Mark Goldberg

Spectrum policy in the race for 5G

Canadian spectrum policy will be taking a high profile this year as the government and industry prepare for the increased bandwidth requirements for 5G deployments.

In December, I wrote “Spectrum policy and auctions”, providing links to the videos and presentations from November’s Spectrum Policy workshop held in Ottawa by the International Telecommunications Society.

This past Thursday, Telecom Review magazine released a special report “Spectrum and regulatory policy in the Race for 5G”, examining what various countries around the world are doing to deliver 5G services. The report concludes with “an urgent plea to the politicians and regulators, if you do not allocate sufficient spectrum to the various operators, then you have delivered 4G and underserved your nation in this global 5G race.”

In the article, Italy is identified as an example of what not to do in releasing 5G spectrum:

Italy in our opinion offers a perfect example of what a regulator ought not to do. Italy’s communication regulator, AGCOM, has set its near-term 5G future up for failure in two ways. First, by offering incumbent fixed licensees a six-year extension of their 200 MHz licenses (in the 3400-3600 MHz band) originally set to expire in 2023, Italy limited itself to only being able to auction 200 MHz of spectrum in its 2018 auction, split among four players. Compounding this shortage was a globally recognized poor auction design which split the 200 MHz into two blocks of 80 MHz and two blocks of 20 MHz (with a 100 MHz cap). In doing so, Italy forced its bidders to fight desperately for an 80 MHz block, leading to exorbitant auction prices and a distribution of spectrum that only supports true 5G deployment on two networks.

Unfortunately, the article continues, saying “In North America, Canada appears set to follow Italy’s example”.

By way of contrast, the article highlights the way some regulators are moving away from auctions, favouring a direct assignment of spectrum.

In April 2019 Japan assigned 100 MHz each to four operators (including a new entrant) based on coverage and investment commitments. Similarly, in the UAE two operators were allocated 200 MHz each with potential plans of making an additional 100MHz available.

ARCEP, the French regulator, has proposed to allocate up to four 50 MHz blocks in exchange for a series of “optional commitments”, and the remainder of the 310 MHz band would be allocated in a traditional auction.

If a hybrid approach like France was tried here, what kinds of potential commitments might be sought by the government in exchange for spectrum?

In September, I wrote “The cost of spectrum policy” and I warned “There is a cost associated with spectrum policy, not all of which is financial. As Canada moves forward with development of auction policy for the next wave of spectrum, it is critical to consider the potential for unintended consequences to have significant impact on consumers.”

The mandate letter for Minister Bains already contains a clause on spectrum set-asides that appears to pre-determine a portion of what normally follows a public consultation: “Award spectrum access based on commitments towards consumer choice, affordability and broad access. You will also reserve space for new entrants.”

If Italy offers a perfect example of what a regulator ought not to do, then Canada should be concerned when an international report warns that “Canada appears set to follow Italy’s example.”

It bears repeating: There is a cost associated with spectrum policy, not all of which is financial.

1 comment to Spectrum policy in the race for 5G

  • Michael Connolly

    The problem with awarding spectrum on the basis on promises to fulfill specified social commitments is twofold.

    First, promises cost nothing so all market discipline gets sucked out of the selection process. Service providers will make the required promises for fear that they will lose market position otherwise even though they are not sure that they will have the means to comply. The temptation to make the promise now and seek leniency from the regulator later if required is irresistable. Alternatively if spectrum is awarded on the basis of best offers to deliver social benefits (a “beauty contest”) then service providers will inflate their offers beyond what they can actually deliver in order to “outbid” rivals. Essentially this is what happened in the disastrous PCS C-block auction held by the FCC back in the day. In that auction bidders were permitted to bid money they didn’t actually have but rather hoped to have in the future. Overly optimistic bidders bid amount beyond reason, bankruptcies ensued and the spectrum was tied up for a decade in the courts generating zero benefit. The award process needs the discipline of bidders having to convince investors (and not the government) that their business plan is sound. The old DOC and Industry Canada held a number of beauty contests that awarded spectrum on the basis of promises and pretty much invariably the licensees failed to deliver as promised.

    Secondly, stipulating the fulfillment of social objectives as a basis for spectrum award or as a condition of auction participation effectively spends public monies (in the form of foregone auction revenues) with no knowledge of the costs. Fulfilling social objectives imposes costs on service providers and reduces the valuation of the spectrum by those costs. Bidders will deduct the cost of compliance from their bids. The government has no way of knowing the cost of the commitments it imposes. How can one know if the achievement of the social objective was worth the price effectively paid by the public? Wouldn’t it be better to simply maximize auction revenues by omitting the commitments and then use the proceeds to fund the attainment of the social objectives? This also gives Parliament the opportunity to decide which of the many priorities it wishes to fund (avoiding what is termed the “hypothecation of revenues” by the regulator) and thus is more democratic. Also, does it make sense to set up a scenario where uneconomic service provision in remote or underserved is required of multiple service providers? It would make more sense to fund service provision in uneconomic areas by selecting a willing service provider on the basis of the least required subsidy as determined by a reverse auction tendering process.

    For some reason we have a tendency to keep repeating the mistakes and failures of the past.