Professor Nick de Roos’ research shows how limiting access to competitor prices in an oligopoly triggers undesired effects, such as higher prices and increased profits for rival firms.
Co-authored with researchers1 from University of Melbourne, Clemson University and Duke's Fuqua School of Business, and funded by the Australian Research Council, the investigation titled 'Asymmetric Information Sharing in Oligopoly: A Natural Experiment in Retail Gasoline', analyses how a shift from symmetric to asymmetric information sharing affects oligopoly prices.
Using a natural experiment from an Australian retail gasoline antitrust case, the study confirms price competition softened when one of the five main retailers lost access to high-frequency and detailed price data from rivals.
The research empirically demonstrates the change led to higher prices and an 8-figure annual profit jump for gasoline retailers, effects opposite to what the antitrust agency intended.
The results provide a cautionary tale for antitrust agencies and industrial organisation (IO) economists on the potential unintended consequences of limiting price information sharing among firms.
The Informed Sources service raises concerns about price collusion
In a sector where consumers are very sensitive to price differences between stations, petrol retailers adjust prices across their networks daily to stop drivers from swapping to a more affordable option.
This underlines the value of having access to complete, real-time and reliable rival price data to quickly identify and respond to adjustments made by competitors.
And this is exactly what the price sharing platform Informed Sources (IS) offered: digital access to station-level price data of all subscribed retailers, updated every 15 or 30 minutes.
The dataset is complemented with non-subscriber station-level price data that the company collects manually at daily, weekly or other frequencies.
Historically, IS has been available to the supply side of the market, but not the demand side, meaning only subscribed retailers can gain 24/7 access to competitors’ prices obtained by the platform.
However, this creates a unique symmetric information sharing setting that can facilitate tacit collusion, an anticompetitive practice where rival firms coordinate to rise prices without having to engage in explicit communication.
Due to the limited number of competitors, oligopolies are particularly sensitive to collusion, raising concerns about the potential role of IS data in facilitating price fixing in Melbourne’s retail gasoline market.
Following a two-year investigation into the IS platform, in August 2014, the Australian Competition and Consumer Commission (ACCC) launched a case that put Australia’s five major gasoline retailers in the spotlight.
The ACCC alleged the IS platform, BP, Caltex, Woolworths, Coles (now Reddy Express) and 7-Eleven, had violated section 45 of the 2010 Competition and Consumer Act, which makes illegal “contracts, arrangements or understandings that have the purpose, effect, or likely effect, of substantially lessening competition.”
In short, the ACCC accused fuel retailers of using IS to propose competitors a price increase and monitor their reaction. If, for example, the response was insufficient, they were able to quickly withdraw the proposal and punish those that didn’t accept the price increase.
The shift from symmetric to asymmetric information sharing
The case eventually resolved in December 2015 with two settlements, with which the ACCC expected to increase competition and allow buyers to make better informed purchasing decisions.
In the first settlement, the largest petrol retailer in network size and market share, Coles, ‘went alone’ and agreed to stop subscribing to IS – or similar service – for five years.
In the second one, BP, Caltex, Woolworths and 7-Eleven, were allowed to continue their subscription, but IS was required to make price data available to third parties, including consumer search apps developers and research organisations.
While the case had no significant impact on app-enabled consumer search (due to a lack of popularity and availability of consumer search apps in Australia), on the supply side, IS began manually collecting Coles’ prices using human spotters immediately after it exited the platform.
Spotters collected one daily price observation per station from three-quarters of Coles’ network, meaning the remaining subscribers could still observe their prices.
With Coles ‘unplugged’ from the platform, the firm faced an informational disadvantage with comparatively limited ability to observe and quickly react to price adjustments made by other retailers.
Limiting price information sharing doesn’t always reduce market power
This real-life setting provides a unique opportunity to examine the competitive effects of transitioning from symmetric to asymmetric price information sharing among oligopolists.
Using pre- and post-settlement daily station-level pricing data (May 2015-December 2017), Nick and colleagues undertook a high-frequency event analysis2 that demonstrates settlement’s effects were the opposite of what ACCC expected.
Not having up-to-date information on what other retailers were doing, meant Coles drastically reduced the frequency of price adjustments and became less likely to modify the cost per litre for consumers when nearby rivals did.
But when Coles adjusted prices, these were higher than when it was on the platform, which allowed competitors to respond with price increases, albeit smaller than Coles’.
As a result, price-cost margins raised by 5.9 and 3.4 cents (AUD) per litre for Coles and rival firms respectively, increases similar to those caused by tacit collusion from the same retailers in Perth between 2010-2015.
Although rival firms had to set prices below Coles’, by offering more competitive alternatives (relative to the new scenario) their share of the market expanded by an average 45%, while Coles’ reduced by 33%.
So, how did this translate into profit for Melbourne’s gasoline retailers?
Coles saw a small (4%) increase in average daily profits, which shows the relatively large price increase was roughly offset by its loss in market share.
In stark contrast, rivals experienced a substantial 38–77% settlement-induced growth in average daily profits, reflecting both the increase in prices and market share.
Annually, Nick and colleagues estimate a combined $33 million (AUD) – £19 million – surge in annual profit due to the IS case settlement, which was intended to promote competition.
Lessons for antitrust agencies and IO economists
At the time, policy discourse was aligned with the idea that restricting the sharing of detailed, real-time price data within a tight oligopoly was procompetitive.
What this research suggests and was not well understood, is that creating asymmetries like those from the IS case settlement, can empower uninformed firms to stick to higher (non-competitive) prices for longer.
As price commitment from one player leads to higher prices across the board, trying to limit price data sharing can have the unintended consequence of bolstering market power.
These are timely and important lessons for IO experts and those involved in the development of antitrust legislation.
The investigation provides valuable insights to consider in price-fixing settlement negotiations and creating safe information sharing harbours that promote efficiency while still limiting tacit collusion.
Nick and colleagues are already looking into the latter, with the initial follow-up work showing information sharing structures can be potentially altered – eg having more than one uniformed firm – to disrupt tacit price coordination and generate consumer benefits.
The study also underscores the power of information sharing platforms to mitigate the effects of case remedies that negatively affect their subscribers.
The authors anticipate these issues will continue to be at the forefront of policymaking in increasingly digitised and data-driven industries where information sharing is a key part of the market structure.
1 The study’s co-authors are Professor David Byrne (University of Melbourne), Professor Matthew Lewis (Clemson University), Professor Leslie Marx (Duke's Fuqua School of Business) and Dr Andy Xiaosong Wu (University of Melbourne).
2 A high-frequency event study design involves analysing the impact of a specific event by examining high-frequency data surrounding the event, instead of relying on daily or weekly information. The goal is to capture the immediate and subtle price changes that might be masked by coarser data, allowing for a more nuanced understanding of the event's impact.
 |
Professor of Economics, Co-Director of Research for Economics Group |