LEXSEE 5 s

Southern California Interdisciplinary Law Journal

Summer, 1997

5 S. Cal. Interdis. L.J. 375

Skewing The Results: The Role of Lawyers in Transmitting Legal Rules

Donald C. Langevoort *

and Robert K. Rasmussen **

* Lee S. & Charles A. Speir Professor, Vanderbilt Law School.

** Professor of Law, Vanderbilt Law School. The authors wish to thank Richard Painter, Jim Cox, and Geoff Miller for comments on earlier drafts.

INTRODUCTION

Business lawyers are often caricatured as worry-warts and nay-sayers, obsessing on risk, burdening their clients' dealings. While this popular portrayal is clearly misleading, n1 it rests on an intuition that we propose to take seriously: that the overstatement of legal risk may be a natural by-product of professional self-interest and self-definition.

The possibility that lawyers are often tempted to act in a manner that is self-serving, rather than in their clients' best interests, is well recognized. Many of the bar's rules of professional responsibility are designed to discourage breach of the duty of loyalty, and the scholarly literature abounds with studies of the temptations to cheat their clients that lawyers face in such settings as the conduct of litigation, n2 the settlement process, n3 fees and billing, n4 and the like. Curiously, however, little if any serious attention has been given to the possibility that self-serving behavior will occur consciously or unconsciously in one of the most basic of the lawyer's roles, that of giving legal advice to a client. The counselling function has received increasing attention [*376] recently, particularly in the law and economics literature. But the question most scholars have chosen to explore is whether the extensive provision of legal advice is socially optimal - how advice is allocated or whether society might not be better off with less of it. n5 The tension posed is between client interest and public interest. The same duality can be found in studies dealing with ethics and professional responsibility. n6 The implicit assumption throughout this literature is that the advice provided to clients, on average, is objective, unbiased and accurate. n7

In this article, we explore the contrary possibility. The question of whether lawyers systematically tend to overstate legal risks is an empirical one, of course, and we do not offer any data one way or the other. Instead, we will approach the problem theoretically (and thus tentatively), with a view toward generating some testable hypotheses about the behavior of lawyers. We will see whether a plausible claim can be made that the popular intuition, however overstated, is based on at least some grains of truth.

If so, the implications would be profound. To the extent that lawyers, on average, do tend to overstate legal risk, then the legal profession should presumably recognize it as a problem of fiduciary irresponsibility and try to discourage it. Much more fundamental, however, are some fascinating questions about the diffusion of legal knowledge in society. If overstatement of risk is endemic, then the information economic actors act upon may well be different from the law as objectively understood, and the efficiency of the "received" law thereby called into question. Excessive caution may be the norm. [*377] The idea that there is an "acoustic separation" between the law as generated and as received by the general population has been advanced. n8 What has not been considered is that such a separation may be in part the product of a filtration bias.

This Article looks at the possibility of such a filtration bias in business law settings. We will concentrate on the sort of advice sizable law firms are asked to provide to their clients. We have chosen this context partly because it is the setting with which we are most familiar. More importantly, it is probably the setting in which elite lawyers are most widely employed in an advisory capacity, and where legal advice seems to be taken quite seriously. Whether our conclusions can be generalized to other contexts - a family law practice, for instance, or a legal services setting - is something we will leave to others.

Why would a lawyer knowingly overstate a legal risk to a client? Obviously, answering this question is the linchpin of our analysis. We posit two main reasons. First, it is often wealth-maximizing. In general, the more legal risk there is, the more necessary and valuable legal services are. True, if a lawyer is asked whether a proposed course of action is lawful and says no, the risks are too high, then the lawyer gains nothing but the fee from rendering the advice. (The same is true, of course, when the answer is yes, there is no risk at all). In the middle ground, however, the lawyer can make the claim that the course of action - or something like it - is possible but only with the careful management of legal risk. In other words, "You can do it, but you'll need my help." The more credible the risk, the more resources that are justified in terms of both legal research and transactional assistance. Now, the wealth-maximizing incentive is palpable.

The second explanation is quite different, but often reinforcing. The nature of the attorney-client relationship is such that clients typically have a hard time measuring how well their lawyers have served them in rendering advice. n9 Clients lack the information and expertise to make such judgments (presumably, that is why they retained a lawyer in the first place). Consequently, if a transaction is foregone [*378] because of the lawyer's warnings, there is usually no way of discovering whether the lawyer was excessively cautious or not. The same is true if the transaction goes forward, layered with excessive and costly precaution, and encounters no trouble or is upheld. The lawyers appear to have done a good job. But one situation where the client will be able to make a rough assessment of the quality of the advice is when the lawyer has given the go ahead (i.e., sufficiently minimized the risk) but the transaction is later deemed unlawful. Here, we suspect, the lawyer can reasonably fear that he or she will pay a substantial reputational (and perhaps financial) penalty - with the immediate client, and probably potential ones as well. n10 And rarely will there be any counterbalancing reward for more accurate calibration. In other words, there is an asymmetry in the observability of good and bad advice that leads naturally to an incentive to err on the side of caution. n11

Part I of this Article employs conventional economic analysis, treating the lawyer-client interaction in the counselling setting as a straightforward principal-agent problem. To this end, we consider whether common contractual or market-based mechanisms are likely to dampen the incentive to overstate risk. As we shall see, the standard method for compensating business lawyers - the hourly fee - actually accentuates the bias. Nor is there reason to believe that law firm structure, competitive influences or client monitoring (i.e., the increased use of in-house legal staffs) will offer a complete check.

Part II proceeds differently. Many lawyers deny that they have observed any significant incidence of consciously biased legal advice. Perhaps the ethos of professional responsibility sufficiently counters the incentive to cheat. But we draw from a variety of the social sciences to create a montage of reasons why lawyers might systematically [*379] overstate legal risk while at the same time believing that they are acting loyally. One sociological explanation is that lawyers' norms somehow operate to legitimate excessive caution, declaring it to be standard or in the clients' or society's best interests. From work in psychology we can see ways that undue attention to risk-positive information can unconsciously be motivated by ego, concern about accountability - even status-seeking. Finally, we argue that the way the legal profession organizes and transmits its knowledge may itself introduce a bias toward risk-positive information, causing a cascade of reactive conformity even by otherwise unbiased legal advisors. Whereas findings from psychology and sociology are often invoked to question the emphasis on rational behavior found in economic analysis, n12 here they tell stories that are supportive and reinforcing.

In sum, we predict that systematic overstatement of risk is a robust, if not universal, phenomenon in the legal profession. Surely there will be situations where the accountability structure overcomes it (as when the advice is open to scrutiny by other lawyers who are in a position to try to take business away from the one who is too cautious) or clients diffuse it (as when a risk-seeking client makes clear to the lawyer that he or she is expected to come as close to the line as possible). Still, large segments of legal advice seem to meet the criteria we identify as likely to give rise to bias. n13

Before beginning our analysis, we should make clear what we mean by the "overstatement" of legal risk. We harbor no illusions that the law is particularly determinate or certain. n14 Quite apart from any of the biases we identify, lawyers will frequently differ in their analysis of a hard legal problem. n15 A rational lawyer will often be [*380] cautious in giving advice, hedging it with qualifiers and assumptions n16 - especially if she senses that the client does not want to risk a lawsuit, even if one might be won. By itself, this is not necessarily overstatement. But by definition, there is some advice that goes beyond simple prudence, where risks are magnified beyond the measure the average reasonable lawyer would attach to them absent the incentives we describe. This is not a very rigorous definition, which admittedly creates a challenge for the design of experimental tests of our hypotheses. n17 But hopefully it suffices for our limited purpose here, which is simply to establish the plausibility of the tendency to overstate legal risk.

I. THE OVERSTATEMENT OF LEGAL RISKS: THE ECONOMIC ACCOUNT

An economic analysis of the lawyer/client relationship shows that the lawyer has an incentive to overstate the relevant legal risks to her client, and that the client does not have sufficient means to ensure that such overstatements do not occur. The lawyer's incentive stems from the fact that she maximizes her income by portraying the risks inherent in a proposed transaction in such a way as to require additional legal services. Such a portrayal, on average, leads to an overstatement of the risks which the client faces. To be sure, constraints on such opportunistic behavior exist. The biggest of these constraints is the market for legal services. Yet, given information asymmetries in this market, competition for such services will not drive out those attorneys who overstate legal risks. While a client may employ devices designed to reduce the opportunity for strategic behavior on the part of lawyers, such as reviewing attorney bills and employing in-house counsel, these devices cannot eliminate the ability of lawyers to increase their payoffs above those which clients would be willing to pay in a perfectly informed and competitive market. Specifically, lawyers can maximize their income by adopting a strategy which overstates legal risks. [*381]

This overstatement of legal risks stems from the fact that the incentives of lawyers diverge from those of their clients. Lawyers seek to maximize their income while clients seek to obtain accurate, cost-justified information. Placed in the language of economics, there is an agency problem inherent in the lawyer/client relationship. Economists have produced a rich literature on the generic problems that arise in a principal/agent relationship. n18 We do not purport to add to this technical, complex body of research. Rather, we seek to apply the general learning in this area to the particular problems facing a client and its lawyer.

We proceed as follows. We first set forth the conflicting incentives of a lawyer and her client. We then look at how these incentives would affect the lawyer's performance where the lawyer is paid on a per-hour basis. We show that even where there is a well-established market for lawyers, which thus prevents the lawyer from charging monopoly rents for her services, the existence of asymmetric information allows the lawyer to charge more than she would in a market with complete information. While the market for legal services and client monitoring both limit the amount which the lawyer can charge, they nevertheless cannot eliminate the potential for opportunistic behavior. Indeed, the market for legal services, which requires lawyers to maintain a reputation for quality work, increases the incentive for the lawyer to overstate the risks inherent in a proposed transaction. Thus, we conclude that in the case of commercial transactions, lawyers who are compensated on a per-hour basis will overstate legal risks.

This conclusion raises the following question: Can clients obtain better information at a lower cost by altering the way in which they compensate their attorneys? Much of the economic literature focuses on the incentives of an agent under differing compensation schemes. n19 Indeed, in other contexts, lawyers depart from a per-hour fee arrangement. In some settings lawyers charge their clients on a contingency [*382] basis, while at other times they work for a flat fee. We show that in the case of business transactions, these alternative arrangements entail substantial agency costs. We tentatively suggest that the durability of the per-hour compensation system implies that the agency costs associated with this arrangement are lower than those attendant with these other compensation schemes.

A. The Basic Transaction Between Attorneys and Clients

We start with the basic transaction between a lawyer and a client. In short, a client purchases service from a lawyer. The nature of this service depends on the matter at hand. A client faced with a products liability suit wants a different service than a client who seeks help in facilitating a merger. Defending a client in a litigation setting is different than attempting to ensure that the transaction which the client wants to enter into does not run into legal difficulties. n20 Indeed, different clients may seek different services when faced with the same legal problem; for example, one client may wish to have a hard-nosed litigator when faced with a lawsuit while another client may seek a less combative advocate. n21 Lawyers simply do not sell a single homogeneous product called "legal services." Thus, any analysis of the incentives of the lawyer and the client must specify the services that are being purchased.

As we stated at the outset, our focus here is on those situations where clients seek advice regarding commercial transactions which they want to enter into. For this reason, we are basically concerned with the incentives of lawyers in firms. Few solo practitioners handle major commercial transactions. n22 When we talk about the incentives of lawyers, we are basically talking about the incentive of the firm as a whole. Where it is necessary to examine how the firm translates its [*383] general incentive to the individual attorneys, we explicitly draw the distinction between the larger entity and those who are its members. With these points in mind, we can now look at the conflicting incentives which arise when a client seeks legal advice on a business transaction.

We posit that the client seeks what we term "optimal information" regarding the legal risks that the transaction may face. By "optimal" information we mean more than "accurate" information. To be sure, clients want all the information that they receive to be accurate. A strategy of basing business decisions on inaccurate information has little to commend it. But clients do not necessarily want to receive all possible information concerning a proposed transaction, regardless of how accurate it is. At some point, the cost of gathering the information exceeds the benefits which the information will bring. For example, clients don't want lawyers spending 100 hours researching the problem of where a financing statement should be filed. If there is any doubt on the matter, it is cheaper to file in all offices which might be the right one. The basic observation is that for every transaction there comes a point where the client wants the attorney to cease her investigation. To put the point another way, the client wants the attorney to act as a perfect agent; it wants the lawyer to act as if she were handling her own affairs. n23 If the attorney would not spend her own money tracking down a tangential point, the client does not want to pay the lawyer to make such an undertaking.

One might be tempted to argue that this desire for optimal information is an artifact of the predominant type fee arrangement between the client and the lawyer, the per-hour billing system. After all, the per-hour fee arrangement ensures that the client both pays for the lawyer's marginal effort and receives the marginal benefit. The client thus wants the attorney to cease her efforts when the additional cost exceeds the additional benefit. We leave for later a full discussion of the ways in which various types of fee arrangements affect the incentives of lawyers in performing their tasks. Indeed, much of the economic literature on agency is concerned with how compensation schemes affect the incentives of the agent. For now, we simply want to [*384] make the limited point that the client's desire for optimal information exists regardless of which compensation scheme is in place. n24

It is readily apparent that clients want to limit the amount of information that they receive to that which is cost-justified when the attorney is being paid on an hourly rate. In such a situation the client directly receives both the marginal cost and the marginal benefit of the attorney's efforts. Yet such an incentive exists regardless of the type of fee arrangement between the parties. Consider first the client's incentives where the client hires the attorney on a flat fee arrangement. Once the lawyer and the client negotiate a fee, the client will want as much information as it could use. Here the attorney bears the marginal cost of her efforts while the client receives the marginal benefits. Looking at the client's motivations only after the fee has been negotiated, however, fails to take into account the client's concern with cost. The amount of a flat fee is not the product of divine mandate. Rather, the flat fee represents an implicit estimate of the amount of effort that the lawyer will spend on the matter. The more hours that the attorney generally spends on a given type of transaction, the greater the fee will be. To the extent that a lawyer spends additional effort where the benefit of such effort is less than its cost, the client will end up paying for this inefficient action. Thus, clients only want cost-justified information even in a flat fee arrangement.

A similar analysis holds for contingent fee arrangements. In such arrangements, the lawyer is promised a certain percentage of the proceeds of a transaction if that transaction is successful. n25 In a competitive market for legal services, this percentage is fixed so that the lawyer's expected compensation is equal to her expected effort. n26 To the extent that lawyers gather information which is not optimal, this is an added cost which will be passed on to the client in the form of a [*385] higher percentage rate. The more effort that a lawyer expects to put into an average case taken on a contingency fee basis, the higher the contingency rate will be. While clients are willing to pay a higher rate to obtain accurate information which is cost-justified, they do not want to pay a higher rate to procure information which does not meet this criterion. Thus, from an ex ante perspective clients seek optimal information even under a contingent fee arrangement.

Indeed, once we assume that there is a competitive market for legal services, it necessarily follows that clients want optimal information when they first select a lawyer. As we noted above, clients want unbiased information. To the extent that the client receives information from the attorney, the client must pay for this information. Regardless of how the particular compensation deal is structured, the attorney will expect to be compensated for her efforts. There is little reason to believe that attorneys will routinely accept fee arrangement under which they will not recover, on an expected basis, the value of their services. Such a price mechanism is an inherent feature of a market system. To the extent that lawyers provide information which is not cost-justified, the clients will pay the cost for this excess. We therefore are quite comfortable with the assumption that clients seek optimal information.

Lawyers do not share the incentives of their clients. We assume that lawyers seek to maximize their fees. Like most other people, attorneys want to have as high an income as possible. People generally receive a higher utility from more rather than less money. Of course, this desire to maximize income is constrained by other desires, such as the desire for leisure and the desire to maintain a pleasant work environment. People usually derive utility both from activities other than work and from the conditions under which they work. Thus, lawyers seek higher wages subject to the constraints that they also want to spend time away from work and that they wish to work under conditions that they enjoy. For the purposes of this paper, we assume that the lawyer has determined the amount of effort she wishes to devote to client matters. We thus treat her as needing to procure enough business to allow her to expend this effort.

The state of affairs just described creates an agency problem. Left unchecked, the agent (in this case the lawyer) would take action which runs contrary to the interest of the principal (the client). Specifically, lawyers would spend as many hours as possible on a given matter. Indeed, if clients had no way in which to monitor the hours [*386] spent by the attorney, the attorney would have an incentive to deliberately inflate the bill. n27 In the extreme case, the attorney would submit a bill based on her view of the maximum number of hours the client would be willing to pay for. While such intentional deception no doubt occurs in some cases, we do not believe that it is the standard practice in the profession. The reason that fraudulent billing is not the norm is that there are substantial constraints on the lawyer's ability to systematically overstate her efforts on behalf of her clients. We therefore set out a fuller account of the relationship between the attorney and the client. In particular, we make the following assumptions regarding the attorney/client setting that we are investigating. We view these assumptions as being a realistic approximation of the attorney/client setting.

First, we assume that there is a competitive market for attorney services. In other words, attorneys are constrained in the prices that they can charge. This assumption comports with reality. A law firm is not a monopoly. It has to compete with other law firms for business. If Cravath, Swaine & Moore charges more than the going rate, they will be undersold by Sullivan & Cromwell. While there may be isolated cases where a law firm has a unique ability for which it can charge above-market prices, n28 by and large commercial transactions of the type we are envisioning can be handled by a number of law firms. The attorney in our attorney-client relationship thus has no market power which it can use to extract monopoly rents from its clients.

This competition does not imply that clients can switch attorneys without incurring any costs. n29 Information provided to a corporate client often turns on the needs of the client. n30 With any new client, the attorney must incur the start-up costs of determining the client's needs. Also, the lawyer must establish a procedure for communicating with the client. These relationship-specific investments must be paid for. We assume that both the attorney and the client contribute to these investments. Thus, if a client seeks to change attorneys, it will both have to search for a new attorney (most likely a relatively low cost), and pay some of the cost of building the relationship with the [*387] new attorney (potentially a high cost, depending on the needs of the client). Thus, when a client switches attorneys, it incurs costs, which may be substantial. Stated differently, the client has an incentive to maintain the existing relationship.

The same is true for the attorney. The attorney, along with the client, makes relationship-specific investments. These investments are costly to the attorney. When an attorney loses a client, it loses these investments. To the extent that the attorney is able to procure a new client to replace the one that it lost, it has to make new investments in the new relationship. It thus follows that, for the attorney, working with a new client is less remunerative than working with an old one.

This conclusion is reinforced once it is recognized that a lawyer incurs costs in getting new clients. The lawyer-oriented press is filled with advice on how to attract new clients. All of the suggested activities require the lawyer to put forth effort. Unless the lawyer receives a tremendous amount of utility from the substance of these activities, the lawyer's overall utility would be higher if she simply had more business provided by her existing clients. Thus, while we assume that there is a competitive market for attorney services, once the client and the lawyer establish a relationship, switching to a new attorney is costly for both the client and the old attorney.

Second, we assume that the attorney has information which the client does not. In other words, there is asymmetric information. Specifically, we posit that the client, unlike the attorney, has no knowledge of the amount of effort necessary to uncover the optimal information which it seeks. This assumption is critical to our analysis. As is well known by now, a competitive market with perfect information leads to socially efficient outcomes. n31 It is thus fair to say that at some level our assumption of asymmetric information drives our analysis. n32 Were the parties to live in a Coasean world, this agency problem could be eliminated by contract. The parties would, without cost, negotiate a contract which required the lawyer to provide optimal [*388] information. The lawyer would not have any incentive to depart from this contractual standard since any deviations from this standard would be observed by client.

Unfortunately, in the real world the agency problem between the client and the lawyer cannot be so easily erased. To be sure, lawyers and clients could enter into contracts which specified that the lawyer was to provide "optimal information." It is reasonable to assume, however, that the client does not know what constitutes optimal information at the time it enters the contract; if it did, it would have no need for the attorney. Nor does simply requiring "optimal information" ensure that it will be provided. There is no Platonic form of "optimal information." What level of effort should be used in investigating any given transaction turns on the legal risks that this transaction poses. Only by starting to examine the circumstances before her can the lawyer begin to make a reasoned judgment as to what constitutes optimal information in this particular case. Even after the lawyer has given the client the requested advice, the client has little idea whether or not the lawyer has acted in a way consistent with its own interest. The client cannot see the choices which the lawyer made in conducting her research, nor, even if it had such information, is the client capable of knowing whether or not the lawyer's decisions were sound. In the language of game theory, whether or not the lawyer has provided optimal information is neither observable (i.e., unknown to the party) nor verifiable (capable of being determined by a third party such as a court). Thus, once we assume that there is asymmetric information, simply requiring the lawyer to provide optimal information cannot solve the agency problem between the client and the lawyer.

Finally, we assume that the client can assess the level of effort put forth by the lawyer indirectly through the hours that the attorney spends on the transaction. This is at best, however, only an approximation of effort. As those of us who have practiced law know, some hours are more productive than others. Nevertheless, as Ronald J. Gilson and Robert H. Mnookin have pointed out, there are reasons to believe that law firms cultivate individuals who put forth high levels of effort for each hour spent. n33 Moreover, even if there is some variation in the amount of effort per hour both for each attorney and among attorneys, we know of no reason to suggest that any one client is systematically given the more productive hours. We thus assume that [*389] although clients cannot directly observe effort, hours spent on a matter is a rough approximation of the effort given to the project.

These rather austere assumptions capture much of the dynamic in the attorney/client relationship. We first examine this dynamic in the context of a per-hour compensation arrangement, the type of compensation arrangement which is ubiquitous in business law practice. We show that this arrangement induces attorneys to give legal advice which, on average, overstates legal risks. We then examine other potential compensation arrangements. Most of the principal/agent literature address the extent to which compensation arrangements can affect the incentives of the agent. n34 We show that no other compensation arrangement is clearly superior to the per-hour fee agreement for reducing the agency costs inherent in the lawyer/client relationship.

B. The Incentive to Overstate Legal Risks in a Per-Hour Compensation Agreement

The most common method of compensating attorneys for their work in commercial transactions is on a per-hour basis. n35 The bills which clients receive are generally calculated directly on the amount of attorney time put into the project. In this section, we examine the way in which per-hour billing affects the incentives of the attorney to communicate accurate information to her client. Specifically, we examine how the lawyer's desire to maximize her income on any given matter and her desire to maintain a reputation for quality affect the advice which she provides.

1. The Income Incentive

The obvious problem with the per-hour fee arrangement is that the attorney has an incentive to put too much effort into the matter at hand. As many have noted, attorneys under this compensation scheme have an incentive to bill as many hours as possible. n36 The [*390] more hours that they bill, the more income they receive. What has received less attention are limits that curb the attorney's ability to spend additional time on any given matter and the effect that such an incentive has on the communication between the lawyer and the client. Indeed, we will show that the two points are related; it is the efforts by clients to monitor attorney performance that induces attorneys to overstate legal risks.

As an initial matter, one might suppose that the incentive to bill more hours has no effect on the lawyer's communication to the client of the relevant legal risks. Instead, the lawyer will adopt one of two strategies. First, the lawyer might simply overstate the amount of time she spent on the project. Indeed, it is a common perception that lawyers, or at least some lawyers, routinely overbill their clients through a variety of devices such as inflating the numbers spent on a given task and double billing. n37 Second, assuming that the lawyer accurately reports the time she spends on a client's work, she may simply spend more time on a matter and then provide accurate information to the client. While the lawyer may not provide optimal information - after all, the lawyer has an incentive to provide as much accurate information as possible, not just that which is cost- justified - the lawyer will supply accurate information. Neither of these two strategies affects the accuracy of the information which the client receives.

These two strategies, however, impose a cost on clients. The first strategy is particularly expensive. The client pays for legal services which it does not receive. It is thus not surprising that there are substantial constraints on attorney overbilling. First, attorneys themselves have deemed such practices unethical. n38 Second, and perhaps more importantly, attorneys who overbill and are caught face significant penalties. Clients routinely monitor their attorneys. If anything, it appears that clients are increasing their efforts to make attorneys [*391] account for their time. Recent years have seen an increase in the number of firm's employing in-house counsel. Such counsel often review the bills of outside attorneys. n39 Recently such counsel have required outside attorneys to submit their bills in a format which groups the hours spent according to the task which the lawyer performed. It is commonly believed that such a format makes it easier to detect overbilling. For those firms that do not have in-house counsel, they may for a given project hire a second outside attorney to monitor the effort of the first. n40 Both the in-house counsel and the second attorney find it in their interest to uncover fraudulent billing because it increases their value to the client. The more abuse that they uncover, the more likely the client is to use their services in the future. Thus, attorneys cannot overstate the hours that they spent on a project with impunity.

To be sure, these mechanisms do not guarantee that all overbilling will be uncovered. No one thinks that review of bills will detect every instance of overbilling. The lack of perfect detection, however, does not necessarily imply that fraudulent overbilling will exist. As is well known in the criminal law literature, optimal fines increase as the rate of detection decreases; otherwise, there would still be an incentive to engage in the prohibited conduct. n41 Thus, if the penalty that attorneys pay when caught deliberately overbilling is large enough, most attorneys will not engage in such practice even though the chance of detection is less than a hundred percent. Indeed, in the case of attorneys caught deliberately overbilling, the penalty they pay far exceeds the gain they received by overbilling. First, they have to give [*392] back all the monies that they overbilled. For example, Harvey Myerson & Kuhn agreed to pay Shearson Lehman Hutton $ 1.1 million for overbilling. n42

More importantly, however, attorneys who are discovered overbilling face a large reputational penalty. The client who was overbilled can switch attorneys. While there is a cost to such action, there is a greater cost to signalling to an attorney that they will pay no penalty if they overbill. Thus, attorneys who intentionally overbill often lose their client. Moreover, future clients would be reluctant to give business to lawyers who have a reputation for overbilling. There is little reason to suggest that a corporation would prefer to do business with someone who has demonstrated a willingness to steal. Lawyers will therefore often avoid a strategy of intentional overbilling.

We see fears of reputational penalties in the recent actions of law firms. Recently, one law firm detected overbilling by one of its attorneys. n43 In what appears to be an attempt to protect its reputation, the firm voluntarily disclosed the overbilling, reimbursed the overbilled client, and disciplined the offender. Given these severe penalties which face those who deliberately overbill their clients, we expect that few attorneys would engage in such action.

This leaves the strategy of increasing the amount of hours spent on any given matter. This strategy is more costly to the attorney than simply padding the bill because she must actually expend effort but it also increases her income. There are reasons to believe, however, that lawyers who adopt this strategy will tend, on average, to overstate legal risks. This conclusion flows from the fact that clients monitor a lawyer's production. A lawyer simply cannot say that she spent a certain amount of time studying a transaction and then give the client little or no information. In other words, at some level she must justify the time spent on a matter. n44 It is this constraint which creates the incentive to overstate legal risks. This incentive derives from two distinct sources. First, by discovering legal risks in a proposed transaction, the lawyer can give the client the sense that the client received something of value for the service. If the lawyer simply stated that there were no problems with the transactions, the client might think [*393] either that the lawyer was not putting in a high level of effort despite the number of hours billed, or that, at least from an ex post perspective, it was a waste of money to hire the lawyer. Either reaction would make it less likely that the lawyer would get future business from the client. Finding a legal risk thus signals to the client that the attorney has indeed expended effort on the client's behalf.

The attorney has an incentive to overstate legal risks for another reason as well. A lawyer who assesses legal risks for a client does not simply state the likelihood that a legal problem will arise. Often times the lawyer can suggest ways in which the risks can be reduced. In other words, some legal risks are manageable. Such management often entails substantial work on the part of the attorney who identified the risk in the original proposal. For example, a transaction may be restructured so as to reduce the risk of it being a taxable event, contracts may be redrafted to cover contingencies not originally addressed, or an internal monitoring system may be instituted.

Overstating legal risks thus creates additional work for the attorney. Indeed, the optimal strategy for the lawyer is to discover legal risks which the attorney can overcome through additional effort on the lawyer's part. This both gives the client the sense that it is receiving a return on the legal fees that it spends, and maximizes the lawyer's income.

This analysis turns on the income of the attorney being tied to the monies paid by the client. Of course, most corporate transactions are handled by law firms. In the firm setting, not all attorneys have a direct correlation between their billable hours and their personal income. To be sure, partners in some firms have their salary determined, at least in part, by the number of hours that they bill. n45 They thus have a direct incentive to maximize their number of billable hours. Traditionally, however, many law firms have used "locked-step" compensation systems; the partner's salary was tied to the length of his tenure with the firm rather than on his contribution to the firm's revenues. n46 In such firms, however, mechanisms exist to ensure that the individual lawyers retain the incentive to maximize firm revenue. Most prominently, all members of the firm, including partners, are required to account for their time. Partners who thus shirk their [*394] responsibility to contribute to firm revenues can be identified and disciplined. n47 Firm culture also stimulates attorneys to increase their productivity. n48 Thus, even in firms where compensation is not based on hours billed, partners retain the incentive to maximize firm profits.

This leaves the incentives for associates. Associates are promised a fixed salary. Their salary does not turn on how many hours they bill. Rather, it is the salary of the partners which is increased by each additional hour which an associate bills. Law firms use a number of devices to align the incentives of the associates with those of the partners. Many firms have minimum billing requirements. At most firms, the number of hours affects whether or not associates will become partners. The greater the number of hours billed, the higher the chance that the partnership decision will be favorable. Thus, it is reasonable to conclude that business lawyers, be they partners or associates, have an incentive to maximize their billable hours. This incentive leads them to adopt a strategy which overstates legal risks.

2. The Reputational Incentive

A lawyer's concern for her reputation also counsels her to overstate legal risks. Few would doubt that reputation is important to an attorney. As we noted above, a client observes the hours that an attorney spends on a matter. It does not observe the actual level of attorney effort, nor does it directly observe the accuracy of the information that it receives. To assess the quality of the advice the attorney provides, the client relies on its belief as to the attorney's abilities. This belief is tantamount to the attorney's reputation. In this section, we show that the process by which clients update their belief as to the accuracy of the attorney's advice induces attorneys to overstate legal risks.

The only information that the client receives regarding the accuracy of the advice it receives is whether or not it was able to go through with the transaction, and, if it was, whether the transaction encountered legal difficulties. This ex post information gives the client some idea as to the accuracy of the legal advice that it received. For example, if the client is told that it cannot do a deal, but then sees another firm doing the same deal without running into legal difficulties, it would conclude that there was a possibility that the advice that it received was inaccurate. Similarly, if a client is told that there is no [*395] legal impediment to a certain transaction, but then discovers that it is sued successfully after it undertakes that transaction, it will question the advice that it received.

Of course, this ex post evaluation of legal advice is not perfect. Law is by no means a precise system. Unexpected events occur. At times, some transactions that were thought fraught with legal difficulties turn out to be perfectly lawful. At other times, transactions which were thought to be immune from suit ultimately get struck down. Thus, the fact that legal advice turns out to be wrong does not necessarily mean that, ex ante, the advice was inaccurate. Nevertheless, clients do draw inferences from ex post events.

To assess the way in which a client draws such an inference, we begin by noting that legal risks come in varying degrees. For example, a reasonable attorney might conclude that a proposed transaction runs a one in three chance of being successfully challenged. Yet, from a client's ex post perspective, the risk either did or did not materialize. The relevant question becomes how the client interprets the outcome. We assume that the client is rational, and thus updates its beliefs according to Bayes' Rule. n49 Bayes' Rule operates as follows. To arrive at an updated belief about the likelihood of a certain event in light of a new piece of information, multiply the preexisting likelihood for that event times the probability that the new information would exist if the event in fact occurred, and then divide that by the overall probability that the information received would occur.

The event which we are concerned about in this paper is the lawyer providing accurate information regarding legal risks. The data which the client receives is whether or not the proposed transaction encounters legal difficulty. In this situation, Bayes's Rule can be written as follows:

Client's updated belief as to Lawyer's quality =

(LR)(CB)/((LR)(CB) + (1-CB)(CRB)

where LR = Lawyer's statement of Risk; CB = Client's preexisting Belief of lawyer quality; and CRB = Client's belief that Risk will occur if lawyer advice is Bad.

For a numerical application of this version of Bayes's Rule, assume the following. The lawyer genuinely believes that there is a thirty percent chance that the transaction will run into legal problems. [*396] The client's preexisting belief that the lawyer provides accurate information is ninety percent. We assume that this belief is based initially on the attorney's reputation. Moreover, the client believes that if the lawyer provided accurate information, there is a thirty percent chance that the transaction will run into legal difficulty. In other words, the client believes that the lawyer accurately communicates her belief. The client also believes that if the lawyer provided inaccurate information in the form of failing to discover a relevant legal risk, there is a fifty percent chance that legal problems will derail the transaction. In this situation, if the transaction encounters legal difficulties, the client's belief that the lawyer provided accurate information declines to eighty-four percent. n50 Conversely, if the transaction does not encounter legal difficulties, the client's belief that the lawyer provided accurate information increases to ninety-three percent. n51

Bayes' Rule illuminates the incentive of the lawyer to overstate legal risk. At the time the lawyer gives advice to the client, the lawyer has no control over the client's preexisting belief as to the attorney's quality. n52 She also has no control over the client's belief as to the likelihood that the transaction will run into legal risk if the advice given is inaccurate. The lawyer does have control, however, over the estimate of risk which the client receives. This is the advice which she gives to the client. By overstating legal risks, the attorney can dampen the reputational penalty she pays if the transaction in face encounters legal problems. For example, assume in the prior hypothetical, the lawyer told the client that there was a forty percent chance of running into legal problems. Now, if legal problems do occur, the client's belief about the accuracy of the information it received falls from ninety percent before the transaction to eighty-eight percent. Conversely, if the transaction goes through without a hitch, the client's belief in the lawyer's accuracy increases to ninety-two percent. Thus, by overstating legal risks, the lawyer decreases the reputational penalty that it will pay with this client if the risk does occur.

To be sure, such overstatement also decreases the reputational bonus that the lawyer will receive if the risk does not come to pass. It is likely that this is a tradeoff that most lawyers are willing to make. The fact that they are currently doing business for this client suggests [*397] that the client has sufficient trust in their work product to continue giving more business. At some point, however, that trust would decline to such a level that the client would lose so much confidence in the lawyer's ability to predict risk that the client would shift to another attorney. Such a shift is quite costly to the lawyer. First, the attorney loses the future business from the client. To be sure, this gives the attorney the ability to take on other clients. Yet, as we discussed above, procuring additional clients itself requires attorney effort. Thus, the loss of a client, even assuming that it did not impair the attorney's ability to attract a new client, reduces the attorney's income.

Perhaps even more importantly from the attorney's perspective, the loss of a client most likely harms the attorney's ability to attract a new client. When a client walks out on a law firm, such a split gets noticed. Other potential clients often learn of the departure. If the disgruntled client publicizes its reason for leaving, prospective clients would decrease their own estimate of the attorney's ability. Indeed, given that the departing client has better information regarding the lawyer's output, outside observers are likely to place great weight on this information. n53 To be sure, the deserted attorney may attempt to convince future clients that the departure was not based on the quality of the work which the client received. Nevertheless, so long as these clients place any weight at all on the possibility that the departure was based on attorney incompetence, the departure will cause these prospective clients to lower their estimates of the attorney's ability. Client departure thus increases the cost of getting new clients.

From the lawyer's perspective, the potential for a marginal decrease in client trust is not offset by the potential for a marginal increase in such trust. By overstating legal risks, the lawyer can decrease the reputational penalty that she pays when transactions go awry. [*398]

There are, of course, limits on the extent to which a lawyer can overstate a legal risk. One of the primary constraints is the fact that clients receive information about lawyer performance not only from the outcome of the transactions that they engage in, but also from the advice which the attorney provides. Just as clients have a subjective probability as to the likelihood of failure when they receive bad legal advice, they also have a subjective probability as to the amount of legal risks in the world. Clients, before they hand a transaction over to an attorney for legal advice, have some idea as to how likely it is that legal problems will require a transaction to be either canceled or restructured. When clients receive advice, there are thus two prior beliefs at work - the belief that the attorney is accurate, and the belief that there is a certain level of risk in the world. In a single transaction, if the client has a high degree of confidence in the lawyer, it may use the advice it receives to update its belief as to the amount of risk in the world. If the client, however, consistently receives negative advice, at some point it will begin to question its estimate of lawyer accuracy. In other words, if the lawyer overstates legal risks by too much, eventually the client will lower its assessment of the attorney's work product.

A second constraint on the overstatement of legal risk is that the client can observe actions taken by its competitors. If a client forgoes an action, such action may be engaged in by one of its rivals. If the rival does not encounter legal problems, this will cause the client to lower its belief as to the quality of its attorney. Eventually, the belief will reach such a level that the client will seek new counsel. Thus, the more common an event, the less likely that the attorney will overstate the risk to such a level that the client fails to take the action.

Indeed, this constraint suggests again that the best course of action for the attorney is to overstate the legal risk in a way that portrays that risk as manageable by the attorney. By overstating the risk the attorney protects her reputation, and by still allowing the deal to go through she ensures that she does not pay the penalty which accrues when she conducts a transaction and the client observes others engaging in that same type of deal. Concerns with reputation thus dovetail with a desire to maximize income and lead the attorney to overstate legal risks in a way which presents such risks as manageable by the attorney.

One objection to the above analysis is that if clients have rational expectations, they will assume that lawyers overstate legal risks, and [*399] thus discount the information that they receive. We have no doubt that such discounting at times occurs. Given the popular image of lawyers as naysayers, clients probably take a jaundiced look at the information which they are provided. There are limits, however, on the extent to which clients can disregard the advice of legal counsel. If a client ignores a lawyer's warning, and the transaction runs into legal difficulties, the fact that the client knew of the risk may increase the client's chances of being found liable. Moreover, even if the client assumes that the lawyer has overstated legal risks, it does not know which legal risks have been overstated. When the client is presented with a manageable legal risk, it has no way of knowing whether there is no legal risk at all, or if the attorney is simply inflating the degree of an actual risk which the attorney can reduce through future efforts. This being the case, the client may find it prudent to follow the advice given, even if it suspects that the legal risk may be overstated.

Once clients expect that lawyers will overstate legal risks, it then becomes incumbent on lawyers to act in accord with this expectation. The more that a client expects that it will be told of all legal risks, no matter how small, the more it will question the lawyer's competence if she fails to identify a legal risk which actually materializes. Given the incentives to overstate that the client believes the lawyer to have, the lawyer's failure to identify a risk will be more likely attributed to a mistake on the part of the attorney rather than a reasoned judgment that it was not cost justified to pursue a certain avenue of research in this particular case.

3. Client Monitoring of Attorney Performance

Both the lawyer's interest in maximizing her income and her interest in maintaining a reputation for uncovering relevant risks leads the attorney to overstate legal risk. This creates an agency cost for the client in that the client both gets a skewed assessment of the transaction it is considering and pays more than it would if it was provided with optimal information. In this section, we assume that the client remains committed to a per-hour fee arrangement and look at devices that the client may use in an attempt to reduce the agency costs associated with such an arrangement.

One potential mechanism that the client can use to reduce its agency costs is to hold a "beauty contest." A beauty contest involves the client showing a project to prospective attorneys, and then having [*400] the attorneys make a proposal as to how they would handle the matter. To date, most beauty contests have involved the selection of attorneys to handle litigation. Yet, there is no reason that such contests could not be used in the commercial setting.

An advantage of the beauty contest is that it may reduce the bilateral monopoly problem that exists when a client has already developed a relationship with an attorney. Having a beauty contest in this situation both signals to the existing attorney some dissatisfaction with the current state of affairs, and also informs the client of its exit options. Telling an existing attorney that you are conducting a beauty contest informs that attorney that it should review its current operating procedure. Moreover, by reviewing the proposals by other law firms, the client knows what services it can expect if it switches attorneys.

The problem with beauty contests is that they are expensive. Attorneys who prepare a proposal and do not win the contest are barred from working on any aspect of the transaction with another client. In other words, there is an opportunity cost to the attorney by participating in the beauty contest. The client must ultimately pay for this cost. Since lawyers operate in a competitive market, they only make a competitive return on their efforts. If attorneys are not reimbursed for their costs of making a proposal, they will have little incentive to enter the beauty contest. Such reimbursement can either be in the form of an upfront payment for the proposal, or in a higher compensation rate if the attorney is in fact the winner. n54 This cost places a limit on how often beauty contests will be used in a commercial setting.

Perhaps the greatest ex post constraint which the client can impose on the lawyer's incentive to overstate legal risk is to have in-house counsel review the lawyer's bills and her output. n55 In-house counsel often have experience in working for private law firms. n56 [*401] They thus have some idea about the amount of time that it takes to complete certain tasks. Indeed, to facilitate such policing, some clients are now requiring that law firms submit their bills with the various time spent on the matter grouped by task. n57 The client thus can readily ascertain the exact nature of the services that it has purchased. It is apparent that this oversight constrains overbilling. The in-house counsel can determine whether or not the outside lawyer is spending excessive amounts of time on the matters that she has been entrusted with.

The use of in-house counsel also constrains the ability of the lawyer to overstate legal risk. Presumably, the in-house counsel knows more law than does the client. Thus, the in-house counsel is more likely than the client to identify intentional overstatement of legal risks. Since the lawyer knows that its work product will be scrutinized by a second attorney, it will tend to be less aggressive in its overstatement of legal risks. Indeed, if in-house counsel were a perfect monitor on the actions of the outside attorney, it would eliminate the agency costs arising in the lawyer/client relationship. n58

There are many reasons to believe, however, that in-house counsel are not perfect monitors of outside attorneys. One reason to think this is that in-house counsel cannot replicate the decisions which the outside attorney had to make. Presumably the outside attorney was hired either because it had more resources to devote to a problem than the in-house counsel or it had an expertise which the in-house counsel lacked. Under either scenario, however, the in-house will not be able to assess with perfect accuracy the performance of the outside counsel. If the in-house counsel has insufficient resources to handle the matter, there is little reason to think that it has sufficient resources to so analyze the product of the outside lawyer that it can ascertain whether the outside lawyer provided optimal information. Similarly, [*402] if the outside lawyer has expertise which the in-house counsel lacks, the in-house counsel cannot determine whether or not the hired attorney gave accurate and cost-justified information.

In addition to the problem of limited information, there also is an agency problem in the in-house counsel/client relationship. Just as there is no reason to think that an outside lawyer will act as a perfect agent for a client, there is no reason to think that an in-house lawyer will so act. To be sure, the in-house counsel has a different set of incentives than does the outside attorney. We do not endeavor to examine completely the incentive structure of in-house counsel. For the purposes of this paper, we assume that, as a rough approximation, an in-house counsel seeks to maximize his importance in the corporate hierarchy.

This desire may lead in-house counsel to scrutinize bills closely. The more money that an in-house counsel can claim that he has saved the corporation, the larger his role may be. Indeed, an in-house counsel, by finding problems in a lawyer's bill, may use these findings to lobby the corporation for even more resources. After all, if he could find these mistakes with his current staff, just think what he could uncover with an even larger staff.

While in-house counsel thus has some incentive to uncover instances of overbilling, it may not have sufficient incentive to discover the overstatement of legal risks. Just as discovering manageable risks may maximize the outside lawyer's income, discovering such risks may maximize the importance of the in-house counsel. If the in-house counsel repeatedly tells the client that there are no risks with its proposed course of action, he has in effect marginalized himself in the corporate hierarchy. His stamp of approval, if routinely given, offers him little chance to expand his domain. Finding legal risks which can be managed, however, affords him new opportunities to expand his empire. The more legal risks that need to be handled, the more staff the in-house counsel needs.

While in-house counsel thus has an incentive to monitor the outside attorney for instances of overbilling, it also has an incentive to acquiesce in the overstatement of legal risk. Thus, such services may reduce the agency costs inherent in the lawyer/client relationship, but they cannot eliminate such costs in their entirety. In sum, the current structure of the lawyer/client relationship creates an incentive for lawyers to overstate legal risks. [*403]

C. The Agency Costs in Alternative Compensation Systems

The per-hour billing arrangement thus does not ensure that clients will receive optimal information. In particular, attorneys will spend too much time on a given project and overstate the risks associated with that project. Clients, however, are presumably aware of this tendency. In this section, we examine whether any other compensation scheme will reduce the agency costs between the lawyer and the client. We focus on two alternative arrangements which are quite common in other areas of legal practice: the contingency fee and the flat fee.

1. Contingent Fees

In theory, the greatest reduction in agency costs in the principal/agent setting occurs where the parties craft a contingent fee arrangement which places all the risk arising from the transaction on the attorney. In the context that we are considering, such a contract would provide that the lawyer would be paid only if the transaction does not encounter any legal problems. Under such a contract, the client would be promised a constant return, and the lawyer's fee would turn on the results that were generated. Such outcome-specific contracts are parieto optimal when the agent, in this situation the lawyer, is risk neutral. n59 The reason for this conclusion is that this contract gives the lawyer the incentive to provide optimal information. Since the principal is guaranteed a fixed return, the lawyer bears all of the marginal costs of her action. Note that this contract differs from the traditional contingent fee arrangement in that the lawyer bears the entire marginal costs and benefits of her efforts rather than a fixed percentage of the overall recovery. She thus has the appropriate incentives to maximize the value of the project by providing all accurate legal information which is cost justified. This contract thus ensures that the lawyer has the incentive to provide optimal information. Moreover, given that there is a competitive market for legal services, the client should be able to receive a promise for the value of the project minus the value added by the attorney's effort.

Despite the theoretical desirability of such contracts, it is not surprising that we do not see them in practice. Simple solutions often do not translate to a complex world. The reason that we do not see such [*404] contracts stems from the fact that under such a contract the attorney effectively buys the project from the client. n60 The result of this purchase is that all the risk of the transaction comes to rest on the attorney. There are reasons to think that lawyers would not readily accept such risk. One is that lawyers are not the only ingredient of success in a commercial transaction. Many different parties contribute their services to most commercial transactions. This results in the fact that if the lawyers' income turns on the success of the project, the lawyers would face an agency problem in which they are the principal. Their agents would be those who actually perform the transaction at issue. It may well be the case that the current clients are much better at controlling the agency costs attendant with the transaction itself than the lawyers would be. Indeed, the reduction in agency costs attributable to having the client rather than the lawyer conduct the transaction may be greater than the agency costs which exist in the client/lawyer relationship. It is not an heroic assumption to assume that the clients are better able to manage the projects that they devise than are the lawyers that they hire. Thus, the project is worth more with the client running it than if the lawyers were in charge. This implies that the lawyers would not be willing to pay a fixed amount to the client which the client would be willing to accept.

Moreover, it is unlikely that lawyers are risk adverse in the type of transactions that we are considering. Sophisticated commercial transactions run into the millions of dollars. For a lawyer to guarantee a client a fixed return in such transactions, the lawyer would have to be willing to risk millions of dollars. It is fanciful to suggest that lawyers would be risk neutral in such a situation. For example, would any law firm have been willing to buy from Ross Johnson his idea to take RJR-Nabisco private? Thus, the contractual solution which places all the risk of the transaction on the lawyer is not a feasible solution to the agency problems inherent in the client/lawyer relationship.

What about a more traditional contingent fee arrangement? Such a fee structure would promise the lawyer a certain percentage of the deal. At first blush, such a contract promises to align the incentives of the attorney with those of the client. Since the attorney participates in the marginal value she adds to the deal, she has the incentive at the margin to provide accurate information. Moreover, unlike the arrangement where the attorney promises a fixed return to [*405] the principal, here the only capital which the attorney invests in the project is her effort. Of course, where the matter is handled by a firm, the partners are investing in the salary of the associates. Nevertheless, the exposure of the attorney is decidedly more modest than it is when the attorney buys the project.

Three problems exist with such an arrangement, however. The first is that the client may be better to bear the risk in the transaction than is the attorney. While the attorney does not invest as much in the pure contingent fee contract as she does when she promises to pay the client a fixed amount, she nevertheless is exposed to the risk that the project may not go through. Moreover, while the attorney may have some control over the legal risks involved, she may not have control over other risks which may derail the transaction. For example, the proposed TCI-Bell Atlantic merger fell apart not because the deal encountered a legal impediment, but rather because the financing could not be worked out. n61 To compensate her for bearing this risk, the attorney will insist on a higher percentage of return for those transactions which are successful. This insistence will raise the expected costs of legal fees for the client. Thus, in this setting, it may be the case that the parties would not reach a contingent fee contract because the client is better able to bear the risks of the transaction not occurring than is the attorney, and the cost required to induce the attorney to bear the risk is greater than the costs inherent in the per-hour billing arrangement. n62

The second problem with a contingent fee in the commercial setting is that the attorney has an incentive to provide less than optimal information. This stems from the fact that the lawyer does not receive the full marginal value of her efforts. Under a contingent fee agreement, the attorney only receives a set percentage of the transaction. The lower this percentage, the quicker the attorney will find that an additional effort is not worth the marginal increase in the value of the deal. This may be a particular problem in large commercial transactions. Typically, attorneys fees represent only a small part of the cost of such transactions. n63 This implies that if a contingent fee were in [*406] place, it would be a relatively small percentage. Thus, attorneys may quickly reach the point where they have no incentive to gather more information. A contingent fee arrangement in this setting may thus lead to the understatement of legal risks.

To be sure, the market provides a constraint on such understatement. As we noted above, lawyers are very concerned with their reputation. n64 If they fail to spot a legal risk which materializes, and if such failures are observable by prospective clients, then such failure may lead to the loss of future clients. The cost of failing to spot a legal risk in terms of decreasing the number of future clients to some extent ameliorates the incentive to understate legal risks in a contingent fee contract. The exact magnitude of such a cost, however, is impossible to determine in the abstract. Thus, we cannot conclude that it only partially offsets the incentive to understate legal risks, perfectly balances this incentive, or overwhelms this incentive so that the lawyer now has an incentive to overstate legal risks. Nevertheless, it is clear that there is a greater chance that the lawyer will understate legal risks in a contingent fee agreement than in a per-hour billing arrangement.

The most formidable problem with the contingent fee arrangement in a commercial transaction is defining the key aspects of the fee arrangement. The difficulty lies in defining "success" and in identifying the pool of money on which the attorney's fee is based. In the case where an attorney represents a plaintiff in a civil suit, defining success is an easy matter - whether or not the plaintiff wins the case. So is identifying the proceeds to which the attorney can look - the judgment. In the case of a commercial transaction, however, things are not so easy. Consider first the hurdles that must be overcome in defining "success." In the commercial setting, there is no dichotomy of outcomes as there is in the litigation setting. There is no judgment setting forth the degree of the client's victory. Rather, there is a host of outcomes which run on a continuum. For example, assume that a client comes to an attorney seeking advice on how to respond to a hostile takeover attempt. Is success in this situation keeping the company independent, getting the highest price for the shareholders (which itself raises questions as to how do you know when you are getting the highest price), or finding a white knight?

Moreover, what is successful from a business perspective may change during the course of the transaction. A board of directors [*407] which at one time thought that the best course for the company was to keep it independent may later decide that it should be auctioned off. The lawyers, however, if they were only to be paid if the company remained independent, would now have an incentive to discover legal reasons why there should be no auction. This problem is particularly acute in areas such as hostile takeovers where one of the jobs of the lawyers is to advise the board of directors as to whether and when there is a legal duty to sell the company. n65 By tying the lawyers' fee to the outcome of the transaction, the lawyers' incentives may not keep pace with those of the client where these latter incentives change during the course of the representation. The legal advice which the lawyer provides would then be shaded toward the outcome which promised the higher return to the attorney rather than to the outcome which the client currently seeks.

Similarly, attempting to define success may create an incentive for the lawyer to give inaccurate information. Whereas a plaintiff always wants to succeed at trial, a business does not always want to go through with a proposed transaction. At times the business needs to know the legal ramifications of what it seeks to do to know whether or not it should enter the transaction at all. Yet if the lawyer's fee were based on the "success" of the venture in that the lawyer would only be paid if the transaction was consummated, the lawyer would be reticent to give information which would reduce the possibility of success. For example, assume that success is defined as successful completion of the merger. In this situation, the lawyer would have an incentive to ignore or at least understate legal risks which would imperil the merger. Thus, a contingent fee arrangement may tie the lawyer's interest too closely to a particular outcome, and thus greatly reduce the chance that the client will receive optimal information.

Added to this difficulty with defining success is the problem of identifying the sum of money on which the attorney's fee should be based. Unlike a civil case, at the end of the deal there may not be a ready yardstick on which the attorney's fee could be based. Consider again a lawyer hired to prevent a hostile takeover. In this situation, there is no ready figure for determining the value of the deal. The same is true if the lawyer helps negotiate a technology sharing agreement or helps set up a joint venture. Even in cases where there is a [*408] readily ascertainable dollar amount for the transaction, it may be difficult to work out a contingent fee. In civil litigation, lawyers have developed norms regarding the amount of the contingency fee. n66 In commercial situations, however, such norms might not develop. Commercial transactions differ widely in both the amount of money at issue and the legal complexities involved. A client with a relatively routine transaction would not want to pay the same percentage as a client which is trying something novel. Thus, there may be added costs as the client and the attorney attempt to find a percentage which provides appropriate compensation for the deal at hand.

These problems with a contingent fee arrangement suggest why we do not commonly see such arrangements in commercial practice.

2. Flat Fees

Another way the parties could structure the lawyer's compensation agreement is for the client to pay the lawyer a flat fee. Prior to the attorney beginning work on the project, the attorney and the client would dicker over the attorney's fee. Once the parties reached agreement, the lawyer would be paid this fee regardless of the success of the transaction and the amount of effort the attorney put in. Like the contingency fee, however, the flat fee creates agency costs which counsel against its implementation.

A flat fee contract is the most efficient method of solving the principal/agent problem where the principal can observe the action taken by the agent. n67 The principal pays the agent a sum which is greater than the agent's next best alternative wage, and directs the action that the agent is to take. The agent will take such a contract because she gets paid more than she otherwise would, and she will conform to the terms of the deal because any defection is immediately known by the principle. Indeed, we sometimes see such contracts in the market for legal services. Consider, for example, the standard home closing transaction. This is a transaction where it is easy to specify the action that the attorney is to take - fill out the proper forms - and it is also easy to determine whether or not the attorney has completed the task. Thus, it is not surprising that attorneys in this setting charge a flat fee. [*409]

Unfortunately, n68 the necessary conditions for the flat fee being optimal are not present in complex business transactions. As we noted above, the lawyer cannot credibly commit to provide optimal information. The client is simply unable to assess whether or not the lawyer's performance meets this standard. This inability to observe attorney action gives rise to a number of agency costs, and thus suggests why flat fees are not a common payment device for corporate attorneys.

One serious problem with the flat fee is the incentives that it creates on the part of the lawyer. After the fee is negotiated, the attorney has the incentive to spend as little as time as possible on the matter. Each additional hour spent on a matter does not add to the bottom line. The lawyer would thus do whatever it took to satisfy the demands of the contract. To be sure, doing nothing at all would probably be a breach of the contract. Just as even the deadest of deadwood in academia still have to teach the classes they are assigned, the attorney still is required to do something (render an opinion, draft documents, etc.) in order to obtain the fee. Nevertheless, the attorney would put forth the minimal effort necessary in order to collect her fee. Presumably, this lack of effort would translate into an understatement of legal risks. The lawyer will not invest the necessary effort to uncover all of the problems the transaction may encounter.

Of course, clients would not be unaware of this incentive on the part of attorneys to put little effort into a transaction for which the attorney is being paid a flat fee. Thus, if we assume that the client is committed to a flat fee arrangement, it would lower the amount that it is willing to pay the attorney. This situation means that the lawyer is not extracting as much from the deal as she could. The client would be willing to pay more if it could be assured that it would receive a better assessment of the legal problems that it faces. Since the lawyer would thus be paid as if she were going to put little effort into the matter, it is in the lawyers interest to find credible ways to commit to expending a greater level of effort. Such a commitment would have the effect of raising the fee which the lawyer could demand.

One way in which a lawyer can give assurances to her client that she will spend more than the minimal amount of time necessary to be entitled to the fee is to develop a reputation for high quality work. As [*410] we noted in discussing hourly fees, lawyers care about their reputation. While one may posit that this is a consumption good in and of itself (most of us care about how others perceive our abilities), a good reputation is also necessary to attract future business. Thus, if a lawyer's failure to do any work on a matter is observable to potential clients, the attorney will make some effort so as to ensure that she can attract future clients. While in any given case it is impossible for a client to determine the exact quality of the legal services provided, if transactions on which an attorney works run into legal problems more than is average, such an attorney will over time develop a reputation for shoddy work. Conversely, if the clients of a particular attorney consistently do not encounter legal problems, then this attorney will develop a reputation for high quality work. The need to develop and maintain such a reputation thus promises clients that they will receive more than the minimal amount of work necessary to collect the fee, and clients will be willing to pay more given such assurances.

This desire to maintain a reputation puts a floor on the quality of the work that the lawyer will perform. Indeed, if a lawyer's reputation accurately reflected all of the attorney's previous efforts, a flat fee would in fact be the optimal compensation arrangement. This is because reputation would be a perfect substitute for direct observation of attorney action. Any deviations from optimal action would, in the long run, affect the attorney's reputation. The reputational constraint would thus induce the attorney to provide optimal information.

Reputation, however, is not a cure-all. One problem is that potential future clients may not have sufficient information regarding the past outcomes of projects in which the lawyer was involved. As we noted above, clients are not able to assess quality of work in anything other than a rough way. Ex post, they are not able to determine with much accuracy whether or not the information that they received was optimal. They only know whether or not they were able to achieve what they sought. While there is some correlation between the level of attorney effort and whether or not the client ultimately received a favorable outcome, the correlation is not that strong. There are many times where a lawyer may miss a legal risk and the risk never materializes; conversely, at times a transaction may encounter legal risks of which even the best attorney would not have been aware. Thus, reputation in and of itself is not sufficient to eradicate the incentive to understate legal risks inherent in a flat fee arrangement. [*411]

A second factor which induces lawyers to put forth a minimum level of effort regardless of the compensation scheme is malpractice liability. At some point, an inadequate assessment of the risks involved in a transaction might expose the lawyer to liability. n69 This threat has two components - as an inducement to put in sufficient effort to discover legal risks and the assessment of these risks that the lawyer gives to her client. As to the amount of effort the attorney expends, one would not think that the threat of liability would induce additional effort. Liability for attorneys is usually based either on fraud (i.e., the attorney knowingly turned a blind eye to client misconduct) or negligent misrepresentation (i.e., the attorney failed to perform an adequate investigation of the transaction). n70 Neither basis of liability would expand the amount of effort the attorney put into the project. Only the duty to avoid negligent misrepresentation has the potential for affecting the lawyer's effort. Given that liability tends to attach only in failed transactions (winners hardly ever complain), this duty to make a reasonable effort adds little to the incentive created by the need to maintain a good reputation. It certainly does not give an incentive to provide optimal information.

The threat of liability, however, might affect the attorney's communication with her client. In particular, it might bias the attorney in favor of overstating legal risks. Once a lawyer discovers that a legal risk exists, the lawyer can reduce her potential exposure by playing up the risk to the client. Thus, the threat of liability, while not inducing the lawyer to spend additional effort, would create the incentive to overstate legal risks to the client.

On balance, the flat fee arrangement gives the lawyer the incentive to put too little effort into any given project.

D. The Effect of the Overstatement of Legal Risks

The obvious question for economists at this point is why can't clients take all this into account? If we assume that clients have rational expectations, why can't they correct for the overstatement of legal risks? The incentives to overstate legal risks are known to the clients. If they know that their attorneys are overstating legal risks, [*412] should they not be able to discount such statements so that the information which they act upon is actually accurate? This question is important because it determines the societal impact of attorney overstatement of legal risks. If clients can discount the information that they receive, then we are talking only about a wealth transfer from clients to attorneys. The attorneys get more and the clients get less, but the transactions which the clients ultimately undertake are unaffected by the distorted information. Conversely, if clients cannot adjust for the attorney's incentive to overstate legal risks, then there are societal allocation questions at issue as well. If clients act based on inaccurate information, they will make worse decisions than they would if they knew the true state of affairs.

At times clients may in fact learn that lawyers are overly cautious. Yet it may be difficult to factor this knowledge into their decisions. The problem arises from the nature of legal advice. Consider the nature of the advice which the lawyer provides. She states that there are problems with the transaction as proposed, and suggests ways in which the risks can be reduced. It is hard for the client to know to what extent the risks in the original transaction have been overstated. Thus, given this uncertainty, it may be rational for the client to agree to the new structure.

This conclusion is reinforced by the nature of the action which the client wants to undertake. Most commercial transactions carry the potential for large gains to the client. Thus, they have a large incentive to have the transaction proceed. At the same time, given the amount of money at stake, the cost of legal liability, should there be such liability, is quite high. From the client's perspective, it is often the wiser course to steer clear of legal risks. This being the case, the client will base its actions on the information which it is provided, even if they believe that such information may contain an overstatement of risk.

II. BEYOND SELF-INTEREST: MORE SUBTLE EXPLANATIONS FOR LAWYERS' BIAS TOWARD OVERSTATING RISK

The foregoing account implies that, on average, lawyers' self-interest would lead them to bias their advice in the direction of undue caution. A predictable response is that this suggestion ignores the sorts of professional and personal motivations that counteract the temptation to prefer self over client. After all, loyalty to clients is [*413] central to the conventional image of what it means to be a lawyer. n71 Perhaps only a deviant segment of the bar would deliberately allow self-interest to affect the objectivity of advice to their clients.

Whether this retort is realistic is an empirical question that we cannot answer. However, we can consider the possibility that even lawyers who consider themselves fully committed to serving their clients might still be biased toward overestimation of risk. There are three complimentary explanations for why this might be so.

A. A Sociological Account: Possible Biases in

Lawyers' Norms

If lawyers do err too much on the side of caution, maybe it is not because they seek to cheat their clients but because that is what the profession teaches them is proper conduct through the socializing processes of training, experience and collegial self-definition. Professional norms - by which we mean the centripetal social forces that actually unite members of the profession, as opposed simply to the bar's official pronouncements - can readily legitimate conduct that might otherwise be subject to question. n72

For instance, lawyers might understand their proper role as one in which the primary duty is to warn clients about the presence of legal risks, to assure that no client ever underestimates the presence of prevailing legal rules or standards. Such an ideology would invoke an image of professionalism, a self-portrayal of the lawyer as one whose special expertise is in giving voice to the law's often obscure dictates, and whose job is keeping the client a safe distance from legal harm. [*414] Stress on identification rather than calibration - an emphasis on the possibility of sanction rather than its probability - could be learned as young lawyers observe more senior ones in client conferences and prepare endless redrafts of legal memoranda. If custom is styled in this way, a bias toward overstating risk would be expected. It might even be rationalized as the moral high road, invoking the image of the lawyer as a "double agent" who serves society while also serving the client. n73

Reference to norms alone, however, is not particularly satisfying. For norms to develop and persist, there must be some reason why. The appeal to professionalism is not particularly persuasive; the image of lawyers more beholden to the law than to their clients is of questionable descriptive accuracy, n74 and it is hardly clear in any event that loyalty to the law would justify cautionary excess. Unless clients somehow want their lawyers to have such a bias, n75 therefore, the more compelling explanation would be that such a bias is in the profession's self-interest. n76 Here, we simply revert to the economics-based account, with self-interest embedded in neutral sounding norms. [*415]

1. Overstating Risks as an Influence Technique

Even if the norm of loyalty to perceived client interests is strong, it does not necessarily imply total candor with clients. As with physicians, n77 the legal profession may unofficially accept a form of client manipulation when done in what appears to be the client's best interests. n78 In particular, overstatement of legal risks may be an acceptable means of influencing a client who otherwise seems insensitive to those risks.

In many circumstances, lawyers must compete for the time and attention of their clients. Especially in an organizational environment, with a noisy information flow driven largely by the stresses of immediate business need, carefully calibrated advice may not always get the deserved response. Standard communication theory suggests that in order to have a message heard, it must often be sharpened in such a way that gives it priority. n79 Fear is a standard sharpening technique. n80 Overstating legal risks may thus be justified as the only means of getting the client to take action, even if the result might be excessive precaution. n81

A simple illustration of this sort of altruistic sharpening can occur when a lawyer helps some entrepreneurs incorporate a business. Business people often resist paying attention to the non-revenue producing formalities associated with the corporate form: properly called meetings, minutes, etc. Inattention to corporate formalities has been invoked by the courts as a mechanism for piercing the corporate veil (i.e., making the entrepreneur/shareholders personally liable for corporate debts), although a careful look at the case law suggests virtually no risk that inattention standing alone will lead to the disregard of [*416] the corporate form. n82 Yet there are ample prudential and more subtle legal reasons why attention to formality is desirable, especially from the lawyer's perspective. Anecdotally, more than a few business lawyers try to instill fear in clients of the horrible consequences of lack of attention to procedural formality, simply to overcome their natural disinterest.

Deliberate sharpening of legal messages is probably most commonplace in the organizational setting. An in-house lawyer faced with the task of getting a highly bureaucratized client to respond to a new government regulation or recent court decision, for instance, probably feels justified in overstating the threat. n83 Of course, that motivation can easily be self-serving as well; the overstatement of legal messages within an organization approaches the blurry line between those intended in good faith to prompt action by the client and those strategically designed to maximize the status and resources of legal players within the enterprise. n84

2. The Nature and Form of Legal Advice

At the risk of digressing, we should observe that the ideological self-definition of the lawyer's role may even affect the form used by lawyers to give legal advice. In general, a rational client seeking legal advice about a proposed course of action should want a probability estimate: the chance that the action will be sanctioned (or the probabilities associated with a range of possible outcomes), and the expected consequences that would follow. With this, the client can make a straightforward calculation of the expected utility of the proposed action. The literature on the efficacy of legal rules assumes that [*417] advice is given in this fashion. No doubt it is the way some lawyers do counsel their clients.

Most, however, resist giving probabilistic advice. n85 Citing long-standing custom, they claim that the process of legal inference is too imprecise to quote odds in mathematical form. Some even raise ethical concerns about equating legal advice with betting odds. In practice, advice tends to be rendered within the framework of a more restrictive set of conventional locutions: sanction of the proposed course of action, for instance, might be said to be certain to occur; highly likely; n86 likely; uncertain; n87 unlikely; highly unlikely; or certain not to occur.

The dissonance between what a rational client should want and what lawyers typically offer raises an intriguing question. Most lawyers would concede that the conventional protocols do not capture perfectly the full range of possibilities - that many legal problems will generate answers that fall somewhere in-between. Indeed, there is not even a clearly defined common understanding within the profession about what the locutions mean (e.g., what degree of confidence is represented by the term "highly unlikely"). Even conceding the difficulty of quantifying subjective legal inference, there is a large body of learning on decision-making techniques, widely employed in business settings, that teaches people to use probabilistic reasoning in situations like these. n88

Then why don't lawyers quote odds? A benign possibility, decision-theory notwithstanding, is that clients in fact have no strong interest in more precise calibration. Studies of the actual risk taking [*418] behavior of business managers have indicated that most have little appreciation of probabilistic reasoning, and see risks largely as possibilities either to be managed or avoided if large enough in terms of potential loss. n89 To these sorts of clients, rough verbal calibration would be perfectly adequate. We should be aware also of a body of literature that supports a preference for verbal rather than mathematical representation of risk when numbers would suggest more precision than is reasonably possible. n90

But it is plausible as well that verbal representation is self-serving. One way is simple: the vagueness of the representations makes it more difficult to second-guess the advice when there has been a bad outcome. There is also a more subtle possibility, depending on how lawyers are trained to translate their particular subjective inferences into one of the customary locutions when there is not a perfect intuitive fit. One way is to construct some mental representation of what these conventions mean (e.g., what degree of confidence is associated with "highly likely") and then simply choose the convention closest to the inference. In probabilistic terms, for instance, if "uncertain" was a fifty percent chance, "likely" a seventy percent chance and "highly likely" a ninety percent chance, then a seventy-five percent assessment would be characterized as likely. This would not introduce any systematic bias toward overstatement: there would presumably be as much rounding up as rounding down.

A bias would arise, on the other hand, if instead the mental process of translating the intuitive sense of risk into one of the conventional locutions works on a threshold basis. Starting with "no risk", the lawyer gradually eliminates each locution that fails adequately to [*419] warn the client of the perceived risk. The cautionary bias would arise here because passing a threshold, even marginally, automatically moves the warning to the next highest degree of intensity. n91 Risk could be overstated, but never understated.

Without knowing more about how lawyers choose the protocols for representing risk in specific cases, we cannot say for sure what function those protocols perform. But it is tempting, in light of the economics-based account, to guess that the form of legal advice that lawyers customarily give is not random or accidental, but rather a norm that serves adaptively as a protective mechanism, mediated by some ambiguity in client demand.

B. A Psychology-based Account: The Possibility of Cognitive Bias

In his ruminations on client counselling, Wall Street lawyer James Freund observed that in his experience, self-serving legal advice is fairly common. But usually, the lawyer is unaware of the bias, sure that the advice is sound and objective. n92 Along these lines, we next consider the possibility that quite apart from any externally generated norms, the mental process of analyzing legal risk can itself distort the estimate.

In pursuing this from a theoretical rather than anecdotal perspective, there is a large body of learning from which to draw, an admixture of psychology and economics under the general headings of behavioral decision theory and social cognition. The behavioral literature operates under the assumption that people act with bounded rationality - that they do not always act as utility maximizers, but rather simply try to do their best with limited time, information and cognitive capacity. n93 One branch of this study has concentrated on the [*420] presence of cognitive heuristics: the mental shortcuts that people unconsciously use in problem-solving that sometimes lead to decisional miscalculation. These biases persist because they are adaptively efficient (strategies for managing too much information with too little processing capability, or in dealing with the absence of information); sometimes, too, they satisfy motivations of ego or emotion. n94

The question for us, then, is whether lawyers might be affected by any of the predictable biases suggested by the research in behavioral decision theory in a way that would lead them to systematically overestimate legal risks. Sadly, lawyers have not been the subject of much empirical study along these lines. n95 But many other expert professions have been the subject of study - especially doctors and other clinicians, but also accountants, managers and others operating in commercial environments often shared with lawyers. By most research accounts, experts show a disturbing tendency to rely on common heuristics, even when diagnostically useful statistical information is available. n96 Although they are frequently called upon to repeat similar inferential tasks, there is not the sort of learning from experience [*421] that causes them to adjust these habits of thinking. n97 We can speculate, at least, that lawyers are no less subject to cognitive processes that lead to predictable biases.

The research in this area is both voluminous and contingent. It does not offer a simple, grand theory of decision-making; instead, there are numerous traits found to be present in statistically significant numbers, under the right circumstances. Researchers generally concede that decision-making is highly task-specific, n98 varying considerably based on the person and the situation.

Fortunately, we need not survey all the predictable tendencies shown by persons engaged in tasks comparable to those performed by lawyers in assessing legal risks. By and large, the research on how experts make decisions under conditions of ambiguity shows only that those choices will frequently be inaccurate, without suggesting anything about the direction of the bias. For example, people have a robust tendency to be overconfident in their judgments. n99 But that by itself would simply suggest that an attorney is as likely to overstate the absence of risk as its presence. Similarly, the inclination to find trends where correlations in available data are illusory n100 says nothing about which way the trends are moving. Giving too much weight to highly salient or available informational cues, or attending too much to the way a problem is framed, n101 are alone equally consistent with over and underestimation.

On the other hand, some segments of this research do point toward a directional bias. [*422]

1. Sources of Underestimation

In some circumstances, there may be a tendency to underestimate risk. For example, the notion of cognitive conservatism suggests that people resist information that suggests change rather than stability. In assessing facts, then, a lawyer might be insufficiently attentive to small changes as a situation evolves, even though those may indicate greater legal risk. More pointedly, a lawyer committed to a client's representation may well be motivated by ego to filter out information that indicates that what he is doing is wrong. Both of these highly situational risk-desensitizing tendencies have been explored in more detail elsewhere. n102

Another situational tendency relates to the so-called "illusion of control" n103 or bias toward overoptimism. People (especially experts) often exhibit excessive confidence in their ability to avoid negative outcomes when they sense some control over the event. This suggests that legal risks that are within the control of the attorney - ones, for example, where the evaluating attorney will also be assuming a significant strategic or advocacy role - might well be underestimated. Indeed, self-serving assessments by litigators of cases they were handling has been documented. n104 Obviously, this situation is not uncommon among counsellors as well.

Here, however, we must be careful. Embedded in this tendency may be a reasonably objective estimate (if not an overestimate) of the legal risk as an external threat, to which the attorney adds a relatively self-serving assessment of his own involvement. The effect, then, may be an inflated cost to the client in the form of greater legal services [*423] than might otherwise be required - something more resembling overstatement of risk than understatement.

More generally, we must also take account of the finding that people tend to be relatively insensitive to very low probability risks, even when they are catastrophic upon occurrence (except temporarily in response to highly salient, vivid warnings). n105 This might suggest that people systematically underestimate those kinds of risks, and leads us to speculate that lawyers, too, might be insufficiently attentive to small legal risks faced by their clients. However, there are two reasons to doubt how significant this bias is in the domain of lawyering. First, the underestimation hypothesis is not the only explanation for such insensitivity; it may be that people simply disregard very low probability risks in order to maintain a constructive, less stressful outlook. Second, the literature supporting this sort of insensitivity does not deal with situations where the actor is motivated by the task itself to search for risks, as lawyers presumably will be. n106

2. Sources of Overestimation

a. Approaching Ambiguity: Stimulating the Imagination:

Decision theory, both behavioral and conventional, posits that the process of inference - such as the assessment of risk - begins with the generation of a hypothesis. The lawyer's first impression when asked for advice, for example, might become the initial working hypothesis; alternatively, the lawyer might have no idea about the answer, and simply begin with the hypothesis that the odds of sanction are even. The hypothesis is then adjusted as new information is considered, until closure. n107 Here, behavioral decision theory departs from the conventional subjective expected utility model largely in its claim that the initial hypothesis - the anchor - often exerts a disproportionate influence on the final decision.

In this light, the first question to explore is how the cognitive task of legal analysis might bias either the adoption of the anchor or a process of adjustment in a risk-positive direction. A specialized body of research addresses specifically how people react to ambiguity when assessing risks or making decisions. Foremost is the work of Hillel [*424] Einhorn and Robin Hogarth, who have constructed a theory of behavior to describe the mental process of confronting ambiguity. n108 The Einhorn-Hogarth ambiguity model deals with the nature of the adjustment process that occurs after a tentative hypothesis is generated by prior experience, expert advice or available information. Adjustment then occurs through a process of mental stimulation or imagination, as alternative possibilities are considered. The amount of imagination depends on the amount of perceived ambiguity: the more difficult the assessment, the more the mind dwells on it. The direction of the adjustment is dependent on the actor's attitude toward ambiguity. Although this variable, too, is highly situational, the common tendency is to dwell on - and thus overweigh - negative outcomes, a bias toward caution that the authors refer to as "defensive pessimism." n109

This model is interesting for our purposes because the task of legal inference is often characterized by high ambiguity. n110 We suspect that lawyers' attitudes toward ambiguity - at least in the absence [*425] of an illusion of control - reflect the common tendency toward caution for two reasons. One, emphasized by Einhorn and Hogarth, is that such an inclination is particularly likely when the decision-maker faces a risk of loss. n111 A transaction found unlawful involves a distinct loss to the client (not to mention the lawyer n112). Second, as we have seen, lawyers' norms are likely to generate a decision frame that prompts a diligent search for risk.

The idea that ambiguity can prompt people to dwell on and thus overestimate risks is perfectly intuitive. People (like lawyers) who are paid to worry will find something to worry about. Once mental activity produces this sort of special attention to risk-positive information, moreover, the drift toward overestimation can accelerate. A fair body of evidence supports the idea that the simple act of imagining a possibility increases a person's estimate of the likelihood that the possibility is true. n113 The mental activity of constructing a causal explanation for why the possibility might be true itself creates a mild bias in favor of the focal hypothesis. This may have a special relevance to the legal [*426] advice setting, since the product of additional thought given to a problem is often increased sensitivity to the indeterminacy of the prevailing legal authority. Dwelling on and testing plausible explanations for the possibility that the client might lose can readily diminish confidence that the client will win.

b. Accountability:

A lawyer's attitude toward ambiguity is also likely to be influenced by the accountability he faces for a wrong decision. Some of the criticism directed against the behavioral research - especially its heuristics and biases branch - is based on the artificiality of the settings in which hypotheses are tested; in particular, that subjects in experiments have little motivation to perform well, and which itself can lead to apparently careless thinking. As a result, more attention has been given in recent years to the role of accountability in decision-making. n114

The predictions generated by the accountability research are not surprising. As expected, a decision-maker who expects to be evaluated is likely to engage in a more careful search of information and alternatives, with a view - perhaps conscious, perhaps not n115 - toward improving the chances that she will be favorably evaluated at the appropriate time. Many common cognitive biases, including the tendency toward overconfidence and the availability/representativeness heuristics, are reduced or eliminated. n116 On the other hand, there is evidence that the ultimate quality of the decision sometimes suffers because of accountability. Actors may overload in their information searches, for instance, thus diluting the effect of the most important evidence. n117 Or, too much attention to what others might think can [*427] distort the process of inference, given the difficulty of making that judgment.

Lawyers who give legal advice face two separate forms of accountability. Their advice may be evaluated on an ex ante basis by other lawyers. This is the case where multiple attorneys in a firm work on or review the same project, or there are other lawyers involved in the transaction. There is also ex post accountability - primarily by the client - once the accuracy of the lawyer's prediction is tested.

Ex ante evaluation may counteract the tendency to overestimate risk. But for this to happen, the reviewing lawyer(s) must (1) be in a position to critically examine the decision (i.e., have enough experience, or do enough separate research to have expertise comparable to the initial decision-maker) and (2) not be subject to the same biases or incentives as the initial decision-maker. So stated, we can guess that this form of accountability will be infrequent. Indeed, within a firm there is often a division of labor among attorneys that diminishes the ability to review except for gross errors, and more importantly, the same incentives will operate broadly throughout the entire team. Under these circumstances, conformity pressures may well exacerbate the bias toward overestimation, n118 not reduce it.

Ex post, the situation is quite different. Here, however, we simply return to the economics-based story. As we have seen, there is a predictable asymmetry in clients' ability to observe errors in the quality of legal advice. From this we would conclude that ex post accountability is a serious concern with respect to underestimation of risk, less so with respect to overestimation. If so, then the learning on accountability readily supports the possibility of a bias toward overestimation.

Asymmetric accountability leads to another prediction. Consistent with the intuitions of most laypeople, there is substantial evidence that knowing the ultimate outcome generated by a set of circumstances alters the way those circumstances are viewed. In what is [*428] known as the hindsight bias, the predictability of that outcome is overestimated. n119

To the extent that lawyers anticipate a hindsight bias, the asymmetry of accountability tilts even further. When advice leads to adverse consequences, it is not unreasonable to fear that the ex ante quality of the advice will be evaluated in an overly harsh manner. The risks will seem more apparent in hindsight. Once again, this sort of fear - of which the lawyer may or may not be consciously aware - can easily affect the cognitive path of hypothesis and adjustment, causing that much more attention to risk-positive information.

c. Self-Interest and the Role of Ego:

The accountability influence brings us to a broader, and perhaps self-evident possibility: that risk perception may unconsciously be biased by self-interest alone. We see what we want to see, and rationalize in objective terms that which is simply desired. n120 In the legal setting, lawyers find risk when they benefit from its presence. That is James Freund's diagnosis, n121 and the behavioral literature offers ample supportive evidence, focusing largely on the role of ego in cognition. [*429]

Ego (or the need for self-esteem) exerts a powerful influence on nearly all cognitive processes. The strong desire to have a positive self-concept - to view oneself as, among other things, rational and responsible - prompts the subconscious tendency to deflect self-critical information and create often illusory accounts of one's successes and failures. n122 High self-esteem, even if supported by a web of self-serving illusions, is associated with numerous positive traits, which helps explain why conceit (at least in its relatively abashed form) is so pervasive and adaptive, n123 particularly among lawyers.

Ego can potentially be biasing in a number of ways. There is tendency - underscored in the research on anticipatory regret in decision-making n124 - to make choices in a way that bolsters both one's external and self-image. In other words, the same subconscious risk-sensitive tendencies predicted by the accountability research could be generated simply by one's own internal fear of being responsible for an observable error. Conversely, lawyers may enhance their own self-esteem (not to mention external reputation) by habitually overstating legal risk and then assuming too much credit for predictably positive outcomes. n125 [*430]

There is another, more speculative possibility as well. Put simply, a lawyer's status n126 in a client interaction is elevated by the assumption of dominance and control in that relationship, n127 and the leverage a lawyer has to achieve that status is the threat of legal risk. By using it, the lawyer can take charge and displace the client's apparent autonomy. Take, for example, a lawyer asked to give a board of directors legal advice in resisting a hostile takeover. Viewed dispassionately, the prevailing law might well give the board substantial discretion to employ defensive tactics of their own choosing. It is not far-fetched, however, to imagine a lawyer motivated to see more risk precisely because of the enhancement to his role that comes with that inference. n128 Here, of course, the dominating incentive is not so much to overstate risk in order to deter client action, but rather to utilize the concern about risk to exert control and cause a modification or restructuring of the client's course of action.

Were any of these instances of overstating risk done deliberately, it would simply be an illustration of the self-serving behavior of the sort predicted by the economics-based story. The point here is the motivation toward self-enhancement may cause a lawyer to engage in precisely the same behavior while maintaining the stress-reducing belief that he is comporting with the norm of loyalty to the client's interests. n129

C. An Information-based Account: Possible Biases in Legal Resources

As we observed at the outset, the process of assessing legal risk in advising clients is a subjective and difficult one. While some questions [*431] posed by clients admit to clear-cut answers, many, perhaps most, do not. Inevitably, legal authority is at least partly indeterminate in that small samples of prior "like cases" may be insufficient to permit extrapolation with complete confidence, unique facts and circumstances often make the identification of "like cases" problematic in any event, and legal decision-makers cannot always be counted on to act in the predicted fashion. Some caution, then, in providing legal advice is perfectly rational, and lawyers who express extreme confidence in their conclusions may be driven more by the need to play the charade of expertise than the desire to convey risk accurately.

In formulating advice, a lawyer will draw on a variety of sources. Personal experience with the same or similar questions, comparable experiences of close colleagues, research into precedent and authority, and information generated by secondary sources (ranging widely from books and treatises to presentations at continuing legal education seminars) are bits of data that will be configured into the cognitive map from which the inference is finally drawn.

This raises the possibility that even a diligent effort to assess legal risk will be distorted if the availability of the underlying data is subtly skewed. A dominance of risk-positive information will make risk-positive inference easier to form and justify. n130 In this section, we look at possible causes of such bias in the resources from which legal risk assessments are drawn.

1. Experience

Personal and collegial experience is no doubt a dominating influence in how lawyers assess risk. n131 In light of the now highly specialized nature of legal practice, experience with somewhat comparable issues and problems is likely; advice previously rendered (and perhaps tested) can readily be recalled and will establish a strong decision frame for the current task. This sort of information is highly salient and available. [*432]

Because of the highly individualized and situational nature of experience, generalization about its potentially biasing effect is difficult. A prior successful experience with a comparable issue (i.e., one in which no harm came) will by itself be risk desensitizing, and visa versa. To the extent that advice has been rendered previously in one direction or another, moreover, the desire for consistency will be strong.

The potential here for a predictable direction to any bias is therefore relatively weak. If anything, it may be risk-negative, since on balance lawyers will probably have had more experience with successful transactions and occurrences than unsuccessful ones. n132 An offsetting factor, however, might be the tendency of people to attribute prior successful experiences to skill even when the dominating influence may be the external circumstances or simple luck. n133 Lawyers whose previous successful experiences involved significant involvement in the matter may be motivated to remember those experiences as ones with substantial risks that were managed well, and import that perception to the problem at hand. Consequently, a lawyer who took a client through discussions with the SEC staff about some disclosure issue might recall a significant risk deflected through careful negotiation and advocacy; the staff might remember the same discussions as routine. When a similar issue then arises, it is the enhanced perception that resides in memory.

2. Primary Authority

Even with the advent of computerized legal research, the data available to the legal analyst is limited in a number of significant ways. The body of published judicial opinions is readily available, of course. But many disputes are resolved without published opinions on the merits. Information about cases brought but dropped, settled or adjudicated privately - a large bulk of the litigation process - is usually incomplete and difficult (if not impossible) to obtain systematically. [*433] A careful risk analyst would also be interested in cases never brought at all, but these are usually wholly unobservable.

As has been well recognized in the literature, the sample of reported cases is likely to be biased toward the "hard" ones, leading some to predict a roughly 50/50 split between dispositions for plaintiffs and defendants on the merits, regardless of the prevailing standard of law. n134 This selection bias can easily disorient the analyst, n135 creating at least the impression of greater ambiguity and risk even where the objective standard might be fairly pro-defendant. n136 Furthermore, many published opinions in civil litigation are dispositions of motions to dismiss or motions for summary judgment. Because these motions are resolved by assuming the plaintiffs' facts to be true, whether they are or not, those dispositions can, in bulk, take on an especially pro-plaintiff (and hence risk-positive) tint. Perceptions may well be distorted toward overestimation of risk. This possibility has been noted specifically, for instance, in the securities law area, where perceptions about what constitutes "due diligence" under section 11 of the Securities Act of 1933 may well have been inflated over time by an extraordinarily small sample of judicial opinions that began with one major pro-plaintiff disposition. n137 [*434]

3. Secondary Authority

The most interesting influence of the available data set occurs in secondary authority. Books and treatises, law review articles and opinion pieces in other media, widely disseminated "To Our Clients and Friends of the Firm" memoranda, and continuing legal education presentations all operate as means whereby the members of the profession construe the law in order to influence the legal risk perceptions of others.

Various biases arise in the professional literature, apart from the simple retransmission of those previously identified. As is often observed, even in the area of academic scholarship, choices as to what is published naturally tend to favor the interesting, and risk tends to be more interesting than its absence. n138 As we have already seen, to attract an audience in a noisy informational environment, messages must be sharpened, ambiguity leveled away. n139 In non-scholarly media, these market-based incentives grow all the stronger.

Economic self-interest also affects what is published, especially in the "practical lawyering" literature. Lawyers often view such writing as a business-generating device, creating an incentive to overstate risk as a means of increasing demand for the particular lawyer's services. n140 Even without conscious realization, lawyers whose writing or speaking deals with the management of legal risk are motivated to inflate the seriousness of those risks in order to justify the claim that substantial skill and expertise is required to deflect them.

This suggests that many bits of information in the body of secondary advice and authority can overstate or overemphasize risk-positive information. What is especially interesting, however, is to consider the dynamic character of the professional construction of law. Lawyers are sensitive to the opinions of others. With law so subjective, and the economic incentives (in terms of both accountability and leverage) so strong, there is a rational inclination for a lawyer who observes another express the presence of serious risk on some issue to [*435] take that seriously, even if he might otherwise come to a weaker conclusion. n141 Identification of risk can be contagious, then. It can lead to a cascade n142 of risk-positive characterizations as such views are retransmitted - perhaps sharpened and leveled - and observable legal advice begins to reflect this information, further enhancing its apparent credibility and authority. Some of this cascading, of course, will be prompted by pure self-interest. Those who benefit from increased perception of risk will find it worthwhile to identify and emphasize its growing acceptance within the professional community. But it can be driven just as much by the more benign process of conformity. n143

One can identify a number of issues to illustrate this cascading risk phenomenon. In a recent study, for example, Lauren Edelman, Steven Abraham and Howard Erlanger analyzed the professional construction of wrongful discharge law. n144 Comparing case law to the tone of warnings about the ostensibly pro-plaintiff state of the law in a sample of articles in law reviews, practical lawyer-oriented publications and personnel management publications, they identified what they considered to be a significant overstatement of risk. (Law reviews were the least biased; the non-lawyer media the most.) Tentatively, their study attributes this to rent-seeking, as both lawyers and personnel managers seek increased status and wealth. Whatever the explanation, it is easy to see how a particular lawyer advising a client, wholly in good faith, would nonetheless be likely to pass on this inflated threat to the client. With so much being written and said about the expansive threat of wrongful discharge law, the lawyer would be unwise to ignore it, no matter what her assessment of the case law. And once some clients begin reacting - by establishing [*436] observable new procedures for dealing with the risk, for example - the pressure for others to follow grows even stronger.

The securities laws, as noted earlier, offer another illustration. One early district court case (arguably somewhat extreme on its facts) under section 11 raised the possibility of substantially heightened due diligence responsibilities in connection with public offerings. n145 That immediately generated a substantial mass of commentary in both the law reviews and practice-oriented publications, most of it somewhat alarmist. n146 One can only speculate, but that professional construction has probably layered substantial additional costs (much of it in legal fees) onto the capital raising process. Later case law, more accommodating of issuers and their associates, n147 has not been met by any comparable cascade of information suggesting a racheting down of perceived responsibilities. And, of course, we must note that such professional construction can easily become a self-fulfilling prophecy. As advice conforms to perception, the resulting prevalence of cautionary activity gradually becomes a standard by which those who fall below are sanctioned.

D. Joinder

If each of these more subtle accounts for the overstatement of risk is plausible by itself, their interplay is likely to be all the more significant. In sum, norms can prompt lawyers to dwell on risk. The cognitive act of dwelling in turn strengthens risk perception, as do the various motivational influences. The legal information base then accommodates these biases by offering an ample quantity of risk-positive information to find. Sometimes, there can be a contagion of risk sensitivity as lawyers look to one another for cues. The whole process is readily rationalized as in clients' best interests, and remains stable because it is so consistent with the profession's economic interest. [*437]

III. CONCLUSION

Hopefully, we have demonstrated that (a) business lawyers have multiple incentives to overstate risks, (b) institutional mechanisms are unlikely to ameliorate these incentives completely, and (c) a host of social and psychological influences create a plausibility structure that allows individual lawyers to avoid feeling disloyal in so doing. In other words, the climatic conditions are right for overstatement to break out with predictable frequency. To be sure, overstatement of risk should occur more frequently in some professional climates than in others. The range of settings in which business clients seek legal advice is so broad and diffuse that generalization is difficult. However, we can suggest a number of factors, under four general headings, that seem particularly likely to affect the incidence of overstatement.

A. The Nature of the Client

The client's attitude toward the matter in question and related risk is no doubt quite important. A client that communicates to the lawyer a willingness to tolerate risk because of the significance of the matter to it (e.g., an aggressive tax shelter promoter) may soften a fear of reputational loss should the project fail. Conversely, the more the client projects a fear of legal sanction, the more the lawyer is able to rationalize overstatement (and the more the lawyer probably has to fear in the event of error). For overstatement to succeed, moreover, the client must not be able to detect it. Thus, the client's ability to monitor is also significant. The more sophisticated the client legally, the less likely it is that overstatement will occur. However, we should be cautious here; when monitoring on behalf of an organizational client is by in-house counsel, the "watchdog's" own incentives may be skewed, and in-house counsel might actually abet excessive caution. n148

B. The Nature of the Lawyer or Firm

Theoretically, a lawyer with a broad portfolio of clients (or in a firm with such a portfolio, if firm-wide profitability is the key element of compensation) has less of an incentive to overstate risk than one dependent for both income and reputation on a small client base. In a [*438] competitive environment, however, these conditions hold with decreasing frequency. n149

C. The Nature of the Subject and the Underlying "Actual" Risk Level

Overstatement will occur most frequently when the underlying actual risk level is at least colorable, but not high. (When the risk is already high, overstatement is of little marginal value to the lawyer.) Presumably, overstatement is easier to accomplish and rationalize when the law is relatively more indeterminate or uncertain. Thus, it will occur more frequently when the law is articulated as a standard than as a rule. n150 As a separate point, the greater the profession's sensitivity is to the matter generally, the more likely it is that the lawyer's informational base will be weighted with risk-positive information. Hot topics will produce more caution than less visible ones.

D. The Nature of the Advice

The fee arrangement with the lawyer will affect the incidence of overstatement; hourly fees create the largest susceptibility compared to flat or contingent fees. So will the financial consequences of a positive or negative risk assessment. Where risk-aversion will result in a loss of business to the lawyer, overstatement is less likely; where it can justify more work, it is more likely. Also, the more private and customized the advice, the less likely the attorney will fear observation by other lawyers who might have the motive and opportunity to expose conscious or unconscious cheating. In contrast, overstatement is less likely when the client is proposing a step that other legal actors are likely to take, so that the effects of differing forms of advice will be observed. n151 Hence, a company considering some innovation that it suspects its competitors may also try will probably get fairly objective [*439] advice about its legality. Similarly, overstatement would be discouraged were a client to conduct a "beauty contest" - asking for competing, confidential submissions ex ante by firms interested in handling the particular matter as to their perceptions of risk and how it might be handled.

Often, of course, these factors will conflict, offsetting each other. n152 Then, prediction is difficult. But one can readily imagine many counselling settings when they will palpably coalesce toward overstatement. A company, for example, wishes to consider a particular liquidation strategy that risks adverse tax consequences because of the nature of the holdings of a dominant shareholder. Other, somewhat less attractive, liquidation options are available. Here, the advice is likely to be highly confidential and fact-specific, making detection of overstatement unlikely. And because of the other options, the lawyer will not face an income-related penalty from foregoing the strategy in question. Absent unusual pressure, we would not be surprised were the lawyer to discourage the course of conduct even if, from a cost-benefit standpoint to the client, the risk was worth taking.

If our analysis holds up to scrutiny and testing, what then? First, as we noted at the outset, we would hope that the scholarly literature would enlarge its focus to recognize the triangular nature of the conflict of interest that arises in the counselling setting, adding the lawyer's distinct self-interest (and resulting self-definition) to the more commonly stressed interests of client and society. We would also hope that the idea of a filtration bias in the transmission of legal knowledge would be taken seriously in assessments of the efficacy of legal strategies, such as those found in the debate over the varying uses of rules and standards in formulating the law.

The message to the legal profession is a bit more sensitive. Nothing in our analysis is in the nature of lawyer-bashing. Our approach has simply assumed that attorneys are human, and thus subject to the same economic, social and psychological influences as people generally. While "debiasing" may be difficult given the complex web of [*440] motivations, n153 it might be useful for the profession (or at least its scholars and educators) to recognize the phenomenon and discuss it openly.

Do lawyers overstate legal risk? We still don't know for sure. But in the end, it would be surprising if they didn't. The presence of legal risk generates both income and status for lawyers. Hence, expertise in divining it becomes a large part of the self-identity of the profession. Legal risk is what we are taught to pay attention to. And when people look for something that they are accustomed to seeing, they are usually able to find it.

FOOTNOTES:

n1. See Ronald J. Gilson, Value Creation by Business Lawyers: Legal Skills and Asset Pricing, 94 Yale L.J. 239 (1984). A popular expression of antilawyer sentiment, making the undue caution claim, is Mark McCormick, The Terrible Truth About Lawyers (1987).

n2. E.g., John C. Coffee, Jr., Understanding the Plaintiff's Attorney: The Implications of Economic Theory for Private Enforcement of Law Through Class and Derivative Actions, 86 Colum. L. Rev. 669 (1986).

n3. See, e.g., Geoffrey P. Miller, Some Agency Problems in Settlement, 16 J. Legal Stud. 189 (1987).

n4. See, e.g., Lisa G. Lerman, Lying to Clients, 138 U. Pa. L. Rev. 659 (1990).

n5. See, e.g., Stephen McG. Bundy & Einer Elhauge, Knowledge About Legal Sanctions, 92 Mich. L. Rev. 261 (1993); Louis Kaplow & Steven Shavell, Private versus Socially Optimal Provision of Ex Ante Legal Advice, 8 J.L. Econ. & Org. 306 (1992); Steven Shavell, Liability and the Incentive to Obtain Information About Risk, 21 J. Legal Stud. 259 (1992).

n6. See, e.g., Robert W. Gordon, The Independence of Lawyers, 68 B.U. L. Rev. 1, 26-30 (1990); William H. Simon, Ethical Discretion in Lawyering, 101 Harv. L. Rev. 1083 (1988).

n7. The problem is also identified, but not extensively explored, in Robert C. Clark, Why So Many Lawyers? Are They Good or Bad?, 61 Fordham L. Rev. 275, 285-86 (1992). To the extent that a problem has been perceived, it tends to be that lawyers will overbill by spending too many hours generating advice, not that the advice itself will be biased. This point is underscored by a recent, thought-provoking symposium on value creation by business lawyers, published in the Oregon Law Review. See Symposium: Business Lawyering and Value Creation for Clients, 74 Or. L. Rev. 1 (1995). In nearly 350 pages, the agency cost problem with respect to counseling was barely mentioned, much less explored. See Ronald J. Gilson & Robert H. Mnookin, Foreword: Business Lawyers and Value Creation for Clients, 74 Or. L. Rev. 1, 12 (1995) (noting in passing that lawyers as agents may be "villains"); Frederick W. Lambert, A Preliminary Inquiry into the Transcendence of Value Creation, 74 Or. L. Rev. 121, 131 (1995) (describing tension between the goals of a lawyer's clients and that lawyer's firm).

n8. See Meir Dan-Cohen, Decision Rules and Conduct Rules: On Acoustic Separation in Criminal Law, 97 Harv. L. Rev. 625 (1984).

n9. This informational asymmetry has been identified in numerous studies of lawyer-client interactions. See, e.g., Ronald J. Gilson, The Devolution of the Legal Profession: A Demand Side Perspective, 49 Md. L. Rev. 869 (1990) (discussing growth of in-house counsel as a response).

n10. The mere fact that a transaction has been struck down does not necessarily mean that the lawyer erred; the decision might be aberrational, for example. But since clients lack any other means of assessing quality, such unfortunate events are the best available - if imperfect - evidence, and hence are likely to have a significant impact on the perception of quality as the information is retransmitted. See Abhijit V. Banerjee, A Simple Model of Herd Behavior, 107 Q.J. Econ. 797 (1992).

n11. In fact, the temptation to overstate risk may well vary in relation to actual or objective risk. We suspect that the temptation is trivial where risk is in fact large: a lawyer gains relatively little either in opportunity for business or reputational protection by overstating an already high-level risk. And presumably there are some risks that are so low that overstatement is difficult. The place where overstatement is most likely is where the risk is perceptible but not probable. Were we to portray this graphically, the curve would have something of an "S" shape.

n12. See Robert C. Ellickson, Bringing Culture and Human Frailty to Rational Actors: A Critique of Classical Law and Economics, 65 Chi.-Kent L. Rev. 23 (1989); see also the articles collected in The Behavioral Foundations of Economic Theory, 59 J. Bus. S181 (Robin M. Hogarth & Melvin W. Reder eds., 1986).

n13. A general survey of the conditions under which overstatement is more or less likely is presented in the conclusion. See infra Part III.

n14. See Anthony D'Amato, Legal Uncertainty, 71 Cal. L. Rev. 1 (1984); Ken Kress, Legal Indeterminacy, 77 Cal. L. Rev. 283 (1989).

n15. For illustrations (which may or may not account for bias), see Douglas E. Rosenthal, Lawyers and Clients: Who's in Charge? 204-05 (1974); Gerald R. Williams, Legal Negotiation and Settlement 113-14 (1983). Self-serving biases have been identified in the litigation and settlement context. See infra note 20 and accompanying text. Thus, these illustrations must be used carefully.

n16. One possible response to the reputational threat identified above is not that lawyers will overstate risk, but that they will dilute their advice sufficiently so that blame can be avoided later on. While this is possible, there is a natural limit on how much dilution can occur without rendering the advice patently unusable. This topic is considered further at infra Part II.B.1.

n17. Presumably, were a control group of lawyers asked to evaluate a problem free of any of the biases we identify (i.e., in a laboratory experiment), some clustering of responses would occur so that a proxy for the "right" answer could be generated.

n18. See, e.g., Sanford J. Grossman & Oliver D. Hart, An Analysis of the Principal-Agent Problem, 51 Econometrica 7 (1983); Milton Harris & Arthur Raviv, Optimal Incentive Contracts with Imperfect Information, 20 J. Econ. Theory 231 (1979); Bengt Homstrom & Paul Milgrom, Multitask Principal-Agent Analyses: Incentive Contracts, Asset Ownership, and Job Design, 7 J.L. Econ. & Org. 24 (1991) (special issue); Stephen A. Ross, The Economic Theory of Agency: The Principal's Problem, 63 Am. Econ. Rev. 134 (1973); Steven Shavell, Risk Sharing and Incentives in the Principal and Agent Relationship, 10 Bell J. Econ. 55 (1979).

n19. For examples of this, see the articles cited in the preceding note. See also Saul Levmore, Commissions and Conflicts in Agency Arrangements: Lawyers, Real Estate Brokers, Underwriters, and Other Agents' Rewards, 36 J.L. & Econ. 503 (1993) ("A common theme in resolving agency problems is that rewards are an important means of reducing agency costs.").

n20. There is an extensive body of literature on the economics of litigation. See, e.g., Lucian A. Bebchuk, Litigation and Settlement Under Imperfect Information, 15 Rand J. Econ. 404 (1984); Avery Katz, The Effect of Frivolous Lawsuits on the Settlement of Litigation, 10 Int'l Rev. L. & Econ. 3 (1990); William M. Landes, Sequential Versus Unitary Trials: An Economic Analysis, 22 J. Legal Stud. 99 (1993); Jennifer F. Reinganum & Lois L. Wilde, Settlement, Litigation, and the Allocation of Litigation Costs, 17 Rand J. Econ. 557 (1986). As Gilson and Mnookin point out, these studies often overlook the role that lawyers play in the litigation process. See Ronald J. Gilson & Robert H. Mnookin, Disputing Through Agents: Cooperation and Conflict Between Lawyers in Litigation, 94 Colum. L. Rev. 509, 510 (1994).

n21. See Gilson & Mnookin, supra note 20, at 541-46.

n22. For a general examination of the forces which encourage the existence of large law firms, see Ronald J. Gilson & Robert H. Mnookin, Sharing Among the Human Capitalists: An Economic Inquiry into the Corporate Law Firm and How Partners Split Profits, 37 Stan. L. Rev. 313 (1985).

n23. For a similar definition of ideal attorney performance, see Earl Johnson, Jr., Lawyers' Choice: A Theoretical Appraisal of Litigation Investment Decisions, 15 Law & Soc'y Rev. 567 (1980-81) (ideal lawyer "will invest additional resources ... in a given case ... until maximum net benefits are achieved for the client").

n24. Of course, in terms of social welfare, we want the attorney to gather only optimal information. Information which is not optimal under our definition represents a waste of societal resources. To be sure, at times a party may have an incentive to gather more information than is socially desirable. For example, when a party seeks advice about a course of action which may lead to tort liability, if a negligence regime is in place the party will gather more information than is socially beneficial. See Kaplow & Shavell, supra note 5.

n25. We explore below the problems inherent in attempting to translate a contingent fee arrangement from the litigation setting to the transactional setting at which we are looking.

n26. See Patricia M. Danzon, Contingent Fees for Personal Injury Litigation, 14 Bell J. Econ. 213 (1983). This assumes, of course, that the attorney has control over the percentage of the recovery the plaintiff receives. If she does not, then she will adjust her effort so that she is paid a competitive wage. See Murray L. Schwartz & Daniel J.B. Mitchel, An Economic Analysis of the Contingent Fee in Personal-Injury Litigation, 22 Stan. L. Rev. 1125 (1970).

n27. For an anecdotal account of such deliberate overbilling, see Darlene Ricker, Greed, Ignorance and Overbilling, A.B.A. J., Aug. 1994, at 62.

n28. One example of this situation may have been Wachtell, Lipton & Katz in the 1980s. They were able to bill their clients based on a percentage of the deal rather than the hourly rate. This suggests that they were not working in a competitive environment.

n29. See Gilson & Mnookin, supra note 22, at 358-60.

n30. Id.

n31. See Kenneth J. Arrow & Frank H. Hahn, General Competitive Analysis 75-106 (1971); Gerard DeBreu, Theory of Value: An Axiomatic Analysis of Economic Equilibrium 94-96 (1959); David M. Kreps, A Course in Microeconomic Theory ch. 8 (1990).

n32. Cf. Eric Rasmusen, Games and Information: An Introduction to Game Theory 133 (1989) ("It used to be that an economist's generic answer to someone who brought up peculiar behavior that seemed to contradict basic theory was 'It must be some kind of price discrimination.' Today, we have a new answer: it must be some kind of asymmetric information.").

n33. See Gilson & Mnookin, supra note 22, at 373-78.

n34. See, e.g., Gilson, supra note 1; McCormick, supra note 1.

n35. See William G. Ross, The Ethics of Hourly Billing by Attorneys, 44 Rutgers L. Rev. 1, 96 (1991) (reporting that 92.4% of corporate counsel surveyed stated that the outside lawyers they hired billed predominantly on a per-hour basis).

n36. See George D. Hornstein, Legal Therapeutics: The "Salvage" Factor in Counsel Fee Awards, 69 Harv. L. Rev. 658, 660 (1956) ("when hours become a criterion, economy of time may cease to be a virtue"); Kenneth Robert, The Hourly Fee System is a "Devilish Creature", in Beyond the Billable Hour 36 (Richard C. Reed ed., 1989) ("Under the hourly fee system law firms have a financial incentive for inefficiency. They can profit from unneeded work."); William H. Rehnquist, The Legal Profession, 62 Ind. L.J. 151, 155 (1987) ("if one is expected to bill more than two thousand hours per year, there are bound to be temptations to exaggerate the hours actually put in").

n37. See Ross, supra note 35, at 93 (40.3% of lawyers surveyed believed that legal bills are "padded" occasionally or frequently); id. at 96 (55.9% of corporate counsel believe that such padding occurs).

n38. The Model Code of Professional Responsibility requires that attorney's fees be "reasonable." See Model Rules of Professional Conduct Rule 1.5 (1995). The ABA has recently stated that this implies that "a lawyer may not bill more time than she actually spends on a matter." ABA Comm. on Ethics and Professional Responsibility, Formal Op. 93-379 (1993). Not only does this cover deliberative overstatement of the number of hours spent on a matter, but also practices such as billing two clients for the same research, and billing one client for time spent in travel while billing another client for research actually done while in transit. See id.

n39. See Abram Chayes & Antonia H. Chayes, Corporate Counsel and the Elite Law Firm, 37 Stan. L. Rev. 277, 292 (1985) ("legal bills are scrutinized and almost universally must be approved by legal departments"); Robert Eli Rosen, The Inside Counsel Movement, Professional Judgment and Organizational Representation, 64 Ind. L.J. 479, 484 (1989) ("inside counsel have emerged as purchasers of outside firm services and monitors and auditors of outside counsel's work").

n40. See, e.g., Darlene Ricker, Auditing Lawyers for a Living, A.B.A. J., Aug. 1994, at 65 (listing several legal auditing firms and discussing their respective activities); Amy Stevens, Companies, in Trying to Cut Legal Expenses, Hire Another Attorney, Wall St. J., June 3, 1994, at B8.

n41. See Isaac Ehrlich, Participation in Illegitimate Activities: An Economic Analysis, in The Economics of Crime and Punishment 68 (Gary S. Becker & William M. Landes eds., 1974); Richard A. Posner, The Economic Analysis of Law 201-12 (3d ed. 1986); David J. Pyle, The Economics of Crime and Law Enforcement chs. 3-4 (1983).

n42. See Laurie P. Cohen, Myerson & Kuhn Is Said to Agree to Pay $ 1.1 Million to Shearson, Wall St. J., June 9, 1989, at B8.

n43. See David Margolick, A Theft Scandal Ravages a Career At a Leading American Law Firm, N.Y. Times, May 13, 1994, at B18.

n44. We discuss the level of justification infra Part I.B.3.

n45. See Gilson & Mnookin, supra note 22, at 318-20.

n46. Id.

n47. See id. at 371-80.

n48. Id. at 374-78.

n49. For an explanation of Bayes' Rule, see Rasmusen, supra note 32, at 58-59.

n50. (.3)(.9)/((.3)(.9) + (.5)(.1)) = .84

n51. (.9)(.7)/((.9)(.7) + (.5)(.1)) = .93

n52. Of course, the lawyer has a long-term control over the client's belief in that that belief is based on prior actions of the attorney.

n53. This outsider-monitoring of the actions of an insider with private information is akin to the market assessing the healthiness of a company by observing whether or not a loan is renewed. See Gur Huberman & Charles Kahn, Limited Contract Enforcement and Strategic Renegotiation, 78 Am. Econ. Rev. 471 (1988). Empirical support for this proposition can be found in Myron B. Slovin et al., Firm Size and the Information Content of Bank Loan Announcements, 16 J. Banking & Fin. 1057 (1992) (positive share price effect reported for loan renewals in small firms); Scott L. Lummer & John J. McConnell, Further Evidence on the Bank Lending Process and the Capital-Market Response to Bank Loan Agreements, 25 J. Fin. Econ. 99 (1989) (similar).

n54. According to the auction theory, the bids received in an auction will be reduced by the aggregate sum of all bidders' costs in preparing their bids. See Kenneth R. French & Robert E. McCormick, Sealed Bids, Sunk Costs, and the Process of Competition, 57 J. Bus. 417, 424-25 (1984); Jonathan R. Macey, Auction Theory, MBO's and Property Rights in Corporate Assets, 25 Wake Forest L. Rev. 85, 88-90 (1990).

n55. A recent innovation along these same lines is to hire an attorney with the sole purpose of monitoring the lawyers handling a given piece of business. See Ricker, supra note 40, at 65; Stevens, supra note 40, at B8. This method of constraining opportunistic behavior on the part of lawyers is best suited for litigation rather than commercial transactions.

n56. See Chayes & Chayes, supra note 39, at 293 ("[General counsels] believed their lawyers to be at least equal to outside counsel in training experience, and in many cases, in results and quality of work product."); Rosen, supra note 39, at 483 ("Inside counsel now are characterized as possessing the knowledge and training necessary to handle complex and important legal matters").

n57. See Amy Stevens, Lawyers Gaze At a Future of Bills That Are Task Based, Wall St. J., July 1, 1994, at B6. Some have questioned the efficacy of such matters. See Ricker, supra note 27, at 64 (quoting Harry Maue, Chairman of Stuart, Maue, Mitchell & James, who stated that task-based billing "may make you feel warm and fuzzy inside, but it is no panacea").

n58. It used to be that only large firms which had sufficient legal work could afford this type of monitoring. Very few small firms could afford to hire an in-house counsel. Recently, however, there is a growing market in firms which specialize in monitoring attorney performance. See, e.g., Ricker, supra note 40, at 65. While this type of service is currently only used to monitor expenses incurred in litigation, see id., future firms might offer to monitor handling of commercial transactions.

n59. See Kreps, supra note 31; Shavell, supra note 18, at 59.

n60. For a similar proposal in the tort context, see Schwartz & Mitchel, supra note 26, at 1154.

n61. See Charles Haddad, The Marriage Is Off: Collapse of TCI-Bell Atlantic Deal May Slow Pace of Similar Alliances, Atlanta J. & Const., Feb. 25, 1994, at G1.

n62. See Shavell, supra note 18, at 66 (suggesting that lawyers do not bill corporate clients on a contingent fee basis because they are risk averse toward the possibility of not receiving a fee).

n63. For example, according to one chief financial officer, the total transaction costs for a $ 10 million unsecured loan are about 75 basis points (.075% of the loan amount). Ronald Mann, Explaining the Pattern of Secured Credit, 110 Harv. L. Rev. 625, 661 (1997).

n64. See Gilson & Mnookin, supra note 22, at 360-71.

n65. On this question, see Revlon v. MacAndrews & Forbes Holding, 506 A.2d 173 (Del. 1986); Paramount Communications v. QVC Network, 637 A.2d 34 (Del. 1994).

n66. See, e.g., Angela Wennihan, Let's Put the Contingency Back in the Contingency Fee, 49 SMU L. Rev. 1639, 1642 (1996) (noting the typical contingency fee "ranges from 25% to 50%, depending on the stage of the case at the time of resolution").

n67. See Kreps, supra note 31, at 590.

n68. At least from the perspective of social welfare; perhaps from the attorney's perspective, the appropriate word would be "fortunately."

n69. On the standard of care in legal malpractice cases, see 1 Ronald E. Mallen & Jeffrey M. Smith, Legal Malpractice ch. 15 (3d ed. 1989).

n70. See Donald C. Langevoort, Where Were the Lawyers? A Behavioral Inquiry Into Lawyers' Responsibility for Clients' Fraud, 46 Vand. L. Rev. 75, 79-94 (1993).

n71. See, e.g., Michael K. McChrystal, Lawyers and Loyalty, 33 Wm. & Mary L. Rev. 367 (1992).

n72. On the interplay between professional norms and professional knowledge, see generally Eliot Freidson, Professional Powers: A Study of the Institutionalization of Formal Knowledge (1986); Magali Sarfati Larson, The Rise of Professionalism: A Sociological Analysis (1977). Relatively less attention, unfortunately, has been given to the ideological self-constructs of the professions than their functional or institutional structures. See Robert T. Nelson & David M. Trubek, Arenas of Professionalism: The Professional Ideologies of Lawyers in Context, in Lawyers' Ideals/Lawyers' Practices: Transformations in the American Legal Profession 177 (Robert T. Nelson et al. eds., 1992). Lawyers define their roles in ways that may even conflict with legal norms. See Susan P. Koniak, The Law Between the Bar and the State, 70 N.C. L. Rev. 1389 (1992). And the public pronouncements of the bar - including its ethical rules - may or may not reflect internal norms. E.g., Ted Schneyer, Professionalism as Bar Politics: The Making of the Model Rules of Professional Conduct, 14 Law & Soc. Inquiry 677 (1989). As Schneyer emphasizes, the bar is by no means monolithic, and it would be a mistake to suggest that there is a single ideology common to the entire bar. Id. at 679-80.

n73. See Abraham S. Blumberg, The Practice of Law as a Confidence Game, in Sociology of Law 321, 328 (Vilhelm Aubert ed., 1969); Robert Gordon, The Ideal and the Actual in the Law: Fantasies and Practices of New York City Lawyers, 1870-1910, in The New High Priests: Lawyers in Post-Civil War America 51 (Gerard W. Gawalt ed., 1984). This idea is very much the subject of Anthony T. Kronman, The Lost Lawyer: Failing Ideals of the Legal Profession (1994). A good rhetorical expression is that of Lon Fuller, who wrote that it is the lawyer's professional responsibility to avoid counselling a client with respect to a matter whose legality was even "doubtful." Lon Fuller & John D. Randall, Professional Responsibility: Report of the Joint Conference of the American Bar Association and the Association of American Law Schools, 44 A.B.A. J. 1159, 1161 (1958).

n74. See, e.g., Robert A. Kagen & Robert E. Rosen, On the Social Significance of Large Law Firm Practice, 37 Stan. L. Rev. 399 (1985).

n75. One could argue that if clients are systematically and strongly risk-averse, then they might prefer that lawyers be biased in the direction of overestimating legal risk in order to offset the predictable errors that inevitably arise when legal advice seeks to be carefully neutral. On general managerial risk preferences from a legal perspective, see Clayton P. Gillette, Commercial Relationships and the Selection of Default Rules for Remote Risks, 19 J. Legal Stud. 535, 552-62 (1990). The idea that clients wish to forego best estimates of legal risk does not seem intuitively appealing, especially given lawyers' conflict of interest. But it is possible that clients would trade off this risk in return for the political protection inherent in shifting responsibility for risk avoidance to the firm's lawyers. Cf. Zur Shapira, Ambiguity and Risk Taking in Organizations, 7 J. Risk & Uncertainty 89 (1993).

n76. Along these lines is the claim of Judge Richard Posner that even the prevailing (but eroding) concept of the law as an objective and knowable body of knowledge is the product of professional self-interest. See Richard A. Posner, The Material Basis of Jurisprudence, 69 Ind. L.J. 1 (1993). The coincidence between norms and self-interest is a standard one in the sociology of the professions. See, e.g., Larson, supra note 72, at 155; Mark Osiel, Lawyers as Monopolists, Aristocrats and Entrepreneurs, 103 Harv. L. Rev. 2009 (1990) (book review). It has heavily influenced the literature on the legal profession in particular. E.g., Thomas D. Morgan, The Evolving Concept of Professional Responsibility, 90 Harv. L. Rev. 702 (1977); Deborah L. Rhode, Why the ABA Bothers: A Functional Perspective on Professional Codes, 59 Tex. L. Rev. 689 (1981). One need not go so far as to posit a complete identity between professional norms and rent-seeking (e.g., Richard L. Abel, American Lawyers (1989)) to recognize that self-interest plays a strong role in the structure of the professions.

n77. See generally Jay Katz, M.D., The Silent World of Doctor and Patient (1984).

n78. For a general study of lawyers' deception, see Lerman, supra note 4. Most of the lies the author identifies are self-serving; some, however, might be considered "altruistic." Id. at 677.

n79. From a psychological perspective, see, e.g., Thomas Gilovich, How We Know What Isn't So: The Fallibility of Human Reason in Everyday Life 91-94 (1991).

n80. It is standard but must be used with care, lest it actually operate as a bar to influence. See Philip G. Zimbardo & Michael R. Lieppe, The Psychology of Attitude Change and Social Influence 345, 355 (1991).

n81. We should be careful here, since it is possible that overstating risk to deal with a reluctant client has as its effect something approaching optimal caution, to the extent that the client predictably does less than the lawyers' threat implies.

n82. See, e.g., Baldwin v. White Inv. Inc., 669 F. Supp. 1054, 1056-57 (D. Utah 1987); K Mart Corp. v. Knitjoy Mfg. Inc., 542 F. Supp. 1189, 1192 (E.D. Mich. 1982). Rather, it is generally assumed that a set of additional factors must be present. E.g., Robert Thompson, Piercing the Corporate Veil: An Empirical Study, 76 Cornell L. Rev. 1036, 1067-68 (1991). This is not to say that informality is unimportant; it may mask fraudulent behavior, and lead to the sort of intermingling of corporate and personal assets that is probative of veil-piercing. But this risk could be communicated and managed much more precisely.

n83. See Lauren B. Edelman et al., Professional Construction of Law: The Inflated Threat of Wrongful Discharge, 26 Law & Soc'y Rev. 47, 78 (1992).

n84. Id. at 74-77. For an insightful inquiry into the sometimes awkward role of in-house counsel, noting the opportunity for the exercise of political power vis-a-vis the client, see Robert E. Rosen, The Inside Counsel Movement, Professional Judgment and Organizational Representation, 64 Ind. L.J. 479 (1989). We might also predict that lawyers' perceptions of what is in the client's best interests can be shaped by unconscious self-interest, blurring this distinction further. This possibility is discussed infra Part II.B.

n85. See Detlev F. Vagts, Legal Opinions in Quantitative Terms: The Lawyer As Haruspex or Bookie?, 34 Bus. Law. 421 (1979).

n86. Or "highly probable."

n87. As a separate matter, lawyers probably are reluctant to advise a client that a risk is wholly uncertain because this undercuts both their self-image and external image as an expert. This, however, does not imply overstatement of risk generally, for it suggests that small risks will tend to be understated in order to achieve greater certainty. For an interesting study of doctors' intolerance of uncertainty, see Martha S. Gerrity et al., Uncertainty and Professional Work: Perceptions of Physicians in Clinical Practice, 97 Am. J. Soc. 1022, 1027-30 (1992). Studies of lawyer behavior have demonstrated a strong desire to avoid at least the appearance of uncertainty. See John Flood, Doing Business: The Management of Uncertainty in Lawyers' Work, 25 Law & Soc'y Rev. 41 (1991).

n88. E.g., Howard Raiffa, The Art and Science of Negotiation (1982); Detlof von Winterfeldt & Ward Edwards, Decision Analysis and Behavioral Research (1986). Indeed, lawyers are frequently taught to make such probabilistic calculations of likely success at trial for purposes of preparing for settlement negotiations. E.g., Gerald R. Williams, Legal Negotiation and Settlement ch. 6 (1983).

n89. See James G. March & Zur Shapira, Managerial Perspectives on Risk and Risk Taking, 33 Mgmt. Sci. 1505 (1987), reprinted in James March, Organizations and Decisions 76, 86-88 (1988). They exhibit overconfidence in their ability to manage or control risks. This sense, presumably illusory, is based largely on their prior successes in avoiding negative consequences (which often is no more than simple luck). Those who have guessed right in the past survive to manage at the next organizational level; those who guess wrong are often weeded out. People often misattribute their good fortune to skill. See infra notes 102-04, 120-25 and accompanying text. To the extent that managers would receive credit for successfully managing a legal risk, they might actually prefer a lawyer who overstates that risk, since the situation is in fact less likely to generate a bad outcome than it would appear. On managerial preferences, see also Kenneth R. MacCrimmon & Donald A. Wehrung, Taking Risks: The Management of Uncertainty (1986).

n90. See, e.g., Thomas S. Wallsten, Costs and Benefits of Vague Information, in Insights in Decision Making 28, 35 (Robin M. Hogarth ed., 1990); Samuel Fillenbaum et al., Some Effects of Vocabulary and Communication Task on the Understanding and Use of Vague Probability Estimates, 104 Am. J. Psychol. 35 (1991).

n91. The direction of the process of elimination is all-important, of course. If the lawyer were to start at the point of certain risk and move downward, the opposite bias would be introduced. It is also conceivable that a lawyer would start in the middle, and move in whatever direction is indicated by the lawyer's sense of whether the risk is greater or less than even. This process would simply be polarizing: understating risks that are perceived as less than even, while overstating those perceived as greater. Even if we assume that the direction is upward, it is possible that one point at which the prediction might not hold is in the area between highly likely and certain. The reluctance to come to unhedged conclusions may trump the bias toward overstatement.

n92. James C. Freund, Advise and Invent: The Lawyer as Counselor-Strategist and Other Essays 40-50 (1990).

n93. The notion of bounded rationality is generally attributed to Herbert Simon. See, e.g., Herbert A. Simon, Rational Choice and the Structure of the Environment, 63 Psychol. Rev. 129 (1956). For good overviews of behavioral decision theory, see Paul Slovic et al., Decision Making, in 2 Stevens' Handbook of Experimental Psychology 673 (Richard C. Atkinson et al. eds., 2d ed. 1988); Robin M. Hogarth, Judgment and Choice (2d ed. 1987); John W. Payne et al., Behavioral Decision Research: A Constructive Processing Perspective, 43 Ann. Rev. Psychol. 87 (1992). The possibility that this literature may be of use in assessing how lawyers and judges estimate the strength of potential cases is raised in Richard L. Wiener & Mark A. Small, Social Cognition and Tort Law: The Roles of Basic Science and Social Engineering, in Handbook of Psychology and Law 435, 449 (Dorothy K. Kagehiro & William S. Laufer eds., 1992). See also Detlof von Winterfeldt & Ward Edwards, Cognitive Illusions and Their Implications for the Law, 59 S. Cal. L. Rev. 225 (1986).

n94. On adaptiveness, see Hal R. Arkes, Costs and Benefits of Judgment Errors: Implications for Debiasing, 110 Psychol. Bull. 486 (1991). A useful survey of the two separate "camps" in this field (i.e., those emphasizing purely cognitive factors, and those emphasizing motivations) is Helmut Jungermann, The Two Camps on Rationality, in Judgment and Decision Making: An Interdisciplinary Reader 627 (Hal R. Arkes & Kenneth R. Hammond eds., 1986). The relatively more unknown motivational influences on risky choice are explored in Lola L. Lopes, Between Hope and Fear: The Psychology of Risk, in 20 Advances in Experimental Social Psychology 255 (Leonard Berkowitz ed., 1987).

n95. For some examples, see George Loewenstein et al., Self-Serving Assessments of Fairness and Pretrial Bargaining, 22 J. Legal Stud. 135 (1993); Elizabeth F. Loftus & Willem A. Wagenaar, Lawyers' Predictions of Success, 28 Jurimetrics J. 437 (1988).

n96. See Colin F. Camerer & Eric J. Johnson, The Process-Performance Paradox in Expert Judgment, in Toward a General Theory of Expertise: Prospects and Limits 195, 203 (K. Anders Ericsson & Jacqui Smith eds., 1991); see also the various surveys (especially chapters 1, 3 and 4) in Expertise and Decision Support (George Wright & Fergus Bolger eds., 1992).

n97. Researchers frequently point out that for learning to occur, feedback must be sufficiently salient and unambiguous to convince the actor, who is motivated to find confirmation rather than disconfirmation, that there was an error attributable to judgment processes. In real life, such feedback is not particularly common. See Slovic et al., supra note 93, at 699.

n98. See John W. Payne et al., The Adaptive Decision Maker: Effort and Accuracy in Choice, in Insights in Decision Making, supra note 90, at 129.

n99. See Sarah Lichtenstein et al., Knowing With Certainty: The Appropriateness of Extreme Confidence, 3 J. Experimental Psychol.: Hum. Perception & Performance 552 (1977).

n100. E.g., Hillel J. Einhorn & Robin M. Hogarth, Judging Probable Cause, 99 Psychol. Bull. 3 (1986).

n101. See Amos Tversky & Daniel Kahneman, Rational Choice and the Framing of Decisions, 59 J. Bus. S251 (1986); Amos Tversky & Daniel Kahneman, Advances in Prospect Theory: Cumulative Representation of Uncertainty, 5 J. Risk & Uncertainty 297 (1992). See also Robin Gregory et al., The Role of Past States in Determining Reference Points for Policy Decisions, 55 Organizational Behav. & Hum. Decision Processes 195 (1993).

n102. See Langevoort, supra note 70, at 98-105 (citing several other sources). In addition, it is quite possible that if a client frames a problem in terms that prompts a search for risk-negative information (see supra notes 20-21 and accompanying text), underestimation could follow.

n103. See Ellen J. Langer, The Illusion of Control, 32 J. Personality & Soc. Psychol. 311 (1975); Hogarth, supra note 93, at 12-14. See also infra note 125 and accompanying text.

n104. See Loftus & Wagenaar, supra note 95 (actual litigators). See also Loewenstein et al., supra note 95 (undergraduates asked to assume a litigation role); Theodore Eisenberg, Negotiation, Lawyering, and Adjudication: Kritzer on Brokers and Deals, 19 Law & Soc. Inquiry 275, 295-97 (1994) (book review). One interesting question is how overoptimism persists when presumably it leads to a greater incidence of observable failures, something which operates as a reputational threat. One answer is that overoptimism generates a self-confidence that may actually lead to greater success (a self-fulfilling prophecy), at least when the opponent is not so imbued. See Loftus & Wagenaar, supra note 95, at 450-51. One might also hypothesize that this is an area in which the art of persuasion plays an important role: successful lawyers, like successful salespeople, are those with an above-average skill to take credit and avoid blame in long-term client relationships. Cf. Flood, supra note 87.

n105. See Paul Slovic et al., Preference for Insuring Against Probable Small Losses: Insurance Implications, 44 J. Risk & Ins. 237 (1977); Paul Slovic et al., Regulation of Risk: A Psychological Perspective, in Regulatory Policy and the Social Sciences 241, 260 (Roger Noll ed., 1985).

n106. See infra Part II.B.2.a.

n107. See Hogarth, supra note 93, at ch. 4; Payne et al., supra note 98, at 102-11.

n108. See Hogarth, supra note 93, at 101-09; Hillel J. Einhorn & Robin M. Hogarth, Decision Making Under Ambiguity, 59 J. Bus. S225 (1986). In this work, the distinction is made between uncertainty (where probability estimates are known, even though outcome is not) and ambiguity (where neither outcome nor the probability estimates associated with it is known).

n109. See Robin M. Hogarth & Hillel J. Einhorn, Venture Theory: A Model of Decision Weights, 36 Mgmt. Sci. 780, 783 (1990) ("Cautious, or 'defensively pessimistic' is generally characterized by 'underweighting' probabilities of gains and 'overweighting' probabilities of losses ...."). Two dominating factors are how important the outcome is and whether the person is contemplating a gain or loss. Where the outcome is important and a gain is contemplated, people typically give greater weight in imagination to estimates below the anchor, and hence adjust the probability of success downward. In complementary fashion, those concerned with a significant loss tend to adjust upward the probability of a negative outcome. In each case, the extent of the adjustment depends on the initial placement of the anchor; if the person starts with a high probability of losing, for instance, the room for upward adjustment is limited. For a good review, see Cynthia S. Fobian & Jay J. Christensen-Szalanski, Ambiguity and Liability Negotiations: The Effects of Negotiators' Role and the Sensitivity Zone, 54 Organizational Behav. & Hum. Decision Processes 277, 278-80 (1993).

n110. Some experimental tests of the model are worth noting. One has to do with settlement negotiations. The Einhorn-Hogarth model predicts that ambiguity will lead potential plaintiffs and defendants to evaluate the risk of going to trial differently. Plaintiffs contemplate a gain through settlement, and hence are generally motivated toward a downward adjustment of the anchor probability of winning. That adjustment will be great for high initial estimates, necessarily less so for low ones. Conversely, defendants will adjust their probability of losing upward, with the greatest adjustment coming when the initial anchor is low. This leads to the prediction that plaintiffs will accept much less to settle a case where there is a high initial anchor probability of winning but also high ambiguity; defendants will pay much more to settle when there is high ambiguity and a low anchor probability of plaintiff winning. Using business students in a controlled experiment, this prediction was confirmed. See Fobian & Christensen-Szalanski, supra note 109. Another study, examining the pricing of insurance by experienced insurance adjusters, confirmed the prediction that increased ambiguity would lead to more weight given in imagination to risks that are above the anchor value. See Howard Kunreuther & Robin M. Hogarth, Risk, Ambiguity and Insurance, 2 J. Risk & Uncertainty 5, 25-26 (1989). See also Howard Kunreuther et al., Insurer Ambiguity and Market Failure, 7 J. Risk & Uncertainty 71 (1993).

n111. The psychological differences in the ways people approach possible gains and losses is a major theme in the behavioral literature. The so-called status quo (or endowment) effect emphasizes that people place greater value on that which they possess than that which they might possess. For a review of the literature from a legal perspective, see Elizabeth Hoffman & Matthew L. Spitzer, Willingness to Pay versus Willingness to Accept: Legal and Economic Implications, 71 Wash. U. L.Q. 59 (1993). The impact of this and related cognitive biases on settlement negotiations is explored in Russell Korobkin & Chris Guthrie, Psychological Barriers to Litigation Settlement: An Experimental Approach, 93 Mich. L. Rev. 107 (1994).

n112. The structure of accountability also tilts in that direction, as discussed below. One interesting study observes that experts are frequently ambiguity-preferring, something that the authors attribute to the fact that they receive credit for their successes in judgment, but not to simple luck. See Chip Heath & Amos Tversky, Preference and Belief: Ambiguity and Competence in Choice Under Uncertainty, 4 J. Risk & Uncertainty 5, 8 (1991). To the extent that the assignment of credit and blame is asymmetrical, as we have argued, this same influence will not be present. See infra notes 115-16, 119 and accompanying text. On the other hand, it also helps explain a different effect where the attorney perceives the risk to be under her control. See supra notes 103, 125 and accompanying text.

n113. See Derek Koehler, Explanation, Imagination, and Confidence in Judgment, 110 Psychol. Bull. 499, 511 (1991); W. Larry Gregory et al., Self-Relevant Scenarios as Mediators of Likelihood Estimates and Compliance: Does Imagining Make It So?, 43 J. Personality & Soc. Psychol. 89 (1982); Ariel S. Levi & John B. Pryor, Use of the Availability Heuristic in Probability Estimates of Future Events: The Effects of Imagining Outcomes Versus Imagining Reasons, 40 Organizational Behav. & Hum. Decision Processes 219 (1987). Once we have identified a predictable direction to the adjustment process, then some of the more generic biases - e.g., illusory correlations - are likely to contribute disproportionately toward overestimation.

n114. See Philip E. Tetlock, Accountability: The Neglected Social Context of Judgment and Choice, 7 Res. in Organizational Behav. 297 (1985). For a good overview, see Itamar Simonson & Peter Nye, The Effect of Accountability on Susceptibility to Decision Errors, 51 Organizational Behav. & Hum. Decision Processes 416 (1992).

n115. The question of how much the influence of accountability is conscious (i.e., knowingly seeking the approval of others) and how much is not, is an open one. See, e.g., Tetlock, supra note 114, at 317.

n116. See Philip E. Tetlock et al., Social and Cognitive Strategies for Coping with Accountability: Conformity, Complexity and Bolstering, 57 J. Personality & Soc. Psychol. 632 (1989).

n117. Philip E. Tetlock & Richard Boettger, Accountability: A Social Magnifier of the Dilution Effect, 57 J. Personality & Soc. Psychol. 388 (1989). See also Arie W. Kruglanski, The Psychology of Being "Right": The Problem of Accuracy in Social Perception and Cognition, 106 Psychol. Bull. 395, 404 (1989).

n118. See Irving L. Janis, Groupthink: Psychological Studies of Policy Decisions and Fiascoes (2d ed. 1982); William R. Ferrell, Combining Individual Judgments, in Behavioral Decision Making 135-137 (George Wright ed., 1985). The conformity pressure is observed in the legal setting in Freund, supra note 92, at 28-33.

n119. See Hogarth, supra note 93, at 142-49; Ed Bukszar & Terry Connolly, Hindsight Bias and Strategic Choice: Some Problems in Learning from Experience, 31 Acad. Mgmt. J. 628 (1988).

n120. In other words, "wishful thinking." The relationship between ego-driven inference and law related behavior is explored more generally in Donald C. Langevoort, Ego, Human Behavior, and Law, 81 Va. L. Rev. 853 (1995). See Gilovich, supra note 79, at ch. 5; Lopes, supra note 94, at 288-91. On rationalization by white-collar criminals that can serve to justify selfish pursuit, see Donald Ray Cressey, Other Peoples' Money ch. 4 (1953); in the corporate context (self-dealing), see Robert C. Clark, Corporate Law 143 (1986). The behavioral literature does not give a clear picture of how subconscious this biasing process is. Much of the research assumes that people are frequently unaware of their biases. E.g., Irving L. Janis & Leon Mann, Decision-Making: A Psychological Analysis of Conflict, Choice and Commitment 95-96 (1977); Susan T. Fiske & Shelley E. Taylor, Social Cognition 228 (2d ed. 1991). On the other hand, this is difficult to prove, since many people also offer disingenuous excuses regarding their levels of awareness, consciously rationalize self-serving behavior, or adopt a posture of avoiding information that would make them confront the self-serving nature of their behavior. See Albert Bandura, Social Cognitive Theory of Moral Thought and Action, in 1 Handbook of Moral Behavior and Development 45, 95 (William M. Kurtines & Jacob L. Gewirtz eds., 1991). For various essays dealing with the nature of this problem, see Self-Deception and Self-Understanding: New Essays in Philosophy and Psychology (Mike W. Martin ed., 1985). For our purposes, it is enough to recognize that the line between conscious awareness and unconscious accommodation of self-interest in decision-making is a fine one, and individual instances will fall at various points along the spectrum.

n121. See supra note 92 and accompanying text. The idea that lawyers overbill by deluding themselves into thinking that it is in the client's best interests is noted in Ross, supra note 35, at 27. On attorney fee petitions in bankruptcy, see generally Theodore Eisenberg, Differing Perceptions of Attorney Fees in Bankruptcy Cases, 72 Wash. U. L.Q. 979 (1994).

n122. See generally Anthony G. Greenwald, The Totalitarian Ego: Fabrication and Revision of Personal History, 35 Am. Psychologist 603 (1980); Roderick M. Kramer, Self-Enhancement Biases and Negotiator Judgment: Effects of Self-Esteem and Mood, 56 Organizational Behav. & Hum. Decision Processes 110 (1993); Ziva Kunda, Motivated Inference: Self-Serving Generation and Evaluation of Causal Theories, 53 J. Personality & Soc. Psychol. 636 (1987); C.R. Snyder et al., Excuses: Masquerades in Search of Grace 31-33 (1983). Much of this derives from work on cognitive dissonance. In essence, the mind filters out information (i.e., that relating to motivation) that is inconsistent with the idealized self-concept, in order to reduce stress. We should note, however, that it may really be the fragile ego, rather than the healthy one, that generates the largest distortion. See also Claude M. Steele et al., Self-Image, Resilience and Dissonance, 64 J. Personality & Soc. Psychol. 885 (1993).

n123. See Loftus & Wagenaar, supra note 95, at 450; Albert Bandura, Human Agency in Social Cognitive Theory, 44 Am. Psychologist 1175 (1989). Among other things, high self-esteem leads to greater persistence and risk-taking, which on average, leads to greater numbers of successes (which in turn leads to increased self-esteem). In addition, high self-esteem is associated with reduced stress and good health, presumably because it enables people to avoid fears and doubts. See Fiske & Taylor, supra note 120, at 212-16. Naturally, ego can also be destructive, sometimes leading to irrational behaviors designed simply to bolster an illusory self-concept. E.g. Ray F. Baumeister et al., When Ego Threats Lead to Self-Regulatory Failure: Negative Consequences of High Self-Esteem, 64 J. Personality & Soc. Psychol. 141, 152 (1993).

n124. Cf. Richard P. Larrick, Motivational Factors in Decision Theories: The Role of Self-Protection, 113 Psychol. Bull. 440, 445-47 (1993).

n125. Here, once again, we see the influence of the illusion of control. See supra notes 103-04 and accompanying text.

n126. On the importance of status in the explanation of economic behavior, see Robert H. Frank, Choosing the Right Pond: Human Behavior and the Quest for Status (1985).

n127. See Langevoort, supra note 120, at 863-64. The dominance and control concept is well recognized in the literature on client counselling though rarely in the business law setting. See, e.g., Robert M. Bastress & Joseph L. Harbaugh, Interviewing, Counselling and Negotiating 287-90 (1990). Yet it is hardly implausible that highly paid elite lawyers are strongly motivated to reverse the subservient role vis-a-vis the client that they are often forced to play. Well-heeled clients often impose upon their lawyers both materially and psychologically; using the leverage of risk is something of an equalizer.

n128. While this question of lawyer bias has not been explored, the idea that bidders in tender offers are affected by bias (the so-called "hubris hypothesis") has. See Richard Roll, The Hubris Hypothesis in Corporate Takeovers, 59 J. Bus. 197 (1986).

n129. We should note that the cognitive and motivational influences on the perception of risk also operate on the way clients perceive legal advice. Thus, for example, a lawyer may affect the client's perception of risk simply by the way he frames the advice (emphasizing loss rather than gain, for instance). This influence is noted in Freund, supra note 92, at 50-52.

n130. Easier to form in the sense that there is an inverse correlation between availability and effort. To the extent that decision-makers are motivated by accountability concerns to find the most readily justifiable decision (see supra note 119 and accompanying text), the skewed availability of information is relevant as well.

n131. On the role of experience in lawyers' perceptions, see Loftus & Wagenaar, supra note 95, at 450. Indeed, personal experience dominates most perception, by itself introducing an egocentric bias. See Dale W. Griffin & Lee Ross, Subjective Construal, Social Inference and Human Misunderstanding, in 24 Advances in Experimental Social Psychology 319 (Mark P. Zanna ed., 1991).

n132. On the other hand, if the tendency to give risk averse advice is strong, that experience will be repeated in subsequent decisions in order to maintain consistency.

n133. This is referred to as the self-serving attribution bias. Although it can be explained simply, because individuals have better access to information about their personal role than external causes, and can thus be expected to overweigh them (see Richard Nisbett & Lee Ross, Human Inference: Strategies and Shortcomings of Social Judgment (1980)), much evidence points in the direction of ego as a motivating factor. E.g., Ziva Kunda, The Case for Motivated Reasoning, 108 Psychol. Bull. 480 (1990).

n134. The seminal article is George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1 (1984). The 50/50 hypothesis is not uncontroversial. See generally Theodore Eisenberg, Testing the Selection Effect: A New Theoretical Framework with Empirical Tests, 19 J. Legal Stud. 337 (1990); Samuel R. Gross & Kent D. Syverud, Getting to No: A Study of Settlement, Negotiations and the Selection of Cases for Trial, 90 Mich. L. Rev. 319 (1991). However, its basic insight is relatively intuitive. That uncertainty can be driven by the selfish desires of the legal profession is noted in Paul Rubin & Martin J. Bailey, The Role of Lawyers in Changing the Law, 23 J. Legal Stud. 807 (1994).

n135. On the other hand, one should note that to the extent that judges (or their clerks) are heavily influenced by precedent, the bias in the data set will also influence outcomes in litigation. If so, the bias in advice may not be entirely erroneous.

n136. To the extent that the 50/50 hypothesis is roughly accurate, one might argue that risk analysts will then underestimate risk with respect to relatively pro-plaintiff claims. If, however, there is a motivated search for risk positive information, there will be an asymmetry here. We should also take into account the possibility that cases vigorously pursued by plaintiffs are those most likely to be based on relatively stronger legal claims, generating a greater number of successful outcomes for plaintiffs - the chief indicator of legal risk - than if the broader data set could somehow be reviewed. To illustrate, a lawyer asked to determine whether a proposed television commercial constitutes deceptive advertising would be able to observe those cases that the FTC staff brought and litigated (some successfully, some not), but - absent private information - not those that the staff considered but did not pursue. That can easily give the observable law a risk-positive coloration.

n137. Janet Cooper Alexander, Do the Merits Matter? A Study of Settlements in Securities Class Actions, 43 Stan. L. Rev. 497, 569-70 (1991). This is closely related to potential biases in the secondary literature, discussed infra Part II.C.3.

n138. See D'Amato, supra note 14, at 21.

n139. See Gilovich, supra note 79, at 91-94.

n140. See Celia Paul, Developing New Relationships, in Developing Your Legal Practice: How to Obtain and Maintain a Client Base 19, 21-30 (PLI Commercial Law & Practice Handbook Series No. 662, 1993). An interesting study of "entrepreneurial" lawyers and their effect on the dissemination of legal information is Michael J. Powell, Professional Innovation: Corporate Lawyers and Private Lawmaking, 18 Law & Soc. Inquiry 423, 427 (1993). And even academics, of course, can be influenced by self-interest. See generally Rebecca S. Eisenberg, The Scholar as Advocate, 43 J. Legal Educ. 391 (1993).

n141. There are a number of reasons for this. Following the attitudes or opinions of respected others is a common heuristic (social learning); sometimes, in fact, we follow the opinions of total strangers. More rationally, the judgment of others is important because professional consensus is frequently a means by reference to which the performance of any one actor is judged. See David S. Scharfstein & Jeremy C. Stein, Herd Behavior and Investment, 80 Am. Econ. Rev. 465 (1990). In these circumstances, pressures for conformity are particularly severe. The rationality of following others' behavior even when private information differs - and the resulting possibility of a large incidence of suboptimal choices - is also treated in Banerjee, supra note 10. See also Candace Prendergast, A Theory of "Yes Men", 83 Am. Econ. Rev. 757 (1993).

n142. On informational cascades, see generally Sushil Bikhchandani et al., A Theory of Fads, Fashion, Custom and Cultural Change as Informational Cascades, 100 J. Pol. Econ. 992 (1992); in the financial area, see Ivo Welch, Sequential Sales, Learning and Cascades, 47 J. Fin. 695 (1992).

n143. On social learning - i.e., the tendency to accept as true the inferences of others - see Robert B. Cialdini, Influence: Science and Practice ch. 4 (3d ed. 1993).

n144. Lauren B. Edelman et al., Professional Construction of Law: The Inflated Threat of Wrongful Discharge, 26 Law & Soc'y Rev. 47 (1992).

n145. See generally Escott v. BarChris Constr. Co., 283 F. Supp. 643 (S.D.N.Y. 1968).

n146. See Harry Heller et al., BarChris: A Dialogue on a Bad Case Making Hard Law, 57 Geo. L.J. 221 (1968); Ernest L. Folk, III, Civil Liabilities Under the Federal Securities Laws: The BarChris Case, 55 Va. L. Rev. 1 (1969); Louis Loss, The Opinion, 24 Bus. Law. 523 (1969) (commentary on BarChris, with the presentation attended by over a thousand lawyers). Plainly, BarChris added a lot of uncertainty in the law, and in that sense did increase risk. For a view that much of the concern might have been overstated, however, see Loss, supra, at 527.

n147. See generally In re Software Toolworks, Inc., 789 F. Supp. 1489 (N.D. Cal. 1992), aff'd in part, rev'd in part, 38 F.3d 208 (9th Cir. 1994); see also Weinberger v. Jackson, No. C-89-2301-Cal., 1990 WL 260676 (N.D. Cal. Oct. 11, 1990); Laven v. Flanagan, 695 F. Supp. 800 (D.N.J. 1988).

n148. See supra note 28 and accompanying text; Part I.B.3.

n149. Separately, an interesting question is the effect of time on the perception of risk. Based on the psychological literature, one might posit that the more time the lawyer has to think, the more risk positive information will be considered. See supra Part II.B.2.a. On the other hand, time pressure - by signalling the potential absence of complete information - may also bias the analysis toward caution.

n150. The idea that increased ex post informational costs are associated with the applications of standards, as opposed to bright line rules, is well accepted. See Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 Duke L.J. 557 (1992).

n151. In addition, situations where multiple clients might be induced to use the services of a lawyer who "invented" an efficient risk management strategy would offer an antidote to excessive caution. See, for example, the invention of the "poison pill" as an antitakeover device - developed for one client but successfully exported to many others. See generally Powell, supra note 140. Here, however, we expect that what is often exported is an inflated perception of risk, with the lawyer's self-serving assessment of his or her ability to manage it.

n152. For instance, imagine a "beauty contest" situation, in which law firms are asked to make submissions prior to retention, but where the issue was of immense topicality, generating a large amount of risk-positive information. We question how many firms would disregard the prevailing risk perception, even if they believed internally that the perception was overstated.

n153. See Baruch Fischhoff, Debiasing, in Judgment Under Uncertainty: Heuristics and Biases 422 (Daniel Kahneman et al. eds., 1982).