How petrol retailers coordinate prices to increase profits

A study of seven years of petrol pricing data from Melbourne has uncovered how the country's largest fuel retailers used a private information-sharing platform to negotiate coordinated pricing arrangements, lifting market profit margins by about 30% over the period studied.

The research, presented by University of Melbourne Economics Professor David Byrne at the 4th Firm and Industry Dynamics Workshop, jointly hosted by UNSW Sydney and the University of Hong Kong, drew on station-level pricing data bought from Informed Sources, a platform that let six major retailers, including BP, Caltex, Coles and Woolworths, monitor each other's prices at every station across the country in near real time.

Prof. Byrne told the workshop that the platform gave the retailers what he called "this officially rich communication media that the companies can actually negotiate over and select among profit-enhancing coordinated pricing structures." He said it took the firms roughly five years to settle on a stable pricing arrangement, a transition marked by repeated price wars, staggered timing disputes and eventual coordination among the major players.

Data shows a shift from a weekly cycle to staggered pricing

The study covered the Melbourne retail petrol market between 2007 and 2013, drawing on daily price records for every station operated by the six major retailers plus hourly data for a critical period, allowing the researchers to identify the exact hour each retailer adjusted prices.

At the start of the sample, Prof. Byrne said, the market ran on a rigid weekly cycle. At the start of the sample, retailers raised prices on Wednesdays every week, a pattern that had held steady since 2005, according to Prof. Byrne. By the end of the sample, that weekly routine had broken down, replaced by a longer and less predictable cycle in which a small group of stations moved first, with the rest of the market following over several days. These are the cycles many Australians in major cities face today.

Learn more: When AI knows what you'll pay: The rise of personalised pricing

Price cost margins at the top of the cycle started at around 10 to 15 cents a litre early in the sample, then rose gradually over time. Margins at the bottom, which started near zero, climbed to almost five cents a litre by the middle of the period before falling back to zero by the end. Prof. Byrne said the shift in overall structure had lifted volume-weighted price cost margins by roughly 30 per cent relative to the start of the sample – an effect he said was comparable to documented cases of algorithm-assisted collusion in other countries.

Retailers used delay as a bargaining tool

Central to the findings was the way the four major retailers negotiated over the timing of price rises, rather than through direct communication. Coles and BP, the price leaders, moved first, while smaller retailers, particularly Woolworths, were allowed to delay their own price increases by a matter of hours in exchange for going along with the higher price. That timing gap effectively served as an unspoken payment, allowing the smaller player to capture extra market share before falling in line.

Prof. Byrne said the arrangement mirrored tactics seen in explicit cartels overseas, where retailers had used phone calls to divide up gains from coordination: "When they phone each other in Quebec, for example, the median number of phone calls when they would do these jumps, up to 65 phone calls within a day for all of them to coordinate a price increase."

"When you have these data-driven algorithms, and firms maybe not with algorithms, but really rich information sharing and potentially able to get to new equilibria, that poses a potential issue"

DAVID BYRNE

In Australia's own petrol cartels of the 1990s, he said, coordination required 40-45 calls. The Melbourne retailers achieved similar outcomes without speaking to each other at all, relying purely on the pricing data shared through the platform.

When the delay granted to the smaller player grew too large, the study found, the dominant retailer would respond by collapsing the cycle altogether, in one case pricing below wholesale cost for four consecutive weeks. Prof. Byrne said the researchers interpreted this as more than punishment: "It's not just that there's punishment, that the prices do double duty when we're trying to communicate on a platform, that I'm sending you a signal that I'm not willing to accept your proposal for two extra hours of delay,” he explained.

Price war spread over 18 months of chaos

The dispute over timing escalated into an extended period of instability, with the day of the week on which prices reset drifting from Wednesday to Thursday, then Friday, Saturday, and eventually Monday, as the smaller retailer kept pushing for more delay and the larger retailer kept withdrawing it. Prof. Byrne described the outcome for consumers during that stretch as unusually favourable: "This is 18 months of chaos. From a consumer point of view, there is a huge price dispersion."

He added that the instability was, in that narrow sense, good for competition, since one retailer stayed cheap while the other stayed expensive, giving shoppers a genuine incentive and opportunity to compare prices.

The retailers eventually resolved the standoff by redesigning the pricing structure itself. Rather than relying on a single fixed day for the whole market to move together, the leading retailer began signalling its intentions with a handful of stations first, then waiting for competitors to match before continuing the increase over several days. This reduced the incentive for smaller retailers to hold out for a better deal, since margins at the bottom of the cycle were kept close to zero regardless of how long they delayed, according to Prof. Byrne, who added that the redesigned structure remained in use in the market today.

Regulation in Perth took a different approach with faster results

The workshop also heard how a separate regulatory intervention in Perth produced a markedly different outcome to the drawn-out Melbourne transition. Western Australian authorities restricted retailers there to setting prices only once a day, a rule intended to make it easier for consumers to shop around. "They only allowed them to do just prices once a day in Perth, thinking that would help people shop because they weren't moving prices around," Prof. Byrne said.

Subscribe for the latest research, analysis and insights from UNSW Business School

He said the Perth market moved to a new pricing equilibrium far more quickly and smoothly than Melbourne did, since the daily limit removed the scope for retailers to bargain over intraday timing advantages, the very mechanism that dragged out the Melbourne dispute for the best part of five years. Prof. Byrne suggested that the contrast raised a caution for regulators designing rules around price transparency and information sharing: measures meant to help consumers can also reshape firms' incentives to coordinate, sometimes in unintended ways.

The research team was also examining the human element behind the data, noting that the pricing decisions studied in Melbourne were made by revenue management staff rather than automated systems. The friction and repeated missteps visible in the data, as staff tried to interpret rivals' intentions solely from price movements, offered a point of comparison for what might happen once pricing decisions are handed over to algorithms, according to Prof. Byrne.

A landmark antitrust action and law changes

The behaviour documented in the study became the subject of a Federal Court case brought by the Australian Competition and Consumer Commission against Informed Sources in 2014, alleging the platform had facilitated anti-competitive conduct. The case was settled in December 2015, with Coles removed from the platform as part of the resolution.

Prof. Byrne noted the finding sat against an earlier government position. A Senate inquiry into record petrol prices concluded that price uniformity among retailers was not evidence of collusion and found that existing law offered no clear path to prosecute coordination reached without direct communication. "It's counterintuitive and irrational to conclude that price uniformity or parallelism is a product of collusion," he said, quoting the inquiry's own findings. Australia later amended its competition law in 2015 to allow regulators to pursue price-fixing understandings reached without explicit communication between firms.

Prof. Byrne said the Melbourne case offered a warning for regulators currently grappling with similar concerns involving data-sharing platforms and algorithmic pricing in the United States' meat, rental, and health insurance sectors, as well as in Europe.

He said existing antitrust frameworks were built on the assumption that natural language was the medium through which firms coordinated, an assumption that no longer applied when firms could reach new pricing equilibria purely through shared data: "When you have these data-driven algorithms, and firms maybe not with algorithms, but really rich information sharing and potentially able to get to new equilibria, that poses a potential issue,” he said.

Similar patterns emerging in other industries globally

Prof. Byrne also noted similar examples from other countries. He pointed to a live case in the United States meat supply chain, where a platform called Agri Stats allows suppliers to share quantities, inventories, and prices with each other, an activity regulators allege has contributed to food price inflation across the country.

"It's counterintuitive and irrational to conclude that price uniformity or parallelism is a product of collusion"

DAVID BYRNE

A separate US case involves RealPage, a platform landlords use to share rent and occupancy data, which feeds into a central algorithm that recommends rent levels. Prof. Byrne said the recommendations were not binding, but research has linked the arrangement to rental prices above levels that competitive markets could explain. He added that a similar pattern had surfaced in the US health insurance sector through a platform called MultiPlan (recently rebranded as Claritev), now under consideration by the Federal Trade Commission and the Department of Justice.

These cases, along with reviews launched by the OECD and European regulators into information sharing among competitors, point to a broader question regulators worldwide are grappling with: whether laws written for a world of verbal agreements can still catch coordination that occurs purely through shared data. Prof. Byrne concluded by noting that the Melbourne case offered regulators a useful, if costly, lesson in what to look for.