Something to prove: reputation in teams.

Agents work for their own reputations when young but for their firms' when old. An individual with an established reputation cannot credibly commit to exerting effort when working alone. However, by hiring and working with juniors of uncertain reputation, seniors will have incentives to exert effort. Incentives for young agents arise from a concern for their own reputation (and the opportunity to take over the firm), whereas older agents work for the reputation of their firms (and the opportunity to sell out to juniors). Thus, the article explains the choice to work in teams. It also exemplifies how type uncertainty in reputation models may be endogenously and strategically introduced.

1. Introduction

* This article aims at a better understanding of professional services organizations, such as law firms, consulting firms, medical practices, architects and so on. The starting point for the analysis is the assumption that in these industries reputational concerns are crucial. Services cannot be inspected prior to sale and it is impossible to make fully contingent descriptions of the product or offer money-back guarantees, so in effect, price is determined in advance and depends on clients' expectations concerning the service provider's ability and effort. (1) Developing and maintaining reputation both at the level of the firm and the individual are crucial. Indeed, previous literature has suggested that the very existence of a firm might arise as a means to manage reputation and that careers are designed to take into account reputational considerations. (2)

I outline a framework in which a young agent is motivated by concerns for her own reputation. However, once she is established, her own reputation is not at risk and cannot act as a motivation. She therefore chooses to hire and work with a junior whose ability is uncertain and, because only combined outcomes are observed, her actions will affect the reputation of her junior. She cares about the reputation of her junior because she controls the client list that the junior needs. She is able to provide incentives for the junior by committing in advance to a price at which she will allow the junior access to the client list (or equivalently the wage that she will pay to the junior if she retains him). Thus, in particular, young agents are motivated by concerns for their own reputations and old, successful agents are motivated by the reputations of the firms that they own (or, more specifically, by concern for the reputation of their employees). Throughout her life, an agent has something to prove, initially that she is competent, later in life that the firm that she owns is competent.

This last observation rings true with casual observation and my experiences and conversations, with lawyers, consultants and other professionals. However, while this article owes much to existing models of reputation, such as recent work by Tadelis (1999, 2002), they are unable to account for such a statement. One stream of literature is built around an assumption that customers cannot distinguish individuals from the firms that they own (Tadelis, 2002; Mailath and Samuelson, 2001), (3) while another branch of the literature assumes that groups are very large (Tirole, 1996; Levin, 2001) so that a single individual's actions can have no effect on the collective reputation. The only way it is possible for agents to have concerns for their own reputations when young, and for their firm reputations when old, is tautologically, to suppose that their actions are able to affect both their own reputations and that of their firms. Thinking of the firm as a small team is a natural way to do this and again accords with casual observation. Many professional services firms are small, and many large professional services firms are organized around small teams in which much of the reputation appears to reside. This is evidenced by the observations that, within large law firms, compensation even of partners is often tied to the performance of the individual departments, and law-firm directories will rank a firm with respect to different specialization (and a typical specialization in a law firm will have a fairly small number of partners). (4) In another industry, investment banking, the frequent observation that small teams move as teams is also revealing.

There are only a few articles that focus on reputation in teams. In particular, in an elegant article, Anderson and Smith (2002) similarly focus on the role of team production in obscuring individual contributions. Their focus is on individual decisions of who to work with--the matching problem--rather than on the choice of team production as a credible commitment to exert effort in production. (5)

This article has a somewhat tangential contribution to consideration of the theory of the firm and the firm as a bearer of reputation, introduced in Kreps (1990) and rejuvenated more recently in Tadelis (1999, 2002). Though much of the discussion both above and below is couched in terms of the firm and its reputation, at heart, the model considers a choice of technology: whether to work alone or whether to work as part of a team, abstracting from all other considerations but the observation that teamwork obscures individual contributions. To the extent that it is natural to think of team members as working in the same firm, the article can be seen as arguing that reputational considerations might affect the technology choice and thereby affect firm or organizational design.

With respect to the application of organizational design for professional services firms, a number of other interesting features arise. (6) In particular, it is shown that teamwork can create rather than dampen incentives because mixed teams of partners and juniors can provide incentives for partners. (7) Moreover, in this framework for an up-or-out mechanism to be effective, promotion must be to partnership--an empirical feature that previous literature has not much addressed, but that arises very naturally in the context of the overlapping generations framework of the central model. (8)

A similar feature arises in Morrison and William (2004), who consider the role of partnerships in ensuring that seniors mentor their juniors. In their article, a partner has an incentive to mentor a junior as only a good junior who has been mentored would be willing to buy a partnership share in the firm; only in this case will the junior be able to maintain the firm's collective reputation and the value of her partnership stake. Thus, the article echoes a number of the themes highlighted here, in particular, a senior is motivated to work because this affects her ability to sell the firm to the junior. However, the issue at the heart of Morrison and Wilhelm (2004) is the incentive for partners to mentor juniors, an incentive that naturally leads seniors and juniors to work together; and they consider no incentive issues for the juniors. Here instead, seniors and juniors are identical and face similar decisions--it is not the case that the senior affects the productive capability of the junior, as in Morrison and Wilhelm (2004), but rather can affect how this is perceived. Moreover, we show that the promotion structure described can be successful in providing incentives for the junior as well as for the senior. Further, their model has no role for individual reputations.

Finally, whereas some have argued that law-firm partnerships might exist to diversify risks for individuals, in practice, one sees groups of lawyers working in the same or related fields. An established argument for this phenomenon is that it allows for mutual monitoring among the partners in a law firm (Alchian and Demsetz, 1972). Note, in addition, that a more homogeneous firm makes it more difficult for clients to identify individual contributions of seniors and juniors and so enables the reputational mechanism highlighted in this article to operate and seniors to credibly commit to exerting effort.

Beyond enriching our understanding of professional services organizations, the article also makes an important theoretical contribution. Any reputational concern relies almost tautologically on current actions affecting future beliefs or continuation equilibria in repeated-game notions of reputation (see Klein and Leffler (1981) or, more closely related to this article, Cremer (1986) and Bar-Isaac (2004a), which contains a lengthy discussion of notions of reputation). In particular, this implies that there must be sufficient uncertainty for reputational concerns to arise because, if beliefs are sure, then little can be done to change them. Loosely, for an agent to be motivated by reputation, she needs to have something to prove. Previous literature has suggested the possibility that this might lead principals to slow down the release of information to sustain reputational concerns (Jeon, 1996, for example) or the important role that exogenous replenishment of type uncertainty might play (Holmstrom, 1999; Cripps, Mailath, and Samuelson, 2004; and the references therein). (9) This is the first article, however, to suggest an approach to strategically introduce a new type of uncertainty and allow an agent, thereby, to credibly commit to exerting effort.