduce the model in Section 2. In Section 3, we specify the contracts and the game for the two scenarios, differing with respect to the extent of hard information agents can supply to the principal. In Section 4, we derive the optimal incentive schemes if agents act non‐cooperatively. In Section 5, we consider the game where collusion is possible. Our conclusions are presented in Section 6. 1. Relation to the Literature Many contributions in recent organisation and productivity literature have emphasised the role of the organisation of functions and tasks as a key factor in explaining productivity differences between organisations (McKinsey, 1993; Milgrom and Roberts, 1992; Womack et al., 1990). Benchmark studies reveal that more than half of the productivity differences between German and Japanese car manufacturers are attributable to differences in the way functions and tasks are organised (Baily and Gersbach, 1995). The organisation of functions and tasks includes the way companies organise internal communication and design incentive schemes to make better use of the special local knowledge that their workers alone possess (e.g. Milgrom and Roberts, 1992; Baily and Gersbach, 1995; Imai, 1989). The underlying incentive problem was first addressed by Carmichael and MacLeod (1993). They suggest that multiskilling of workers will enable a firm to reap the gains from enlisting the full cooperation of workers for labour‐saving techniques. Our paper is complementary to this line of work. We discuss how improvements in the organisation of functions and tasks in public organisations can be achieved by appropriate incentive schemes, such as tournament and rotation schemes. Whereas a broad range of insights has been derived for private organisations using explicit and implicit incentive schemes with agents acting non‐cooperatively2 or colluding,3 only a few contributions deal with adequate incentive schemes for public organisations. Private and public organisations differ with respect to the set of possible incentive instruments (Wilson, 1989). Dewatripont et al. (1999a, b) provide insight into the role of implicit incentives in the form of career concern within a generalised version of the career concern model developed by Holmström (1982b), which can also be applied to public organisations. In our paper we focus on task and budget assignment as a tool for motivating agents in public organisations, where career concerns cannot provide sufficient incentives. Examples are the education and health services. We provide insight into the role of job rotation and the use of tournaments in motivating agents to reveal unproductive tasks and to improve productivity in the organisation. Our analysis may also have some bearing on private organisations when authority to allocate budgets and to determine compensation differs. For instance, a manager in an R&D division can allocate budgets, but is not authorised to change his worker compensation, and thus may act under similar constraints as in public organisations. 2. Model We consider an organisation with n units n > 1, indexed by i, j or k. Each unit is managed by an agent and performs a variety of tasks. The agent is indexed by the unit he is governing. Therefore, we call agent i the unit manager i. To simplify the analysis, we assume that there are only two types of tasks. Tasks can be either productive or unproductive. The return of a task is either rg or rb with rg > 0 and rb < 0. The share of unproductive tasks is denoted by Ai and we assume that half of the tasks of a unit may be unproductive at the most, i.e. . Depending on the share of unproductive tasks Ai, the average productivity or return of a unit is: Unproductive tasks can be beneficial for unit managers however, because of a desire for power, control, empire building or because of private benefits from tasks that may include perquisites for the job, the acquisition of human capital and the possibility of signalling ability (Aghion and Tirole, 1997).4 Thus, unproductive tasks can occur when individuals want to expand as much as possible or when agents focus on generating private benefits from tasks which may make such tasks unproductive for the principal. Therefore, there are a variety of reasons why unproductive activities are beneficial for managers. Naturally, we think that unit managers also have an influence on the productivity of their units. Our model can capture this in the following way. Suppose that managers have an unobservable talent or ability to impact on productivity in their unit. Very talented managers will make most activities productive, i.e., Ai will be low. Less talented managers end up with a large share of unproductive tasks, and thus have a high value of Ai. Such an interpretation naturally implies that the principal neither observes the ability of unit managers, nor the share of unproductive tasks Ai. This will be precisely the central assumption explored in the next few paragraphs. The principal is responsible for the whole organisation. We assume that the principal cannot use the average productivity in each unit to set up incentive schemes. Either he does not directly observe ri because output is not measurable, or because it is only measurable with long delays, as is the case in the education or the health sector. To derive the potential for task reallocation, our central assumption is that standard stick and carrot incentive schemes cannot be used at all. Therefore, the principal's instruments are restricted to budget or task assignments. This assumption, which rules out any monetary incentives, is crucial for our analysis and deserves further comment. First, there are real‐world organisations where monetary incentives play little or no role. In many employee categories in the public sector in Germany, individuals cannot increase their wages through excellent performance. Examples of this are all those civil servants who have reached their highest possible career position. In many cases, since the wage profiles are set by law, civil servants reach the highest possible position quite early in their careers. Therefore, monetary incentives play a limited role for these employees. Predominantly, wages depend on job descriptions, the university degrees necessary to get the job and seniority. Detailed examples can be found in the pay regulations for the public sector in Germany.5 Anecdotal evidence suggests that similarly rigid wage structures also operate in the public sector in France. However, even in the rigid wage structures of public organisations, there are career opportunities associated with an increase of wages. But, as mentioned above, wage increases for promotion are rather small in public organisations and seniority determines wage rises. In some cases, wages do not respond to promotion at all. The latter is the case in Germany, when a professor becomes the head of a faculty or even of the university.6 Therefore, we believe that monetary incentives in public organisations, e.g. in Germany or France, play only a limited role for most of the employees. Second, the absence of monetary incentives can be justified by infinite risk aversion in wealth or by non‐contractability of the principal's benefit as discussed extensively in Aghion and Tirole (1997) and Tirole (2001). Concern is appropriate that the absence of monetary incentives in a public organisation may be due to unmodelled behaviour that also acts as a constraint on the schemes presented in this article. Such unmodelled behaviour might be connected with multi‐task issues and the influence of unions. In the first case, we know from Holmström and Milgrom (1991) that measurement problems severely hamper the use of monetary incentives. In the second case, unions may be prompted to restrict incentive schemes in order to compress the wage distribution of its members (Fitzenberger and Franz, 1999). We cannot exclude the possibility that such constraints affect the ability of the principal to implement the schemes presented in this paper. In recent years universities in Germany have tried to use tournament‐like schemes to allocate tasks and budgets in accordance with the revelation of less productive tasks.7 From the perspective of industrial organisation literature, we could interpret such moves as a response to increasing global competition, in the sense of exposure to best practice in areas where public ownership plays, and continues to play, a large role in continental Europe (Baily and Gersbach, 1995). Examples are universities, health care and vocational training schools. To sum up, monetary incentives play a very limited role in many public sector organisations. Moreover, we suggest that lack of competition has permitted such organisations to refrain from using incentive schemes such as those indicated in this article. Therefore, our analysis has both a positive and a normative flavour. A further assumption of our analysis is that tasks can be reallocated with no productive loss. There are various potential justifications for this assumption. First, the role of managers initially is to organise a unit and to identify unproductive tasks. After units have been organised and unproductive tasks have been identified, tasks can be reallocated across unit managers without costs. Second, the assumption is made for simplicity of exposition and serves as a benchmark case. In principle, we think that managers can add value continuously to their unit, which will be lost when tasks are reallocated. In such a setting, the application of the schemes in our article requires that such losses due to reallocation are not too large. Third, we could imagine that each manager leads a unit consisting of standard and special operations, e.g. teaching and research activities at universities. The budgets for the special operations are distributed equally among managers initially. Our schemes are then applied to the special operations whose budgets can be freely redistributed. However, each unit always requires a person to manage the standard operations but some tasks can be reallocated with no productive loss. The principal faces the following informational asymmetry. The unit manager i observes the productivity of the tasks performed in his unit. Therefore, he could in principle eliminate unproductive tasks and raise productivity in the organisation. The principal, however, does not observe the productivity of the tasks. He has a priori knowledge about the extent of unproductive tasks in unit i. From the perspective of the principal, Ai is assumed to be uniformly distributed over the interval . Hence, unit i may have a small or a large share of unproductive tasks. We further assume that Ai and Aj are uncorrelated across units for any i, j, i ≠ j. An important assumption is that the principal can verify and eliminate a set of unproductive tasks if this set of tasks and its productivity is revealed by the unit manager. Hence, we assume that the unit manager can provide the principal with hard information about the productivity of tasks.8 For instance, he may reveal the inefficiencies involved in certain stages of activities in his unit directly to the principal. Our assumption is in line with the practices in management consulting which suggest that reductions in slack in business units require information from the employees of the unit (Baily and Gersbach, 1995). We use αi to denote the share of tasks revealed as being unproductive by the unit manager of unit i. The principal can verify whether these tasks are indeed unproductive. Hence, announcements of a portion of unproductive tasks larger than Ai could be verified and only unproductive tasks would be eliminated. Thus, it is never worthwhile for the unit manager to announce productive tasks. The principal cannot, however, observe whether the unit manager has disclosed all the unproductive tasks in his unit, because he does not observe the productivity of the residual tasks. The rationale for this assumption stems from the productivity and organisation literature on continuous improvements (Womack et al., 1990; Baily and Gersbach, 1995). Only workers and managers directly involved in the execution of activities acquire knowledge enabling them to increase productivity. The principal's utility, denoted by UP, is monotonically increasing in the net returns. The principal is assumed to be risk neutral. Thus, UP is given by: The manager of unit i derives utility from two sources: the income and the size of the organisational unit. Since flexible wages play a limited role in public organisations, we assume that wages are fixed exogenously by a predetermined and rigid wage structure. We normalise the utility from the fixed wage to zero. We assume that the utility unit managers derive from the size of their unit is positive, regardless of whether the tasks are productive for the principal. Examples are the desire for power, control or empire building often mentioned in the organisation literature or the possibility of the unit manager using some of the resources in his unit for activities that are not in the interest of the organisation. The unit manager's utility before contracting depends solely on the size of his unit and is normalised to 1, i.e. Ui = 1. Since unit managers derive utility from the size of their unit, and hence also from tasks that are unproductive, the principal must provide his unit managers with incentives to reveal their knowledge. 3. The Game In the following, we consider two cases distinguished by the unit manager's knowledge about unproductive tasks. Knowledge refers to the extent to which a unit manager can provide hard information on unproductive tasks. In the first case, the unit managers are supposed to have organisation specific knowledge. In this case, unit managers are able to reveal the share of unproductive tasks of each unit in the organisation to the principal and therefore the overall slack in the whole organisation. This assumption can be motivated by the similarity of the units or by cross‐sectional training of the unit managers at the beginning of their careers. In the second case, we assume that the unit managers have only unit specific knowledge, i.e. they know only the share of unproductive tasks in their own unit and thus they are only able to provide hard information on unproductive tasks in that unit. Which case is more plausible is not a priori clear. In the example of German universities one could argue that unit specific knowledge is more plausible. However, organisation specific knowledge cannot be excluded because of communication and observations among agents prior to restructuring. 3.1. Organisation Specific Knowledge If the unit managers have organisation specific knowledge, they are able to report the unproductive tasks in each unit. denotes the share of unproductive tasks in unit j revealed by the manager of unit i. Thus, the unit manager i announces a vector containing the share of unproductive tasks in each unit he wants to reveal to the principal. The unproductive tasks will be eliminated by the principal. The principal offers a contract with the following interpretation. Depending on the revealed knowledge, unit manager i receives a share βii of the residual tasks in his unit, i.e. tasks that are not eliminated by the principal, and a share βij of the residual tasks in unit j (j ≠ i). Two possible interpretations of βij fit our model. First, βij is the amount of tasks (or the budget) not deleted in unit j and reallocated to unit i. For this interpretation, the residual activities of unit j needs to be broken up, since we do not consider joint control over units. In a second interpretation, the shares βij could be probabilities of obtaining control over the undeleted activities in unit j. As unit managers’ utility is assumed to be linear in the size of their units and ∑iβij = 1, such an interpretation is equivalent. We think that the second interpretation is somewhat less plausible from an empirical point of view, at least when considering restructuring efforts in public organisations in Germany. Complete reallocations of units occur but in most cases control over units remains in place, with some redistribution of activities or budgets.9 We assume that the announcements are nested, i.e., the maximum share of unproductive tasks in unit i the principal can eliminate is given by: The assumption can be justified technologically by assuming that there is a natural order in revealing unproductive tasks to the principal. Or the principal may only need to use in first‐best incentive schemes. As we will see, the latter is true and, hence, we proceed directly on the nested assumption that the maximal share of unproductive tasks the principal eliminates in unit i is given by . Under this contractual arrangement, the utility of unit manager i is given by: The game between the principal and the unit managers is given as follows: Stage 1: The principal offers the incentive scheme Stage 2: Each unit manager announces a vector of the shares of unproductive tasks across all units. Stage 3: The principal and all unit managers observe the announcements. Task reallocation represented by βii, βij, ∀i,j, i ≠ j is executed. The principal's objective is to eliminate all unproductive tasks. Thus, he must choose the incentive coefficients βii, βij such that at least one unit manager will fully reveal his private knowledge of the slack in the organisation. The principal maximises the sum of the average productivities subject to the individual incentive constraints (ICi) and the task constraints (TCj). Note that we can neglect the participation constraints since we assume that wages are sufficiently high to motivate managers to work in the public organisation. Our problem is an exercise in implementation theory; see Moore (1992) for a comprehensive survey. We examine the implementation of productivity improvements in strictly or weakly dominant strategies, where possible, and for Nash implementation otherwise. Hence, the principal's problem is given by: The principal maximises expected returns by eliminating unproductive tasks. The incentive constraints require that an announcement vector for unit manager i provide utility that is weakly higher than any other feasible announcement vector , given the announcement vectors of the other unit managers. The task constraints represent the fact that no more than all residual tasks can be distributed to the unit managers. 3.2. Unit Specific Knowledge If the unit managers can only reveal the share of unproductive tasks in their own unit, the principal offers the contract Di[βii(α1,…,αi,…,αn), βij(α1,…,αi,…,αn)]∀i,j, j ≠ i. αi denotes the share of unproductive tasks in unit i revealed by unit manager i with αi ≤ Ai. Under this contractual arrangement, the utility of unit manager i is given by: The structure of the game is similar to the case of organisation specific knowledge. The only difference is the contract offered by the principal. The principal's problem in this case is given by: 3.3. First‐Best We complete our model by the characterisation of the first‐best solution, which follows immediately from the principal's profit function. Proposition 1 (i) An incentive scheme under organisation specific knowledge is first‐best if for every i . (ii)An incentive scheme under unit specific knowledge is first‐best if for every i αi = Ai. At this point it is obvious that the first‐best solution can be achieved by standard incentive schemes if the principal is able to use monetary incentives or the threat of firing; see e.g. the surveys by Hart and Holmström (1987); Prendergast (1999). Our novel element is the design of non‐monetary incentive schemes to achieve first‐best solutions when standard stick and carrot instruments cannot be used. 4. Optimal Incentive Schemes Without Collusion In this Section, we derive the incentive schemes for implementing the first‐best solution when unit managers do not collude. 4.1. Organisation Specific Knowledge If the unit managers have organisation specific knowledge, the principal can implement a first‐best solution by a tournament as shown in the next Proposition. Proposition 2 Suppose the unit managers have organisation specific knowledge. The principal can implement a first‐best solution as a unique equilibrium of the announcement game with the following tournament where m denotes the number of unit managers revealing the highest amount of unproductive tasks: 10 Every unit manager i announces (A1,…,An) and expects utility The proof is given in the Appendix. The tournament scheme implements a first‐best solution since unit managers only have the chance to obtain tasks if they reveal as much as other unit managers. This induces unit managers to reveal the maximum possible. Besides the tournament scheme there are other possibilities of implementing the first‐best solution. The principal can also use incentive schemes that eliminate the unit managers’ costs of announcements instead of rewarding large‐scale announcements of unproductive tasks with additional tasks and utility as discussed above. For instance, the principal can use a specific form of job rotation. Proposition 3 Suppose the unit managers have organisation specific knowledge, then the principal can implement a first‐best solution through job rotation: Proof: Under job rotation, the unit managers’ utility after rotation does not depend on the announcements of unproductive tasks in their own unit. Thus, the unit managers are indifferent with respect to the revelation of unproductive tasks in their own unit. Hence, every combination of announcements is part of an equilibrium of the announcement game. In particular, the announcement vectors constitute a Nash equilibrium of the announcement game. The main disadvantage of this scheme is the non‐uniqueness of equilibria in the announcement game.11 Hence, the principal can only implement the first‐best solution by job rotation with certainty if indifferent unit managers act in the interests of the principal. In contrast to the tournament scheme, in which the unit managers completely reveal their private knowledge, the rotation scheme may not induce unit managers to reveal the maximum share of unproductive tasks in each unit, since unit managers could lose by rotating to the next unit. 4.2. Unit Specific Knowledge We now consider unit specific knowledge where the unproductive tasks in the manager's own unit can only be demonstrated to the principal as hard information. We will show that the principal can always design an incentive scheme, which implements a first‐best solution. This scheme requires that unit managers must be compensated for the loss of tasks in their own unit by a transfer of tasks from other unit managers. Proposition 4. Suppose n > 2. Suppose the unit managers have unit specific knowledge. Then, an incentive scheme exists which uniquely implements a first‐best solution. The scheme is given by: The proof is given in the Appendix. Under the proposed incentive scheme, unit managers have a strict incentive to reveal their knowledge given the announcements of the other unit managers, because the loss of tasks in their traditional unit is compensated by a higher share of tasks obtained from other units. Since unit managers receive a share of their traditional tasks and a share of the other unit managers’ traditional tasks, such schemes can be characterised as partial job rotation. Note that the announcement of all unproductive tasks is a dominant strategy. Therefore, the incentive scheme works independently of whether unit managers have the same limited information about unproductive tasks in other units as the principal or whether unit managers may observe the extent of unproductive tasks in other units but at the same time be unable to provide hard information to the principal. Therefore, the limitations of the revelation principle (Haller, 1992) do not apply in our context. The schemes in Proposition 4 and in Propositions 2 and 3 imply that some managers end up with more tasks and others with fewer. This could occur through yearly reallocations of budgets without influencing the job status and the pay of unit managers. For instance, professors in German universities are endowed with a certain budget, which can change every year, and thus professors can undertake more or fewer tasks. Allocating more tasks can also be interpreted as promotion, while fewer tasks could mean demotion. Such an interpretation is more problematic, since in Germany, for example budget reductions for managers in public organisations are widespread but they are not real demotion with associated pay reductions. As in the case of organisation specific knowledge, job rotation can yield a first‐best solution. 5. Collusion Among Unit Managers In this Section, we allow for collusion among unit managers. We believe that unit managers have strong incentives to go for collusion agreements. Since, under the incentive schemes in Propositions 2 and 4 for instance, activities will be eliminated, the aggregate utility of managers declines. If the unit managers have the same information structure, they can write binding side contracts which can hurt the principal. This is obvious in the case of organisation specific knowledge. In the case of unit specific knowledge, the possibility of collusion through contracts requires unit managers to have the same information about the proportion of unproductive tasks across units. However, as they do not share the same knowledge, they can only give hard information about unproductive tasks in their own unit. For instance, the unit managers know the extent of unproductive tasks in their own unit but cannot indicate the unproductive tasks in other units to the principal because their knowledge is not sufficiently detailed. Without this assumption, collusion under unit specific knowledge does not occur and therefore our scheme in Proposition 4 can be used to achieve the first‐best solution. In the following, we assume that collusion is possible in both cases in order to explore how collusion can be broken. 5.1. Side Contracts Between Unit Managers We examine collusion in which agents are not able to transfer utility via side payments in order to remain consistent in excluding monetary incentives. Hence, a side contract is only characterised by the announcements of the unit managers, i.e. the announcements unit managers agree to supply to the principal as hard information. In the following we introduce side contracts for unit specific knowledge without side payments. In Sections 5.2. and 5.3. we discuss collusion for organisation specific and unit specific knowledge in detail. Let us denote side contracts under unit specific knowledge by C(α1,…,αn) where (α1,…,αn) is the vector of announcements all n unit managers agree to deliver to the principal. Collusion agreements for a subset of unit managers are defined accordingly. Obviously, the set of side contracts the unit managers will write under a given incentive scheme fulfills the condition that no other side contract exists with and As a tie‐breaking rule, we assume that unit managers do not engage in side contracts if they are indifferent between the non‐cooperative outcome and the outcome under a side contract as defined above. That is, if they are indifferent, unit managers will act in the principal's interests by not colluding. In the case of unit specific knowledge, the game between the principal and the unit managers when side contracts can be written is given as follows: Stage 1: The principal offers the incentive scheme Stage 2: Unit managers may write side contracts C(α1,…,αn). Stage 3: Unit managers make their announcements. Stage 4: The principal and all unit managers observe the announcements. Task reallocation as determined by βii,βij is executed. 5.2. Organisation Specific Knowledge If agents can collude, the tournament scheme in the last Section is not collusion‐proof. The unit managers can raise utility with a side contract in which they commit themselves to revealing nothing. Such a side contract C[(0,…,0),…,(0,…,0)] implies where is the utility when collusion is possible and the utility under no collusion when the principal uses a tournament scheme. However, the principal can modify the tournament scheme to make it collusion‐proof. Ex ante the principal can determine a unit manager who will be treated specially by the incentive scheme if other unit managers reveal the same share of unproductive tasks. Such an incentive scheme is discriminatory but it helps to prevent collusion and to implement a first‐best solution. To simplify notation we introduce the set of managers L who reveal the largest improvement in productivity: We denote the number of elements of L by m. We obtain: Proposition 5. Suppose collusion is feasible and that the unit managers have organisation specific knowledge. Then the principal can implement a first‐best solution through a discriminatory incentive scheme. The incentive coefficients are given by: with 0 < ɛ < 2. Collusion does not occur and every manager i announces (A1,…,An). Utilities are: The proof is given in the Appendix. Note that the scheme is discriminatory as the manager with the lowest index is treated differently if several managers reveal the highest productivity improvements. As before, the unit managers have an incentive to win the tournament. To prevent collusion, the principal must, however, modify the tournament in two ways. First, he has to provide a positive share of tasks for all unit managers who reveal their knowledge completely. Thus, subcoalitions entering collusion agreements are not attractive. Second, the principal has to single out one of the unit managers under consideration with respect to the design of the incentive coefficients. In the tournament scheme, this is achieved by assigning a special share of tasks to the unit manager with the lowest index value if more than one unit manager reveals the maximum amount of unproductive tasks. The manager with the lowest index value in this case is called the special unit manager. Moreover, while the utility of the special unit manager decreases with the announcement level in the discriminatory incentive scheme, the other unit managers’ utility increases with a higher share of tasks revealed to the principal. Therefore, collusion by the grand coalition is not attractive either. The special unit manager obtains the largest share of tasks if the principal sets ɛ very small. If ɛ is higher and closer to 2, the special unit manager can be better or worse off than the other unit managers depending on (A1,…,An). 5.3. Unit Specific Knowledge Given the non‐cooperative solution derived in Section 4.2., the unit managers can increase their payoffs through a side contract under unit specific knowledge. For a given difference in the shares of unproductive tasks across two units, the unit managers have higher utilities if they commit themselves to announcements with a lower absolute level but without changing the difference between them. To illustrate the possibility of collusion in this case, we consider two unit managers. Suppose that the shares of unproductive tasks, A1, A2, are given by A1 = A2 + Δ with Δ > 0 and are sufficiently small. Since utility under the incentive scheme in Section 4.2. depends on the relative announcements of the unit managers, the unit managers can raise their payoffs by lowering the absolute level of the announcements, leaving the relative levels unchanged. Suppose the unit managers write a side contract in which they commit themselves to the announcements α1 = Δ, α2 = 0. Then, the utilities under the side contract are higher than without collusion for both unit managers because: To implement a first‐best solution, the principal must design an incentive scheme, which satisfies the following intuitive requirements. First, in order to prevent collusion, at least one unit manager must gain a higher utility from the noncooperative outcome than from a situation with side contracts. In the following, we denote this individual no‐collusion condition for unit manager i by NCi. Second, the individual incentive constraint must hold for each unit manager, which is most difficult to fulfil in the case of the unit manager who would benefit from side contracting. As demonstrated in the following proposition, a first‐best incentive scheme that can overcome the collusion threat does not exist. Proposition 6. Suppose collusion is feasible and that the unit managers have unit specific knowledge. Then, no incentive scheme exists which implements a first‐best solution. The proof is given in the Appendix. Colluding unit managers can agree to reduce their announcements without changing the order of revealed announcements, which under any feasible incentive scheme make the managers better off.12 6. Conclusion In this article we have analysed incentives schemes in public organisations aimed at improving the productivity by motivating agents such as unit managers to reveal unproductive tasks. Since these organisations are often characterised by non‐verifiable output, wage rigidities, lifetime employment guarantees and limited career perspectives, other incentive schemes than standard stick and carrot incentives need to be applied. If an agent's utility is influenced by the sphere of control, as is often the case for unit managers, task or budget assignments are the principal's last resort in motivating agents. Depending on the knowledge unit managers can give to the principal as hard information, simple tournament and rotation schemes implement a first‐best solution if unit managers act non‐cooperatively. In both schemes, unit managers must be compensated for the loss of the announced tasks by being given tasks from other units. To prevent collusion in the case of organisation specific knowledge, stronger requirements need to be fulfilled. In the case of organisation specific knowledge, the principal must reward unit managers differently even if they make the same announcements. Thus, there is a trade‐off between efficiency and fairness. In the case of unit specific knowledge, an incentive scheme that can break collusion does not even exist. Potential extensions of the article include changes in the assumption on the symmetry of the units. While the incentive schemes for the case of organisation specific knowledge will still hold, adjustments must be made for incentive schemes in the case of unit specific knowledge. Suppose, for instance, that the units differ with respect to their size. This would reduce incentives to reveal unproductive tasks for the manager of the larger unit because he cannot be compensated sufficiently by the residual tasks in the smaller unit. Overall, asymmetry of units tends to lower the power of partial rotation schemes. A further fruitful extension of our article is to enrich the framework by drawing upon the model and the considerations in Aghion and Tirole (1997). In particular, we could allow agents to screen projects on behalf of the principal where each project is associated with a verifiable monetary gain for the principal and private benefit for the agent. The agent can communicate a project proposal to the principal. The principal could either overrule an agent or order another agent to take over. In such a framework, the principal must provide incentives to search for information and its communication. The real question is whether our schemes remain optimal in such circumstances. While it appears that schemes such as the one in Proposition 4 tend to generate appropriate incentives for the search of beneficial projects, a full‐fledged analysis of this point would be a fruitful task for future research. Although the derived schemes implement a first‐best solution in our model, the use of such schemes must be complemented by additional considerations. In particular, the use of discriminatory tournament schemes could be problematic. Several authors emphasise that there is a positive relationship between fairness and agents’ motivation (Akerlof and Yellen, 1990), since agents compare their performances and rewards with those of other agents. Moreover, the use of discriminatory incentive schemes may result in lobbying activities aimed at influencing the choice of the special unit manager, which may harm the principal in other respects (Milgrom and Roberts, 1988, 1990). Hence, the question whether discriminatory incentive schemes are applicable in public organisations merits further research. Another useful line of research would be the investigation of the possibilities to achieve productivity improvements by job design. In particular, to which extent public organisations should ensure that unit managers know the fine details of the production process in other areas than their own units. Our analysis shows a trade‐off between the gains of specialisation and the possibilities of providing incentives if people are specialised. Our analysis suggests that, in general, task and budget assignments can be used as an incentive device motivating unit managers to improve productivity in public organisations. However, due to the difficulties in preventing collusion, task and budget assignments may not always be an equivalent substitute for standard stick and carrot incentives. Nevertheless, their use is indispensable in public organisations for achieving productivity improvements. Appendix A Technical Appendix is available for this paper: http://www.res.org.uk/economic/ta/tahome.asp Footnotes 1 " See e.g. Ministerium für Wissenschaft, Forschung und Kunst Baden‐Württemberg (1998). More examples are discussed in Section 3. 2 " See e.g. the surveys by Hart and Holmström (1987); Holmström and Tirole (1989); Prendergast (1999). Important contributions to this literature include Holmström (1979; 1982a); Grossman and Hart (1983); Lazear (1989); Holmström and Milgrom (1991). 3 " See e.g. Tirole (1986; 1992); Holmström and Milgrom (1990); Faure‐Grimaud et al. (1998). 4 " Technological change may also make certain activities obsolete and thus create unproductive activities too. 5 " See e.g. Bundesbesoldungsgesetz, Bundesangestelltentarifvertrag in the survey of Bundesministerium des Innern (2002). 6 " This has been confirmed in private communication. 7 " For instance, departments that have revealed a significant share of less productive tasks may be rewarded by the allocation of funds in future investment programmes, or by temporal guarantees for their remaining budgets. See also Frankfurter Allgemeine Zeitung, 11.2.2002. 8 " The hard information paradigm was introduced by Tirole (1986; 1992) and used in Laffont and Tirole (1993); Laffont and Meleu (1997); Kofman and Lawaree (1993). 9 " See e.g. Ministerium für Wissenschaft, Forschung und Kunst Baden‐Württemberg (1998). 10 " Note that only one manager will be chosen from the urn since tasks can only be allocated once. 11 " The same problem occurs if the principal simply asks each unit manager about unproductive tasks in other units without using an incentive scheme at all. 12 " One could use outside managers in order to break collusion by assigning tasks to them. References Aghion , P. and Tirole , J. ( 1997 ). ‘Formal and real authority in organizations’ , Journal of Political Economy , vol. 105 , pp. 1 – 29 . Google Scholar Crossref Search ADS WorldCat Akerlof , G. and Yellen , J. ( 1990 ). ‘The fair wage‐effort hypothesis and unemployment , Quarterly Journal of Economics , vol. 105 , pp. 255 – 83 . Google Scholar Crossref Search ADS WorldCat Baily , M. and Gersbach , H. ( 1995 ). ‘Efficiency in manufacturing and the need for global competition’ , Brookings Papers on Economic Activity, Microeconomics , pp. 307 – 58 . OpenURL Placeholder Text WorldCat Bundesministerium des Innern (2002 ). Der öffentliche Dienst in Deutschland . Carmichael , H. and MacLeod , B. ( 1993 ). ‘Multiskilling, technical change and the Japanese firm’ , Economic Journal , vol. 103 , pp. 142 – 60 . Google Scholar Crossref Search ADS WorldCat Dewatripont , M. , Jewitt , I. and Tirole , J. ( 1999a ). ‘The economics of career concerns, part I: comparing information structures’ , Review of Economic Studies , vol. 66 , pp. 183 – 98 . Google Scholar Crossref Search ADS WorldCat Dewatripont , M. , Jewitt , I. and Tirole , J. ( 1999b ). ‘The economics of career concerns, part II: application to missions and accountability of government agencies’ , Review of Economic Studies , vol. 66 , pp. 199 – 217 . Google Scholar Crossref Search ADS WorldCat Faure‐Grimaud , A. , Laffont , J‐J. and Martimort , D. ( 1998 ). ‘A theory of supervision with endogenous transaction costs’ , mimeo, Toulouse. Fitzenberger , B. and Franz , W. ( 1999 ). ‘Industry‐level wage bargaining: a partial rehabilitation–the German experience’ , Scottish Journal of Political Economy , vol. 46 , pp. 437 – 57 . Google Scholar Crossref Search ADS WorldCat Frankfurter Allgemeine Zeitung ( 2002 ). ‘Raus aus dem Elfenbeinturm und rein in die Märkte’ , 11th February, p. 14 . Grossman , S. and Hart , O. ( 1983 ). ‘An analysis of the principal‐agent problem’ , Econometrica , vol. 51 , pp. 329 – 69 . Google Scholar Crossref Search ADS WorldCat Haller , H. ( 1992 ). ‘Limitations of the revelation principle’ , Economic Letters , vol. 39 , pp. 409 – 13 . Google Scholar Crossref Search ADS WorldCat Hart , O. and Holmström , B. ( 1987 ). ‘The theory of contracts’ , in ( T. Bewley, ed.), Advances in Economic Theory, Fifth World Congress , pp. 71 – 155 , Cambridge: Cambridge University Press . OpenURL Placeholder Text WorldCat Holmström , B. ( 1979 ). ‘Moral hazard and observability’ , Bell Journal of Economics , vol. 10 , pp. 74 – 91 . Google Scholar Crossref Search ADS WorldCat Holmström , B. ( 1982a ). ‘Moral hazard in teams’ , Bell Journal of Economics , vol. 13,. pp. 324 – 40 . Google Scholar Crossref Search ADS WorldCat Holmström , B. ( 1982b ). ‘Managerial incentive problems: a dynamic perspective’, in: Essays in Economics and Management in Honor of Lars Wahlbeck , Helsinki: Swedish School of Economics. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Holmström , B. and Milgrom , P. ( 1990 ). ‘Regulating trade among agents’ , Journal of Institutional and Theoretical Economics , vol. 146 , pp. 85 – 105 . OpenURL Placeholder Text WorldCat Holmström , B. and Milgrom , P. ( 1991 ). ‘Multi‐task principal‐agent analyzes: incentive contracts, asset ownership, and job design’, Journal of Law , Economics and Organization , vol. 7 , pp. 24 – 52 . Google Scholar Crossref Search ADS WorldCat Holmström , B. and Tirole , J. ( 1989 ). ‘The theory of the firm’ , in ( R. Schmalensee et al. , eds.), Handbook of Industrial Organization , vol. 1 , pp. 61 – 133 , Amsterdam, North Holland. Google Scholar Crossref Search ADS WorldCat Imai , M. ( 1989 ). Kaizen: The Key to Japan's Competitive Success , New York: McGraw‐Hill . Kofman , F. and Lawaree , J. ( 1993 ). ‘Collusion in hierarchical agency’ , Econometrica , vol. 61 , pp. 629 – 56 . Google Scholar Crossref Search ADS WorldCat Laffont , J.‐J. and Meleu , M. ( 1997 ). ‘Reciprocal supervision, collusion and organizational design’ , Scandinavian Journal of Economics , vol. 99 , pp. 519 – 40 . Google Scholar Crossref Search ADS WorldCat Laffont , J.‐J. and Tirole , J. ( 1993 ). A Theory of Incentives in Procurement and Regulation , Cambridge: MIT Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lazear , E. ( 1989 ). ‘Pay equality and industrial politics’ , Journal of Political Economy , vol. 97 , pp. 561 – 80 . Google Scholar Crossref Search ADS WorldCat McKinsey Global Institute ( 1993 ). Manufacturing Productivity , Washington, DC: McKinsey & Company, Inc . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Milgrom , P. and Roberts , J. ( 1988 ). ‘Employment contracts, influence activities, and efficient organization design’ , Journal of Political Economy , vol. 96 , pp. 42 – 60 . Google Scholar Crossref Search ADS WorldCat Milgrom , P. and Roberts , J. ( 1990 ). ‘An economic approach to influence activities in organizations’ , American Journal of Sociology , vol. 94 , pp. 154 – 79 . Google Scholar Crossref Search ADS WorldCat Milgrom , P. and Roberts , J. ( 1992 ). Economics, Organization and Management , Englewood Cliffs: Prentice Hall . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ministerium für Wissenschaft, Forschung und Kunst Baden‐Württemberg ( 1998 ). Abschlußbericht der Hochschulkommission , Stuttgart. Moore , J. ( 1992 ). ‘Implementation, contracts, and renegotiation in environments with complete information’, in ( J.‐J . Laffont, ed.), Advances in Economic Theory, Sixth World Congress , vol. 1 , pp. 182 – 282 , Cambridge: Cambridge University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Prendergast , C. ( 1999 ). ‘The provision of incentives in firms’ , Journal of Economic Literature , vol. 37 , pp. 7 – 63 . Google Scholar Crossref Search ADS WorldCat Riesz , F. and Sz.‐Nagy , B. ( 1956 ). Vorlesungen über Funktionalanalysis , Berlin: VEB Deutscher Verlag der Wissenschaften . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tirole , J. ( 1986 ). ‘Hierarchies and bureaucracies: on the role of collusion in organizations’ , Journal of Law, Economics and Organization , vol. 2 , pp. 181 – 214 . OpenURL Placeholder Text WorldCat Tirole , J. ( 1992 ). ‘Collusion and the theory of organizations’, in ( J.‐J. Laffont, ed.), Advances in Economic Theory , vol. 2 , pp. 151 – 206 , Cambridge: Cambridge University Press. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tirole , J. ( 2001 ). ‘Corporate governance’ , Econometrica , vol. 69 , pp. 1 – 35 . Google Scholar Crossref Search ADS WorldCat Wilson , J. ( 1989 ). Bureaucracy: What Government Agencies Do and Why They Do it , New York: Basic Books . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Womack , J. , Jones , D. and Roos , D. ( 1990 ). The Machine that Changed the World , New York: Rawson Associates . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Author notes " We thank Volker Hahn, Martin Hellwig, Till Requate, Eva Terberger‐Stoy and Jean Tirole, seminar participants in Basel, Heidelberg and at the annual meeting of the German Economic Society 2000 in Berlin, two referees and the editor for their helpful comments. © Royal Economic Society 2005.
TI - Productivity Improvements in Public Organisations
JF - The Economic Journal
DO - 10.1111/j.1468-0297.2005.01014.x
DA - 2005-07-01
UR - https://www.deepdyve.com/lp/oxford-university-press/productivity-improvements-in-public-organisations-pQHrf2zsa3
SP - 671
VL - 115
IS - 505
DP - DeepDyve
ER -