HD28 .M414 DEWEY 1^3 ALFRED WORKING PAPER SLOAN SCHOOL OF MANAGEMENT P. Power in Profit Maximizing Organizations Julio J. Rotemberg, MIT School of Management June, 1993 Working Paper No. 3582-93 MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE, MASSACHUSETTS 02139 Power in Profit Maximizing Organizations Julio J. Rotemberg, MIT School of Management June, 1993 Working Paper No. 3582-93 MASSACHUSETTS INSTITUTE OF TECHNOLOGY NOV 1 4 2000 LIBRARIES Power in Profit Maximizing Organizations Julio J. Rotemberg Current version: June * 15, 1993 Abstract This paper considers the question of how profit maximizing firms make decisions. recognizes that members It of these organizations tend to have incompatible preferences over decisions but that willingness to pay for decisions plays a very limited role in actual decision making. Instead, a sizable empirical literature "important" to their organization, i.e. who documents that people who are provide critical services, are hard to replace, or deal effectively with the shocks that affect the organization, are powerful in that they have disproportionate influence over decisions. This paper shows that power over decisions to those who are The recison is willing to pay the most, can be that the right to shape the firm through attractive as an employer. Thus, decision this, its decisions and not giving profit maximizing. makes the firm more making power should be given to those employees that the firm wishes to retain for a long period of time. *Sloan School of Management, M.I.T. I wish Management Workshop for valuable comments. to thank Tom Kochan and the participants at Stanford's Human Resource Power in Profit Maximizing Organizations Rotemberg Julio J. Organizations provide their members with a great variety of non-pecuniary sources of makes a superhuman The problem is differ across members. So, decisions that give a great deal of that, unless the organization less utility to others. install While all members might agree that it carpets, preferences over the details differ. Even new effort at utility to homogeneity, preferences some members would be nice when it utility. comes to often give expand a building or to decisions that affect the long term success of the organizations, such as personnel decisions, people differ in the kind of person they would like to be surrounded with. Employees are thus willing profit to pay to affect the firm's decisions. maximizing firms take decisions they should take into account the reductions compensation that are possible as a result of these decisions. of This implies that, when two decisions which produce the same revenues, a decision which allows it to reduces employee profit In particular, confronted in employee with a choice maximizing firm ought to choose that compensation the most. These considerations suggest that firms would benefit from schemes that ask employees for contributions and which pick that decision for which employees are willing to contribute the most. In practice, decisions inside firms are very decision making rights or on the basis of any other mechanism that are three possible explanations for this absence. that the firms are not profit maximizing at The second sometimes is find seldom made on the basis of either auctions The first, which is elicits contributions. embraced by March (1962) that auctions present special problems inside firms. Finally, the third it There and that the process of decision making proves all for is is it. that firms advantageous to make decisions which do not correspond to those for which their employees are willing to pay the most. I explore this third possibility here. employees by taking decisions that they In particular, like is I construct a model where compensating not a perfect substitute for raising their current wages. The result that the firm will find is it advantageous to compensate certain types of indi- viduals with high wages while others are compensated disproportionately with decisions that are taken according to their wishes. Wages differ from influence in one crucial respect. The employee does not have to remain with the firm at t + to enjoy 1 contrast, the utility from having an influence at time gives may give one pleasure that makes utility some t+ more it from the from attractive. all the benefits from his time ( wage. By t is less portable. feeling in charge but it also changes the organization in a The problem from an employee's point fact that the organization of the utility from the decision at is t is more attractive collected only if when he Influencing decisions at of view is t way that he only gets actually works there. Thus, the individual remains an employee at 1. This has two implications. First, individuals are, all else equal, willing to pay more who expect for the right to to stay at t+ 1 have decisions made according to their wishes. the firm were just trying to mimic the outcome that would prevail with an auction Thus, even if they would let these individuals have a large say. Second, and more importantly the firm will benefit by implementing the wishes of those employees that high probability, even The reason is if these are not necessarily the employees who it what amounts to a my purposes, wants to retain with are willing to pay the most. form of deferred compensation. The motivation giving influence to this subset of employees would vanish that specify future wage compensation in to specify future wages wants to keep, even if all if may benefit cis it is not possible from giving influence to employees it others are willing to pay more. becomes which types of employment relations do firms want prolong by these means. By knowing the answer to this question, one knows who firms in the sense that they have a disproportionate say in firm decisions.^ This question in part because there who have for the firm could write complete contracts states of the world. But, as long in this fashion, the firm this point, the question employees for that, by following these individuals' wishes, the firm raises the probability that they will stay in order to collect At with higher probability is to has power inside is an important a substantial empirical literature documenting the features of the power. 'This is essentially the definition of power proposed by Dahl (1957). He says A has power over C if he is likely to affect C'a behavior. Moreover, A is regarded as having more power than B if y4 is more likely to affect C'a behavior than is B This corresponds in my setting to the idea that A has more power if he is more likely to succeed in having the firm adopt his favorite course of action. In particular, influence First, as suggested role. seems to be associated with two characteristics of a group or employee's by the "resource dependence" theory of power (Emerson 1962) those who control important resources tend to have power. Second, as implied by the "strategic contingency" theory of Hickson et large and arise (1971), those ai. randomly tend who can help their organization cope with problems that are to have power. Evidence for the importance of control over resources (1978) large who show amounts provided by Pfeff"er and Salancik that those academic departments within universities that either bring in the of grant monies or that teach a great many students have power. Further evidence provided by several papers which build on who occupy is Tushman and Romanelli (1983) and is show that individuals central positions in the communications networks of organizations have large influence. Evidence for the view that power flows to those that can cope with uncertainty can be found in Crozier's (1964) observations of maintenance workers in French tobacco plants. These were able to cope with machine breakdowns and, as predicted, had a great deal of power. evidence for the strategic contingency theory I will is More systematic presented in Minings et al (1974). argue that the findings of this empirical literature are consistent with other words, I will suggest that my model. In groups and individuals that either have control over important resources or help the firm cope with uncertainty are likely to be ones whose presence the firm What wants to prolong. more, a closer reading of this empirical literature gives further credence is to the notion that firms give power to those it is keen to keep on the payroll. This paper thus takes issue with March (1962) and Pfeffer (1981). the evidence on decision (1962) views firnris making inside firms is instead as political coalitions. inconsistent with profit maximization.^ In this of owners are not ^This study all as well as of the coalition" (p. that important in decision making. many In other words, people are members March view "the focus of attention shifts from the owners to the actual, operating organizers of the coalition... loose constraints on the active Both authors argue that My (Stockholders') demands form 112). In other words, the objectives aim is to show that this de-emphasis others including Perrow (1970), Minings et aJ.(1974) measures influence by reputational variables. to have power if others believe that they do. See Pfeffer (1981) for evidence that people who deemed power tend to have a lot of power in practice. power extended well beyond pure maintenance issues. Croiier reports "Finally, any change in the organisation of production itself, any move by the assistant director in his own domain, will be subject - because of the position of the maintenance workers as natural leaders in the shop - to the interference of the maintenance department and its all-powerful chief." (p. 126) This shows that, unlike what happens in the theory of Jensen and Meckling (1992), power does not necessarily flow to those who have the best information about what decision is best. Pfeffer (1981) contains several other examples where power goes to people other than the "experts". *They actually take issue with the broader idea that Arms pursue any superordinate goals. are viewed as having ''Their of owners not warranted by is much of the existing evidence. Similarly, Pfeffer (1981) dismisses the idea that decision he means "goal consistency and congruity" ular goal making and he views (p. 20) is profit assumed by economists. He contrasts these perspectives by saying "The rational model that parochial interests and preferences control choice" (p. 22). model implies that "choice should be groups" He thus regards . in predicting who it relatively He goes on to say that the ratio- uncorrected with the preferences of the same as invalidated because, aa we saw, group as follows. In Section 1, I characteristics are very helpful present the basic structure of the model: two employees have divergent preferences over some decision. the profit maximizing solution make model presumes political has power. The paper proceeds to rationality, maximization as the partic- presumes that information and value-maximization dictate choice; the nal By "rational". is to let the the decision. In section 2, I set employee who I is show that, without other distortions, willing to pay the most have the right up a model that has a distortion which can be ameliorated by the selective granting of power. In particular, employees have outside periods that they work for the firm. The a high compensation in the future. The reason result employee both today (because he knows that he is is off"ers in each of the two that the firm would like to be able to promise that this disposes of the danger of losing the will receive a high level of income in the future) as well as in the future period. But, it seems plausible to assume that firms are not able to commit to the future compensation of the employee. The most important reason for this is that there are good reasons to terminate the employment relations for exogenous reeisons (such as changes needs). Moreover, the constellation of reasons for termination cannot plausibly be written into a contract. Rather, whenever future I it wage show it sufficiently makes sense to let employment complex that they the firm fire the employee wishes. But, in the presence of this freedom, any contract that specifies the employees will be renegotiated. in Section 3 that this absence of commitment power to the more valuable employee even though he as other employees. But, will is in their let firms if the firm is free to fire the is implies that the firm might wish to give not willing to pay eis much for this power employee then any promise of a future wage be renegotiated. Section 4 focuses on groups rather than on individuals. What makes groups from individuals different what payment they will receive in the future) sometimes equilibria that groups tend to consist of both young employees (who are unsure is exist that provide and old employees. The result some wage security to employees. I is that reputational show that, nonetheless, power might be given to the more valuable group. Section 5 returns to issues of individual power. employees will shows that the desire to keep valuable It sometimes lead the firm to distribute power very unequally. It will do this even certain cases where the employee benefits from this concentrated power are smaller than the of benefits employees would obtain if in sum power were distributed more evenly. Section 6 departs from the earlier sections by treating the case where individuals use their power only to increase their income (as opposed to using income raise one's will it to obtain non-pecuniary benefits). tend to flow to individuals who show that even I are valuable in the here, power to Section 7 first place. concludes. A 1. static setting In this section affects its I with divergent preferences consider a one period model where the firm has to employees' welfare. The purpose of this section is to make one demonstrate that, distortions, employees' willingness to pay plays a key role in optimal decision for is what and is follows, I then turn to a discussion of the ways in decision which in the absence of making. As a prelude which the empirical literature on power not consistent with this model. Consider an owner entrepreneur whose firm has two risk neutral employees two jobs a and 6 respectively. Both employees work for it makes sense to imagine that that this preference arises only because A prefers a. gives and B that fulfill one period and, before the beginning of period, the entrepreneur has to choose between two decisions which notation, A a whereas A more B I will label prefers a and /?. Given In this section fi. I this this assume non-pecuniary benefits, and similarly for B. Moreover, I assume that issues considered in Jensen decision making decisions. The rights idea is by more. The reason if this decision has no other effect on I thus abstract from the and Meckling (1992). They emphasize that employees will be given they are particularly well informed about the consequences of various that better informed employees are able to I profits. ignore these considerations 5 is make decisions that raise profits that observers of power and decision making inside organizations such as Crozier (1964) to the best informed.^ That and Pfeffer (1981) stress that power is often not allocated not to suggest that expertise plays no role in decision making. is It is rather that the benefits from giving power to experts will have to be traded off against the benefits from allocating power I a suppose that, if in the a is way that B units of compensation while much as A chosen, is suggested by the models knows a and ^. will consider. gets non-pecuniary benefits that are worth as gets no non-pecuniary benefits. Similarly, units of compensation and has no /S I He should then choose eff"ect /? on A's utility. Suppose and lower B's salary by to him as raises B's utility as that the entrepreneur first exceeds if ^^ ^^ /? much a and should choose a and lower A's salary by a otherwise. This strategy for the entrepreneur leads him to gain either a or ^, whichever is larger, In practice, the entrepreneur and has no is effect unlikely to on the employees' know the values of the decision, the entrepreneur could auction the right to English open outcry auction (in which the decision pay the most make the in the sense that the other decision himself) otherwise. The exceeds a, B would let employee B make difference with the case who decision-making. number if is the decision to the two employees. made by the employee who is willing to not willing to pay more to get the right to if /? exceeds a and would this right /? is let that, A make it assuming ^ and would thus personally gain /3 — a.^ value the non-pecuniary consequences of decisions more should The question is whether this simple model of deci- certain individuals and groups have a disproportionate influence on In the one-period of decisions so that the employees where the entrepreneur knows a and be given the right to make these decisions. making can explain why is make the decision would only have to pay a to get In conclusion, individuals sion a and ^ Supposing that each employee knows only their own private value of will gain informational rents. An utilities. model considered so far, individuals the value they assign to the non-pecuniary benefits possible origins for such a high value. The first is make disproportionate is high. that these individuals get There are two more utility from getting their way. This seems like a rather tenuous basis on which to explain the empirical distri- bution of power. The reason is that power appears to be systematically related to variables such as the degree to which the individual and groups's action are important in buffering the firm from An example from Croiier (1964) is discussed in footnote 3. *This solution works so well because the value of the decision is purely private to the individual Matters would be more complex if the utility of several employees were affected by each of the choices available. Nonetheless, mechanisms with characteristics similar to an auction which do achieve efficient outcomes exist in this case as well (see Groves and Ledyard (1977). external shocks and the degree to which substitution of these particular employees by outsiders difficult (see Minings et a/. (1974)). It is is hard to see why these factors would be associated with a large utility for particular decisions. The second reason an fits stemming from many decisions in their the make be is that they cissign a relatively low value to marginal increments money income. They are then willing to give up more Such a low marginal their well-being. in individual or group could assign a high value to the non-pecuniary bene- first utility of income is to obtain decisions income likely to arise if one's The theory would then imply that a disproportionate number made by high-income employees. This can potentially explain have power. After all, it are critical (Emerson (1962)), or cope with the firm's uncertainties (Hickson et ai.(1971)) and are hard to replace (Minings tion.^ large of decisions would why employees which to the firm in the sense that they either control important resources The is Thus, people with higher income should be willing to pay more for the right to place. decisions. which increase et aJ.(1974) seems plausible to suppose that such valuable employees get high incomes. distribution of power can thus be derived from the distribution of income in an organiza- Mowever, this analysis fails to account for two sets of facts. First, it does not rationalize the absence of explicit side payments in the process of making decisions. Second, workers, it cannot explain the existence of employees, such as Crozier's (1964) maintenance who have Nor does relatively less it a great deal of power and relatively limited income.^ account for cases where highly paid professionals power than outcomes such as these, I line employees whose income is fulfilling staff lower (Dalton (1959)). construct a model where equilibrium wage income ^This analysis assumed implicitly that the firm was already important caveat because some of the empirical studies of power harder to be certain that the wages paid by these institutions are summarized in Pfeffer (1981), Pfeffer and Salancik analyzed the is functions have To not set so as to paying individuals the profit maximizing wage. distribution involve optimal in the first deal with This government-owned institutions and is it an is place. For instance, in a series of studies power of various academic departments within the confines and the University of California systems. They showed that departments which brought in grants were powerful in that they tended to be represented on important committees and that the promotions of their faculty were more rapid. If one imagines that these universities were unable to pay these valuable faculties more because of bureaucratic rules then it would make sense to give them non-pecuniary favors instead. Thus power in these institutions can be a substitute rather than a complement for income. 'At this point it might be argued that, in spite of the absence of auctions or other market mechanisms inside firms, the political process produces the same outcome as "as if" the decisions were made through a market mechanism. But, particularly with the collapse of communist regimes, we generally regard explicit market mechanisms as superior to central planning even though, to some extent, central planners do in fact emulate markets. * Crozier reports that management "would like [the maintenance workers) to apply for supervisors' jobs, but they consistently refuse what they consider an unattractive offer." (p. 105). This suggests both the supervisors are paid more [since otherwise management's offer would not make sense) and that the amenities (including power) of the supervisors are smaller Thus power and income are not in direct proportion. of the University of Illinois . The ensure a fully optimal employment relationship. certain employees to remedy important 2. A I in the then that power will be allocated to is this equilibrium deficiency in wages. In the next section equilibrium wages in such a model when there then demonstrate result no opportunity is for giving power I derive the to employees. I subsequent section that giving power to valuable employees can play an role in this context. model -where wages tend to be now suppose in later periods that the firm operates for two periods rather than one. employee to have outside opportunities both The reason is and in the current that the firm would benefit from promising the future. low inefficiently that such promises, credible, if do this to allow the in the future period. employees that they its I will receive large result payment is in would reduce the employees' temptation Before dealing with outside opportunities, to leave today at a relatively low cost. The I specify the other determinant of the employees' wages, namely their value to the firm. I assume that employees more valuable job.^° A and B have some firm-specific human capital and, as a result, are same to the firm than potential outside employees that can in principle carry out the The revenue produced by their outside replacements these employees in period can produce only R*^ R\ and R'g respectively while and wl respectively. As a w'^ A and B vf^Rf-R^ + w^ equals and Rl respectively. The outside employees, were to be hired, would require wage payments firm would be willing to pay employees t in period = vf i2f - t i?* result, the bent employees. Without loss of generality, I It is will they most the equals + u;* . = 1,2 (1) These are the differences between the value produced by the incumbent employees and the the firm can extreict from their replacements. if profits the value contributed to the firm by the incum- A assume that employee produces more value so that vf exceeds vf I It is will show in the next section that individuals with high thus worth discussing the sources of such high v's. v's. The will obtain disproportionate power. In particular, characteristics of employees that have been associated with be thought of as raising employee's t; power I want to show that those in the empirical literature root cause of the difference in v's is '"This lack of ready replaceability has been shown to be an important determinant of power by Hickson Hinings et ai.(1974). It plays a crucial role in my analysis as well. 8 human can capital et ai.(1971) and by i.e., the acquired or innate characteristics that make superiority can be the due to intrinsic abilities, but positions where the employee is the employee superior to it is new employees. This often the result of being placed in a specific able to acquire information whose possession valuable to the is firm. It seem plausible that control over important resources allows employees to mation. Employees who collect such infor- control resources presumable can prevent others from learning how them. Thus, Crozier's (1964) maintenance workers prevented others from knowing how to to use fix the machines by making manuals disappear. '^ This ensured that they alone had the requisite knowledge, and power in made them valuable to the firm. explain Pfeffer's (1981) evidence concerning academic departments within universities along similar which bring in large quantities of of value in the employees is likely to Iccist many be much more costly groups and individuals in (in the sense that departments that teach few students and bring who can cope with new employees income or student happiness et ay.(1971) uncertainty have high to perform as well as old employees and Hinings The reason v's. if in little is difficulties in the past are more valuable than new the degree to which a subunit copes with uncertainty. which the subunit is They do money. et ai.(1974) that it is If, who have coped Hinings et a/.(1974) measure this by combining the extent to is in fact uncertain. ^^ Scoring highly on measure of coping would seem to require a great deal of specialized knowledge which, implies a high much involved in activities that deal with unforecastable events^'^ and information on whether the environment facing the particular subunit this ones. will fall) they face a routinized task. instead, the task involves coping with a great deal of variability, then employees with such students tend to some incumbent employees provide a great deal seems plausible that, as suggested by Hickson easier to train He shows that departments form of money and happy students. Replacing these proven incumbents with new than replacing incumbents also lines. grant monies and departments that teach have power. These are departments where at It One can in turn, v. Finally, there is a large literature originating with Tushman and Romanelli (1983) which shows that the extent to which one communicates with others in the organization affects one's power. In "See Croiier (1964) p. 153. '^They measure whether the group tries to prevent unforecastable events from happening, whether the group tries to anticipate these events and whether the group tries to "absorb" them so that they have a minimal effect on the firm ''Depending on the unit the relevant uncertainty involves unforecastable movements in the level of sales, its composition, the quality or quantity of its inputs or the credit-worthiness of the customers. particular, being central to this "interaction network" raises one's power. The reason could that being central involves an extensive knowledge of the organization which raises one's who know focusing on networks inside firms, Krackhart (1990) shows that those communications better, its i.e. well be Also v. the network of those whose personal depiction of the network parallels more closely actual links, have power. Knowledge of one's organization is both something that takes a long time to acquire (so that new recruits would lack such knowledge) and done. Thus, such knowledge too would give a high v to In addition to the employee's internal value, there its is is crucial for getting things possessor. a second determinant of his wage. This The key assumption my is the employee's opportunity set in the outside labor market. is that individuals have access to outside offers both in the current and the future period. These outside offers do not present difficulties would then separate trast, the firm to do so. But, if more outside options. This in his relies critically (and thereby keep the employee) when it is very hard to component know the value is con- efficient of compensation is some- the employee assigns to the on the lack of observability of the first period offer non-pecuniary benefits from is better informed about first offer the employee receives observable or not does not decisions.^'' Therefore, model by assuming that the employer knows the characteristics of the that assuming the employer it is By offer. second period of employment. Whether the affect the analysis in the case of firm and the employee individual's value to the firm. true not only because the monetary is non-pecuniary benefits of the outside argument offer more than the The analysis assume that the firm does not know the value of individual's realistic to times unverifiable but also because My the firm knows their value. the outside offer pays would match the outside it is if for period offers first period is realistic. I simplify the ofTer. It may In particular, be it seems reasonable to suppose that young people are treated more symmetrically by the job market (so that their outside offers are relatively similar). Moreover, it that young people in general are more adaptable so that they are to move. By suggest that there show seems plausible to suppose more equal in their willingness contrast, only a subset of older employees have attractive outside options and, in addition, only a subset of '*I also in Section is less them is ready to change their lives midstream. Both these arguments information about the outside options of older employees. In any event. 6 that, in the case of pecuniary benefits, the lack of observability of 10 first period offers plays a role as well. the main reason for assuming that the first period offer common knowledge is that is it simplifies the analysis. To are A further simplify the analysis, drawn from the same and B have available two periods. If A I distributions. I I assume that both employees have access to offers that thus assume that the outside options that employees period pay them a present discounted value of z over the in the first remains at the firm variable with distribution function F. same also in period one, his second period outside offer B's second period offer is a random drawn independently from is this distribution. eissume that the firm cannot precommit at the beginning of the current period. j a wage equal to W^ After this wage . I start with the second period. it sets the employees' current At the beginning of period is the firm offers t, its announced, each employee receives an outside W^ in exchange Employee A leaves decides whether to stay and collect the wage wages, wages so that its his employee offer and work. To study equilibrium for his when wages wage W^ is smaller than his outside opportunity and similarly for B. Thus, the firm sets wages to maximize Assuming there a positive probability that outside offers are below than is vi^, optimal wages satisfy f{wi){vi - Wi] - F{WI) =0 j = A,B (2) This implies that the wage of employee j must be smaller than v^ so that employees sometimes leave to take jobs where their productivity Hall and Lazear (1984). between the weige and The its is lower. This inefficiency arises here for the firm benefits in spite of the inefficiency because reservation value in all it same reason as in gains the difference the states of nature where the employee remains with the firm. Another implication of (2), and one that will play a key role below, for a given distribution function of offers, raises the probability can see this from the fact that the derivative of (2) Wj W^ is is One positive while the second negative. so that the probability that employee j stays rises as well. 11 that an increase in v that the employee stays. with respect to Vj order conditions require that the derivative with respect to raises F is Thus an increase in V2 This suggests that employees who have a lot of specific human capital (so that their v is high while their outside offers tend to be low) have a high probability of remaining with the firm. human Consider by contrast the case of employees whose general high but the probability that they will get outside offers below v that the probability that these employees will remain In period 1, capital is leave. high so that their v Equation low. (2) is then implies low. is the employees do not stay unless they can expect the whether they stay or is Thus period one wages must be Wi = z-pl F{W^)Wi + y"" same present value of income at least equal to xdF{x) (3) J The resulting profits of the firm are '^"'= I it \v{-z + p\F{Wi)vi+ assume that both the term pertaining were negative first E for either A to r xdF{x)]\ (4) and that pertaining to in (4) B are positive. employee, the firm would be better off letting that employee leave in the period. It is worth contrasting the outcome A to pay prespecified future wages. (4) itself. The first with what would occur in (4) firm with the ability to if the firm could precommit would set W2 commit to itself maximize order conditions for this problem are f{wi){vi-Wi)^0 This implies that the second period wage that period. This in period offset by 2. may seem The reason its ability Because is the firm does this to pay a lower (2) implies that the wage is wages themselves fall (5) set equal to i^ so that the surprising since the firm is effectively giving up its is efficient in monopsony position is in period one. second period wage is like to over time. Indeed, as long as u^ set below commit is Vj, we can conclude that second itself to. This does not mean that sufficiently larger consistent with rises over time in each employee's wages even without any rise or not, the firm outcome that the resulting increase in second period wages period wages are lower than what the firm would they If would prefer to commit itself to 12 than v[, the model is commitment. Whether pay the same present discounted value in the form of a compensation profile that rises even more shall see that letting individuals make benefits similar to those of such wage commitments. The 3. I power role of in the basic now introduce power The the relevant decision. issue decision in question a discrete decision as I is An K is taken before the beginning of the in Section firm. them power, achieves made decisions are according making as officially in charge of A or B. period. Rather than assuming first assume that the owner must choose a parameter I 1, by his own income minus with works at the if then whether the firm acquiesces to is assumed giving i.e., owner of the firm A's utility /c/3. The that the non-pecuniary benefits and costs that as he or she period individual has power treat the I K which can be either positive or negative. utility is given first we model into the model. To keep matters simple, to his wishes. The decisions in the steeply over time. In the next section, it This seems natural is given by his income plus /ca while B's crucial confers if assumption concerning the choice of must affect each employee for as long one thinks of the decision as involving the purchase of a long lived asset or the hiring of a co-worker. Insofar these make the employee better off, they do so for whatever period the employee actually works with the capital or the co-worker. It is commit worth noting itself to at this point that decisions will be make employees. a exceeds the decision that maximizes the while increasing B's. it the firm should set k to /3 If it knows the preferences would do best by setting keeping the difference. I will Section 2, particular, now show /c 1, its will To make of all utility as V2. 1 if the firm can There is then no the non-pecuniary benefits of largest possible value and reduce A's wage employees and the largest value of k reducing A's wage by a in all each period, raising B's by is one, /9 and ^^ when that, the owner cannot pre-set future compensation in the model of demonstrate that the owner higher) even in cases where his way. = of sum its willingness to pay will sometimes play a I just as in Section paying a second period wage that gives the same reason not to If made this point, B I other employee incurs no losses, it will make assume from now on that of its all exceeds a. is I A willing to pay will show (whose v more In is to get that, nonetheless can once again conduct an auction for the right to make the win the auction but, if the firm wants to make sure that the the resulting wage savings to the other employee. employees, it for the decision will have to give /3 role in the decision. the decision favored by has a larger stake in the decision and '*When it does not know the preferences The employee with the most value decision. will much more muted 13 make k the owner will choose to Given a particular choice of = n, Assuming {vt when va positive /c, wages as before that the solution is sufficiently high. second period are chosen to maximize in the - W^)FiW,'' + is Ka) + (vf - W,^)F{W,^ - k^) (6) interior, the first order conditions for this maximization are: f{W^ + ^&)[v^ - W^] - F{W^ + The second order /ca) = /{Pyf maximum conditions for a is K. To see this, I One issue that arises at this point is (8), equation (9) implies that larger than /) or that in spite of it The that result — ^2 /(v2 I is - f'[v^ W^ + na how A stays higher is would exceed the cost F and obtain - W^] ^ This means that increases To understand W^ + employee who is is willing to that the firm chooses maximization /c to A likes not, by itself a sufficient pay more. To see K&)[v^ rises wage so slightly this, suppose that we ignore period one and assume profits in (6). - W^] - pf[wf 14 more valuable for not allocating the decision to the The first is: + na constant as well. raises the probability that the argument maximize the second period a/(Vy/ — W^^] in koc raise the this result note that lowering the F ' of doing so. have shown that making the decision that this when k the probabilities of departure depend would be that, since the wage would be lower, the benefit from raising the wage ) (7) (8) kept constant would keep the probability of staying is = A's wage either rises with k (in the case where /'[v^ the subsequent adjustment of the wage. employee stays. But this 2/ k^) = A,B declines by less than the increeise in /ca. In either case, probability that the employee will stay. much j diiTerentiate the first equation in (7) dKa Given - W^] - F(Pyf - larger, equation (7) implies that the probability that is set equal to zero. on K0)[v^ then requires that /'K'-O-2/<0 Because va + /c^)[vf - W^\ = order condition for which, using (7) becomes aF[W^ + Ko) - 0F{Wf + This means that a should be positive (so that probability that for an increment A stays exceeds k in is equal to A a times small amount ensures that the winner which in is k, = should make the decision) when a times the B times the probability that 13 stays. But, A's willingness to pay the probability that he stays and analogously for B. This means that an auction between the two employees model K^) is for the right to increase (or decrease) the person that ought to make by a /c the decision. Thus, the why willingness chosen only with second period profits in mind does not explain to pay plays such a small role in firm decision making. my therefore turn I attention to the first period. As in my analysis of the second period, start I by treating k as known and derive the equilibrium wages. These are particularly straightforward because if I assumed that the outside offers of the them a wage is employee assures himself a present value of z. the firm wants to keep the employees, indifferent By W^ + than two employees where known it must offer Thus, in this period. them at least sufficient to keep between staying and leaving. leaving, an Kct in the first period. this. If In the second period, he gets the second period's outside offer the wage that keeps him is W^ + higher, he leaves By /cq staying, employee if his outside offer and gets the outside A gets is lower offer. Thus indifferent satisfies. roo W^ = z and similarly for Ka - pF{W^ + Ka){W^ + Ka] xdF{x) (10) B. By paying these wages, the present discounted value TTp - = vf +vf -2z + /c(a - of the profits from A and B ^) roo roo + p\ F{W^ + Ka)[W^ + + [4 - Wt)P[W^ + Using Leibnitz's K becomes /ca] K~0) + + / (vf xdF{x) + F{wf - k0){W^ - - W^)F{wf - rule, the derivative of this present /c/3)} k0\ + xdF{x) (11) discounted value of profits with respect to IS {a[l+pF(Py2'' + «a)l-^ll+pF(tvf-/c^)]}+p[a(v2^-PV2'')/(<+«")--^(^f-^^2^)/(^^2^-'^^). (12) 15 to pay for an increment in k Employee A's willingness Thus for B. and only A is for B an increment than is + pF{W.^ + pay willing to and similarly /ca)] decrement for a if if + pF{W^ + a[l is more willing to pay d(l is positive. This expression is Ka)] - ^{l term identical to the that the term in square brackets of (12) which is + pF{W^ - in curly k^)] (13) brackets of (12). Equation (7) implies the difference between the expression in (12) and the expression in (13) equals p[aF{W^ + - PF{W^ - /ca) k^)] (14) Therefore, as long as A's probability of remaining in the second period exceeds B's, (12) can be positive even when (13) probability of remaining t; higher is when k We for a negative k. is The economic must pay in A A wage will is B effects is equal to the difference But, there in k. remain at the firm The main motivation the same human offers is in an additional as B's. capital his v high probability. ^ where B in his expected k, specific A The instead of a. less it t;"^ raises the difference between between the two employee's willingness to pay effect of raising /c. Doing so is that his v human capital. is pay A is for for raises the probability more valuable, changes this effect in k. large while his distribution of offers If, instead, A would be higher but he would also tend to get higher result willing to lowers the wage that the firm disutility. in the willingness to power to Thus A has more The is then the higher value of /3, period 2 (and lowers B's). Because for giving were very unlikely to pay choose a = if the following. Raising by the increase can be sufficient to overcome any difference is this larger value of period one by the expected benefits that thereby accrue to A. Similarly, a small change that Thus, gets his way. rationale for this result wages that must be paid to these two larger. that A's thus have the possibility that willingness to pay will be overridden by the desire to keep the more valuable employee. Indeed, immediately implies that is to a positive value even in cases /c Remember willing to pay more). zero because his v is can be sufficient to induce the firm to set pay more B negative (so that is had only more general offers. than v^, the firm would set a wage so that A If these outside would leave with would be that expression (14) would be negative and the firm would Thus power comes from specific rather 16 than general human capital. This implication is supported to some extent by the findings of Fombrun (1983) and Ibarra (1993). They show that tenure on the human capital, is which job, likely to is itself be positively associated with having specific positively correlated with influence over decisions. ^^ They also very highly correlated with centrality in the interaction network which, as is has been shown to be correlated with power The previous Given (12) and (13) earlier, papers. as It is A k raises the probability that in will stay and reduces both A's willingness to pay for an increment to k and this raises further the firm's benefit from doing so. to an employee. many mentioned analysis shows conditions under which the firm prefers a slightly positive value of K to setting k equal zero. But, any increase 5's. a great in I show that tenure there were economies of scale from conferring power if Doing what an employee wishes raises his probability of remaining with the firm thereby raising the desirability of giving him what he desires. This means that the approach taken in this section, where the derivative of B are given power, In subsequent sections is profits with respect to k appropriate only when k I constrained to is is used to determine whether lie in A or a small interval around zero. thus return to the approach of section one where the firm must choose between two discrete decisions. The setting is result in this section can be given yet another interpretation. The firm's problem that high wages are a very costly are costly because the employee shift is power benefit him What makes it it The first period. for a long time. retains the employee in the second period. Giving This sets it advantage of giving an individual power in clubs, the makes with Power thus leads to deferred (non-pecuniary) compensation. a particularly good form of deferred compensation the deferred compensation. They firm would prefer to helpful in this regard because the decisions that the employee makes the employee happy has been made memberships of retaining the employee in the leave in the second period anyway. compensation towards the future so that an employee power his may way in this is in the past, it is is that, since the decision that impossible for the firm to renege on apart from renegotiable offers of future wages. Another that, unlike other contractual obligations such as lifetime employee only gets to receive the compensation if he remains with the firm.i^ '* Unfortunzitely these studies, like many other empirical studies of power, do not analyze wages Thus one could interpret it does in much of the labor economics literature) and that these regressions as showing that tenure raises wages (which employees whose wages are high "spend" some of these wages on influence. '^Lifetime membership in a club would have this property if and only if potential employers were located in its vicinity. 17 it were not transferable and there were no other An 4. Overlapping Generations Model with Reputations In the previous section, 1 because doing so makes him literature on power showed that a valuable individual employee might be given power in organizations, this is consistent with the empirical important to realize that much of power is it While to depart. less likely really is group power rather than individual power. Thus, designing engineers have a firms where the product that changed often. Similarly, Crozier reports that maintenance workers as a group had power and is sold is in the factory distinction between individuals and groups some sense the B as constituted of groups of people with similar tastes. This members not. The they all result remain of one group acquire a lot of specific would then be that designers, in the company. The only have both old and young workers in many power in certain he studied. In v's if lot of human is One can immaterial. group of people can capital while members think of A differ in their in the other do example, are given a great deal of leeway so that for difficulty with this interpretation of their groups. As is that firms tend to Kreps (1990), the existence of such in overlapping generations would, under certain conditions, allow the firm to establish a reputation for paying high wages to old workers who belong to valuable groups. This would then avoid the need to give power to these groups. In this section I consider an overlapping generations model with reputations that Klein and Leffier's (1981) model of product markets. to receive high power) wages (or power) in the past. in the future if In other words, the In this past. still I show if payment of high wages that, in the context of this model, giving it is is (or the granting of power) builds destroyed so that employees expect power to groups whose members are valuable profits. essence of the argument group power than model members of a group expect the firm has stopped paying high wages (or taken power away) in the has the potential to raise The is that it is easier to maintain a reputation for giving the valuable to maintain a reputation for giving them high wages. In other words, the incentives to deviate are smaller in the case of a reputational equilibrium with power. is I that deviations in which the firm cuts wages raise the current profits of the firm. will based on they have always received such high wages (or a reputation for doing so in the future. But, this reputation low wages (or no power) is show that deviations where power is taken away from the valuable employees 18 The reason By contrast, may actually lowers current profits. The section divided in two subsections. is The first deals with reputational wages in the absence of power considerations while the second incorporates power explicitly. Wages 4.1. an Overlapping Generations Model in consider a setting where employees in two groups I employee in group j provides a value A employment each employee has an outside option which z^. is With probability (1 two periods. young) is In the first period of his old). is Each worth z to him in present value. In the a probability / that the employee receives an outside offer whose value — is him only m. /) the employee's outside opportunity pays v\ > and Z2 + vl pv^ so that the firm would keep both types of employees in 22 in On — the second period and z the other hand, (1 The for assume that I them is work employment (when he v\ in the first period of his while he provides a value Vj in the second period (when he second period, there B and - I assume f)i^2 pz2 in the ^ j ^^ all periods = B ^, if it (15) could commit itself to first. also that -m)> v^- or Z2 /v^ < Z2 - {1 - f)m (16) inequality in (16) implies that, ex post the firm prefers to pay the employee the lower rather than 22- This is profitable even though setting the stays only with probability (1 in this discrete — paying wage equal to wage m m implies that the employee /) rather than with probability one. Condition (16) is necessary model to ensure that second period wages without commitment or reputation are inefficiently low. A reputational equilibrium employees of type wages. _; in the its firm pays a high wage, W2 to stay in spite of the low wages because they believe that they too will wages when they are high wages to The the following structure. second period and thereby convinces young employees to accept low The young employees receive high heis old employees, old. it On the other hand, loses its reputation if the firm ever deviates from paying and the young employees come to believe that they will receive low wages in the second period. The firm hcis nothing to gain from paying wages between equilibria with wages higher than 22 are m and 22 in the second period and even harder to sustain than equilibria with second period 19 wages of pays 22. thus investigate only whether there I employees 22 to its — accept a wage of 2 in the pz^. If a reputational equilibrium such that the firm is By doing second period. so, it convinces such an equilibrium exists, and there is its young employees to number of no growth in the employees, the firm's expected present discounted value of profits starting at an arbitrary date ^ VJ+VJ- Z-[l- ^r P)Z2 l-p were to deviate, the firm would pay If it expect to be paid m in the second its ^ old employees period so they would demand m}^ Young ' employees would then a salary of (2 — p(l — f)m — pf zi). subsequent periods, young employees would demand the same wage so that there would be no In reason to pay the old employees any wage other than m. deviation is The deviating reputation its d = vf - When 22. ^ + - (1 /)(v^2 - sis (19) f and p, is m (18) < Z2 —-{I + satisfied, the firm profits is 22. - p){z2 There then do not Whether is in the now assume '* how even though V2 exceeds satisfied p, exist equilibria of this sort f and unclear. m and equilibrium wages are sufficiently low in practice to rule out What is required is that there be a long period offers are relatively unlikely. Overlapping Generations Model that a discrete decision must be firm must choose in each period between gain a. is (19) by deviating from any reputational equilibrium whose where employees can be exploited because good outside I by maintaining - m) are sufficiently small this condition second period equal m. Power p/22 if reputational equilibria in the real world 4.2. - P)m) + (1 profits in (18) exceed the profits in (17) that the firm receives second period wage in the of profits starting at a ^—^ vi As long The present value therefore . A is In addition, all the a and employees in /?. A who made each If period. the firm chooses work there As a in the first section, the all current employees in in the next period gain a as well. Following Klein and Leffler (1981), I assume that young employees react in the same manner to any deviation no matter small. Given this reaction, deviating by paying is more profitable than deviating and paying a second period salary m m reason is that the higher wage has no effect on the probability of retaining the employee. Because v'^ which would induce exceeds 22 which exceeds m, paying is also more profitable than paying a second period salary below all old employees to leave. between and 22- '^he m m 20 Thus, choosing a generates non-pecuniary benefits A that future young employees in Employees in B utility of employees an employee in B gain lose ^ when any utility when a the firm chooses 0. Z2 — between is 2a. The m — 2a and — 2d z^ this is 22 - I /? is on the in stays with probability (1 if his — 25 wage is — both the /) if his equal to or greater than if (20) consistent with (16), the condition under which the firm prefers to pay Thus, as in Section 2, m making the choice prefers raises the probability that he stays at the firm. will suppose that (20) holds and that the same is true of the analogous condition for B when chosen /vf > My final assumption fv^ < is 22 much more 22 - (1 equilibria, the firm prefers that These equilibria outcome I same group firm always lets the - f)m - 2f0 (1 - /)m - fa it of people 22 - (1 (or 2/9) - f)m - is replaced by d //3 (or /9) (22) worthwhile to pay the second period employees so when they have been now show will (21) when 2a fvf < that they stay with probability one only one of the two periods. his favorite - that neither (20) nor (21) hold In other words, the firm does not find in eff"ect - /)m - 2fa (1 rather than 22 in the absence of power considerations. A A while he stays for sure fv^ > that Instead, current and next This latter choice has no in firm will choose to set the wage equal to Z2 Note that is A. These effects on utility change the wage that the firm must pay to retain current and the previous period. Then, an old employee wage chosen. is second period. In particular, suppose that the firm has chosen a in the different is gain as well.^^ do not gain or period's employees in two periods, as before. What for given their favorite outcome that there exist reputational equilibria where the make the decision. I will also show that, among these which gives power to A. exist as long as young employees believe that whatever group has been given be given their favorite outcome in the subsequent in the current period will also "The model can easily be generalized so that value of past a'a is different from that of current one's One can even distinguish between the value to an old employee and that to a young employee form having had a chosen in the previous period What matters for the analysis is that there be some advantage to future A employees of choosing a in the current period. On the other hand, I also assume that exceed d ^"The condition may appear more stringent because v^ u^ is lower than v^ Wj and that helps to ensure that (21) holds . 21 period. I chosen start by computing the chosen today, and in the past, is when the analogous profits profits that the firm chosen /? is would earn at such an equilibrium expected to be chosen is (22) is wage satisfied, the Expecting to receive this wage was bigger than the previous period and chooses in for the old wage Current period profits in the next period, - in 2(1 A + While current period tt" If, = vf and - p)q p{z2 - 2a) = - p)22 - z + v^ - 2 - (1 vi^ B would remain vf - + (1 - (1 Comparison of A would in Because 22 is " 2a. stay as long as their - la pz2 (23) z - /)(vf - (1 - p)m) + + (1 - (24) pjZ2 (25) (25), total profits are + v^ + vf + + in this period. + 4d - (1 /)vf + 4a - \lz i)v^ /? \2z + (1 - (26) with (27) reveals that the firm always choosing a even when ^ /? in the future, the expression + vf + vf + 4^ - exceeds a as long as v^ + p{\ instead of always choosing a, the firm has chosen = grants power. that maximizes current profits young employees profits in therefore expected to choose ;r^ A a again it I would then be vf (24) in compute I or equal to z Adding employees a was After computing these steady state profits, in perpetuity. consider deviations in which the firm changes the identity of the group to which Suppose the firm chose a Then in the future. if p{\ j))zi + (1 in the past, - p)(l chooses analogous to (26) + /))22 + (1 - /)m] - ^ today and sufficiently higher is is p){\ - /)m] would gain higher steady state is (26) than Uj . (27) profits Once by again, the firm can gain by giving power to the more valuable employee. For there to exist an equilibrium where deviate at the point at which assuming the firm heis it makes A (or its decision. been choosing a and is B) always gets power the firm must not want To analyze the incentives to deviate, expected to continue doing 22 so. I I to start by then compute the current profits from deviating by choosing 0. young employees 22 — 2/3 in in B I suppose that, as a result of this deviation, the will expect the firm to choose again in the next period. Expecting to be paid /? the next period, they are willing to work for z- PZ2in the current period. in B only worth m will stay as long as their still - vf {z2 — d). B wage this, A whose outside With /9). + Z + pz2 + il- m expect to be paid between staying and leaving z The resulting current profits in vf Adding n-^ = vt The (29) and - 2 +d+ wage their if (29) is in the following period between (m (m — is and — d). Similarly, so, conditional a- A - pfz2 - f)m p{l on To be workers must thus receive in the current period, these - d) and (30) are pfzi + p{l - f)m + (1 - f){v\ -m + a) (31) (31), total profits in the period of the deviation are then + vf+{l- f){vt + vf + ) (2 difference between these profits ^2 Assuming that these -m + ^) f){vf staying in the current period, they expect to leave with probability / in the next period. indifferent the equal equation (22) implies that the profit maximizing wage currently young employees in (m — inside the firm equals employees would leave with probability / Given ^ because to leave with probability / because of (22). However, those wages, current profits in activity In A, old a were equal if of utility on the job (as opposed to 2d). Old employees is (28) This expression would be larger than (23) young employees receive only offer ^ is - (1 - /)(d and w^ + ^) - [2z - p{l (2 - /)(d + + (2 - p)(l - f)m] (32) is - f)m - fv^ - 2//3 + no smaller than d, (21) implies that (33) profits in the period of the deviation are smaller f)z2 is negative. than steady state profits when period after the deviation, profits are given by n^. 23 (33) /3) This means that is chosen. In the p It thus follows that, this deviation is if the gap between v^ and vf The reason deviations that power only works in retaining employees when are so unattractive in this setting is who have been powerful in the past and not sufficient to retain those that used to be powerless. Now consider the reputational equilibrium where show that the profits in the period in possible that even in this case this deviation than TT^ Thus, for sufficiently right left hand hand side of this expression is not profitable. The reason + -^n'^ < 1- is I may find Nonetheless, . that n°'^ is lower (34) the present value of profits after the deviation while the 0. If (34) holds, there is an equilibrium This is particularly firm might have traditionally given power to one group of employees When because they were the most valuable. valuable, the firm x^ - This suggests that firms can be stuck giving power to the "wrong" people. A as well. This even though profits are higher when the firm always chooses a. where the firm always chooses change. is n'^'^ -^ from remaining with side represents the profits likely if conditions straightforward to low p n^^ The It is sufficiently high, exceeds if u'^ is its is . always chosen. is which the firm deviates by choosing a equal deviation leads future profits to equal n" which, it more conditions change so that another group becomes more profitable to continue to give power to the traditional group even though giving power to the other group yields higher steady state 5. is applied twice. Deviations that change the it is balance of power for one period lead to defections by those is sufficiently large that n°' exceeds tt", is not profitable. Profits in the period of the deviation are lower than n^ which lower than steady state profits. itself p profits. Concentrated versus Diffuse PoAver In the next two sections arise in the case of I return to issues surrounding individual power. While similar issues group power, it is analytically more convenient the overlapping generations aspect of group power. existed a single decision that appear more equitable is to instead of choosing either which offers some is taken following either compromise and a which give each favors only A, or Up A to this point, or B's wishes. employee a /?, I An little bit have assumed that there alternative, of what he which might desires. Thus, the firm might prefer a third alternative 7 benefits for both employees. In particular, 24 to deal with these by ignoring I suppose that this alternative leads B to get utility equivalent to 7^ A units of cash while gets the equivalent of 7^*. In addition, I assume that 7'^ + 7^ > max(Qj) so that, in addition to being equitable, alternative 7 (35) efficient; is it many provides at least as non-pecuniary benefits as the other alternatives. One To make interpretation of this third alternative suppose that employees this concrete though they have sometimes is A and A what B By 7^ + 7^ is common In a with Section or 2, in separate physical locations Alternative 7 then lets each employee o; . can be thought of as But, because getting what one desires /3. I assume that the firm operates just for two periods. section 3, however, in assuming that the employees outside offer in period one second period is worth either m or salaries in the first period it would choose 7. It offer in the Z2 — y^ while B's equals 22 and similarly for — sometimes affected is than getting one's wishes at one's main location, visits rarely is less valuable larger than either work a exceeds 7^ because A he desires in both locations. Thus one as involving "local" decision making. contreist, alternative by B's typical environment. Similarly, ^ would exceed 7 at the location it typically each other's locations. visit determine some local characteristics of the job. giving to think of 7^. A's 22- If is the firm can precommit would then let its z while the second period A's second period payment equal period payment would then equal 2 first worth follow I — pz2 — (1 + p)t^ B. The resulting profits would equal E E p^'-%i + in-2z (36) t=l,2j=A,B which, given (35) exceeds the profits the firm can obtain under precommitment with either a or Given the is earlier development, profitable for the firm The period. firm finds it should be clear that the question of whether the 7 alternative depends on whether it profitable to keep fvt > which is analogous to (20). Because 7^ equivalent condition for B is /?. it A induces the firm to keep the employees in the second with probability one by paying him Z2-{1- f)m is 22 ~ "; 2/7^ smaller than a, it is if (37) possible that (37) as weU as the violated even though fv^> Z2-{l-f)m-2fa 25 (38) and the equivalent condition for B are satisfied. will I assume what this in follows. It implies that the choice of 7 induces each employee to leave with probability / in the second period whereas the choice of a or ^ would keep one employee with The choice m— 7^. This wage of 7 then leads to a in of m probability one. — turn means that the required -y'^ A for A period wage for first replaced by y^ while B's wage can be computed analogously. is the second period while B's equals in The a given by (30) with is resulting present value of profits then E = 'r-r + ('^i f ) + />(! - + f){^2 1') - 2{z - pfz,) (39) i=A,B It is (38) is straightforward to establish using similar reasoning that profits from choosing a satisfied equal vf + vf These differ from (39) and a gain of pfv2 consistent with (1 loss of + p)a +a+ in that Note that profits is (1 v^ - /)vf ) - (1 + /?(! — is (2^ - pfz^) when a (40) chosen. is larger than the latter, +7 /))(7 ), it is The result and because is (35) a is possible for the expression in (39). from choosing sufficiently high value of +Q+ stays with probability one being larger than by (40) as long as one switches the a A p{v^ pf Z2. Because the former exceed the expression in (40) to if when A fi when and relative to B the condition analogous to (38) superscripts and replaces a by vf implies that giving power to A is satisfied are given 0. Thus, once again a is more attractive even smaller than $. This section has demonstrated that A even the power in interprets 7 as giving local decision making power. thought of as the only available option Employees If 7 is interpreted in this way, it all efficiency better to implement 7. This can be viewed as a benefit of organizational size it is h. possible for the firm to be better off concentrating though from an equity viewpoint as well as from a complete-contracts viewpoint or activity it is if one should be for single activity firms that specialize in either activity o in these single activity firms still have to interact with people carrying out the other activity. But, in the case of independent firms, the identity of the firm with which a particular employee interacts together. It out activity then makes 6 in little is likely to change over time since nothing binds two particular firms sense to give A control over a particular firm. 26 the environment of the employee carrying The the choice of same firm a thus an attractive option only is is the two activities are carried out within or between firms subject to a long term contract. a {A) be confident of the where a if more environment with which he 6 profitable than 7, we expect Only then can the employee doing will interact in the future. So, in the case same firms to either carry out activity 6 within the firm or at least to have a long term contract with a firm that does A that one can thereby extend the environments over which The benefit of doing so is has power and thereby retain him in b. the firm. My demonstration that a can be more attractive than 7 even when 7 would be chosen with complete contracts does not mean that the incompleteness of contracts specifying future compensation creates a bias towards the concentration of power. Suppose in particular that (37) and the analogous condition for B are satisfied so that both employees stay with probability one 7 is from giving power to those whose departure is if chosen. Then, profits from 7 equal j=A,B which exceeds (39) and (40) as long as (35) As is satisfied. in the earlier sections, the firm benefits thereby averted. The choice of the equitable alternative 7 to retain employees than a choice that concentrates 7 become more attractive 6. Power The with the to raise one's if more employees are is thus less attractive if it is less likely power more. By the same token, the choice of likely to stay cis a result. income decisions considered so far gave individuals non-pecuniary benefits. In this section Ccise enacted, put of pecuniary benefits. them in The idea is I deal that individuals prefer decisions which, once a position to ensure that their salary is high. A rather obvious example is the case where individuals seek promotion for themselves. Almost equally obvious are the cases where individuals seek to have the firm embark on an investment project that makes particular use their expertise. Quite generally, decisions different individuals are in the future. seniority, I will assume made by the firm affect how critical to the firm's success Thus, instead of assuming that the in this section that of v's depend only on they depend also on the decision taken at the beginning of the first period. 27 The to individual employee gains from having the firm increase his v because that leads the firm pay him more. The firm reacts employee as he becomes more way because in this becomes more concerned about it Thus, the firm's choice of project involves critical. the probability that one employee stays. I show that the employee whose internal value In this section, I a B to the firm stays equal to Vj implies that, in the second period, the value of On . second period to vf while that of A A this stays equal to v^. commit period while the second implies that they would receive The outcome with commitment It should thus choose and should choose or probability one. ^3 a (15) Z2 to in the raises the value of B in the and (16) both hold. The both employees in the second absence of power considerations. straightforward. Since both employees stay with probability if otherwise. Without commitment, The it is once again firm will indeed choose to keep the employee fx/s If m /? critical whether choosing leads the firm to pay an employee Z2 in the second period and thereby keep /3 which assume that the choice any event, the firm should choose the project which increases the value of an employee by in more. is assume I paying itself to I to the firm rises to v^ while the value of the other hand, the choice of implies that the firm would like to first first demonstrate high will get the firm to raise his value is modify the model of the previous section as follows. of a I further so as to raise his wage.^^ still one part raising Since the firm extracts more surplus from A, the firm thus tends to prefer the project which raises A's internal value. In this section formally. in losing the I will the > Z2 - (1 - f)m, j whose value increases him with if = A,B (42) assume from now on. first period off'er is known to equal 2, eis I have I assumed until now, the firm sets a period wage that ensures that both employees stay with probability one in this period. This '^'Another reason the firm will tend to raise the employee's wage is that the choice of project conveys information to the labor market. Choosing the project which uses intensely the skills of A suggests that A is particularly adept (see Waldman (1984). I ignore this in what follows. ^^ In a model with more stages, this increase in value also leads directly to increases in power. This suggests that a multiperiod version of this model would show entrenchment in the sense that individuals with large power (due to high initial v's) would gain further power (because the firm would make their v's grow. For evidence of entrenchment, see Morck, Shieifer and Vishny (1989). 28 implies different wages depending on whether the firm chooses A implies that expects his wage period wage equals {z first — the second period to equal in pz^)- he must be paid a wage equal to By — [z B contrast, — pfz2 a or /3. He Z2- will m can only expect the firm chooses a, (42) If thus stay as long as his second period so that in the p{l — f)rn] in the first period. p {v^ + (1 The resulting profits are vf + vf An -2z + Thus a is more profitable in curly brackets higher. Once as well. the edge more fz2 } fz2 } if - 4) is the i^l same choosing a in the case of commitment. values increase by the + + vf -2z + p[v^ + {l- f)4 + {{^S The term /)vf from choosing ^ equal identical derivation implies that the profits v^ - - ^2)} + - f[^2 t^f ) > (43) as the expression in (41) which equals the net gain from Thus (43) says that, in the case same amount, the firm should choose a is where both employees' A's second period value again, the firm favors the employee with higher internal value. Even though the decisions when A is raise the value of more valuable because is initiaOy similar both individuals by the same amount, a has raises the probability that it The reason is A stays and A produces value. Note also that willingness to pay plays no role in this decision. Because having the firm make the choice that favors a particular employee raises that employee's future wages to z^, employees are willing to give willing to give One slight up some of their first period drawback of the model considered so maximum is — is that the employee surplus from the decisions that favors them. This benefit for the employee far exchange. in up the same amount, namely p[\ — f){z2 each employee's wage by the whose main wage is m). that, in equilibrium, the firm does reduce is particularly implausible in the case of decisions that they raise his wage. period offer received by the employees is known 29 Thus, employees get no willing to pay. nothing in this model, as well as in the models of earlier sections, first However, both employees are to the firm. The reason is that I the employee gains have assumed that the I now will relax this assumption so that, for reasons essentially identical to those in Roberts (1988), the employees are Thus, as in their increase their strictly better off paper, employees would find own the firm will tend to favor the employee whose period value) is wages, it when that, both the firm makes with probability — (1 employees, however, first know their outside offer its it What equals (z + this profitable. I of (z not necessary for the conclusions, Here which obviously requires that A be a present value oi z chooses a when (41) now by period. + + z), the firm I (1 period that each employees will let A The earlier be larger. commit The — A) to a path working at then ensures that employees stay in if they are (if still the firm offers a present value of focus on the special case where the former is + prj{ -z)> X, strictly positive. j = A, B is z. more (44) The consequence x to both employees no matter what of (44) its decision. is that the firm pays Thus the firm once again satisfied. contrast this I I first thus assume that A(t;{ I is the effect of this modification in the case where the firm can By promising a present value is determines from the beginning. ^^ both period whereas they stay only with probability While it i) with probability A. of future wages so that both employees earn zi in the second period the firm). has the implication that does know it equal to zero. is when decision and offers. A) whereas model thus involves the special case where A Consider will entertain also period value (as opposed to only the second first does not know the employees' outside offer equals z I in influence activities to high. now assume I firm chooses their favored outcome. advantageous to engage it The modification internal value. when the Milgrom and precommitment outcome with the equilibrium when wages are set period concentrate on the case where X{v{+p{lThis inequality implies that, additional i to the employee who f)vi in the is + pfz:i-z)<x, j = A,B absence of commitment, the firm not favored by the decision. The is (45) better of not paying the reeison this is not worthwhile ^•'Milgrom and Roberts (1988) assume that the employees learn the value of their outside offers after the firm has made its In this case, the firm could extract the expected surplus that employees get from the decision by charging them for adopting the course that they prefer. Indeed, if both employees assign the same value to the decision that they favor, the firm can extract this surplus in full by conducting an auction. decision. ) in spite of (44) is commitment, the firm only that, without probability /. Indeed, (45) is gets V2 in the second period with consistent with (44) as long as Z2 is sufficiently smaller Inequality (42) implies that the firm does pay Z2 to the employee that than Vj. favored by the decision is so that this employee stays with probability one in the second period. Condition (44) then implies By that the firm will pay the additional x to this employee. employee can only expect a present value of 2 contrast, (45) implies that the other he stays at the firm. if Therefore, profits from choosing a equal + Vi pvs while those from choosing difference A= The term of - X){vf + p{l - /)vf - z + pfzz] z-x+{l- A)[vf + p{l - f)vt - z + pfzi] x + {I between these two expressions X{vt ) a with commitment. The A is - vf + p{v^ -vt-{vl-v^)} + p{f+il- in curly brackets incentive to choose - z equal +pvf - vf The - once again, the expression is, last term says that, just as because v^ exceeds Kj . in (41) that Finally, the first additional incentive to favor the employee that provides higher term first commitment first desirable to choose A when The reason is for this v^ exceeds Vj creates an in (47) says that there period value. Thus, even both (41) and (43) are zero so that both employees provide the same values it is (46) determines the desirability the lack of in (43), - vf f)X){vt in the because A is when . the following. Choosing B an second period, a leads the firm to pay A a higher wage in the period which, in turn, ensures that he stays with probability one in that period. This than doing the same for is more valuable in the first period. To gain is better intuition for the result, consider the choice between two investment project which use either A's or S's talents intensely once they are completed. Choosing yl's project induces the firm to keep the gestation period. This benefits A in the form of a higher to choosing the project that uses B's talents because period. 31 A is A for sure during salary. It also benefits the firm relative more valuable than B during the waiting ) A Notice finally that, as long as makes him He strictly better off. that his outside offer in the + now sure to collect x instead of receiving this only in the event period pays {z first having an expectation of [z is does not have to pay for the decision directly, the choice of a + x). Thus But, the question remains whether the firm him for the decision. If A for B one ignored The reason extract anything from A. is and that charging anything true, the firm is probability one. B Because is is also benefits much if the expression in (46), B his own amount A is would be vf + pv2 is that equal to A 2. can if his if own in principle Suppose in exceed The last that A would pvf be unwilling to bid anything is + outside offer equals z A If it is, a x. - for + 2) A x. But, would he be would not bid anything even for sure Suppose that A Then, attractive. in this expression ensures that The reason he does not bid of a. 7. term + He would be not willing to pay anything). is if B and v^ /(I - is a even though he a because both equal fj is /3. while 22 zero so that, with equal to - vf X){vf can exceed x for were to bid x for addition that the expression in (41) precommitment, both decisions are equally A(vf he is x. wage were z? The answer depends on whether his alternative larger than x or not. A= willing to pay for offer equals only z (otherwise that would be unnecessary: the firm would implement now show But, precisely A. from ^, the firm might be able to gain more by having the two not willing to bid anything willing to bid a positive I to * willing to bid this A amount between zero and x and guaranteeing that the employee stays with better off paying x employees bid against each other. The most that Obviously by charging to employee with probability this A would not be able clear that the firm is it /?, x). able to extract this surplus from is implementing a leads to the departure of because (44) + Xx) to being equal to {z the present value of his salary goes from in spite of (45). is Thus strictly better off that the firm has no better choice available to it is possible with the choice it. Conclusions This paper ha^ tackled the question of with what employees do ex ante ^*By for gets in order to achieve if the employee does not yet know having the firm pick a. contrast, exchange who his outside offer 32 power ex post. power. when the It has not dealt explicitly Because power flows to those who decision is made, he is wiMing to pay (1 - X)x in supply the firm with the largest value relative to that provided by newcomers, seeking power should try to maximize this internal value. One, which is less desirable, involves category fall activities information First, There are two ways of achieving incumbent employee can obtain The for the firm. this. human other, which reducing any potential replacement's productivity on the job. In this make essential (such as the purposeful misplacement of repair manuals) that difficult to obtain. Thus, the ways. follows that those desirable from the firm's perspective, involves the accumulation of specific capital that raises the profits the is it efforts of employees to change their informational environment can be seen an employee might acquire information to improve two in his decisions thereby raising his internal value relative to that of an uninformed outsider. Second, he may reduce the information available to outsiders so the employee's value increases without a corresponding benefit to the firm. Similarly resistance to a change in formal information systems proposed by top management can be seen in these two ways: as an effort to maintain high personal productivity and avoid unprofitable disruptions or as a prevention of changes that facilitate the acquisition of information by potential new employees. Thus, the paper made consistent with the evidence of Markus and Golden Triangle company derailed changes divisional accountants in the that would have is information systems information more easily available to the corporate accounting change would have taken power away from the divisional profit commit to granting This maximizing ex post, nonetheless distort employee actions away from profit maximization ex ante. the extent to which the firm can staff. staff. These examples show that, while the granting of power may be may in Pfeffer (1983) that The issue is it then power only to those employees whose value has risen for legitimate reasons. The firm may, for example, establish an ethical code of conduct and commit itself to firing to the firm. This show to be useful to It even is if is employees who violate this code regardless of the value that they provide somewhat combat important to similar to the bureaucratic rules that influence activities. stress, power were allocated Milgrom and Roberts (1988) however, that these distortions of employee behavior would arise efficiently. For example, they would arise in my model even if firms could determine in advance a schedule that determines each employee's wage as a function of that employee's internal value. This is so even though, in this case, power would be allocated to those 33 most willing to (no matter The pay for The reason it. how they do so) is that employees who raise their internal value in this setting can expect to receive a higher income. money use of power as opposed to for the purpose of compensating important employees does not create particularly strong incentives for raising one's value at the expense of firm efficiency. One should thus interpret this paper as showing that the composition of employee compensation and, in particular, the extent to which they are compensated with power or with be profit maximizing. That does not mean that the firm could not do better if it money may could prevent individuals from taking actions that raise their total compensation without contributing to profits (See Holmstrom and Milgrom (1990) Another extension of this examples of such for activities). work that would seem worthwhile is to consider individuals will be given authority in addition to power. In this paper on the ability of a person or that favors. it itself is for To analyze made by is this question I What have not dealt with are cases where the authority I delegated. formal authority; one can be in wishes without being officially in plays no role. all, Thus, it will i.e. have considered a very simple model where the decision As has been noted repeatedly (See Thompson After have focused on power, group to have top management implement the organizational change the owner of the firm. making decisions I when groups and (1956)), one's power need not coincide with a position to have a particular decision charge of making the decision. That is made according to one's not to say that authority those in authority act as the referees between the competing interests. sometimes prove administratively expedient to put in charge of making decisions those individuals whose desires will ultimately prevail. But the delegation of authority even to the employees whose wishes will prevail might be problematic. In particular, the firm must ensure that those in authority do not turn around and choose that decision for which employees are willing to pay them the most. this form are this type One between illegal. In practice this may not be a serious problem because overt side payments of However, a more thorough analysis of the delegation problem in models of seems warranted. crucial question that remains my model and open concerns the the view that power than profit maximization. My hope is is feasibility of distinguishing empirically allocated to particular groups for reasons other that this paper will give impetus to a search for facts on 34 power distribution that could distinguish between these two views. For instance, the model clarifies who control that the economic view of the firm important resources get a lot of is not contradicted by evidence that individuals power. But, it would be contradicted employee can be replaced by an outsider played no role. In if the ease with which the other words, evidence that showed power flowing to readily replaceable individuals with access to important resources would be inconsistent with the model. 35 8. References Dahl, Robert A.: "The Concept of Power," Behavioral Science, Dalton, Melville: Men Who Manage, New 2, 1957, 201-215. York, Wiley, 1959. Emerson, Richard: "Power-Dependence Relations," American sociological Review, 27, 1962, 31-41. Fombrun, Charles: "Attributions of Power across a Social Network," Human Relations, 36, 1983, 493-508. Groves, Theodore and J. Ledyard: "Optimal Allocation of Public Goods: Free Rider Problem," Econometrfca, 45, 1977, 783-810. Hall, A Solution to the Robert E. and Edward P. Lazear: "The Excess Sensitivity of Layoffs and Quits Demand," Journal of Labor Economics, 2, 1984, 233-57. to Hickson, D.J., C.R. Minings, C.A. Lee, R.E. Schneck, and J.M. Pennings: "A Strategic Contingencies Theory of Intraorganizational Power," Administrative Science Quarterly, 16, 1971, 216-229. Hinings, C.R., D.J. Hickson, J.M. Pennings and R.E. Schenck: "Structural Conditions and Intraorganizational Power," Administrative Science Quarterly, 19, 1974, 22-44. Holmstrom, Bengt and Paul Milgrom: "Multi-Task Principal Agent Analyses: Incentive Contracts, Asset Ownership and Job Design," mimeo, 1990. Jensen, Michael C. and William H. Meckling: "Specific and General Knowledge, and Organizational Structure," in Lars Werin and Hans Wijkander eds. Contract Economics, Oxford: Basil Blackwell, 1992. "Assessing the Political Landscape: Structure, Cognition and Power in Organizations," Administrative Science Quarteriy, 35, 1990, 342-69. Krackhardt, David: Benjamin and Keith B. Leffler: "The Role of market Forces in Assuring Contractual Performance," Journal of Political Economy, 89, August 1981, 615-41. Klein, Kreps, David: "Corporate Culture and Economic Theory," in J. Alte and K. Shepsle eds. Rational Perspectives on Positive Political Economy, Cambridge, Cambridge University Press, 1990, 90-143. March, James 662-678. "The Business Firm as a Political Coalition," Journal of Politics, 24, 1962, Reprinted in James G. March Decisions and Organizations, New York: Basil G.: Blackwell, 1988. Markus, M. Lynne and Jeffrey Pfeffer: "Power and the Design and Implementation of Accounting and Control Systems," Accounting, Organizations and Society, 8, 1983, 205-218. Milgrom, Paul and D. John Roberts: "An Economic Approach to Influence Activities ganizations," American Journal of Sociology, 94 (Supplement), 1988, S154-S179. in Or- Morck, Randall, Andrei Shleifer and Robert W. Vishny: "Do Managerial Objectives Drive Bad Acquisitions?," mimeo, 1989. Perrow, Charles: "Departmental Power and Perspectives in Industrial Firms" in Mayer E. Zald ed. Power in Organizations, Nashville: Vanderbilt University Press, 1970. Pfeffer, Jeffrey: Power in Organizations, Marshfield MA:Pitman, 1981. and Gerald Salancik: "The External Control of Organizations: Dependence Perspective," New York: Harper and Row, 1978. Pfeffer, Jeffrey D.: "Authority and Power of Sociology, 62, 1956, 290-301. Thompson, James Tushman, Michael L. decision Making: in "Identical" A Resource Organizations," American Journal and Elaine Romanelli: "Uncertainty, Social Location and Influence in A Sociometric Analysis," Management Science 29, January 1983, 12-23. Waldman, Michael: "Job Assignments, Signaling, and EflRciency," 15, 1984, 255-67. 37 Rand Journal of Economics, Date Due MIT LIBRARIES 3 9080 02237 3622