CNU - Chorus

[Zaphod]

My thoughts on Starlink and other technologies are that they probably aren't a direct threat. They fill a need where fibre isn't available or cost prohibitive (e.g. rural or remote areas). Fibre is likely to be best choice for consumers that need reliable internet

Reliability, stability and low latency is quite important for a few use cases (including gaming and video/audio calling).
A extra few milliseconds in internet delay has a dramatic impact on video/audio calling and gaming. You know how awkward it is if you've been a call where there is an even small delay in audio or video.

Completely agree with what you've written above.

Unfortunately there are a number of people convinced that wireless is naturally better than wired. Wires are old tech, it's awkward and should be discontinued. Wireless is the future. They can't be swayed even when provided a highly detailed technical rundown on how the wireless (sat, cellular, and Wi-Fi) operates, and why that's not the case.

[Snoopy]

Currently Chorus charges a regulated price per user that is passed through to the retailer.

HYP2021 Slide 7 shows coming wholesale 'price changes' for Chorus wholesale customers.

Broadband Plan	FY2020 Pricing	FY2021 Pricing
50Mbps	$42.50/m	$43.56/m (from 01/10/2021, +2.5%)
100Mbps	$46/m	$47.15/m (from 01/10/2021, +2.5%)
200Mbps	$55/m	$56.38/m (from 01/10/2021, +2.5%)
1 Gbps	$60/m	$56/m (from 01/07/2021, -6.7%)

What struck me about these price changes is that from October 1st the 1 Gigabit scheme (that's 1,000Mbps) is cheaper than the 200Mbps scheme! That seems very odd. These are wholesale prices of course. So it isn't a given that all retailers will pass these price changes on. Have any forum members out there been informed of upcoming price/speed changes to their broadband plans?

SNOOPY

[Snoopy]

I am becoming convinced that increasing ARPU will be a core determinant of of the profitability of Chorus going forwards. But strangely, although Chorus themselves are keen to throw this acronym around in their presentations, I can only find one mention of what this figure actually is. In slide 15 of PR2020 we learn that

"Growing fibre uptake and ARPU: June FY20 $48.42* vs June FY19: $47.50"

This is a note specifically referring to Fibre GPON (Gigabit Passive Optical Networks) revenue of $393m (FY2020) to $294m (FY2019). The little asterisk in my quote leads to another reference which says that the figure for FY2020 has been altered

"*FY20 ARPU adjusted to exclude COVID-19 related industry credits"

The FY2019 figure seems clean though. So I was curious as to how Chorus are doing their ARPU calculations:

ARPU = Broadband Revenue / Broadband Users

To convert that to a monthly figure, divide by 12. Using the figures on pages 21 and 22 of AR2019 I get:

ARPU = 1/12 x ($294m / 0.599m ) = $40.90 (too low)

If I add 'fibre premium (P2P)' to the 'fibre broadband (GPON)' revenue I get

ARPU = 1/12 x (($294m+$74m) / 0.599m ) = $51.20 (too high)

That figure doesn't take into account the growth in the number of broadband users during the year. If I back out the 'fibre premium' P2P revenue and use a linear average of the number of subscribers at EOFY2018 and EOFY2019 then I get:

ARPU = 1/12 x ($294m/ (0.5x(0.599m + 0.433m)) = $47.48 (low, but very close)

That previous calculation assumes a single straight line growth 'curve' represents the change in broadband accounts over the year. AR2019 gives us the half year figure as well. Using that as well will give a better linear interpolation of the growth in broadband accounts:

ARPU = 1/12 x ($294m/ (0.3333x(0.599m + 0.517m+0.433m)) = $47.45 (extra bit of linear interpolation gives a slightly worse result)

I am satisfied that second to last calculation is 'close enough'.

SNOOPY

[Snoopy]

My current understanding is that 'Fibre Premium' is a business offering to allow their different business premises to be directly linked up (peer-to-peer). It might be useful for complex applications where reliability or security may be important. Imagine if a business wants to establish a private network between their different sites without going through the internet. My view is that this is a very niche product/service and small/medium business may not require this service..

Some more information from AR2018 when the term 'Fibre Premium' was introduced as a revenue reporting category. The revenue figures I have compared with the equivalent figures from AR2017 when the now superseded revenue naming conventions were in place.

	Revenue FY2017 (from AR2017)	Revenue FY2017 (from AR2018)
Basic Copper	$450m
Copper Based Voice		$163m
Enhanced Copper	$248m
Copper Based Broadband		$501m
Data Service over Copper		$32m
Fibre	$198m
Fibre broadband (GPON) (2)		$123m
Fibre premium (P2P) (1)		$79m
Value Added Network Services	$36m	$34m
Infrastructure	$23m	$23m
Field Services	$76m	$76m
Other	$9m	$9m
Total Revenue	$1040m	$1040m

Notes

1/ P2P means 'where two parties or devices are connected point to point via fibre'.
2/ GPON means 'Gigabit Passive Optical Network'. GPON does not imply that everyone connected to it has a Gigabit speed connection. It just means that with suitable switching and routing equipment hooked up to the network, they could. So GPON simply means the fibre FTTP network that all Chorus customers are hooked up to.

The first point to note is that what was once classed as 'Fibre' ($198m), does not equal the sum of 'Fibre Broadband' and 'Fibre Premium' ($202m). Furthermore the sum of 'Basic Copper' and 'Enhanced Copper' ($698m) does not equal the sum of 'Copper Based Voice', 'Copper Based Broadband', and 'Data Service over Copper' ($696m). However that $2m 'copper' difference could be rounding reclassification anomaly, whereas even 'rounding error' could not explain the $4m mismatch in 'Fibre' categories. It is possible that some of the 'value added services' have been reallocated to fibre.

The 'old' Fibre classification included:

1/ Something called HSNS or 'High Speed Network Services', a section of dedicated data bandwidth that may transfer over copper or fibre.
2/ 'UFB Backhaul', which can deliver connections between mobile tower sites.
3/ 'Direct Fibre Access Services', which I think is what is subsequently classified as 'Fibre Premium' P2P.

This is useful in reconfirming what 'fibre premium' revenue is not: A premium product to provide 'hyperfibre connections' to upscale customers.

SNOOPY

[Entrep]

I'm taking a look at CNU. This is from Craigs in May:

Prima facie this appears positive as our estimates did not include depreciation tilting so the Commission's estimates would represent a lift in allowable revenue. However, the devil is in the detail of how the Commission arrived at its MAR estimate. The key things to note are: 1) Depreciation of the core fibre assets is based on Straight Line, but the Commission allowed depreciation of the financial loss asset based on Diminishing Value. So essentially has allowed some depreciation tilting, whereas we had previously assumed Straight Line for both. 2) Offsetting this the Commission reduced opex by around 10% relative to CNU’s submission. While the depreciation methodology affects the timing of cash flows but is NPV neutral, we assume the reduction of the opex allowance is permanent and effectively reduces the value of the residual Copper asset by lumping additional cost into that bucket. We estimate a 15-20 cps impact on our valuation.

This MAR estimate was based on CNU’s RAB submission of $5.5b. The Commission's view on RAB is due in August so this has the potential to reduce the MAR range further.

Forecast changes but DPS largely unchanged: We have increased earnings estimates for FY23 and FY24 to reflect a higher MAR and smoothing, but longer-term this is temporary as the front-loading of depreciation will reverse and become a drag on earnings. This temporary impact and the relatively small scale of the earnings changes means we have made no changes to near-term DPS estimates.

Price Target of $6.55 (prev $6.70). Neutral rating remains: Our Price Target remains based on the average of our forward DCF valuation (average WACC 6.3%, Tg 2.5%), and our forward DDM valuation. Our Price Target implies a cash yield of 4.0%. Key up/downside risks include regulation, UFB connection volumes and performance, and connection loss to fixed wireless.

[Snoopy]

I'm taking a look at CNU. This is from Craigs in May:

Quite a bit of jargon in that report from Craigs. 'MAR' stands for 'Maximum Allowable Revenue' permissible for the regulated monopoly broadband wholesale network. The FY2020 total broadband revenue through Chorus was $466m. This is still well under the $689m (FY2022) to $786m (FY2024) of revenue that the commerce commission propose as an 'upper bound' for revenue in the future. Although the Craig's report does suggest that the 'commerce commission ceiling' may yet be further lowered.

The report takes a short to medium term view looking at FY23 and FY24, with no mention of the massive debt repayment schedule that starts to kick in around that latter date. UFB connection volumes is listed as a 'risk'. I would have thought that UFB connection volumes were the primary driver of earnings, and it is IMO disgracefully misleading of Craigs to skirt this core issue. A proper analysis should IMO model the copper being 'backed out', with particular reference to the part of the broadband roll out that Chorus do not control (greater Christchurch, much of the Central North Island and Greater Whangarei) as well as the timing of of more and enhanced service connections via Chorus.

A very poor analysis I think would sum up what Craigs has done here. Making such a statement, I guess, puts an onus on me to come up with something better. My analysis is still a work in progress, but it is progressing....

SNOOPY

[Snoopy]

I am becoming convinced that increasing ARPU will be a core determinant of of the profitability of Chorus going forwards. But strangely, although Chorus themselves are keen to throw this acronym around in their presentations, I can only find one mention of what this figure actually is. In slide 15 of PR2020 we learn that I can

"Growing fibre uptake and ARPU: June FY20 $48.42* vs June FY19: $47.50"
I can
This is a note specifically referring to Fibre GPON (Gigabit Passive Optical Networks) revenue of $393m (FY2020) to $294m (FY2019). The little asterisk in my quote leads to another reference which says that the figure for FY2020 has been altered

"*FY20 ARPU adjusted to exclude COVID-19 related industry credits"

The FY2019 figure seems clean though. So I was curious as to how Chorus are doing their ARPU calculations:

ARPU = Broadband Revenue / Broadband Users

To convert that to a monthly figure, divide by 12. Using the figures on pages 21 and 22 of AR2019 I get:

ARPU = 1/12 x ($294m / 0.599m ) = $40.90 (too low)

If I add 'fibre premium (P2P)' to the 'fibre broadband (GPON)' revenue I get

ARPU = 1/12 x (($294m+$74m) / 0.599m ) = $51.20 (too high)

That figure doesn't take into account the growth in the number of broadband users during the year. If I back out the 'fibre premium' P2P revenue and use a linear average of the number of subscribers at EOFY2018 and EOFY2019 then I get:

ARPU = 1/12 x ($294m/ (0.5x(0.599m + 0.433m)) = $47.48 (low, but very close)

That previous calculation assumes a single straight line growth 'curve' represents the change in broadband accounts over the year. AR2019 gives us the half year figure as well. Using that as well will give a better linear interpolation of the growth in broadband accI canounts:

ARPU = 1/12 x ($294m/ (0.3333x(0.599m + 0.517m+0.433m)) = $47.45 (extra bit of linear interpolation gives a slightly worse result)

I am satisfied that second to last calculation is 'close enough'.

The following revenue figures for 'fibre broadband' and 'copper broadband' are largely taken from the 'Management Commentary' section of the respective annual reports. The 'In summary' table lists the number of connections, whereas the 'Revenue Commentary' lists the revenue attributable to each type of broadband.

Year	Broadband (GPON) Revenue	Broadband (GPON) Connections	ARPU (F)	Copper Broadband Revenue	Copper Broadband Connections	ARPU (C)
FY2015	$29m	75,000	$?	$268m	1,132,000	$?
FY2016	$60m	167,000	$37.19	$456m	1,059,000	$34.69
FY2017	$123m	292,000	$44.66	$501m	894,000	$42.75
FY2018	$198m	433,000	$45.52	$421m	754,000	$42.58
FY2019	$294m	599,000	$47.48	$344m	597,000	$42.44
FY2020	$393m	740,000	$48.92	$271m	466,000	$42.49

Notes

1/ Sample ARPU calculation for Fibre Broadband (GPON) for FY2019:

ARPU = 1/12 x ($294m/ (0.5x(0.599m + 0.433m)) = $47.48

2/ Fibre GPON customers for FY2016 = Fibre 'Mass market' Customers for FY2016
Fibre GPON Revenue for FY2016 and FY2015 taken from Appendix 3, Slide 36 , in PR2017.

3/ Fibre GPON customers for FY2015 estimated as: 88,000 - 13,000 = 75,000

4/ Copper broadband numbers for FY2016 and FY2015 are from Slide 5 of the Chorus PR2016.

5/ Copper broadband revenues for FY2016 and FY2015 are from Appendix 3, Slide 36, in PR2017

Why is this table important? The first ARPU column is representative of the business being gained as fibre broadband is being rolled out across the country. The second ARPU column represents business being lost as copper broadband is gradually retired and replaced with fibre broadband. It is heartening to see the ARPU being gained is greater than the ARPU being lost , even if such a superficial statement does not tell the full story (more on this later).

I talk of copper broadband as a 'legacy technology'. Yet it is surprising how recently the likes of copper VDSL was the latest in broadband technology and still in its growth phase (revenues sharply rose between FY2015 and FY2017). The calculated ARPU is consistent with the wholesale /Dial Up (remember that?)/ ADSL/VDSL/ bill being fairly steady over the time-frame of this table. By contrast the ARPU for Fibre GPON is steadily rising. This is consistent with, on average, a higher number of higher specification Fibre Broadband plans being sold as the years go by.



SNOOPY

[Snoopy]

The FY2018 annual report lead to a recategorisation of Chorus customers into different reporting classes.

	Connections FY2017 (from AR2017)	Connections FY2017 (from AR2018)
Baseband Copper	976,000	313,000
Naked Unbundled Bitstream Access (Broadband using Chorus Equipment)	213,000
Naked Unbundled Copper Baseband IP (Voice) (using Customer Equipment)	18,000
Copper VDSL		244,000
Copper ADSL		650,000
Unbundled Copper Local Loop (Voice & Broadband using Customer Equipment)	81,000
Sub Loop Extension Service (Connecting Customer Equipment to Chorus)	1,000
Unbundled Copper		82,000
Data Service over Copper	8,000	8,000
Fibre (mass market / GPON)	292,000	292,000
Fibre (Premium Business) / Premium (P2P)	13,000	13,000
Total Connections	1,602,000	1,602,000

The main disconnect between the two reporting regimes is how the copper customers are distinguished. The purpose of this post is to try and figure out how to disentangle 'copper broadband' figures when they are reported under the old naming regime (In Annual Reports FY2017 and earlier). However, I now find this information has been 'retrospectively disclosed' in PR2017 Appendix 3. That takes away the purpose of this exercise. So I will leave this post as an 'unfinished symphony'.

SNOOPY

[Snoopy]

Year	ARPU (Fibre)	ARPU (Copper)
FY2016	$37.19	$34.69
FY2017	$44.66	$42.75
FY2018	$45.52	$42.58
FY2019	$47.48	$42.44
FY2020	$48.92	$42.49

Why is this table important? The first ARPU column is representative of the business being gained as fibre broadband is being rolled out across the country. The second ARPU column represents business being lost as copper broadband is gradually retired and replaced with fibre broadband. It is heartening to see the ARPU being gained is greater than the ARPU being lost, even if such a superficial statement does not tell the full story (more on this later).

Wholesale Broadband Operator	%ge Nationwide Connections	EOFY2020 Connections {A}	EOFY 2020 Premises Passed {B}	%ge Regional Take Up {A}/{B}
Chorus	72%	725,000	1,209,000	60.0%
Northpower	2%	19,000	33,000	57.8%
WEL Networks	14%	145,000	237,000	61.2%
Enable	12%	117,600	187,000	62.9%
Total	100%	1,006,600	1,666,000	60.4%

Notes

1/ The end of the financial year is, for all four operators, on 30th June.

The most important figure in the above table, for Chorus investors, is the one in bold. Chorus owns 100% of the legacy copper network. So as users progress to fibre, they will eventually lose all of their copper legacy network, but only retain around 72% of the replacement fibre network. This means that although transitioning from the copper network will overall mean a boost in funding for pure fibre network owners, the expected 'gain' in revenue at Chorus is a little different.

Weighted Average Fibre Monthly Revenue Gain Per User	72% x $48.92= $35.22
less Weighted Average Monthly Copper Revenue Lost Per User	100% x $42.49= $42.49
equals Weighted Average Overall Monthly Revenue Change Per User	-$7.27

Yes you did read that right. As the fibre transformation wave wafts over the country, Chorus can expect to lose on average $7.27 MRPU for every customer transitioned to fibre! In Australia they are forecasting around 70% of old copper network premises will eventually transition to fibre. The remainder will transition to fixed wireless or have no fixed line at all (relying instead on mobile communications not associated with any physical address).

If we use this 70 percent transition figure in an NZ context (an incremental 10% over where we are now), and assume we are sitting at at 60% take up rate right now, we can work out how much revenue Chorus can expect to lose from here on in, as 'NZ as a whole' transitions to fibre:

$7.27 per month x 12 months x 120,900 users transitioning = -$10.5m over a year, but echoing down all future years as well, from the remaining copper users expecting to transition from copper to fibre.

Based on last years NPAT figure, that represents a permanent NPAT profit decline of (0.72 x10.5m) / $52m = 15% ($7.6m in dollar terms).

This is in sharp contrast to the other broadband providers who are only making gains in the copper to fibre transition. To combat these impending ongoing revenue losses, Chorus will be under pressure to shut down the copper network where they can to avoid duplicating capital spend and maintenance.

SNOOPY

[Snoopy]

5/ Forecast fibre revenue from 26-03-2021 'Initial Asset Value' Presentation, slide 3. (FY2021 and FY2023 values interpolated).

Chorus have not provided any forecast as to where they expect their non-fibre revenue to go over the next few years. I have inspected the five year revenue trends for:

1/ 'copper connected revenue' and
2/ 'field services, value add network services and infrastructure'

as two groups. The latter group I am forecasting constant revenue of $120m over FY2021 to FY2024 inclusive. The 'copper connected revenue', comprising 'copper based broadband', 'copper based voice' and 'data services copper' have over HY2021 already declined on an annualised basis (that means no further deterioration over FY2021) of $70m. So I am forecasting an $80m decline over FY2021, $70m over FY2022, $60m over FY2023 and $50m over FY2024. The diminishing decline rate I am modelling to take account of a slowing trend as easy conversions to fibre have happened already. My table of forecast 'total change of revenue' over the period of interest is as follows.

	FY2016	FY2017	FY2018	FY2019	FY2020	FY2021F	FY2022F	FY2023F	FY2024F
Fibre Revenue	$133m	$202m	$276m	$368m	$468m	$592m	$715m	$735m	$755m
Fibre Revenue Increment		+$69m	+$74m	+$92m	+$100m	+$124m	+$123m	+$20m	+$20m
Fibre Revenue Increment Percentage		+51.9%	+36.6%	+33.3%	+27.2%	+26.5%	+20.8%	+2.8%	+2.7%
Non-Fibre Revenue	$875m	$838m	$714m	$602m	$491m	$409m	$339m	$279m	$229m
Non-Fibre Revenue Decrement		-$37m	-$124m	-$112m	-$111m	-$82m	-$70m	-$60m	-$50m
Non-Fibre Revenue Decrement Percentage		-4.3%	-14.8%	-15.7%	-18.4%	-16.7%	-17.1%	-17.7%	-17.9%
Total Revenue	$1,008m	$1,040m	$990m	$970m	$950m	$1,001m	$1,054m	$1,014m	$984m

Points to note on the above table

1/ The last four columns are 'forecasts' which is a sophisticated way of saying 'guesses'. The guesses are educated (the fibre bit by Chorus and the rest by me) but they could be wrong!
2/ Net migration of people to New Zealand from April 2020, thanks to Covid-19, has been near zero (previously around 5k per month), which has removed a tailwind of probable new customers.
3/ Spark and Vodaphone, now joined by Two Degrees, have stepped up their plans to sell 'fixed mobile' services over their mobile networks. These are lower cost competitors for Fibre Broadband for light broadband users. There is also a transition underway to 5G mobile networks. 5G is a more serious competitor to fibre broadband.

I believe that:

3i/ the aforementioned fixed mobile competition, PLUS
3ii/ the not as yet widespread adoption of 8k streaming video AND
3iii/ the slow take up of virtual reality gaming AND
3iv/ the fact that the fibre broadband roll out is coming to an end (the supply of customers just waiting to jump on has gone).

....means there is little justification for another incremental jump in access speed by consumers (and consequent higher broadband prices) right now. Hence the reason behind the forecast slowing in the growth of fibre broadband revenue.

4/ The carrying value of copper cables at EOFY 2020 was $360m and last year depreciation on those cables was $60m. This points to an accounting life going forwards (assuming no additions or disposals) of 6 years, starting from EOFY2020. Of course 'accounting life' and 'useful asset life' are not always the same thing. But we can say that if the copper network was effectively decommissioned by EOFY2026 (just five years away) that would not cause a ripple in the accounts. My somewhat aggressive forecast reductions in copper revenue up until EOFY2024 reflects this. Yet I do expect some legacy copper network which is not economical to replace with fibre or wireless systems to remain. That means I am not assuming the copper network will be completely dead by EOFY2026.

So how is my forecast going so far? Q4 2021 operational figures were released by Chorus today. So I will discuss that.

SNOOPY