See also, my summary of the book.
When you measure any indicator of performance, you incur a risk of worsening that performance.
We believe this is a book that needs to be on the desk of just about anyone who manages anything.
Some books on measurement so strongly advocate its use that they look almost exclusively at success stories. They profess to tell you how to get it right, but they supply little or no detail about the consequences or likelihood of getting it wrong.
trust, honesty, and good intention are more efficient in many social contexts than verification, guile, and self-interest.
performance measurement “the most powerful inhibitor to quality and productivity in the Western world”
One troubling possibility is that one person’s functional system is another person’s dysfunctional system—that the difference is solely in the eye of the beholder.
General Electric began working to determine “key” performance measures in 1951 and that forty years later the measures generally used are not satisfactory.
dysfunction is violation of the spirit and not the letter of stated intentions
Dysfunction’s defining characteristic is that the actions leading to it fulfill the letter but not the spirit of stated intentions.
common measures of business performance, such as quarterly profits or stock value, often distort the incentives of U.S. business managers by emphasizing short-term over long-term objectives
Employees’ true output (such as value to the organization) is often intangible and difficult to measure; in its place, organizations choose to measure inputs (such as the amount of effort devoted to the task as measured by counts of interviews performed).
Informational measurement must be careful not to affect behavior, because the information conveyed by measures is likely to be most representative of actual events when people being measured behave as if the measurement system did not exist
The employment contract, for example, seems at odds with economists’ idealizations.
in a departure from the usual economic portrayal of exchange, this transaction is not clearly defined or prespecified. The employee agrees not to specific terms, but rather to act in a general way on the employer’s behalf,
The exchange between employee and employer becomes complicated as terms are ambiguous and verification of contract performance is difficult.
process refinement measurement, provides information that reveals the detailed structure of organizational processes. Detailed accounts of the internal workings of processes are useful in designing improved processes, thereby making the organization function more efficiently.
Coordination measurement provides information that allows short-term (sometimes real-time) management of organizational flows and schedules.
Even where performance measures are instituted purely for purposes of information, they are probably interpreted as definitions of the important aspects of that job or activity and hence have important implications for the motivation of behavior (p. 247).
Campbell’s (1979) law
The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor (p. 85).
The link between management practice and research in economics is tenuous at best.
the assumptions underlying many economic models are so simplified that it is difficult to take them seriously as descriptions of real organizational situations.
Managers often fail to consider the increasing costs of additional effort from employees, especially when those cost increases are implicit, such as increases due to lower quality when employees work faster.
Other costs that get left out of managers’ mental accounting will be important later in this book, especially the costs of creating and maintaining a system to verify agent effort expenditures.
Design of the incentive system amounts to solving the problem of discovering the optimal point, at which marginal revenue and marginal cost just offset each other.
The principal can only determine the agent’s effort level by looking at the level of revenues that result.
Increasing the level of agent effort increases the likelihood of higher revenues but it does not guarantee higher revenues.
Depending on the agent’s attitude toward risk, he may demand a considerable surcharge in exchange for taking on a job in which the potential revenue outcomes are partially beyond his control.
The optimal payment schedule is one in which payment to the agent increases as revenues increase. This result can be interpreted as a recommendation to design incentive systems in which rewards are increased as measured performance improves.
The conclusion should be disturbing because much of common practice seems based on drastic oversimplification.
If you believe in pay-for-performance, consider whether your underlying opinions about organizations and individual motivation contain oversimplified notions such as those that underlie the R-H model.
how smart employees work matters at least as much as how hard they work, especially in jobs that involve a lot of mental activity.
The problem facing the principal is how to arrange payment schedules to encourage appropriate magnitude and mix of effort by the agent.
in examples of dysfunction, the principal seems to set payment schedules that cause bad behavior by the agent. The principal is usually harmed by the agent’s bad behavior when customers become displeased.
as long as the agent is not strictly effort averse—that is, as long as he finds it pleasurable to expend some effort that yields value to the customer—the principal will prefer to pay a fixed fee that does not depend on measured performance.
Might agents be persuaded to enjoy doing more work that is beneficial to the principal, regardless of promises of monetary reward? Might internal motivation be called on to solve incentive problems?
Even output measures such as revenues are vulnerable to manipulation by firm managers who can change accounting methods.
value is provided in a non-additive manner as agents devote effort to tasks.
the potential for dysfunction arises when any critical dimension of effort expenditure is not measured.
measuring only easy-to-identify or easy-to-measure areas is a flawed practice.
many critical effort dimensions (for example, quality-related dimensions) have been observed to be especially difficult to measure (Ishikawa, 1985).
the customer acts as the final judge of the value of agent effort allocations.
Frequently, more effort can be extracted from agents only at the price of some incentive distortion. Whether the increase in value to the customer caused by the agent’s increased effort outweighs the loss of customer value due to incentive distortion is central to motivational measurement issues.
Distortion, then, is the property of an incentive system that causes the agent’s effort allocation to leave the best-mix path, thereby seeking a less than ideal effort mix.
The Crucial Test of a Motivational Measurement System: Does the value to the customer of the extra effort induced by a motivational measurement system more than outweigh the detrimental effects of the incentive distortion it causes?
Specific determination of the agent’s effort allocation under no incentive system allows comparison with allocations that can be realized under an incentive system.
organizations that rely solely on macro-level measures such as revenues are vulnerable to exploitation by individuals in the hierarchy who know that an individual performance will be lost in the crowd of individual performances of everyone in the hierarchy.
The main problem for most incentive systems in use in real organizations is not noise in measures of performance but, rather, bias intentionally introduced by those being measured.
It is often argued that software developers have different criteria for the value of their product than do their customers.
uncertainty about what the customer wants has the effect of reducing the utility the agent gets from satisfying the customer.
software developers frequently demand very specific instructions and are discomfited by uncertainty about what the customer wants.
the shape of customer preferences is assumed to be known by principal and agent. In real circumstances, it is unlikely that even the customer knows his preferences fully.
Customers do not mention needs that they assume will be met by all products of the type in question. And they do not mention product qualities that they have not thought of but would like. They mention only qualities that they do not expect but know that they want.
One way the principal can manage the agent is to leave him alone.
ask whether the effort allocation induced by an incentive system provides more value to the customer than the unsupervised allocation.
the situation in which everything that matters can be measured easily is very rare, if it ever occurs.
Some distortion is always induced by the incentive system under partial supervision;
The mode of supervision is not something that a profit-maximizing principal gets to choose. Rather, the mode of agent supervision is dictated by the structure of measurement costs in a given environment, which the principal has only limited ability to influence. In other words, a manager cannot simply choose to fully supervise employees and thereby rule out dysfunction. Some dimensions of effort will usually be too expensive to profitably measure.
When important dimensions are ignored, dysfunction can result from policies that encourage the agent to maximize on measured dimensions
Partial supervision will be effective when targets are set at a moderate level on measurable dimensions and agents are discouraged from exceeding targeted levels of performance.
Under partial supervision, dysfunction will result eventually if levels of effort are permitted to increase without appropriate boundaries.
you may succeed in achieving quality in the sense of meeting specification and still the customer may not be satisfied.
Quality is, by its nature, ephemeral; it is whatever the customer says it is, regardless of how vehemently some quality experts may cling to definitions like “conformance to specification.
In partial supervision, however, increases in effort beyond a certain point reduce value to the customer. In such a case the principal should set targets more moderately and perhaps even penalize overperformance in a particular activity.
Whatever specific thoughts inhabit the mind of an honest salesman, what is most important is that he has abandoned formal definitions of performance in favor of his own values and discretion, often to his immediate material detriment.
External motivation might be defined as a tendency to act in response to promises of rewards for performance according to mutually known criteria that are set down in advance by the promisor. In contrast, internal motivation might be defined as a tendency to act as compelled by private expectations of performance according to the action-taker’s own personal criteria. When targets are external, performance is objectively discernible. Concern with objective definitions of performance is, then, a trait of those who are externally motivated. Conversely, when motivation is internal, performance is a matter to be determined by the action-taker’s personal judgment. Concern with personal, subjective codes of behavior is a characteristic of those who are internally motivated.
At Semco, a Brazilian manufacturer of pumps, mixers, valves, and catering and other industrial equipment, most employees decide their own salaries. Their bonuses, which are tied to the company’s profits, are shared out as they choose. Everyone, including factory workers, sets his own working hours and groups of employees set their own productivity and sales targets. There are no controls over travel or business expenses. There are no manuals or written procedures. Workers choose their own boss and then publicly evaluate his performance. All employees have unlimited access to the company’s books and are trained to read balance sheets. Everyone knows what everyone else earns, and some workers earn more than their boss. Big corporate decisions, such as diversifications and acquisitions, are made by all employees … (“Diary of an Anarchist,” The Economist, June 26,1993, p. 66).
“It’s really very simple,” says [the firm owner]. “All we’re really doing is treating people like adults” (The Economist, loc. cit.).
The style of management that relies on internal motivation, and that contrasts sharply with measurement-based management, is here called delegatory management.
The agent is made to feel disutility for effort less acutely, inspired by enthusiasm for the mission of the organization, or by the persuasiveness of its leader. Or, the agent is convinced to value customer satisfaction more highly than he did previously, for similar reasons.
As weight is shifted away from disutility for effort toward utility for satisfying the customer, the unsupervised allocation moves outward from the origin in a way that improves value to the customer. Notice also that the unsupervised allocation always moves along the best-mix path (see Fig. 10.1). As long as the agent considers all effort expended to be equivalent (that is, as long as he does not, for example, dread interviewing more than referral activities—a possibility that was discussed in Chapter Six), and as long as agent utilities are aligned with customer value, his unsupervised allocation will be on the best-mix path.
delegatory management increases value to the customer without inducing any distortion at all.
Empirical work on human motivation (Frey, 1993; Deci and Ryan, 1985; McGraw, 1978) has shown that external motivators often crowd out internal motivation. This means that measurement-based management is in conflict with delegatory management. There is a negative interaction because of the implicit message of distrust that a measurement system conveys by the fact of its existence.
DeMarco said of performance measurement: It’s demeaning… . People are motivated by extrinsics and intrinsics and the intrinsics are much more powerful—the pride and workmanship, the enjoyment of doing things with your colleagues… . But the extrinsics shove the intrinsics aside. Say you give me an extrinsic reason to do this. I lose track of whether I’m having a good time working with my colleagues on this goal.
Alfie Kohn (1993) observes that measurement connotes comparability which, in turn, fosters competition and even unfriendly feeling between workers, thereby undermining any sense of common purpose.
measuring worker activity within one area has a detrimental effect on internal motivation in other areas.3
The negative interaction that results in “crowding out” is attributed to a shift in locus of control from agent to principal that reduces the agent’s sense of self-determination
the conflict between delegation and measurement means that the mere introduction of measurement-based controls moves the unsupervised allocation toward the point (0, 0), opposite of what is desired.
delegation compares favorably to partial supervision because delegation cannot cause dysfunction, or even distortion, and the movement along the best-mix path that it causes is more efficient in producing value for each unit of agent effort expended.
Situations that call for no supervision, although they are more common than is usually conceded,
Table 11.1. Recommended Management Methods.
marginal costs of delegation are lower and that resulting unsupervised allocations are of more value to customers when agents and principals are members of the same organization (because of organizational identification).
Measurement should be more attractive when principal and agent are members of different organizations than when both work for the same organization.
external incentives tend to be used more often between client and contractor than between employer and employee.
There are numerous organizational and task features that affect measurement costs. Some of these Note: See the list that follows this highlight. Edit observers of work being done by a group have difficulty distinguishing how much effort each member is contributing. Tasks in such settings are inherently difficult to observe, hence full supervision of them is unlikely to be attained.
There are also numerous organizational and task features that affect delegation costs. Note: See the list that follows this highlight. Edit recommendations for some specific scenarios, Note: See the list that follows this highlight. Edit a large and probably growing portion of important productive activity can be described as having high delegation and measurement costs.
In journalism or medicine or software development, it is not clear how to compare what is measured on any single production cycle with anything else.
converting the situation into one more suitable to delegation. This sort of conversion can also be effected in various ways, including Note: See the list that follows this highlight. Edit Steps taken to make employers and employees more trustful of each other can also lower delegation costs.
Managers can be trained in management styles that emphasize inspiring and trusting employees rather than analyzing and coercing them.
conditions that allow delegation costs to remain low are fragile;
Placing trust in employees may seem as if no one is minding the store if things go wrong.
Management-By-Objectives, or MBO (Drucker, 1954), as it is often practiced, seems especially misguided. MBO prescribes setting objectives, establishing quantitative criteria for measuring progress, and linking rewards with achievement as indicated by the measurements. The practice of MBO is obviously a form of measurement-based management, but it is mostly used in managerial and professional settings where tasks are not well-suited to measurement. For similar reasons, merit-based pay systems that depend on measurable criteria for success also seem badly conceived, despite their growing popularity.
Deming’s Fourteen Points
software development seems particularly poorly suited to measurement-based management.
where purely informational measurement is realized, it often reverts to having motivational features after a short period of time.
Much enthusiasm about measurement in organizations is probably based on a vision of this ideal state where true output is known, incentives are perfectly aligned, and measurement information accurately portrays reality.
agents become sensitive to a competitive dynamic that is not represented in models that feature one principal and one agent.
Ouchi (1981) notes that the employment practices of Japanese firms are designed in just this manner, to reduce competition between coworkers. Interestingly, many firms in the West are moving in the opposite direction, toward merit-based systems of compensation (Tully, 1993). Western firms’ fondness for merit-based systems is partly due to historical trends that are difficult to reverse.
much of the managerial diagnostic capability envisioned by many advocates of measurement is lost when a manager cannot see the measurements corresponding to different areas within her responsibility. What managers have instead is a self-assessment tool.
self-assessment is the eventual endpoint of a successful measurement program.
Although informational measurement enforced by strong safeguards is antithetical to motivational measurement, it is entirely consistent with delegatory modes of control.
Managers are transformed—some would say reduced—to communicating direction and providing help to workers.
The challenge for managers is to become more trusting, able to inspire and communicate, and willing to help rather than be helped.
economic tradition seeks to explain behavior as the joint outcome of concurrent optimization by all involved parties.
Scientists in behavioral traditions are troubled by the assumptions of the economic tradition, which imply that people always try to optimize, always know how to optimize, and make no mistakes in optimizing that are significant enough to impact enduringly on observed phenomena.
The mistake made by earnest measurement system designers can be simply stated in the terms of the model in previous chapters: A principal who commits dysfunctional acts mistakenly believes she is in a fully supervised situation when she is, in fact, in a partially supervised situation.
the difference between full and partial supervision is crucial in that prescriptions for successful principal behavior in the two cases are contradictory.
There are two ways a principal might be able to distinguish between full and partial supervision:
the principal should eventually be able to tell, based on outcome feedback, which mode of supervision is being realized.
the need for control is often interpreted narrowly as a need for measurement-based control.
The inclination to interpret control narrowly is due to what might be called a standardization reflex.
Redesigns for measurement tend to fail when the setting and product are not particularly suited to measurement. Note: Such as in Scrum, agile estimation for software development. Edit “escalation situations” in which decision-makers remain committed to a course of action that has been shown to be unproductive.
commitment to failed courses of action can arise from unwillingness to admit failure, even when the task that was taken on was impossible to complete satisfactorily.
the principal is guilty not of refusing to face reality but rather of super-zealousness or unclear thinking about measurement.
Output measurements are dangerously declared to be “actual performance results”; they are, of course, unlikely to be “actual” for an activity that is as complex and difficult to observe as software development is. Note: E.g., effort estimates and subsequent “actuals” recorded. Edit The attractiveness of mechanistic analogies derives in part from a human weakness for what Abraham Kaplan (1964) calls “the mystique of quantity,” which he defines as “an exaggerated regard for the significance of measurement just because it is quantitative, without regard to what has been measured or to what can subsequently be done with the measure” (p. 172).
what is quantified often seems more attractive to some people than what is not quantified, even when what is quantified does not convey any more information.
mythical numbers, by which he means the tendency of a number, once produced, to take on a life of its own, to endure despite well-founded complaints about the method used to derive it.
there are dire consequences of not accomplishing comprehensive measurements. The analogists tend to regard measurements as unbiased signals. The flaw in this conception, which is shared by agency theorists, has been discussed in previous chapters. The lack of understanding that leads to the mistake may derive from a well-documented human tendency to misunderstand conditional probability. A principal who believes with sound basis that favorable outcomes imply measurements in favorable ranges, may reason that therefore the reverse is true, that measurements in favorable ranges imply favorable outcomes. But the conclusion does not follow.
designers of software production systems are forever redesigning, replacing old modes of control, and substituting new but structurally similar modes, with predictable lack of success.
in real situations, the principal has perfect knowledge of very little. To gain knowledge she must experiment.
In software development, for example, evidence of a failed control system may only come to light after release of the software to customers, which could be years after the control system is introduced. When outcomes are revealed so slowly, learning is difficult.
A principal who learns experimentally will not gather data needed to compare delegatory and measurement-based alternatives, if she is not inclined to try the former.
Designers of measurement systems sometimes seem to continue in measurement-based control efforts despite knowing that their efforts are unlikely to work.
in a hierarchical organization every manager is an agent as well as a principal.
Even CEOs are agents, strictly speaking.
while a manager within a hierarchy might be responsible for design of a system of measurement and serve as principal to a set of agents, her own performance is judged mostly by how well her organization—that is, her agents—does according to the very measurement system the principal installs. The principal has an interest, then, in installing easily exploitable measurement systems.
The lone agent who does what is good for the customer by refusing to meet dysfunctional targets not only loses any target-meeting bonuses but is also at risk for absorbing blame for the unfavorable outcomes caused, ironically, by the actions of everyone but him.
It is interesting to contemplate why there is no real principal in such an organization. The nesting of principals and agents in a hierarchy clearly provides incentives for collusion to produce measurements that can be subverted. What is puzzling is why the true principal, the owner of the organization’s stream of production, does not insist on better measurement systems, or even no measurement system if no system is appropriate to the activities of the organization. Stockholders in a corporation surely want management to be effective—why do they not force this? One reason is that owners often have few tools at their disposal to influence the design or to force the abandonment of measurement systems. Their power to make changes is spread thinly and constrained by legal precedent that places primary responsibility for running a company with professional managers. Moreover, because they are not close to the activity of the organization, absentee owners are easily misled by the kinds of flawed analogies described under the “earnest explanation” heading in Chapter Fourteen. A situation that is probably common finds the owners of a firm, who are collectively the true principal, making earnest dysfunctional mistakes as their own hired managers actively lobby to reinforce mistaken impressions about the effectiveness of measurement.
1) What is happening when organizations succeed at organizational control? and 2) Shouldn’t organizations that are more successful at organizational control crowd out other, less successful organizations, in keeping with survival of the fittest?
internal motivation deteriorates in the presence of external incentives.
the very act of installing a measurement system makes it more likely that managers will behave with guile.
the Japanese president of the American subsidiary of a Japanese bank refuses to give his American managers the performance targets for which they are begging. The Japanese bank president’s reasons are revealing: If only I could get these Americans to understand our philosophy of banking. To understand what the business means to us—how we feel we should deal with our customers and our employees. What our relationship should be to the local communities we serve. How we should deal with our competitors, and what our role should be in the world at large. If they could get that under their skin, then they could figure out for themselves what an appropriate objective would be for any situation, no matter how unusual or new, and I would never have to tell them, never have to give them a target (pp. 40-41).
The anecdote provides insight into Deming’s claim (Gabor, 1989) that performance measurement is one of Western management’s most misguided policies.
In the non-attributional culture, systems of measurement do not motivate employees who fear for their career prospects and future rewards. Attribution and blame is unthinkable within the culture. Motivation is derived from wanting the group to succeed. Measurements are a means of obtaining information and their limitations are widely acknowledged. Measurements are not the central obsession of managers, nor are they the main tool for managing. They are, in fact, beside the point somewhat, secondary to discovery of information through walking around and talking to people doing the work about what the real problems are. In the non-attributional culture, people are inclined—as is urged in a Japanese proverb—to fix the problem, not the blame. This organizational climate adjusts the preferences of workers so that aversion to effort is minimized and desire to help the group by doing well for the customer is maximized.
Interviews conducted with renowned experts reveal what is considered to be state-of-the-art thinking about the use of measurement in software development. Computer software development is an intriguing case for two reasons. First, interest in measurement is high among software practitioners, so the issues raised here are relevant to practice. Second, the model developed here suggests that software development is usually poorly suited to measurement-based control.
All firms that trade within the European Common Market are legally required to be ISO 9000 certified;
The purpose of the ISO 9000 series of standards is to provide a means by which customers can be assured that suppliers are using quality business practices without having to perform individual audits of each prospective supplier.
A Software Capability Evaluation is an assessment of an organization’s process for developing software as it relates to a particular parcel of work for which the U.S. government plans to contract (SEI, 1991). It is a tool intended to help a government agency determine the organization’s ability to produce a particular product on time, within budget, and with high quality.
The CMM is a distillation of what SEI scientists, based on their research and experience, believe to be an ideal form of software development.
There is limited empirical data that shows a correlation between evaluated maturity levels and quality in software product or effectiveness of development process as determined by some external measure.
80-90 percent of the companies going through the [ISO] certification are just wasting their money” if their intention is truly to improve quality (Fouhy et al., 1992, p. 45).
The net benefit of measurement systems is not self-evident.
The existence of this difference between end and interim customers is apparent upon noting that customers do not often seek to impose production monitoring on the firms from whom they intend to buy.
Whenever the end customer is also coordinating the production process, he or she begins to experience the problems of the principal in evaluating value. For example, someone coordinating the building of his own house would not be a pure end customer in the sense of the model, because he suffers from the same attributional difficulties that a principal does.
To the extent that a system of laws provides imprecise incentives (think of the tax code), it even encourages dysfunctional (that is, uncivil) behavior.
From a purely external motivation standpoint, given that the incentives created by laws are almost certainly imperfect, civility in social settings (if one believes it is really there) seems to arise not because of laws but in spite of them. Viewed more broadly, however, taking internal motivation into account, laws must surely play a role in codifying the criteria that establishes identification with a social group. Laws, then, may function less as interlocking threats and more as prescriptions of social ritual. They express not so much the terms of a social contract as they describe the steps in a social dance in which people tend to want to participate. What happens, then, to a society and its laws when people stop wanting to dance the same dance or belong to the same groups? The obvious answer—that the society becomes uncivil, even violent, and the laws, which never worked as deterrents anyway, ineffective—is chilling because it seems confirmed by the daily newspapers.
what might happen if the high price of observing hard-to-observe dimensions were occasionally paid.
even the small probability that measurement will occur creates an expectation in the minds of agents that might cause them to allocate effort to that high-cost dimension. External incentive effect might be realized even on dimensions where cost of measurement is very high.
probabilistic measurement might also be seen as a way of rescuing motivational measurement from the disrepute bestowed on it by the model. If agents were inclined to behave as if full supervision were realized out of fear of a low probability check on high measurement cost dimensions, then systems of measurement would work much more as they do in airplane cockpit analogies, and the like. The simplistic notions of measurement-based control might seem vindicated. The problem with this position, though, is that most systems of measurement explicitly do not threaten investigation on difficult-to-measure dimensions. In some cases, there are explicit or implicit guarantees that the agent will be measured only in accordance with the agreed-upon criteria. Such guarantees may be legally enforced (such as in wrongful dismissal proceedings). The whole idea behind a measurement-based merit pay system, for example, is that the agent is told exactly what constitutes performance. Such systems seek to drive out the sort of ambiguity that might serve as a deterrent against opportunistic behavior in the minds of agents.
The cynical explanation of dysfunction suggests that consultants, who are not members of the group and are therefore less motivated internally to achieve group objectives, benefit from installation of measurement systems that are conceptually attractive but actually dysfunctional. Conceptual attractiveness makes a consultant’s suggestions sellable—it gets him in the door. Dysfunction permits consultants to control measures of their own performance, thus preserving their credibility and their flow of income. This is a much stronger statement than the common observation that organizations and their consultants have incentives that are not exactly aligned. The cynical explanation suggests that wares peddled by consultants are more likely to be both conceptually attractive and dysfunctional than similar wares that arise from other sources (for example, from workers within the firm).
It remains at least mildly embarrassing for the science of economics that its main theories provide so little justification for the existence of one of the most common features of the economic landscape—the large, multi-division firm.
Why does simply organizing into a firm provide advantages that could not be realized by constructing similar arrangements between firms?
Arguments that seek to explain firm integration inevitably point to features of organization that supposedly lower costs in some way. But it is virtually always possible to argue that these features could be replicated in an inter-firm arrangement.
Williamson argues that within firms agents are less able and less apt to take advantage of colleagues. The reason for lesser ability is not clear because explanations of it suffer from the discussed broad weakness. But the assertion that agents within firms are less apt to seek selfish advantage is more interesting. Here is where the model and explanations in this book can come into play by offering a suggestion: The advantage that organizations have over markets may be due to their increased ability to delegate and their ability to eliminate distortion by convincing agents to forego opportunities for self-interested guile.
This advantage of organizing is different in nature from the advantages proposed by economic theories because it flows from sociological and psychological characteristics of the situation, not tangible structural features.
There is one certain difference between being within a firm and being outside it that can never be replicated by a long-term contract—the difference of belonging to the organizational group.
The remarkable findings of Dawes et al. (1988) about the power of group identity to induce cooperation and the unlikely factors that lead to group identity are a potential cornerstone for a new theory of firm integration.
The organizational landscape is littered with the twisted wrecks of measurement systems designed by people who thought measurement was simple.
organizational leaders will have to work twice as hard as they might like to establish a culture conducive to measurement, in which measurement is seen as a useful way to learn but not as the be-all and end-all of performance management.
A good test of whether you are succeeding in creating the right kind of culture is to ask yourself what seems to be driving the people around you to do a good job. Is the motivation of workers primarily internal or external?
what is best for the organization almost never is exactly the same as what is best for the worker’s measurement performance.
you can succeed in producing a culture conducive to measurement. There are organizations in which people seem to have given themselves completely to the pursuit of organizational goals, at least temporarily, organizations in which members hunger for measurement as a tool that helps get the job done. In these settings, there is nothing special about measurement; measurement seems neither remarkable nor threatening. To use measurement inappropriately would betray a sacred trust, and no one would consider such a betrayal.