jalanlong wrote: ↑Tue Jul 06, 2021 4:43 pm
D1984 wrote: ↑Sun Jun 27, 2021 1:52 pm
3. The third thing I want to mention (and then I promise I'll stop as this post is rather long already) is that much of Intl ex-US's crappy performance since 1989 or 1990 is due to it weighting Japan at such a high amount at the peak of the Japanese asset price bubble....in other words, this is a flaw of a purely cap-weighted index;
I do think that is a huge issue but I dont know a way around it. I have researched at least 50 broad based international etfs and every one of them has approximately 40% in Japan, UK and Germany. So the fact that Finland or Italy has such great returns during a specific time has no bearing on my possible returns as a US investor. The only way I see around this is to buy single country ETFs and equal weight them which seems like a lot of work. Plus even those funds are cap weighted so that the largest components are the multinational firms that tend to follow the US market anyway.
This is one of those situations where you wonder why no major ETF issuer has offered a GDP-weighted (or at least an "EAFE Lite" type of index fund like I mentioned in my post) EAFE or EAFE+Canada ETF. The costs to run it wouldn't be very much with commissions as low as they are today; plus an ETF only needs a minimum of around $24 or $25 million to be self-sustaining in terms of paying for itself without an absurdly high expense ratio. There is an index fund (MAIIX) for the regular cap-weighted EAFE that has an ER of 0.09 and another one (VTMGX) for the regular cap-weighted EAFE + Canada that has an ER of 0.07; it should be possible to do either an equal-country weight fund, an EAFE Lite (or EAFE+Canada Lite), or a GDP-weighted developed markets fund for maybe a 0.10 or 0.12 ER tops (even a true equal stock weighted developed international markets ETF could be done for probably under a 0.23 or 0.24 ER). Maybe this is a circular chicken-and-egg case of "the market doesn't provide it because it doesn't perceive anyone wants it since no one seems to be demanding it, but no one is demanding it because they don't perceive the market wants to provide it." The difference in returns from 1989 or 1990 to today would've been fairly significant. Until an ETF issuer comes up with such an ETF (or even a mutual fund) I'll just stick to actively-managed funds for international exposure....at least within my tax-sheltered space.
Using the IShares individual country ETFs isn't that great of a solution either; they have ERs of 0.51% or higher...if I'm paying that kind of fee I'll take my chance with an actual active manager, thank you very much.
PS - Just some data for EAFE (or similar ex-US developed country index) based on
ANY form of weighting besides naive regular cap-weighting. I have ran backtests for all of the following:
1. A nominal GDP-weighted EAFE index (actually, MSCI itself provides this one back to 1970) based on nominal GDP for each country.
2. A "true constant $ at PPP" GDP weighted EAFE index (using constant 1990 Geary-Khamis Int'l PPP weighted dollars from the Angus Maddison database for 1970-2010 and at-PPP constant 2017 Int'l dollars from the World Bank's database for 2011 to 2020); doing this helps remove any overweighting or underweighting of countries that might occur over several years just because a country's currency strengthens/weakens relative to those of its peers while its economy--and thus its underlying actual real GDP--grows at roughly the same rate or so as those peer countries.
3. An equal-weighted EAFE index (by stocks, not by country) with data provided from MSCI from 2020 back to the mid-1990s and from a gentleman who goes by the handle "NovelInvestor" before that back to 1974 (using data he obtained from MSCI).
4. An equal country-weighted EAFE index (where each country gets an exact equal weight and the index is rebalanced every December 31st for the next year....in other words, if there were, say, 12 countries in the index then each would get an 8.33333% weight regardless of the countries' actual market cap or GDP)
5. A slight variation on #4 where I--in order to answer any potential accusation of "well of
COURSE the equal country-weighted index did better....it only had Japan at a tiny portion of its actual cap-weight since there were upwards of 16 countries in the EAFE even back in the 1980s.....you knew Japan would underperform from 1989 onwards so this is cheating"--
deliberately added a 20% Japan weight to the equal country-weight EAFE index each year. By this I don't mean I set Japan's weight at just 20% each year; what I did was set Japan's weight at 20% PLUS whatever it's actual index weight was; in other words this index each year consisted of an 80% amount of the equal country weight EAFE (which obviously included Japan at whatever its weight was in said index) and then the other 20% of the index was the MSCI Japan Index. This means that (for example) if in a particular given year there were 20 countries in the equal country-weighted index (19 of which were not Japan and one of which was Japan) then Japan would get a weight of 5% in this index; as such, Japan's total weight in the "EAFE Equal Country Weight plus 20% Japan" index for that year would be 24% (since 5% of 80% is 4% and 4% + 20% is 24% ); obviously Japan's weight in the index would vary a bit year by year based on how many other countries were in the EAFE index for any given year but adding 20% extra in Japan regardless of that helps to answer the charge that I am deliberately skimping on Japan's weighting just because I know in hindsight that its stock returns stunk for 20+ years.
6. The same as in #5 above but putting Japan at 25%
plus whatever its actual index weight was rather than 20% plus whatever its actual index weight was.
7. The RAFI Developed Ex-US Large Cap index; this is a fundamentally-weighted (earnings/revenues.dividends) index rather than being cap-weighted.
8. Various "EAFE Lite" or "Japan Light" indices that fixed Japan's weight at the LESSER of either:
A. Its actual index weight, or
B. 20%, 25%, or 30% of the total index weight.
So for example, for the "EAFE Japan Lite 25" Index if Japan's index weight at the beginning of a given year was, say 18.22% (or indeed any number between 0.01% and 25.00% ) then the weighting--and the returns--would be the same as for the regular cap-weighted EAFE index; if Japan's weight was, say, 45.85% then it would instead be capped at 25% and the rest of the index countries would be proportionally increased in weighting as per their cap weights. Since MSCI provides an EAFE Ex-Japan Index (the EASEA Index) back to 1970 it was relatively easy to set up a spreadsheet to calculate all this using the known returns from the EAFE, EASEA, and MSCI Japan indices.
I would also like to add that before anyone says this is "cheating" because I am targeting only Japan by capping its index weight and that I am only doing this since I know that Japan underperformed from 1989 to the mid 2010s and thus have the unfair advantage of hindsight bias...well, to be frank the only country (at least after the 1970 inception of the MSCI EAFE index) that this would've affected would've been Japan simply because no other EAFE country was ever in a giant bubble big enough to make it upwards of 30% of the index (a bubble which I might add put Japan's cap weight at far greater than its GDP weight or weight relative to the world economy at large); Japan was at around 65% of the EAFE at its peak weighting in late 1988 or early 1989!
Before 1970 the only effect of such and index cap weight for any one country in a hypothetical pre-1970 EAFE or ex-US developed index (at least if at 25% or 30% level) would've likely been to underweight Britain somewhat (vs all the other countries in the index) in comparison to the pure market cap weighted version of such an ind..this would have probably been in effect from the early 1900s to maybe the early or mid-1960s. The net result would have been (and to be fair this is just an estimate--and a very crude one at that--from data taken from "Triumph of the Optimists", from the Jorda macrohistory database, from various individual country equity return datasets for France, Denmark, Austria, Switzerland, Australia, the UK, and Canada, and from an excess equity returns by country by decade study from Bridgewater) that the "Britain capped at 25 or 30%" index would've outperformed the cap-weighted index in the 1900s and the 1930s, would've barely underperformed the cap weighted index in the 1920s, would've underperformed the cap weighted index in the 1910s and 1940s (albeit it would've outperformed the cap weighted index in 1940 and 1941 when it was most needed for a US investor---when US equities had two negative years....it was the almost all of the rest of the 1940s where it would've underperformed), and would've outperformed (vs the cap-weighted index) in 1960, barely outperformed the cap-weighted index in 1961, underperformed the cap weighted index pretty hard from 1962-1966 (led by poorer-than-the-UK returns for France, Switzerland, Italy, Germany, and only barely equal to the UK returns for Australia for these years overall), probably lost by a tiny bit to or roughly equaled the cap-weighted index in 1967, and then creamed the cap-weighted index for 1968 and 1969 (led by somewhat-better-than-UK returns for Canada, Italy, France, and Germany, and by much better to in some cases stunningly better returns than the UK for Japan, Australia, and Switzerland....given that Japan by this time would've been a part of the developed markets index since at least 1966 or 1967).
Most of these returns would (at least for the decade as a whole) not differ much from the cap weighted return; they might be plus or minus it by maybe half a percentage point a year or less for the given decade as a whole (albeit some individual years would see some dramatic differences one way or the other....see the part about the 1960s above.
The one exception would likely be the decade I didn't mention above....the 1950s. British equities didn't do badly during this decade, to be sure, but
any kind of non-cap weighted index (be it GDP weighted, equal weighted, equal country weighted, or a "cap any one country at 20/25/30 percent of the index regardless of actual market cap weight" ) would've beaten the cap weighted index soundly because it would've allocated relatively more to countries that either barely beat or roughly equaled Britain in returns for this decade (Canada and Sweden), lost to Britain by a hair for the decade as a whole but would've beat it during many years that the alternate weighted index/es would've had a relatively higher allocation to said country (Australia comes to mind) or countries that absolutely pulped (or rekt/pawned/murked/clobbered/insert your choice of adjective for "killed" here LOL) the UK in equity returns from 1950-59; places like Austria, Italy, France, Netherlands, Finland, and most of all Germany (which managed a roughly 25% real CAGR for this decade).
Of course, after 1-1-1970 I don't need to guess how these alternate indexes would've done because I have data (either from MSCI, from NovelInvestor, or from my own calculations using actual indices from MSCI and weightings from other credible sources).
9. One other "alternate EAFE index weighting" bears mentioning here as well (although strictly speaking it technically isn't an actual index per se)....I think it's been mentioned on the Bogleheads board--by Rick Ferri if I recall correctly but don't quote me on that as I'm not 100% sure he's the one that posted it--that simply splitting EAFE into Europe and Pacific and equal weighting them 50/50 and rebalancing each year beat the EAFE pretty nicely; I tried this myself from 1966 to 2020 (1966-69 data for Pacific being roughly cap-weighted Japan and Australia rebalanced each year--I didn't include any other Pacific countries markets' because Singapore and Hong Kong weren't in the developed markets category until 1972 or 1973 and New Zealand wasn't even added until the mid-1980s; Taiwan and South Korea are still classified to this day as emerging markets by MSCI so they weren't included either--and 1966 to 1969 data for Europe being taken from the annual "Europe stock returns" from Ibbotson and Sinquefield's equity risk premium paper published in 1983 or 1984; from 1970 onwards actual MSCI data for Pacific and Europe were used); actually, if I recall correctly from my own experiments in with this in Excel doing
ANYTHING from 90/10 to 10/90 (for either Pacific or Europe) beat the EAFE...I think I still have an Excel file with this data on it if you want to see for yourself; this "beating the EAFE despite how exactly it weights Europe and Pacific over a very wide range" phenomenon is likely because it didn't overweight Japan at the height of the bubble! Note that including Canada at a fixed percent (say 10/45/45 Canada/Pacific/Europe, or 16/42/42, or 15/40/45, or 20/50/30, etc) doesn't change much; the "split index" still beats the MSCI EAFE+Canada; again, this is because the index isn't stuck overweighting one country at the arguable height of that nation being in the biggest equity bubble in modern history (I say "modern" because the South Sea Co bubble and Companie Mississippi bubble might've pushed up British and French markets to higher valuations than than Japan in 1988 or 1989; I'll leave that for economic historians to debate)
I would also like to point out that alternate-weighted indices (whether GDP, equal country weight, equal weight, or capping country weighting at a certain fixed percent) aren't a new idea; State Street actually came up with the first "Japan Light" index in mid-1988 when they wanted a more reasonable index to compare their foreign equity portfolios to (they were underweight Japan because they guessed--quite correctly in hindsight--that Japan was in a massive stock bubble....they were just a little early in their guessing) so they simply weighted Japan at half its market cap weight; MSCI soon followed with its own "Japan Lite" index which did virtually the same thing and then for a few years after that even provided what they called "EAFE Lite" indices that--rather than just capping Japan alone (although the practical effect was similar to capping Japan alone since no other country in the EAFE had its cap weight at market cap at such a greater value than its actual GDP-weighted economic weight)--had a hard cap on any country of the lesser of either its market cap or the given percentage; IIRC they discontinued publishing these in 1998 or early 1999. MSCI has also produced GDP weighted indices since at least the late 1990s (and backtested them to 1970 for the EAFE and the early 1990s for the MSCI EM). Equal weighting of course has an even longer history; the first S&P 500 index (the one that Samsonite used for its pension fund circa 1971 or 1972) was equal weighted....but they had to give it up after a few years since equal weighting required much more rebalancing than cap weighting and this was rather expensive given the stock market commissions in effect at the time.....which obviously would be a virtual non-issue today.
Come to think of it, a fixed cap weight for a given entity in an index has at least some history as well; in the late 90s and early 2000s pretty much every Canadian mutual fund--whether index tracking or active-- at the regulatory behest of the Canadian equivalent of the SEC was made to cap exposure to any one stock at 10 percent despite Nortel making up some 34% of the TSX 300 (and almost 45 percent of the TSX 60!); this meant that they missed some of the runup in Nortel during the dotcom boom but also didn't get hurt nearly as badly when Nortel got cut to ribbons during the mid 2000 to late 2002 dotcom crash and eventually traded at a few cents a share and IIRC either went bankrupt or got bought out at a pathetic price. Another example would be the NASDAQ index committees in late 2014 or in 2015 capped exposure to any one stock at a certain amount; this ostensibly applied to any an all companies in the NASDAQ-100 but the practical effect was simply to reduce Apple's weighting so the NASDAQ-100 would consist of a fairly good variety and weighting of stocks rather than just "a bunch of Apple and not quite as much of everything else". Finally, one place where such a maximum fixed weight for any one company (or any one sector), or even an equal weight index, that as far as I know was not tried (but that would've saved investors' bacon big-time if it was) was Iceland in the late 2000s; as of January 2008 just under 85 percent of its market-cap weighted index was in a few big financial stocks and the end result was the Icelandic stock index losing more in one day in late 2008 than US stocks (or for that matter a total world stock index) lost during three whole years in the Great Depression!
Anyhow, I guess that's about it. The link contains a file (in Excel format) for the alternate EAFE indices (equal weight, equal country weight, equal country weight plus Japan 20%, equal country weight plus Japan 25%, GDP-weighted, constant PPP real GDP weighted, RAFI Developed ex-US, and the various 20/25/30 percent country capped EAFE indices; it also has some graphs and charts showing how extreme Japan's weighting in the EAFE was circa 1988 and 1989.
Link:
https://easyupload.io/m/vd9na9