« Bonuses & productivity | Main | Economics for politicians »

April 24, 2015


Feed You can follow this conversation by subscribing to the comment feed for this post.

Roy Lonergan

It wasn't just the Tirole point in the last post, there's also Stuart Gulliver saying, “Can I know what every one of 257,000 people is doing? Clearly I can’t.” Well, of course he can't. But he's paid as if he can. In reality everyone is in a small business more or less consisting of all the people they talk to in a day. I suspect that an organisation doesn't have to be very big before it hits the Hayek Threshold (I'll claim that as a coinage if it's new!)


Isn't Hayek's "insight" just a little trite?


I think Roy's "Hayek Threshold" is an interesting concept, but I've got 2 potential caveats.

1. The Hayek Threshold measures a single variable which is likely inseparable from other variables related to size/growth.

So, you could cross some standardised Hayek Threshold but still see positive returns to size if the business benefits from economies of scale...

2. It's likely to be a measure of communication efficacy, on a spectrum, rather than a threshhold related to company size.

It's part of why (great) startups are focused in their early days in direct customer feedback -- there is an opportunity to build a better product by winning the information battle (against incumbents) through more direct communication with, or insight about, the target customer.

As a non-economist, I'm not well enough versed in Hayek (or for that matter, Coase) to say much about the firm as a system of information-coordination, but it does seem that a well-run small company should have certain advantages against a larger competitor because it can gather and act on information more efficiently, all things being equal -- only, all things aren't. :)


@ Roy: you might be right about the Hayek threshold (good phrase!) being quite low:
Only a tiny fraction of firms are big - only 0.2% have more than 500 employees:


Right, but you don't have to actually know what everyone's doing in order to be a good CEO. Capital allocation is a much more important part, and generalised cost-control. Both can be achieved without "knowing everything". As an investor that's what I look for in a CEO. Much the same in govt. Same as a football manager.


@ Uberlibertarian - that's a good point. It's consistent with Boris Groysberg's work, which shows that what matters isn't so much management skill per se as a match between the manager's particular skill and the job requirements:
1. This implies that talk of "good" managers misses the point: what matters most is the match, not the person.
2. There's a danger of overconfidence: a once-good CEO can become dizzy with success - eg Goodwin's successful takeover of NatWest emboldened him to buy ABN Amro.


Hayek's quite a significant figure in legal philosophy, too. I think he's generally seen as wrong and a bit outdated, but the name still has a few solid ideas attached to it - and they're not necessarily ideas which prove the moral superiority of the nightwatchman state, as much as he might have liked them to be.


Has anybody formalized the knowledge aggregation by markets? Anybody knows about some info-theoretic view on the markets?

E.g. if we have a market with Bayesian agents, each one having a posterior and trading based on it. What is the market posterior?

What if the agents are not just speculators (but e.g. sell just because they need liquidity?).

Thank you!

Steven Clarke

Thanks for the mention!

One aspect of this that you haven't talked about in this post is when the knowledge problem is systemically important or not.

If a many individuals manage their wealth badly, it may be bad for them, for tolerable overall. If many firms are managed badly, it will be tolerable. "There is a great deal of ruin in a nation".

But if 'too big and interconnected to fail' financial institutions are badly run, it can be disastrous. If a government is badly run, that also imposes huge costs.

On another point, I have a question about the active/passive management debate.

I'm also inclined to the belief that "you can't beat the market". But the market is ultimately made of individuals who use their private information to get ahead.

It seems slightly paradoxical - if everyone invested in passive funds, no one would be gathering any information for the market to process at all.

Roy Lonergan

Re posts later than my last, the Hayek Threshold might be:

1. Something like the size of the observable universe, so s multiple of c.
2. Something lile the uncertainty prínciple with a limit on certainty of conjugate variables.
3. Something related to entropy and the second law of thermodynamics.

Or whatever. The point is that this would be a limit on anyone's possible influence on anyone else. Possibly there is an answer in information theory. I don't know, but I'm happy with the idea that some management is outside threshold and do hasn't a clue.


UberLibertarian asks:

What as an investor do I look for in a good CEO?

He then makes the great leap and equates this with a good CEO in a general sense. In other words a good CEO is dictated by investors.

Well, may I interject here and say bollocks to all that!

For example, utility companies may make shit loads of money because the market doesn't work for consumers, and the average investor will think, "What a CEO, I want his children", while the customer will think "You fucking robbing bastards, they should fucking hang that thieving CEO, they estimated my meter reading at 3700 and it is only 3500, they bastards have had an interest free loan, give me the address of that CEO twat so I can give him a piece of my mind!"

On information, the CEO is all knowledgeable when things go smoothly but when the shit hits the fan and the fraud investigations start they suddenly claim to know absolutely nothing.

From Arse To Elbow

Dmcdougall says "a well-run small company should have certain advantages against a larger competitor because it can gather and act on information more efficiently". But the evidence, in terms of profitability and productivity, is that large companies are more efficient - that's why they get to be large.

We shouldn't be distracted by the ideological fluff of "nimbleness" and "disruption". The economy is dominated by large businesses that have become larger over the last 40 years not because of globalisation (which is symptomatic) but because they have become more efficient as bureaucracies.

Hayek (like von Mises) is of declining relevance because his pessimistic worldview was the product of Hapsburg bureaucracy as much as a fear of communism. He was unable to appreciate the impact that information technology would have on (commercial) central planning. Instead of Gosplan we got Google.

The "Hayek threshold" sounds like a variant on Coase, given that the transaction costs of a firm are largely the costs of information-aggregation - i.e. knowing what is going on in a timely enough manner to take action. When businesses talk about "focus" or "control" when making out/in-source decisions, they are usually talking about the cost of information.

As technology has driven the cost of information down, and allowed the exploitation of whole new classes of data, so the size of big business (in terms of reach and market dominance) has increased. Ironically, this has also allowed the boundaries of the business (in terms of beneficiaries) to be reduced, which means that a global corporation can still pretend to be an upstart.

Steven Clarke


I may be very wrong, but weren't higher transactions costs the reason why things are done in-firm rather than through the market?

In which case, lower transactions costs would lead to more transactions between firms rather than within them - leading to smaller firms rather than bigger ones.

gastro george

"As technology has driven the cost of information down, and allowed the exploitation of whole new classes of data, so the size of big business (in terms of reach and market dominance) has increased."

As somebody noted (it might well have been FATE), the market has changed. In retail especially, money was made selling goods to consumers. Now the big money is made selling consumers to suppliers - this applies across Google, Tesco, etc. It's only big companies with the data that can do this.

From Arse To Elbow

@Steven, there are two tendencies at work here.

The fall in traditional transaction costs has allowed many corporations to shrink by making outsourcing and sub-contracting easier. But the same forces that have enabled this (IT and the legislative framework of globalisation) have also made it cheap to gather the sort of information that was once prohibitively expensive (e.g. customer preference data).

A good example of this is Google Street View. The camera-car drivers are subcontractors; Google now has a dataset of images that hitherto would have been unfeasible to create. In terms of reach and market dominance, Google is a lot bigger than GE, Ford or US Steel ever were.

Bob Roddis

"The reason that a socialist economy cannot calculate is not that it is socialist, but because a single agent owns and directs all resources. Expanding on this point in his 1976 essay on “Ludwig von Mises and Economic Calculation Under Socialism,” Rothbard explains:

There is one vital but neglected area where the Mises analysis of economic calculation needs to be expanded. For in a profound sense, the theory is not about socialism at all! Instead, it applies to any situation where one group has acquired control of the means of production over a large area—or, in a strict sense, throughout the world. On this particular aspect of socialism, it doesn’t matter whether this unitary control has come about through the coercive expropriation brought about by socialism or by voluntary processes on the free market. For what the Mises theory focuses on is not simply the numerous inefficiencies of the political as compared to the profit-making market process, but the fact that a market for capital goods has disappeared. This means that, just as socialist central planning could not calculate economically, no One Big Firm could own or control the entire economy. The Mises analysis applies to any situation where a market for capital goods has disappeared in a complex industrial economy, whether because of socialism or because of a giant merger into One Big Firm or One Big Cartel. (Rothbard, 1976, p. 75)

The Mises analysis thus applies to any situation where the market for a particular capital good disappears because a firm has become so large that it is the unique producer and user of that capital good. As we have seen, such a firm will not be viable."

"Economic Calculation and the limits of Organization" page 15-16



I've said it before, but it bears saying again. Information is fragmented, but a collapse into a single dimension variable (price) is a profoundly limited (and inherently inadequate) response to the fragmentation.


«perhaps competition will eventually weed out egregious incompetence, but market forces might not grind so finely as to eliminate all inefficiency - and might even in some cases actually select in favour of quacks and charlatans.x#

That's about as good as it gets in practice. Also "eliminate all inefficiency" is both unattainable and undesirable. There is going to be an unavoidable lowest level of mistakes, and all we get to choose is whether they are going to be type 1 or type 2.

«[ ... ] Hayek's message has shifted. Whereas it used to support dominant western institutions, it now undermines them.»

Someone at The Economist (before it became largely a propaganda funnel) opined more or less that in open economies competitive markets operate among businesses in which the most efficient business wins, but businesses are soviet style organizations inside and it is the best manager at politics who wins.

That's an important dialectical tension...


«a collapse into a single dimension variable (price) is a profoundly limited (and inherently inadequate) response to the fragmentation.»

If you have a better idea, tell us :-).

Also note that there is another dimension: preference.

What "price" does (when it works) is to set a boundary to preference, but within that boundary choice is arbitrary.

The "Information is fragmented" that is collapsed into price is largely about cost. (when price works well as a scarcity indicator, that is when rent does not happen significantly), Disregarding information about cost is usually not a good idea, however crude the information about cost has been collapsed.


«For in a profound sense, the theory is not about socialism at all! Instead, it applies to any situation where one group has acquired control of the means of production over a large area»

That is part of one of my common observation and it is that because of clever propaganda that the political system "capitalism" (where political power is held by big business property owners), the institutional system "markets", and the production system "industry" have been conflated, and the propaganda bit is that the benefits arising have been attributed to "capitalism" rather than "competitive markets" or the "industrial system" of production.

Socialism could well be based on "competitive markets" and the "industrial system", and vice-versa capitalism can well operate with soviet style managerialism and with less productive modes than the "industrial system" (and it does in many fascist or authoritarian states).

Therefore for those who think that "competitive markets" result in better coordination though the "wisdom of crowds" than managerialism the latter is a bad idea whether the political system is capitalist or whether the production system is industrial.

My understanding of economic history is that of the three above it is the industrial system that gives the most benefits, and competitive markets comes second, and the political system does not matter much, as long as (a big qualification) it provides an environment for industrial production and *competitive* markets.

But that overall I think all that is not very important compared to the really big determinant of relative prosperity, which is a reliable institution of bankruptcy. My impression is that regardless of the political system, mode of production, exchange institutions, cultures where bankruptcy is avoidable or difficult or complicated become rather poorer than otherwise.

Because it is ultimately bankruptcy that leads to better use of resources and allows taking advantage of the "wisdom of crowds" or whatever else in allocating resources.

Some people think conversely that the institution of diffuse and enforced private property is the key, but I think that it is not that essential.

PS JK Galbraith pointed out that most of a capitalist economy is actually planned soviet-style, inside huge corporations well before MN Rothbard published his paper.


Hayek was a self described medievalist. I think that says enough.

Steven Clarke

@FATE Is there not perhaps an interesting distinction between the likes of Tesco and Google?

They both benefit from getting bigger to gather more data about their customers.

But for Tesco to reach more customers, the company must scale itself up - more stores, more staff - becoming a more complex firm to manage. The knowledge problem and diseconomies of scale set in.

For Google's core business of search, it can scale itself to more customers without becoming itself a bigger, more complex organisation.


"But for Tesco to reach more customers, the company must scale itself up - more stores, more staff"

Tesco can actually utilize Google in this respect and reach more customers by using web based technology and gather more data from online questionnaires and the like. This increasing knowledge won't change the fundamentals though will it?

They can also get into an up and coming shopping phenomenon called internet shopping.

If they don't I bet their competitors will. Soon they will all have perfect knowledge but that knowledge will not be held in the brain of some overpaid CEO but in the computer database which generates useful information.

Not that any of this will change the fundamental basis of capitalist exploitation we should understand!

Hayek and Mises seem to be missing one important aspect of the market at its freest, namely that it is a total disaster and results in so much social chaos that only deep and colossal amounts of conscious human intervention can stabilize it. So there is only one thing worse than conscious planning, and that is a lack of it.

The nature of capitalism, as private property, means that it can never deliver us fully from the iniquities and failures of the market, which is why the apologist of capitalism would have us believe that whatever capitalism is on an given day is the best that humans can ever aspire to.

From Arse To Elbow

@Steven, the comparison between Tesco and Google is revealing.

As gastro george noted, the driving force in retail is selling customers to suppliers. This was the dynamic behind the introduction of loyalty cards in the 90s and Tesco's notorious charges to suppliers. Their problem was that they were simultaneously persisting with an older and more expensive form of information-gathering, i.e. opening up new superstores, while the competition were diverting their capital to online shopping.

Though home deliveries and click-n-collect are more expensive for supermarkets (estimated to be about £15 a load), the quid pro quo is a huge improvement in customer data and thus the potential for greater lock-in. The recent decline in loyalty card schemes, like the replacement of superstores with "dark stores", shows the extent to which the lock-in dynamic is shifting from loyalty (i.e. customer familiarity with a retailer) to expectation (i.e. retailer knowledge of a customer).

The cost of retail logistics continues to fall, as does the cost of acquiring customer data. In parallel, the market for goods (i.e. the global middle-class) continues to expand. For this reason, the future actually looks as rosy for Tesco as for Google. Both are data mediators. Tesco's recent mistake (perhaps polluted by UK ideology) was to think it was also a property magnate.


Of course, libertarian socialists like Bakunin and Kropotkin had already flagged up the problems with centralised forms of socialism long before Hayek. Kropotkin, for example, was stressing the importance of local knowledge in "The Conquest of Bread" published in 1892 (in French -- based on articles from the later 1880s).

For those with a market focus, Proudhon in the 1840s had already criticised the likes of Louis Blanc and his centralised, non-competitive socialism by stressing the only competition could produce accurate values and prices. And not to mention that threat to liberty of such centralised systems. Proudhon argued for market socialism -- a federated system of worker controlled associations which would sell the product of their (collective) labour.

So in terms of knowledge, information, etc. -- von Hayek and von Mises were at best repeating debates which had already taken place within the left (but linking them to the right and genuflection to the capitalist elite).

The notion that a centralised version of socialism was possible or desirable had already been questioned by anarchists (libertarian socialists) since before Marx become a Marxist. But, like the term libertarian itself, the right try and present themselves as having invented it...


"In this sense, Hayek's message has shifted. Whereas it used to support dominant western institutions, it now undermines them."

More like a full circle. At the time of his first brush with fame, the planned economy was the dominant idea and Hayek was dismissed by contemporary mainstream thinkers - the Road to Serfdom is practically the classic cry from the wilderness. Then decades later, a brief currency under Thatcher/Reagan, but only because they were dismantling the relics of the previous wave of planning. Now (as you say) belief in omniscience is back firmly in fashion.

Marketeers get seduced by the idea that the market itself will be smart, and will take care of us; that it knows what we want, not just because it is made of us, but because it somehow transcends our petty impulses and does the information processing (the thinking!) for us.

The truth is that while it does act as a process that handles incomprehensible amounts of information, and it does transcend us as it encompasses us, even so it does NOT know what we want. It's much more like the weather than we'd prefer. Or like natural selection: under no obligation to produce more pretty butterflies rather than poisonous worms.

I don't think its a coincidence that the classic "I, Pencil" ultimately resolves to a call to faith in God... "limited rationality", indeed.

Steven Clarke


"Tesco's recent mistake (perhaps polluted by UK ideology) was to think it was also a property magnate."

Perhaps that's partly it. But just as IBM didn't dominate home computing, Microsoft mobile devices, Kodak digital cameras - it's hardest for the dominant player to adapt when the ground shifts under it.

The comments to this entry are closed.

Why S&M?

Blog powered by Typepad