Long Review: Economic Hierarchies

post by SebastianG (JohnBuridan) · 2022-04-23T12:43:05.258Z · LW · GW · 4 comments

Contents

  Central Planning Games as Gateway Drug 
  The Nature of the Firm
  Information Flow
  Span of Control
    3a. The Big Span of Control Model
  Governance and Incentives and Committees
    Committees
  Back to the 900 Prosperan Engineers
  Puzzles Remain, while Good Governance remains a Scarce Resource
None
4 comments

Naked and Afraid from the Discovery Channel didn’t live up to its potential. To be fair, a handful of scraggly naked people trying to make it in Kenya’s wilderness made for interesting television, as they scratched themselves, got infections, and looked pretty uncomfortable. But the interpersonal drama seemed contrived since their goal was mere survival, and the division of labor was not highly interesting. I didn’t learn what I wanted--too much complacent nakedness, not enough competence-porn.

The show I'd like to pitch is one about progress and knowledge. 900 scraggly people, they don’t have to be naked, but for the sake of argument, let’s say they are naked, are plopped in the wilderness with a bunch of raw materials and the mandate to build the highest level of civilization possible in three years, outcompeting another group of 900 scraggly naked people. Boom! Instant natural experiment: knowledge, society, organization, bottlenecks on development. From it we could dream up better models of how to bounce back from a civilizational setback, settle charter cities, and craft efficient institutional structures. We could recruit some of the best minds in hundreds of fields not to consult but to build publicly, with all data and streams tracked and uploaded to the internet for analysis. Of course, our prediction markets about the show would be filled with bets about what milestones would and would not be reached. Plus, we would be entertained and given insights at the same time.

That’s my pitch for how we will inspire people about progress, productivity, and the mysteries of society’s organization. I want the world to see bureaucracy and technology stripped down to their barest essentials, not contrived nakedness in the wilderness. The details of the show might help us understand what governance and incentives would make for the fastest civilization building.

A sense of wonder about these things and a desire to cultivate comparative advantage drove me to read Economic Hierarchies by Gordon Tullock, published in 1992.

Central Planning Games as Gateway Drug 

When I was a kid, I wondered what the bare minimum number of people required to make a moon landing is. Firstly, for the rocket program itself, but then for the entire supply chain that makes a moon landing possible. How many janitors, cooks, and truckers are required? How many computer scientists, farmers, and sheet metal cutters? Is this number lower or higher than it was in 1968? How much of this could be automated? The question grabbed me for the same reason I think Rollercoaster Tycoon, Minecraft, Civilization 5, and Mini-Motorways grab people: central planning is fun for the central planner! An ecosystem in a bottle, a town in a snow globe, one of those hobbyist train tables which the old guys built, Warhammer 40k, these things all somewhat scratch the itch. That same itch is scratched by Wikiciv, the wiki for building things from scratch, for it is an attempt to store the deep, practical how-to guides of civilization. Wikiciv is on board with this TV show idea, or they would be if I had sent it to them.

As an adult, building a business or organization also scratches that itch, though it is far more unwieldy, and the central planner fantasy proves to be just that. Do we know how many people with what skills are needed to do certain civilizational feats? Rockets capture the imagination, but if we want to take our organizations to the moon, we are going to need something more than rocket fuel; we will need operational excellence, the thing we are all waiting for -- a fully fleshed out theory of bureaucracy!

Gordon Tullock’s Economic Hierarchies is a hunt for that Red October of theories. As an opening salvo, it provides just enough to titillate the future scraggly nudist who will be working in front of cameras while building the Prospera Space Agency from scratch. Yes, I know space flight is not feasible for 900 people, (SpaceX alone has 9,500 employees; but because rockets are cool, 900 naked aerospace engineers will serve as my prime motivating example while we dissect Economic Hierarchies. So, let’s imagine, that like a new age Gregor MacGregor, we are sending our 900 to Prospera to build the Prospera Space Agency from scratch.

The Nature of the Firm

Tullock starts with Robert Coase’s essay “The Nature of the Firm”.

To what extent does our Space Agency need to be a single agency at all? Why not just a series of individual transactions which ultimately leads to the rocket ship and crew? The government/TV show producers want to buy a moon landing from a group of 900 scraggly naked people. Everyone in their barefoot society rushes to fashion raw materials into intermediate goods which will allow them to produce even more complex intermediate goods, and so on, up the supply chain until eventually the government can purchase the complete package of rocket, lander, and crew, etc.

Obviously, our mosquito-bitten space engineers will not spontaneously get to work on individual projects for a few reasons. Even if there were a means of exchange or price system rewarding individual tasks, which in this scenario there is not, we still have the traditional economic problems.

There are transaction costs in figuring out who is doing which things and how to coordinate actions with others. If Geffen, the satellites expert, starts producing copper wire without any outside deliberation, he is risking inefficiency, perhaps taking up a task which someone else has absolute and/or comparative advantage in completing. Therefore, it is more beneficial for the skilled worker to communicate and coordinate with people before deciding on a course of action.

So, they will create organizations for metal working, plastic molding, food production, manufacturing, and so on. Organizations are necessary, because of high transaction costs. An organization internalizes and reduces those costs. That’s Coase.

Tullock’s guiding question for the entire work is: what determines the size, scope, and structure of these organizations? Are there any broadly applicable principles by which a firm determines whether something will be done in-house or contracted out? To put it scholastically, how many choirs of bureaucrats does the entrepreneur require? Or more to the point, what determines institutional structure?

Tullock posits two possible views for why organizations set up their span of control, governance, information flow, and internal incentives the way they do.

Option 1) It doesn’t matter. They just have to pick something not obviously counterproductive. Within each industry and potential firm, there is an immense amount of freedom for how to organize the hierarchy. The varying structures of retailers, wholesalers, franchises, reverse franchises, consultants, in-house work, bonuses, salary increases, and penalties simply do not follow from any generalizable principles. Sorry, science!

Option 2) It does matter based upon some key details of the firm’s inputs, outputs, and human resources available.

Do we reject option 1? Tullock rejects it without much of a wince. Some organizations are far, far, far less competent than others and able to successfully organize are around the goals of the firm. When an organization is conquered or bought out, or merged, sometimes its structure is greatly improved. Changes in organizational structure change outputs, sometimes dramatically so! The fact suggests that organization structure matters and can be figured out. Yet at the same time, there is still an immense diversity of firm types. Diversity of organization types within competitive markets should indicate either niche strategies, or key variables which determine ideal type.

Studied confusion about the various determinants of organizational structure grabs Tullock for almost 300 pages of text.

[R]etailing is sometimes carried out by individual stores, purchasing through wholesalers, and so on, sometimes by big, centralized organizations like Safeway, sometimes by a sort of combination, which we call franchising, sometimes by the kind of thing that is represented by Ace Hardware, which I have called reverse franchising. Nor is this a complete list of all the different ways these organizations can be set up.

 

The existence of this vast number of different organizations presents a puzzle. The simple solution is that organizational form does not matter very much. If this is true, there are a considerable number of professors in business and public administration who should be encouraged to seek more constructive activity. On the assumption that it is not true, that that there is an alternative explanation, which is that certain details of the various problems met by these organizations dictate one particular structural form as most efficient. That seems to me the most likely explanation, but as the reader has discovered, I have been unable to put my finger on those detailed reasons.

Moreover, we all know about free-riders, defection, Malthus, and Moloch. Clearly, some organizations are filled with an especially disturbing quantity of rent-seeking employees, office politics, immiserating incentives, and inadequate equilibria. (Tullock himself invented the term “rent-seeking” in his 1967 book Politics of Bureaucracy, which is in the same volume as Economic Hierarchies, for firms which use their power to decrease competition rather than improve themselves. In Economic Hierarchies he extends the idea to employees within a firm, who use their position to make their lives more comfortable and convenient rather than maximizing the firm’s good or their superiors’ wishes.)

Of course, there is separate and wide literature for dealing with these other issues of incentives and efficiency. Economic Hierarchies, though, seeks out the relationship between incentive constraints and structure. 

After our survey of hierarchical forms, we can say a great deal about the way to run them most efficiently. As stated in the last chapter, the use of incentives and well-functioning incentives is important. The basic structure, however, remains a puzzle.

What we have then is a society in which a large number of different organizations, some of them small, some of them simply individual people are interacting. A goal should be a structure in which the individual organizations are motivated to behave efficiently and the interaction is efficient. Over the history of the human race, these objectives have been a preoccupation of truly great minds. It cannot be said that they have been finally solved.

Tullock stuffs his book full of abstract and concrete examples. The abstractions and toy models illustrate his basic conjectures and his concrete anecdotes leaven the text with humor and additional insight. Without these examples, the text would be dry, dense, infuriating, and cynicism-inducing. 

Information Flow

Let’s start with information flow. Information seems like a key determinant in how we will organize our society.

How will the entrepreneur learn if something is not going right? How will the departments coordinate with each other? Who is really in charge of this meeting?

From the perspective of the mid-level bureaucrat there is a cloud of people in the organization -some higher, some lower. In the figure below our mid-level bureaucrat is the x; everyone in the oval has some immediate effect on his actions. The dots are constantly moving in and out of the oval. The purpose Tullock has in mind is to motivate a meditation on the difference between the formal, official and informal, organic structures of the organization and attempt to reconcile them. He complements the diagram with a motivating excerpt from War and Peace in which the character Boris Dubretskoy discovers the difference between the formal and informal.

 

At the moment when Boris entered, Prince Andrey, drooping his eyelids disdainfully (with that peculiar air of courteous weariness which so distinctly says, “If it were not my duty, I would not stay talking to you for a minute”), was listening to an old Russian general with many decorations, who, rigidly erect, almost on tiptoe, was laying some matter before Prince Andrey with the obsequious expression of a common soldier on his purple face.

“Very good, be so kind as to wait a moment,” he said to the general in Russian, with that French accent with which he always spoke when he meant to speak disdainfully, and noticing Boris, Prince Andrew to no further notice of the general (who ran after him with entreaties, begging him to hear something more), but nodded to Boris with a bright smile, as he turned towards him. At that moment Boris saw distinctly what he had had an inkling of before, that is, that quite apart from that subordination and discipline, which is written down in the drill-book, and recognized in the regiment and known to him, there was in the army another and more actual subordination, that which made this rigid, purple-faced general wait respectfully while Prince Andrey—of captain’s rank—found it more in accordance with his pleasure to talk to Lieutenant Boris Drubetskoy. Boris felt more than ever determined to follow in future the guidance not of the written code laid down in the regulations, but of this unwritten code. He felt now that simply because he had been recommended to Prince Andrey, he had become at one step superior to the general, who in other circumstances, at the front, could annihilate a mere lieutenant in the guards like him.

There is nothing like the moment in life when we realize that roles and titles are not indicative of power, influence, intelligence or capacity. Oftentimes, I find forcing formal structures upon a small organization too early does violence to the organic natural hierarchy, which develops through the nature of relationships. For those who need the clear signals offered by titles, leaving things informal is a painful and confusing situation. But my hunch is that it is efficient, and carries little risk when an organization is young.

This idea of the cloud of people makes sense for all organizations. If we think of it first as a cloud and secondly as a formal hierarchy, we probably are hitting closer to the truth of things.

An individual within the bureaucracy faces a cloud of other people, some of whom are clearly more important and can affect his career. Others are clearly inferiors; a large number cannot be classified either way. If this individual wants to get ahead, he must either please the superiors or form attachments to some of them, so it does not matter if he displeases others. Most large hierarchies have a sort of quasi-feudal structure as much higher officials have a number of followers who do their bidding and not that of their organization. One particular case is the “ugly genius.” This is a brilliant man who has no real prospect of getting ahead on his own because of his personality. A shrewd, smooth manipulator would regard such a person as a treasure, since his only real role in life is following someone else. The genius will provide bright ideas without having the savoir-faire to double cross his superior.

I know the type: an official in Health and Human Services, who is known for his object-level competence and “ugly genius” qualities; fortunately, he and his coterie have been promoted to high enough ranks that they can be a more competent hook in the side of the other slow incompetents who make their careers inside a leviathan.

One difference between corporate structure and government structure, I notice, is that the government’s official rules try to make absolute rank more vital, by having more depend upon it, especially since seniority is easier to monitor than competence. Nonetheless, it still seems that there are ways around this through various “lateral” moves that open up higher ceilings. If anyone knows more about this, I am happy to hear it. Tullock implies that information ultimately is the currency for those inside the bureaucracy who are not in immediate relationship with one another.

I doubt that absolute rank is vital. Here you may know people who are either much higher or much lower than you in terms of absolute rank; they are so removed from your immediate structure that you need not to give them your attention. They are in essence, equals even though they do not appear to be. In some cases, such a relationship may be useful to both parties. Exchange of information may be jointly beneficial, even if the information is always somewhat indirect and only of general interest to the recipient.

I believe though that the general interest provides a sense about the competence and efficiency of other department managers, their goals and foibles, their talented inferiors, and stalled projects. All these at some point in the future open up the possibility of advantage. Meetings of this sort are in the explore category. Poke your head out and see what’s out there; something might prove very useful.

The individual should regard his superior as only partially informed about his behavior. But as knowledgeable about aspects of it that are unpredictable. Controlling what information is passed on to the individual can influence his behavior positively. Although the degree of influence also depends on how well the superior assimilates these additional channels of information.

In other words, the superior-individual relationship has a lot of fog-of-war, and so, perhaps, the lack of perfect knowledge about what each other knows keeps tighter coordination than otherwise would be the case? This is a fascinating conjecture and similar to the idea of using random audits and selective check-ins to keep an organization working without much direct oversight. In a situation of imperfect oversight, one would want to consistently report good things to the superior.

Let us apply this idea to our middle level Foreign Service officer who faces various decisions regarding his activities in an embassy. He wants his junior officials to assist him, whether he is trying to carry out the orders of his superior or trying to implement what he thinks is the ethically superior course of action, (even if he is embezzling funds). He may also want to organize an apparatus of propaganda that communicates to his superior systematically biased positive data about his own activities.

Of course, if one is successful, the individual can then engage in more rent-seeking activities.

General Taylor, the commander of the Eighth Army during my last few days in the embassy in Korea, was barely interested in what went on in his army and immensely interested in what went on in the Pentagon and in the executive office of the president. His efforts relating to his command were geared mostly towards seeing that nothing happened to get him bad reports at those higher levels.

As a boss, the formal hierarchy is usually not the sole way to obtain information for precisely these reasons. Relationships with others within the organization can reveal a great deal.

People who saw that wonderful movie 9 to 5 recall that there was one unsympathetic character who was listed as a “spy for the boss.” She passes along to higher officials things that the lower officials don't want them to get. One understands perfectly why other people didn't like her. But from the standpoint of the efficiency of our total economy, or that corporation only, she was performing a social function, and when the other people succeeded in sending her off to Paris to prevent her from reporting a fairly major sin to their superiors, they were lowering efficiency.

Anybody who's been in an organization knows about efforts to prevent information which would be of use to higher officials from getting to them. For example, I was in a rather small organization, in which one of the secretaries characteristically arrived late. It did not affect her getting the work done; she was indeed very efficient. Nevertheless, the office supervisor reported her to the head of the organization. The other secretaries regarded this as a crime almost equivalent to rape.

Like a medieval king teaming up with the rich men of the towns to force the lords into submission on some issue, the hierarchical pincer move is not merely a tactic, but a method by which all actors can try and improve compliance. 

“You went over my helmet!?” 

“No, sir, more like around.”

Span of Control

The move from information to compliance brings us to the major issue Economic Hierarchies addresses, span of control, i.e. the number of subordinates an individual is responsible for. The coordination of subordinates drives us right into that most feared of demons, bureaucracy.

Tullock offers a sneaky definition of a bureaucrat: someone whose activity in the organization is not judged by market forces. So bureaucrats, instead of doing the thing which is rewarded by revenue, ensures that the thing is done which is rewarded by revenue. Within all sufficiently large organizations, there is a bureaucracy, somewhat hierarchical, which is supposed to ensure that something gets done to minimum standards. What makes this definition so sneaky is how much activity falls within the name ‘bureaucracy.’

We cannot take a broad view of market forces here. Tullock means evaluation of job performance will not be primarily in dollars. Rather, evaluation of job performance will rely upon attempts at objectivity other than dollars.

[William Niskanen] told me “A bureaucrat from the government moving into Ford Motor Company is all set to hit the ground running.” In other words, the structures are much the same.

Anthony Downs produced a sensible terminology, he suggested that we use the word bureaucracy, for an organization whose output is not evaluated in the market, the word bureaucrat would be applied to any individual whose output is not evaluated in the market. This pair of definitions means that it is possible to be a bureaucrat in a non-bureaucratic organization, and to be a non-bureaucrat in a bureaucratic organization.

Sometimes entire firms are purely bureaucratic. Accreditation agencies, for example, are purely bureaucratic. Using Tullock’s definition, the skewer of bureaucracy will run through any complicated workflow. Since each person only contributes a small part to the final product, much of their time will be spent in sharing information with each other, and only some time in working on the product itself. Bureaucracy for the Prospera Space Agency will not only be necessary, but likely ubiquitous.

Tullock does not say that bureaucratic agents exist to ensure the work gets done, but rather to ensure that what pleases the superior gets done. This difference makes dark, depressing sense. The output of work is evaluated by market forces, but the bureaucrat is evaluated by other means, since his influence on the work is hard to evaluate.

I remember one man who went in about the same time I did into the Department of State and rose very much faster. He had just been appointed counselor minister of the embassy, a small one, at the time that I was contemplating resignation. He kept emphasizing the necessity of improving his bridge game because he had heard that the ambassador liked bridge. Efforts to get him to comment on the political situation, a fascinating one, in his assigned country were completely frustrated. He simply was not interested.

In his review of Seeing Like A State Scott asked why do these people keep getting promoted even though their harebrained bulldozer plans and neat rectangular grids maim and destroy society? I think the answer lies somewhere in this. Le Corbusier can propose the same crazy plan for Moscow and for Paris, and then spend all his time schmoozing and endearing himself in high society.

The superior must create a structure to make sure his will is carried out towards the desired goal(s). This structure is called the span of control. When we assume that with each additional intermediary there is a percentage loss in the execution of the desires of the superior, then the superior has to make choices about how to organize the bureaucracy so that his will can be done. There are endless variations on this in the book, laborious yet insightful.

3a. The Big Span of Control Model

Tullock sharpens his machete on the twin issues of span of control and rent-seeking, then sets off into the Honduran jungle looking for recurring patterns. Here I go through his most magnificent toy model, which he found in the ancient Poyais Tomb of some magnicifent Cacique buried with 39 servants.

Suppose we have a society of 40 people, one of whom is a fairly dominant type, with a good chance of becoming the dictator of an organization. Each of these 40 people in the average day performs 10 acts; these citizens then affect the outside world by 400 total acts each day.

One of them is a potential dictator, and under our present circumstances, 10 out of the 400 acts are those he favors [since he is the one doing them], the other 390 nearly are randomly related to his desires. Suppose that he now becomes a complete authoritarian dictator, using previous three-for-one structure: i.e. he will have three immediate inferiors, each of whom supervises three, and each of them also supervises three. With this structure there are now 27 people at the bottom having acts within the outside world. The other 13 are our dictator and his supervisory personnel. The net effect is that there are 270 acts relating to the outside world, whereas before there were 400.

The point is that these 270 actions are more closely aligned to the dictator's will. But how closely? Tullock continues:

Assume that at each stage there's a 25% deterioration. Therefore, of the acts performed by the top-level inferiors who are simply supervising 75% will be in accordance with the real wishes of the dictator and 25% not, and so on down the line. In this situation, the 27 people at the bottom will end up with 43% of their acts carrying out the wishes of the dictator, or a total of 113 acts.

This is far better than the dictator's 10 original acts. But I am confused; if 390 acts were originally randomly related to his desires, we probably want some assumption about how many of the random 390 he would have willed anyway. Tullock presumes that uncoordinated actions neither cancel each other out nor interfere with the dictator’s plans; but rather such actions are just really long lunch breaks, or something innocuous like that, rather than, say, email correspondence with journalists.

On the other hand, 156 of their acts will not be in accordance with the dictator's preferences, but will be randomly distributed. Certainly, from the dictator standpoint, having roughly 114 acts in accordance with its desires and effect on the outside world is better than 10. He has gained; in fact, this is the largest number of acts affecting the outside world that he can have.

Is this unrealistic? If the majority of acts affecting the outside world cannot be in accordance with the dictator’s wishes, shouldn’t we expect to see corporations behaving in strange and contradictory ways all the time? What’s keeping them coordinated? It is practically a truism that most companies and organizations appear fine on the face, but behind the curtain are always on the verge of collapse. Software stacks are like this too; so, it seems, is everything.

But let us suppose that there are some things he wishes to control in which his interests are more serious or more important than others. Assume that he establishes a smaller organization for the purpose of dealing with fewer things –specifically, one organization in which he has three supervisors under his control. And they each control three people who actually affect the outside world. In other words, we drop off the bottom rank. In this case, there will be 270 acts that are not controlled, because they are performed by people who are not part of the hierarchy, 90 acts performed by the bottom nine people in the hierarchy; so, again, the total number of acts performed is less than the original 400, and not all of those bottom 90 acts are in accordance with the desires of the dictator. Using our 25% deterioration, we find that 56% of the acts at the bottom level are performed in accordance with the desires of the dictator, or 50 together. The people at the bottom also perform 39 acts that are not in accordance with his desires, and the total number of acts thus is therefore 309.

I found this section confusing at first. In other words, Tullock is imagining a scenario in which the one dictator takes control of 3 people who together coordinate 9. The remaining 27 people are left on their own to do whatever. Thus he gets a consistent set of 50 acts that follow his dread bidding. Why would he prefer this structure to the previous one in which 114 acts were in accordance with his will?

Obviously, this situation is less desirable from the standpoint of the dictator unless the 50 acts that are now subject to control are more important than the others. There is no reason why that could not be so, and, as a matter of fact, most dictators do follow this kind of a pattern or some system under which they have even less control. In other words, the dictator has reduced his total control over the environment in order to concentrate his control over some things that he finds very important.

This situation of a leader attending only to those aspects he finds very important reminds me of Dominic Cummings. Cummings' complained that Boris Johnson was more concerned with his girlfriend than with nuclear security issues. But although describing a salient dynamic, I think Tullock is missing something here. Surely, part of the reason some things can be ignored is not merely that they are less important, but also they can be relied upon to operate acceptably smoothly without additional interference. I take issue with Tullock's model for not explicitly including any assumptions about the nature of the 270 acts which are not according to the dictator’s wishes. Clearly, they are not exclusively rent seeking activities, such as embezzling company funds for flights to the Prospera Space Agency, or clearly at cross-purposes with the dictator, which would, I imagine, provoke his attention. Instead, Tullock models them as orthogonal and irrelevant at current margins, whereas in reality some acts are subtle sabotage unlikely to be caught and others are of the helpful-but-kind-of-missing-the-point variety.

The leader can decrease his span of control and yet still create a more effective organization if he knows what things to target. Tullock is driving at this lesson. 

Our final model here is one in which the dictator has not two but three levels of importance for acts. The things he thinks are important, he does himself; we will assume he uses one-third of his time and therefore gets 3 contacts with the outside world perfectly in accord with his preferences. There is another one-third of the acts which are dealt with through a very small bureaucracy. Since he is now devoting one-third of his time to controlling that bureaucracy, he does not get fifty contacts with the outside world in accordance with his preferences but only twenty. The rest of the society is also organized in a gigantic hierarchy with him at the top, and he devotes only one-third of his time to supervising that, with the result that 20 of their acts are not in accordance with his preferences. For a grand total, he has 43 acts in accordance with his preferences.

Okay, so he is taking some control of the part of the 27 ignored people! I was getting nervous and a bit clammy about that.

Clearly, this situation is worse than either of the other two, but he has stratified things so that the most important acts are certainly in his control, the second level of importance are to a considerable extent under his control, and the ones that are least important are only modestly controlled. Whether this is better than the earlier systems depends on whether he has strong feelings about which policies are most important.

The crystalline insight here is that the control structure reveals the different levels of importance attached to different activities by the leader. We can take this insight and apply it to our organizations and lives. Does the control structure match the relative importance I attach to various aspects of the work?

The assumption that some things are more important and have greater weight than others seems obviously true. But for those of a high modernist bent, or those who wish for well-functioning bureaucracy, this is deeply depressing. Tullock makes it seem that there is no way to achieve highly coordinated efficiency in a large, formal organization. The entrepreneur or CEO will always be unable to have his vision or wishes transmitted faithfully and then performed.

This fabricated structure is actually quite realistic. Supervisors, whether they are dictators or branch managers of Safeway, do make some decisions themselves for direct action to the outside world; they supervise some through a narrow hierarchy; they allow many aspects to be dealt with by a wide hierarchy. Even this system with three levels of importance is a simplification since many things are seen as not significant enough to supervise.

I see this passage and thought experiment as a culminating one. Many of the insights scattered throughout the work are included here. Span of control depreciates down the hierarchy. Leaders have developed different methods for handling the depreciation rate and limiting rent seeking, such as random audits and administrative shake-ups. The search for good incentives and the constant tinkering with them reveal the extent to which span of control is the defining issue of the firm. Yet, like in much of the book, Tullock fails to tie his insights about various methods into a coherent framework for determining how any such an organization should be set up. I feel so close to a visceral insight too, but so far I’m drawing a blank.

I do want to meet “the Napoleon of Safeway”; that’d be a treat.

Governance and Incentives and Committees

Our final section is on governance.

Prospera Space Agency might select a dictator to rule and administrate, then the dictator might select and sort other people to lead teams in departments. The key thing is that the dictator accurately judges what is most important to keep track of and knows what can be left to other departments to figure out on their own.

You know, sometimes supervision is easy. It doesn’t have to be hard.

Much depends of course on how easy the task is to supervise. The situation in the traditional army in which soldiers stood in long straight lines is unfortunately exceptional.

Frederick the Great watches from the hilltop to see which regiments of the 10,000 men are marching forward at the right pace.

Prospera Space Agency is going to have to be lean and intelligent to do so much with so little. For example, it is more likely to organize itself into functional departments, each carrying out radically different tasks, not able to be supervised by the Frederick the Great method.

Less common method is one in which the government is divided, not geographically, but functionally. The ruler maintains an army with garrisons everywhere, a police ministry will have offices scattered throughout the country, an engineering division will build roads all over, and so on. Normally, most governments have been geographic with some aspects of specialized functional division. The central government of the United States is not dictatorial, but like many other governments in federal areas, it is divided functionally, rather than geographically. Of course, that is compensated for by the fact that the states make up a geographic division that is not part of the federal government.

In our age, the equivalent for neat military lines is financial statements. These are a nice assessment tool.

The accounting system provides a fairly decent way of measuring the impact of various divisions of the larger enterprise on the larger enterprise's profits. Something like the general counsel's office, to a large extent, escapes this measurement process. Because although the costs can be easily evaluated, the benefits are hard to put a price on. Aside from the advantages of having a simple objective, and a method of measuring its success, corporation officials must be continually concerned with the accounting evaluation of their behavior. In other words, they must please their superiors.

So since accounting measures are a standard that knowably and reliably pleases the superiors, we should expect individuals to try to meet at least the minimum threshold of departmental profit.

Accounting measures calculate how much profit is made, they do not tell you how much is due to the external environment, the competitive pressure, and so on. In most cases today however, the accounting system is a great boon to managers of large companies attempting to deal with their subordinates. Any manager who uses the accounting system and nothing else will be beaten by a manager who's able to use the accounting system as a guide, but then improve on it by considering the degree of competitive pressure and other factors.

Fine, they aren’t perfect. We didn’t say market forces were perfect, but that they are way better than the alternative.

Even though managers of profit seeking enterprises are in a much better situation than the managers of the Department of State and controlling their subordinates, the problems are still real.

Tullock is not so sanguine about market force incentives sweeping throughout the company, making the bureaucrat behave. Remember, the bureaucrat is not assessed by the market!

For now, we will assume that the official’s objective is to follow the course of action that maximizes his likelihood of promotion, or at least retention. This assumption is not arbitrary. People who maximize their career goals rise fastest. Hence, they're the topmost people in a bureaucracy once they get to the top, however, they no longer have anywhere to go. At that point, they may change to a heavy consumption expenditure.

One can definitely feel this pull when one finds a comfortable position in an organization. I have at times. You say to yourself, “I have risen as high as I need to. I am respectable and am in a respectable job doing well and doing good. I deserve some time for complacency.” I suppose one way to avoid this is becoming power hungry, but that seems like it might bring out some of your less flattering traits. Another method is to have more kids until it is hard to afford them, then you’ll have to find ways to keep rising!

Phil Tetlock, in Superforecasting, mentions that forecast skill also has to take into account difficulty due to external factors. Predicting rainfall in Phoenix, AZ accurately is easy. In Springfield, MO, it's most difficult. The same goes for assessing departments: it is difficult to assess how much success or failure is attributable to internal or external factors. Since it is so hard to compare corporate departments, even with the pure accounting mechanism, it is especially difficult for those departments whose value is not monetary in nature.

I mentioned earlier the impossibility of determining whether a given embassy is doing a good job. And, to give away a secret, the Department of State does not make any significant effort to find out. If all the routine steps are taken, the various reports filled at the right time, and Congressmen pleasantly entertained, higher officials can expect a standard rate of promotion.

Sometimes it is abundantly clear that a department or a bureaucrat is ineffective, and yet…

Here's a true horror story. A political appointee in the Reagan administration discovered that one of his high-ranking civil servants was never in the office. When he inquired about the man, he learned he was a dipsomaniac and was home drunk. Since the appointee had some fervor, he decided that the civil servant should be fired, if he did not cure himself. The lengthy procedure for getting rid of a civil servant requires that his supervisor have several interviews with him, in which the defects of his work are pointed out. Since the man was almost never in the office, this approach proved difficult. However, when it was finally accomplished, the man was fired. He probably sued the government for reinstatement and the Civil Service Commission not only reinstated him but compelled the government to pay his lawyer’s fee.

Excuse me, while I bang my head against this palm tree.

That was some extremely effective rent-seeking. I pray every morning that this was a fluke example and if anyone in the comments provides their own horrifying example, know that I will not forgive you should it cause me a mental breakdown.

We are concerned about rent seeking in the interior of the corporation. Let us elaborate on our previous discussion of leveraged buyouts. Suppose that we have a large corporation and there appear to be profits from taking it over and reorganizing it. The method of taking it over has changed over time. Long ago, it was proxy fights, then it became the activities of corporate raiders, and now it tends to be leveraged buyouts.

In general, Tullock says, cutting 1/3 of a bloated middle management seems to work reliably for increasing efficiency, for a while at least. He posits two reasons for this. Firstly, remaining management works harder to compensate for the cost of acquiring the company. Since they are now in debt with interest, if the company does not turn out profits the investment might yield a loss. Secondly, management now has less time and ability to seek out rent-seeking opportunities after the reorganization.

As a last bit about governance, Tullock explores the nature of rewards and punishments. In the spirit of free market theorists, he emphasizes that freedom of exit means that all so-called ‘punishments’ within the organization are actually reductions in benefits.

It used to be said that when J. Edgar Hoover did not like an official, they would send them off to the FBI offices in Butte. This was regarded as punishment. But it really was not, since the person could freely leave if he wanted. What it was really was reducing his rewards sharply. It is easy to mix up the choice between a cut in the rewards a person receives and actual punishment. From the standpoint of the person's situation before the change is made, there's no great difference, except that it is always possible for the person to refuse to accept the reduction in reward and quit. You cannot do that with a true punishment.

It is one of the characteristics of capitalism or, for that matter, any kind of open economy that the basic method of getting obedience is the carrot. The stick is restricted to preventing crime, and things of that nature. A man given a sentence of 10 years cannot just decide to quit. One of the characteristics of communism as it was, not as it was claimed to be, was that it depended heavily on the stick rather than the carrot.

I find this an important and interesting reminder for political and ethical reasons. Society in its governments and organizations confers benefits and incentives on us, but when we make rules which compel action, it is likely best, insofar as possible, to remember that freedom for exit is an important check. People might choose to incur the high costs of a reduction in benefits, but that is not the same as a punishment. Offering people these options seems to be a hallmark of a non-tyrannical society.

Committees

Anyway, getting back to governance, we need to talk about committees. The endless supply of committees. The never-ending meetings of committees. Why do they torture us so?

The committee itself may in a special way reduce rent seeking activity. Personal goals are harder to push through a committee than through an individual because of the number of people to convince and offsetting factors that individuals do not have as much riding on the matter and, therefore, may put less energy into it.

If you want to stop Lex Luthor from rising up the ranks too fast, try committees!

This indicates that the innumerable committees which we observe in large corporations, and meeting on all sorts of subjects actually serve the function of making rent seeking somewhat less severe.

This seems to cost in efficiency, though. Committees are not known for moving fast, are they? People, though, are even less willing to take responsibility for their actions than committees are.

A relatively risk-neutral person would probably be more likely to take chances with his own corporation than with a corporation by which he is hired. I have a friend who for a time was a high official in one of the more powerful and older American corporations. He said one of the real functions of the committee structure in this corporation was that it permitted people to take large risks without any responsibility. You never could trace any losses back to anyone. This protection may have set the other phenomenon [of creating inefficiency].

Frankly, I never thought of this. I am very surprised by this idea! Committees take risks where individuals won’t? I want some more evidence before I can be convinced. Maybe the idea is that they can in some circumstances be a type of bravery support group so that collectively the institution can pass the bravery test and make tough and risky decisions? Or maybe, the average individual is more afraid of being blamed for a decision gone wrong that will take them out of the game, and so, on average committees dominate dictators? I’m not sure of this, though I’ve written on related issues before. [LW · GW]

Why then is a committee a good vehicle? We have mentioned that individual committee members will not produce as many ideas as they would if the same individuals were dealt with separately. Assume that five junior officials produce some particular quantity of new ideas individually. And that when you subtracted the duplication, there would be a total of say 25 ideas. The committee would produce fewer ideas, say fifteen, as many as three individuals operating by themselves. They then examine this collection of ideas and write them according to merit, passing on one to their superiors as the principal recommendation with possibly several alternatives. They are fully prepared to switch to another if the superior does not approve of it. The process will not be perfect. However, it is likely that the top idea passed on to the superior is, say, one of the three best out of fifteen.

The committee then combines a process of generating ideas with a process of winnowing. It does neither with perfect efficiency but performs the combined task better than any other system. It particularly eliminates genuinely poor ideas early in the process so as not to waste time on them. Careful attention should be given to the opportunity the structure provides for the strategically maneuvering and rather unscrupulous junior officials. The committee is not a good area for him to operate, because his actions will be promptly noticed by rivals. He may of course, be an expert in committee manipulation, and may operate very well in committee. Hence, his promotion is promotion for merit. A committee by guaranteeing much publicity for almost anything that is done, makes the more devious types of political maneuvering by officials less likely to pay off.

So committees, without having the best ideas, may still propose consistently high quality ones. By this process of generating and winnowing, they are also keeping eyes and ears on each other’s intentions and actions, creating the social cohesion, or pressure, if you will, to keep everyone on the same page. This can be good for keeping outlying members in check, but bad if the outlying members are the only competent ones. From an organizational standpoint, I guess you would want to gauge whether your goals are better served by a normal distribution of talent which tends towards being typical, slower, methodical, and echo chamber-y, or if the mean performance doesn’t cut it, by handing the power of the committee over to a dictator.

This then is my explanation of why we see all these committees in profit making corporations, even those subject to considerable pressure. Normally, committee structure is cut back when the organization is in trouble. But that may simply indicate that they are taking a relatively short-range approach to economy rather than long-range. A company on the verge of bankruptcy is hard pressed to make money immediately and may be willing to give up some potential for future money making. These remarks about the committee, however, are only a specialization of some more general remarks about the nature of corporate headquarters. Anyone who has had contact with them will be impressed with not only the number of committees there, but the number of people, many of whom do not have highly specified jobs. Indeed, the professional management advice teams that go around the world spend much of their time bemoaning this fact and suggesting that the jobs be sharply defined, and that people stopped getting outside of their formal jurisdiction. In spite of all this advice, the phenomenon continues.

I really like this framing of committees. It allows that they have use for ensuring long-run continuity of whatever but, like the Roman Republic, need to be suspended in a crisis. And furthermore, informal roles and interactions dominate human affairs given half a chance. This is something that the inner James Scott always insists on, quite rightly. Trying to reduce what people, heck, even a company is trying to do to one or two variables simplifies too much. Given half a chance, humans crave imperfect, messy, highly networked systems, with strange and distributed feedback loops. The highest and best officials harness and love this jungle, which flies in the face of Big Brother.

Well run organizations, however, do not simply accept the top recommendation of this administrative swamp. The higher officials look over several of the decisions, selecting the one that he likes best. Sometimes the individual at the top will invent his own idea, rather than accepting the ones presented to him. Even in this case, however, he will usually turn to the administrative swamp for some processing of his basic idea.

Back to the 900 Prosperan Engineers

What does this all mean for the Prospera Space Agency?

Self-selection means that we can ignore all sorts of normal bottlenecks that throttle normal organizations. For example, we can expect all of our members to be fanatics, dedicated to the mission of technological progress.

The cameras and public attention likely mean that rent-seeking - of the egregious sort - will be nearly non-existent. In fact, many participants in the experiment will likely see themselves as being in a giant, public marshmallow-test style job interview.

Assuming some pretty standard functional departments, I doubt anything very unique organizationally will emerge from the “Prosperan Naked and Looking to Tech Up” experiment. They will likely just create a jungle of interrelations the way any high powered network does, with some formal structures for guardrails, and get to work. But I would like to be surprised by something different.

We started out with a dream of progress through natural experiment, but by this point it has unraveled. The natural experiment of 900 engineers was contrived. Its lessons are unlikely to transfer. While it is still perhaps a great idea for cultural-impact reasons, for the purpose of coming to a deeper understanding of bureaucracy it’s no better than the default option: getting a job or starting a company.

I am sad that my original game plan of relating everything to my original thought experiment has not worked out. Many of the problems of a bureaucratic organization are contingent upon the environment. And in this case, the artificial environment of a prestigious natural experiment changes too many variables to draw general lessons or implement a first principles state-of-nature approach to organizational creation. Once again, high modernism has been foiled, because we overdetermined the environment, and wished to use simple models, not just to describe, but to prescribe human activity.

Puzzles Remain, while Good Governance remains a Scarce Resource

Tullock ends in aporia.

Having frankly confessed that I am puzzled by [why we see so many structures], I could put this manuscript in a drawer, and hope that in another 20 years of thought, I might solve it. Since I am 69, however, that does not seem feasible.

The alternative is to present the problem to my readers and hope that they will solve it, preferably in less than 20 years. Detective mysteries contain a challenge to the reader. I'm following that precedent. Can you produce an explanation for these many different organizational forms? If you can, you should become both wealthy and famous. You would also confer a great benefit on society. With all those incentives, surely, you will try.

I am a sucker for challenges like this. Someone posts a bounty, and I am interested immediately. Tullock has, in a way, placed a bounty on bureaucratic organizations. And I fell for it by writing this review. Scott placed a bounty of sorts on writing compelling reviews, and I fell for that too. I want to work a little less hard at my job, and so here I am, rent-seeking, i.e. writing a book review.

Economic Hierarchies has provided me with a handful of tools that I will certainly use and think about in exercising organizational control: the cross-span structure, like the U.S. Army uses, seems to provide a lot of safeguards at the cost of a large administration. This structure seems right for organizations with very low error tolerances. Leaner structures, like the final structure in the “Big Span of Control Model” rely on a lot more trust of less controlled members of the bureaucracy while administrative focus is on other important aspects. I found the how and why of different types of administrative shake-ups something to keep in mind as my organizations grow. Tullock's explanation of the good side of committees has helped me overcome my committee allergy. Methods of random allocation and functional grouping allow one to optimize audits and assess departments against each other. 

These little strategies amount to one core take: each hierarchy exists to overcome some limitations, but it cannot do so costlessly, and always does so imperfectly, managing these limitations, costs, and tradeoffs is what the leader does in an effort to achieve some ultimate goal.

Another general hypothesis I formed while reading Economic Hierarchies is that bureaucratic structure is highly path dependent from the moment of its establishment. Initially the entrepreneur tries to expand capacity to manage different subsidiary projects. The way this happens likely depends on the external environment and resources open to the entrepreneur at the moment, and then follows path dependency for some time. Committees and heavier bureaucracy act as insurance against the worst happening against the leader’s wishes, but they limit the freedom for action of the most talented - creating a tradeoff between the longevity of the organization and its capacity to act.

As a result, I have also become more agnostic about general structure mattering so much. Although Tullock sought a guiding principles, I am suspicious that they cannot be found. Similarities will be Tolstoyan, in that “all happy companies are the same, and all unhappy companies are unhappy in their own way," but with the caveat that there are no happy companies. And important differences in structure are based, not upon some simple, rational schema, but, at best, a 70x70 flow chart of situations one might find oneself in.

As a meandering meditation with anecdotes Economic Hierarchies, is a great lubricant for bureaucratic story swapping. Gordon Tullock’s favorite examples involved Safeway, Napoleon Bonaparte, the Department of State, Chinese government, and Fairfield Ltd. a real estate construction company. If he were successful in working some deep insight into my mind, I would be proposing we go out and test a few ideas right now.

As it is, I, like him, submit myself to aporia and leave the adventure to you. 

 

4 comments

Comments sorted by top scores.