Broadband and Internet

East Africa: The Race to The Terabit Backbone

In the last few months, two major operators in Kenya announced their plans to upgrade their traffic backbones. Safaricom announced it’s partnership with Huawei to upgrade its Mombasa-Nairobi backbone to 400Gbps while Liquid Telecom announced its partnership with Nokia to upgrade its East African backbone to an initial capacity of 500Gbps using OTN/DWDM technique. This as you would imagine is a massive upgrade from the current capacities.

With the increase in the use of Internet and mobile phone use in Kenya and the greater East African region, these backbones will ensure that the region does not suffer from bandwidth constraints due to lack of capacity to carry the traffic. Also, to enhance the user experience, Internet giants such as Facebook and Alphabet (parent company to YouTube) have set up powerful cache servers within the country. These cache servers store most popular content locally in Nairobi at the Africa Data Center in Nairobi and iColo in Mombasa. How these work is for example is when a person watches a YouTube video and shares the video link with others in Kenya, the next person won’t get the video from the YouTube servers in the US or Europe, but will get the video from a server in Nairobi where a copy will be temporarily stored. What this means is that the video will load faster with no buffering and also ease up capacity on the international backbone that would have been used to access the video by everyone to whom the video was shared. Same case for Facebook videos and photos. With the increase in smart phone adoption and multimedia content sharing on social media such as WhatsApp and the fact that over 70% of East African internet traffic is social media and video streaming, the amount of traffic is bound to increase exponentially if the region starts to use broadband and cloud more productively and not just for recreation, these backbone upgrades are therefore happening at the right time.

Fiber-Optic Upgrades 101

Other than the high capacity they provide, the other main advantage of using fiber optic cables for data transport is that once the optical glass core is buried in the ground, any upgrade of capacity does not necessarily need additional cores buried (unless better cores with much less losses per km come to market or the existing core is extensively damaged). The upgrade only needs to happen at the end point equipment.

Fiber optics use light to carry information, in the early days, a single color of light (a.k.a wave length) was used to carry traffic. However, with the advancements in physics, it became possible to use several colors of light (several wave lengths) to carry traffic in a single glass core; a technique known as DWDM. All advancement in optical transport is simply based on how many color shades or hues that a fiber network equipment can detect. Just like in the 5G arena, there has been a silent war between European and Chinese OEM’s on whose equipment can detect more hues (more wavelengths). The more hues the equipment can detect at the same power levels, the more data can be carried on that channel.

5G and Multimedia Traffic

With the approval by the 3GPP of the 5G-NR standards, it is expected that 5G networks will deliver speeds of up to 20Gbps (in lab conditions), real world speeds will hover around 1Gbps which is still high by today’s metrics. With the increase in consumption of multimedia and streaming content, the demand for even bigger data backbone pipes will continue on an upward trend. Just to give an example, in mid 2017 Liquid upgraded their EA backbones from 10Gbps to 100Gbps and less than two years later to 500G because of the demand. In the next 3 years or so, Liquid’s 500G and Safaricom’s 400G won’t be sufficient. There will therefore be the need to scale up to the Terabit space. This will allow the operators meet capacity demand in the region for both data and voice traffic. This is especially critical at a time when many Internet companies are setting up Content Delivery Networks (CDNs) and cloud presence in Kenya. Facebook is already in Mombasa, Google in Nairobi and Mombasa, Netflix and Akamai in Nairobi and many others. Microsoft is finalizing the setting up an Azure Stack in Nairobi as Amazon Web Services (AWS) announced that they are setting up local cloud presence in Africa with an availability zone in Nairobi and South Africa. These locally based cloud points of presence will enable cloud service providers offer lower latency to end users across Sub-Saharan Africa and will enable more Africans to leverage advanced technologies such as Artificial Intelligence, Machine Learning, Internet of Things (IoT), mobile services, and more to drive innovation. With the current broadband use not being productive use; with most traffic being recreational and social media, the availing of low latency cloud services from within the continent will drive up our efficiencies and help the region to develop.

Government’s Role

I happen to have attended the National Broadband Strategy II (NBS II) stakeholders consultations and review of the proposed strategy. There seemed to be a shift from the strategic direction of NBS I which mainly focused on creating an enabling environment for private investors to roll out broadband networks in the country. In the proposed NBS II that covers the period 2018-2023, government seems focused on using tax payers money to roll out broadband. This to me is not the way to go as previous government investment in similar projects such as the National Optic Fiber Backbone (NOFB) haven’t been well executed and currently stands at 12%. The percentage completion of the other projects under NBS I such as digitization of core government registries where several registries have been digitized was not documented. For some other projects, there was no indication whether they were undertaken or not; an example is the development of county management information system. All these projects were allocated KES 250B under NBS I. The NOFBI’s current performance and availability is far below the closest competitors offering. The government should leave telecom infrastructure development space to capable private sector hands. Government can aid the private sector with incentives and tax breaks on equipment and other costs associated with broadband roll-out and creating a level playing field through proper policy and sector regulation.

Mobile Telephony, Regulation and Law

My Thoughts on the Proposed Retail Tariff Controls in Kenya

telecoms-13The Communication Authority of Kenya is mulling on introducing retail tariff controls in the Kenyan mobile telephony sector. As you would imagine, this has elicited mixed reactions depending on who you ask. This move it is said is part of a raft of recommendations by the recently released report that studied the Kenyan telecommunication competitive landscape. The introduction, according to the regulator is aimed at correcting market failures that are as a result of one operator being dominant over the rest. The proposed retail price controls aim at limiting Safaricom’s freehand in determination of loyalty schemes and promotions, prohibition of on-net discounts, mobile money fees charged to unregistered and cross-platform money transfers. Tariff  control is a regulatory mechanism that can be employed to correct market failures in a market. The main motivation for price control is to protect consumers’ rights and interests in circumstances where market forces alone have been unable to do so.

The competitive landscape report implies that Safaricom’s price differentiation between on-net and off-net calls leads to tariff-mediated network effects. By this I mean that subscribers on a network find it cheaper to call others who are on the same network than calling people on a different network, the effect of this the emergence of the ‘club’ effect where subscribers decide which network to join based on which network the people they call mostly are on. If most of my family and friends subscribe to a network X, I am more likely to subscribe to network X because by being on network X and not on network Y, my overall calling costs will be lower due to cheaper on-net pricing. This it is argued is abuse of its dominance. But this is only true if the interconnection rates are high between Y and X meaning it will cost me more if I chose network Y where none of my relatives and friends subscribe to. As we speak, interconnection rates are already regulated by CA. This fact alone means that the regulator found and enforced an interconnection rate that was cost based and reflected the market realities. The Safaricom on-net call rates are above the regulator-set interconnection rates  and are similar to the rates its competition provides its customers to call into the Safaricom network, in short, Safaricom’s subscribers pay the same rate as competitors subscribers when both call another Safaricom subscriber. CA would be justified in imposing tariff controls if indeed Safaricom’s subscribers paid less to call their counterpart than when competition subscriber calls. The fact that the report found that each operator was dominant on its own network also means that the club effect can easily be replicated by competition through aggressive marketing and innovation.

Individually tailored tariffs

Another proposal by the regulator is the prohibition of individually tailored loyalty schemes and promotions. The CA 2010 Tariff regulations specify that all promotions or loyalty schemes must be approved by the regulator prior to commencement. Any promotion, offer or scheme currently in operation has the direct approval of the regulator, the details of each promotion or loyalty scheme structure were available to CA before approval and these include individually tailored schemes and promotions. When an operators behaviour due to regulation causes it to become successful, it cannot be punished for successes it encountered while operating within the laid down regulatory framework. If there is any discriminatory behaviour, then CA failed in detecting and stopping this even with the existence of regulations. When that happens, it is clear that regulation is not the solution to the perceived market failures and the only remedy is let competitive forces determine customer choice.

What would happen if call tariffs are regulated?

Most people assume tariff regulation leads to the lowering of retail rates. This would only be true if CA had a way of accurately measuring the costs involved to provide service to additional customers on their network. At the moment, due to the fact that Safaricom has invested more in network infrastructures geographical and population reach, it costs them less to bring in a new customer and provide service to them than its competitors. The proposal to use Long Run Incremental Costs (LRIC) method would mean that because of the massive economies of scale enjoyed by Safaricom owning a large network and technology convergence, the cost of offering an extra minute of termination on the network is virtually zero in the short-term and LRIC is applied to give a realistic cost over the long term. How long is ‘long term’ will determine the price point.
A regulated tariff price should imitate the prices that would have arisen in a market with effective competition, both because this provides incentives to produce the requested service at the lowest possible cost and because the operators requesting the service have the incentive to optimize their own investment decisions. In this way price controls can contribute to efficient utilization of social resources. The problem however is that LRIC works very well in legacy circuit switched networks where it was easy to link network utilization with the provision of a service (if a switched circuit is in use, it is providing a directly attributable service), in a converged environment, the relationship between network resource utilization and provision of services is not that straightforward and an LRIC model applied to it would fail to efficiently allocate costs. As networks evolve into Next Generation Networks (NGNs), LRIC approach to service cost allocation becomes inefficient and LRIC starts to resemble the Full Allocated Costing (FAC) model. The result? higher prices to the consumer.

Replicable retail tariffs

One of the recommendations by the study is that Safaricom’s Tariff must pass the ‘replicability test’.  In order for a product or service to be considered replicable, it must be commercially and technically capable of being replicated by Safaricom’s competitors. By this we mean that any new tariffs that Safaricom comes up with should take into consideration the capability or state of competitors infrastructure and services especially at the wholesale level. If for example, Safaricom comes up with a tariff that takes advantage of a new innovation they developed, then competition should have similar innovations on their networks or should be allowed by Safaricom to access this new innovation at wholesales prices to enable them offer similar tariffs. In order for competitors to be able to compete with Safaricom in retail markets, it is necessary for them to have access, either via their own networks or through access to Safaricom’s network, to the wholesale components that enable those retail prices to be offered. This as you can imagine will slow down the pace of innovation by Safaricom or change the main focus of innovation from the customer to competition, anytime Safaricom wants to innovate, they will think of ways through which competition will not benefit much from it as opposed to spending their time and knowledge on how the innovation will first benefit the customer.

Tower sharing

The report found that Safaricom is dominant in the tower market owning about 65% of telecommunication towers in Kenya and recommends that it shares it towers with competition. To do this effectively, Safaricom has to let a 3rd party manage the towers on its behalf. Safaricom owns its towers while competition has a mix of owned and outsourced/leased towers. With rapid advancement in technology, there is a saying in the industry that it is always cheaper to build a network tomorrow. Any new entrant or existing operator rolling out new retail services and infrastructure such as towers will do it cheaper today than those who did it in the past. See how easy it was for Finserve Africa (Equitel) to get market reach by the flick of a switch riding on an MVNO license? Older operators didn’t get to enjoy these benefits and had to put in sweat and blood to get where they are.
In the Safaricom 2017 sustainability report, the operator spent close to 10 million litres of fuel and about KES 48,000 per month per tower location on energy costs. The report also shows that fuel and power costs are coming down and that Safaricom has also embraced green energy initiatives to power the active components at the tower stations, it is becoming cheaper by the day to setup and manage tower locations but the increase in the number of towers makes the management of the same a challenge.  Outsourcing of tower management is used by some operators to shift a large element of their fixed costs to the variable costs column of their financial books, in doing so the operators financial health improves. Tower outsourcing has little to do with improved efficiency in actual operations and management of these towers but more to do with improving the books. Due to this, the global trend in the telecom sector is the outsourcing of tower management to independent 3rd parties, something Safaricom should consider for its own benefit and not due to regulatory requirement. Safaricom should then decide to lease its towers to competition on a purely commercial basis.

Conclusion

Although well meaning, the CA needs to explore non-tariff based remedies to the perceived dominance of Safaricom. Tariff control is a very intrusive approach to handling abuse of dominance whether real or perceived. As markets evolve to become competitive, ex ante regulation should reduce as the regulator forebears regulation in favour of competitive forces. The challenge however is that customer inertia can mask the existence of competition in the market leading to the regulator applying regulatory tools to correct this perceived market failure. Market forces should be let to dictate the market tariffs.
The tariff control ocean floor is littered with shipwrecks of tariff control attempts by various countries that  tried to correct market failures by  employing it. One of these countries is Mexico who in 2012 introduced tariff controls by baring America Movil from charging interconnection fees to its competitors. Movil controls 70% of Mexico mobile market. As of 2015, mobile call charges had dropped by about 17% but investment in the sector dropped 30% during the same period, for a developing country like Kenya, a drop in ICT investments will have far reaching effects across multiple sectors of the economy. Whereas the intended outcome of tariff control is to  protect the consumer from a dominant operators actions, the question that remains is whether tariff control is the only viable market correction tool available.