The great financial educator William Bernstein said: “You have to understand what market history looks like. What market history tells you is that the very, very best investments are made when things look the worst.”
It’s for similar reasons that I write so often about the past. I want to try to understand what fleeting or lasting horrors my investment choices might inflict even before any rewards come due.
This means examining as fully as possible the asset classes that comprise today’s investing mainstays.
First-world problems
Most Monevator readers’ portfolios are dominated by World equities – that is, developed world stocks.
But there’s a problem if you want to know how the World index has performed over the long-term.
Which is that the two benchmarks that stretch back farthest are pay-walled.
Fair enough, I suppose. Professors’ Dimson, Marsh and Staunton’s DMS database and Global Financial Data’s indices are both based on exhuming stock returns from fusty old journals and ancient newspaper archives. Someone’s got to keep the wonks fed and watered.
But that doesn’t help the investor in the street. People like us who are keen to avoid becoming investors out on the street, by educating ourselves in the ways of the investing world.
True, you could simply use the MSCI World’s easily-accessed tale of the tape. Its data runs from 1970.
But in my view that paints too benign a picture.
No Great Depression, no World Wars, no decade of deflation, no deglobalisation.
While 50-odd years sounds like a long time, we can only really see how equities responded to a wide set of conditions by retrieving the greater part of the 20th Century.
Introducing a new world index
We need more open-source data. And I’ve found it!
Enough to create a World index reaching back to 1919:
- I’ve taken historical country-level stock market returns from the Macrohistory database.
- I then weighted each country using stock market capitalisation data from the paper, The Big Bang: Stock Market Capitalization in the Long Run.
- Then I currency-converted all the results to GBP1 using exchange rate data from the Macrohistory team.
This process enabled me to assemble a World index in GBP that begins in the aftermath of World War One. At the other end of the timeline, the new index segues into the MSCI World GBP from 1970.
The resulting World equity index is not perfect (and I’ll explain why further down) but I believe it’s good enough.
So I’ll use this index to represent the World equities portfolio in future Monevator long-term performance articles.
In the meantime, the rest of this article will chart how world equities have fared from 1919 to 2023.
Then I’ll briefly pop the bonnet on the index as a treat for the hardcore at the fag end – I mean the grand finale – of this piece.
Investing returns sidebar – All returns quoted in this piece are real annualised total returns. That is, they’re the average annual return (accounting for gains and losses) realised in a given time period. These returns include the impact of reinvested dividends, but strip out the vanity growth delivered by inflation that does nothing to boost your actual spending power. Local currency returns have been converted to GBP.
World index: long-term equities growth
Here’s the World equities growth chart using our new index versus two rival long-term benchmarks: US and UK equities:
Data from JST Macrohistory2, The Big Bang3, MSCI, Aswath Damodaran, and FTSE Russell.
August 2024
The graph reminds us again that the rest of the advanced world has struggled to keep pace with US equities since the mid-1990s, aside from a brief panic room huddle during the Global Financial Crisis.
We can also see that home bias cost UK investors dearly throughout – even though the UK has remained one of the world’s top-performing markets over time.
World index annualised returns in GBP (% per annum)
Let’s now look at the long-term average real return numbers with dividends:
| 2023 | 10 years | 20 years | 50 years | 105 years | |
| World equities | 8.9 | 8.4 | 6.7 | 5.5 | 6.7 |
| US equities | 16.5 | 11.6 | 8.3 | 7.5 | 7.7 |
| UK equities | 0.6 | 2.3 | 4 | 6.2 | 5.6 |
The US wipes the floor with the rest of the world across every timeframe. Particularly in the last ten years as the ascendency of Big Tech – and its concentration in US stock markets – has left competing sectors looking like yesterday’s news.
It would be interesting to see whether the US still dominates in an alternative world with the Big Tech winners stripped out. We’ll save that for another time.
World index: annual returns
Annual World index results resemble any other crazy equity returns chart. They look like an abstract cityscape of soaring skyscrapers and deep shafts boring into negative space.
Happily however the towering returns outnumber the dark days lost in bunkers.
Thus somehow our long-term financial wellbeing emerges from this profile of sky-dwellers and underlanders.
Annual returns: World vs US vs UK stock market indices
A question: does diversifying across the world take the edge off those trips to the bargain basement?
This chart indicates that the World index might provide some downside protection relative to single country markets.
The cyan bars seem to punch shallower holes than the USA’s red. Though also notice how dynamically America tends to bounce back.
Drawdowns: World vs US vs UK stock market indices
This is the trauma room chart: a raw record of loss and terrible stock market slashes. All the same, you can see how the Great Depression is mitigated by the World index versus the US during the 1930s. (The impact of the Great Depression was not so severe in the UK, for one thing.)
World War 2 and subsequent recessions were also typically blunted by a World stock assemblage.
A notable exception is the early 1990s slump when the Japanese stock market bubble burst. The Tokyo stock exchange comprised over 40% of the index in 1989 but it made up only 11% ten years later.
Holding the World portfolio also exacerbated the Dotcom Bust of the early 2000s, as Japan continued to sell off and the UK piled on the pain too.
The risk-adjusted view
All told, our eyes do not deceive us. The numbers show that the World index has inflicted less volatility on investors over the long-run (1919-2023):
| Index– | Annualised return– | Volatility– | Sharpe ratio |
| World | 6.7% | 17.4% | 0.38 |
| US | 7.7% | 19.7% | 0.39 |
| UK | 5.6% | 20.5% | 0.27 |
The higher your Sharpe ratio, the better your risk-adjusted returns. That is, the more return you get per unit of risk as measured by volatility.
From this we can conclude that the World has proved about as worthwhile a buy as the US when returns are costed against the volatility you endured to attain them. (This is the essence of the Sharpe Ratio measure.)
Viewing the benchmarks on the single dimension of returns would imply that world equity diversification has proved sub-optimal, compared to if you’d gone all-in on the US.
But taking that broader view reveals how the rest of the world offers good reason not to pin all our hopes on perpetual American exceptionalism.
World index market share
The MSCI World is utterly dominated by the US stock market these days. It currently weighs in at a 71.7% share of the index:
Source: MSCI. August 2024
Our investing fate is inevitably reliant on the world’s most important capital market, though that’s nothing new.
This next chart compares the market capitalisation of each of the major developed world stock markets:
Source: The Big Bang. August 2024.
We can see that the US has almost always been the biggest player – offset to a greater or lesser degree by the UK, Japan, France, Germany, and the plethora of smaller fish known as ‘Other’.
Since 1919, the US share of the world market has ranged from 31% (1988) to 73% (1951).
For what it’s worth, the US is close to its historical ceiling right now.
Inside the World index
I want to emphasise that the World index presented here is not the global index.
I’m relying on MSCI World figures from 1970 onwards. That index excludes the emerging markets. Its Asian representatives are limited to Japan, Singapore, and Hong Kong.
Pre-1970, I use Macrohistory’s country list. That is limited to the Anglosphere, Japan, and Europe.
Macrohistory’s research omits Austria, New Zealand, Ireland, and Eastern Europe.
Indeed, it’s the absence of Austria and Russia that enforced our 1919 cut-off. Those two imperial stock markets weighed about 5% each before World War One intervened (by the light of the DMS database).
South Africa is the other notable no-show. Its stock market accounted for a couple of percentage points of the whole during most of the period.
Every benchmark makes some exclusions for reasons of practicability. Ours are imposed by the limits of publicly available data.
Even so, we’re happy that our numbers are a credible representation of the historical World index. The loss of fidelity versus commercial alternatives doesn’t change the lessons we can learn.
Finally, I’d just like to thank the academics responsible for the Macrohistory database and The Big Bang research. They have created an immense resource and been incredibly generous in freely sharing it with the world.
Thank you Òscar Jordà, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, Alan M. Taylor, and Kaspar Zimmermann.
Take it steady,
The Accumulator
- British Pounds Sterling. [↩]
- Òscar Jordà, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, and Alan M. Taylor. 2019. “The Rate of Return on Everything, 1870–2015.” Quarterly Journal of Economics, 134(3), 1225-1298. [↩]
- Dmitry Kuvshinov and Kaspar Zimmermann. 2021. The
Big Bang: Stock Market Capitalization in the Long Run. Journal of Financial Economics,
Forthcoming. [↩]
Good job gentlemen.
@TA
Some excellent research all round. Thank you. For your efforts -Lee.
Excellent job dredging up this older data and linking it to current series. Many thanks.
I guess it tells the story all us passivistas already cling to – we know nothing about the future, so global equities makes as much sense as anything else in the equity world and probably minimises the risk of anything horrible happening.
Thankyou. I love a long term chart. It’s great for getting a bit of perspective. Excellent.
@TA – can I add my thanks to those above?
A really great piece of work.
> MSCI World figures from 1970 onwards. That index excludes the emerging markets.
I was going to grizzle that I am sure there is some EM in VWRL which is Lars’ goto world tracker. Nearly 10%. Looks like that is the difference between MSCI World and the FTSE All-World Index. A different interpretation of ‘World’.
Worse still I swap ‘twixt VWRL and HMWO to use the CGT allowance, presuming equivalence. HMWO uses the MSCI world index which is inferior IMO (if you are going to accept Lars’ premise) although I am not going to press that whinge too far at the moment as the concentration in HMWO is to my favour.
Nice piece of financial archaeology. I guess historically the EM sector being missing didn’t matter so much. Running this forward, while being 90% accurate is probably OK the divergence could add up over the next few decades
Fantastic work @TA. Very impressive. It’s good to see that the long-term returns hold up in local currency. The UK is a peculiar case as the world has changed from GBP to USD being the world’s reserve currency over the period.
There will be some survivorship bias in the data from excluding the likes of Russia and Eastern Europe, total wipe-outs after the communist revolution.
Some risks are political and country-specific, therefore I want more diversification than the cap-weighted world indices offer. It can only happen in a banana republic that a convicted felon, who has previously staged an insurrection, can become the leading contender for highest office on the main conservative party’s ticket, right?
The MSCI-ACWI is a start, but it’s still ~65% US.
@TA:
Nice work and a very useful post.
I’ll bet you can guess what I am about to ask, specifically: have you used RPI to represent UK inflation?
@sparschwein I’ve always based my investments on Lars Kroije’s view.
If you are overweight or underweight one country compared to its fraction of the world equity markets, then you are effectively saying that a dollar invested in the underweight country is less clever/informed than a dollar invested in the country that you allocate more to.
You would therefore be claiming to see an advantage from allocating differently from how the multi-trillion dollar international financial markets have allocated.
But you are not in a position to do that unless you have edge.And we agreed we don’t have edge…
However, looking at the worlds political positions, and the number of passive investors I question how efficient marks are at the moment, and if some tweaking is, indeed, required.
I once came across a paper by an American academic who claimed that looking at US equities since 1919, or since 1900, was too short a time frame. He dug out earlier data. His conclusion was that equities looked like less of a sure thing than many people argued.
Alas I can’t find the bookmark. But if I do in future I’ll bung it in a comment on your esteemed blog.
Mind you, you can come to much the same conclusion just by ignoring data from the US, the UK, Australia and South Africa. Elsewhere equities have had less than magical properties. So: do you think that the future of the USA will be more like the past of the USA or the past of the Austro-Hungarian Empire, Russia, Germany, or Japan?
@NOP86 thanks for bringing this up. I think this is worth unpicking.
In fact I agree with Lars’ argument (and I believe that markets are mostly efficient) BUT I disagree with the conclusion.
Lars’ argument only considers expected (average) returns. If return is all that matters, then sure, go all-in on stocks and use market weights.
But in reality, we don’t get to live the average of all possible futures. We only get to live one path. So I care about risk, and I care especially about such risks that would send us working at Tesco in retirement. Even if the probability is low. So I am happy to forego some expected return to avoid the all-eggs-in-US-basket situation.
A typical 60/40 of MSCI World and Bloomberg Global Aggregate Bond index is ~58% invested in US stocks and bonds.
@dearieme – Scott Cederburg maybe? He has published a series of papers on long-term historical stock market returns across many countries. There are a couple of good interviews with him on Rational Reminder.
One of the very interesting conclusions was that “risk” changes a lot depending on the time horizon. In the very long run (decades) international diversification within the stock allocation becomes *more* important. And stocks-bonds diversification becomes *less* important.
Betting against America has been a losers game for many years now-certainly all my investing life
Looks to stay that way for a while yet
The power of that country in so many fields including finance is remarkable
xxd09
Thinking about this a bit more, there are some assumptions in my previous post (#11) that are maybe worth spelling out.
1. Markets are crap at anticipating major disruptive events. Wars, pandemics, even disruptions in the financial system (see GFC). I’ll add civil wars to the list.
2. Even if markets can anticipate such events, the risk is very hard to price, because probability, timing and impact are all unknown.
3. Even if markets could price such risks accurately for the average investor, our personal exposure may be very different.
Where this leaves us, I suppose, is that markets are mostly efficient in the sense that it is very difficult to make a consistent profit of their inefficiencies. So we are better off with index funds than active investing in the same asset class, in most cases. Unfortunately we still need to think for ourselves and manage our own risk.
Incredible work pulling this together @TA. Thank you.
Can’t help but wonder if 1919 gives a flattering start date to the returns?
Although it does include a 1919-20 chunky drawdown of (eyeballing it) around 10% (US), 20% (World) and 30% (UK); it does also miss out WW1 and the disappearance of the Russian stock market from 1917 to 1992 (MICEX) and 1995 (RTS). Maybe that would shave off a few tenths of a percent from the World returns from 1919 to 2023. OTOH, IIRC DMS have Australia and South Africa as top performers since 1900, so maybe including them would even things out again.
Interesting how things go wrong for the UK on a relative basis from the 30s and 70s.
Great article.
TA, out of interest what are the UK total real returns from the market peak in the late 90s to now the present day? It’s a pretty flat period on your graph for the UK! As are those periods between 1935 to 1955 and again quickly after between 1960 to 1980.
@Sparschwein: “Scott Cederburg maybe?” Not the chap I meant but a useful tip. Thank you.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3594660
@deariemie: Prof. Edward F McQuarrie’s 2023 work perhaps?
Abstract here:
https://doi.org/10.1080/0015198X.2023.2268556
Joachim Klement summarises McQuarrie in a three part series here:
https://open.substack.com/pub/klementoninvesting/p/stocks-and-bonds-for-the-long-run
and here:
https://open.substack.com/pub/klementoninvesting/p/stocks-and-bonds-for-the-long-run-1c9
and here:
https://open.substack.com/pub/klementoninvesting/p/stocks-and-bonds-for-the-long-run-394
Great work to put this together!
I think it would be strengthened by some cross-checking against the Dimson et al. work to understand the potential impact of not sampling all countries and time periods.
I think Dimson et al. find a substantially smaller long-term real global equity return (5.2%/year from my notes, with a forward projection of ~4%/year). As others have said, this may be due to your index omitting WW1 and the preceding years and some major markets that collapsed. So this index potentially gives too optimistic a view of the historical long-term return, which seems worth noting.
@ Delta Hedge and PeteR – the real annualised return for this index back to 1900 is 6.1%. Thinking about your comments, it’s probably better to include WW1 than worry excessively about the absence of Austria and Russia. I’ll do that in the future and update this article when TI’s punishing content schedule allows 😉
I still think using UK historical returns is a reasonable proxy for a more cautious assessment of the possibilities without veering off into Imperial Russia type catastrophising: https://monevator.com/uk-historical-asset-class-returns/
The last return I can find for DMS is in their Global Investment Returns Yearbook 2023. 5% annualised USD for the World and 5.1% for Dev Markets. 3.8% for Emerging Markets.
One question I have for you all (relating to dearieme’s and Sparschwein’s points about long-term risk) is how great an idea investing in the emerging markets seems now?
Given the evidence for EM’s lagging long-term returns and their relative lack of shareholder protections up to and including large-scale state interference and theft.
Are we going through the motions on emerging markets by owning a global tracker featuring 8% EM, more like 5% in a 60:40 portfolio?
@ Lenahan – UK real total returns peaked in Dec 2021. Less reputable outlets will try and drum up a story about lost decades by touting nominal price return graphs of the FTSE 100 and whatnot. We would never do that 🙂
That said, the real annualised return was 2.7% over the last 25 years, so not great for home bias fans.
@ Al Cam – I use Macrohistory’s UK inflation rate. They cite:
1870 – 2016 from Hills, S, Thomas, R and Dimsdale, N. “A millenium of macroeconomic data – version 3.1”, Bank of England. Series: Headline consumer price index (cpi). Level.
2017 – 2020 from Office for National Statistics (ONS), series “CPI annual rate 00: all items 2015=100”
After 2020, I carry on using CPI.
Although Macrohistory label their inflation data as CPI, it’s RPI during the time periods you’d expect.
@TA and your question for us all on whether to bother with EMs: On 8th June 2023 Golman Sachs updated their “Path to 2075” report with this to say on EMs:
“We expect EM equities to outperform DM in the longer run, due to stronger earnings growth and, as risk premia fall, multiple expansion. However, the most important dynamic underlying EM capital market growth in our projections is the equitisation of corporate assets, the deepening of capital markets, and the disintermediation that takes place as financial development proceeds (processes that do not, by themselves, imply EM equity outperformance).
Our projections imply that EMs’ share of global equity market capitalisation will rise from around 27% currently to 35% in 2030, 47% in 2050, and 55% in 2075. We expect India to record the largest increase in global market cap share – from a little under 3% in 2022 to 8% in 2050, and 12% in 2075 – reflecting a favourable demographic outlook and rapid GDP per capita growth. We project that China’s share will rise from 10% to 15% by 2050 but, reflecting a demographic-led slowdown in potential growth, that it will then decline to around 13% by 2075. The increasing importance of equity markets outside the US implies that its share is projected to fall from 42% in 2022 to 27% in 2050, and 22% in 2075.”
If you don’t include include EMs (and also Frontier Markets and Small, Micro and Nano Caps) then IMO you are diverging from pure global cap weight.
Maybe that’s a good idea and maybe it isn’t but it’s not pure passive, even if Lars might disagree 😉
EM performance is inversely linked to $ strength as money flows to the US to buy American equities and US Treasuries in good times and bad respectively.
Arguably this can lead to systemically undervalued EM assets.
And IIRC Callum Thomas from Topdown Charts had a chart recently showing that EM Small Cap Value was something like 5 times cheaper relative to US Large Cap Growth compared to 2007/8 (might be remembering it wrong though).
Seems too good not to have some exposure.
@TA (#20):
Thanks for the reference. I have located the original spreadsheet on the BoE website, and it seems to be the [annual] data behind the long term BoE inflation calculator. This BoE calculator incidentally goes all the way back to 1209!
AIUI, the inflation data from 1950 is derived from original RPI data, but has been re-cast [and in some areas enhanced] by the ONS to represent CPI inflation. Fuller details are available at the ONS link given in the BoE calculator. To quote the BoE calculator “CPI estimates before 1988 are modelled based on data collected for the Retail Price Index (RPI)” and please note [from Wikipedia] “RPI was first introduced in 1956 replacing the previous Interim Index of Retail Prices that had been in use since June 1947.”
The BoE spreadsheet also ‘explains’ the pedigree of the pre 1947 data. Albeit, that it took me a some time to work that out – hint I ended up looking at the contents of CPI index cells in column D (& column F from 1914) in tab A.47.
AFAICT a total of seven different sources are used to populate the BoE spreadsheet, and the source data for these reside in columns BG, BI, BJ, BK, BW, CA & CB in the same tab.
In a nutshell, pure RPI does not seem to be have been used. I could, of course, be wrong – as it is not the easiest spreadsheet to follow – but that is how I see it today.
@Delta Hedge: Apart from dabbling with a bit of VHYL to cover running costs in my Vanguard account, I have for a while chosen to stick with Developed World ETFs on cost and transparency grounds. I may be overly cautious on the accountability front, but the cost saving is undeniable. The usefully enhanced performance to date is a nice bonus and it will take a major leap in Whole World performance to wipe out the accrued advantage.
@Sparschwein. All asset classes tend to have survivorship bias in their return numbers. It’s somewhat ridiculous that the passive fanatics complain about survivorship bias in active funds but never mention that the index numbers have the same issue.
The answer is diversification. Don’t back one asset class, one country etc. Those US who are 100% in domestic stocks (and sometime just the S&P500) are making one huge bet.
re EM – I think it’s difficult to allocate based on historical returns. At least, we also need to consider valuations. Historically, high valuations correlate with low expected returns and vice versa. People seem comfortable with US stocks at CAPE 35, and everyone hates China at CAPE 11 (from https://indices.cib.barclays/IM/21/en/indices/static/historic-cape.app ).
I’ve always included EM for diversification, and I’ve recently overweight China because I see more opportunity than risk. Which could turn out very wrong, of course.
Re #22:
Further reading suggests to me that only CPI data from 2005 is classed as a National Statistic and anything earlier in the series back to 1948 is classified an official statistic. Also, and according to the ONS “there are limitations with constructing a historical series and, as such, users should treat these estimates with some caution.” So whilst, the extended CPI data is probably better than pure RPI (in terms of accuracy, etc) it may well inherit some of the RPI flaws too.
@Sparschwein #25: China CAPE 11x v US now on 35x:
Difficult one to try and call because China’s a one party state moving away from a free market and, like the French from the 1950s to the 1980s, in effect pursuing development without shareholder gain.
Plus there’s the Chinese housing market disaster and demographic time bomb issue, as well as the ever present risk of sanctions and trade wars in an economy very dependent on exports. Geopolitical risk must be very high.
OTOH the US may have the Magnificent 7 etc, unlike China, but it too has a looming demographic and an already all too visible debt crisis brewing.
So maybe it’s a bit of an apples to apples as well as an apples to oranges situation.
@Delta Hedge.
That must be the work, I think. Thank you.
And thanks to TA for igniting such a useful comments thread.
@TA If including WW1 brings the long-term return to 6.1%, wiping out ~10% to represent losses in Austria and Russia would only bring the return to about 6.0%. So it seems to give a figure ~20% higher relatively than Dimson et al., which is perhaps a reasonable estimate of the error compared to the truth.
Whether that’s “good enough” depends on what you want to use it for. I probably wouldn’t use it to judge historical stock returns from 1900, because we have the Dimson et al. results for that. It feels unclear to me how large an error to expect in aspects besides the long-term real return – drawdowns, returns over sub-periods etc. Perhaps it’s fair to think there’s a ~20% error on these things too, but there’s more uncertainty. Would it be expected to be closer to the truth in more recent years? I’m not sure of a good way to validate it on shorter periods. Anyway, it seems like a fun dataset, but it just looks to me like we shouldn’t take results too literally yet and slap at least a 20% error bar on them without further analysis.
Also, re “using UK historical returns is a reasonable proxy for a more cautious assessment of the possibilities”, the long-term UK return is just above the average for developed markets in the Dimson et al. data. So if by “more cautious” you meant something below the average, it would seem to need a worse case.
@TA (#20)
Interestingly (for those of us interested in such things!), there are at least 4 curated time series of inflation for the UK prior to 1948 (some of which overlap).
1) Long term retail prices index at ONS
2) Measuring worth
3) Barclays equity gilt study
4) Hills et al (2016), as used in the macrohistory.net database (and by you in this post)
The values can be significantly different (e.g., during the WWII period the annual inflation can differ by more than 3 percentage points). Consequently, the ‘safe’ withdrawal rate, understandably, also depends on the inflation series used. For example*, with a 60/40 portfolio (UK stocks/bonds) there is a small, just over 15 bp, difference in the calculated SAFEMAX. However, for individual retirement start years, the SWR varied by up to 1.5 percentage points.
* My own calculations using returns from macrohistory.net and the various inflation series
@ dearieme, Delta Hedge, Sparschwein – a long time ago William Bernstein wrote a prescient column about the lack of shareholder protection in emerging markets and China in particular. IIRC, he questioned whether investors would be rewarded in these markets as other actors siphoned away the lion’s share of the profits. Given the extraordinary rise of China in GDP terms but relatively disappointing stock returns, I wonder whether Bernstein’s forecast explains much of the gap. Ultimately there’s no point diversifying into an asset class when the dice are loaded. I really haven’t investigated the issue enough to come to a firm conclusion but this is where I’ll look if I get around to it. I ignored Bernstein’s column and overweighted into emerging markets at the start. These days I’d be happy defaulting to the All-World weighting. +1 Delta Hedge’s comment.
@ Al Cam and Alan S – Very illuminating! Thank you Al Cam for digging deeper into Macrohistory’s source and sharing what you’ve found. Alan S – that really is quite eye-opening re: UK safemax. Personally it made sense to me to consistently use Macrohistory figures unless there’s a good reason not to. Do either of you have a sense of what the default inflation figure should be, or is that a moot question? In other words, any historical inflation benchmark is always going to be a compromise? Much like any historical stock index…
@TA. I would say the biggest driver of EM outperformance/underperformance is economic globalization. EM outperforms during those periods where we get rapid globalization. It underperforms as globalization stagnates or declines.
I’d take a look at the KOF globalisation indices. https://kof.ethz.ch/en/forecasts-and-indicators/indicators/kof-globalisation-index.html.
The “End of History period” from 1990 through to 2010 was a period of incredibly strong globalization. EM loved that environment. As that stagnated in the 2010s, so has EM equity performance. Protectionism is bad for everyone in the long term, but initially it’s lower income countries that suffer. Those of us in higher income countries will get our just deserts eventually for being so short-sighted but right now onshoring and friendshoring don’t favour EM.
That must have been quite some effort, well done. Personally I would treat any equity index not created in real time with some caution and scepticism. Even with the best of intentions it is all too easy to miss information that would lead to survivorship bias. I do appreciate you have been careful to avoid this as best you can though and as you say is probably the best you can do with publically available information.
I had not realised that MSCI now make their entire back history available for free. Strictly speaking, their data was only created in real-time from 1986, the previous history created from historic information, so again some level of caution is justified. It is good that MSCI release this data for free though and it’s a shame FTSE Russell don’t do the same.
@TA(#31):
No worries, it was quite interesting to try and follow it all through. And wrt my comment at #26, I am now fairly certain that the extended CPI (and CPIH) series (extended back to 1948) do have some of the RPI flaws*. As I understand it, the only way to fully eradicate the “formula effect” from these periods is to use the original raw price data, which AFAICT is not how the ONS extended these series back in time.
My preferred UK inflation measure is CPIH reported monthly.
WRT [very] long term inflation series (ie starts before 1948) today I would possibly favour using the BoE data, albeit is only annual**. I am not sure if this is exactly what Macrohistory use as I notice they call up a reference from 2016, but the extended CPI data in the BoE tool (back to 1948) was actually last updated in 2022 by the ONS. During that 2022 update the ONS also extended the CPIH back to 1948 too. IIRC, the 2022 update did lower the historic CPI (vs its previous incarnation) by about 0.2pp on average – which might be important.
@Alan-S (#30):
The ONS paper (O’Donoghue et al 2004) “Consumer Price Inflation since 1750” [your Option 1] gives their reasoning as to how they selected between the various (sometimes overlapping) options available to them at that time.
OOI:
a) do you know if the 15 bp impact on the SAFEMAX is more or less than the effect of including estimated costs, etc?
b) how far back do your UK SWR calcs go?
* although I am not too clear of the likely impact, as all the quantifications of the formula effect I have found relate to periods that start much later than 1948; although apparently what is now known as the “formula effect” has been a worry since, at least, the 1970s!
**I have often used an alternative source for monthly “RPI” that dates back to 1915; AFAICT, HMG did start collecting data and producing some sort of price index in 1914
P.S.
The “previous incarnation” I refer to (in #34) above dates from 2018 (ie also later than 2016) and refers to revisions made to the “Consumer Prices Index (CPI) historical modelled estimates between 1989 and 1996”.
The point being that there seems to be quite some churn in what are easily assumed to be historic records. That is, what is a good “historical modelled estimate” today may not be so good tomorrow.
On the other hand, AIUI contemporaneous CPI/CPIH are as a matter of policy (perhaps as a consequence of being designated National Statistics) never revised, so this level of churn with their historic analogues may well be doubly surprising!
TA (#31) and Al Cam (#34)
I think all inflation figures have a certain level of ‘it depends’ about them even modern indices. For example, the basket of goods/services/housing etc. is critical and variable (which is why the ONS periodically review the items included). One reference (IIRC it is the ONS one mentioned by Al Cam), mentions that one early inflation measure excluded beer prices (presumably because they didn’t want to encourage the working classes to drink). I guess, ultimately, we have to settle for what is available, but should bear in mind the sources and magnitudes of errors.
As for the impact of fees on SAFEMAX, Kitces calculated that (very roughly) SAFEMAX is reduced by about half the fee level (so a 40 bp fee level would result in a drop of 20 bp in SAFEMAX), so the error bar introduced by inflation measures is similar in magnitude.
My SWR calculations go back to 1870 using the macrohistory database for stocks and bond returns, although I note that the maturity of those in the macrohistory database, which are, as far as I can tell, largely drawn from the Barclay’s equity gilt study is least 15 years. I have calcukated (and made available a database of gilt returns at different maturities and found the difference in the SAFEMAX between the ‘best’ and ‘worst’ gilt sector is about 15 to 40 bps (depending on the stock allocation).
All in all, I am beginning to think that SAFEMAX values should probably be quoted to the nearest 0.5 percentage point (I think it was Bernstein who suggested something along these lines), particular when future values are completely unknown and unpredictable.
@Alan S (#36):
Yes, it is that 2004 ONS paper that mentions the apparent lack of beer for the working man in days of yore! IMO, the paper is well worth digging out (it has been archived by the ONS but can still be located via Google) as it is rather short (with: 4 [A4] pages of text/diagrams; one third of a page for refs; and a further 4 pages of data tables) and is packed with good stuff, such as: “It should be noted that in general the relevance and quality of the primary sources diminishes the further one goes back in time. This means that comparisons further back in time and over long periods should be regarded as more approximate than comparisons over short periods in more recent years.”
FWIW, I happen to agree that “the basket of goods/services/housing etc. is critical”. The practical limitation with the approach is that it explicitly assumes that all folks have the same basket of goods & services, but just use differing amounts (including zero) of each component part. Clearly this is not the case, but the alternative of some 26 million separate UK household baskets is entirely impractical. So, if anything, the selection of the weights is even more important.
IMO, it is useful to bear in mind* that inflation is fundamentally a latent variable. That is, you cannot buy an ‘inflation meter’ and plug it in somewhere and take a reading. Inflation is a useful economics concept whose utility (pun intended) is generally well understood, but it is no more than that!
*especially so for physical/natural scientists/engineers
I took a closer look at the “historical modelled estimates” for both CPI and CPIH published at the ONS website earlier today. And concluded the following:
The published indices run from Jan 1949 to December 1988 at monthly intervals; and hence the CPI/H inflation rates run from Jan 1950 to December 1988 at monthly intervals. Both indices assert 1965 = 100.
If so desired, these data could be manually spliced together with the regular CPI/H indices (with 2015=100) to form a contiguous series of monthly CPI/H values all the way back t0 1949/50*.
If an even longer sequence is required then data from the 2004 ONS Paper (O’Donoghue et al 2004) “Consumer Price Inflation since 1750” could be spliced in to suit*. Note that the data in the 2004 paper is annual only, but a stylised monthly break-out could be added if so required.
*albeit with the many issues and qualifiers highlighted above and the general guidance that “in general the relevance and quality of the primary sources diminishes the further one goes back in time. This means that comparisons further back in time and over long periods should be regarded as more approximate than comparisons over short periods in more recent years”.
Perhaps, this might be an acceptable approach to the 100 year plus challenge that @ZX threw out there some months back, see: https://monevator.com/weekend-reading-five-graphs-that-justify-the-gloom/#comment-1748771
@ZX – cheers for the link! Sometime ago you mentioned that emerging market bonds had outperformed emerging market equities. Do you think that was a historical one-off or is there any reason to believe bondholders could be better placed than shareholders to profit from EM?
@ Al Cam – thank you for that assessment and the time you’ve put it into it. The one fly in the ointment appears to be that the O’Donoghue paper is pro-RPI. They even seem to prefer RPI as a measure up to the present day (or 2003 as per their publication date). See section ‘1947 to current day’. It’s a great paper but doesn’t appear to provide a solution pre-1949 that’s any more reliable than RPI as far as I can tell.
@TA:
The O’Donoghue (O’D) paper, like every other paper, is a product of its time.
Some key things to bear in mind relevant to when the O’D paper was penned:
a) The Consumer Prices Index, or CPI, was first officially published in 1997 to measure inflation consistently across all European Union states*;
b) the United Kingdom has tracked CPI since 1996, but IIRC only UK CPI data from 2005 onwards is classed as a National Statistic;
c) RPI was classed as a National Statistic until 2013;
d) in December 2003, the BoE inflation target was changed to CPI of 2% from the previous target of RPIX of 2.5% (RPIX is RPI excluding mortgage interest);
e) CPIH was introduced in 2013
So, simply put, before the late 90’s there was nothing [in the UK] better than RPI. And whilst the CPI is mathematically more correct than RPI its scope is less than the RPI. The CPIH addresses the scope shortfalls of the CPI which are possibly perceived as more important in the UK than elsewhere in Europe. I would say the UK is fortunate to have back-casted CPI and CPIH data to 1948.
Hence, I suspect if the O’D paper were to be re-written today the Section called ‘1947 to current day’ would probably change the most.
Hope this helps.
* The primary goal of the European Central Bank (ECB) is to maintain price stability, defined as keeping the year on year increase HICP (CPI in the UK) target on 2% over the medium term. The HICP is also used to assess the convergence criteria on inflation that countries must fulfil in order to adopt the euro.
@Delta Hedge, @TA – I’d agree with all the reasons to dislike Chinese stocks. But beware of groupthink. When stocks are cheap, there are always good, real reasons why they could become even cheaper. Asset classes are easily dismissed when they had a recent bad run. On the other hand, there is little attention to the high valuations, low equity risk premium and significant concentration risk with US stocks.
@Sparschwein: Agree with you in terms of what I actually do. Back in May, following @TI’s “What to do if you’re queasy about the US stock market” article I tilted ~10% away from the US overall (compared to its global weighting) with half (5% of portfolio) going towards more EM stocks, Euro stocks, Small Caps, Value stocks and Small Cap Value and, at the same time, the other half (5% of portfolio) went equally into EMSD (SPDR MSCI Emerging Markets Small Cap UCITS) ETF and the DGSE (WisdomTree Emerging Markets SmallCap Dividend UCITS) ETF. That makes me significantly overweight China for the reasons that you highlight and which I tend to agree with.
At the same time though EM generally and China particularly must be cheap for a reason.
Accordingly, I have every apprehension that if something does eventually go very badly wrong in the US-China relationship then I’ll be taking an outsized battering on the non-US side of my portfolio.
Reading history is a bit of a pass time for me and, personally, I try as a result of that not to underestimate the ideological dimension with China.
The CCP is now a rather strange hybrid: a somewhat nationalistic ‘Market Leninist’ entity which runs a mixed economy (socialism with Chinese characteristics) and which sees China as at once both a traditional Great Power and also as being ‘anti-revisionist’ (the Party is no fan of either Khrushchev or Gorbachev) and ‘anti-imperialist’.
The CCP certainly has not embraced capitalist values, even though, for purely pragmatic and tactical reasons, it has temporarily adopted and utilised capitalist forms and mechanisms.
As such, I can’t see it ever being very sympathetic towards the interests of foreign shareholders from the ‘West’.
Update: 105 year annualised return of World index is 6.7% not 6.8%. Growth of £1 now £886 1919-2023. Volatility 17.4% instead of 17.3%.
Disappointing stock returns from EM and China might also be down to poorly made and constantly changed indicies. And these changes are influenced primarly by politics. I’ve stopped investing in EM index funds after I realised that MSCI (and other index providers) started removing a ton of Chinese stocks, just because the US government doesn’t like them. Even the providers located in the EU, like Stoxx or Solactive are complying with this and no thought is given about the people actually invested with their products. This continues and they’re blacklisting new ones every couple of months. Many people want to invest in China, despite it being an autrocracy. And they do it for many different reasons. A market cap weighted index should surely be a great choice, given that EM markets in general are not very cheap to trade. But then suddenly you see a bunch big companies removed from your index for purely political reasons. I will give an example with companies like CNOOC (oil) or China Mobile, which are among the biggest and they have done well for their shareholders since they were removed from the western indicies. I wholly expect Chinese semiconductor companies to do very well in the future and to become major competitors to the likes of Intel, Nvidia, Tsmc. Will investors in EM or China equities really have no exposure to them, just because the American government doesn’t like it? Knowing the answer to this question, I decided to go ahead and invest in EM stocks myself. I am sorry if this has gotten too political, but I want people to give it a second thought.
Everything in the end is political unfortunately
Are you really surprised that investing in a totalitarian police state like China(Russia etc) is fraught with difficulties?
Corruption is widespread and there is no serious rule of law
Investors have a chance to make a fortune or get wiped out which is much more likely
Free markets ruled by laws against investing misdemeanours are why investors flock to the US ,U.K. etc
It’s all relative of course but I am afraid not all countries (most of them in fact)do not play by our rules
Investors should bear this basic fact in mind when investing outside stockmarkets of the US,U.K.
xxd09
Thank you for sharing your thoughts @SAM, and for setting out a well reasoned position, one which does not often get an airing.
I would agree with the points made by @xxd09 above about the issues and shortcomings of investing in China.
I would add a few caveats though, plus an additional specific downside
Firstly. like the scientific enterprise, the endeavour of investment should be apolitical.
Some investments are seen as ethically questionable (in SRI/ESG terms) but have performed brilliantly.
In 2015 the Dimson-Marsh-Staunton Global Investment Returns Database in the Credit Suisse Global Investment Returns Yearbook showed a $1 investment in 1900 in tobacco stocks grew in nominal terms with all income reinvested to $6.3 mn over the next 115 years which is about 14% p.a. and well above any other sector.
So, it’s not a moral/ethical/SRI/ESG framework which puts me off of China here.
Secondly, there are other perspectives on China.
Personally, and I suspect unusually for the readership of this site, which seems to lean right/libertarian, I’m a leftists at heart and, notwithstanding the disasters of the Great Leap (1958-61) and Cultural Revolution (1966-1969) and the recent issues with the Uyghurs etc, I can see (not agree with mind you, but nonetheless understand) the PoV of the PRC.
In the CCP’s eyes, in 1949 they inherited/won a country decimated:
– By cynical British imperial actions during the Opium Wars of 1839-42 and 1856-60.
– By Western interference leading to the concession system (which, in effect, reduced much of of late Imperial China to the status of a colony) and to the Boxer rebellion of 1899-1901.
– By massive internal instability, especially from the Taiping Rebellion of 1850-64, in which as many people lost their lives as from the whole of WW1; the Warlord Period following the collapse of the last dynasty, the Manchu, in 1911-12; and by the Chinese civil war of 1927-1949.
– By the Japanese invasions of Manchuria in 1931 and of China as a whole from 1937 to 1945.
In the CCP’s self image, they survived the Long March of 1934-36 and, after uniting the mainland in 1949-50, they then were forced to face off against the USA from 1950 to 1953, following its forces reaching the Yalu in the autumn of 1950, before pushing them back to more or less the 1948/9 boundary between the two Koreas by 1951.
From the perspective of the CCP, from 1949 to 1971 the USA prevented the Security Council of the UN from even recognising the PRC and the US did not grant the PRC full diplomatic recognition until 1979.
After 1978, with Deng’s reforms, in the eyes of the CCP it was the PRC Party-State “socialism with Chinese characters” model which rebuilt the country from one of the world’s poorest to the largest economy by Purchasing Power Parity and with a middle income PPP GDP per capita.
This is the PRC/CCP perspective. One can disagree with it vehemently, but it doesn’t change that Xi and CCP Central Committee completely see the world through a prism of what I’ve just described.
They think that the US is an inevitably declining imperial hegemon which is doomed, and that Japan and the US are desperately trying to encircle the PRC and lock it in within the First Island Chain.
Again, one can disagree with all of this, but that’s how the Central Committee of the CCP actually sees the world.
And Xi and his colleagues are definitely very much still Communists. Deng may have found it necessary for the PRC to take “the Capitalist Road” temporarily, but the CCP remains fundamentally an anti-revisionist Marxist-Leninist party. Have a look, for example, at Xi’s 2018 speech to a Party event to mark the bicentennial of Karl Marx’s birth. The CCP sees the events in Eastern Europe and the former Soviet Union from 1989 to 1991 as a disaster and the Party will exact any price to ensure that they are not repeated in the PRC.
Once again, one can disagree totally with the CCP/PRC perspective here, but this is how they actually see the world and the historical situation, and they see it that way with a complete conviction and a total confidence.
A leopard does not change its spots in this case.
If you’re looking to invest in the PRC as a Western investor, then this is what you are up against. If anything kicks off with Taiwan, then expect no help whatsoever from either the PRC government or your own in getting your money out. It’ll be gone totally and forever.
Thirdly, you might say, well this risk is all in the price. But is it really? How many business people from the West investing into China really understand what I’ve sketched out at my second point above? If it’s not fully and widely understood, then how likely is it to be integrated into price discovery?
Fourthly, there’s a much higher fraud risk in China than many realise. Look at how even very experienced and expert outside investors can get burnt, like Anthony Bolton did with Sino-Forest Corporation in 2011.
Fifthly, war over Taiwan is very on the cards. Xi wants reunification to occur between the centenary of the founding of the PLA in 2027 and the 20 anniversary of him becoming General Secretary of the CCP and Chairman of the Central Military Commission in 2032. If war breaks out expect that foreign investments held in Chinese entities will become worthless overnight.
Sixthly, most investments into Chinese companies have to be made via ADRs or into Cayman or such like vehicles. These vehicles claim to have a contractually enforceable right to dividends etc from the underlying Chinese entity which actually owns the IP and any other businesses assets. I wouldn’t like, as an outside investor, to have to put that proposition to the test.
Seventhly, whilst the market cap of Chinese equities has actually soared these past twenty years, the massive dilution of shareholders has meant miserable returns. Commercial success may not translate well or at all into shareholder success.