5G Auctions? Costa Rica and Brazil

5G Auctions? Costa Rica and Brazil

I have been amazed by the results of 5G spectrum auctions in Latin America because these processes were designed with significant room for improvement. It's essential to deeply understand the economic context of the sector and develop a public policy that not only aims to close the digital divide but also ensures the sustainability of the sector and promotes economic growth in the short, medium, and long term — clearly oriented toward transforming economies based on raw materials into ones that take full advantage of the opportunities offered by the digital economy.

In recent days, I wrote a LinkedIn post explaining why the 5G auction process in Costa Rica, designed by SUTEL, had a high probability of creating conditions that could facilitate effects similar to collusion among participants. During a webinar organized by 5G América and SUTEL, I listened to the explanation provided by the head of the auction, which sparked my interest regarding several key points in the auction’s design.

Let me now get straight to the point, focusing on phase 1 of the auction in which Claro and Liberty Costa Rica participated. SUTEL explained that prior to designing the auction, it consulted mobile operators about their interest in participating and the amount of spectrum required in each frequency band. Therefore, SUTEL was fully aware—quantitatively—that the amount of spectrum available for auction exceeded the total requested by both operators. Additionally, the third mobile operator, *ICE*, did not participate due to a decision made by the regulator itself.

Furthermore, SUTEL acknowledged that it knew in advance that the auction process would not generate any real bidding or competition between Claro and Liberty. Both companies were equally aware that sufficient spectrum was available for each of them, meaning there was no incentive for either to outbid the other—they simply didn’t need to compete.

Let’s recall: auctions are processes designed for markets where there is actual competition or at least where the number of bidders exceeds or equals the number of spectrum blocks being allocated. The purpose of an auction is to establish conditions under which bidders submit offers so that these offers compete with one another, with the winner being the one whose bid surpasses the set conditions—such as offering to pay more than the base price or submitting the highest number of Access Infrastructure Units (UIA), as defined by SUTEL.

Spectrum auctions are a model widely used by many countries to assign spectrum since the early days of mobile technologies—even recognized by Nobel Prize-winning economists. This model also acknowledges the possibility that spectrum may not attract interest from participants, which is why auction terms often include provisions allowing some portions of spectrum to remain unassigned. That is, when the number of bidders equals the number of spectrums blocks available—a situation that indeed occurred in the auction conducted during the Licitación Mayor organized by SUTEL. Although not explicitly stated in the auction rules, the process effectively turned into a “desert” auction.

This outcome should have been easily predictable by SUTEL. So why didn’t they implement additional measures sufficient to foster a truly competitive process? Or why did they choose to use an auction system when they already knew the necessary conditions for competition weren’t present?

The result was that neither Claro nor Liberty had to compete. Instead, both companies will deploy the same number of Access Infrastructure Units (UIAs), with the investment falling below the actual value of the spectrum.

Technically speaking, although collusion involves bidders agreeing in advance to coordinate outcomes in a competitive process, SUTEL’s auction design facilitated a result resembling collusive behavior. Even though we can assume Claro and Liberty had no need to coordinate—the process itself lacked competitive incentives—SUTEL failed to establish conditions that would promote the expected level of competition in an auction, which relies on bidders competing against each other.

Now, both Claro and Liberty must install 305 UIAs in the 700 MHz band, which corresponds to 90% of the spectrum value, equivalent to USD 28,736,793. In other words, they will invest USD 98,218.99 per UIA. For the 2500 MHz band, each company will install 293 UIAs valued at USD 27,579,150, or USD 94,126.79 per UIA. In the 3500 MHz band, each committed to installing 916 UIAs worth USD 86,482,764, equating to USD 94,413.49 per UIA. And in the high bands, each will install 38 UIAs valued at USD 3,541,023, or USD 93,184.81 per UIA.

Another inconsistent element, both technically and economically, is that the valuation of each UIA was nearly identical across all frequency bands, despite the fact that equipment costs vary significantly depending on the frequency used. Particularly in the 26/28 GHz bands, more base stations are required to cover the same area—like a stadium or concert hall—compared to lower bands like 700 MHz, 2500 MHz, or 3500 MHz. Also, infrastructure setup costs—such as site development, electricity consumption, and rental per square meter—are higher. For example, antennas used in the 26/28 GHz bands are much smaller and lighter than those in lower frequency bands.

Therefore, there’s a high probability that the number of UIAs was miscalculated relative to the actual value of the spectrum. Moreover, the numbers may be insufficient—particularly in the 2500 MHz and 3500 MHz bands—to cover the 354 additional districts where SUTEL expects to improve service quality. For instance, 293 UIAs in the 2500 MHz band would provide coverage equivalent to just 17–27 UIAs in the 700 MHz band, depending on terrain ruggedness, whether coverage is indoor or outdoor, and the presence of buildings or obstacles. Similarly, in the 3500 MHz band, both Claro and Liberty must deploy 916 UIAs, which equate to roughly 49–83 UIAs in the 700 MHz band.

In a well-designed auction, the process should have included additional clock rounds, during which bidders would incrementally offer to deploy more UIAs. This would reduce the unit cost per UIA until its value reflected 90% of the spectrum’s true worth. Only then could the spectrum’s assigned value align with reality—likely varying significantly by frequency band.

As I’ve made clear, SUTEL did not properly assess market conditions or select an appropriate method for allocating access frequencies. As a result, Costa Rica has not received the real value of the spectrum, neither through cash payments nor through infrastructure deployment that reflects the assigned value per band.

Costa Rica is not the only country to experience this distortion in 5G spectrum allocation. As SUTEL noted, the process followed was modeled after Brazil’s approach, implemented by ANATEL. In Brazil, the auction also yielded similar results: the phase designated for national operators saw no real bidding, and the number of infrastructure units required to be deployed fell far below the spectrum’s actual value.

By June 2024, ANATEL reported the installation of 23,098 radio base stations—an insufficient number for a country the size of Brazil. By comparison, South Korea—a country with fewer people and less territory—has deployed over 170,000 5G base stations in the 3500 MHz band alone. China announced it had installed over 600,000 new base stations per month as of July 2023, while India reached 300,000 base stations.

In my opinion, both Brazil and Costa Rica have adopted a flawed spectrum valuation model and auction design unsuitable for their telecommunications markets. Brazil, in particular, faces a major challenge deploying 5G nationwide using only mid-band frequencies while operating in a market where average revenue per user (ARPU) does not exceed USD 6.

Finally, there are ways to design an auction process that is truly competitive or at least ensures that bids exceed the base price. For example, in many auctions, it is common to establish that the opening bid — the first offer — must exceed the base value by a certain percentage. This would have allowed Claro and Liberty, in the case of Costa Rica, to bid for a greater number of Access Infrastructure Units (UIAs) than initially defined, even without having to compete directly with each other.

We also presented another approach during the meeting organized by 5G América and SUTEL — a method to foster competition in an auction process where the number of bidders is equal to or fewer than the spectrum blocks being offered in each frequency band.

Our SpectrumInvest services have been developed taking into account all the elements involved in deploying a technology like 5G in markets as challenging as those in Latin America — from a financial perspective, coverage requirements, and other key factors. If you'd like to learn more about our SpectrumInvest services, please contact us:

📧 contact@artifexconsulting-us.net