For anyone else contemplating such a move, here is how I did it and where I encountered tricky bits.
The basic idea was to build a portfolio with the highest return and the least volatility/risk through selecting assets that are as a non-correlated as possible, i.e. they do not move and down together - one zigs while the other zags. In addition, the assets themselves are not single companies, they are many companies held together as a mutual fund or Exchange Traded Fund (ETF), in order to remove the effect of single company events.
The end results of my many hours spent on the planning of the renovation are shown in the extract from the spreadsheet I used that is included with this post.
Here are the key features:
- no individual companies eligible for the portfolio, only ETFs and mutual funds to ensure diversification; however, for my 27% Canadian bond allocation I decided to keep the bond ladder I have built; the range of maturities from 2007 to 2026 probably suffices to be diversified and I don't sacrifice yield from management fees e.g. XBB has an MER of 0.30%
- the range of asset classes is both geographical, covering more or less all the main economic areas of the world - Canada, the US, Europe, Japan, Pacific nations, emerging countries such as Russia, China, India, Korea, and sectoral, with large cap and small cap companies, value companies (low price/book and other measures), real estate, commodities; these asset classes have been chosen based on multiple sources that demonstrate their un-correlated returns over time - books such as those of Richard Ferri and of Roger Gibson, online resources such as Bylo Selhi and Index Fund Advisors
- the diversifying sectoral asset classes each only have a small percentage of the total and where possible this is spread amongst the various geographical areas, in order not to be too concentrated in any one asset class;
- no asset class has less than 2% of my total asset allocation and the total number of holdings is 16; this is to ensure that any holding is sufficiently large that when rebalancing comes along in a year, the dollar amount to be rebalanced will be large enough that trading costs will not chew up the amount; as a rough rule of thumb I'll want the rebalancing amount to be at least $1000 since my trading cost is $10/trade (1%); Ferri recommends no more than 15 holdings to keep the rebalancing exercise from being too complicated and I fully agree based on my own experience of doing rebalancing amongst only six ETFs the last few years (XIU, XGD, XIT, XRE, XFN, XMD); given that I have four accounts (open, RRSP, LIRA1 and LIRA2), none of which can transfer money to another, the limit on number of holdings is a practical necessity. It was at this point that a number of candidate asset classes, seen in the bottom chart with the ETF ticker symbols, disappeared from the final allocation.
- the reduction of the impact of adverse currency shifts, especially against the US dollar, is the concern that led me to buy more than half of my US large cap holding and US small cap as the currency hedged XSP and new iShares Canada XSU; in the international holdings there is some currency risk but it spread over many countries' currencies so it is more acceptable - whether this is truly an acceptable risk I don't know as I have not been able to find clear answers in the finance research I have read so I will keep looking. In the meantime, however, there really isn't much choice to find ETFs that will fill the asset classes in my chart except to buy unhedged ETFs on US markets, so I've decided that diversification is most important. The only exception might be XIN, the hedged counter-part of EFA, the iShares EAFE fund, but since the global whole market class did not make it into my portfolio it isn't relevant. Of course, it is true that despite the funds selected being traded on US exchanges in US dollars, the currency risk is not the US dollar but those of the under-lying holdings; the US$ is only a pass-through currency.
- Vanguard funds appear often in my final selections - VGK, VPL, VNQ, VBR, VV, VWO. As my chart shows, there are other good funds (I didn't put any in the chart that weren't low cost, passive index funds) but Vanguard comes out on top pretty well every time they are available in a class. Why? Vanguard ETFs have the lowest tracking errors; a lower Price/Book in the value category; lower turnover than the avg fund in a category = lower transaction costs, less cap gains distributions; more holdings than the avg fund in a category =more diversification, less risk; the lowest management fees. It may interest some to note that I compared a few TD e-Series funds (TDB900, 904 and 911) but their higher management fees offset their plus of not having trading costs. My fund selections are highlighted in the greeny/yellow colour on the chart.
- The case of the US dollar denominated holdings caused me something of a conundrum. All foreign dividends are taxed as ordinary income, which suggests US$ holdings should go into registered accounts while Canadian holdings that can benefit from lower dividend tax rates should go into the open account. However, it is an unfortunate fact that my broker BMO Investorline won't hold US dollars within a registered account. BMOIL still requires US dollar sales to be converted into Canadian dollars and then back again into US$ for a purchase. There is a spread of about 1.8% between the buy and sell rates on the US$ at BMOIL so when I came to do my rebalancing next year I'd get dinged plenty, probably more than the tax differential. So I will keep as much as possible of my US$ holdings in my open account where I can keep a US$ cash balance. BMOIL could fix this deficiency since there isn't any legal impediment to having US$ in a registered account; I'm sure as soon as another of the major bank brokers does it they will too but it rankles in the meantime to pay extra FX costs for no necessary reason.
- my portfolio is treated as one across all the four accounts I have - open/taxable, RRSP, LIRA1 and LIRA2; where I put individual holdings is influenced by other considerations such as taxes but taxes are subservient to doing proper asset allocation. All the fixed income/ bond holdings are in the registered accounts for instance. Dividing up the holdings amongst these accounts proved to be a significant challenge just to make things balance. In addition, I tried to anticipate the rebalancing by placing investments within the same account that might end up needing to be rebalanced up or down e.g. I put a big chunk of the DJP commodities holding into the RRSP along with VGK Europe equity and VPL Pacfic equity as they are historically negatively correlated and the former should go down when the others go up. Similarly, in the open account I put VNQ REITs and VV US large cap, which are also negatively correlated. Hopefully, that will enable easy rebalancing within an account next year but we shall see.
- my asset allocation is NOT driven at all by any income needs, though I am starting into that period of life when withdrawal is beginning (for why this is the correct way to approach things, I rely on experts such as Moshe Milevsky). If I will need to withdraw funds from any account, It will be based on the asset allocation - whichever is above its target will have a portion sold.