Arnold Kling  

Thoughts on Spontaneous Order

PRINT
One Take on Strauss's Crazines... Book 1 Podcast...

In a comment from a troll Bryan Caplan entry, I find,


I've long since lost all patience with Hayek. His original, true ideas could have been five good blog posts

Personally, I think that the concept of spontaneous order takes more than five blog posts. The fact that Cato Unbound is devoting a whole issue to it is indicative. In it, John Hasnas writes,

constructed orders have a designated final decision maker; spontaneous orders do not. ..What makes the state the state is its power to stop evolution. In short, spontaneous orders are systems of individual choice. Constructed orders are systems in which there is an official organ of collective choice.

The question this raises is whether a designated final decision maker can defer to a spontaneous order. This relates to a difficult question from Robin Hanson, which is why would a small firm ever have a comparative advantage. If a firm happens to be large and complex, the CEO has the option of running the firm in a decentralized way. Instead, large firms are usually run in centralized fashion, with consequent disadvantages, or so it would seem. Some possible explanations:

1. CEO's are biased to think that they can outperform decentralized decisions. Just as Progressives are biased to think that government can fix imperfect markets. Or just as people feel safer driving cars than taking planes, because they are in control when they drive cars. Or just as people believe that they can outperform market index funds by making "smart" stock market picks.

The advantage of this story is that it is quite plausible to believe that people, particularly people as egotistical as CEO's, over-rate their competence. A disadvantage is that one has to explain why the market does not punish such overconfidence.

2. Firms are primarily status games, not profit-maximizing entities. Large firms give people an opportunity to compete for globally high status positions with a relatively low probability of achieving those positions. Small firms allow people to obtain locally high status positions with relatively high probability. People differ in their tastes for these games, as well as in their tastes for stability and security (large firms) vs. novelty, risk, and opportunity (small firms). Thus, both small and large firms can exist within the ecosystem. Presumably, large firms will tend to be found in industries where large firms have a comparative advantage. It is not in the interest of managers of large firms to try to run divisions in a way that replicates the behavior of small firms, because that would be throwing away their comparative advantage.

When I started at Freddie Mac in late 1986, the company was at that inflection point where it had passed the size of a small tribal entity. The CEO noticed and mourned the fact that it was attracting more people who were focused on job titles. This was a sign that as the company grew, the status game had shifted to something broader, in which formal titles held more meaning..

3. Risk issues.

If a big company decentralizes, there is no effective way to control the risk that an individual unit creates. Look at what happened to AIG as a result of its Financial Products unit. One can argue that AIG suffered from too little bureaucracy. A better-managed entity would have put some bureaucratic roadblocks in front of the folks who wrote all those credit default swaps.

Robin asks why a firm cannot simply give a division the same set of incentives that it would face if it were a separate entity Well, if Financial Products had been a separate entity, without AIG's AAA rating to trade on, it would never have gotten started. Perhaps that is the right outcome. But one can also envision circumstances in which a division could have exploited AIG's AAA rating without taking excessive risk, and in that case making decisions as if all entities have to be able to operate as if they were separate from the parent company would yield the wrong answer.



COMMENTS (7 to date)
baconbacon writes:

I don't think that any of your explanations work well because you ignore political influence

"A disadvantage is that one has to explain why the market does not punish such overconfidence."

"When I started at Freddie Mac in late 1986, the company was at that inflection point where it had passed the size of a small tribal entity. The CEO noticed and mourned the fact that it was attracting more people who were focused on job titles. This was a sign that as the company grew, the status game had shifted to something broader, in which formal titles held more meaning.."

The market clearly tried to punish Freddie Mac- only to have to government rein in that punishment. It is very surprising to see you ignore this obvious difference between large and small firms during a time in which large firms are having all of their flaws exposed and then papered over by the gov. If large firms have an advantage in political power then they would naturally gravitate toward CEOs who specialize in political influence and not toward efficiency maximizers. It is also not difficult to see why such individuals must often be involved in more minutia than one might think. If a CEO wants to influence a Senator he might agree to have a new factory in division X built in a core constituency area of that Senator (and naturally have the CEO shaking the hand of the Senator at the opening). In doing so 9 times out of 10 this would not have been the location that the subdivision would have chosen for their new factory and thus they are going to take smaller profits (or larger losses) so that the whole company can enjoy the benefits of the Senators influence. In this way we can see that selection will be toward those CEOs who are more involved, more likely to make decisions for individual pieces of the company as well as for the broader company.

george writes:

Your first explanation is correct. You question whether the market punishes CEO overconfidence - and the answer is: of course! The very existence of the small companies - usually in niche markets - is proof. If the large corporations could, via centralized control, perfectly respond to all business opportunities, small firms would never exist. The punishment is usually in small amounts of market-share loss (small relative to the size of the firm). Occasionally, large firms get punished by eventually losing large amounts of market share: an example is IBM not responding to demand for mini-computers, thus spawning the likes of DEC, etc., and, later, DEC not responding to and developing microcomputers.

Alternatively, if decentralization could achieve the same, we would eventually evolve to a single decentralized corporation servicing all possible business opportunities - an event whose probability is exactly zero.

TomB writes:

There are varying degrees of the centralization of decision-making in companies. Some are run more entrepreneurially, giving groups less direction and more space to solve problems in their own way. I have seen such a structure at some consulting companies.

To be a single entity, some degree of centralized decision-making is necessary. The company needs to have a cohesive strategy. Homogeneity of product offerings is a benefit. Without sufficient direction, groups or divisions could start competing with each other instead of cooperating.

Once an entity becomes large, it becomes less nimble because it has lots of capital invested in long term projects. Smaller firms are more nimble and can move into new markets more quickly. Larger firms frequently acquire these new firms rather than take on the risk of attempting to enter new markets.

bbb writes:

A quick Coasean answer to Robins "difficult" question: organization costs.

Longer Hayekian answer:

The rationale for large companies allowing their divisions a lot of freedom lies in the greater utilization of dispersed knowledge which decentralization allows. Decentralization begs the question of how much freedom the divisions of the large company are to be allowed, and to what extent they are to be allowed to use the companies common assets (financial resources, brand name and good will, etc) given their degree of freedom. Total freedom of action for the division with no restrictions on the use of the companies common assets would lead to an costly overutilization of these commons.

The reason for the inefficiency (higher organization costs) of divisions in a large decentralized company lies in the fact that also the decision of the CEO on how much freedom to grant and what limits to impose on it can only be based on the CEOs own limited knowledge.
Even if the division tried to convince the CEO of the profitability of permitting certain actions causing costs for the whole company, the CEO will only be able to decide on an issue based on his own knowledge. Because of the imperfect articulation of information (tacit knowledge) this knowledge must always be different (smaller) than that of the divisions members.

So the institutional form of small firms (not necessarily the small firm itself) is more efficient because it has lower market entry barriers for dispersed valuable knowledge and less inefficiencies based in asymmetric knowlegde (limited knowledge of the CEO central planner) compared to the form of a large decentralized company.

Eric Rasmusen writes:

(1) Yes, Hayek can be summarized in five lines. So can any good idea-- a good short abstract is the sign of a big idea. One of course needs more lines to persuade and elaborate.

(2) The diseconomies of scale topic is hugely important and neglected. Two thoughts:

(a) One firm with many divisions has only one stock price. Thus, owners of a big firm have a coarser view of the criticism of the market and can't incentivize the division managers as well as it could if they were separate firms.

(b) One can of course completely decentralize a firm--- even having separate corporations and stock prices. But then it isn't really a firm anymore--- it is a holding company.

mark writes:

Why is "spontaneous order" called that and not "organic order"? I think the latter would sell better since "organic" is universally perceived as good in current parlance whereas "spontaneous order" has echoes of "spontaneous combustion" in it that trigger a reflex of skepticism.

Ryan Vann writes:

Mark,

It's funny because it is true. I've always been keen on emergent order, as it implies a sense of unpredicatability. Complex adaptive systems is even more precise, but runs the risk of sounding like verbose technocratic jargon.

Comments for this entry have been closed
Return to top