India 700-MHz Spectrum Reserve Prices Will Discourage Bidders

Minimum prices set for an auction of 700-MHz spectrum in India are so high (two to four times higher than prior auctions)  that many of the leading mobile companies will not bid. And, according to Fitch Ratings, the eventual spectrum winners might well regret having won. That “winner’s curse” has happened before, often with 3G spectrum auctions.

India's telecom regulator recommended a reserve price of INR115bn (US$1.7 billion) per MHz for nationwide 700MHz spectrum.

Fitch Ratings “believes that efficiency gains from deploying 4G services on 700MHz will be insufficient to offset the relatively high price.”

The reserve price is about twice the price set for 800-MHz spectrum, 3.4 times the reserve price for 900-MHz spectrum and four times the minimum prices set for 1.8 GHz spectrum.

Winning therefore “could exert further pressure on participating telcos' balance sheets and cash flow, and limit their ability to invest in capex over the medium term,” say Fitch Ratings analysts.

In fact, the top four telcos, including Bharti Airtel, Vodafone, Idea Cellular and Reliance Communications may hesitate to bid, as balance sheets already are “stretched,” while available cash is expected to become an issue once Reliance Jio enters the mobile market in the spring of 2016.

Instead, the leaders might choose to rely on spectrum they already have acquired. Bharti Airtel will use 900 MHz, 1.8 GHz and 2.3 GHz.

Reliance Jio, after having invested about US$15 billion on spectrum and networks, will use 800MHz and 850MHz spectrum.

Every single cost in any telecommunications or other ecosystem ultimately is paid for by customers or business partners, it goes without saying.

That is true of all infrastructure costs, including the cost of spectrum, which is why many observers favor increased use of shared spectrum and unlicensed spectrum as a way of expanding capabilities while minimizing end user cost.

The cost of licensed spectrum, typically a major expense, now is an issue in India, which is preparing to issue licenses for 700-MHz spectrum. High costs were an issue in earlier spectrum auctions as well.

After a massive auction where incumbents essentially had to pay whatever was required to retain use of frequencies they already were using, “the industry simply does not earn enough to support the bids,” some have argued.  

In that spring 2015 auction, India’s mobile companies bid Rs 1,10,000 crore for spectrum across four bands that were substantially above the reserve (minimum) prices.

Buyers paid 1.79 times the reserve price for 800MHz spectrum and 1.95 times the reserve prices for  900 MHz.

Operators paid less of a premium for higher-frequency spectrum, just 1.16 times the reserve price for 1800 MHz spectrum and 1.05 times the minimum for 2100 MHz spectrum.

John Giusti, Chief Regulatory Officer, GSMA, is among observers urging the Indian government to lower minimum prices for the 700-MHz auction.

“The GSMA is very concerned by TRAI’s recommendation to set a starting price of US $1.7 billion per MHz for pan-Indian 700MHz spectrum,” said Giusti.

source: Live Mint

India has one of the lowest average revenue per user (ARPU) metrics globally (US $2.45 at the end of 2015).

High spectrum costs, combined with limited revenue contribution from data services, new competitive pressures and high capital investments to upgrade to 4G, mean it will be difficult to sustain business models, if high spectrum prices hold, Giusti argued.

“The more mobile operators have to pay for a spectrum licence, the less capital is available to roll out new mobile networks,” Giusti said. “We encourage greater focus on the long-term benefits of connecting more people in India to affordable mobile broadband, rather than on short-term financial gain.”

“High reserve prices and an unrealistic predetermination of spectrum value could also reduce the willingness of potential bidders to buy the spectrum,” he noted. “For example, in Australia, an unrealistically high reserve price resulted in a valuable portion of the 700MHz spectrum left unsold and unused.”

“Setting reserve prices at reasonable levels will be key to achieving the Digital India objectives, allowing operators to focus their resources on building the necessary infrastructure to deliver high-quality mobile services for Indian citizens,” Giusti said.

5G Can Improve Rural Connectivity if Spectrum is Free

It is far from clear how well 5G networks might help improve or supply quality internet access in rural, mountainous and thinly-populated parts of the United States. Almost by definition, such areas do not support sustained business models for fixed networks, and require subsidies.

That is why hundreds of independent wireless internet service providers now are the way many people in rural areas get internet access services. Those firms also tend to rely on use of unlicensed spectrum to make the business model work.

The same sort of economics work for Wi-Fi. Low entry cost, as spectrum is available at no cost, is key.

source: Saankhya Labs

One might argue that should continue to be the case in the 5G era: networks will have to be wireless, and will require unlicensed spectrum access.

It probably matters less whether the unlicensed spectrum is gotten using shared or dedicated mechanisms (Citizens Broadband Radio Service, which is shared; or new unlicensed spectrum in the millimeter wave bands). The point is that tough business models require no-cost access to spectrum.

Whether 5G networks can make a big difference in providing good rural coverage hinges on access to no-cost and low-cost spectrum.

Over the past few years, some have worried about the cost of 5G spectrum, although spectrum prices are dropping, generally speaking, in part because there is a huge increase in supply, and because mobile operators must now more carefully weigh the cost of new spectrum against expected financial return.  

Also, firm strategies now vary. Some firms believe use of unlicensed spectrum will be more important. Others substitute small cells for additional spectrum. Some need additional spectrum more urgently than others, based on present holdings.

Recent auctions of 3.5-GHz spectrum have no clear pattern, especially since the various auctions featured different amounts of total spectrum and different license allotments (bigger or smaller amounts of spectrum per license), and there always are local market drivers (some contestants have greater needs for spectrum).

The point is that supply and demand issues affect price, as always. Finland offered the most 3.5-GHz spectrum, at 390 MHz. Spain and Italy each sold 200 MHz. U.K. regulators auctioned 150 MHz.

Finland’s prices wound up at 0.04 euros per megaHertz pop (a MHz POP represents one megahertz of bandwidth passing one person in the coverage area).  Spectrum sold for 9.07 euros per MHz POP in Spain, but a whopping 0.51 euros per MHz POP in Italy. U.K. spectrum sold for 0.17 euros per MHz POP.

On the other hand, at least one Australian official worries that a recent big merger between TPG and Vodafone will reduce demand and lead to lower prices.

source: GSMA

One might simply argue that supply and demand will  work. Whatever the limits on new spectrum at 3.5 GHz, regulators simply must make more spectrum available in other bands. More supply takes care of pricing pressures. Releasing more unlicensed spectrum, spectrum sharing, spectrum aggregation and additional spectrum in the millimeter wave bands all will help ensure there is plenty of 5G spectrum and that prices will not be onerous.

Supply in the 3.5-GHz auctions will be something of an issue, in most countries, as there is not lots of spectrum available there.

source: GSMA

French regulator Arcep’s chief Sebastien Soriano announced it will be challenging to keep prices low for a 2019 spectrum auction, especially when supply cannot keep up with demand, but said it wants to find a way to do so.

Traditional government views of spectrum auctions as easy ways to raise government revenues also are issues. Policymakers have to balance the need to make lots of spectrum available with the desire to raise revenues.

As you would expect, firms that have paid high prices justify their actions by arguing the new spectrum will help reduce costs per gigabyte, as well as supporting all the new end user demands for capacity.

High spectrum prices, though, have been big problems for mobile operators. Recall the high prices paid for 3G spectrum in many countries, which nearly bankrupted several firms in Europe, the high prices paid in India and in some other countries.

More supply will help keep prices within reasonable ranges.

If There is a Spectrum Bubble, Does it Martter?

Spectrum is not the only cost input for a mobile service provider, nor is it the largest cost input. As a rule of thumb, operating expense might be in the range of 45 percent of revenue, while all network related capex might be in the range of eight percent of revenue. So spectrum acquisition costs are not a huge driver of overall costs, typically.

What really matters is revenue. Still, the cost of spectrum matters in an environment where service provider costs threaten to exceed revenues earned from such spectrum.

Depending on the country and the population density and terrain, a fully functioning 3G network, for example, might cost between several hundreds of dollars per customer, to a few thousands of dollars per customer.

Spectrum costs are a fraction of that. Assume a service provider has 20 MHz of spectrum in an area of three square miles, paid for whether all of the capacity is being used or not (spectrum reuse is necessary to avoid signal interference).

Assume people reached by signals from that one tower number cover three square miles, where the density is 600 persons per square mile. That implies a population of 5,400 people, each representing 20 MHz per pop. At prices of 10 cents per MHz pop, that works out to $2 per person, or $10,800 in spectrum costs in that area.

At 33 percent market share, implying 1782 paying customers, the cost per customer is about $6 per customer. Even paying interest on such an investment is a small part of the total cost of providing service.

But you can see the sensitivity to price per MHz pop. At $1 a Mhz pop, the spectrum would represent $108,000 in spectrum costs, or $61 per customer. At $4 per MHz pop, spectrum costs would represent $244 per customer.

Amortized over 10 years, with monthly revenue of $50, that still is not unworkable, at reasonable market share. But as with any fixed cost, market share really matters. At 16 percent share, spectrum cost grows to about $488 per customer.

If revenue earned from the leading services sold by mobile service providers is dropping, and if market share is fragmented, the cost impact of spectrum acquisition is magnified.

In that sense, revenue per MHz pop, though not a metric anybody uses, likewise will drop. In the end, that is the key issue: revenue per MHz pop, not cost of spectrum per MHz pop, at a high level.

Whether Long Term Evolution 4G auctions will become a spectrum bubble is anybody’s guess, at the moment. But industry observers with long memories will recall that vast overbidding nearly bankrupted leading European mobile service providers when 3G auctions were held.

There are some signs of price inflation in the Netherlands, Ireland, Taiwan, in Austria and in the Czech Republic, for example. In Taiwan, bid prices were about three times what regulators set as the minimum price. The Czech Republic suspended its auction when prices grew too quickly.

The U.K. 4G spectrum auctions generally are considered reasonable, compared to Czech prices before the auction suspension. The May 2013 U.K. auction raised around EUR0.18/MHz/pop. The Czech auction had reached EUR0.25/MHz/pop, about 30 percent higher than the actual U.K. prices.

MHz per pop is a way of measuring capacity per person, and cost per megahertz per pop is a way of measuring spectrum cost, per unit of capacity, per person.

Even that pales in comparison to 3G auction prices in some markets, where past prices have been measured in tenths of cents or cents. In some cases, Western European 3G prices were measured in dollars.

Of course, the “right price” for spectrum hinges on any number of business and market factors. The value of Clearwire spectrum provides a recent example.

Some recent 700-MHz spectrum in the U.S. market has sold for dollars per MHz pop, a “high” price by world standards. But that spectrum also has coverage and wall-penetrating advantages bidders believed justified the price.

Whether spectrum was acquired at prices “too high” can be determined only after the capacity is put into service and revenue generated by that spectrum can be assessed. Prices of dollars per MHz pop might be quite reasonable if the new spectrum allows a service provider to gain customers, raise profit margins or gross revenue, cut churn or create uniqueness.

In other words, 3G prices were an order of magnitude to two orders of magnitude above spectrum prices paid before, or after. To be sure, the value of spectrum generally is affected by the actual frequencies: lower frequencies are more valuable than higher frequencies.

That is a function of signal propagation, not bandwidth potential. Signals at lower frequencies attenuate less, and hence travel further, with better ability to penetrate walls. On the other hand, signals at higher frequencies are capable of providing much more bandwidth, using any specific coding technique.

Still, prices for 3G spectrum awarded in more recent auctions also was measured in the dollars per MHz pop. The 3G auctions in India provide a recent example.

Also, some spectrum, licensed for backhaul applications rather than end user services, generally costs less than spectrum enabling actual end user services.

For example, 3.5 GHz spectrum intended to support  fixed wireless access applications, rather than mobile applications, often was sold at prices an order of magnitude less than spectrum for mobile apps.

3.5 GHz Spectrum Band Pricing Examples

Country	Band	Price of 10 MHz	Per MHz-PoP
Italy	3.5 GHz	€ 10,793,651	€ 0.0189
Germany	3.5 GHz	€ 4,325,397	€ 0.0053
UK	3.5 GHz	£1,750,000	£ 0.0030
UK	3.8 GHz	£744,048	£ 0.0012
Netherlands	3.5 GHz	€ 500,000	€ 0.0030
Switzerland	3.5 GHz	CHF 1,416,667	CHF 0.0178
Canada	3.5 GHz	$2,877,402	$ 0.0049

On the other hand, 2.5 GHz spectrum made recently available in many European countries (and Canada in 2004 and 2005) cost more than 3.5 GHz spectrum, but they are much lower than prices fetched in the 800 MHz spectrum band (which range between € 0.5 – € 0.8 per MHz-PoP in most countries).

2.5 GHz Spectrum Band Pricing Examples

Country	Band	Price of 10 MHz	Per MHz-PoP
Sweden	2.5 GHz FDD	€ 14,867,475	€ 0.159
France	2.5 GHz FDD	€ 66,866,394	€ 0.106
Italy	2.5 GHz FDD	€ 35,996,667	€ 0.059
Belgium	2.5 GHz FDD	€ 5,025,455	€ 0.046
Belgium	2.5 GHz TDD	€ 5,002,222	€ 0.045
Italy	2.5 GHz TDD	€ 24,678,367	€ 0.041
Sweden	2.5 GHz TDD	€ 3,416,868	€ 0.037
Spain	2.5 GHz FDD	€ 12,334,753	€ 0.027
Germany	2.5 GHz FDD	€ 18,412,643	€ 0.023
Germany	2.5 GHz TDD	€ 17,303,600	€ 0.021
Netherlands	2.5 GHz FDD	€ 2,627,000	€ 0.0012
Canada WCS	2.3 GHz WCS	$ 6,136,598	$ 0.018

Looking at spectrum pricing in the higher spectrum bands, 2.5 GHz assets typically sell at a discount of up to 92 percent that of 800 MHz spectrum, while 3.5 GHz spectrum sells at around 82 percent discount to that of 2.5 GHz.

3.x GHz Spectrum Band Pricing Examples

Country	Band	Price of 10 MHz	Per MHz-PoP
Italy	3.5 GHz	€ 10,793,651	€ 0.0189
Germany	3.5 GHz	€ 4,325,397	€ 0.0053
UK	3.5 GHz	£1,750,000	£ 0.0030
UK	3.8 GHz	£744,048	£ 0.0012
Netherlands	3.5 GHz	€ 500,000	€ 0.0030
Switzerland	3.5 GHz	CHF 1,416,667	CHF 0.0178
Canada	3.5 GHz	$2,877,402	$ 0.0049