Plural subjects and a puzzle

The core of Gilbert’s theory is formed by her idea that when a collective is performing a joint action, it has to be a plural subject. Gilbert defines a plural subject as a set of human beings among whom there is common knowledge that each of its members has openly manifested to all other members a quasi-readiness to perform some joint action. Let me explain Gilbert’s technical terminology. First of all, the members have to be quasi-ready to perform the joint action. This means that if you are a member, then you are individually willing to perform your individual share of the joint action, provided the other members also perform their shares. Quasi-readiness is a kind of conditional willingness in the sense that, for example, you are quasi-ready to go to the opera with someone else if you go to the opera on the condition that the other person joins you. Secondly, the members of the collective must have expressed their quasi-readiness in such a way that all other members notice this (unless they do not pay attention, fail to interpret the utterances in the right way or fail to draw standard logical inferences). You could express your quasi-readiness to go to the opera with someone by saying such things as ‘Let’s go to the opera tonight.’ And if the response is positive – ‘Yes, let’s. Parsifal had a rave review in the Guardian last week!’ – the other person reveals quasi-readiness to go to the opera as well, and you have established a plural subject with respect to going to the opera. Common knowledge refers here to the fact that not only do you know about the other person’s quasi-readiness, but also that you know that the other person knows about your quasi-readiness, and so on. It is, in other words, completely open to the two of you that you are both quasi-ready to go.

This is an admittedly rough sketch of plural subject theory, but no more details are needed to understand Lahroodi’s puzzle. To recall, Lahroodi discredited individualism about corporate epistemic virtues, arguing that individual epistemic virtues are not always sufficient to obtain corporate epistemic virtue. Rather than defending holism about corporate epistemic virtue, he promised a puzzle about holism, based on plural subject theory. He considers a group of people incorporating open-mindedness. For the group to possess this virtue, it must have the disposition to perform open-minded group activities. Now for activities to count as genuine group activities they have to result from the group’s members having expressed their quasi-readiness to perform their parts of these activities. This, Lahroodi seems to claim, means that the ultimate requirement is that the group’s members are individually ready to perform activities that result from their individual disposition to proceed in an open-minded manner.

Suppose this were right. Then plural subject theory, which Lahroodi considers to be a holist theory par excellence, would lead us to conclude that corporate epistemic virtues reduce to individual epistemic virtues. Holism would paradoxically support individualism. The plausibility of this puzzling conclusion, however, depends on the assumption that the only view of corporate epistemic virtues consistent with holist theories is one according to which a corporate entity possesses an epistemic virtue V precisely when its members are jointly committed to acting according to virtue V. In a reply to Lahroodi, Todd Jones casts some doubt on this assumption, claiming that corporate entities can derive their epistemic virtues from their members instantiating what he calls ‘knowledge-enhancing tools’ and from their ‘simulating’ epistemic virtues.¹⁷ Jones hints at a number of ways in which collectives accomplish such simulation. The gist of his contribution lies, however, in exploring the conceptual space between individualism and holism, without offering hands-on tools enlarging our understanding of corporate epistemic virtues in business. This is not to diminish the importance of the problems Jones hints at. Quite the contrary, it is when we home in on collective entities as we know them in business that Lahroodi’s puzzle and Jones’s critique become pressing. Were organizations to possess epistemic virtues only to the degree that management and employees possess them, the relevance of individual virtues would be overrated and the value of the precise structure of the organization undervalued. Individual virtues are important, but if they are a precondition for corporate virtue, hardly any virtuous corporation exists.

Corporate internal decision structures

Where Jones speaks about the simulation of virtue this can be taken as a suggestion to pay more attention to the internal organizational structure of the company. Plural subject theory provides little in the way of explaining a corporate entity’s internal workings; we have to consider alternative views. It is only natural to turn our attention to the highly influential theory of corporate entities developed by Peter French.¹⁸ This theory, which is part of a view of corporate moral agency that need not detain us here, holds that, unlike people, corporate entities have corporate internal decision structures. These structures consist of two elements. First of all, a responsibility flowchart. This flowchart sketches a hierarchical command structure in the organization as well as positions and levels of management. The corporate internal decision structures also determine the relationships of subordination and authority between these positions and levels. Secondly, there are corporate decision recognition rules allowing us to distinguish between genuinely corporate decisions and individual ones. Even though very different sorts of corporate entities fall within French’s definition, one may suspect that it owes much of its inspiration to the concept of the public limited company or corporation that we encountered in Chapter 1. Despite fundamental disagreements with Ronald Coase and other economists working on the theory of the firm, French’s concept of responsibility flowchart may be seen as reflecting insights on the efficiency of hierarchical corporate governance structures, whereas the recognition rules offer mechanisms tailored to allowing corporate entities to conform to corporate law and become legal persons with limited liability.¹⁹

The responsibility flowchart of a corporation is described in such things as the memorandum and articles of association or the certificate of incorporation and bylaws, in the terms of employment of directors, managers and employees, and in many other official and internal documents. They determine the rights and duties of directors, management and employees, the hierarchical structures of authority and command among them, and the principles and methods of operation. Typically, these documents also contain provisions about the way the board and its members represent the company, and how these powers of representation can be delegated to a corporation’s employees. Such provisions function as corporate decision recognition rules determining the conditions under which an employee’s actions are to be conceived as corporate decisions. A loan officer sending an email to a client with an offer for a particular loan performs a corporate action; a loan officer using a corporate email account to send an email to a friend does not.

How does French’s theory of collective entities allow us to develop a view of corporate epistemic virtues that steers clear of Lahroodi’s puzzle? The problem was to develop a view of epistemic virtue that is sensitive to the fact that for a firm to have the virtue of open-mindedness, say, not all employees need to be open-minded individually. What makes a firm open-minded is whether as a firm it displays open-minded behaviour springing from a stable acquired disposition towards open-mindedness.

Describing the hierarchical command structure of a firm, corporate internal decision structures clearly suggest themselves as a locus of corporate virtue. Or so I argue. A firm is open-minded whenever its responsibility flowchart and corporate decision recognition rules together lead to a tendency to make open-minded decisions. Decision structures have to ensure that evidence from different sides is weighed, that the firm’s own preconceptions are not unduly privileged and so on. This requires a hierarchical command structure in which open-minded people are assigned to specific tasks and allocated specific powers. More generally, for a firm to incorporate epistemic virtue its employees have to possess the epistemic virtues that are necessary to fulfil the particular functions they occupy within the firm. Open-minded people, for example, have to be employed in positions that require open-mindedness, or current employees occupying such positions have to acquire open-mindedness. Management has to safeguard what I call virtue-to-function matching. Incorporation also requires employees to be in a position to act on virtues they possess. The firm must warrant that whenever open-mindedness is required, employees can act open-mindedly without obstruction. Management structures of command and control must, more generally, encourage and support virtue instead of frustrating it. There has to be organizational support for virtue. Furthermore, wherever open-mindedness is absent, responsibility flowcharts and recognition rules must help to counter epistemically unvirtuous behaviour. There have to be organizational remedies against vice.

Structures, functions, cultures and sanctions

To make this more precise, I turn to a third view of collective entities that looks into the internal workings of collective entities with an even sharper eye for detail than French’s, and with closer connections to some of the Aristotelian themes of epistemic virtue theory. Subsequently, I discuss virtue-to-function matching, organizational support and organizational remedies. Seumas Miller analyses organizations as systems of interdependent roles determined by four characteristic elements: structure, function, culture and a system of sanctions.²⁰ To begin with, Miller views an organization as an institution in which individual human beings play roles defined in terms of the specific tasks they have to carry out and the specific rules and procedures determining the performance of these tasks; in his own words, this is an ‘embodied [structure] of roles and associated rules’.²¹ Roles are typically interdependent. Performing tasks requires cooperation between various roles, and the way roles interrelate is generally one of hierarchical command. The structure of the organization, for Miller, comprises the roles that constitute the organization as well as the relationships between them, and is quite similar to French’s responsibility flowchart. In a business context, the Cadbury Committee offers a good illustration of this point. The Cadbury Committee was set up in 1991 in response to a number of accountancy scandals in the United Kingdom. It promoted the separation of the positions of chairman and managing director as well as the placing of independent directors on the board, and its description of the role of non-executive board members is still influential: ‘Non-executive directors should bring an independent judgement to bear on issues of strategy, performance, resources, including key appointments, and standards of conduct.’²² This is a clear element of a responsibility flowchart determining the structure of public limited companies.

This description, however, depends on the role that directors are meant to play, which in turn depends on the corporation’s aims. Unlike French, Miller therefore devotes a significant part of his theory to the function of organizations. Functions give rise to the tasks associated with the roles of an organization’s individual members. For Miller, the maximization of shareholder wealth is only one of the many goals a corporation can have. Short-term financial self-interest may be what inspires most shareholders, but markets and firms are there to satisfy more fundamental societal goals, Miller believes, such as increasing material wellbeing. Coming close to Robert Solomon’s virtue ethical position, which I dealt with briefly in Chapter 1, Miller defends the claim with the example of privatized jails, the goal of which is to contribute to retribution, rehabilitation, security and deterrence, even though as a for-profit organization they also aim to maximize gains. Similarly, Miller hints at the idea that finance firms operating on the capital market have as a goal the provision of finance to enterprises in the production industry.

But more is needed. Structure and function together determine the formal tasks and rules that characterize the organization. The third element, institutional culture, captures the norms, values, beliefs and attitudes that informally run through the organization. Culture may conflict with the rules and procedures determined by an organization’s structure and function; see, for instance, the frequently mentioned case of Enron, the American energy company that collapsed under the weight of ethical misconduct to which a culture of dishonesty and epistemic vice had significantly contributed.²³ Miller’s view entails that subgroups within an organization, however, may have significantly different, possibly competing subcultures. The last ingredient of an organization is, finally, a system of formal and informal sanctions. Informal sanctions are an immediate consequence of an organization’s culture inasmuch as organizations (or parts of organizations) have ways to express disapproval with non-conforming members. Particularly in business, however, formal sanctions are an important part of the organization, incorporated in internal disciplinary measures, dismissal procedures, incentive schemes and so on.

Back to the puzzle

Recall that Lahroodi observed that there may be a group of individually open-minded people who as a group lack open-mindedness. A real-world example of this phenomenon was a situation in which a firm’s directors individually possess love of knowledge, but not as a group. This shows that mere individual epistemic virtues are not sufficient to incorporate epistemic virtue. We have also seen that individual epistemic virtues are not even always necessary for group epistemic virtue. The argument here was the feasibility constraint that if corporate entities are to be open-minded only if all of their members are, then hardly any corporate entity would count as open-minded.

That individual epistemic virtues are neither necessary nor sufficient is not altogether to deny the correlation between individual and corporate virtue; rather, the correlation is more complex than has been described above. To understand the complexities it is useful to look at the concept of function as it occurs in Aristotle’s derivation of eudaimonia and virtue.²⁴ A textbook example to explain the concept of function considers the function of a tool – a hammer, say. A hammer has a function for a carpenter to drive nails into wooden objects to join them. For the hammer to fulfil this function well, it should be made of the right materials and be of the right size and weight. This is what sets it apart from a pair of tongs or a croquet mallet. A carpenter’s hammer that optimally meets these requirements can be called an excellent hammer, or a virtuous one, because it fulfils its function excellently.

Aristotle famously extends functions to natural phenomena, animals and human beings, defining the function of human beings as an activity of the soul that conforms with or not – ‘unconforms’, he writes – with reason.²⁵ This extension of function to human beings has prompted the response that function presupposes a form of design that, the objection goes, is absent in everything not man-made.²⁶ Neither Aristotle’s cryptic formulation nor the critique should detain us here, however, because it is my intention to consider roles and functions in the context of business only; and these are paradigmatic cases of human design, for what a company should aim at, or what an employee should do, is up to human beings to determine.

The idea of exploiting Aristotle’s concept of function as a tool to gain a deeper understanding of corporate virtues is not entirely novel. One line derives corporate virtues from the function of the firm. Assuming it is the function of a firm to generate sustainable profit, this line considers such corporate virtues as efficient production, resource management and correct pricing.²⁷ Another line derives individual virtues from the function of the firm, holding to the view that for a firm to fulfil its function well, its employees, managers and directors also have to fulfil particular functions, and from these particular functions particular virtues follow. This view is defended by Robert Solomon, among others.²⁸ My derivation of epistemic virtues follows Solomon’s line. The idea is that just as the different functions of hammer and tongs within the carpenter’s workshop give rise to different ‘virtues’ for these tools, so too do directors, managers and employees require different epistemic virtues within the firm.

Board roles

To get a better understanding of how virtue-to-function matching works, let us look at the roles and functions of members of the board of directors of a firm. As a rough point of departure, the function of the chief executive officer (CEO) is that of the managing director, determining the company’s strategy and acting as an agent for its owners, to whom fiduciary obligations are owed. The chief financial officer (CFO) acts as a controller and financial director of the firm. CFOs are in charge of a firm’s financial reporting, they interact with the capital markets and they play an increasingly significant role in strategic decision making. Besides these two directors, boards may feature chief development officers, chief governance officers, chief information officers, chief internal control officers, chief operating officers, chief risk officers and chief technology officers, not to forget the non-executive directors, sometimes collected in continental-style supervisory boards, whose function it is to monitor and oversee management.³⁰

That virtues have to match functions becomes clear if we contrast the CEO and the chief risk officer (CRO). Unlike the CEO, whose appetite for risk is sometimes necessary to realize long-term innovative projects, the CRO has to safeguard the company from risks, but not excessively, for that could endanger the firm’s expected profits. In a telling statement, Stefan Schmittmann, CRO of Commerzbank, Germany’s second-largest bank, explains that this requires not only that CROs have actively to strive to uncover hitherto hidden risks, but also to be open to discussion and disagreement.³¹ Explorativeness, temperance and justice are the primary epistemic virtues for the CRO. This contrasts with CEO virtues in many ways. The CEO is the firm’s main decision maker and communicator, bearing the main responsibility for the company’s strategic decisions and management. Embodied in the recognition rules of the firm, CEOs represent the company to shareholders and other stakeholders and are the main source of information to investors, journalists, consumer organizations, employee representatives, suppliers, governments and non-governmental organizations. The ensuing epistemic virtues form a diverse group. CEOs need epistemic generosity because they are an important source of information to the company’s stakeholders. They need humility and justice because they must pick up signals from the market and the employees, from suppliers and customers, even though they may consider these sources of information to be subordinate. They also need courage to admit that in many cases they are much less knowledgeable on particular company details than their subordinates.

Which particular epistemic virtues are required of particular functions is often a relational matter. Epistemic virtues are required of particular roles in their dealings with other particular roles. Consider another example. For a company to function well and remain profitable, it is important that no trade secrets and other sensitive information leaks to competitors, and as a result of this, epistemic generosity towards competitors is a vice. Equally it is a virtue towards non-executive directors, who have to supervise the company. Non-executive directors depend for their knowledge of the company on the willingness of executive directors to provide sufficient, adequate and understandable information about the company.³² That non-executive directors increasingly turn to other sources of information reflects the fact that terse, even reticent CEOs form important obstacles to adequate monitoring of the firm.³³ CEOs often limit the amount of information they provide and it is often impossible for non-executives to discover whether what they receive has any value to them.³⁴ One director put it thus:

Virtues of overconfidence

CEOs should share information more generously with non-executives, but one thing should not be demanded from them: epistemic temperance. Or so it might seem. Empirical and theoretical research suggests that some functions and some relationships require not so much epistemic virtue but rather epistemic vice. A recent article by David Hirshleifer, Angie Low and Siew Hong Teoh, for example, addresses what they call the ‘[t]he biggest puzzle raised by existing research on managerial beliefs and corporate policy’, namely, that companies frequently hire overconfident managers, allowing them to ‘follow their beliefs in making major investment and financing decisions’.³⁶ Companies should hire managers who lack epistemic temperance.

Or should they? Hirshleifer and his co-authors complete their puzzle by showing that overconfident managing directors are better innovators, which they think is a good reason to hire them; their prime example is the late Steve Jobs, former CEO of Apple. The connections between epistemic virtue theory and these forms of behavioural research are worth investigating because, as I indicate at several junctures in this book, such research helps reveal where epistemic virtues are needed and why realizing epistemic virtues is difficult, and also what forms of education or training may assist people to acquire virtue. If CEO overconfidence is instrumental in realizing corporate aims such as innovation, epistemic virtues may not be as relevant as I claim.

But matters are not entirely as they may appear. First of all, we have to pay close attention to the way behavioural researchers measure or ‘operationalize’ overconfidence and other epistemic vices. We cannot measure people’s overconfidence in the same way we measure their height, and that is why researchers attempt to develop proxies for overconfidence. An approach in behavioural finance pioneered by Ulrike Malmendier and Geoffrey Tate is to consider options exercise as a proxy.³⁷ CEOs typically receive large allocations of options on shares in their own company as part of their compensation package, often surpassing their base salaries several times over. In 2012, for instance, Jamie Dimon, CEO of JPMorgan Chase, the US multinational bank, received options worth $5 million (with a base salary of $1.5 million and $12 million in shares). Only after what is called a vesting period, and only if various pre-determined performance criteria are met, will Dimon be able to exercise these options, in conformity with general rules of corporate governance. It is this feature that Malmendier and Tate exploit to define a measure of overconfidence. CEOs are overconfident when, very roughly, they hold these options even after the vesting period has expired.

Why does this make initial sense as a measure of overconfidence? The next chapter explains how diversification removes a certain sort of risk from a portfolio of assets: it is better to have shares in different companies and different sectors than in one only (except if the company in which one invests is going to do better than the diversified portfolio). A CEO who does not act in line with the tenets of diversification, sticking to the options after the vesting period, is markedly irrational unless, for sure, the CEO holds the belief that the company is going to beat the market. Malmendier and Tate assume that this belief betrays overconfidence. If you think that your company is going to do better than a diversified portfolio, then you are overconfident.

In another article, Malmendier and Tate put forward an alternative measure of overconfidence, based on press coverage.³⁸ They scanned newspaper articles in the New York Times and a handful of finance periodicals for the company name and the word CEO and for the following words that, the researchers believed, relate to overconfidence and its contraries: confident, confidence, optimistic, optimism, reliable, cautious, conservative, practical, frugal and steady. CEOs using the first four words more often than the last six count as overconfident, according to this alternative measure.

The exercise of options and press coverage may have their use as proxies in finance; it is rather less clear whether they adequately capture the concept of epistemic vice and virtue. It may be the case that the decisions of CEOs to retain their options stem from beliefs about the prospects of the company that they gained in epistemically unvirtuous ways. It may also be the case that journalists use the word confident to describe what is more accurately called an overlyconfident director. But optimistic beliefs about the company are not necessarily the result of epistemic vices, and journalists do not always use understatement (probably quite the opposite). Behavioural researchers readily admit that a subtler measure of overconfidence has to capture the way CEOs use information, interact with other directors, management, employees and others; significant progress may be expected here as behavioural finance is a rapidly growing field. My remarks, then, are meant to caution against deriving hasty conclusions about epistemic virtues from empirical research, rather than to criticize the research itself.

Secondly, what behavioural economists call overconfident may in reality be a virtue that matches a function. If CEOs are expected to spur innovation and if this requires a certain form of bold confidence, then overconfidence may be a virtue for them. Consider again the difference between the chief risk officer (CRO) and the CEO. Epistemic temperance is, as we have seen, typically required of the CRO, but it is not particularly a character trait that follows from the job description of a CEO. What we condemn as excessively confident or overly optimistic in a CRO we may praise as adequately risk-loving in the CEO. (Just as we call overly precise the baker working like a research chemist.) This observation is backed by empirical work showing that boards are more likely to fire not only CEOs embodying the one extreme of temperance,diffidence, but also those showing the other extreme, clumsily called ‘excessively overconfident’ in the literature.³⁹

This brings me to a third point. The view I defend in this chapter is that specific epistemic virtues that employees ought to possess should originate from the firm’s corporate aims. Given such a point of view, only research that relates virtues or vices to the realization of corporate aims is relevant to an evaluation of the empirical plausibility of epistemic virtue theory. Innovation is absolutely essential to almost all firms, but it cannot be the ultimate goal of any firm. Innovation for its own sake is senseless. For our purposes, it is perhaps more relevant than the work of Hirshleifer and his co-workers to consider research on overconfidence and the generation of shareholder value. To begin with, firms run by overconfident CEOs are thought to pay out smaller dividends.⁴⁰ This is an obvious negative side effect to the owners of the firm (except perhaps if it is offset by capital gains). Moreover, when overconfident CEOs engage in mergers and acquisitions they are more likely to fail due to the fact that, for example, they have inflated ideas of their own capacity to generate profit.⁴¹ Overconfidence is, more generally, claimed to have an adverse impact on corporate investment decisions.⁴² These are plain disadvantages to the shareholders, but other authors again claim that managerial overconfidence contributes to the solution of coordination problems and that it mitigates moral hazard, which is good for shareholders.⁴³ Yet all in all, though the results are mixed, little evidence shores up the claim that overconfidence is an essential epistemic character trait that CEOs ought to possess to fulfil their function as the prime steward of the firm and its owners.

This critical excursion into behavioural finance is intended as an example of the sort of empirical and conceptual argumentation that we should engage in when it comes to applying epistemic virtue theory to business, finance in particular. But it also points to a number of methodological hurdles that any such application encounters. Instead of stressing these difficulties even more, I return now to the question of how organizations can acquire epistemic virtues at the corporate level, in particular by providing organizational support for virtue.

Structure

Seumas Miller’s four-pronged analysis of organizations in terms of structure, function, culture and sanctions helps to introduce a certain systematization in the various forms of organizational support. I start with structure. Relevant here is research on knowledge management. The useful concept of knowledge management cycle sees firms start by creating and capturing knowledge, continue with sharing and disseminating knowledge, and finally applying the knowledge.⁴⁸ Knowledge management research provides insights in ways that, at any of the three stages of the cycle, help or hinder employees to proceed in epistemically virtuous ways.

Consider epistemic generosity. This virtue can make or break a business. Robert Grant and others have pioneered what could be called an epistemic theory of the firm, advocating the view that firms are means by which managers coordinate and integrate the knowledge that resides in the individual employees with the end of developing certain products or services.⁴⁹ Knowledge sharing is found to increase a firm’s competitive advantage, to help firms turn abstract ideas into concrete products and services, and to improve problem-solving capacities.⁵⁰ The need for epistemic generosity is further underlined by the estimate that around half of the knowledge within a company resides ‘in employees’ brains’, only leaving those brains if they share.⁵¹ This is reinforced by the following estimates, derived from a study by the International Data Corporation.⁵² Knowledge workers spend around 15 to 35 per cent of their time seeking information, but they are successful in less than half of that time. Most of the information employees search for is readily available from colleagues; but the searchers do not know that. The explanation is not that the colleagues lack epistemic generosity. Knowledge sharing does not happen because there is no effective organizational support. Many firms have set up an intranet, which employees can use to post messages and requests for information. As Don Cohen and Laurence Prusak observe, however, knowledge sharing may work much better around the coffee machine than on an intranet.⁵³ This has led knowledge management scholars to stress the importance of endowing employees with sufficient time for knowledge sharing as well as a physical location for face-to-face meetings. It also shows that the optimism concerning the New World of Work or the officeless company may be misplaced.⁵⁴

Another example of structural support for generosity is this. LabMorgan, a business unit of JPMorgan Chase, invested in web-based financial services.⁵⁵ Using Intraspect Software it created in 2001 a knowledge management site called Deep Thought. Deep Thought is a form of cloud computing, consisting as it does of nothing more than a set of web-based folders in which employees store business plans received from customers. It contains the fairly intriguing trick, novel at the time, that the folders have email addresses; users can upload documents by sending the document as an attachment to an email message. While the idea of Deep Thought is simple, it drastically changed the way employees shared information about business plans. It allowed users to go through all previous business plans, feedback and decisions, and to hold up new proposals against them. The system did not make employees more epistemically generous; rather it offered organizational support to epistemically generous employees to act on the virtue.

Culture

Culture being roughly a set of beliefs and values characterizing a corporation (or part of it), it impacts negatively on epistemic generosity if a corporation is infused with the idea, for instance, that knowledge belongs to particular employees rather than to the entire organization, or in organizations where employees receive recognition primarily for the knowledge they possess instead of for their knowledge sharing. If teamwork is almost entirely absent, if only particular people are accepted as authoritative sources of information, if employees feel that they are not allowed to make mistakes and reveal ignorance, or if they do not speak the same language, sharing knowledge will not get off the ground even if all employees are individually epistemically generous.⁵⁶

How can organizations support generosity? Sometimes a change in culture works. People tend to consider knowledge as personal property, in particular when they have played a significant role in acquiring the knowledge. In many business enterprises, however, knowledge is rather seen as the property of the organization; knowledge in such organizations remains anonymous, not attributed to individual employees or work groups. The combination of the individual private property view of knowledge and the organizational anonymity view clash, with the result that employees keep their knowledge to themselves rather than generously sharing it with others. It is rather difficult to change individual views of property. Organizational support for generosity, however, can be gained when we slightly change the culture in the organization and attribute knowledge to particular employees and start making authorship of knowledge explicit. An example can be found at Xerox, the American document management multinational. Xerox engineers help customers by developing innovative solutions for technical problems of printers and other devices. Ideally, when all engineers share their solutions freely, no engineer has to reinvent the wheel. At Xerox, however, knowledge sharing was found to be relatively poor. The company implemented a knowledge management strategy and set up a database to which engineers could contribute their solutions. An important incentive for the engineers to contribute to the database turned out to be the fact that the names of the engineers figured prominently in the entries they contributed. A minimal change in culture was sufficient to obtain the desired result: Xerox acknowledges authorship of knowledge.⁵⁷

Sanctions

Knowledge may not only be hoarded as private property; employees may also think of knowledge primarily as a means to increase their power. To support generosity here a change in sanctions may be needed, that is, a change in systems of punishment and reward. Knowledge-as-power views are exacerbated by a stick-and-carrot system praising people for the knowledge they possess rather than for the knowledge they share. Epistemically virtuous organizations have alternative systems of sanctions in place. As I have already noted, under CEO Jack Welch of General Electric, knowledge sharing became an important target for the firm. A knowledge sharing warehouse was, for example, developed for information and experiences with customer complaints about quality issues, which employees could access from the intranet. This was part of a larger move towards an epistemically generous culture. Kimiz Dalkir, from whose monograph on knowledge management this case is taken, describes the importance of knowledge sharing at General Electric thus:

The knowledge management literature is awash with examples of companies rewarding employees for their knowledge sharing with shares or options or with gift cards they can redeem in online stores. Consider an example from Hill and Knowlton, a global public relations company. It takes a very unadorned approach against hoarding knowledge and for sharing it. Realizing that employee performance reviews, due to their low frequency, are unsuited to providing employees with effective feedback on their contributions to knowledge sharing, this company decided to give employees immediate gratification. It changed the sanction system and decided to award employees sharing their knowledge with coins in an electronic currency they could redeem for cash when shopping at participating online stores.⁵⁹ Organizational support does not need to be difficult.

By rewarding some behaviours and sanctioning others, compensation schemes can influence epistemic virtue in other ways as well. Bonuses are an example. They have been criticized by the popular press, attacked on account of their having provided employees with perverse incentives that led them to work not with the customer’s but with their own interests in mind. Some scholars have also pointed out that incentive schemes can be ‘gamed’ in such a way that it looks as though the employee has done a lot of work, but in reality has not. Here I show that they also have an epistemic side, the main claim being that it is very hard to set up performance-based remuneration contracts with employees working as portfolio managers in the financial services industry in such a way that senior management is able to find out whether the employees have done their job, or gamed the system.

An article by Thomas Noe and Peyton Young makes this more concrete.⁶⁰ Noe and Young invite us to imagine a trader in an investment bank whose remuneration package includes the following bonus. His investments will be compared to the S&P 500 index, and if he does better his bonus will amount to 20 per cent of the difference between his investments and the index. The trader controls $500 million in assets, which he chooses to invest in the S&P 500, gaining some dividends, and in addition selling asset-or-nothing put options to a third party. Such options are to the effect that if at the end of the month the S&P 500 is below a certain price, the option holder receives all shares in the trader’s fund. Noe and Young specify the option in such a way that the likelihood of this event is 5 per cent. Consequently the trader can sell them for 5 per cent of the price of the total number of shares in his portfolio; after selling the options the trader can buy new shares in the S&P 500 for 5 per cent of $500 million. In the words of the authors, he then awaits the end of the month, ‘keeping his fingers crossed’.⁶¹ The probability that the trader will not lose is 95 per cent, and if that happens he can sit still for the rest of the year, claiming a bonus of 20 per cent of the 5 per cent of $500 million by which he has outperformed the index. His personal gains in that case amount to $5 million. No doubt, the options will be called with a probability of 5 per cent, but Noe and Young emphasize that this

A trader calling the options loses the shares. But this only happens with 5 per cent probability. An epistemically virtuous manager wants to know whether the trader’s success in beating the index should be ascribed to brute luck or to skills and talent. Firms do not want to dole out lavish bonuses to employees whose success is merely a matter of luck. What the authors point out, however, is that performance-based compensation schemes make it very difficult for managers to distinguish talented from lucky traders. In the example, the trader’s investments may collapse in the first year, and management may fire him; but management may also blame the collapse on ‘bad luck’. The trader may, on the other hand, beat the market during the first year; but then it is difficult for management to find out if his success is due to skill. The trader’s books contain information about the positions and thereby allow the firm to uncover the underlying investment strategies, but though this is relatively easy in the case above, traders can design highly complex strategies, which it will be very difficult and costly to discover.

The literature on incentive schemes has primarily investigated the effects of bonuses on individual behaviour, and has in particular defended the claim that they distract professionals from paying sufficient attention to their customers’ interests. We have now seen that bonuses also make it more difficult for management to act virtuously. Even managers with high levels of inquisitiveness, attentiveness and perceptiveness likely miss manipulation where it occurs. Performance-based compensation contracts such as the one above erect smokescreen inhibiting epistemic virtue.

Macro-level remedies

Suppose a bank develops a new product. Knowledge about the product may originate from a product design team, but also from marketing, risk management, compliance, or even from the corporate social responsibility, sustainability or ethics department. Suppose that ultimately the board decides whether the product will be marketed. The decision is an easy one when all units agree about the product’s potential. Generally, however, boards are confronted with conflicting views before making the final decision. They have to weigh more ‘optimistic’ product design and marketing views and more ‘pessimistic’ risk management, compliance or sustainability views.

Research on judgement aggregation makes abundantly clear that the order in which boards or other bodies deal with particular items on an agenda radically influences the ultimate decision. To take one striking example, had the German parliament in 1991 changed the order of proceedings during the voting that determined the future capital of unified Germany, Bonn not Berlin would likely have come out the winner.⁶³ Research in psychology has shown, moreover, that the mere order in which one receives information influences the beliefs one ultimately forms. This effect is an interplay of confirmation bias, the sunk cost fallacy and other phenomena. People pondering which doxastic attitude to adopt towards some proposition p, on the basis of available evidence, are rather likely to interpret the evidence as emphasizing their prior belief about p, thereby displaying the confirmation bias. Suppose that at t₀ I am neutral with respect to whether a product should be marketed, and that at t₁ I receive ‘optimistic’ information concerning the product. Then I am on the whole more likely to disregard ‘negative’ information about the product obtained at a later point in time t₂, or to misconstrue such negative information by interpreting it as support – this is the most striking aspect of the confirmation bias – rather than counterevidence for the optimistic views I adopted at t₁. The conclusion is that ‘pessimistic’ risk management, compliance and sustainability views are set at an epistemic disadvantage when they are deferred to the end of the epistemic decision procedure, which they usually are.

This is exacerbated by the sunk cost fallacy.⁶⁴ When a firm has so far spent £1 million on designing a new product, these costs are sunk, owing to the fact that they have been incurred and cannot be recouped. To determine whether the product shall be put on the market it is rational only to consider the future cash flows and other future costs and benefits of the project. The decision should not depend on how much money has already been spent on the project, but only on what we can expect the project to deliver in the future.⁶⁵ Human beings, however, find it hard to do this; they find it hard to ignore sunk costs. One readily spends £200 on a ticket for a concert, realizes that the concert is dreadful, but one does not leave because the money has been spent, as the traditional textbook example has it.⁶⁶ Similarly, if after spending £1 million on the design of a product, risk management, compliance or sustainability gives a more ‘pessimistic’ evaluation of the product, the ‘pessimists’ are down 1–0 at halftime. A risk manager contributing to The Economist in 2008 explained:

Obviously much research remains to be done to find ways to remedy confirmation bias, sunk cost fallacy and similar psychological effects. Much can be gained from fairly simple procedural adjustments, though. One example is to avoid unnecessary temporal divisions. Risk management, compliance and sustainability, for instance, should participate actively in the design process from the very start, rather than at the end.⁶⁸ Boards should actively solicit not only evidence in favour of a prospective project, but also against it. It may well be the case that Goldman Sachs came through the crisis relatively unscathed because it was permeated with organized opposition. Simplifying somewhat, the idea was that instead of asking the authors of a business proposal to investigate not only the advantages but also the disadvantages, a proposal was, at Goldman Sachs, critically examined by a different team, independently of its authors. It should not come as a surprise that independent critics are better positioned to detect potential flaws than a project’s advocates.⁶⁹

Micro-level remedies

Organizing opposition in order to counter groupthink or sunk cost fallacies is a proven strategy for remedying epistemic vice at the macro-level of the organization. I now turn to micro-level remedies, which are probably useful more frequently because they require a lesser degree of intervention with organizational structures, cultures or sanctionary systems of reward and punishment. To defend this claim, I discuss a topic that has attracted a great deal of attention in academia and elsewhere: rotation policies. This is also a fine example of micro-level design that is interesting in its own right. For the purpose of introduction, consider the accountancy profession. If accountants and their clients develop too close a relationship, the objectivity and independence of the accountants’ views of their clients’ financial positions may decrease. Hence mandatory rotation policies have been suggested in which firms must change their accounting house after three, six or eight years.⁷⁰ Accountancy is not alone here. For similar reasons, the rotation of non-executive directors has been proposed in Britain and South Africa, requiring such things as that a third of all non-executives retire every year. Mandatory rotation has been recommended for tax officials in Bulgaria, and a widely publicized suggestion of a rotation policy was made by Michel Barnier, the European Commissioner for Internal Market and Services, when in 2011 he proposed that issuers of securities should be required to change their credit rating agencies every three years. The idea is simple. When accountants, non-executive directors, tax officials and credit rating agencies are swapped from time to time, their views about the firm, its products, its taxes or its debt come closer to the truth.

I defend the claim that rotation policies are micro-level remedies for epistemic vice. I focus on epistemic justice; other policies remedy other vices. I should tread carefully here. Most of the research on mandatory rotation has focused on accountancy, and even then, the results are mixed. They reveal an increase of independence, but one may fear that will be accompanied by a loss of knowledge and information. A long-term relationship with a client risks an accountant’s independence, but the knowledge an accountant gradually gains about the firm is greater than what a fresh accountant knows; and the increased depth of knowledge may well outweigh the decrease of independence.⁷¹ I do believe, though, that effective rotation policies help incorporate epistemic justice. The mechanism by which rotation policies accomplish this feat is not that of virtue-to-function matching ensuring that individual employees possess this virtue, nor is it that of providing organizational support for already epistemically just employees. Rather, rotation policies are ways to guarantee epistemically just belief formation practices in an organization populated by employees lacking epistemic virtue or only possessing it to a slight degree. Though an individual accountant or firm may not be a gem of virtue, interchanging them guarantees objectivity and independence at a macro-level.

With the exception of director rotation, the policies mentioned above do not concern employees within one firm, but rather relationships between firms. To defend the claim that rotation policies help firms incorporate epistemic virtue, I use an example from banking examined by Andrew Hertzberg, Jose Liberti and Daniel Paravisini.⁷² This is a rotation scheme at an Argentinian branch of a large US multinational bank among loan officers dealing with small and medium-sized enterprises. The function of loan officers is to screen loan applicants by determining the risks attached to the projects for which they request finance. In addition, loan officers have to monitor the loans, both actively and passively. Active monitoring is a prospective form of monitoring that starts right after the loan has originated. Its aim is to raise the net present value of the project, that is, to exert a positive effect on the future cash flows or value of the borrower’s project. This requires various means to diminish the probability that borrowers will encounter problems with repaying their loans, ranging from the recommendation of adjusted conditions for new lending to downright interference in the borrower’s affairs. Passive monitoring, by contrast, is retrospective and aims to measure rather than affect the project’s value. It provides continuously updated estimates of the probability that the borrower will pay back in time.⁷³

Why is monitoring a possible site of epistemic injustice? A problem may arise when one loan officer monitors a borrower both actively and passively. A dismal picture generated by passive monitoring reveals that the borrower’s project is in bad health, but may also expose the loan officer’s inadequate active monitoring because the officer had not been able to intervene constructively in the borrower’s project. A loan officer downgrading the probability estimate of a borrower’s repaying the loan therefore endangers her reputation, which often results in demotion or a cut in the number of assigned clients. This gives the loan officer an incentive to withhold bad news about the project and provide dishonest, or at least deceptive, reports.

The aggregate result of individual dishonesty and deception is corporate epistemic vice. It affects love of knowledge in the first place. As we saw in Chapter 3, love of knowledge motivates one to search for information; it also leads one to adopt beliefs only if one possesses sufficient evidence to justify the adoption of the beliefs; evidence-based practices in marketing provided an example. It is crucial to note that individual dishonesty and deception do not lead to corporate lack of information search; information is searched after all. The problem rather is lack of justification. If individual retrospective judgements concerning the net present value of a borrower’s project are upwardly biased, the aggregate corporate judgement is based on what looks like justifying evidence (evidence obtained from individual loan officers) but in reality is not. The corporate view is skewed in the same way as individual views are skewed when they are based on biased evidence. There is a lack of justification.

It may be argued that as long as the problem lies in the loan officers’ failure to pass on sufficient information to other positions in the firm, the corporate problem arises due to lack of individual epistemic generosity; and in the light of the above, two solutions may suggest themselves. One is to match function and virtue by hiring more epistemically generous loan officers; the other is to provide organizational support for generosity. From a theoretical point of view, the first solution is plain. Let me therefore focus on the second. Organizational support for generosity may result when management separates active and passive monitoring and makes loan officers responsible for one only. Yet this change in the structure of the organization is likely to go against business economic considerations. Banks want to stick to combining active and passive monitoring on the grounds that this exploits economies of scale. Because information garnered from active monitoring benefits passive monitoring (and conversely), separating these roles leads to inefficiencies, which banks want to avoid. The suggested form of organizational support is, then, theoretically very elegant and simple, but it is too expensive to catch on. Hence banks ought to consider other ways to incorporate virtue.

The rotation policy studied by Hertzberg and his colleagues is an example here. According to the policy they examine, a loan officer may at any time be withdrawn from a particular client (borrower), but the chance that this happens will change over time. During the first thirty-three months of a relationship with the client, the probability that the officer is replaced is below 5 per cent. As soon as the relationship has survived the thirty-fourth month, it is terminated with a probability of 58 per cent during the next three months, after which the termination rate decreases again. The details may sound arbitrary. The precise design of the policy, however, is of considerable importance because it ensures that loan officers know that they are very likely to be withdrawn from the project after around three years; the rub is, they are unable to predict precisely when.⁷⁴

All this would not be an effective way to encourage corporate epistemic justice were it not for a very important finding about loan officer reputation that Hertzberg and his co-authors uncover. Identifying a loan officer’s reputation with the volume of assets she has under management (as measured by the number of firms and amount of debt managed), the authors first stipulate that absent a regime of rotation one should expect loan officers who downgrade a firm always to suffer a loss of reputation. Confirming these expectations, the study subsequently shows that with the three-year probabilistic rotation policy such effects can only be seen in the third year. No reputational effects occur during the first two years. The cause of this difference is important. The authors suggest that during the first two years the loan officer’s superiors believe that it is not the loan officer who should receive the blame for inadequate active monitoring, but her predecessor. Officers reporting bad news about a firm they have actively monitored for more than two years, however, typically suffer a significant change in assets managed.

This is all fine, but it is in the most significant reputational damage a loan officer may face that we can find an explanation of why the rotation policy is an effective way to increase corporate epistemic justice. If a loan officer’s successor reports bad news about a client, the loan officer is confronted with a decrease in managed assets that may be up to four times as large as had the loan officer reported the bad news herself. Though bad news about a firm one has managed for more than two years never benefits one’s reputation, it is, in other words, much more preferable to do the reporting oneself than to let one’s successor be the bearer of the news.

To summarize, the rotation policy examined here increases epistemic justice at the corporate level along two lines. Loan officers report bad news during the first two years of their relationship with a client because they do not face any reputational damage; and they report bad news during the third year because, although they incur reputation loss, the loss is greater if they leave it to their successor to report. It is important to underscore that this is a phenomenon of corporate rather than individual epistemic justice. The loan officers display no epistemic injustice themselves; they lack epistemic generosity, as we saw, and may be dishonest, but I see no indication that in the process of passive monitoring they disregard particular items of information, fail to engage in particular forms of inquiry or otherwise act in epistemically unjust ways. Individual lack of one virtue entails the lack of another virtue at the aggregate corporate level. Absent rotation policies, the bank has too optimistic a picture of the health of its borrowers’ projects, and proceeds to act on this mistaken picture. Though the individual loan officers have adequate views of the value of their projects, they do not communicate them truthfully. Remedying these vices, the rotation policy helps to incorporate epistemic justice.