In both Indian and U.S. markets, there is a growing expectation that spectrum prices are vulnerable to a tumble, despite expectations of growing need for capacity. In some cases, the issue is simply high reserve prices. That is the issue for Indian 700-MHz spectrum.

In other cases, the issue is supply. Auctions of 600-MHz spectrum already have one of four major U.S. firms not bidding, while a second firm will be able to bid in many cases without competition from the two biggest suppliers. That should mean lower prices.

U.S. mobile operators also know that much more spectrum will be released in the millimeter wave band, small cell approaches will largely satisfy demand in high-density areas.

And then there is the growing and successful use of unlicensed spectrum, soon to be augmented by shared spectrum. Meanwhile, Dish Network has amassed a hoard of spectrum that it either must deploy or sell. More now believe Dish Network will be unable to satisfy network build requirements, and will either sell–or have to give back–its licenses.

The point is that spectrum acquisition options and substitute network architecture solutions are improving, a trend that should lead to lower spectrum prices.

In India, as much as 2,400 MHz of spectrum–and a minimum of 2,000 MHz–is scheduled to be put up for auction by the late summer of 2016.

Compare that to a total former allocation of 200 MHz to 300 MHz for the entire Indian mobile industry.

Some new spectrum lies in the 700 MHz band, though most of the capacity consists of presently-used spectrum that has to be re-licensed.

The total new 700-MHz capacity will add but 35 MHz of incremental bandwidth.

One is left to guess that the 2,000 MHz (total spectrum actually might even rise to as much as 2,400 MHz after harmonization measures) figure is the total amount of spectrum across all 20-odd telecom “circles,” as the serving areas are known.

Some analysts have suggested that spending just for the 35 MHz of 700-MHz spectrum will represent as much as 74 percent of total auction proceeds.

And spectrum spending, if reserve prices are paid, will be quite substantial.

The reserve prices represent an amount about double all prior investment in spectrum in India, across every auction, according the the GSM.

In the U.S. market, observers disagree rather sharply about how much latent spectrum held by Dish Network spectrum is worth.  

Analysts at Kerrisdale Capital continue to argue that demand for Dish Network’s spectrum are wildly optimistic, and that spectrum prices are headed dramatically lower.

Others have argued that the spectrum represents most of the equity value of Dish Network as a whole. Some have pegged the mobile spectrum licenses as 80 percent of Dish equity value, based on a valuation of $35 billion to $50 billion for the spectrum licenses.

The notion that spectrum licenses create equity value for mobile firms is logical enough.

By some estimates, facilities-based U.S. mobile operators, plus Dish Network, own about $368 billion worth of spectrum licenses.

AT&T now holds spectrum licenses worth more than $91 billion, estimates Goldman Sachs analyst Brett Feldman, while the value of Verizon’s spectrum is $79.4 billion.

In all, AT&T now holds spectrum licenses worth more than $91 billion, estimates Goldman Sachs analyst Brett Feldman. He also estimates the value of Verizon’s spectrum at $79.4 billion.

The current equity value of all AT&T stock is $176.5 billion, implying that spectrum alone represents 51.6 percent of AT&T’s total equity value.

Verizon’s market value is $207.9 billion, implying that Verizon’s spectrum represents 38 percent of total valuation.  

Bloomberg Intelligence, in fact, estimate the total value of Sprint’s 2.5-GHz spectrum alone at $115.1, about 2.4 times Sprint’s enterprise value of $48 billion.

In fact, some argue that T-Mobile US spectrum accounts for more than 100 percent of its total market value.

But the market value of any particular block of spectrum, or Dish Network’s holdings, hinge on several assumptions. First, one has to assume that the potential buyers (AT&T or Verizon, principally) “need” that spectrum, and need it soon.

One has to assume that other viable options–possibly costing less–do not exist. One has to assume that cash available to buy Dish Network spectrum will be plentiful after commitments made in the 600-MHz auction.

And one has to assume would-be buyers can balance all that with capital investments in the coming 5G network, plus any other strategic investments that might be made.

Finally, longer term, as the U.S. spectrum licensing regime shifts to spectrum sharing and release of huge new swaths of spectrum in the millimeter wave region, one has to anticipate that spectrum licenses, though still valuable, might be worth less.

Nearly everyone agrees that much more mobile spectrum will be needed in the future (a few might argue that future needs mostly can be met by use of smaller cells). The issue is how much, and when, that might happen.

Not so soon, Kerrisdale analysis suggests. Looking at use of the 4G networks, for example, Kerrisdale says that with average (mean) smartphone usage of 3 GB per month, with a median of 1 GB, there is an implied load of 0.009 megabits per second, per device.

Assuming that 10 percent of a day’s usage occurs during the peak hour, about 0.02 megabits per second, per device, at peak. Average LTE throughput in the U.S. mobile market, per carrier,  is 10 megabits per second.

Also, for a number of reasons, Kerrisdale expects spectrum prices in the 600-MHz auction ot fall dramatically from the last AWS auction, which Kerrisdale says Dish Network was able to “game.”

For starters, Sprint will not bid at all, and Dish Network is unlikely to have the resources, much less the appetite, to do so. AT&T and Verizon probably will try to win about 20 MHz each, while T-Mobile US is expected to win most of the 30 MHz of reserve spectrum for smaller carriers. That amounts to demand of about 70 MHz, of an offered 126 MHz of so of capacity.

In other words, less demand than supply; hence the likelihood of lower prices.

In a regulatory environment that seems now to have shifted away from large mergers, there further is the issue of whether roadblocks might now be put in the way of any sale of Dish Network spectrum to either AT&T or Verizon, for antitrust reasons.

Nor, contrary to some popular belief, is the mobile business an especially lucrative business. Profits on invested capital range from a high at Verizon of 16 percent to a low for Sprint of just one percent, with T-Mobile US earning three percent and AT&T eight percent.

The point is that, no matter now valuable, none of the providers can afford to overspend on spectrum, especially if new options are going to emerge (new spectrum, lots of shared spectrum, better ways to use even more unlicensed spectrum).

“The industry’s mediocre profitability not only makes it foolhardy for a new player to aggressively enter the market; it also calls into question the rationality of paying ever higher prices for spectrum,” Kerrisdale analysts say. “When an entire network can only generate a single-digit return on capital, there is a limit to how valuable a single input to that network can be, especially at the margin.”

Kerrisdale researchers note that 84 percent of the U.S. population lives in suburban and rural areas with typical population densities in the hundreds per square kilometer. The clear implication is that 16 percent of the population lives in highly-dense areas where spectrum is most needed.

Across most of the land mass, there simply are not enough people, served by any single carrier, to exhaust capacity. With the caveat that it is peak use, not average use, that matters when designing a network, the “typical” cell tower coverage area does not suggest especially demanding capacity supply.

As an example, assuming a typical tower coverage radius of 2.5 km, a single cell tower could cover about 20 square kilometers containing 5,000 people.

If a single carrier has 35 percent market share, that means supporting 1,800 users.

At a usage rage of 3 GB per month per user, the tower would see 177 GB of average daily traffic.

Assuming only 10 MHz of downlink bandwidth, three sectors, and 1.5 bps/Hz/sector of average spectral efficiency, the site’s capacity would be 45 Mbps or 486 GB per day. Not a crisis, in other words, on average. Peak demand is perhaps another issue, but it is worth noting that the bulk of a consumer’s smartphone usage tends to happen at home, presumably therefore not at “peak” hours (drive time).

Dish Network always has been a firm willing to invest in spectrum. But now it faces a clock: unless it has built out a network reaching 70 percent of U.S. consumers by 2020, it will lose a batch of licenses. Other holdings will need to reach similar levels of coverage by 2022 and 2025, or the licenses are forfeit.

And it appears those deadlines will hold even if Dish Network sells the licenses to another firm. So there is a possibility that potential buyers can simply wait until Dish loses the licenses. The assumption is that the spectrum will eventually be put back up for auction, at which time prices will be lower, if the assumption about declining spectrum prices is correct.

The key point is that spectrum prices might fall, in the future. There will be so much new supply, even with increasing demand, that unit prices should fall.