My last post, reviewing 2020, observed that my performance is, superficially at least, very similar to Vanguard’s WoRLd equity tracker ETF VWRL. Despite my portfolio involving a helluva lot more complexity/faff. My post elicited this comment from Bob:
Thanks for sharing, intriguing as always. As someone who recently (18 months ago) simplified my portfolio into three holdings: 1 VWRL seven figures, 2 [single megacap tech stock] six figures, 3 Vanguard Global Bond six figures. I find myself reading about your complexity and not feeling jealous one bit. So the question is, why do you dislike VWRL (or similar global tracker) so much? You mention the comparison several times, what is stopping you making the change? That is after all what reviews should lead to e.g. insights, and change.Bob, commenting on 2 January 2020
Bob’s challenge is a good one. Why wouldn’t I just swap out my entire portfolio for, say, holding only a single world equities tracking ETF like VWRL or its non-Vanguard equivalents (see Monevator’s updated list of alternatives here, or the SRI alternatives listed on my ETFs page)?
How exactly does my performance compare to VWRL?
As it happened even before this comment I had started a deeper dive into this question. The graph below plots my monthly performance, since Jan 2013, versus VWRL (share price only; this ignores the ~2% annual dividend yield). As you can see, there is a tight correlation – with an R^2 of 0.869 (R^2 is a measure of correlation between 0 and 1, where 1.0 means completely correlated, 0.0 means no correlation at all). In contrast, my R^2 versus VUKE and versus VUSA are both a lot lower, at 0.74 and 0.50 respectively. So VWRL is indeed a close fit with my portfolio. I owe blog-izen zxspectrum48k my thanks for first bringing this to my attention a couple of years ago.
The other analysis I did here is to consider the ‘unitised’ return, per unit of risk. My favourite measure of this is the Sharpe ratio; higher equals more return for a given level of risk (volatility), where anything above 1 is generally considered decent (professional, hedge fund quality, ahem). A first order approximation of Sharpe suggests my Sharpe is 1.06, VWRL’s is statistically-similar 1.08. VUKE’s, by contrast, is a dismal 0.07 – reflecting more volatility than my portfolio, and far lower returns. But the standout Sharpe performer is the S&P 500; IUSA’s Sharpe is almost 1.3 – off, incidentally, high volatility but even higher average returns. A S&P index tracker has beaten almost everybody for years (though the MSCI USA index has done markedly better of late, due to Tesla).
So, based on this initial scrub, I would do well to replace my entire portfolio with VWRL. That set me thinking.
What are the sources of my return?
When you dig into any portfolio, its returns come from different components. Classic portfolio theory, for instance, divides a return into the market’s performance (beta/passive) and the return over and above (or, often, below) the market (alpha/active).
The starting point for breaking down any portfolio though is the markets it is investing in – what its beta is, in effect. This is the asset allocation question. My mental model is that asset allocation accounts for 80% of a portfolio’s return. The year 2020 makes it easy to see why – here are the returns of the obvious geography/asset markets I look at, in 2020:
Your ability, with any UK-focused equity portfolio, to beat almost any US-focused equity portfolio, in 2020, was practically nil. And, likewise, if you held UK/US bonds in 2020, you more than likely did OK – but not as well as if you had a broad basket of US stocks.
My asset allocation is, in fact, significantly different from VWRL. We are both fairly similar with our US weighting. But my portfolio is 21% UK weighted, deliberately far higher than the UK’s 4% global weighting these days (down from 6% not too long ago, sigh) – reflecting my home being in the UK. VWRL’s international (i.e. non USA) exposure is correspondingly far higher than mine, at 36% compared to my 22%. My Australia weighting is also deliberately high, at 6%, compared to what I estimate is a 1.5% VWRL weighting.
And this difference in weighting makes a big difference. My 5x greater exposure to the UK, in a year when the UK stockmarket fell by 11.5% and other key markets all rose, hurt me. VWRL’s weighted average markets rose by 10.4%, in fact. Whereas my weighted average markets rose by only 6.8%.
Against VWRL’s benchmark rise of 10.4%, VWRL’s actual performance of about 9.5% does not look quite so good. I assume dividends are the difference. But in the meantime, my own return of 8.5%, compared to my market’s average of under 7%, looks quite a lot better.
This suggests that, rather than owning VWRL, perhaps instead I should simply mirror VWRL’s asset allocation? In fact I think my asset allocation is a better fit for me than VWRL’s, with my life being orientated around the UK far more than an ‘average global stock market investor’. This year, my UK exposure cost me, but imagine if the opposite had happened – the UK had risen by 17% where the US had dropped by 11%. In that case I would be feeling pretty sour if I was down almost 10%, while all the ‘nearest’ assets were up 17%, and my relative financial firepower was down by 25% relative to my home market.
In fact my allocation includes a significant weighting towards bonds, whereas VWRL has none. But in 2020, the weighted equity return in my geographies was +6.6%, and my weighted bond return was +6.8%. There was negligible difference. This isn’t always the case, though bonds have done surprisingly well for surprisingly long.
While I am considering where my return comes from, I have three experimental portfolios to consider. Each has been running for most of the last 8 years, albeit not tracked as carefully as my overall portfolio:
- High Yield Portfolio. This miserable portfolio is a clear dog. I have for some years owned a handful of ‘guaranteed income’ holdings, and leveraged them up cheaply. In theory this produces a nice high single digit return with relatively little risk. In practice the holdings have always disappointed, and my returns suck – I don’t know exactly but I think annual returns are under 5%, less than half of my portfolio average, with approximately similar volatility. I keep holding on, telling myself “one more year”, mainly because I have never felt I understood clearly what is going wrong. In my defence, the total size of this portfolio is less than £50k – about the same as many of my single holdings. Nonetheless, after 8 years, I’m pulling the plug.
- Dividend Growth Portfolio. Another subaccount I run is a ‘Dividend Champion’ portfolio – of mostly USA stocks that have a long track record of increasing dividends. This portfolio has been a solid performer; I don’t have exact numbers but I would say it tracks S&P minus the tech component. But at the end of the day I think I’d be better with an S&P500 index tracker.
- Tech growth. My other subaccount contains only tech stocks – mostly USA, big and small – and needless to say it has done very well indeed. Again, no numbers, but it is in danger of lop-siding my portfolio. It has grown to be about 10% of my total equity portfolio. And I also have significant tech holdings mixed into my other accounts. Given that NASDAQ grew over 40% in 2020, it is reasonable to think that my tech investments alone drove all my outperformance in 2020.
Active vs passive
The other obvious analysis to do here is to compare my ‘beta’ with my ‘alpha’. This is not an easy analysis to do properly, so I haven’t. But if I break out my performance based on type of holding, by geography, I get a rough feel for the breakdown. Green represents outperformance versus the equivalent index tracking ETFs (shown in bold), pink represents underperformance.
The analysis above is approximate. It ignores dividends/coupons, for instance, which makes my UK individual bond holdings look a lot worse than they are (as they yield 6-7% each).
But some patterns in 2020 stand out:
- My passive holdings tracked effectively, more or less. The bond ETFs look a bit skewiff but don’t allow for coupons. In some countries my equity ETFs outperformed the markets, in others they lagged – but they were definitely on trend.
- My stock picking did OK, in English speaking markets, in the UK, USA and USA. This is primarily about my tech holdings – my Australian portfolio includes a significant holding in XeRO, for instance, which almost doubled last year. In the UK, where I don’t have any tech holdings, my direct shares lost about 7%, but the market and ETFs lost more.
- My stock picking elsewhere did not work. This makes sense to me. I have a handful of European stocks, none of which have done very well. I also made a significant error by investing in Airbus in January, which promptly halved. I cut my losses. I don’t know these stocks well; I am buying large megacaps like Adidas, Heineken, Airbus and others. I think also that whereas the UK FTSE-100 has some clear duds in it (banks, Vodafone, BT, the list goes on), so by buying large UK businesses that are not duds gives you a reasonable chance of outperforming, if you go for Eurostoxx 60 you are less likely to own the walking dead. This argument is basically that the top 60 European companies (which will include 5-10 UK companies) are better, on average, than the top 60 UK companies, which sounds a bit unpatriotic but reflects the ‘pollution’ I perceive in the FTSE from banks and extractors.
- My active equity funds are generally doing OK, albeit not in the USA. Again, there is some logic for using fund managers to pick stocks in emerging markets, and even continental Europe. They know them better than me.
- My bond funds are losing to ETFs. I have been sceptical about the ability to index track bonds, so have swallowed the relatively high fees that come with bond funds. But the evidence above is fairly clear – I should stick to bond ETFs, not the funds.
While these conclusions are coming from only one year of data, I would say that they feel more robust than that – I am pretty sure they would hold over 5+ years too.
Tax efficiency & fees
For many portfolios, when you look at the return you see tax and fees make a material difference. Both are, mathematically, equivalent – they are deductions from the gross portfolio return.
In my case, my analysis effectively ignores tax. I pay tax ‘offline’, so to speak. And even where I make withdrawals from the portfolio to pay the tax bills, by unitising my portfolio I strip out the effects of additions or withdrawals – so what you see on my blog is analysis of just the underlying returns.
As to fees, I have reduced my blended average fees to just 0.41% of my investment portfolio. This is not negligible, and will reduce my returns accordingly, but it is within the margins of error of my analyses in this post. VWRL’s fees of 0.22% are lower than mine, but by less than 20bps. I could, perhaps, compensate for that with superior stock picking.
This leaves one more potential component of my returns to last. My leverage. I’ve been targeting a 12% level of leverage, though for most of 2020 I was also running tactically overweight cash – reducing my typical net leverage to 8-9%. In principle this boosts both returns and volatility (but holds Sharpe constant, I think). By comparing Sharpe ratios I think I am stripping out the effect of leverage. But in absolute return terms it will have an effect.
My leverage of about 12% in theory amplified my return, but came with a financing cost too of about 1.5%. This has helped my returns by a bit, but is not a big factor relative to the geographic asset allocation and the tech overweight.
What are the sources of complexity?
Having considered my sources of return, I’m now turning to the sources of my complexity. Whereas I’d like as much return as possible, for a given amount of risk, I would like as little complexity as possible! Hence the thought experiment about replacing my entire portfolio with one holding of VWRL.
The first driver of complexity for me is having multiple brokerage accounts. Even within one brokerage, if I have a General Investment Account as well as an ISA account as well as a SIPP account then this is three accounts, and even holding just VWRL in each account would represent three lines in my tracking spreadsheet.
I have veered from too many accounts to having too few. My main argument now for having several is the government guarantee, via the FS Compensation Scheme (FSCS), which has a cap of
£50k £85k per investment account. I kind of assume that this limit will in practice apply to per-person-per-brokerage. And on that basis I have accounts with 6 brokerages – providing me with, I hope, at least £500k of guarantee. Mrs FvL has accounts with 3 of the same, plus a 4th different one. So 7 platforms to track, and 16 different unwrapped/wrapped accounts across them – 9 of which are taxable. I am comfortable with these numbers.
If I held simply VWRL and GBP cash in each account, I would have 32 rows in my spreadsheet. Add in some USD/EUR/AUD holdings, and my effective minimum is more like 38 rows. Of this 9 are taxable, there would be 9 tax reportable holdings to deal with – across me and Mrs FvL. This would be a very welcome improvement in the current reporting, it must be said.
Diversity across fund providers
The next source of complexity is my desire to avoid having all my eggs in one ETF provider’s basket.
In principle though, I could ‘alternate’ my holdings between VWRL and another alternative (which needs thought – as iShares don’t have an obvious one), so each account had only one ETF holding in it. On this basis I could theoretically keep my number of spreadsheet rows unchanged, at fewer than 40, and hold up to 16 different ETFs across, in theory, as many as 16 different providers.
Not wanting to realise capital gains
A key driver of complexity is my longstanding aversion to selling holdings. This has left me with over £1m of unrealised capital gain.
However right now there is so much media chatter about capital gains tax rates increasing that, if anything, my unrealised gains are providing quite a strong incentive to take action now. If I were to realise £1m of gains now, crystallising a £200k Capital Gains Tax liability, I could reset my costbase. If CGT rates went up to, say, 40%, I have halved my liability on any subsequent sales. Against this there is the risk that CGT rates don’t change, or change a lot less than mooted in the media, and I have paid the tax for ‘nothing’.
Overall I think the risk of significant hikes in CGT are real enough that they make me relatively sanguine about enduring a six figure CGT bill. I also feel that after the strong returns over the last few years, my portfolio ‘can afford’ a significant tax bill – especially if that tax bill is smaller than a single month’s swing several occasions over the last year.
Other sources of complexity
Leverage, an unusual aspect of my portfolio, adds negligible complexity. In fact it adds flexibility, and reduces the need to sell holdings when I need access to liquidity.
Obviously my desire to do some active stock picking is a significant source of complexity. Right now I have about 60 individual holdings. This is higher than I need to be an effective stock picker, but is not absurdly large – based on other portfolios one reads about in the blogs.
Some of my active stock picking is in the form of ‘sub strategies’. These have between 6 and 15 holdings each. I quite like these from a reporting point of view – it makes it easy to see my monthly performance by strategy.
A key source of complexity for me is Experimentation. I have been running, for example, a small High Yield Portfolio for years – with half a dozen small holdings in it. This HYP is in an unwrapped account, so it is landing in the tax return. I like the ability to Experiment, and think this aspect is worth allowing 20 or so rows in the spreadsheet for.
Edging towards a more refined strategy
In all of this discussion I am mindful of my Investment Policy. This policy hasn’t changed, materially, in the eight years I have been tracking my portfolio rigorously. And its principles – diversify, use tax shelters, prefer passive, etc – remain the right ones for me. But at the same time the policy has a lot of flexibility – it would support me splitting my money equally across VWRL and five near equivalents, in my existing wrapped/unwrapped accounts, for instance.
Moving forward, I am tempted to adopt the following approaches – but would very much appreciate feedback on this post before I follow through completely!
1: Stick to the existing asset allocation. My UK overweight has cost me in relative performance to VWRL, but it suits me and my personal circumstances. I wouldn’t feel comfortable having as little as 4% of my invested wealth in the UK, the market that I know best and where I do have at least the potential to see some outperformance from active investing. Likewise being underweight on Europe and Emerging markets works for me. My professional and personal life are entwined with the UK, USA and Australia – so I’m happy concentrating there.
2: Dump the substrategies. I intend to axe my existing substrategies. However I will keep the accounts I use for them, and continue to use them as testing zones, value £50k, with max 10 holdings in total.
3: Adopt a new ETF diversity policy: keep as few ETF holdings as possible, subject to my asset allocation and my existing accounts, and capping any single provider (iShares, most likely) at 25% of my total portfolio.
4: Focus single stock picking on English language markets – i.e., in my case, UK, USA, Australia. In these markets there is at least some plausible evidence that I have ‘edge’ – my individual holdings outperform the index. These securities to be held in unreported accounts (ISAs/ SIPPs) if at all possible, and if not then my Ltd company portfolio.
5: Focus active funds on sectors/geographies that I don’t know so well. For markets where I do not have ‘edge’, the question is ‘index track’ or ‘give it to professionals’. I tend to think the professionals don’t earn their keep in the UK and USA, where markets are closer to efficient, and thus index tracking here is better. But in less liquid markets I suspect professionals can do better – and a couple of my holdings reinforce that belief. So I’m OK using active funds in E.g. continental Europe, China, other emerging markets.
6: Unwrapped accounts should be for a few large, passive holdings. Or, at a push, core holdings that don’t pay dividends – like big tech stocks such as AMZN and GOOG. My numerous single line stocks, especially those that pay dividends, should be ideally in my wrapped-and-unreported accounts (ISAs/SIPPs).
Targets for 31 March 2021:
No more than 50 unwrapped holdings
I want to own no more than 50 holdings in unwrapped accounts, i.e. that aren’t in a tax wrapped account like a SIPP or ISA. Fifty still sounds like a lot, I hear you think. After all, why can’t an unwrapped account look like this one?
And not like this one (truncated at the letter E!):
For starters, I currently own 39 ETFs listed below. I am sure this can be pruned down to 30, but am not so sure about getting down to 20.
Then I have twenty or so holdings that I regard as ‘Core’ – including individual stocks such as Amazon, Disney, Nestle, Xero, and some funds I am happy to own such as Blackrock European Dynamic Fund, the City of London Investment Trust, and so on. A couple of these are not holdings I would buy into an wrapped account these days but they have significant unrealised gains which I am not minded to crystallise, possibly ever.
I also have a couple of illiquid holdings, in the (unwrapped) private bank account, which I can’t sell.
So, getting my unwrapped holdings down to 50 is a tough target. It does however feel like the right next step for me.
In terms of ‘unwrapped’, I in fact have different layers of wrapping to contend with:
- 10/10 Fully unwrapped – in my name. My personal name, in a ‘general investment account’. These funds are taxable at my marginal rate, which is 45% income tax and 20% capital gains tax. And this activity is all reportable on the tax return. My initial goal is to get the number of holdings in this category down to max 50.
- 6/10 In a lower tax name. Funds that are in Mrs FvL’s unwrapped accounts are at a slightly lower tax rate than mine. Her marginal income tax rate is around 40%, and her marginal capital gains tax rate may even be lower than 20% (i.e. we don’t always use her £12k annual allowance in full). And funds that are in my Ltd company pay tax on both income and gains at corporation tax rates, which are currently 19% but rumoured to go up probably around 25% before long. The activity here is also all reportable to the tax authorities.
- 3/10 In an offshore bond wrapper. I have a small offshore bond wrapper. Income in this account is free of tax, and I can withdraw the original capital tax free over 20 years. However any subsequent distributions, of any sort, are taxable at my marginal tax rate. I think of the effective tax rate here as being about half my own tax rate. Until I take any distributions, none of this activity is reportable to the taxman.
- 2/10 In a SIPP. Funds in the SIPP are tax free, and not reportable, until I take distributions or cross the lifetime allowance of about £1.1m. However 75% of distributions are fully taxable in my name. And I am on course to hit the lifetime allowance in the next few years, at which point any gains above that level are fully taxable.
- 0/10 In an ISA. The best of the best, certainly for any funds above/beyond any pension limits. No tax or reporting obligations at all, ever. Under current policies.
Max 20 holdings duplicated in more than one account
I have a lot of holdings that are duplicated across accounts. I know how this is happening – I reinvest dividends in each account, usually against the same underweight allocation, often aiming for the latest ‘watchlist’ stock, or if in doubt a core ETF like VUSA or INXG (both of which I have in 6 separate accounts). I have 78 holdings that are in more than one account, and six holdings that pop up in five+ accounts.
The complexity cost is significant, with more reporting, more logging than is necessary. My investment tracking spreadsheet gives me the ability quickly to see my consolidated picture, and removes any need for accounts to be somehow diversified or balanced within themselves (though I do have a couple of accounts that I like being ‘self contained’, which both have about 9 holdings in). So part of my streamlining will be to try to identify the best account/wrapper for each stock and concentrate my holdings there. I don’t know quite what the right target is here but I think I should be able to pare down duplication to max 20 holdings, all of which I will own at least £50k of.
I have started this streamlining already. My brokers are going to have a bumper month, as I am doing a lot of transactions. I will report back after the end of this UK tax year – on the 5th April 2021.
I’d be interested in comments. Have you considered which ETF most closely tracks your portfolio performance? How much duplication do you have across accounts? How do you trade off diversification vs simplicity?