Algorithmic price fixing

by Henry Farrell on January 9, 2017

This FT article is pretty interesting:

The classic example of industrial-era price fixing dates back to a series of dinners hosted amid the 1907 financial panic by Elbert Gary, then chairman of US Steel. In a narrow first-floor ballroom at New York’s Waldorf Astoria Hotel, men controlling 90 per cent of the nation’s steel output revealed to each other their respective wage rates, prices and “all information concerning their business”, one attendee recalled. Gary’s aim was to stabilise falling prices. The government later sued, saying that the dinner talks — the first of several over a four-year period — showed that US Steel was an illegal monopoly.

Algorithms render obsolete the need for such face-to-face plotting. Pricing tools scour the internet for competitors’ prices, prowl proprietary databases for relevant historical demand data, analyse digitised information and arrive at pricing solutions within milliseconds — far faster than any flesh-and-blood merchant could. That should, in theory, result in lower prices and wider consumer choice. Algorithms raise antitrust concerns only in certain circumstances, such as when they are designed explicitly to facilitate collusion or parallel pricing moves by competitors.

… a German software application that tracks petrol-pump prices. Preliminary results suggest that the app discourages price-cutting by retailers, keeping prices higher than they otherwise would have been. As the algorithm instantly detects a petrol station price cut, allowing competitors to match the new price before consumers can shift to the discounter, there is no incentive for any vendor to cut in the first place.

“Algorithms are sharing information so quickly that consumers are not aware of the competition,” says Mr Stucke. “Two gas stations that are across the street from each other are already familiar with this.” This episode suggests that the availability of perfect information, a hallmark of free market theory, might harm rather than empower consumers. If the concern is borne out, a central assumption of the digital economy — that technology lowers prices and expands choices — could be upended.

The argument here, if it is right, is twofold. One – that even without direct collusion, firms’ best strategy may be to act as if they are colluding by maintaining higher prices. Firms have a much weaker temptation to ‘defect’ from an entirely implicit bargain by lowering their prices so as to attract more customers, since there are unlikely to be significant gains from so doing, even in the short run. The plausible equilibrium is something that might be described as distributed oligopoly. Harrison White once defined a market as being a “tangible clique of producing firms, observing each other in the context of an aggregate set of buyers.” With super-cheap information, it doesn’t have to be a clique any more to be tangible.

The second is that where there is direct collusion, the information burden on regulators is much higher. For example, one may plausibly imagine that oligopoly-type outcomes might emerge as a second-order outcome of the aggregated behavior of automated agents. One might also imagine that it might be possible artfully to tweak these agents’ behavior in such a way that this will indeed be the most likely result. However, proving ex post that this was indeed the intent will likely at best require a ton of forensic resources, and at worst may be effectively impossible.

NB that both of these can happen entirely independently of traditional arguments about concentration and monopoly/oligopoly – even if Amazon, Google, Facebook, Uber etc suddenly and miraculously disappeared, these kinds of distributed or occulted oligopoly problems would be untouched. If you take this set of claims seriously (the evidence presented in the FT piece still looks tentative tentative), then the most fundamental problem that the Internet poses is not one of network advantage, increasing returns to scale and so on advantaging big players (since, with a non-supine anti-trust authority, these could in principle be addressed). It’s the problem of how radically cheaper communication makes new forms of implicit and explicit collusion possible at scale, squeezing consumers.



divelly 01.09.17 at 4:52 pm

Buy less stuff!


bob mcmanus 01.09.17 at 6:09 pm

1) Cue Wilder and “administered prices”

2) Firms are always price-makers. All firms. Relative and marginal Costs are about the distribution of profits.

3) Anwar Shaikh, Capitalism 2016


Michael 01.09.17 at 7:05 pm

Fascinating. It’s worth recalling in this regard the old argument from Friedman, according to which monopolies may be acceptable in free markets because monopolists have incentive to behave “as if” there is competition (since overcharging may create openings for new competitors).


Collin Street 01.09.17 at 8:13 pm

As I’ve pointed out a couple of times… profitability depends on the exclusion of competition. The consequences of this now that we can actually get “fully competitive markets” with fully flexible prices and effectively-unlimited numbers of participants… this isn’t something we’ve had to really deal with before on this scale.

I mean, what’s the alternative? If the firms used the same technology to “compete” they’d bankrupt themselves over the medium term.

Our approach to market competition is fundamentally misguided, shaped by intuitions that only worked when geography and communications imposed competitive barriers we could ignore, let businesses be profitable without our overtly protecting them from competition. We can’t do that any more. If we want businesses to be able to deliver a profit, if we want private-sector investment to be a thing, we’re gonna have to take some of the barriers-to-competition that modern communications etc eliminated and engineer them back in deliberately.

We can’t enforce anti-competition laws, not as written. They’ll destroy the economy. But the problem they were designed to fix still exists: we need a new approach.


Jeff R. 01.09.17 at 8:43 pm

I frequently see pairs of gas stations across the street from each other with consistent, significant price differences. (And not just in places like downtown San Francisco where you have to drive seven blocks to make a left turn, but in places where there’s a signalled light right there.)


Peter Dorman 01.09.17 at 8:56 pm

I haven’t read the original FT article, but color me skeptical. (1) What determines the price at which these algorithms arrive? Is it simply a question of reinforcing whatever the status quo is? (2) The algorithms, by automating mimetic pricing behavior, make it more difficult for participants and potential entrants to infer competitors’ costs from their prices. That could lead to a lot of repricing as a form of discovery. Drawing inferences from this requires some formal modeling, I think.

It is true, however, that in repeated collective action problems, the shorter the timespan associated with each stage game the lower the imputed discount factor and therefore the greater likelihood of a cooperative solution.


bruce wilder 01.09.17 at 11:04 pm

Firms are not always price-makers; some firms in a system may be quite passive. Most U.S. gas stations are convenience store operations and I’m guessing that their gas is priced and sold as a pass-thru. The convenience store wants the price to be “competitive” because that’s what brings in people to buy sugar and caffeine, but they do not want to be drawn into subsidizing the gas price, so most (but not all) will tie their own hands.

Traditional arguments about concentration and monopoly/oligopoly were always a lot of handwaving rubbish and untested storytelling. Defining a “market” where none existed, as Harrison White did, was never a path of enlightenment, as it overlooked the needs and necessities of strategically organizing and managing a technically and socially complex and dynamic production and distribution process.

This link might be of some interest tangential to the OP:


bruce wilder 01.09.17 at 11:29 pm

Collin Street @ 4


We do need a new approach. Or, an old approach.

The structure of American banking and finance during its long Frank Capra period might be a good example: A structure of many diverse and mostly small institutions, drawing on minor sources of economic rent for structural stability, such as prohibitions on interstate banking or the slightly higher interest rates savings and loans could pay on deposits, as well as general prohibitions on common strategic direction of commercial banks, investment banks, insurance companies, credit unions and stock brokers. Not to mention usury laws and state audits of commercial bank lending and public institutions like Fannie Mae (when it was dedicated to a public purpose).

A very complex structure and not a laissez faire blob — it represents a different way of thinking about economic structure. The idea of a self-regulating market economy that can be relied on to deliver the equilibrium best of all possible worlds is basically a seductive lie, that undergirds a lot of patently ineffective or futile ideas about economic policy toward economic structure.


mjfgates 01.10.17 at 3:09 am

Jeff R@5: On a multilane road with a lot of commuters, the station on this side of the road gets all the business in the morning and the station on that side gets all the business in the evening. I have seen a pair of gas stations that basically swapped prices at two in the afternoon and then again at some ungodly hour of the morning, because people are willing to pay an extra nickel per gallon to avoid turning across traffic twice.


drouse 01.10.17 at 6:16 am

The very low cost of information cuts all ways. It’s easy to determine buying behavior and if an algorithm decides that if a certain population don’t aggressively pursue price or leave a particular ecosystem to buy, special pricing could be applied. I’ve heard rumors about this concerning Amazon.


George McKee 01.10.17 at 6:20 am

I can’t read the paywalled FT article, but I’ve been noticing the quoted behavior in airline ticket prices for a long time. You used to regularly see news stories about some airline or other making a move to raise prices on one route or another, speculating on whether other airlines would follow along, and whether the price move would spread to other routes. No back-room plotting was implied or needed.

I developed a first cut of a general hypothesis about the existence conditions for this kind of “virtual cartel” behavior, based on the ability of “competing” firms to evaluate price changes and reset their own prices more rapidly than customers could discover those changes and respond to them with actual purchasing behavior.

This doesn’t seem to happen as much nowdays, though. I attribute the change to improvements in customer ability to find and react to pricing changes using search tools like you get with Expedia and Carlson Wagonlit Travel, and to the emergence of discount airlines like JetBlue who always “defect”, and don’t play the cartel game, making it impossible to obtain a coordinated price rise. The emergence of monopolistic single-airline hubs probably has an effect, too. When your choices of routes between Wichita and San Francisco involve stops in Denver(United), Dallas(American), and Atlanta(Delta), there’s no real need for those airlines to worry about more sophisticated anti-competitive measures.

I never found time to simulate the situation or develop any math to determine the exact parameters for communication bandwidth, search efficiency, maximum number of participating firms, etc. that would lead to sustained above-minimum prices.

The forensic resources that would be needed to obtain convincing evidence of “echoing” price changes would be akin to those needed to detect front-running and other stock market shenanigans – regulators would need to track every price on every item from every vendor, and run big data analytics on those records. Certainly feasible, but not inexpensive.


ZM 01.10.17 at 8:13 am

“The argument here, if it is right, is twofold. One – that even without direct collusion, firms’ best strategy may be to act as if they are colluding by maintaining higher prices. ”

Slightly off topic, but I wonder if this sort of thing has also impacted the media over the last 20 years with the rise of the internet.

Editors and journalists can all look at the other newspapers and journalists work, and see what news they are printing, and what news they are not printing.

The logic of media ownership rules was meant to prevent this — but if all the media are in effect colluding over what news and opinions to print, and what not to print, then the media ownership rules become more and more redundant.


MisterMr 01.10.17 at 9:45 am

It seems to me that this has nothing to do with algorithms, and everything to do with the determination of the profit share.

Given a country with a total production of 100, workers will eat a certain percentage of the products, let’s say 70%, and “capitalists” will eat the remaining 30% (abstracting from other social categories).
But what determines that capitalists get 30% and not 40% or 20%?
Orthodox theory says that in a competitive market prices will fall towards their cost of production, but in these costs of production is included the “opportunity cost” of the use of capital, and this opportunity cost is just the expected “normal” profit from capital, so that all the orthodox theory says is that all capitals will converge towards the same rate of profit, but it doesn’t say what this rate of profit (or the corresonding profit share) is.

In other words we know that competition will push all capitalists to make 20%, or all capitalists to make 40%, or all capitalists to make 30% (adjusted for the different levels of capital intensity), but we don’t know exactly what level the profits will converge to, nor why (this is the essence of Sraffa’s “neoricardian economics”).

This determination of the wage share (or profit share) happens at 3 different points:
1) at the moment of employment, the worker will demand a certain wage from the capitalist, depending on his/her bargaining power; people who speak of falling union power or of a weak job market speak of this point.
2) but on the other hand we can see the profit share as the markup businesses do on their live production costs; people who stress competitons and/or rising monopolies stress this different point.
It is worth noting that points 1 and 2 refer to totally different causation mechanisms, but in pratice the effects cannot be distinguished, so that we can’t actually distinguish between the effects of “oligopoly” and the effects of “bad job market”.

3) there are also other factors that redistribute income, like taxations or various forms of rents, and in my opinion these things can have a retroactive effect on distribution (e.g. if the marginal income tax falls rich people have more money that they invest in housing, the cost of houses goes up, and this increases rents, thus influencing before-tax income).

Since the supposed effects of competition, oligopoly etc. are in my view undistinguishable from other effects that influence the wage share, I’m very sceptical about about explanation of the falling wage share (or increased profit share) based on a lack of competitivity (why would our economy be less competitive than the economy of the ’60s)?


Trader Joe 01.10.17 at 12:55 pm

I’d agree that collusion via algorithm is possible, but all evidence points to the contrary. Retail margins are falling for virtually all major retailers in the US and UK (can’t say broader than that, but would be surprised).

At least so far, the ability to price discover has been a more valuable tool for consumers in arbitraging to find a deal than it has been for sellers to prevent price erosion.

Equally, fuel selling is a massively competitive market, its all but impossible to get all actors to follow the script. For many outlets, fuel prices at convenience stores are routinely set as a derivative of cost, not as a function of market. Large scale franchises often control the prices centrally – not locally – so are set without knowledge of the ‘station across the street’ and designed to insure that no-one can accuse of fuel being called a loss leader. Beyond that, margins are razor thin on fuel. Most of the money is made in the store, not at the pump.


reason 01.10.17 at 3:49 pm

You’ve got to be kidding. The Guardian is the same as the Daily Mail?


divelly 01.10.17 at 4:09 pm

re 5. A premium on brand loyalty,as well?
re 8. If I get 5% on a passbook acct., won’t I be charged 8% for a mortgage?


stevenjohnson 01.10.17 at 5:15 pm

It’s been a long time since I had access to a university library. And in most of that I didn’t have time to skim the shelves and randomly find things like G.C. Harcourt Some Cambridge Controversies in the Theory of Capital (or Human Action or Wilkinson’s Poverty of Progress.) But the last time I looked there wasn’t really a generally accepted theory of the level of profit. Whatever agreement there was seemed to be that profit was both exogenously determined by unspecified forces and endogenously determined by consumer preference. That would be nonsense were it not so convenient. Presumably then my impression is my mistake. What is more or less the current orthodox thinking on the general rate of profit and dynamics in changes thereof?


bruce wilder 01.10.17 at 6:57 pm

stevenjohnson @ 17


Collin Street 01.10.17 at 7:14 pm

Orthodox theory says that in a competitive market prices will fall towards their cost of production,

Marginal cost of production. This is less than the actual cost of production. Competitive industries are unprofitable.


chris s 01.10.17 at 8:37 pm

Isn’t just this a specific example of the general tactic of price signalling in markets dominated by oligarchies (which essentially petrol stations are at a local level).


anymouse 01.10.17 at 9:56 pm


Because people only travel one way to work? Gas stations on the same side of road can have significant price differences for very long periods of time.

Also you really do not need an app to tell you the price of gas at the station across the street. When they change the price they post it on big sign for everyone to see.

Basically we should be extremely skeptical of these claims.


hix 01.11.17 at 4:55 am

The app was designed/legislated to increase competition by the way.


hix 01.11.17 at 4:59 am

“special pricing could be applied. I’ve heard rumors about this concerning Amazon.”

Very likely. Whats certain is that many internet retails charge higher prices on the mobile site and/or in particular if one uses a more expensive phone/tablet.


MisterMr 01.12.17 at 10:19 am

@Collin Street 19

“Competitive industries are unprofitable”

If we assume that orthodox theory is correct, and competition draws prices towards the marginal cost of production. But obviously this doesn’t happen, so we have two choiches:

a) There isn’t enough competition;
b) Competition doesn’t actually drive prices towards the marginal cost of production, even if we had “perfect” competition, and the orthodox theory is wrong.

My opinion is that “b” is true but, since it is very weird from our normal point of view (influenced by “orthodox” theory), we tend to believe that what is actually happening is “a”.

My point is that, when we speat of “the economy” as a whole, the supposed effects of

a) low competition between businesses (rents, monopolies etc.)
b) bad conditions for workers (high unemployment, low protections from firing etc.)
c) indirect redistribution such as taxes

all end up in a low wage share, so we cannot actually distinguish them and, in fact, maybe we are speaking about the same thing from different points of view.


Z 01.13.17 at 1:36 pm

I don’t quite understand the specifics of the article (for lack of technical expertise) but it occurred to me recently that Hayek’s critical cognitive argument in favor of markets-according to which markets ensure a distillation of complex information through the price mechanism-was somewhat weakened by the fact that markets are populated by agents (namely, firms) with a direct interest in distorting the information available (as explained in a particular case in the article, but if we are talking about new technologies see also ranking and rating algorithms on gatekeepers site like Amazon or Yelp) and so with a direct interest in the termination of the cognitive mechanism that made markets valuable in the first place.

If this is correct, the main weakness of market-based systems compared to democratically supervised ones (for instance in the context of the relevance of privatizing public services) might somewhat ironically be described in purely Hayekan terms: in the middle-term, there is a strong presumption that agents in the former will cave in to the incentives to produce faulty information.


notsneaky 01.14.17 at 8:42 am

This is really just the old “we will meet any competitors price” coupons… except on the internet. And these have been subject to anti-trust scrutiny for awhile. Just a new medium.

19 – *Marginal cost of production. This is less than the actual cost of production. Competitive industries are unprofitable.*

Huh?? It could go either way. Theoretically, in the long run, it’s the same as the actual cost of production (marginal cost = average cost). Not saying this is true, just that this is what the theory says.


hix 01.14.17 at 10:37 am

Does anybody know about the legalilty of inidividual algorithm based pricing in major jurisdictions?

Comments on this entry are closed.