Thursday, October 9, 2014

Spectrum valuation and a case for Verizon and against wannabe players like Sprint and desperado/promotes like GSAT

This post has been in the back of my mind for a long time. It is also talking my book - long Verizon - big, boring with a *lot* of very high value spectrum and short the spectrum wannabe players like GSAT.

As for timing, the post is prompted by the beginning of GSAT's collapse following the research report by Sahm Adrangi and Kerrisdale Capital.

We have another motivation. For 18 months or so Verizon has been one of our biggest (and sometimes our absolute biggest) long position and it has not worked. We haven't lost money - we just have not kept up with the bull market.

This might be (and I hope it is) just the symptom of a wild bull market where "grandmother-safe" stocks (like Verizon) get left behind. But it might also be because I am wrong.

I sincerely hope for some really knowledgeable spectrum engineers to tell me where I am wrong on this. Comments are appreciated either on the blog or by email.


Spectrum usage - the basis for valuation

Imagine we are sitting in the room and I have a purple flashlight. I can transmit information to you using that flashlight - by flashing it.

In the old days I might have used morse-code to do it - and I could probably get this to work at something under typing pace.

As computers have improved - and the light flashes faster and faster - I have been able to get more and more information into my signal. With telephone signals you get about a 10X improvement by going from analog to (2G) digital and get another large lick between 2G and 3G and then 4G. By the time we get to LTE for everything (including voice) this will largely have played itself out.

The purple light is flashing nearly as fast and efficiently as it can now.

Now suppose I have a room full of people and I wanted to flash my purple light to send everyone some information.

One way is that I could "share the channel" by sending a bit of my signal to everyone in the room usually with some kind of program whereby they can pick out their bit of the signal and everything else gets discarded. One example of this was "code division" as in "Code Division Multiple Access" or CDMA.

The problem is that as I share the channel everyone's gets their information a bit slower. I need to wait whilst information is sent to everyone else.

And if I try to do it too much I just cram up the signal.

I could try a different approach - which is to use multiple purple flashlights and flash signal at full pace to everyone separately but I will have another problem.

The room will be flooded with purple light.

And that purple light will degrade everyone else's signal.

That degradation has a name: "interference". Too much interference and the speed at which I can get information slows down - often dramatically.

There are a few solutions to the interference problem.

One solution is that instead of having a big tower flashing purple light to everyone I have lots of tiny little towers that flash a low powered purple lights that has an effective transmission distance of a few metres. In an ideal world everyone has their localised purple light which gives them information and does not interfere with their neighbours.

Alas that idealised solution requires lots of capital expenditure (you have to build a tower every few metres).

Another solution is to make the purple light into a beam and beam it at every individual person - so they all get their own signal. This has a name - "beam forming" and it the basis of many proposed 5G phone systems.

Another solution is to use the mathematics of interference to my advantage - design my system so that only my signal survives - and everyone elses' is cancelled as white noise.  There are variants on that - but the most famous is MIMO and it winds up being very computer intensive. However you can reasonably get another half order of magnitude of wireless efficiency via this method. It is also the basis of Steve Perlman's wireless technology. I am deeply skeptical of Steve Perlman's project - but I am sure there are people out there more knowledgeable than me.

Another solution is to encase my purple light in glass and send it individually to each person. You can get a vast amount of information this way (fibre-optics) but you lose mobility.

By far the cheapest solution though is to just use another colour.

So I flash in red, violet, indigo, yellow and a bunch of colours you can't see (but are still really colours).

Flashing my signal in multiple colours is an alternative to building more towers. This is a pretty direct trade-off. If I am allowed to use many colours I can get much more information out of a single tower.

When somebody buys "spectrum" what they are really buying is the right to flash light in a different colour. These colours are radio frequencies (say 700 MHZ) rather than visible light (say 700 THZ for the colour purple).

The right to flash information in a different frequency (that is a different colour) is an alternative to the practical obligation to build more towers.

As towers have an identifiable capital and running cost (a cost structure well known to phone companies) spectrum has a definable value - defined by avoided cost.

The point here is that if you know the cost of the phone company and the amount of capital expenditure avoided you can work out the value of the spectrum to a phone company. The phone company can either buy more spectrum or build more towers. [To work this out accurately you would need a map of the US, demographic and phone usage data by small region and try to work out how you would serve them.]

If you have a lot of information to transmit you can save a lot of money buying spectrum rather than building towers - and hence your spectrum is worth a lot of money. It is worth more if (a) the people are difficult to serve or (b) there is a lot of people willing to pay for speed or reliability.

This is where the value of spectrum in billions of dollars comes from. It comes from avoided cost.


A phone company with a good spectrum position will be able to avoid a lot of costs and still provide a very good service. This will make it profitable. It will be able to charge more (as it has reliable coverage) and will not have to build a lot of towers.

A company with a bad spectrum position will have a cost structure that - well frankly - sucks. It will be competed away.

When we think about the value of spectrum we have to think about the amount of cost that it avoids.

If it avoids lots of costs it is worth a lot.

If it avoids very little cost it is worth very little.


The rise of smart phones has increased demand for information a lot. Okay a huge amount. We all have the internet in our pocket these days.

And that has obliged phone companies to spend lots of money to keep up with demand. If they don't spend all that money their phone system will break down. [Lots of readers will remember AT&Ts problems in 2009 and if you do not here is Jon Stewart to remind you.] After the iPhone was introduced demand for data went through the roof (especially at AT&T who had a monopoly on iPhones). AT&T capital expenditure in the last five years has been roughly $100 billion - mostly on fixing this problem. These are not small sums and as spectrum is an alternative to capital expenditure AT&T most certainly wishes they had more of it.

The rise of smart-phones has driven up the capital expenditure requirements of the carriers - and as spectrum is an alternative to capital expenditure it has driven up the value of spectrum.

That rise in the value of spectrum (which really has been quite extreme) is the main reason for owning Verizon. They have lots of good spectrum which means good coverage at low costs and that means big profits. My view is simple here. The trend will continue and is Verizon's friend. Profits will be *much* higher in five years. The stock starts at a low teens PE ratio with a 4.4 percent yield and it is only going to get better.

Verizon wireless grows revenue at about 7 percent and will grow earnings faster.


What sort of spectrum avoids costs?

Spectrum has value because it allows you coverage and capacity at lower costs.

But not all spectrum is equal.

There is really good spectrum. That comes at low frequency and with very few restrictions as to the power output.

And there is poor spectrum which is either very high frequency or has restrictions as to the power output or both.

Low frequency spectrum is good because it propagates really well. It can go through buildings so you can use your iPhone whilst sitting on the toilet in the centre of your concrete office block. It can travel decent distances with limited regard to obstacles. In the wireless world low frequency tends to mean 600-900 MHZ.

In every country in the world there was a big block of this "beach front real estate" taken by the TV channels. Why? Because when they were allocating spectrum in the 1950s there were not many demands and the TV stations asked for and got the best bit.

In every country in the Western World the TV channels are being forced to go digital and their extra spectrum is being sold off. That sell-off has been the basis of most of the big spectrum auctions that have taken place.

In the last auction AT&T and Verizon were the big purchasers - and those blocks are where the LTE/4G networks have been rolled out.

However this is not just "low frequency" spectrum but it also has very few restrictions as to the power of transmitter allowed. Typically Verizon is allowed a tower hundreds of feet high on which they can put a 1000 watt transmitter. This allows Verizon the ability to cover a large amount of area (often hundreds of square kilometres) from a single transmitter. The most valuable spectrum at the last big auction covered the affluent suburbs of Chicago - middling population density and thick insulating walls meant that low frequency spectrum was required. Affluence meant that people would pay for the premium service.

You would never use the spectrum that way in Manhattan. There is simply too much demand for it. However you can still use smaller, lower towers and get good coverage inside buildings.

The two things that make this spectrum value is frequency (low) and power restrictions (very few).

High frequency spectrum is less good generally. It doesn't propagate as well. At very high frequencies (say 2500 MHZ) it can't get into a building and so you lose coverage as you go into elevators or behind trees. You certainly lose coverage when you go into the concrete core of your building (ie the toilet) and try to use the phone.

High frequency spectrum does however have one big advantage.

There is more of it. Much more of it.

And so it gets extensively used for capacity when you really need to move a lot of data.

If you are going to build a wireless network in Tokyo or Seoul or New York you are going to need a lot of spectrum (because the density of use is high) and some of that will need to be high frequency (simply because there is not much low-frequency spectrum and there is a lot of high frequency spectrum).

The poor propagation of high frequency spectrum can be used to your advantage. High frequency spectrum has a hard time getting into a building resulting in poor indoor coverage. A solution is to "light" the building from the inside - putting a transmitter say in the air conditioning duct. This makes for very good internal coverage.

Moreover as the high frequency spectrum can't normally get into the building in this case it can't get out of the building. This means the spectrum used in the building does not interfere (much) with the spectrum used outside the building allowing reuse of spectrum.

Very high frequency spectrum (say upper-band WiFi at frequencies of around 5000 MHZ) can stay inside individual rooms of a home. Low interference is good and means the WiFi never gets congested. The downside is low propagation even within your home. [If you try using your iPad in bed on the other side of the house you will understand the problem.]

Verizon's spectrum mix (unequivocally the best in America) consists mostly of low frequency 700 MHZ "beach front" property with another amount of high-frequency spectrum (the so-called AWS bands) which are used primarily for urban infill. They have some frequency at higher levels too however they have not traditionally had any interest in very high frequency spectrum (eg 2500 MHZ).

Power restrictions

Driving in the country at night you know that a big city radio station (even one a very long way away) can drown out a country station. Radio - especially AM radio - is very low frequency (say 0.6 MHZ) and can propagate a long way (hundreds or sometimes thousands of kilometres).

There is not much spectrum this far down the dial so you can't have a lot of stations. And they interfere with each other a great deal. The big city station has a power output that will simply dwarf a regional station.

Not everyone will be allowed to broadcast with high power because something with high power will (badly) interfere with something at lower power a little way up or down the dial.

All rights to use spectrum come with a power restriction. That power restriction can be for a trivial amount of power or can allow you to use 1600 watts from a 300 metre high antenna.

Obviously being able to pump out a large amount of power from a high tower with a great line of sight will be cheaper than putting a smaller transmitter every 50 meters.

The higher the amount of power you can pump out the higher the value of spectrum.

Lightsquared and power restrictions

Power restrictions can completely neuter the value of your spectrum. LightSquared had some spectrum that was originally designated for satellites. Phil Falcone bet his hedge fund and his reputation on the possibility of getting that spectrum approved for terrestrial use. If he could it was very valuable.

The only problem: that spectrum was near (in frequency) to the spectrum used for the economically important GPS satellite system.

The Federal Communications Commission (FCC) said that was fine so long as independent tests verified that there was no interference.

Phil Falcone took that as the go-ahead - the right to use the spectrum.

Alas the independent tests came back nasty - a phone tower interfered with an aircraft navigation system 20 kilometres away. Approval was denied.

Now I am going to make the obvious point. GPS signal is broadcast by a satellite running on solar energy and batteries hundreds of kilometres away. A phone signal comes from a large tower hooked to mains power.

There are 11 orders of magnitude difference of power between the GPS signal and the phone signal. If the colours were not very different (say a slightly different shade of purple) then there is simply no way that the GPS signal could be seen. The way to think about it: try spotting a 1 watt LED versus the entire power load of a nuclear power station converted to light when the light is a similar colour. It can't be done.

LightSquared demonstrated one thing: Phil Falcone is so egotistical he thought his will could out override the laws of physics. It cost him his fund and his reputation. It cost his clients an awful lot of money.

Alas this seems common enough on Wall Street. I have seen the most absurd things being said about spectrum as if power restrictions do not matter. Power restrictions absolutely determine the value of the spectrum holding.

Quick summary as to what is valuable and what is not

Low frequency spectrum with limited power restrictions is hugely valuable.

High frequency spectrum with only modest power restrictions is less valuable - but is very good for urban infill. Once you have covered most the nation with low frequency spectrum you need to do mid-range or high frequency infill. This is the stage Verizon is now at - using AWS spectrum for infill.

High frequency spectrum with dramatic power restrictions is worthless. A high frequency spectrum with a binding power restriction can't get into a building and can't go very far and you simply can't cost-effectively build a network using that stuff. Besides it competes with WiFi - and WiFi is free - and competing against free is difficult.


 There is a lot of WiFi spectrum, some at 2400 MHZ but most around 5000 MHZ. It is unlicensed and anyone can use it. It is high frequency spectrum which makes transmission through the walls of your home difficult (as anyone who has tried can attest).

There are some rules for using it - concerning how you are meant to share channels. The most important rule however is a power restriction. Power restrictions on WiFi are binding and a WiFi signal has an effective range of about 70 meters. [The range will depend on how many walls you have and also on other WiFi signals and how good your antenna is.] Whatever - it is useless over long ranges.

The power restriction is absolutely critical to WiFi. The reason why everyone in a street manages to use WiFi in their home without causing critical interference is that everyone abides by a power-restriction. [You cannot legally sell a WiFi access point that does not abide by the power restriction.]

WiFi carries a lot of data and it carries it by the mechanism of giving everyone a very low powered purple light which they put right next to them. But instead of just purple it has about twenty colours (known as channels). You can carry a *lot* of information by WiFi - multiple high-definition TV channels even - but you can't carry it very far. And for the most part the consumer pays for the infrastructure [ie the little purple light] which is nice for the carriers.

Point here: low-powered high-frequency spectrum competes with WiFi and you can't do that very effectively because WiFi spectrum is fairly abundant (caveat in the next section) and free.

The caveat: WiFi in the Superbowl Stadium versus WiFi in Times Square

There are about twenty useable WiFi channels and it is not often that twenty people want to be downloading something simultaneously within say a 100 foot radius.

When this happens you run the risk of WiFi congestion.

There are prominent places when far more than twenty people want to use the same spectrum simultaneously. The Superbowl stadium at half time has 70 thousand people in it all wanting to film, take pictures and upload the stuff to the internet. The Superbowl stadium at half-time is probably the most congested spectrum in the world.

The WiFi (and phone systems) in the last Superbowl worked. [It had to work - it was sponsored by Verizon. Verizon would have had very bad press if it had not worked.]

In Times Square at the queue for half-priced tickets everyone seems to want to download theatre reviews (as much as anything to see what to see).

WiFi does not work. I know. I have tried it.

It is worth explaining why.

Even though there are rules unlicensed spectrum suffers from the problem of the commons. Too many people in an uncontrolled fashion want to get onto it. The commons however are not very big (a couple of hundred feet radius mostly) and most commons (say the one in my suburban street) have a limited number of people who want to share. So "commons" is not a big problem.

In Times Square "commons" is a big problem. Way too many people fit into my small radius and the network is congested.

At the Superbowl that is also true - but for one thing. There is only one party managing the network in the Superbowl stadium. Signal from outside the stadium can't get in. (The walls are too thick!)

So within the stadium the unlicensed spectrum behaves as if it were licensed spectrum. Sophsiticated WiFi equipment (such as Meraki) include "rogue network detection" as as standard feature. If you were setting up a wireless network for (say) Stanford University where there were a gazillion devices you would want to ensure that rogue networks were prohibited and to ensure you own the spectrum in those places. Rogue devices are jammed and "deauthorized".

The FCC has recently pronounced against jamming rogue networks - at least if you want to charge for an alternative network. The FCC recently fined Marriott Hotels $600 thousand for jamming a rogue WiFi hotspot and then forcing people to pay for hotel-provided WiFi.

My guess is that jamming remains legal (it is certainly supported by equipment manufacturers) but it is only legal if you give away the WiFi signal on your controlled network.


GSAT - the latest victim of Sahm Adrangi's precision analysis - is the owner of two blocks of spectrum - both originally allocated as satellite spectrum.

One lot is clearly worthless as it even closer to GPS spectrum than the Lightsquared spectrum. Interference is a given. There is no way they will be allowed to operate at any meaningful power level.

The other lot of spectrum is adjacent WiFi and will never be allowed by the FCC to be substantially more powerful than WiFi spectrum for fear of drowning out WiFi. GSAT proposes using their spectrum with a 4 watt power limit. [Verizon spectrum is low frequency and often has a 1000 watt power limit.] This will mean it has a range commensurate with WiFi.

What GSAT are proposing is a private WiFi channel.

But for the most part there is enough WiFi channels.

In football stadiums or Grand Central where rogue networks can be removed there is plenty of WiFi capacity. [This was demonstrated by the fact that your phone worked during the last Superbowl.]

This applies in hotels too (provided hotels give up charging for WiFi and hence don't fall foul of the Marriott ruling).

In places where rogue networks can't be excluded like Times Square another WiFi channel will be nice. But there are not that many such places. Moreover one channel will get congested for sure in those places... after all if just two people want it you have a difficulty...

Simple summary: GSAT has nothing in the spectrum game. High frequency spectrum with a binding power restriction is worth something near zero. It is a non-starter.

It is hard to see what capital costs GSAT's spectrum helps a carrier avoid. And if it doesn't help them avoid costs then the spectrum is worthless. [Sorry guys.]

That said there are plenty of hedge fund managers who hold GSAT on some vague notion that spectrum is scarce and hence valuable.

I know you read this blog. I suggest you start your next letter as follows:
Dear Limited Partner 
As you know we hold GSAT even though their spectrum is restricted to low power output and is of a high frequency and hence will not have a range of more than a few hundred feet. It also competes with WiFi spectrum within that range. 
We however think that the laws of physics do not apply to us because...
Low frequency spectrum - and the incredible value it offers

AT&T and Verizon have the lion's share of low frequency spectrum in the US. [Both Sprint and T-Mobile have narrow slices.]

These large holdings of low-frequency spectrum mean that Verizon and AT&T can offer better coverage at lower capital cost than their competitors.

The low frequency spectrum however still has limited capacity. If you want lots of capacity you have to use high frequency spectrum (simply because there is much more of it).

If you have low frequency spectrum you advertise yourself on network reliability ("Can you hear me now") but you offer limited and often small data plans.

If you have a high-frequency network your reliability is going to suck (sorry Sprint) but you can offer very large (even unlimited) data plans.

To the extent that you can charge for reliability and you do not have to carry huge amounts of data you should have pricing power AND more modest capital expenditures. You should make pots of money.

We have almost pure comparison between a high-frequency player and a low frequency player in the US - an that is between Verizon Wireless and Sprint.

Verizon Wireless was a separately reporting subsidiary until the end of 2012 and reported its data (courtesy CapitalIQ) can be found here. You can see it did something like $76 billion in revenue in 2012 and the capital expenditure was 8.8 billion (which was its all time high). The capital expenditure was the all-time-high because of the roll-out of the LTE network. Since then Verizon Wireless revenue has risen. Here is wireless revenue by quarter:

(dollars in millions)
Operating Revenues
Retail service$ 15,538$ 15,786$ 16,169$ 16,422$ 16,776$ 16,967$ 17,246$ 17,288
Other service616607559656740744741790
Total Operating Revenues19,02419,99419,52319,97620,39921,12520,87921,483 

You can find the original here.

The growth rate is about 7 percent. It is pretty good.

Operating income for wireless is growing faster because there is relatively fixed costs - and thus operational leverage. Operating income for wireless is now running over 7 billion per quarter and the growth rate was 14 percent in the first quarter but only 7 percent in the second quarter. [There were some unusual expense shifts - my guess is the growth rate is about 10 percent.]

Capital expenditure for wireless is currently running a little over 10 billion per year (it has gone up - all that urban infill is not cheap). They are guiding it down in a couple of years which makes sense (but I guess has to be seen to be utterly believed).

By contrast here are Sprint's numbers. [Again you can find the original courtesy of CapitalIQ.]

Revenue is only 35 billion per year - less than half Verizon - and it does not grow very much.

More importantly capital expenditure is running at roughly 6 billion per year. Again they are guiding for it to fall somewhat - but it can't fall much because of the lack of coverage. What we have is less than half the revenue, roughly 60 percent of the capital expenditure and a lack of growth. It is not pretty.

Capital expenditure has been SUBSTANTIALLY larger than operating cash flow for many years. By contrast Verizon Wireless has substantial operating cash flow after capital expenditure. The spectrum holdings of Verizon have avoided vast capital expenditures and provided better service and pricing power.

Sprint by contrast has large and increasing funding needs. They needed a huge capital injection from Softbank. My guess is that they will need another one.

T-Mobile whose spectrum position is similar to Srpint also has capital expenditure substantially larger than operating cash flow. Alas their parent is not as rich. My guess is that contrary to the popular perception Sprint and not T-Mobile is the ultimate survivor - but only because Sprint is controlled by Softbank and Masa Son will decide at some point he prefers owning a solvent number three phone company than owning a large stake in Alibaba.

Some people (and guidance) think that T-Mobile becomes cash flow positive (after capital expenditure) when it finally turns off old 2G networks. I do not believe it as I think their unlimited buckets now will force them to infill for a very long time. [The 2G network thing will also apply to Verizon.]

The Verizon question

Verizon wireless grows revenue at about 7 percent, income at about 10 percent and according to management is likely to drop its capital expenditure.

It is priced at a teens PE ratio.

Of course it includes a wireline business but that is no longer shrinking. Why? Mainly because it is the opposite end of all those WiFi access points.

What is is worth?

My guess is a *lot* more than the current price. I can't see any reason why the demand for bandwidth will not be a lot higher in a decade and the ability to solve problems by digital compression (ie 4G over analog) is nearing an end. There is likely going to be a true crunch and that crunch will provide pricing power.

But bluntly you don't need that. All you need is the revenue growth to continue at 4-7 percent for half a dozen years, profit growth is thus 8-14 percent and the free cash is overwhelming. Verizon is very cheap it continues. It starts with an almost 4.5 percent dividend. If it gets better at that rate for half a dozen years (still an open question) Verizon is really cheap. In order for Verizon not to be good value that revenue growth needs to stop or the capital expenditure requirements need to blow out.

Some people state the revenue growth will stop because there is a price war going on in US telephony. I have spent a lot of time looking at it and think it is mostly a phoney war (forgive the pun) but it is clearly impacting Sprint whose revenue is stationary but whose customers

I would love (off-the-blog) a decent conversation with anyone who has a decent idea on how to value Verizon's spectrum. I have a list by county of the US of their spectrum holdings (its a 25 megabyte file).

Is there any more spectrum?

It is worth asking whether this can be solved by making more spectrum available. Are there great "fallow fields" of spectrum?

The short answer is no.

The long answer is a little more complicated.

The available low-frequency spectrum has by-and-large come from freeing up the spectrum previously used for TV. In most countries (though not America) all the auctions have happened.

In America there are two major releases of that spectrum. The first was in January 2008 - and the second will be in 2015 (the so-called incentive auction).

The 2008 auction was bid just after the introduction of the iPhone (announced 29 June 2007, introduced a little later). The bidding prices were not high. Spectrum demand and prices have gone up massively since the introduction of smart-phones.

The 2015 auction will be the last auction of high-value low frequency spectrum in my lifetime. There are no more huge blocks to be had. The FCC has reserved 30 MHZ for minor players (not AT&T or Verizon). That reservation is a huge gift from taxpayers to the minor carriers and will allow at least one of Sprint or T-Mobile (or maybe Dish) to prosper. For the minor players the 2015 auction is their last chance to get some high-powered low-frequency spectrum. It is do-or-die.

Because Sprint has a richer parent (Softbank) my guess is the winner will be Sprint.

At high frequency there is not really a shortage of spectrum. There is a lot of WiFi spectrum if you want to build a very dense network (at very high cost). But more realistically Sprint owns (courtesy of Clearwire) huge swathes of very high frequency spectrum with much more realistic power limits than WiFi. Sprint is able to offer a lot of capacity - but as the spectrum is high frequency they might have trouble offering it over a broad market.

Future research

I am seeking anyone who can help me value Verizon's spectrum. I want to develop a physics-based model of Verizon's competitive advantage. Hope dear readers that you can help.



Will said...

I tend to think that all cell networks just have too many similarities to how landlines were 20 or 30 years ago. The large infrastructure costs and the threat of emerging technologies makes the whole space vulnerable.

E Kvendseth said...

John, thanks for a great technical introduction to US telecom (very interesting for a stupid financials guy). While you outline why VZ is certainly in an enviable situation when it comes to spectrum, wouldn't it also be reasonable to address the biggest risk that is facing telecom and cable providers in the US - the unbundling of services? If we see a similar development to what's already happening in Europe, isn't it fair to assume that VZ profitability could take a pretty big hit? Would love your thoughts on this.

Dan James said...


I'm curious about how you weigh T-Mobile's move to essentially allow all services over local WiFi.

If you consider coverage from less of a geography consideration and more of a use case scenario - people predominantly use their phones at home, at work, at the gym, while commuting, at the coffee shop.

All of these places (with the exception of most commutes) are wifi enabled. My parent's house in Charlotte, NC has terrible tower coverage by AT&T and TMobile, but with WiFi my blackberry will soon work just fine there.

There are key use cases left over (getting directions, looking up restaurants while on the road), but there is a reasonable belief that as telecoms start to allow local WiFi to supplement as coverage, spectrum become less valuable.

Essentially, the avoidance cost for towers goes down as companies will be able to utilize crowd-sourced infrastructure instead of shouldering the build cost themselves.

Anonymous said...

Does it matter if Verizon is very difficult to deal with as a customer? Could this be a hidden liability that would be hard to lose if a formidable competitor showed up (Google Fiber or other technologies)? Not trying to make a point, but asking for thoughts of OP in this regard.

Anonymous said...

I believe you meant to write "Sprint" not "Verizon" in the following sentence.

"T-Mobile whose spectrum position is similar to Verizon ..."

Anonymous said...

you say L2 spectrum is worthless. yet Ergen empties his daughter's entire trust fund to buy up the LP debt. and meanwhile JPMorgan is proposing to inject $1.1 BILLION in form of debt and EQUITY ($450 million in equity to be exact) as part of the new reorganization plan. so either those parties are complete idiots - or to me this seems like just because there isn't a currently identifiable use case for certain bands of spectrum, doesn't mean it has no value like you suggest.

and btw, VZ currently has the LOWEST downlink spectrum per service revenues out of all the carriers. and NO you can't build towers to infinitely and beyond to get around this issue. so your long case is probably going to end up wrong, especially if someone like Softbank decides to commit serious capital to the US wireless market

Anonymous said...

Even with the threat of new technology, the cost of implementing it to cover the United States will be enormous. Only VZ and AT&T will have the cash flows to fund such a project. Bronte, have you graphed the capex % to revenue for V and compared them to all their competitors? VZ is spending more on capex in dollar amount, but it is shrinking in terms of revenue %. This is a runaway effect; VZ will always be able to afford better equipment and better service without reducing free cash flow to investors.


Anonymous said...

Great thought provoking post.
As a Verizon customer with 4 family iPhones, I can tell you that Verizon is the high cost carrier in the USA. I am considering switching my family plan to Tmobile in 2015 as with them you buy a chunk of high speed 3G data, when it's done, it throttles down to 2G for free). Verizon either hits you with overage charges or you have to agree to cut off your data at the limit.

You are right, Verizon has the best and most reliable voice/txt coverage, but I expect Sprint to initiate a huge priceware with Masa-san's cash, which will make the savings to great to NOT switch.

Colin Lee said...

I have verizon in las vegas.. Tmobile is much better than Verizon here.

Tom L said...

I agree that Wifi is not a perfect solution for all situations, but it is a great solution for the vast majority of situations.

For example, I recently dropped my Verizon call/text plan at $45/month for a Republic wireless plan at $10/month that has unlimited call/text/data over wifi. It works great for me, and I believe it will solve most users needs most of the time. I acknowledge that it will not work well in Times Sq, but that is a rare event.

This seems like a big risk. This solution is only beginning to play out.

Les said...

One way to increase capacity is to use existing copper infrastructure bonded to increase bandwidth create macro small cell LTE
( disclosure- I'm a shareholder)

Anonymous said...

Low band (<1GHz) spectrum values are higher because of superior radio propagation, but you can mimic coverage quality by building more sites at higher frequencies (of which some sites have to be in building to get around the problem of penetration through glass, steel, concrete etc at ~2GHz). The problem is that the later entrants did not have cash flows to afford the much greater site densities.

Sprint now has a small amount of 800MHz, T-Mobile is buying 700MHz and eventually they will both get reserved 600MHz (assuming they can afford), so the playing field will be more balanced.

While VZ was very smart in buying up spectrum early on and created its market share lead, it now looks light on spectrum for its traffic compared to everyone else. They cannot densify networks at 700MHz because of interference, and are using AWS spectrum for that now, but there will come a time that they need a lot more spectrum as traffic keeps doubling every couple of years (or accelerates).

Spectrum valuation is irrelevant except in bankruptcy or if you are buying or selling. Once it is put into use, it is what you do with it, not what it might theoretically be worth. Hence would not value VZ that way, but think about spectrum as a liability for VZ going forward to top up their capacity over time. Similarly for say DISH, spectrum is seen as separate asset today but if brought into use (eg via spectrum host or merger with Say T-Mobile) simply becomes a productive asset like machinery in a manufacturer and drops out of the valuation to be replaced by its value in use.

Unknown said...

Contact your local IEEE and sponsor a junior engineers program project. They are very cheap, ver appreciative and very competent.

PSC said...

A few points:

2.4 Ghz (and 5.8Ghz) are really crap in part because of microwave oven interference. Microwave ovens emit at this frequency because it's absorbed by water. So to some degree water (e.g. rain) attenuates this frequency.

I'm very happy with your mundane "cash generating bemoth with nice growth at a lowish P/E" thesis.

But I think you're overestimating capex for wireless. Implicitly you're saying wireless capex is very high for any carrier, . But whole world LTE equipment sales was around $30bn for 2013, and Verizon total capex was around $16bn - which says that not too much of that $16bn is for LTE gear. I'd suspect it's in the order of 2-3bn for Verizon.

Ben said...

John, while I agree Verizon has scale advantages, Sprint just put in place a former handset distribution entrepreneur (Brightstar) in as CEO. They are going to cut costs dramatically and compete on price. T-Mobile is also competing on price. Verizon's biggest spectrum advantage is outside of urban areas, they have a quasi-monopoly or duopoly in most rural areas. However, in cities, since they have more customers and Sprint and T-Mobile have more towers than Verizon and AT&T, they can effectively compete with higher band spectrum. Also, Masa's plan in Japan was to introduce a huge number of distributed antenna systems that aren't on towers to make use of his 2.5 MHZ spectrum. I think your post oversimplifies the situation massively. While Verizon may be an ok investment, they have to contend with pricing pressure from aggressive competitors and their network is decent but actually in most urban areas Sprint and T-Mobile are pretty good too. Sprint's network has been massively disrupted by their rip and replace strategy to deploy their Nextel and 2.5 GHZ spectrum but should improve now.

I think the idea that Masa would want to own a 3rd player in wireless rather than Alibaba is absurd. First of all he doesn't need to choose, his Japanese business is going to throw off tons of cash and Sprint is going to improve its margins through cost cutting. Second of all I think his plan is to wait until a new administration and then try to buy T-Mobile. Third, Alibaba is a way better business than Sprint (and Verizon) and has a lot more upside over time.

Anonymous said...

I have no ideas whether Verizon is cheap or not but I wonder how much impact the terrible customer service they offer has on the business. From the sidelines, it seems to me that T Mobile under J Legere has strived to be customer friendly and is actually on the way of getting liked by customers - not an easy feat in this market.

The rumoured Iliad (Free) offer to buy TMobile may be a long shot, but looking at how Iliad totally disrupted the French telecom market seems to me to provide a clear roadmap for TMobile (and also Sprint) on how they could take on the duopoly AT&T/Verizon. Iliad essentially hard discounted its services, basically accepted a lower rate of return, in the process destroyed its competitors' businesses. As an externality: the French now have some of the best and lowest cost telecom services in the OECD. Given how Americans are being ripped off by their cable and telecom firms, isn't there a large hole for a competitor to do the same?

Also - Google Voice got updated, MMS is now working, on pretty much all North American networks, EXCEPT Verizon. I dont know if this is a Tech issue, but if not, this is the sort of situation which antagonises customers and has the potential to cause damage.

Last point: I seem to have read that it is now possible to carry 4G network over existing Wifi routers - again I dont know the tech side that well- but the point was that provider you could get enough broadband customers, that by itself would give you a mesh on which to carry/relay your cell network, this reducing your tower/spectrum costs.

Anonymous said...

Hi John - what am I missing?
Is there not a wealth of literature on how to value VZ spectrum?
Easy DCF and bob's your uncle...

Anonymous said...


This is completely off-topic.

Here is a recent article which might be of interest:

The author seems to suggest that your hero is not quite the hero he pretends to be.

Perhaps the author is being a little unfair to single out your hero for criticism.

You see, if you look at the broad sweep of so-called "tax reform" in Australia since the mid-1980s, it is difficult to escape the conclusion that all the "tax experts" in Treasury are clowns.

No need to publish this little contribution on your blow.

Coleman said...

Won't the trend towards seamless switching between WiFi and Mobile spectrum kill hi freq fill in?

I don't know if this feature already exists with others, but it's just been added to the new iPhones.

Anonymous said...

What is the point of valuing Verizon's spectrum? It will never be sold, and even if it was, it would be to someone who would monetize it just like verizon is doing. The "value" of the spectrum is the value of the cash flows it can generate - which is exactly what Verizon is doing.

I don't understand the point of the exercise. Even if you thought the value of the spectrum was 2x the value of Verizon, now what? Verizon will always be valued off of cash flows and the dividend.

just mike said...

you make a number of errors with regard to GSAT in your thesis.

GSAT does not propose to use its spectrum for wifi as you suggest. they will instead offer a service called TLPS which is 5x faster and has 5x the range. hence 5x the value. do the math.

wifi is full. this is according to the FCC. As they are the head honcho in charge of managing this resource I think they are right and you are wrong.

Any crowded public place needs more wifi. your anecdotal experience means nothing compared to industry, military, and government reports.

there may be enough wifi channels for you, but there are not enough wifi channels for industry, military and government. see above.

lastly you ignore the roll TLPS will play in a layered wireless network, this may be because you do not understand how this technology really works or may be some other agenda you wish to promote. regardless, an omission of fact is still an error.

duaneg said...

FWIW I think you are wrong when you say, with respect to the Marriott fine, that jamming is legal if you give away bandwidth.

My understanding is that jamming is always illegal under FCC rules. As you say, inside the superbowl no jamming is needed for WiFi signals as they will not penetrate.

If you wanted to setup a portable hotspot tethered over your cellular connection, however, you can do that and no-one has any right to stop you.

There was a recent monster thread on NANOG about this, if you want to learn more from the people who quite literally run the Internet in North America:

Not that any of that is important to your thesis, I suppose.

Om that point a friend of mine is an analyst with Telco 2.0, who deal with this sort of stuff as their bread and butter. If you want expert opinion I'd recommend contacting them, or someone similar (the IEEE would be a very good option, as recommended upthread).

Anonymous said...

John, there are several local universities that have very competent staff and research students in electrical and telecommunications engineering as well as physics. Find the head of the department at the uni of your choice, ring them up explain what you want, then get them to recommend a bright student to you for a casual job over the summer. I'm sure a broke university student would not only be grateful for the work but also the experience. Uni of Wollongong, Newcastle, UTS and Sydney uni all have well regarded engineering faculties. Most university websites also have a find an expert search function.

John Hempton said...

Just Mike - there is very little you can do with that spectrum you can't do with WiFi and it will get congested.

But I love incoherent bulls. It makes me much happier with regard to my short!


John Hempton said...

As for the person who thinks Judith Sloan is the person to comment on the Henry legacy...

She always was a clown.

Remains so to this day.


Anonymous said...

If I understand your thesis right, it is:

low freq spectrum + high power limit = less towers/infrastructure = lower capex = higher free cashflow

I was chatting with a friend in the industry and he siad this is a part of it and maybe more of a case in less dense areas. However, in dense cities, you still have to put up more infrastructure (i.e microcells, etc) just to ensure the quality of the access you provide. Thus, the spectrum is just one leg of the stool but it does not necessarily automatically confer you an enormous advantage.

My other thought is that I am not convinced spectrum is limited since I don't see a reason why the government can just take something like FM spectrum and force it to go digital like the TV spectrum. So wouldn't that just create more supply?

At the same time, I am also long VZ but since I think there are too many examples of countries where a 4th provider has not survived. Therefore the industry has a high chance of ending up with just 3 competitors. One other reason I am long is that I think as people use more wifi, they are conditioned by habit to want high-speed internet anywhere they go so it will create an attachment to the service.

Jay Goldberg said...

A couple things to keep in mind.
1) Carriers really do not fully understand their costs. Verizon is better than anyone at this, but even they struggle to measure something that is fundamentally difficult to measure. (What is the marginal cost of infrastructure?)
2) Capex figures from the carriers are a bit mis-leading, more than 50% of reported capex are siting fees (i.e. permitting) and labor (the cost of the installers). Siting costs are a variable cost and fit with your thesis, but the labor costs are somewhat fixed and heavily unionized. None of the carriers (including Verizon) have as much leverage as you might think.
3) Another problem with the fixed cost leverage you mention is that the carriers have considerable opex associated with adding coverage, not just capex. The big costs here in maintaining each cell site are things like rent for the location (big number in urban areas) and backhaul which is the connection from the base station to the core network (big number in less urban areas). And as we move to 4G, these costs go up because the carriers cannot rely on installed copper but have to upgrade (or lease) fiber.
4) You compare Sprint and Verizon’s capex numbers, but this is somewhat misleading. Sprint is a combination of multiple large acquisitions. As a result, they are operating three different networks – CDMA (Sprint), iDEN (Nextel), WIMAX (Clear). There are big economies of scope possible by consolidating down to one network, and having all those legacy costs with distort capex. They are still spending on the labor to maintain those networks, even as they migrate customers off of it and revenue declines. Revenues will go to zero on the old networks before capex does. New management team should not be penalized for past bad decisions especially when they seem committed to shutting down old networks.

So while I agree with your overall assessment of Verizon’s spectrum position, I’m not sure the financial model works as you lay out.

JJ said...

John did you consider the effect of carrier frequency on antenna size? rule of thumb is that if you increase the wavelength you proportionately increase the dimensions of the antenna. I am not in this field any more but if you move from am antenna that has a lower frequency bound of 850 MHz to one which is using a 700 MHz carrier, it may not be trivial to develop devices of the same form factor without compromising the sensitivity of the device.

PSC said...

Thinking more - there's something funky about your AT&T $100bn over 5 years capex number. Have a look at:

See page 3. These guys (PWC) put the total mobile capex worldwide at $35-45bn roughly over the years in question.

These guys:

Put US mobile capex for 2012 at $10bn. If the US is roughly 25% of world mobile capex (which sounds about right) then these numbers square up.

So either AT&T is burning up half mobile capex worldwide and something like double[!?] US mobile capex .... or there's something funky with AT&T's capex .... or there's something funky with your 100bn over 5 years mostly on mobile arithmetic ....

just mike said...

j. where I worked(building satellites btw) a snarky rebuttal generally requires logic and reason to support it. you say there is very little you can do with gsats tlps that can't be done with currant wifi. can you explain you reasoning?

btw, long or short has nothing to do with physics.

as I said do the math: if you are familiar with the bands in question you know that there are three channels and an adjacent buffer zone. gsat owns 1 channel and the buffer, this gives gsat 50% of the current capacity.

since tlps has been shown to run 4-5x faster than wifi it can easily double what wifi does. and don't forget that is at 4-5x the distance. the only thing required is current equipment and a software upgrade.

let me spell it out, lower build-out costs due to higher capacity/greater distance = more money. they never said they were doing anything different, just more of it.

further everything you say flies in the face of the fcc and leading providers

who should we believe? someone who is asking for help to value verizons wifi? or a group of professional engineers that have spent over a year testing and verifying the systems performance?

(btw everything I have said is all published on the fcc website and is part of the public record, it has nothing to do with me. if you have an argument with it you should collect your data and post you comments to chairman wheeler as he has said TLPS is the only immediate solution to the wifi shortage, if you can show him cause to change his mind before approving this I am sure he would be grateful.).

John Hempton said...

The person who thought there was something funky about the wireless capex numbers does not get that most the capex is in siting, approval, wiring, powering and not in the box.

China Mobile Capex is alone larger than the number you link to.

HSNYC said...

Have you separated out the capex funding handset purchases?

Phil said...

Just Mike: I can't find any definition of TLPS which isn't GSTAR marketing material in one form or another, but they appear to be claiming to be able to use existing wifi chipsets with a simple firmware change to allow the use of the GSAR channels.

I strongly suspect that their "5x faster + 5x range" is marketroid speak for "we compared a single receiver + transmitter using our empty, unused channel with one using a congested wifi channel and saw a difference 5x the bandwidth and 5x the range"

This would be a) entirely unsurprising and b) uninteresting since it doesn't mean that GSTAR can achieve 5x the aggregate usage of a public wifi channel once they start selling bandwidth to all and sundry.

If you have any counter-evidence, do quote it - I'd love to see it.

Anonymous said...

Hi John,

I work with an telecom/infrastructure industry focused bank that can walk through some of the spectrum valuation discussions.

How would I contact you privately?


Australian Institute of Polish Affairs said...

Could I ask for clarification: if you have a US phone plan, does the company also include wifi access and have public wifi access points (a la what Telstra is starting to do, and I think BT has done for a while)? Otherwise, I don't understand the Times Sq/SuperBowl example....

Anonymous said...


You're exactly right, the 5 x 5 thing is a comparison of apples to oranges. Their spectrum butts up to the top end of the ISM band and if you use their 11.5MHz combined with 10.5MHz at the upper end of the ISM band you get another 22MHz WiFi channel that doesn't overlap with the three existing non overlapping available WiFi channels (1, 6 & 11) in the ISM band.

This spectrum is available for WiFi use in some countries as channel 14 and that's why a firmware update would make it available easily on many devices as it is already supported in the chipset.

Skimming though one of the documents they submitted to the FCC their argument seems to be that because they will control access to the spectrum they can perform a co-ordination and management function to deal with interference that is not available in the ISM bands today.

John Hempton said...

My blog is email is brontecapital at -

Please Alex - would love to hear from you.

Colin Lee said...

Do you have any thoughts on Apple's E-Sim card? VZ is not on the roster of wireless providers in an attempt to keep control of its customer. Interesting to see how it'll play out.

Unknown said...

"700 Mhz" is what i'd like to call "dumb spectrum".

For dense cities, it is a more a matter of interference in the network structure, more than anything else. Distance & power is an issue that can be resolved by modifying the deployment method, as long it is within the (2.5Ghz to Potentially 13Ghz).

The LTE-A model will shift past 700Mhz very quickly.

Happy putting a value on 700 Mhz. Telstra & Optus spent money on "dumb spectrum".

Unknown said...

On a second note, anyone who thinks 2.5Ghz & 5Ghz is crap clearly didnt do their homework.

Anonymous said...


Another great post! I understand that there's no technology at the moment to accommodate this increase in usage thus leading to a crunch. Why are you betting that such a technology will not be created in the future? As you can tell my technical understanding in this field is modest but technological innovations do tend to happen sometimes as a need for them arises, i.e. as we get closer and closer to such a "crunch". Would love to hear your thoughts.

Unknown said...

Not sure if this question is still relevant since I’m almost a year late on this. Nevertheless, I think the following would be a reasonable generalised approach to valuing spectrum ownership based on avoided cost:

At a high level, this appears to be an optimisation problem with a computational complexity of O(m x n) for m regions and n tower types.

Characterise tower types of different spectral frequencies:
* Coverage Area <- maximum penetration of spectral band
* Maximum Data Load <- lower frequency = less data
* Cost (PV of construction and maintenance, plus some discount for flexibility around upgrading)

Get a list of regions (e.g. counties) with data for:
* Population (as a proxy for data usage -> demographic data would improve this)
* Region area

For each region:
1. Eliminate all towers where the data provision density (max data load/coverage) is less than data consumption density (population * per capital data use / region area)
2. For each remaining tower type: find the cost of tower coverage as the number of towers required (region area/tower coverage area) times the cost per tower
3. Select the tower type with lowest total coverage cost
Aggregate costs across regions

You can then compare the cost implications of owning different spectrum regions by restricting the available tower types for each network provider.

I think the implication here is that the marginal value of owning low-frequency spectrum is only generated in regions where the data provision density of low-frequency towers exceeds the data consumption density of the area covered (as some of the earlier comments pointed out).

Potential Improvements
* Finer regions (e.g. by townships rather than counties)
* Circular coverage areas don’t tessellate well. To get full coverage with optimal circle packing, you need about ~21% more towers. Alternatively, you might get more cost effective coverage with a combination of low spectrum and high spectrum towers in each region
* It might be fun to strap a revenue model on top of this

Sandymount said...

General disclaimer

The content contained in this blog represents the opinions of Mr. Hempton. You should assume Mr. Hempton and his affiliates have positions in the securities discussed in this blog, and such beneficial ownership can create a conflict of interest regarding the objectivity of this blog. Statements in the blog are not guarantees of future performance and are subject to certain risks, uncertainties and other factors. Certain information in this blog concerning economic trends and performance is based on or derived from information provided by third-party sources. Mr. Hempton does not guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based. Such information may change after it is posted and Mr. Hempton is not obligated to, and may not, update it. The commentary in this blog in no way constitutes a solicitation of business, an offer of a security or a solicitation to purchase a security, or investment advice. In fact, it should not be relied upon in making investment decisions, ever. It is intended solely for the entertainment of the reader, and the author. In particular this blog is not directed for investment purposes at US Persons.