ICASA sets off down the spectrum road

Barely skipping a beat, ahead of schedule, ICASA on 1 November released its promised information memorandum on the licensing process for high-demand spectrum.

ICASA’s document sets out the roadmap going forward for the auction of spectrum that the sector has been clamouring after for some 10 years. It even provides an unusually generous timeframe for stakeholder comment ‒ 90 days ‒ albeit ICASA’s generosity is likely occasioned less by an onrush of early Christmas spirit than by the need to await the outcomes of the ITU’s World Radio Conference 2019, currently on the go in Sharm el-Sheikh, Egypt.

Whilst the roadmap will see a dramatic increase in spectrum availability for mobile services ‒ from 566MHz to 958MHz ‒ it nevertheless begs a number of procedural questions and leaves several key issues hanging.

ICASA’s approach to the licensing of International Mobile Telecommunications (IMT) spectrum, as anticipated, goes bigger than the ministerial policy direction that triggered it. It includes some frequency ranges being eyed for 5G: 2 360-2 400MHz (IMT2300) and 3 400-3 600MHz (IMT3500).

But it stops short of contradicting the minister’s moratorium on 5G spectrum, making fault-finding from that quarter difficult.

ICASA is careful to ground its approach in the enabling legislative and regulatory framework, and to push all the right buttons: the National Development Plan, SA Connect, the 2016 White Paper, universal access and service, and the cost to communicate.

But its careful regulatory history behind high-demand spectrum is somewhat sanitised. There is no mention of the regulator’s own culpability in the withdrawal of the 2011 ITA. Nor is there any mention of the 2016 litigation between ICASA and the minister that led to the demise of the regulator’s previous attempt to get high-demand spectrum out into the market.

ICASA’s document invites stakeholders to add their own shopping lists and hobby horses to the mix.

The core of the roadmap centres around ICASA’s presentation of no fewer than five alternatives for how to slice and dice the core high-demand spectrum in the 703-790MHz (IMT700), 790-862MHz (IMT800) and 2 500-2 690MHz (IMT2600) bands. Careful canvassing of the options is likely motivated by the howls of complaint generated by the lots in ICASA’s previous spectrum ITAs.

Existing licensees and potential bidders alike will certainly have firm views on which of the five options ICASA should implement, and the final choice is likely to be a substantial bone of contention. Confusing to the layperson, grist to the engineer’s mill, it is nonetheless a mixed box of apples and pears.

Note: FDD = Frequency Division Duplexing
TDD = Time Division Duplexing

Further, the spectrum in the IMT700 and IMT800 bands is, as ICASA acknowledges, subject to the completion of the stalled digital migration.

One of the slices (Lot A) is set aside for the WOAN in all options, with slightly varying configurations of spectrum in either the IMT700 or IMT800 bands, and a slice of IMT2600 spectrum.

Three of the lots (Lots B, C, D) look to be earmarked for mobile incumbents, with equalised allocations of spectrum in either IMT700 or IMT800 and in IMT2600 in almost all cases. It is these lots that Vodacom, MTN and Telkom Mobile are most likely to vie for.

One of the lots (Lot E) appears in only three of the options, and then with considerably less spectrum attached than for the three prize lots above. In the other two cases, the remaining spectrum is held over to be auctioned at a later date. Lot E is clearly less desirable than the others, which is why no universal service obligations are attached to it ‒ although it is puzzling why it has not more specifically been tailored to attract a new entrant, since it is unlikely the deeply troubled Cell C will be able to afford to bid.

In all cases, bidders will be subject to a limit on spectrum acquisition, ensuring those with the deepest pockets are not able to squeeze others out of the market by bidding for multiple lots.

In two of the lots, Rain retains its incumbent spectrum, and in two others it is required to migrate in-band. One of the options, however, makes no mention of Rain, leaving the fate of its IMT2600 spectrum open to speculation in that case.

It is apparent that ICASA intends to finalise the auction for that spectrum still available in the 2 360-2 400MHz (IMT2300) and 3 428-3 544MHz (IMT3500) bands only once WRC-19 has reached its conclusion at the end of the month.

This is partly because South Africa is currently lobbying for the ITU to earmark IMT3500 primarily for mobile services.

The ICASA roadmap provisionally sets out lots in these two bands, but takes the process no further for now.

There are a number of public policy objectives, universal service obligations in particular, attached to the first round of the auction (IMT700, IMT800 and IMT2600).

In particular, ICASA has stated it intends to apply the kind of universal access and service provisions favoured under the spectrum auctions in Germany. The model requires successful bidders to achieve certain minimum levels of coverage and bandwidth in under-serviced areas before being permitted to offer services in well-serviced areas.

This obligation is likely to be met with strenuous opposition from potential bidders. It will, of course, reduce the value attached to the spectrum in question, and lower likely revenues from the auction because of the additional cost burdens it will impose on licensees.

Although billed as successful, the recent 5G auction in Germany was criticised for imposing “unreasonable and unrealistic” licence conditions, for undercutting rollout because of the cost of the spectrum, and for being excessively drawn out.

It is also unclear why ICASA has decided to push the MVNO model as part of its USO requirements. Whilst this may stimulate service competition, and diversify the number of licensees hosting MVNOs, it may be overly prescriptive in terms of its business model, and is reminiscent of the now-obsolete community service telephone obligations imposed on the mobile licensees.

There are also very necessary provisions covering BBBEE and open access, and passive infrastructure facilities leasing.

Additional social service obligations may yet be added to the ITA once it is issued. ICASA’s document invites stakeholders to add their own shopping lists and hobby horses to the mix.

Some of the provisions dealing with obligations towards the WOAN are difficult to understand. This commentator has yet to fathom what precisely is meant by the “offtake” requirement, namely that a “minimum of 30% national capacity must be procured from the WOAN collectively”, or indeed how this requirement will be operationalised.

ICASA may have similar difficulties, which is why the minister’s wording is reproduced verbatim.

It is also unclear why roaming provisions for the WOAN have been added, given that it has historically been conceived of as a wholesale operator.

However, the inclusion of roaming likely reflects the fact that the licensing framework of the ECA does not cater for such a wholesale restriction on service licences.

One of the problem areas in the roadmap is its failure to deal with the question of what constitutes an under-serviced area for the purposes of coverage and rollout obligations.

The current definition of such areas was promulgated by ICASA way back in 2012, on the basis of a rather arbitrary arithmetic exercise applied to data drawn from Stats SA.

It resulted in 199 out of a national total of 251 municipalities and district municipalities being declared under-serviced. That list is no longer helpful in the broadband era. But it has never been updated, despite ICASA’s undertaking to do so bi-annually.

The ICASA roadmap kicks for touch on this issue, asking stakeholders to propose a revised definition of under-serviced areas.

Despite some weaknesses and problem areas, the ICASA roadmap marks a crucial first step in getting spectrum rubber out on the ICT road.

Much of the debate is likely to focus on the determination of the lots for the first round of the auction, and on the public policy obligations to be attached to the ITA.

ICASA will need to focus on attaching an appropriate value to the spectrum on offer in order to set realistic reserve prices, and on developing the parameters of the simultaneous multi-round ascending auction it says it favours, in order to minimise gaming strategies from bidders and achieve a fair auction price that balances monetary value with social objectives.

A number of commentators will be watching the touchline to see what determination of under-serviced areas ICASA will adopt, and how it intends to proceed with the second round of the auction (for IMT2300 and IMT3500 spectrum) once the outcomes of WRC-19 are known ‒ either as an add-on to the current process or as a separate second stage.

The spectrum auction, therefore, is a process unlikely to see completion within the six months optimistically predicted by ICASA.