TEL v TLS

[Snoopy]

Telstra are quite up front about their mobile network in Australia covering 2,500,000km2 (PR2019, slide 15)

Spark are less forthcoming about area covered, preferring to speak about the percentage of the population that their network reaches.

The government of New Zealand is currently building a new emergency services communication network.
https://ngcc.govt.nz/public-safety-n...s-and-answers/

In rural areas this will piggy back off existing cell towers, the ones originally owned and operated by Spark and One (or Vodafone as it was then).

The above link tells us that roam across the Vodafone and Spark networks will create a 5% uplift in coverage (a 16,500 square kilometre uplift). We can assume that Spark are already covering the area served by more urban emergency response, and that their coverage is matched by 'One'.

0.05 x (Urban Coverage Area) = 16,500km2 => Urban Coverage Area = 330,000km2
=> total area covered by Spark Mobile Network = 330,000km2 + 16,500km2 = 346,500km2

That means the Spark mobile network covers 350,000km2 in round figures.

SNOOPY

[Snoopy]

A tabular comparison between the network investments made between Spark and Telstra and of the dollar values of 'capital spending' and 'depreciation' is notable.

	Spark {A}	Telstra	Telstra / 5 {B}	{B}/{A}	Reference
Transformative Assets Added Composition (prior to FY2022)	4G Mobile, Data Centres	4G Mobile, Inter city data highway	N/A	N/A	Posts 169 and 170
Transformative Expenditure Program (prior to FY2022)	$1,300m	$9,300m	$1,900m	+46%	Posts 169 and 170
Depreciation FY2022	$234m	$2,572m	$514m	+119%	Post 166
Intercity Fibre Highway (FY2022)	1,640km	5,210km	N.A.	N/A	Post 170
4G Network Area Coverage (FY2022)	350,000km2	2,500,000km2	500,000km2	+43%	Post 171
Network Cell Towers (FY2022)	1,273	8,000	1,600	+26%	This Post

In all cases, the population adjusted relative spend and the population adjusted size of the network is greater at Telstra. The figures are not all directly comparable though, as the differing Transformative Asset Composition' shows.

The common factor of each company's 'transformative change' is the building of an up to date 4G mobile network, a kind of network that is eminently suited for data transfer. A very notable difference is that Spark is keen to 'emphasize to we shareholders' (via the AR and presentations) , spending on their own 'in house' data centres. By contrast, Telstra emphasizes their connectivity spend between major city data centres by waxing lyrical in their AR on their spend on 'the fibre pathway'. This is the 'fibre pathway' it owns 'in house' (nothing to do with nbn) to connect the likes of data centres, whether they are company owned (InfraCo branded) or not (e.g. public cloud providers, like Amazon and Microsoft).

In fact Telstra, since FY2021, has operated an in house InfraCo branded Cloud data centre in both Sydney and Melbourne, where customers can install their own hardware if they so choose (meanwhile, Spark has several such 'private' data centres in New Zealand). In equal contrast, Spark, as well, own a 'fibre backbone network' between major centres, loosely comparable to Telstra's 'fibre highway' (Note: Spoark's fibre is a completely separate network to the Chorus fibre that serves end line customers on an individual premises basis).

What I am highlighting in this comparison is that the announced emphasis on where Spark and Telstra were spending their 'transformative investment' money was subtly different.

One explanation for Telstra's apparent 'overspend' on PP&E (compared to Spark, see above table) is that Australia has such a scattered population to cover, outside of the metropolitan areas. This is particularly emphasized by the recent sales of each company's respective cell tower networks in their respective countries. From Telstra's AR2017 p29:
"15 percent of investment in our mobile network has gone to provide services to the most remote two percent of the population in Australia."
With the sale of 49% of the Amplitel business, over 8,000 physical towers (800 of which were government grant subsidised to cover nationwide black spots), mast, large pole and antenna mount structures were transferred to the new holding company. By contrast, Spark sold 1,273 mobile tower structures when it divested those physical assets to the Ontario teacher's pension fund. Equalising those two networks on a population basis -using a factor of 5-, the 1,273 mobile towers in New Zealand is equivalent to: 5 x 1,273 = 6,365 towers in Australia. That works out to be 26% more towers on an equalized population basis for Telstra in Australia.

Capital spending and depreciation charges are eventually linked. Incremental capital spending over and above the norm will lead to higher depreciation charges going forwards.

SNOOPY

[Snoopy]

There is a problem with 'poor disclosure' amongst the Telstra depreciation regime. We know the total. But we don't know anything about the make up of the assets being depreciated, nor the depreciation schedules of those items.

	Spark {A}	Telstra	Telstra / 5 {B}	{B}/{A}	Reference
Transformative Assets Added Composition (prior to FY2022)	4G Mobile, Data Centres	4G Mobile, Inter city data highway	N/A	N/A	Posts 169 and 170
Transformative Expenditure Program (prior to FY2022)	$1,300m	$9,300m	$1,900m	+46%	Posts 169 and 170
Depreciation FY2022	$234m	$2,572m	$514m	+119%	Post 166

At Spark, Telecommunications assets are being depreciated along the following lines (there are no declared equivalent figures published for Telstra, although I am guessing they would be similar):

Links and cables	10-30 years
Network transport	3-15 years
Mobile radio Access Network	5-25 years
Customer Premesis Equipment	3-5 years
International Cable and Satellite	10-15 years

One way to interpret the higher population normalised depreciation at Telstra is to say that 'on average' the telecommunications equipment at Telstra must have shorter lifespans. How does that thesis fit in with the following observations?

i/ Telstra has a director for 'Network Transport and Routing Engineering'. From this, I can guess that the 'Network Transport' class of assets, referred to in the depreciation schedule, is the switching hardware and routing equipment that 'lights' the dark fibre. It is possible this category includes data centre electronics as well. I don't think that, once adjusted for population, there would be any significant difference between the 'Network Transport' depreciation schedules of Spark and Telstra.

ii/ "Mobile radio access network" assets are connected to mobile towers. The first jump in a mobile call is a radio wave signal from your personal handset to the nearest mobile radio receptor/mobile tower of the network you signed up to. That message received by the nearest tower then goes from that tower to the nearest network base station or "cell'. Furthermore, not every mobile tower on 'Mt Obscuresville ' has a cable connection running up to it. The path a mobile phone call follows can be entirely over the radio spectrum, or it might connect to a local cable node for part of the journey. 'Mobile radio access network' equipment must be 'available for use at any time', to make a viable network. So it makes sense to me that 'Mobile radio access network' depreciation is directly linked to the footprint of the mobile tower network. I need to mention here that although both Spark and Telstra have sold down their 'mobile tower networks', what has been sold down is the undifferentiated physical structures. The clever call handling and brand differentiating technology remains in the full ownership of the telco. In summary, don't expect "Mobile radio access network" depreciation to be less relevant going forwards, now that the mobile tower assets have been sold down. We know that there are proportionately more towers needed to cover a certain population in Australia, compared to New Zealand. So Telstra will have inherently higher depreciation of this asset class than Spark.

iii/ With fibre broadband in New Zealand, the on-premises equipment is provided by the wholesaler Chorus. By contrast, in Australia, Telstra make a big thing about their 'Telstra Smart Modem' (post 170), which has a relatively short depreciation life (see customer premises equipment schedule above). So it looks like Telstra has a much higher 'Customer premises depreciation' footprint than Spark.

iv/ Undersea cables. Spark (Southern Cross cable 34.8%) - AND -
Telstra (Reach 50% (investments in 40 submarine cables), Southern Cross 25%, INDIGO 'small stake' (to Indonesia and Singapore), AJC 46.9% (Australia to Japan) PIPE pacific cable (Australia to Guam) Telstra Endeavour 100% (Sydney to Hawaii) SEA-ME-WE-3 'small stake' (Perth to South East Asia, the Middle East and Europe), Dacom Crossing Corporation 49% (Korea, japan, Hong Kong, Singapore, Taiwan, Phillipines China).

- We can see that both Spark and Telstra have shareholdings in intercontinental cable networks. However, Telstra have more undersea cables to look after.

The 'dividing by 5 population equalizing trick' does not apply in the case of undersea cables,. That is because the capacity of a 'dark fibre cable' is dependent on the electronics lighting it up. So with 10,000 or 10,000,000 users,, the cost of laying the cable itself would be approximately the same.

The original cost of the Southern Cross cable (the one cable company in which both companies have a shareholding) was estimated at $US1.5billion. The most recent upgrade cost of the Southern Cross cable to 72TBps is budgeted at $US300m, of which Spark would supply $70m to $90m of equity. So it is clear that the total costs of these cables are in the billions of dollars. The length of the Southern Cross cable is just over 30,000km. Undersea cables are capital intensive and majority shareholders will see heavy annual depreciation charges as a result. And that includes Telstra (via their 50% or more ownership of Reach and Endeavour) but not Spark.

Nevertheless, high depreciation will reduce the profit on a Telco's minority stake equity accounted investment entities, as accounted for by Telstra and Spark too. So while the effect of high depreciation of cable networks will be there in any NPAT result, the effect of that depreciation, as recorded in the Spark/Telstra accounts, is indirect. That means depreciation expenses of minority stake equity accounted investments will not appear as depreciation in the books of the investment parent entity (be that Spark or Telstra).

In summary, as far as depreciation accounting charges go on international cable assets, Telstra will have a far greater direct weight to carry than Spark.

Summing the whole post up

Bullet points ii/, iii/ and iv/ would suggest that Telstra has an inherently higher depreciation expense structure than Spark, and there is very little 'good management' at Telstra can do about that.

SNOOPY

[Snoopy]

This is not a subject I know much about. But I found the transcript from the Telstra's 12th November 2020 investor day contained some fascinating insights on this subject. The following are selected excerpts from the talk delivered by Nikos Katinakis, the Group Executive of Telstra, for Telstra Network and IT. Nick is talking about what frequencies -and why- that Telstra are eyeing up in the deployment of their mobile 5G network, and how those frequencies will work around their existing 3G and 4G networks to do so. The talk is equally relevant to what Spark may or may not be doing with their own roll out of 5G in New Zealand.

-----------------

"5G is a lot more efficient than 4G. Our data consumption continues to grow by 30 to 40% per year, and we are always keen to use the next generation of technology that allows us to deliver more data at a lower cost per bit."

"Last year, we walked you through how the spectrum situation plays out in 5G. I would like to give you an update on this. We have three layers of spectrum that we are making available to 5G. Low band; great for coverage over long distances and in building penetration from the outside. We are selectively using our 850 [Mhz] spectrum, that is also used for 3G, and 700 [Mhz], that is used for 4G."

"We have announced that we are shutting down our 3G network in 2024, and in the meantime, we are deploying some cool technology called dynamic spectrum sharing that allows us to use the same radio for both 3G and 5G simultaneously. Mid-band is perfect for coverage improvements and additional capacity, and then high band or mmwave, excellent for massive capacity uplifts. We are using the mid-band, the 3.5 [Ghz] spectrum, as the foundation of 5G everywhere we deploy. "

"Two points of differentiation in the way we’re deploying our network, is that we have focused on deploying 5G in a contiguous fashion, and also by using low bands for inbuilding penetration."

"One important point that I want to make on this; it’s around speeds. As you can easily understand, big speeds start to become irrelevant. Most telcos around the globe, including ourselves, have been publishing big speeds that indicate the progress of the technology. We recently announced 4.2 gigabits per second on a 5G test. With the introduction of mmwave, the numbers get so large, that they really become irrelevant. Instead, we have started focusing on the actual experience that our customers are getting, wherever they are, and wherever they use their devices and their services."

----------------

While I accept everything that Nick is telling us above, there is a lot I do not understand.

Yes I get that 5G will deliver more data. But at a lower cost per bit? How does that work? I thought (perhaps wrongly) that 5G was going to require more receptors than 4G (mainly to improve latency) and so it would be a higher capital and power cost network to run. Of course it is possible for 5G to be higher cost to run, but lower cost per bit of data processed, simply because so much more data is able to be processed. Perhaps that is what Nick is referring to?

The velocity of an electromagnetic wave can be expressed in terms of wavelength and frequency as follows:
v=fλ, where 'f' is the frequency of the wave and λ (lambda) is the wavelength.

The speed at which electromagnetic waves travel is the speed of light, a constant value represented by the letter 'c'. So the more familiar form of the above equation is: c=fλ. What this equation is telling us, is that the higher the frequency of the electromagnetic wave, the lower the wavelength. This is why Nick is telling us that low band (i.e. low frequency) signals are better for longer distances,, because they are accompanied by a longer wave length (easier to cover more distance). I think the longer wavelength is also what allows these lower frequency signals to leap through walls more easily.

The energy of an electromagnetic wave is proportional to its frequency. The so called 'high band' radiation is 'high frequency'. So this is why I thought 5G was more power hungry, as more power was needed to capture more data. That means I am a little perplexed by Nick saying Telstra will use 'medium frequency' as their base broadcast range for 5G. Yet right at the beginning of the talk, Nick is even saying he has three 'layers of spectrum' (presumably low, medium and high) available for 5G? What is going on there?

I guess anyone who has read this far will have come to the conclusion that I know a little bit about what I am talking about, but obviously not enough to take in the full impact of what Nick is saying. So anyone with a better understanding of what is being proposed here, please feel free to enlighten me.

SNOOPY

[Snoopy]

	Spark {A}	Telstra	Telstra / 5 {B}	{B}/{A}	Reference
Transformative Assets Added Composition (prior to FY2022)	4G Mobile, Data Centres	4G Mobile, Inter city data highway	N/A	N/A	Posts 169 and 170
Transformative Expenditure Program (prior to FY2022)	$1,300m	$9,300m	$1,900m	+46%	Posts 169 and 170
Depreciation FY2022	$234m	$2,572m	$514m	+119%	Post 166

Capital spending and depreciation charges are eventually linked. Incremental capital spending over and above the norm will lead to higher depreciation charges going forwards.

Some comments here to expand on my quoted comment above in bold. Capital spending and depreciation would exactly match if the rate at which equipment wore out was equal to the new expenditure spend each year needed to replace that equipment that had worn out. As a general rule this is not what happens, as capital expenditure tends to be 'lumpy'.

In the case of the quoted table above, the depreciation expense is real. But my figure as to what the dollar capital expenditure is at Telstra towards their 'developing 4G network' and 'cross country city node fibre connections' in particular is an estimate.
This is in contrast to when I did a similar exercise on Spark (post 769), which was more accurate due to better disclosure. I have taken the total declared capital expenditure at Telstra over FY2017 and FY2018 to represent the capital spending on this 'technology transformation' effect. But this period of expenditure at Telstra will have also included expenditure on existing 2G and 3G networks, expenditure on the existing legacy fixed network, and 'Land and Buildings' not tied to communication assets as well as 'other plant and equipment', BUT excluding overseas funded assets (which I am guessing includes the jointly funded but minority owned stakes in 'international joint venture and fully owned undersea cables'). My underlying assumption was that all capex at Telstra over FY2017 and FY2018 was towards 'transformative projects', when I knew that in practice this was not true. To make up for this 'overestimate', I assumed that zero capital expenditure over FY2019 at Telstra went towards the transformative projects, when in fact I knew that some 'finishing off' expenditure on the transformative projects did spill over into that year.

If a company was to make a big capital investment over a short period, then I would expect the current year annual depreciation expense to be quite a small proportion of the new investment. So what do we actually see?:

Spark: $234m/$1300m = 18%

Telstra: $524m/$1900m = 27%

We see the Telstra annual depreciation, as a proportion of 'transformative asset investment', is incrementally larger than at Spark. How to explain that? One explanation is that my estimate of Telstra's transformative investment was too low, despite being 46% higher than Spark in inter company comparative terms (possible). Another explanation is that depreciation at Telstra has been extra high, because of the closing down of the legacy fixed asset network. As users migrate to nbn, this has meant an accelerated depreciation schedule' on those soon to be shut down legacy network assets. With this process largely complete, we will have to wait until the FY2023 results are posted to see if this is part of the explanation. A look at the FY2023 half year result suggests depreciation is down 4.9% for that interim period, although how that will translate to the full year period at Telstra is unknown.. If 'accelerated legacy network depreciation' is a factor, this would imply that the remaining asset value on the balance sheet relating to that network is significantly more than the capital on balance sheet related to the much more recently deployed 4G mobile network (quite possible).

SNOOPY

[Snoopy]

	Spark {A}	Telstra	Telstra / 5 {B}	{B}/{A}	Reference
4G Network Area Coverage (FY2022)	350,000km2	2,500,000km2	500,000km2	+43%	Post 171
Network Cell Towers (FY2022)	1,273	8,000	1,600	+26%	Post 172

Plenty of speculation from me (post 175) on why depreciation at Telstra is whatever way you look at it higher than Spark. But let's not forget a possible simpler explanation. Namely that Australia being such a vast country needs more infrastructure to service a given number of people. And furthermore, Australia being such a harsh country, -weather wise-, the equipment deployed in Australia does not last as long as if that same equipment was deployed in New Zealand. The extent in the geographic disparity between Australia and New Zealand is very evident in the table above.

SNOOPY

[Snoopy]

The last in my series of posts considering why depreciation is higher, by whatever measure you use, at Telstra compared to Spark concerns the in country fibre highway between main centres that both Spark and Telstra have built, but that only Telstra 'really trumpets'.

	Spark {A}	Telstra	Telstra / 5 {B}	{B}/{A}	Reference
Intercity Fibre Highway (FY2022)	1,640km	5,210km	N.A.	N/A	Post 170

First an admission. I don't know the exact path these fibre highways take. The total distances I have used approximate the sum total of main road highway distances between adjacent major centres in the two respective countries. Furthermore these totals do not consider any ''branch lines' (for example Sydney to Canberra or Queenstown to Dunedin) that may exist. My reasoning here is that although the fibre cables may not exactly follow the road path, it would make sense to have them reasonably accessible by road to allow storm or earthquake damage to be quickly repaired. Branch line (if they exist) lengths are likely small compared to main trunk line distances. So I think just comparing my estimate of the main trunk fibre line length within each country makes for a fair comparison.

You will notice in the above table that all the mathematical scaling that I could have employed I have deemed N.A. or 'not applicable'. This is for the same reason I discussed in post 173 when I was talking about international inter-continental fibre cables, namely

"The capacity of a 'dark fibre cable' is dependent on the electronics lighting it up. So with 10,000 or 10,000,000 users,, the cost of laying the cable itself would be approximately the same."

So just like the international fibre cable situation, the number of users is entirely secondary to the shear number of kilometres covered in determining the fibre cable laying cost. Put succinctly, Australia is the bigger country. so 5210/1640= 3.2 times more length of cable is needed to cover it. More cable = more depreciation cost. It is as simple as that!

SNOOPY

[Snoopy]

This is not a subject I know much about. But I found the transcript from the Telstra's 12th November 2020 investor day contained some fascinating insights on this subject. The following are selected excerpts from the talk delivered by Nikos Katinakis, the Group Executive of Telstra, for Telstra Network and IT. Nick is talking about what frequencies -and why- that Telstra are eyeing up in the deployment of their mobile 5G network, and how those frequencies will work around their existing 3G and 4G networks to do so. The talk is equally relevant to what Spark may or may not be doing with their own roll out of 5G in New Zealand.

-----------------

"5G is a lot more efficient than 4G. Our data consumption continues to grow by 30 to 40% per year, and we are always keen to use the next generation of technology that allows us to deliver more data at a lower cost per bit."

"Last year, we walked you through how the spectrum situation plays out in 5G. I would like to give you an update on this. We have three layers of spectrum that we are making available to 5G. Low band; great for coverage over long distances and in building penetration from the outside. We are selectively using our 850 [Mhz] spectrum, that is also used for 3G, and 700 [Mhz], that is used for 4G."

"We have announced that we are shutting down our 3G network in 2024, and in the meantime, we are deploying some cool technology called dynamic spectrum sharing that allows us to use the same radio for both 3G and 5G simultaneously. Mid-band is perfect for coverage improvements and additional capacity, and then high band or mmwave, excellent for massive capacity uplifts. We are using the mid-band, the 3.5 [Ghz] spectrum, as the foundation of 5G everywhere we deploy. "

"Two points of differentiation in the way we’re deploying our network, is that we have focused on deploying 5G in a contiguous fashion, and also by using low bands for inbuilding penetration."

"One important point that I want to make on this; it’s around speeds. As you can easily understand, big speeds start to become irrelevant. Most telcos around the globe, including ourselves, have been publishing big speeds that indicate the progress of the technology. We recently announced 4.2 gigabits per second on a 5G test. With the introduction of mmwave, the numbers get so large, that they really become irrelevant. Instead, we have started focusing on the actual experience that our customers are getting, wherever they are, and wherever they use their devices and their services."

----------------

I am coming increasingly to the view that it is the growth in mobile networks that will drive the incumbent telecommunications market players growth going forwards. So which company is doing the better job up to now, Spark or Telstra?

	FY2019	FY2020	FY2021	FY2022	FY2023	latest 2 year gain	4 year gain
Telstra Mobile Revenue	$10,084m	$10,130m	$9,310m	$9,470m	$10,258m	+10.2%	+1.73%
Telstra Mobile Revenue (divided by 5)	$2,017m	$2,026m	$1,862m	$1,894m	$2,052m
Telstra Mobile EBITDA margin	34%	34.7%	39.2%	42.2%	44.9%
Spark Mobile Revenue {A}	$1,271m	$1,288m	$1,311m	$1,351m	$1,470m	+12.1%	+15.7%
Spark Mobile Product Cost {B}	$496m	$459m	$474m	$447m	$486m
Spark Mobile Unallocated Cost allocation, {C} (Note 1)	$318m	$328m	$322m	$320m	$367m
Spark Mobile Total Cost allocation, {B}+{C} ={D}	$814m	$787m	$796m	$767m	$853m	+7.16%	+4.79%
Spark Mobile EBITDA margin, ({A}-{D})/{D}	36.0%	38.9%	39.3%	43.2%	42.0%

Notes

1/ I am using the table below to apportion unallocated labour and other operating costs to the 'mobile' operating unit at Spark. This information is needed to convert Sparks unconventional metric of 'product margin' (See AR section 2.1 on 'Segment Information') into the more universally measured 'EBITDA margin'.

	FY2019	FY2020	FY2021	FY2022	FY2023
Spark Mobile Revenue {A}	$1,271m	$1,288m	$1,311m	$1,351m	$1,470m
Spark Total Revenue {B} (Note 1)	$3,518m	$3,588m	$3,565m	$3,694m	$3,875m
Spark Mobile Cost Allocation Ratio{A}/{B} = 'R'	0.3613	0.3590	0.3677	0.3657	0.3794
Spark Total Labour Cost (Unallocated) {C}	$475m	$511m	$491m	$495m	$511m
Spark Total Other Operating Expenses (Unallocated) {D}	$404m	$402m	$385m	$381m	$456m
Spark Mobile Unallocated Cost allocation, R({C}+{D})	$318m	$328m	$322m	$320m	$367m

Telstra in their 'Full Years results and operations review" have a table labelled 'Product EBITDA margins'. It is these figures I have put in the above table. I did wonder if they had 'done a Spark' and left labour and other costs out of their figures. But after I had calculated the sum of EBITDA earnings implied by that table, it seems this is not the case. The reason the table is headed 'Product EBITDA earnings' is to differentiate it from the segmented analysis where earnings are grouped under business header units, not product groups.

--------------------------

[Snoopy]

This thread has been dormant for 10 years, but time to fire it up again. The thread title should now be SPK vs TLS. But I hope Vince will sort that out when he returns from hibernation. A lot has happened in 10 years. But after growing apart with what was TEL shedding their fixed network arm Chorus twelve years ago, TLS has (finally) followed with their own fixed line network going to nbn, with the associated transfer payments now rounding down. Now we are left with what were the two former incumbents, still the largest telecommunications retailers in their respective countries. I feel a comparative stand off coming on.....

The population figures I have for both countries as of 2022 were 5,185,288 (New Zealand) and 26,124,814 (Australia). That is a ratio of 5.0:1 in favour of Australia. I have added a row to the table to normalise for this population mismatch where appropriate by dividing the TLS figure by 5.

	Spark (SPK)	Telstra (TLS)	Telstra (TLS) (population normalised)
Operational Sector	Telecommunications	Telecommunications
Total Employees	4,976	28,889	5,778
Geographic Market(s)	New Zealand	Australia
Share Price 15-05-2023	$NZ5.22	$A4.35
Market Capitalisation 15-05-2023	$NZ9.732b	$A49.92b	$A9.984b
Capitalised Dividend Valuation per share (2019 to 2023)	$NZ5.24	$A3.09
Declared earnings (FY2022)	$NZ410m	$A1,814m	$A363m
Normalised earnings (FY2022)	$NZ397m	$A1,643m	$A329m
Normalised eps (FY2022)	NZ20.9c	A14.2c
Normalised eps growth over 4 year period (FY2018 to FY2022)	+7.18%	+2.90%
Historical PER (SP@15-05-2023)(FY2022)	25.1	30.6
dps (paid during FY2022)	NZ12.5c+NZ12.5c	A5c+A6c
Earnings Payout Ratio (excluding DRP)	120%	77.5%
Gross dps (paid during FY2022)	NZ17.36c+NZ17.36c	A7.14c+A8.57c
Historical Gross Dividend Yield (using Share Price 15-05-2023)	6.63%	3.61%
Shareholder Equity (based on equity at EOFY2022)	$1,475m	$A16,837m	$A3,367m
ROE (based on equity at EOFY2022)	26.9%	9.77%
Sales (FY2022)	$3,720m	$A16,837m	$A3,367m
Net Profit Margin (FY2022)	10.7%	6.98%
Total Drawn Term Debt (last balance date EOFY2022)	$1,526m	$A10,982m	$A2,196m
MDRT (Based on Drawn Term Debt at balance date EOFY2022)	3.84 years	6.05 years

At this point I was planning to have a short summary on the relative merit and demerit points of investing into Spark and Telstra. But going through the comparative columns line by line, it is apparent that Spark is outperforming Telstra in all measures. There is nothing we can learn from the Aussies, and there appears to be everything they can learn from us!

Operational metrics show how a company is performing. But for investors, the key statistic to look for is the value you get for the price you pay. Looking at both through a 'no growth' capitalised dividend lens, you would be buying SPK as a no growth prospect on the market today at fair value. OTOH the premium you are paying for TLS ($A4.35 / $A3.09 = +40%) indicates a lot of forward growth is already built into the share price.

It is rare that such a comprehensive comparison is as one sided as this. Buy Spark on share price weakness and sell TLS into any share price rally looks to be the message this table is telling us.

discl: hold SPK and TLS, but a lot more of the former.

Another year, another head to head contest. The thread title should now be SPK vs TLS. Vince is still on his cryogenic break by the look of it, so the thread title remains frozen. The two former telecommunications incumbents, still the largest telecommunications retailers in their respective countries, face off.....

The population figures I have for both countries as of 2022 were 5,185,288 (New Zealand) and 26,124,814 (Australia). That is a ratio of 5.0:1 in favour of Australia. Table column 3 normalises figures for this population mismatch where appropriate by dividing the TLS figure by 5.

	Spark (SPK)	Telstra (TLS)	Telstra (TLS) (population normalised)
Operational Sector	Telecommunications	Telecommunications
Total Employees	5,432	31,761	6,352
Geographic Market(s)	New Zealand	Australia
Share Price 18-10-2023	$NZ4.99	$A3.87
Market Capitalisation 18-10-2023	$NZ9.206b	$A44.71b	$A8.943b
Capitalised Dividend Valuation per share (2019.5 to 2023.5)	$NZ5.32	$A3.15
Declared earnings (FY2023)	$NZ449m	$A2,051m	$A410m
Normalised earnings (FY2023)	$NZ425m	$A2,128m	$A426m
Normalised eps (FY2023)	NZ23.0c	A18.4c
Normalised eps growth over 4 year period (FY2019 to FY2023)	+5.99%	+52.1%
Historical PER (SP@18-10-2023)(FY2023)	23.1	21.0
dps (paid during FY2023)	NZ12.5c+NZ13.5c	A7.5c+A8.5c
Earnings Payout Ratio (excluding DRP)	113%	76.2%
Gross dps (paid during FY2023)	NZ17.36c+NZ18.75c	A10.71c+A12.14c
Historical Gross Dividend Yield (using Share Price 15-10-2023)	7.24%	5.90%
Shareholder Equity (based on equity at EOFY2023)	$1,940m	$A17,816m	$A3,563m
ROE (based on equity at EOFY2023)	21.9%	11.9%
Sales (FY2023)	$3,875m	$A23,173m	$A4,635m
Net Profit Margin (FY2023)	11.0%	9.18%
Total Drawn Term Debt (last balance date EOFY2023)	$1,052m	$A12,675m	$A2,535m
MDRT (Based on Drawn Term Debt at balance date EOFY2022)	2.48 years	5.96 years

Discussion

In an industry renowned for cost cutting, I was surprised to see the workforce of each protagonist grow by around 10% over FY2023. In the case of Spark, this was primarily explained by an in-sourcing of field services, no doubt assisted by growth in the company's contracting arm Entelar (which includes the old Connect 8). Added to that was an increase in staff at subsidiary MATTR. MATTR is the 'exciting' standalone Spark subsidiary company. It provides infrastructure for verifiable data and digital trust. MATTR operates over a Software as a Service (SaaS) Platform. Meanwhile over FY2023 Telstra expanded into new markets, acquiring Digicel Pacific (1700 employees). Furthermore Telstra recalled outsourced overseas call centre jobs back to Australia, resulting in what were indirect contract jobs being returned to full time employment status. Finally organic growth in the 'Telstra Purple': professional services, managed services and cloud would have been a sure bet to add to the head count in that specialist division.

There are differences worth mentioning. But my overall impression is how similar our two protagonists are from an investment perspective on those raw table numbers. Telstra is much bigger, courtesy of the much bigger country in which it operates. But when you scale some of those figures by 5, to take account of the relative population size of Australia verses NZ, the similarities become more obvious. In the case of normalised earnings, almost spookily so: $425m vs $426m in their respective currencies!

In a reversal from last year, it is now Spark that trades on the higher PE ratio, possibly supported by the higher dividend yield on offer. The return on shareholder equity continues to be significantly higher at Spark as well. This could be due to the vast size of Australia, with more equipment needed to connect a more thinly spread population. It could also be due to the 'not inexpensive to set up' Telstra Purple division, synonymous with highly paid 'computer techs' which have been added to with acquisitions like 'Power Health' and 'Medical Director'. These two acquisitions cost over half a billion dollars for the pair in FY2022. but actual profits from these venturesfor their part in the 'Telstra must be present' market space of 'Telstra Purple' are signalled to be years away. Spark, by contrast, have much more modest ambitions in their own in house software space, preferring to sell solutions developed by others. That means more money in the pockets of Spark shareholders today, but losing the opportunity cost of one or more of these big software projects 'coming right' in future years (with the exception of MATTR). Returning to Telstra, I think there is $800m worth of revenue within the 'Telstra Enterprise' 'business unit envelope', NAS applications in particular, which is loss making. A quick 'rule of thumb' on valuing such software assets is to 'value them at a multiple revenue'. A multiple of 5 has some historical precedence, albeit this is a 'rough measure' as it doesn't individually consider:

● Annual Recurring Revenue (business size)
● Growth rate (momentum)
● Net revenue retention (product quality)
● Gross margin (profitability)

By this measure, $800m in revenue equates to: 5x$800m/$11,554m= 35cps

Operational metrics show how a company is performing. But for investors, the key statistic to look for is the value you get for the price you pay. Looking at both through a 'no growth' capitalised dividend lens, you would be buying SPK as the better no growth prospect on the market today at fair value. OTOH the premium you are paying for TLS ($A3.87 / $A3.15 = +23%, or 72cps) indicates a lot of forward growth is already built into the share price. But looking at the cumulative four year normalised growth rate at Telstra over over 50%, verses just over 5% at Spark, is this evidence that the Telstra growth premium really is justified?

Comparing the 'product performance' disclosures at Telstra in AR2019 and AR2023, indicates from where this improved profit growth has come from. The EBITDA margin expanding from 34% to 44.9% at the largest product grouping mobile has helped. But the big change has been at the InfraCo business unit, a business unit wasn't even recognised as such in AR2019. Back then 'infrastructure' was simply regarded as a 'necessary cost'. InfraCo and the associated mobile tower company Amplitel, combined now represent more that 10% of Telstra's total revenue. And what is more they now have by far the highest EBITDA margins of 65.1% and 79.3% respectively of any business unit. But most of their 'customers' are other divisions of Telstra. That means the day to day operational parts of Telstra are operating with a much higher input cost base in FY2023 than in FY2019. Yet still the EBITDA profit margins in those other product groupings are steady or increasing on what some would say are 'artificially increased input costs'. What a brilliant way to ramp up profitability across all the front line divisions without appearing to be too greedy! Corporatising a cost centre like Infraco was a brilliant accounting trick. But it is an ace that once played, cannot be played again. So I would not read too much in to that spectacular normalised growth figure at Telstra over four years of 52.1%.


discl: hold SPK and TLS, but a lot more of the former