Rethinking Hierarchy

For all the hype and promise swirling around the idea of eliminating management to create agile, flat organizations, bosses and corporate hierarchies have remained extremely resilient. As we argued in the pages of MIT Sloan Management Review in 2014, under the right conditions, having such hierarchies in place is the best way to handle the coordination and cooperation problems that beset human interactions.1 They allow human intelligence and creativity to flourish on a larger scale. They provide a larger structure, with predictability and accountability, for specialists to do their work.

But that doesn’t mean traditional, command-and-control organizations are right for today’s environment. We see a confluence of business and social trends influencing the development of new kinds of hierarchies. Rapid technological progress, instant communication, value creation based on knowledge rather than physical resources, globalization, and a more educated workforce require us to rethink how we wield managerial authority. Meanwhile, individual views on politics, religion, and culture also inform our attitudes toward hierarchies — such as whether we value autonomy or admire authoritarian figures. All of these factors point to a new, different role for hierarchy to play in meeting the challenges of the 21st century.

The key challenge for designing and operating hierarchies today and tomorrow is to balance two opposing forces. The first is the desire, common to us all, for empowerment and autonomy, which helps companies mobilize employees’ creativity and exploit their unique knowledge and capabilities. The other is the need — particularly in environments characterized by rapid change and interdependent activities across the enterprise — to exercise managerial authority on a large scale.

Companies need clear, fairly enforced policies and procedures that achieve coordination and cooperation while respecting employee desires for empowerment and relative autonomy. Managers have to figure out when to intervene and when to let employees handle problems themselves.

These are tough issues without easy solutions. Which decisions should be decentralized (or delegated)? How much discretion should employees have over the decision areas delegated to them? How are these employees incentivized and evaluated? How do executives make sure that all these decentralized decisions mesh together? A central lesson of theories and evidence on organizational structure is that there are no universally “best” answers to these questions, only trade-offs that depend on the contingencies facing the company. Identifying and acting on those trade-offs — not decentralizing everything, everywhere — is the key to successful leadership.

Some proponents of the so-called bossless company pretend that these trade-offs don’t exist. They hold that only full decentralization or delegation — “make everyone a boss” — makes companies innovative and adaptable.

When COVID-19 was declared a pandemic in March 2020 and lockdowns began, few leadership gurus proposed making everyone a boss to address the challenges that companies suddenly faced. Rather, managers had to figure out how their companies could survive, which meant changing their business models, enabling remote work for employees, and adjusting to massive demand shocks and supply chain disturbances. Of course, the need for many companies to shift to a remote work model required that many tasks and decisions be delegated. The point is that these changes had to be coordinated, with key decisions made from the top.

To better manage the tension between hierarchical control and individual autonomy, we need to rethink managerial authority.

Rethinking Authority

There is compelling evidence that many companies are removing layers of management and that more decisions are being delegated to employees. Organizations are still hierarchical, but hierarchy is changing its form. There is also evidence that the exercise of authority is changing.

Authority has many faces: It may mean the right to hire, fire, instruct, supervise, intervene, and sanction. But the exercise of managerial authority is also associated with other behaviors: leading, creating structures and processes, forging consensus, aligning behavior around shared goals, and fostering change.

We call these two types of authority Mark I and Mark II, respectively.2 They roughly correspond to the distinction between management and leadership. Clearly, the same manager can exercise both types of authority, but it is crucial to understand how they differ and when each is most appropriate, because using one when the situation demands the other can be disastrous.

The traditional view of authority (based on the work of Ronald Coase, Oliver Williamson, and Herbert Simon) assumes that the boss can select the appropriate task, knows all the possible ways to perform it, and can observe the output, so that rewards and punishments can be administered appropriately. In other words, the boss doesn’t want to perform the task directly (perhaps because she is too busy with other tasks), but she knows as much about the task as the worker does, and she may or may not observe exactly what the worker is doing as he performs it.

In the modern, knowledge-based economy, however, it is increasingly unlikely that bosses will know everything their workers know. A tech CEO might be skilled at finance and marketing or good at strategic planning and human relations but might not know how to code. A sales manager might understand the product but be unfamiliar with a specific sales territory.

In his later work, Simon described a second notion of authority (perhaps reflecting a changing view of what kind of authority is important in more modern conditions): The role of the boss is to decide what decisions should be delegated.3 That is, the boss selects a target outcome, decides which workers are best suited to achieve it, chooses how much discretion to give to those workers, and steps aside. Authority in this sense is not about choosing specific tasks and making sure those tasks are performed, but about setting goals, writing job descriptions, selecting people, and evaluating results. This is our Mark II authority — not micromanagement, but macromanagement.

Organizational norms currently demand that leaders shift toward Mark II authority and away from Mark I, because most workers don’t want or need a manager telling them what to do and when to do it. Rather, managers have to design the system in which empowered, autonomous workers can flourish.

Certainly, exercising authority via the choice of delegation can add new challenges. Ironically, some workers may prefer less delegation, perhaps because they are risk averse. Another drawback relates to interdependencies. An organizational structure with lots of delegation is more complicated than one based on simple command and control, and managers have to make sure employees are exercising the discretion given to them under Mark II authority in ways that are coordinated with each other and the company’s goals. Moreover, company decision-making may be slower under Mark II authority, especially if workers need to discover solutions independently. So while this authority results in a system that is typically more rewarding for workers and more motivating than the traditional form of authority, it might not be the best choice in all situations.

How Not to Use Authority

It is hard to design a hierarchy well, particularly when employees are empowered and the company is delayered. Managers are increasingly facing this challenge.

One of us (Foss) studied the implementation and use of agile software development practices in a large telecommunication firm, which we will refer to as Company E, across multiple sites over five years.4 Company E adopted the agile methodology as a part of its efforts to be less bureaucratic and more flat; its goals included being able to respond more quickly to market needs, reducing development time and costs, and supporting organizational learning. While the change sped development and cut software maintenance costs, the emphasis on meeting deadlines hampered learning and innovation: Team members reported having less time for reflection, bouncing ideas off each other, and sharing knowledge. And despite the implicit intention to free employees from bureaucracy, the agile method gave project leads the authority to closely track performance, leaving many team developers feeling stifled by an overbearing managerial hierarchy rather than self-managed.

Enabling Authority

One of the most important ways in which managers exercise authority is in how they delegate it to others. To enable authority in subordinates, managers must first convince them that they really do have autonomy over certain decisions and actions and earn their trust that authority will not be summarily withdrawn. And when managers do, on occasion, need to override an employee’s decision, they need to proceed carefully to avoid breaking that trust. Putting employees in situations where leadership is shared, or where it rotates among team members, can be a good way to maintain this trust. Another good way to foster trust is to have explicit and transparent rules and procedures, which can reassure employees that their authority is real and will not be compromised by opportunistic, micromanaging supervisors.

Consider extreme action teams, which are made up of specialists who perform urgent and interdependent tasks with highly consequential but often unpredictable outcomes. Examples of such teams include U.S. Navy SEALs carrying out special operations or a medical team in a hospital receiving an influx of emergency patients during a natural disaster. These teams use dynamic delegation, an organizational structure under which roles and responsibilities are moved around quickly as needed; in this scenario, employees sometimes find that their decisions are overridden and that responsibility is taken away from them.5

A recent study found that dynamic delegation helps extreme action teams perform more reliably while also providing training and experience for less experienced members.6 The broader implication is that this kind of improvisational process works — but only because it happens within an established hierarchical structure. Moreover, overruling may not be perceived as a bad thing when team members think that the person doing the overruling actually knows best. If you are a medical intern treating a patient in critical condition and the senior physician overrules you, thus saving the patient’s life, then you are probably quite happy to have been overruled.

Despite such cases, there is little doubt that psychological contracts are often broken by managers engaging in unnecessary and intrusive micromanaging and overruling. Employees may also come to see their position in the workplace and the resources they can use as entitlements. Organizational changes such as restructurings can thus be perceived as breaches of the psychological contract, though the manager would not see it that way. Analyzing data from Spanish manufacturing companies, one of us (Foss) found that such breaches in fact hurt employee productivity.7 Morale and creativity decline, and employees start to leave.

The ability to establish, maintain, and signal good psychological contracts will be an increasingly important part of the “enabling” exercise of authority. This is one reason why highly decentralized organizations need a lot of structure.

Formalization and Transparency

When job descriptions are well understood and distinct, with little functional overlap, problems coordinating work or making decisions are less likely to arise between employees. But at companies with complex, interdependent processes, where two or more functions might have a stake in a decision point, it becomes critical to proactively clarify roles, decision rights, and the organizational priorities underlying those choices.

For example, at many technology companies, developers’ autonomy runs up against the demands of the cybersecurity team, which expects to vet any software before it is put into production. In order to keep coordination problems from becoming more poisonous cooperation problems — which may damage internal relations and be hard to repair — managers must weigh the need to move quickly and innovate against the imperative to maintain a certain level of security, and assign decision rights accordingly.

Many coordination problems are less straightforward than the example just described, and here the classic management instrument of clearly defining roles helps avoid turf wars. Sure, formalized job descriptions can be constraining — and can become perceived entitlements that obstruct organizational change or lead individuals or units to overstate their right to control particular organizational decisions. But these risks can be managed; choosing not to define roles and jobs and formalize processes and procedures is not a good alternative.

If we all know that Paul sings lead and plays the bass, John sings harmony and plays rhythm guitar, George plays lead guitar, and Ringo plays drums, and if all these parts fit together smoothly, the result is great music. Everyone knows his role and how the roles complement one another. Such transparency can work in a small consultancy, a repair shop, or a subunit in a bigger company with just a few employees whose clearly defined roles add up to a coherent whole.

Problems arise when there are conflicts among roles. Who is going to sing lead? Who gets songwriting credit? The larger and more complex the organization, the greater the likelihood of conflict. There are a lot more roles to define and fill. Employees come and go. It’s difficult to figure out who is doing what, how well tasks are being performed, and whether changes need to be made. Full transparency in a large organization may lead to information overload. This is partly why larger organizations are typically divided into departments, divisions, branches, project teams, or other subunits, all of which can be managed like smaller units.

Should the activities of one subunit be transparent to another? On the one hand, it might seem better to have each group focus on its own tasks and not be distracted by what’s going on in other units. Let Apple’s iPhone division focus on making great phones without worrying about how Apple Music or Apple TV is doing. Let the executive team be concerned about that. On the other hand, some knowledge of the company’s overall strategy and performance is necessary for any group to perform well; if Apple is investing more in content creation and integrating media properties with its hardware devices, the teams making these devices might want to design them differently. Finding just the right level of transparency between units is a key management challenge.

Exercising Authority Smartly

Every business model, strategy, organizational structure, or management style has its strong and weak points. Exercising authority smartly means figuring out what decisions to delegate, who to put in key positions, and when to intervene, as well as deciding whether the system needs to be revised in response to changing conditions.

The advantages of delegating a decision or an action may outweigh the costs under some contingencies or conditions but not others. Good managers know this intuitively, but that doesn’t make it easy to make the call. Getting delegation right is a perennial management challenge. Moreover, besides the “hard” aspects of organizational design, like formal rules, incentives, and monitoring, there are the “softer” parts from psychology and ethics to be considered: Employees expect rules to be fair, they take pride in being empowered, they want their jobs and roles to feel significant, and they are subject to the biases and judgment errors (such as jealousy, overconfidence, and motivated reasoning) to which everyone is prone — even managers. Taking all this into account is truly difficult.

Delegation is not simply a matter of empowering workers to do their best. Companies like Spotify and Valve, where software projects are mostly independent efforts with little need for coordination between them, can leave team size and composition, project requirements, and even some budgets up to the teams. Bottom-up organization — not literally bossless but flat, flexible, and organic — makes sense for these companies. But it might not make sense for a large manufacturing company that has a portfolio of highly interdependent products featuring shared designs and components and that sells bundles of complementary products. For such a company, more centralized control is needed. In deciding what, when, and how to delegate, managers can begin by asking a few simple questions to determine what will work best in their organization.

Is it more important to make the best decision or a faster, good enough decision? If there’s a high degree of urgency around decision-making, it’s often best for higher-level managers to make the decisions without dialogue and consensus.8 Of course, quick decisions may turn out to be wrong, but if the costs of delay are high enough, they are often worth the risk. Does this need for quicker decisions justify the potential for a demotivating effect on employees? That depends on the reason speed is needed, the capability of top managers, and the attitudes and expectations of employees.

Speedy decisions are often made in response to threats — typically a sudden and unexpected drop in company performance. In the face of an existential threat, a company’s decisions become highly consequential. Under these conditions, a decentralized approach is likely to perform less well, because subordinates are less willing to assume responsibility in the face of such risk and may make slow or poor decisions. In such cases, senior managers are likely to concentrate and retain control over key decisions.

What knowledge is needed, and which employees have it? Is the most critical knowledge for a given decision held by senior managers (who know the company’s overall strategy), by middle managers (who have a bird’s-eye view of their department or division), or by lower-level employees (who know their customers best)? Can knowledge at lower levels of the company be neatly summarized (say, in a customer relationship management database) and made available to higher-level managers? What about the reverse? Can strategic priorities be communicated to lower-level employees to explicitly inform certain decisions or actions? It is crucial that managers ask themselves these questions.

At a critical point in Intel’s history, when it transitioned from mainly producing memory chips to mainly producing microprocessors, its middle managers had a better grasp of the right strategic direction than top managers did. Still, Intel’s senior managers were often better positioned to grasp the key issues at the company, industry, and economy levels, as CEO Andy Grove demonstrated.

The fundamental task of a company’s top decision makers is to synthesize all information that could have a significant effect on the company’s current and future performance and make the big decisions on that basis. They can rely on advisers, experts, consultants, and what we earlier described as extreme action teams. But ultimately, the buck stops with the CEO. So while some decisions should be delegated to employees with superior knowledge of local conditions, overall authority — including decisions about which decisions to delegate — cannot be passed down the line.

What do employees feel they currently own or control? Many employees naturally feel a sense of ownership over what is currently in their control, such as their budgets and decision rights, and making changes to those decision rights poses particular challenges.

One of us taught a course for many years with complete control over topics, readings, and classroom activities, then moved to another university where the curriculum for that course was controlled by the department to ensure uniformity across instructors. Losing that autonomy led to a lot of frustration and resentment.

The behavioral economics concept of loss aversion comes into play here: People tend to value things they had and lost more highly than things they never had. Any choices that centralize decisions that were previously decentralized are going to cause some pain. Managers must convincingly explain the rationale for change to affected employees.

Is it fair? Organizations must consistently apply fair and transparent principles when resolving disputes — that is, they must demonstrate procedural justice. If employees believe that the processes are fair, they are more likely to accept managerial intervention, whether that be occasionally overriding an employee’s decision or centralized authority itself. Can managers explain when and why autonomy must be curtailed? Do employees feel that their concerns about autonomy and responsibility are heard and taken seriously? If so, then choosing not to delegate certain decisions, or to intervene when necessary, is likely to work. If not, employees could become disgruntled.

The need for hierarchy isn’t going away, but the form it takes is changing: deciding how things will be done rather than telling people what to do, and designing and enforcing the rules of the game rather than making everyone play it in a certain way. As Haier Group founder and CEO Zhang Ruimin put it, “Leaders of other enterprises often define themselves as captains of the ship, but I think I’m more the ship’s architect or designer. That’s different from a captain’s role, in which the route is often fixed and the destination defined.”9

In redesigning managerial authority and hierarchy for the 21st century, leaders must realize that they don’t need to know everything, but only just enough, and they need to consider what their employees want and think is fair in designing structures and systems.