Satellite Auctions - How can we enhance the scalability of the Revenue Share Model?
Auctions for Satellite Spectrum may not be the optimal solution, but a modified % AGR model could serve as a viable alternative if an auction becomes the sole option.
Auctioning Spectrum for Satellite is NOT a good idea. No model will work without artificially limiting the number of prospective winners - leading to wastage of spectrum. Why? The reason is that the Satellite Operators aren’t capable of using spectrum efficiently (compared to terrestrial players) due to a limited number of satellites visible over the geography at any time vs. lakhs of Base Stations installed terrestrially. This is why satellite spectrum can’t be auctioned to players for “exclusive use” as we do for the terrestrial spectrum, and therefore, most advanced economies of the world assign spectrum for satellite services administratively.
India might want to follow a different approach, given the past issues that it has faced when assigning 2G spectrum administratively - leading to a scam that became problematic for the government. Hence, TRAI is analyzing different models for auctioning satellite spectrum. One of the possible models being discussed is the auctioning of satellite spectrum using % revenue share/annum as a metric.
I had analyzed this model in the past and found many issues. The most important ones are - 1) the opportunity of arbitrage; and 2) the need to treat all spectrum bands at par. The former I have discussed in my article titled - Why is auctioning satellite spectrum based on a % of AGR a suboptimal auction design?
The purpose of this note is to suggest an improved revenue share model that is robust enough to prevent the opportunity of arbitrage and also turn the later issue (No.2) into an advantage by increasing the regulator’s flexibility in using the bands that are targeted for auctions for rolling advanced wireless technologies in the future like 6G.
The Basic Principles
Before we dive into details, let’s first spend some time discussing the basic principles.
Principle 1 - Since the intent is to auction spectrum and NOT license, any model that we plan to use, should reflect that intent. In other words, the price the bidder pays (% Revenue share) should be proportionate to the quantum of spectrum held by him.
Principle 2 - The model should scale (without the possibility of arbitrage) for all auctions held in the future in the same sets of bands or the new ones.
Principle 3 - The model should be fair for both kinds of bidders - existing and new, and should NOT create any disadvantages to either.
Principle 4 - The model should be extremely simple to administer without the need for using complex formulas.
Principle 5 - The model should allow enough flexibility for the regulator to intervene to resolve conflict arising on account of different approaches towards sharing (similar to the practice followed by the FCC in the US) and also enable the refarming of any spectrum bands in the future, in case the band becomes a mainstream band for deploying 6G.
The Modified % Revenue Share Model
The modified % Rev Share Model is described below.
First - The reserve price should be fixed as % AGR/MHz and NOT just % AGR (Adjusted Gross Revenue). The benefits of doing so will be evident as we discuss the model in detail in this note.
Second - All bidders should be allowed to take spectrum overlapping with each other at the auction price but with clearly defined sharing rules in advance. Also, the regulator should hold the power to refine these rules in the future if needed to prevent any player from facing any disadvantage.
Third - No bidder should be allowed to bid for a spectrum less than a minimum quantum, with a clearly defined yearly outflow to be paid in cash, OR actual % Revenue Share, whichever is higher. This payout should be proportional to the quantum of spectrum held by the player.
Finally, DOT should retain the flexibility to move satellite players into different bands if needed some time in the future. This is a must to protect the larger interest of the country as some of the targetted bands might end up becoming mainstream bands for 6G. This the DOT can easily do as the auction design is aimed at treating all bands at par for all rounds of the auction held even in the future. This will be in alignment with the reference made by FCC in its consultation (July 2023) is reproduced below for ready reference.
“…Additionally, SpaceX's “authorization is subject to modification to bring it into conformance with any rules or policies adopted by the Commission in the future.”
How will the bidding work?
The DoT should announce in the NIA the total number of bidders who will be declared winners in the auction. Let’s say this number is restricted to 4 in phase 1. Assume that the number of bidders registered for bidding is 7. The auctioneers will start with the reserve price of let’s say 0.0015 % /MHz (DoT has to decide this number). If the number of bidders who agree to this price is greater than 4 then the auction proceeds to the next round with a price increment of let’s say 10%. As soon as the number of bids becomes equal to 4 the auction ends. This is described in the following table.
The % AGR/MHz discovered in the auction above is 0.00292. The final quantum of % AGR to be paid by the bidders will depend upon the quantum of spectrum they end up demanding from the DoT. While they do so the winners will also give their bands of preference. As an example, this is listed in the table below.
How does the Model Scale?
Let’s assume another auction is held after a few years. Such an auction will include the existing players as well as a new set of players. Also, assume in this auction winners 1 and 2 of the current auction also bid and win. Additionally they and demand spectrum of 500 & 1000 MHz each. Let’s say the winning price in the next auction is less than the current auction at 0.00182 %/MHz.
Therefore, the total revenue share for Winner1 will now become = 500*0.00182 (Auction2) + 4.38 (Auction1) = 5.29%. Similarly for the winner 2 the new value will be = 1000*0.00182 (Auction2) + 5.84 (Auction1) = 7.66%
Now reason this model scales (and the one using weighted average aggregration does not) is because even though the bidders have chosen to bid different quantums of spectrum, the final % AGR of the bidders is just a linear addition of the two % AGR numbers emanating individually from the auctions held separately.
Key Benefits of this Model
Firstly, the modified revenue share model scales without opening the possibility of arbitrage. In other words, it prevents the bidders from pulling down their Total Revenue Share by seeking a higher quantum of spectrum in a future auction at a lower price (% Revenue Share).
Secondly, the model motivates the winners to use spectrum efficiently and prevents them from demanding a huge quantum of spectrum unless they absolutely need it.
Thirdly, it empowers the regulator with full flexibility to move operators to new bands (if required) in case the current band becomes the mainstream band for 6G (For context refer to my earlier article titled - India & US 6G Alliance).
Finally, the model is simple to administer and empowers the regulator to intervene in case of disputes in the sharing of the same block of spectrum between two players simultaneously, thereby protecting the interest of the weaker player.
Key Shortcomings of this Model
Firstly - The model treats all spectrum bands at par from the point of view of price, which might not be the case as some spectrum bands might be more valuable than others due to less noise in lower-frequency bands. Note that high-frequency bands are susceptible to more phase noise.
Secondly - The auctioning of spectrum based on % Revenue Share is not the best thing to use as a metric for auction, as revenues are uncertain and can’t be assured.
Thirdly - For the model to work, the revenues emanating from satellite services can’t be mixed with those emanating from the terrestrial without linear adding the component of the revenue share of satellite with that of the terrestrial business - making the interworking unviable (revenues of terrestrial business will be huge). Because of this reason, the model will find it difficult to support flexible use of spectrum, i.e. the same spectrum to be used for both satellite and terrestrial services. Also, we will encounter a similar problem for those operators who choose to use satellite services to backhaul traffic from their terrestrial BTSs located in remote locations, as the revenues emanating from the same will be subject to arbitrage even if the entity decides to provide such service by a separate held company under the same promotors even at arm’s length.
Finally - The model creates artificial restrictions on the number of winners, even when a lot of capacity will lie wasted, as satellite operators will not be able to use assigned spectrum resources fully.
Conclusion
Satellite spectrum is not a finite resource, and therefore auctioning it might not be the most optimal thing to do. But in case the auction is the only choice available, then the Modified Revenue Share Model (% of AGR/MHz) looks to me a lesser evil to choose and might work scale to support future auctions.
But for this model to work (without the possibility of arbitrage) the revenues emanating from satellite services can’t be mixed with those offered terrestrially, without aggregating the revenue shares of both businesses linearly.
In addition, DOT should arm itself with clearly defined powers to intervene to protect the interest of players who might find themselves disadvantaged with respect to sharing spectrum with other players operating in the same block of airwaves.
Also, to protect the interest of the country for using any of the bands for future 6G, the DOT should also be armed with powers to shift satellite players to alternate bands as the FCC is in the process of doing.
After taking all these into consideration, India has to make a conscious decision in the best interest of the country to iron out all constraints (past and future) as much as possible, to truthfully serve the interests of the stakeholders and the consumers.
(Final Note - This model is limited to ONLY user links and the spectrum for gateways will have to be given administratively bundled with the airwaves assigned through the auctions)