Smartshares NZB New Zealand Bond Fund: 12m return 11-08-2023
Quote:
Originally Posted by
Snoopy
Excellent question Nor. Here is the link to the information page on the NZ Bond trust
https://smartshares.co.nz/types-of-f...sh/nzbondtrust
(Edit 15/08/2023): The above web-page reference is dynamic and appears to be updated daily. (This means that some of the numbers referenced in this post will no longer correspond to the numbers on that page).
If you go down to the 'portfolio characteristics' section, that shows a Bond Yield of 5.66%, for an average yield duration of 3.44 years. Fund charges at the top of the page indicate the 'annual fee' is 0.54%. So I would expect the gross return to unit holders to be something like:
5.66% - 0.54%= 5.12%
Yet historical gross dividend yield is a pitiful 1.66% ('Fund Details' top of page). Something is not adding up here! (Edit: see post 5 for the answer to this)
Returns for the year, after fees but before tax, were 1.78% over the last twelve months (as listed under NTA unit performance graph). According to the one year interactive chart, the unit price on 8th August 2022 (Actually the price for the trading day before the weekend which was 5th August 2022) was $2.99272. That means the value to unit holders one year later including distributions as at 8th August 2023 must be:
$2.99272 x 1.0178= $3.05055
The unit price on the interactive chart dated 8th August 2023 (actually the figure for the NTA on 7th August 2023) was $2.86737. The difference, which must be the distributions over the last 12 months, must therefore have been:
$3.05055-$2.86737=$0.183180, or 18.3c.
18.3c gives an historical gross yield based on the 7th August 2023 unit price of: 18.3c/$2.86737= 6.38%
That is very different to the 1.78% gross return declared. Would appreciate if you could double check my maths Nor. But I agree with your sentiment and expectations which do not appear to have been met.
The link to the information page on the NZ Bond trust
https://smartshares.co.nz/types-of-f...sh/nzbondtrust
The above web-page reference is dynamic and appears to be updated daily. This means that some of the numbers referenced in this post will no longer correspond to the numbers on that page). The numbers have changed a lot since I did these calculations four days ago. So I am redoing the calculations using 'today's' values.
If you go down to the 'portfolio characteristics' section, that shows a Bond Yield of 5.59%, for an average yield duration of 3.31 years. Fund charges at the top of the page indicate the 'annual fee' is 0.54%. So I would expect the gross return to unit holders to be something like:
5.59% - 0.54%= 5.05%
Yet historical gross dividend yield is a pitiful 1.66% ('Fund Details' top of page). Something is not adding up here! (Edit: JeffW has since answered this in post 5. The 1.66% represents four months of gross historical 'dividends in 'cpu'', divided by the current unit price. So no capital movements are included in this unit holder 'rear vision mirror' return).
Annual Return Reconciliation
Returns for the year, after fees but before tax, were -0.02% over the last twelve months (this statistic may be found under the NTA unit performance graph header). According to the one year interactive chart, the unit price on 11th August 2022 (Actually the price for the preceding trading day, 10th August 2022) was $2.99205. That means the projected value (using the benefit of hindsight because we know what the one year return to unit holders after fees but before tax turned out to be) one year later including distributions as at 10th August 2023 must have been:
$2.99205 x (1-0.0002)= $2.99145
Note that I am assuming that all quoted returns are 'before tax'. Why is that? Because a bond is a 'financial arrangement' for NZ income tax department purposes. That means an individual's tax bill depends on the timing between when that individual bought the bond and when they sold it. And the market value of these bonds fluctuates on the secondary market with time. Since these two dates are liable to be different for each individual investor, it would be misleading for a fund like this to publish a chart on 'after tax returns', while even hinting that is useful from the perspective of an individual holder. The only 'after tax return' the likes of Smartshares and their NZB fund could usefully chart, would be to assume that all bonds purchased were held to maturity.
The unit price (declared NTA) on the (linked url above) interactive chart dated 11th August 2023 (actually the figure for the NTA on 10th August 2023) was $2.86666.
a/ The difference between this figure $2.86666 AND
b/ the (with hindsight) projected and therefore actual fund balance after one year ($2.99145 as calculated above),
MUST BE the distributions paid out over the 12 month period under review. So comparing the projected return as measured by projected NTA (calculated with the benefit of hindsight) @10th August 2023, and the actual NTA (which is after the preceding twelve months of dividends has already been taken out) -also @10th August 2023- we see a difference:
$2.99145-$2.86666=$0.12479, or 12.5c.
Given what I have discussed in the paragraphs above, this figure is the likely gross return 'after fees' but 'before tax', which for a PIE entity would be levied at a rate of 28%. So the net income over the year would be 12.5c x (1-0.28) = 9c
12.5c gives an historical gross yield based on the 10th August 2023 unit price of: 12.5c/$2.86666= 4.36%
This doesn't tie in with the 1.66% after tax historical dividend yield quoted above. Could this mean that an extra charge that I have not identified is being deducted from the fund? In case you were wondering 'if I had 'got it wrong' and the 0.54% management fee had not been deducted from the graphed figures after all.....
0.0054 x $2.86666 = 1.5c
If I take 1.5c off my calculated distribution (12.5c-1.5c=11.0c) that is nowhere near enough to reduce my calculated yield down to the reported 1.66%. So this fund looks to have incurred some other loss which I am yet to identify.
The overall gross return after fees and before tax, (a loss in this case), was declared after one year to be -0.02%. But after tax this 'declared loss' reduced to 0.0%. This indicates that there must have been a tax refund at the NZB fund level.
The graph immediately above where all of these statistics are presented is labelled 'Net Tangible Assets per Unit'. IF this graph was showing an 'annual return' net of all fees after tax, THEN this would imply the unit price exactly one year prior and the unit price of 10th August 2023 would be the same number. But the numbers aren't the same ($2.99205 does not equal $2.86666). This is another hint that tax has NOT been deducted (or refunded) from the unit price chart.
SNOOPY
The hit from bond sales to meet redemptions: Part 1
Quote:
Originally Posted by
JeffW
Potentially there is zero taxable income, if there's been the sale of bonds at a loss, so no tax payable, but they're still distributed the cash interest received. Not sure, but a possibility
I am interested in investigating this theory further.
If I had my figures right, the annual return on the NZB fund (after fees but before tax) dropped from a miserable 1.78% (07-08-2023) to an absolutely pitiful -0.02% (10-08-2023) over three days. That three days was in the run up to the RB governor Adrian Orr's outlook for interest rate statement (on 16-08-2023) admittedly. 'The market' took that to mean interest rates would remain at an elevated level for longer. That could have caused some of the longer dated commercial bonds that NZB holds to fall in value. But there is another possibility that JeffW has mentioned above. That being sales of bonds at a loss to meet fund redemptions. So did that happen?
Looking at the daily statements on movements in the NZB fund, it did. On 10th August there were 145,647,077 NZB units on issue. On 7th August 145,717,077 NZB units on issue. That represents a drop of:
145,717,077 - 145,647,077 = 70,000 units
Sure enough on the 9th August 2023, a sale of 70,000 NZB units at a price of $2.86779 was reported. That represents:
70,000 x $2.86779 = $200,745.30
in dollar terms removed from the fund.
As at 31st July 2023, we were told that the NZB fund had a yield of 5.59% and a duration of 3.31 years. The five year government bond rate is the closest match to the 3.31 year duration of NZB bonds. Five year government bond rates rose from 5.37% (07-08-2023) to 5.42% (10-08-2023) over the three days period of concern. (Source https://www.rbnz.govt.nz/statistics/...interest-rates). That kind of interest rate raise should reduce 5 year bond values on an annual basis to:
5.37/5.42= 0.9908
of what they were before the interest rate rise. However, the NZB fund had an average 3.31 year duration. So the reduction in value over the whole fund due to that interest rate rise I estimate as:
0.09908^3.31 = 0.9698
Leaving aside the question of bond sales for the moment, this is the 'capital hit' I would expect the NZB fund to take as a result of interest rate rises. But what was the capital reduction that actually took place over those three days?
2.86665/2.86732 = 0.9998 (the capital loss was lower than expected. I will discuss this result in a later post)
Did the sale of 70,000 bonds, probably at a loss, to meet redemption requirements affect this? 70,000/145,717,077 = 0.05% of the portfolio. What was the coupon value these bonds were sold at? My rule of thumb is that company bond rates are a couple of percentage points above government stock rates. So I am saying these company bond rates probably sold for 5.42%+2%= 7.42%. The portfolio interest rate was 5.59% and average bond duration 3.31 years. I can therefore estimate the capital loss (actually fractional capital retained) on these bond sales to be:
(5.59/7.42)^3.31= 0.3916 =0.4
That means in round figures, 60% of the capital value of those bonds sold to meet repayments may have been lost. That sounds a lot. But for this one transaction, once you relate that to the capital value of the whole fund, it is not.
70,000 x $2.86779 x 0.6 = $120,447 (unit bond sales loss to meet redemption)
145,647,077 x $2.86779 = $417,685,230 (fund size)
=> Percentage of fund lost through selling at a loss = $120.447/$417,685,230= 0.028% (i.e. not material)
We have calculated that the sale of bonds over a three day period was not material to the fund value under the much larger downward value shadow of rising interest rates. But this was only considering a period of 3 days. Over 365 days, it might well be a different story.
SNOOPY
The hit from bond sales to meet redemptions: Part 2
Quote:
Originally Posted by
Snoopy
We have calculated that the sale of bonds over a three day period was not material to the fund value under the much larger downward value shadow of rising interest rates. But this was only considering a period of 3 days. Over 365 days, it might well be a different story.
I have been working my way through a whole year of NZB unit redemptions as laid out in post 10, and finally this exercise is finished: Result 8,020,000 units withdrawn.
The annual single year return for the NZB fund (after fees but before tax) was -0.02% (year to 10-08-2023). I now want to re-examine JeffW's idea of a possible explanation for this poor return: That being, sales of bonds are being made at a loss to meet fund redemptions. But this time I will look at data covering 365 days, not just 3.
On 10th August 2023 there were 145,647,077 NZB units on issue. On 10th August 2022 there were 102,951,618 NZB units on issue. That means overall there was a large inflow of money into the NZB fund over the year under consideration. Nevertheless that 'net increase' masks the fact that 8.02m units were withdrawn from the NZB fund during the year (for time line details of the withdrawals see post 10).
The underlying theory I am working on is that if a fund manager receives an inflow of new funds, then they can use their judgement as to when is the best time to invest those new funds. But if a unit-holder wants their money back, then the fund manager has no choice. They have to sell down immediately, whether the market timing is advantageous or not, to reimburse the unit holder.
For the purpose of this exercise I am assuming a 'worst case' scenario: Each and every one of those 8.02m units returned were sold at a disadvantageous price (a loss). In practice the NZB fund does run an ANZ 'NZ dollar current account'. That 'ANZ cash balance' may have offset the need for some of those disadvantageous price point sales that i am speculating exist. But bear with me. The idea is not to calculate an 'exact dollar value' of money lost. Rather, we want a ballpark figure of 'the possible money lost', just to see if such an amount comes anywhere near explaining why the NZB fund 'had a zero return' over the 10th August 2022 to 10th August 2023 year under examination.
As at 31st July 2023 (the latest available reporting record date as I write this), we were told that the NZB fund had a yield of 5.59% and a duration of 3.31 years. The five year government bond rate is the closest match to the 3.31 year duration of the total NZB bond weighted average portfolio. So I am using the five year government bond rate as a comparator. The five year government bond rates rose from 3.26% (10-08-2022) to 5.42% (10-08-2023) over the 365 days period of concern. (Source https://www.rbnz.govt.nz/statistics/...interest-rates). That kind of interest rate raise should reduce 5 year bond market spot capital values -on an annual basis- to:
3.26/5.42= 0.6015
of what they were before the interest rate rise. However, the NZB fund had an average 3.31 year duration. So the reduction in residual value over the whole fund due to that interest rate rise I estimate as:
0.6015^3.31 = 0.1859
Leaving aside the question of bond sales for the moment, this is the 'capital hit' I would expect the NZB fund to take as a result of interest rate rises. But what was the capital reduction per unit that actually took place over that year in question?
2.86665/2.92205 = 0.9810 (i.e. the actual capital loss was lower than expected, given my assumptions above.)
Did the sale of 8,020,000 NZB units, probably at a loss, -to meet redemption requirements- affect this? Well, 8,020,000/145,647,077 = 5.5% of the NZB portfolio. So I think the sale of those bonds potentially really does matter.
What was the coupon value these bonds were sold at? My rule of thumb is that company bond rates are a couple of percentage points above government stock rates. So I am saying the capital of these company bond rates probably sold on the secondary market for a yield of 5.42%+2%= 7.42%. The portfolio interest rate was 5.59% and average bond duration 3.31 years. I can therefore estimate the capital loss (actually fractional capital retained) on the residual capital from these bond sales to be:
(5.59/7.42)^3.31= 0.3916 =0.4
That means in round figures, 60% of the capital value of those bonds sold to meet repayments may have been lost. That sounds a lot. But how does this amount of cash relate to the total value of the fund? (Note that I have cast my eye over the unit values for the August 2023 year and settled on a unit value of $2.86 as typical, to use in this exercise).
8,020,000 x $2.86 x 0.6 = $13.762m (unit bond sales loss to meet redemption)
145,647,077 x $2.86779 = $417,685,230 (fund size, 10th August 2023)
=> Percentage of fund lost through selling at a loss = $13.762m/$417.685m = 3.29% (i.e. this is material)
Now go and have a look at the yields of the fund constituents 'as issued' (post 14). I haven't done a weighted analysis of those. But it does look conceivable that the weighted average interest rate received during the year (unadjusted for subsequent interest rate movements) was in the ball park of 3.29%. If that were true, we now have a credible explanation as to why this fund made a near zero return over the whole year ended 10th August 2023. I am not saying this is definitely the complete explanation. But it is at least a partial possible explanation that I intend to ruminate on overnight!
SNOOPY
The hit from bond sale redemptions: Part 3
Quote:
Originally Posted by
Snoopy
As at 31st July 2023 (the latest available reporting record date as I write this), we were told that the NZB fund had a
yield of 5.59% and a
duration of 3.31 years. The five year government bond rate is the closest match to the 3.31 year duration of the total NZB bond weighted average portfolio. So I am using the five year government bond rate as a comparator. The five year government bond rates rose from 3.26% (10-08-2022) to 5.42% (10-08-2023) over the 365 days period of concern. (Source
https://www.rbnz.govt.nz/statistics/...interest-rates). That kind of interest rate raise should reduce 5 year bond market spot capital values -on an annual basis- to:
3.26/5.42= 0.6015
The yield and average duration of the NZB fund, as quoted above, are a collective weighted average of the fund constituents. I was wondering if there was any single bond out there that would mimic this funds duration (3 years and 4 months from the 31st July 2023, meaning a maturity date of 30th November 2026). I figure a 'comparative study' of such bonds, over the year ended 10th August 2023 period, might provide a comparative template to what happened to NZB over that period. There will be a key difference though. Unlike NZB, when these comparative bonds are traded nothing is redeemed. The ownership of the bond simply passes from one owner to another. So there would be no 'redemption effect' on the bond price. That means by comparing the chart performance of such bonds with the chart performance of NZB, I can get a visual picture of whether the 'redemption effect' that I am hypothesizing is happening is real.
Here is a complete list of the 'fixed rate binds' available to be traded on the NZX
https://www.interest.co.nz/bonds-data/issues
Useful 'comparator bonds' are as follows:
Company & Bond |
Maturity Date |
Coupon Issue Rate |
Rating |
Argosy Property ARG020 |
29-10-2026 |
2.90% |
Not rated |
ASB Bank ABB110 |
16-11-2026 |
5.52% |
AA- |
Auckland International Airport AIA240 |
17-11-2026 |
3.29% |
A- |
Christchurch City Holdings Limited CCH030 |
05-11-2026 |
3.01% |
AA- |
Contact Energy CEN060 |
19-11-2026 |
4.33% |
BB+ |
Kiwibank KWBHA |
02-11-2026 |
4.93% |
Ba1 |
If you put any of these bond tickers into an investment charting tool like 'Stocknessmonster', the plot produced will have a vertical scale marked 'interest rate'. This is counterintuitive to those used to looking at share prices. Becasue the higher the interest rate goes, the lower the market value of the underlying bond. To preserve capital a consistently low interest rate, somewhere near the coupon issue rate of the bond is what an investor wants to see in their 'bond investment charts'.
Thus these bond charts are plotted in the opposite way to the collective NZB fund, which is priced on 'unit value'. For an NZB investor, they want the unit value as high as possible.
As a starting point I am looking at ABB110. With a coupon yield of 5.52%, it is close to the quoted yield of NZB at 5.59%. One interpretation of those two numbers is that Mr Market has assigned a similar 'risk profile' to the AAB110 bonds and NZB. The link to an ABB110 chart is here:
https://stocknessmonster.com/charts/abb110.nzx/
And the link to the comparative NZB chart is here:
https://stocknessmonster.com/charts/nzb.nzx/
This is the point where you have to twist you investment head a bit. For consistency, we are looking for the NZB chart and the ABB110 charts to behave in opposite ways. Both entities exist in the same NZ interest rate space. The chosen metrics that are quoted means that if you see the NZB chart going up, you would expect the ABB110 chart to be going down, and vice versa. That is what we expect. But what is actually happening?
Before I answer that question I need to draw attention to another 'comparator problem'. The range on each of those 'Stocknessmonster' charts is different:
i/ For NZB (in the last year it oscillates between $2.89 and $2.82, a variation of about 2.5%)
ii/ For ABB110 (in the last year it oscillates between 4.9% and 5.9%, a variation of about 17%)
This means if you were to overlay the NZB chart on the ABB110 chart, you would have to imagine the NZB chart 'squashed vertically' so that the variation in height of the graph is only 2.5/17= one tenth (approximately) the visual variation shown.
Comparator Observation
If I look at the ABB110 chart from about the beginning of April 2023, then I can imagine a steadily riding line or near constant gradient from a vertical axis figure of 5% up to a peak of 5.7%, with some variation like the sharp saw tooth peak of 5.8% reached in July 2023. Interestingly this July peak is matched by a saw tooth trough in the NZB chart over the same time period, which is exactly what I would expect (remember for equivalence these two comparator charts have to move in opposite directions). Yet something very strange is happening in this comparison show down. I would expect the rise in the AAB110 indicator in the period I am talking about, to be matched by a similar fall in the NZB unit price over that same time-frame. But I do not see that. Instead the NZB unit price is largely steady, starting out at about $2.87 and declining barely perceptibly to $2.86. IOW the NZB unit price is largely flat, when it 'should' be declining. Can anyone come up with a plausible explanation for this observation?
SNOOPY
P.S. This stuff is really testing me. I am finding it very difficult to understand how NZB works.