Common Managerial Mistakes in Decentralized Organizations

Preface: A Brief Review of Decentralization

An organization is a group of individuals with a common purpose.

A decentralized organization structure is one in which subordinates are permitted a degree of autonomy to decide or act upon certain matters as representatives of the organization.

Strategic and managerial control involve the determination of the optimal level of decentralization (or delegation) and choices of related policies and mechanisms like budget processes, performance measures, and reward schemes.

By “delegation” and “autonomy” we do not mean the absence of controls.  Instead, we mean the absence of close or constant monitoring or supervision and the presence of devices, like incentive pay, that influence behavior.  Our emphasis is on the fact that the choices of managerial controls induce certain behaviors—both desired and undesired—and like Frank Sinatra sang in a slightly different context, “…you can’t have one without the other.”

Update: if you are interested in the following, we strongly encourage you to read our long post on defense department procurement, When Duplication of Effort Saves Money. Decentralized organizations will always duplicate some amount of effort; however, the fact that one cost is higher does not mean that all costs are higher nor does it say anything about benefits. One should consider the total, marginal (or relevant) benefits of decentralization versus the total marginal costs of it, and compare that difference with the net benefits of a centralized organizational structure.

Incentive Problems

Without close or constant monitoring (or their technological equivalents), whenever an organization permits autonomy, there is a potential for incentive problems—the possibility that subordinates will prefer outcomes (and therefore make decisions or take actions) that are inconsistent with organization’s goal.

These problems arise because of information asymmetries, which are often categorized as either (1) private information problems or (2) hidden effort problems. ¹ In the former, the private information could be about the internal or external environment or about the person himself, i.e., his type.  Often it is necessary to elicit the private information to solve coordination problems among multiple individuals or departments.  In the latter, the hidden effort usually involves some task that the organization chooses not to observe—usually because of the cost of monitoring—and this leads to motivation problems.

The organization’s losses due to these problems are know as agency costs because the subordinate has the authority to act as an agent on behalf of the organization.  Note that on financial statements agency costs are not categorized as such; instead, they appear as additional compensation expense or reduced output and revenue or higher production or investment costs.  Our focus in this essay in on the fact that agency costs are natural and predictable byproducts of delegation.

Clearly, no organization would permit someone to act on its behalf if the costs of such delegation were greater than the benefits.  So, when we observe subordinates with authority, we can reasonably infer that the benefits of delegation exceed the costs, including the agency costs.  ²

In certain environments, there are many good reasons to provide a subordinate with a high level of autonomy.  These include the cost savings of extensive supervision and monitoring; the opportunity for the organization to benefit from faster and more informed decisions (using the subordinate’s private information and unique expertise), and the improved job satisfaction for certain types of workers.

In most real settings of interest, subordinates have both private information (about themselves or the environment) and take effort that superiors cannot observe.  A reasonable manager—a thoughtful and empathetic one—has the foresight to predict the undesirable actions that the organization’s structure and policies may induce and therefore be able to estimate the costs associated with them. ³

Thus, if autonomy is optimal, then there is an economic level of dysfunctional behavior (with an associated level of agency costs).  This optimal level may vary as environmental factors vary.  First, if the organization’s strategy or structure changes as a reaction to changes in environmental factors, the level and cost of dysfunctional behavior may change.  Second, managerial or technical innovations may change dysfunctional behavior or the associated costs.

Regardless, with all things equal, the organization desires the minimum possible level of dysfunctional behavior and agency costs without losing the benefits of autonomy.  ⁴

A Slight Detour

In a former life as a professor, we lectured on decision making within firms, which we define as organizations with profit motives.  We would use mathematical problems as models of the firms, and discuss the general firm problem of profit maximization and the production problem (or specific action plan) where the firm’s goal was to minimize costs given a certain desired level of activity or production, e.g., quantity.  In fact, we would even force our advanced MBA students to use Mathcad to solve such problems.

We would ramble on about various assumptions and characteristics that were needed to solve the math problems and at the end the discussion would ask the following question:

Below are two statements.  Are both, either, or neither true?

• Profit maximization implies cost minimization.
• Cost minimization implies profit maximization.

While it may be unfair to pose these same questions to the reader without the six hours of lecture, the PowerPoint slides, the graphs, and the math, we ask anyway.  What does the reader think?  Yes, it is a rather vague setting, but please give it a quick try. ⁵

Profit maximization does imply cost minimization.  If the firm’s plan of action maximizes profit then it must also minimize costs (given that plan).  We will prove it by contradiction, which is a very typical method used in economics.  So, we will assume that the statement is false, and then show that it cannot be false.

Suppose the firm could find a cheaper way to act or implement the same plan.  In that case, the plan wasn’t profit-maximizing as we had supposed, and that is a contradiction.  So, unless the environment changes—say, prices, costs, or technologies—if one is maximizing profit, there is not a cheaper way of producing the planned output (or providing the planned service level).

The second statement is false.  Cost minimization does not imply profit maximization.  The fact that total costs are minimized via a certain plan of action says nothing about the benefits derived from that plan or about the net difference between the benefits and costs, i.e., the profit.  Thus, minimizing costs does not guarantee profit maximization.  In fact, in simple, stark economic models, with increasing concave revenue and convex cost functions, the global cost-minimizing plan of action is usually… inaction. To minimize costs, the firm should not operate and, therefore, not consume any resources.  Total cost is zero, and you can’t get lower than that.  Unfortunately, and of course, unless you are a farmer, revenue is zero, too.

The Baby and the Bath Water…
Or the unintended consequences of eliminating agency costs.

What’s our point?  Except in extreme cases, the profit-maximizing choice generates a non-zero, economic, level of costs that cannot be eliminated without eliminating the benefits, too.

Now, why are we especially long-winded about this issue?  Partly because it is our nature but mainly to emphasize the point that in decentralized organizations, many managers ignore this maxim and try to eliminate all agency costs when, in fact, agency costs are normal costs of doing business in a decentralized manner.  That is, these managers lack the discipline and willpower to implement truly decentralized controls.

The word “willpower” in the preceding sentence was purposefully chosen because permitting employees discretion and autonomy often requires the same level of personal discipline as maintaining a diet or forsaking other temptations.  In fact, due to our own personal weakness, we can’t avoid the temptation of another aside on Strategic Consistency and Managerial Discipline, but we will hide it over here.

So, what can a poor manager do?  Economics provides a little guidance.

The Revelation Principle

In economics the Revelation Principle states that in certain environments—under certain assumptions—one can learn the truth from others.  Per our side essay on managerial discipline, this usually requires some type of commitment by the superior to elicit the private information.  That means that while one can learn the truth, one cannot learn the truth for free. (Obviously, if one could learn the truth for free, there wouldn’t have been a problem in the first place.)

Unfortunately, in most real corporate settings, all the conditions needed to get truthful reporting are rarely present, and even if one could get it, it isn’t necessarily worthwhile.  So, the superior must trade-off the value of the information versus the costs of its acquisition, where the major cost is the informational rent that the subordinate demands (for knowing something the superior doesn’t). The cost may be a payment similar to a brib or it may involve the opportunity cost of allowing the worker to shirk or consume perquisites in exchange for the information.

A Couple of Non-business Examples

Prosecutorial Offers of Immunity – consider criminal cases—either in real life or on television—where a district attorney or other prosecutor provides an indicted accomplice with immunity in exchange for his truthful testimony against a co-conspirator.

The benefit is that the prosecutor increases the probability of convicting the partner-in-crime.

What are the costs?  First, a (presumably) guilty person goes free and is not penalized for his or her crimes.  Secondly, at the margin, morally-challenged individuals are more willing to join conspiracies knowing that if they are caught their penalties can be reduced by cooperating with authorities.  This is equivalent to shirking effort within a firm. (Of course, operators of criminal organizations know this, too, and attempt to develop their own control mechanisms that generally promise retribution for squealing, e.g., omertá or code of silence.)

Teenage Drinking – To reduce the probability that their teenager is in an alcohol-related automobile crash, some parents adopt the policy of no retribution for honest revelation.  In other words, the parents instruct their teen to call home if the child or his or her driver has been drinking or is intoxicated.  The parents commit not to punish the child despite the fact that the behavior is illegal.

On one hand, conditional upon the fact that the teen has been drinking, this policy may reduce the possibly severe, adverse consequences of driving while impaired or riding with an impaired driver.

On the other hand, committing to eliminate the punishment for such anti-social behavior also signals to the child the parent’s (at least) tacit approval.  Like the prosecutor who realizes that making deals increases the likelihood of crime, parents should expect an increased likelihood of their child consuming alcohol. Unintended consequences arise when the child is too impaired to think clearly enough to call or the impaired state leads to other behavior and negative consequences; take your pick with the ways that drunken teenagers can get into trouble.  ⁶

In our examples, the information seeker must be able to credibly commit to his or her policy.  If the district attorney reneges on his pledge after the fact and attempts to prosecute the informant, then he forsakes the use of the tool for the duration of his term.  In most locales, there aren’t that many defense attorneys, and that type of information travels quickly among them.

If a drunken child telephones his parents and is then punished, the parents should not expect the call the next time the teenager drinks.  (Of course, the child may be less likely to drink because of the expected punishment.)

These examples provide evidence that there are no simple solutions.  Such problems are rarely eliminated (without the often prohibitive cost of full centralization).  At best, they may be mitigated.  So, we are left with the basic question.

Does one seek information or wish to motivate?

The optimal choice of whether to elicit information or motivate effort (and to what degrees) depends upon the various costs and benefits of each alternative, including long-term implications, and these costs and benefits are determined by certain characteristics of the organization’s environment and strategy.

It is worth noting that it certain settings, a single piece of information—a single relevant fact—may be worth multiples of a subordinate’s annual salary, and the person may get a bonus for seemingly “doing nothing.”

So what’s the solution?  We recommend hiring a consultant and have a particular one in mind: someone who realizes that there is no single solution for all settings and that all decentralized solutions to the control problem will induce some economic level of agency costs with which one must learn to live.  Costs that mature, thoughtful, and disciplined manager must permit for the chosen solution to the control problem to remain optimal.

Epilogue

In the spirit of charity and goodwill towards the reader, we present characterizations of two types of problem managers in the decentralized (or partially decentralized) structures.  Based upon our observations, we will label them “The Idealist” and “The Extremist.” ⁷

These two types differ by intensity but also by temperament; the Idealist is the kinder, gentler version and generally has a sincere demand for the information and effort.  The Extremist is often a misanthrope but tries to disguise by considering himself a hard-charging, high-intensity manager.  There are energetic, decent folks trying to do the right thing.  We are not describing those good people.

The Idealist: Wouldn’t It Be Nice if…

We often observe superiors who wish they possessed their subordinates’ private information or could observe their subordinates’ hidden efforts.  Often such sentiments are not expressed as these positive desires; instead, they are communicated as complaints like “I know he is lying to me;” “He says he doesn’t know, but that’s not possible;” or “He is so lazy.”  (Is he lazy or is he behaving optimally for himself and the organization?  What could he know that you don’t?)

Usually, the idealist is attempting to manage a decentralized structure in a centralized way, and spends too little time thinking about the design and effects of controls and too much time focusing on their subordinates’ observed behaviors, which any reasonable person may admit might be annoying.

In extreme cases, we have seen idealists attempt to eliminate all dysfunctional behavior.  Such “management” is almost always self-defeating.⁸

Now the clever idealists may ask rhetorically, “What are you writing about?  There is no problem.  Near the beginning of the essay (and elsewhere on the web site), you, yourself, mention that the optimal level/degree of autonomy will change as environmental factors changes.  When we ferret out the information and direct workers regarding specific tasks, we are responding to changes in the environmental and (implicitly) changing the organizational structure.”

If such direct management results from a conscious decision to centralize, so be it.  However, attempting to eliminate all dysfunctional behavior is de facto monitoring, and it eliminates the benefits of delegation and autonomy.  Moreover, if the structure is decentralized and you—as a superior—are attempting to monitor rather the design efficient mechanisms and policies, then you may be the incentive problem—not your subordinates, e.g, that knee-jerk reaction to punish opportunistic behavior—though satisfying—need not be optimal, and that makes the Idealist the incentive problem.

Idealists need to be trained about the nature of their managerial responsibilities, which extend beyond supervision.

The Extremist: No pain, no evidence of management.

Before describing the Extremist we note that in dynamic environments, without incentive problems, the goal to operate at 100% capacity is almost never optimal.  The basic idea is that a little bit of slack allows one to take advantage of profitable (and possibly ephemeral) opportunities as they randomly arise.  This isn’t always true, but in the absence of additional facts, it is a reasonable working premise.

Now, Extremists are often unaware of this or willfully ignore it; subordinates will just have to worker harder or longer or both.  Hey, they should be happy that they have jobs.  They don’t deserve them, after all.  Extremists push for maximum effort, regardless of the costs (or benefits) to the subordinates or the organization.

Unfortunately, Extremists never identify themselves as such.  Worse, yet, is the fact that their superiors may not identify them either.  Instead Extremists tend to pool themselves—see adverse selection—with earnest folk and refer to themselves are hard-charging or cost-conscious or efficiency-minded, but their motivation is rarely positive.  They are often busybodies and prospective empire builders, and for them schadenfreude is a way of life.

We could go on, but there is no need.  The chance that such a person will manage effectively in a decentralized organization is about zero.  The chance of redeeming them through training is very close to zero, as well.  They are incentives problems and will create additional problems for their organizations (while trying to blame others for any failures).  Unfortunately, for organizations lacking strategic consistency the Extremists’ absence of managerial discipline may go unnoticed as they may produce short-term gains at the current expense of their subordinates and the future cost of the organization’s ability to function effectively and efficiently.

Copyright © 2008 Spero Consulting

---

Footnotes: