SPK- Spark NZ (TELCO)

[mondograss]

A lot of good questions there Snoopy. A couple of things to bear in mind:
1) Data sovereignty, most if not all govt data isn't allowed to reside offshore. Good for Spark you'd have thought, yet they've made very little headway in convincing the govt to let them host it. Yet Azure, AWS and CDC will almost certainly pick up a lot of work in this space. FWIW when Waikato DHB was hacked in that ransom-ware attack, it was using the Spark data center in Takanini. Though I'm not saying that was necessarily Sparks fault it will have tarnished their reputation somewhat.
2) Scale really matters in this business. AWS and Azure and even CDC have a buying power that Spark will always struggle to match when it comes to hardware and on the Azure\AWS front they put a lot of software development effort into their offerings, which Spark will never be able to match.

So I think Spark will be stuck mostly doing private cloud hosting of legacy systems that can't easily be migrated to public cloud and a bit of consulting. The big end of town will mostly go to Azure\AWS to get the global interconnectivity it needs and the govt will mostly go to CDC and probably public cloud for the less sensitive stuff now that there's NZ located data centers.

[Snoopy]

FWIW when Waikato DHB was hacked in that ransom-ware attack, it was using the Spark data center in Takanini. Though I'm not saying that was necessarily Sparks fault it will have tarnished their reputation somewhat.

Very interesting. I can't say I am pleased (being a Spark shareholder). But it is important for us to know these things.

SNOOPY

[676767]

This 'private wire' argument rings a bit hollow to me. I understand you can have a private connection at both your end and the data-centre end. But what about in between? Does each private connection have their own private wire that they hang themselves from telegraph pole to telegraph pole? Wouldn't the 'private wires' go into a common all enveloping common plastic conduit?

SNOOPY

Its a bit more technical than that, there isn't a 'private wire' per customer straight through to the Datacenter, they all run down the same lines. It's more a logically separated network just for you that is considered "trusted". This is separate from the Internet and other customers private networks so no one but you and you service provider can route through it. When using public cloud you would usually (although not always) need to access it via the Internet which would be running it over an "untrusted" network. Certain companies will try and avoid this for obvious reasons (security, data sovereignty, ddos, etc)

I am not fully on board with your definition of a 'hybrid cloud' Clip either. The whole idea of splitting private company held information information into two baskets. one where it doesn't matter too much if it is hacked, and another where it does, sounds problematic.

SNOOPY

Its less about 'this data is less important, lets send it to the cloud' and more about 'what is the most appropriate solution to host this'

Very few of the people moving to azure / aws will be doing a straight pickup and dump of their infrastructure from a private cloud provider to a public cloud one. If they are, they are doing it wrong and most likely are going to be spending more in the long run. The big change is moving from things like an on premise email server to O365 (I.e IaaS to SaaS). Or moving a self hosted Database server from Iaas to PaaS, where they now only need to administer a Database, rather than looking after and scaling a whole server that hosts a database as they previously had to. This is where Spark can't really compete with the big public cloud providers as they don't offer these solutions (nor should they).

However, there will pretty much always be a need for physical infrastructure which is where the 'hybrid cloud' strategy is coming in for a bunch of businesses.

Networking, Firewalling and having some compute or IaaS in a local datacenter is advantageous.
Big scalable workloads (Saas or PaaS) are pretty much always better in the cloud (webservers, databases, email, storage)

Designing, building and managing these networks are where the big bucks are for Spark...

[Snoopy]

eps = (Normalised Earnings) / (Total Shares on Issue, EOFY)

FY2017: ($396m - 0.72($20m) )/ 1833m = 20.8cps
FY2018: ($365m - 0.72($10m) )/ 1835m = 19.5cps
FY2019: ($409m - 0.72($15m) )/ 1836m = 21.7cps
FY2020: ($427m - 0.72($35m-$2m) - $10m -$7m)/ 1837m = 21.0cps
FY2021: ($384m - 0.72($28m - $16m) ) / 1867m = 20.1cps

Notes

1/ Figures for FY2017 and FY2018 are derived from the re-reported profit figures as presented in the December 4th 2018 Analysts Briefing, titled 'Updates to External Reporting'. These updates alter the financial reporting for the FY2017 and FY2018 year as though the subsequently applied new accounting standards NZ IFRS15 (on apportioning revenues and costs) and NZ IFRS16 (on the balance sheet representation of leases) were already in force over FY2017 and FY2018. Doing this means that all calculated results are compared under the same set of accounting standards.

2/ For FY2017, I have removed the $20m of profit that resulted from the one off sale of surplus land on Mayoral Drive.

3/ For FY2018, I have removed the $10m of profit resulting from a 50% sale of formerly 100% owned subsidiary 'Connect 8 Limited' (an infrastructure civil construction business).

4/ For FY2019 I have removed from profit $2m from the sale of a long term investment/business, $11m from the sale of Property Plant and Equipment and $2m from a gain on lease modifications and terminations, making a grand total of $15m to be adjusted for.

5/ For FY2020 I have removed $35m of 'other gains' (that includes $5m from the sale of a long term investment or former subsidiary, $28m from the sale of Property Plant and Equipment and $2m from a gain on lease modifications and terminations). I have offset this with $2m of rent concessions that would not have been granted outside of a Covid-19 environment. Furthermore I have removed a one off $10m downward adjustment to the tax bill that was a result of a law change reinstating the depreciation allowance on commercial buildings. Finally I have brought in a retrospective adjustment of $7m from FY2021. This adjustment relates to "reflect a reduction in net earnings of $7 million for the amortisation of reacquired rights that were previously regarded as indefinite life and therefore not amortised."

6/ For FY2021 I have removed $28m of 'other gains' (that includes $1m from the sale of a minority shareholding in long term investment 'Now New Zealand' (a boutique broadband retailer), $9m from the sale of Property Plant and Equipment (primarily mobile plant and equipment) and $18m from a gain on lease modifications and terminations). I have offset this against a one off refund of $16m of historic wire and maintenance charges that was charged to some fibre broadband customers.

Original Concluding Remarks: Some of the changes in profit from year to year could almost be seen as rounding errors. I don't want to make a judgement just on that. I need to sleep on this result before I decide what to do.

Conclusion: Average normalised earnings over the past five years was 20.6cps. The high of 21.7cps was 5% higher than average and the low of 19.5cps was 5% less than the average. This is just about as flat as corporate earnings get in the real world. My BT4 test (post 1810 below) has discussed the reduced likelihood of cost cutting in the future being able to maintain profits. My growth initiative research (post 1812) does not give any hint of a significant boost in top line revenue going forwards. Looking at the trend in the last five years of earnings, as used in the 'Buffett Model', is meant to be the indicator of whether a company is able to continue positive 'eps' growth into the future. The spreadsheet part of the Buffett valuation model does not work unless this is the case. There is no clear historical pattern of increasing earnings per share here. Furthermore it looks like increasing 'eps' into the future is going to get harder. For these reasons, I cannot see any reason to overturn the actual 'eps' numbers calculated because of unusual market conditions (e.g. the Covid-19 environment). The numbers do tell the story. The result of this test for Spark is a 'fail'.

Buffett Test 1 may be found in post 1806. Not enough has changed year to year to warrant me rewriting it. So we are straight into Buffett Test 2.

eps = (Normalised Earnings) / (Total Shares on Issue, EOFY)

FY2018:	($365m - 0.72($10m) )/	1835m	= 19.5cps
FY2019:	($409m - 0.72($15m) )/	1836m	= 21.7cps
FY2020:	($427m - 0.72($35m-$2m) - $10m -$7m)/	1837m	= 21.0cps
FY2021:	($384m - 0.72($28m - $16m) ) /	1867m	= 20.1cps
FY2022:	($410m - 0.72($26m) ) /	1872m	= 20.9cps

Notes

1/ Figures for FY2018 are derived from the re-reported profit figures as presented in the December 4th 2018 Analysts Briefing, titled 'Updates to External Reporting'. These updates alter the financial reporting for the FY2018 year as though the subsequently applied new accounting standards NZ IFRS15 (on apportioning revenues and costs) and NZ IFRS16 (on the balance sheet representation of leases) were already in force over FY2018. Doing this means that all calculated results are compared under the same set of accounting standards.

2/ For FY2018, I have removed the $10m of profit resulting from a 50% sale of formerly 100% owned subsidiary 'Connect 8 Limited' (an infrastructure civil construction business).

3/ For FY2019 I have removed from profit $2m from the sale of a long term investment/business, $11m from the sale of Property Plant and Equipment and $2m from a gain on lease modifications and terminations, making a grand total of $15m to be adjusted for.

4/ For FY2020 I have removed $35m of 'other gains' (that includes $5m from the sale of a long term investment or former subsidiary, $28m from the sale of Property Plant and Equipment and $2m from a gain on lease modifications and terminations). I have offset this with $2m of rent concessions that would not have been granted outside of a Covid-19 environment. Furthermore I have removed a one off $10m downward adjustment to the tax bill that was a result of a law change reinstating the depreciation allowance on commercial buildings. Finally I have brought in a retrospective adjustment of $7m from FY2021. This adjustment relates to "reflect a reduction in net earnings of $7 million for the amortisation of reacquired rights that were previously regarded as indefinite life and therefore not amortised."

5/ For FY2021 I have removed $28m of 'other gains' (that includes $1m from the sale of a minority shareholding in long term investment 'Now New Zealand' (a boutique broadband retailer), $9m from the sale of Property Plant and Equipment (primarily mobile plant and equipment) and $18m from a gain on lease modifications and terminations). I have offset this against a one off refund of $16m of historic wire and maintenance charges that was charged to some fibre broadband customers.

6/ For FY2022 I have removed $26m of 'other gains' (that includes $10m from the sale of Property Plant and Equipment (primarily mobile plant and equipment) and $16m from a gain on lease modifications and terminations)

Conclusion: Average normalised earnings over the past five years was 20.6cps (the same as FY2021). The high of 21.7cps was 5% higher than average and the low of 19.5cps was 5% less than the average. This is just about as flat as corporate earnings get in the real world. Next a review on input costs.

Financial Year	Labour Expense	Finance Expense
2018	$513m	$77m
2019	$475m	$85m
2020	$511m	$94m
2021	$491m	$81m
2022	$495m	$74m

The table above shows labour expenses have been held down, and interest costs are at cyclical lows. But I would expect upward pressure on both these cost staples as we look to FY2023. Growth initiative research (post 1982) does not give any hint of a significant boost in top line revenue going forwards. Nevertheless mobile service revenue growth continues to exceed expectations, which largely covers the 'new class' growth initiatives and cloud services not performing to expectations

Looking at the trend in the last five years of earnings, as used in the 'Buffett Model', is meant to be the indicator of whether a company is able to continue positive 'eps' growth into the future. The spreadsheet part of the Buffett valuation model does not work unless this is the case. But it is clear there is no clear historical pattern of increasing earnings per share here.

Conclusion: Fail test.

SNOOPY

[Snoopy]

ROE= (Normalised earnings) / (Shareholder Equity EOFY)

FY2017: $382m / $1,601m = 23.9%
FY2018: $358m / $1,483m = 24.1%
FY2019: $398m / $1,465m = 27.2%
FY2020: $386m / $1,493m = 25.9%
FY2021: $375m / $1,503m = 25.0%

Conclusion: Pass Test

ROE= (Normalised earnings) / (Shareholder Equity EOFY)

FY2018: $358m / $1,483m = 24.1%
FY2019: $398m / $1,465m = 27.2%
FY2020: $386m / $1,493m = 25.9%
FY2021: $375m / $1,503m = 25.0%
FY2022: $397m / $1,475m = 26.9%

Conclusion: Pass Test

SNOOPY

[clip]

Confirmed. From slide 5 of 1HYP2023
"Cloud mix-shift trend continues, with volume growth in lower-margin public cloud and co-location being offset by lower private cloud volumes and repricing."



I understand the drive to send production of goods offshore to save labour costs. But I fail to see why putting your data into large overseas data-banks should be cheaper than putting that data into the same electronic hardware run from a data-centre in New Zealand. You don't need to look far to see that there is not a strong economic argument to run datacentres offshore. If it was cheaper to do this overseas, why are Azure/AWS and Microsoft planning to set up data-centres of their own here in New Zealand?

Why do Spark consider they 'cannot match' these overseas providers? Their 1HY2023 report indicates, without spelling it out in those words, that they are suffering from corporate cost bloat in their data-centre business. If they can fix that, why can't they go toe to toe with the overseas players?



This 'private wire' argument rings a bit hollow to me. I understand you can have a private connection at both your end and the data-centre end. But what about in between? Does each private connection have their own private wire that they hang themselves from telegraph pole to telegraph pole? Wouldn't the 'private wires' go into a common all enveloping common plastic conduit?

Furthermore, if you are truly worried about privacy, why would you entrust your data to a US controlled corporate that operates outside the envelope of New Zealand law? I guess many NZ businesses are not worried about that, which is why they are moving to Azure/AWS and Microsoft. But it seems to me such businesses are giving away a lot, to save very few -if any- dollars. At least if you have a problem with Spark, you can phone them in New Zealand and speak to a New Zealand operator and they can send a New Zealand technician down to sort out a New Zealand domiciled electronic box. I think Spark has a real trump card to play here in the service aspect of what they do.

I am not fully on board with your definition of a 'hybrid cloud' Clip either. The whole idea of splitting private company held information information into two baskets. one where it doesn't matter too much if it is hacked, and another where it does, sounds problematic.

SNOOPY

Mondo and 67 have basically covered all responses. Pricing - economies of scale. Spark's Takanini datacenter can supposedly support up to 1200 racks IF full based on this article. Amazon datacenters have 2000 racks based on this article and they have 87 data centers. With scale everything becomes cheaper - power, cooling, computer components, land/lease space etc.

Racks is a VERY rough unit of measurement to compare scale. A rack is basically a cabinet you put servers and networking components in - they are units of physical space. They could be more or less full. E.g. I have a previous customer where we rented 1 whole rack of space in a data center - however, in that, we only had 3 physical servers.
In a standard size rack, you could physically fit more than 40 servers.
Spark may be renting out physical space e.g. 1 whole rack of space to a customer, whether it is fully utilized or not. Whereas Amazon/AWS rent out services hosted on their own servers in that space, meaning AWS racks are generally full and are utilizing all available space/resources. I don't know what Spark offer in terms of renting physical space, if they keeps rack full etc.
Spark's cost for power, cooling, lease does not decrease much whether they are 50% utilized or 100% utilized. So if they don't have good uptake, their profit margins are smaller, or they have to raise prices to cover costs, further pushing customers towards cheaper public cloud options.

Re. private wire - and what about inbetween - it's the same wires, but a router at each end creates a "private tunnel" between the 2 endpoints, so the data sent between them is encrypted and can't be accessed by others. For example, my house is the office, and your house is the data center, and I am sending mail (data) to you. When mail goes from my house to your house, it has a very strong password and giant padlock in it, so the data sent between us is private. Doesn't mean that we have a physical road/wire between our houses that nobody else can use.

A big decider of whether you go public/private cloud comes to data sovereignty, as others said, you're not allowed to store govt data outside of NZ, previously this was the case also with medical data although this has changed over the past years and Azure/AWS is now OK. There may be other companies with specific policies around where data can be stored - e.g. if you have commercial secrets you may require that to be hosted in NZ. Or your password vault may need to be stored on your own premises, while your data itself that is accessed with those passwords is on public cloud - meaning you are running a hybrid cloud environment.

[Snoopy]

Net Profit Margin = (Normalised Profit) / (Operating Revenue)

FY2017: $382m / ($3,505m - $20m)	= 11.0%
FY2018: $358m / ($3,533m - $10m)	= 10.2%
FY2019: $398m / $3,518m	= 11.3%
FY2020: $386m / $3,588m	= 10.8%
FY2021: $375m / $3,565m	= 10.5%

Notes

1/ Turnover across FY2017 and FY2018 has had revenue from asset sales removed from the revenue total.
2/ Turnover from asset sales is not included in the revenue that I have quoted for FY2019, FY2020 and FY2021.

I want to drill down a bit more into that big profit margin lift [ (11.3%-10.2%)/10.2% = 10.8% in one year, which is >2% inflation] between FY2018 and FY2019. If I look at the segmented result (AR2019 p52),

Product Category	Operating Revenue (FY2019)	Product Margin (FY2019)	Product Margin %ge (FY2019)	Operating Revenue (FY2018)	Product Margin (FY2018)	Product Margin %ge (FY2018)
Mobile	$1,271m	$775m	61.0%	$1,237m	$732m	59.2%
Voice	$486m	$310m	63.9%	$573m	$369m	64.4%
Broadband	$685m	$344m	50.2%	$665m	$350m	52.6%
Cloud, Security & Service Management	$400m	$327m	81.8%	$370m	$315m	85.1%
Procurement and partners	$365m	$43m	12.3%	$357m	$40m	11.2%
Managed Data, Networks & Services	$197m	$104m	52.8%	$207m	$111m	53.6%
Other Operating Revenues	$114m	$51m	44.7%	$114m	$49m	43.0%

I can see the product categories with the highest 'Product Margin' are:

1/ 'Cloud, Security & Service Management' ' and
2/ 'Voice'.

But 'Voice' was on the decline (-$87m in lost revenue on a year vs year comparison). When a high margin business unit declines like this, it puts pressure on the up and coming 'product unit revenues' to fill the gap. Cloud, Security & Service management is too small to do that on their own (+$30m - equivalent to +$30m x 81.8/63.9= +$38m in revenue indicative of profit substitution terms). But combine that with the growth in mobile revenue, also very profitable, at: +$34m x 61.0/63.9=$32m and for broadband +$20m (equivalent to a profit equivalent revenue offset of $20m x 50.2/63.9 = +$16m) and we are covering those lost 'voice unit' profits: $38m + $32m + $16m = $86m. This means profit indicated revenue offset from the remaining business units is minimal (actually slightly negative). Yet none of this explains the increasing profit margin 'year on year'.

It looks like the biggest contributor to the profit margin increasing was a very significant labour cost saving (a $475m-$513m=$38m drop in the wage bill over the year- see table below). There is a lot of automation behind the scenes that will be contributing to this. But can it continue? Here is what has happened to the wage bill in the five years under review.

Financial Year	Labour Expense	Finance Expense
2017	$550m	$75m
2018	$513m	$77m
2019	$475m	$85m
2020	$511m	$94m
2021	$491m	$81m

This table suggests to me that the big savings in labour have already been made. Alongside of this, I have recorded the annual finance expense (the second column of the table).

A fall in the 'under 3 month' NZD Commercial debt ($228m-$155m=$73m), both in capital being borrowed and interest paid on that capital, looks like the explanation for interest payments saved over FY2021 (see AR2021 p89). There are $100m in domestic notes maturing in each of FY2022 and FY2023 too. Refinancing those at lower interest rates should take the pressure off the interest bill in the next two or three years. Yet the picture that is emerging here is that taking cost out of the business looks to be an exercise that will get more and more difficult going forwards. That means it will be increasing revenue on the 'top line' from the up and coming revenue sources going forwards that will be required to 'fuel growth' going forwards.

In summary, I think there is room for 'surprise on the upside' if some of the new business units start to get traction in their specialty markets. I think 5G could yet prove to be a profit margin game changer too. Yet doing this cold hard numbers exercise has reinforced to me that Spark is not a sure bet. The fact that there has been some quite good execution of the business plan over the last five years, while the net profit margin has shrunk - albeit not significantly so - seals the fate of Spark for me when lined up against this criterion.

Conclusion: Fail test

Net Profit Margin = (Normalised Profit) / (Operating Revenue)

FY2018: $358m / ($3,533m - $10m)	= 10.2%
FY2019: $398m / $3,518m	= 11.3%
FY2020: $386m / $3,588m	= 10.8%
FY2021: $375m / $3,565m	= 10.5%
FY2022: $397m / $3,694m	= 10.7%

Notes

1/ Turnover across FY2018 has had revenue from asset sales removed from the revenue total.
2/ Turnover from asset sales is not included in the revenue that I have quoted for FY2019, FY2020, FY2021 and FY2022.

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

How much do the level of interest rates affect normalised profit? I don't normalise for those! But I do note that while the interest payments on debt have decreased year on year ($81m -> $74m), the amount of capital borrowed has increased ($1,403m -> $1,526m) (see AR2022 p108)! And this in spite of Foreign Currency Medium Term currency borrowing going down on paper (in fact the FMTNs have not changed in their home currency - this is a snapshot NZD exchange rate effect.) The increase in debt is termed a 'temporary working capital impact' (PR2022, slide 22) .

One reason for the interest paid drop is the switch to new lower rate 'green bond (sustainability related)' funding ($365m dollars worth). This replaced the $160m of borrowings formerly taken out with the Hong Kong and Shanghai bank and Mitsubishi UFU Financial Group Bank. (I believe this $200m jump in overall debt will be reversed when a 'planned surplus' of $200m is banked from the TowerCo proceeds.)

Short term debt (due within 12 months), -mostly on variable interest rates-, represents $293m/ $1,526m = 19.2% of the total (down from 26.6% in the previous year). If short term debt rates rise an incremental 2 percentage points, this will increase the company interest bill by: 0.02 x $293m = $5.9m. That is back to where the interest bill was at EOFY2021, so not a massive deal. And I haven't adjusted that figure for the $200m reduction in debt, once the TowerCo surplus proceeds are absorbed.

Of course, all this excludes the effect of 'lease interest' due on the sale and leaseback of Spark's network cellphone towers to 'TowerCo'. I would regard such money as a 'network running expense', rather than a true finance cost.

In summary, you don't need a mathematicians eye to describe the profit margin history of this company. It is flat.

Conclusion: Fail test

SNOOPY

[Snoopy]

If the odd one off tax adjustments were taken out of the FY2021 Spark result, then the underlying tax bill for the year was even less, only $155m. $155m is $181m - $155m = $26m short of the actual underlying tax bill paid by Spark. So:

1/ IF $181m was the actual tax paid to give 100% imputed dividend AND
2/ $155m was the normalised requirement for tax to pay THEN

the underlying dividend imputation rate at Spark over FY2021 was only $155m/$181m = 85.64%. This is more in line with earlier imputed tax rates of 80.6% and 75% paid over FY2019 and earlier.

Hi Snoopy

I did a bit of a dive into the Spark tax numbers. Keep in mind that IC's are deducted from the ICA when they are paid, not declared so the recent final dividend will not appear in the ICA until the interim report next year (being FY22H1).

Imputation Credit Account: Transaction Summary	Balance	Reference
ICA Account 01-07-2020	$0	(AR2020 p94)
add Taxes Paid	$188m	AR2021 p65 (AR2021 Cashflow Statement)
less IC's attached to Dividends	-$179m	(1)
less Other	-$27m	(balancing figure) (2)
equals Closing Balance 30-06-2021	-$18m	(AR2021 p101)

Notes

1/ $179m being the total of: FY20H2 Div $230m (AR2021 p91) + FY21H1 Div $231m (AR2021 p91):
= $461m x 0.389 = $179m

Sub Note: 0.389 is the number you apply to the declared dividend to get the IC's where they are fully imputed (.389 = .28/.72). Declared dividends paid during FY2021, being FY2020H2 (paid September 2020) and FY2021H1 (paid March 2021) were both fully imputed.

2/ Full reconciliation on current and deferred tax all stacks up, except for the -$27m above.

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

The fact they are showing tax payable of $23m (AR2021 p63, Balance Sheet) suggests they have not paid more tax than they should have. A reconciliation of the current tax account shows:

Tax Payable (Current Liabilities)	Amount	Reference
Opening Balance	$44m (3)	(AR2020 p55, Balance Sheet)
add Current Year Tax Expense	$172m	(AR2021 p100)
less Current Year Adjustment	-$4m	(AR2021 p100)
less Taxes Paid	-$188m	(p65 AR2021, Cashflow Statement)
equals Closing Balance	$23m	(p63 AR2021, Balance Sheet)

(3, Snoopy clarification) To make the column addition work, this figure needs to be $43m. This indicates the $44m quoted may be subject to rounding error.

Instead of having paid more tax than they should have, they owed (and paid) tax for the previous year, being $39m (AR2020 p94). So the $188m paid was $39m for the prior year and $149m for the current year ($188m - $39m = $149m). Note the opening balance was $43-$44m, less I'm assuming the $4m current year tax adjustment to give a payment for the prior year of $39m.

Tax Expense:
Tax Calculation (AR2021 p100) & Profit & Loss Statement (AR2021 p62) both show $169m

Current Year Tax Expense	$172m	(AR2021 p100)
less Year Tax Adjust prior period	-$4m	(AR2021 p100)
less Deferred Tax Expense	-$4m	(AR2021 p100)
add Deferred Tax Adjustment prior period	+$5m	(AR2021 p100)
equals Tax Expense Classified	$169m	(AR2021 p100)

Rather than being funny business with tax, I view the amount payable as being real (ie Spark owes $23m) and that for whatever reason they went OD on the ICA account. It could be that included in the $188m of tax payments were overseas tax payments that do not count towards the ICA. Notice in the Reconciliation of Income Tax Expense (AR2021 p100) there are two adjustments for $6m, one being taxes paid overseas and the other being for a different tax rate.

Either someone miscalculated the available IC's or there was an unknown tax receipt/refund (4) that put a spanner in the works. Notice the prior year had $1m tax receivable (AR2020 p55, Balance Sheet) but AR2021 p63 (Balance Sheet) has $0 tax receivable and does not show any tax receipts - I suspect any such receipts have been netted off against the payments made so we do not have 100% visibility. I am curious as to what penalties they will cop for the negative ICA balance.

(4) Maybe the ICA was good when the dividend was declared but there was a tax receipt/refund between the payment date and year end. Or a combination of all 3.....receipt + stuff up + non-NZ payments.

FERG

P.S. Don't use the $155m - that is a notional tax figure and, other than in the AR note/reconciliation (AR2021 p100), does not appear in any journals made by Spark or in any amounts payable or paid or tax returns etc.

I am still intrigued as to how over a considerable period of time Spark can pay fully imputed dividends of 25cps seemingly significantly ahead of their core earnings (refer post 1994). So I have reformatted into tabular form Ferg's work he did on this topic in 2021 (see reformatted quoted post above), and will now perform a similar exercise on the FY2022 results.

Imputation Credit Account: Transaction Summary	Balance	Reference
ICA Account 01-07-2021	-$18m	(AR2021 p101)
add Taxes Paid	$160m	AR2021 p82 (AR2022 Cashflow Statement)
less IC's attached to Dividends	-$182m	(1)
add Other	$24m	(balancing figure) (2)
equals Closing Balance 30-06-2022	-$16m	(AR2022 p120)

Notes

1/ $182m being the total of imputation credits attached to: FY21H2 Div $233m (AR2022 p110) + FY22H1 Div $234m (AR2022 p110):
= $467m x 0.389 = $182m

Sub Note: 0.389 is the number you apply to the declared dividend to get the IC's where they are fully imputed (.389 = .28/.72). Declared dividends paid during FY2022, being FY2021H2 (paid September 2021) and FY2021H1 (paid March 2022) were both fully imputed.

2/ Full reconciliation on current and deferred tax all stacks up, except for the $24m above.

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

One way to pump up your imputation credit balance beyond what might be expected is to pay your tax in advance, before it is due. The fact Spark are showing 'tax still payable' of $40m (AR2022 p80), on the Balance Sheet. suggests they have not paid more tax than they should have. A reconciliation of the current tax account shows:

Tax Payable (Current Liabilities)	Amount	Reference
Opening Balance	$23m (3)	(AR2021 p63, Balance Sheet)
add Current Year Tax Expense	$177m	(AR2022 p119)
add Current Year Adjustment	+$1m	(AR2022 p119)
less Taxes Paid	-$160m	(p82 AR2022, Cashflow Statement)
equals Closing Balance	$40m	(p80 AR2022, Balance Sheet)

(3) To make the column addition work, this figure needs to be $22m. This indicates the $23m quoted may be subject to rounding error.

Spark owed (and subsequently paid) tax for the previous year, being $19m (AR2021 p101). So the $160m of tax paid in the cashflow statement over FY2022 was $19m for the prior year and $141m for the current year ($160m - $19m = $141m). Note the declared opening balance was $22m-$23m before adding back a $1m current year tax adjustment to give the 'IRD book closing tax payment' for the prior year of $23m.


Tax Expense Statement
Tax Calculation (AR2022 p119) & Profit & Loss Statement (AR2022 p79) both show $171m

Current Year Tax Expense	$177m	(AR2022 p119)
add Year Tax Adjust prior period	+$1m	(AR2022 p119)
less Deferred Tax Expense	-$8m	(AR2022 p119)
add Deferred Tax Adjustment prior period	+$1m	(AR2022 p119)
equals Tax Expense Classified	$171m	(AR2022 p119)

Summary Conclusions

A/ At the close of FY2022, Spark owes taxation authorities $40m. There is nil offset taxation recoverable to set against this figure (tax losses are held in Australia, but are not on the books - presumably because there is little chance they will ever be utilised).

B/ For reasons undeclared, the New Zealand imputation credit account had turned negative at both previous balance dates (by -$16m at EOFY2022). Nevertheless, we are told that the imputation credit account was restored to an unspecified positive balance on 31st March 2022, in accordance with IRD requirements (see post 2001 for an explanation of this).

C/ The 'other' imputation credit account adjustment which I have highlighted in 'red' in both my post and the post I have referenced by Ferg, remain opaque. I accept these net adjustments must have happened, because the imputation account transactions have to tell the story of the year on year change in imputation account balance. But, despite how official these figures look in the tables, as presented, they are 'fudge factors'. I remain concerned as to how imputation credits of the order of -$27m and $24m respectively can 'disappear' and 'appear' like this. The only 'saving grace spin' I can put on this, is that over a two year window these mysterious changes have almost cancelled each other out (-$27m+$24m = only -$3m).

I raise the matter of 'imputation credit balance', not because it is an obscure accounting point. The reason for my close attention is because the imputation credit balance available determines Spark's ability to pay fully imputed dividends. And if, for example, 20% of the dividend were to suddenly becomes not imputed, that would significantly affects Spark's value proposition as an investment.

SNOOPY

[Sideshow Bob]

https://www.nzx.com/announcements/409262

Spark New Zealand commences on-market share buy-back

Further to previous market announcements, Spark New Zealand Limited (Spark) has announced today that it is commencing an on-market share buy-back of up to $350 million on 6 April 2023.

[Snoopy]

The following post is talking about the FY2021 tax year......

Spark's accounting year ends 30 June

IRD EOY for Imputation credits is 31 March

Spark will report for their year to 30 June - reported to IRD - but considered 31 March year for tax purposes.

(For a dividend paid in April 2021, Spark) will have all of 2021/22 year (IRD tax year ending 31-03-2022) to make up those credits attached to the dividend, in tax payments.

Thanks for that information.

Spark's interim dividend was paid also on 9 Apr 2021 - in the subsequent 31 March 2022 IRD Imputation year (Snoopy note: although this payment was still in the Spark FY2021 reporting year that ends on 30th June 2021). So this may result in a negative ICA account at 'Spark balance date', if credits are attached to dividends paid as they were in the Interim 9 April 2021 Div pay-outs.

I am not sure I fully appreciated what you were saying in this post from over a year ago nztx, when you wrote it. Sorry I am a bit slow .

But now I see that it is:

A/ The mismatch in the IRD tax reporting year (ending 31st March) and the company reporting date (30th June) - COMBINED WITH
B/ Actual dates that provisional tax payments to the IRD are due

- that has caused the 'negative imputation credit issue' in the Spark Annual Report for FY2021 Please allow me to explain in my own words, that which you already know......

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

The IRD were happy, all the way through this scenario, because -as far as they were concerned- for the company tax year ended on 31-03-2021.....

Both of the dividends paid in the twelve months prior to that 31-03-2021 IRD balance date, -being the dividends paid on:

i/ 03-04-2020 (the interim dividend for the Spark FY2020 year) AND
ii/ 02-10-2020 (the final dividend for the Spark FY2020 financial year)

- were fully matched off against tax paid by Spark, and the consequent imputation credits, available on the IRD books.

Now we move to the 09-04-2021 dividend payout. This was paid out of earnings that are subject to provisional tax payable over the IRD FY2022 year (even though this payment was made during the Spark FY2021 financial year). Provisional tax can be due on different dates, depending on what provisional tax system is opted into.

https://www.ird.govt.nz/income-tax/p...rovisional-tax

The three provisional tax system choices are either :

1/ the 'Ratio Option',
ii/ the 'Standard Option' or
iii/ the 'Accounting Income Method'.

We don't know which of these methods Spark has chosen. But we don't need to know that. Because the earliest that any provisional tax payment is due under any of the three methods is 28th May 2021. That in turn means that if Spark chooses to pay a dividend 'early' on 09-04-2021, then the IRD are not 'worried about the Spark imputation credit account going negative'. As far as the IRD are concerned, tax payments from those earnings are not yet due, That means the imputation account for those 09-04-2021 dividend feeding earnings did not even exist on 09-04-2021, at the time this dividend was paid! As long as the IRD get their tax due on the appropriate provisional tax date, the earliest being 28-05-2021, then Spark has fulfilled their fiduciary tax paying duty, and the IRD are happy.

Any negative imputation credit balance in the Spark balance sheet, might be best thought of as a reminder of an IRD tax bill that is coming up. Yet it is not negative from an IRD perspective, because it represents a tax bill not yet due.

Contrast that to the Spark perspective, where not only have the earnings that enabled that 09-04-2021 dividend to be paid been banked. They have already been passed on to shareholders as a dividend. And consequently the associated tax liability for Spark to eventually pay to the IRD tax on those profits has been 'set in stone'. It is this that has been recorded in the Spark balance sheet as 'income tax due' in the form of a 'negative imputation credit'. So perhaps it is best to think of those negative imputation credits as a reminder to Spark that the dividend associated with those imputation credits has already been paid, and they had better not miss paying these imputation credits to the IRD when due!

SNOOPY