Over the last month, I’ve been playing around with an intriguing returns dataset developed by researchers Jorda, Knoll, Kuvshinov, Schularick, and Taylor (“JST” Study for short). It includes estimates of annual returns for stocks, government bonds, government bills (a proxy for cash), and
The JST dataset covers a remarkably tumultuous period in modern history including two world wars and the Great Depression. And in this time, particular countries also experienced episodes of massive inflation, other stock market crashes, bond defaults, as well as less severe episodes of inflation, stagflation, and deflation.
Why Worry?
When I write about investing, I usually start from the pragmatic assumption that governments, economies, and markets will continue to function at some minimal level. I’ve never concerned myself with investing disasters like world wars because I figure in those situations people will be more concerned with saving their lives than growing their money.
Perhaps having some gold socked away might help during the next zombie apocalypse. But investing pragmatists argue that when civilization completely breaks down, the winner will be the person with the most guns and ammo not the person with the most gold, and certainly not the person with the most stock and bond certificates.
However, the guns-and-ammo argument assumes one type of disaster: a complete breakdown of civilization everywhere. In contrast, the JST dataset documents a variety of historical investing disasters, some more severe than others. As an example, consider a Jewish person living in pre-WW II Germany. In that time and place, even a closet full of guns and ammo would be pretty useless against organized platoons of Nazis with tanks and bombs at the ready. But using saved gold (or other tangible assets like diamonds, jewelry, art, etc.) was key to some people escaping from Germany before the worst atrocities occurred.
Given the variety of possible investing disasters, it makes sense to at least consider which assets might help withstand such events, even if they are very unlikely to occur.
Deep Risk
Bill Bernstein wrote a fascinating booklet about just these sorts of investing disasters, which he calls “Deep Risk”. Using historical data similar to the JST Study, he found four flavors of deep risk:
- Severe or prolonged inflation – when the purchasing power of your money decreases over time.
- Prolonged deflation – when purchasing power increases (negative inflation).
- Confiscation – most often seen when new or higher taxes are enacted, but it can include governments freezing assets or even converting to communism.
- Devastation – the destruction caused by wars, revolutions, and other widespread violence.
Deflation sounds innocuous, but it usually occurs during economic recessions and has pernicious effects like increasing the value of debts (which is bad for the stocks of companies that carry debt), discouraging consumer spending, and increasing inflation-adjusted (real) interest rates.
Historical Examples
Bernstein’s booklet made me curious about how stocks and government bonds have historically reacted to the four types of deep risk, and the JST dataset provides some answers to that question.² Let’s go back to the example of Germany again because it offers a particularly dire example of deep risk as shown in this graph from 1900 to 2015 of the inflation-adjusted growth of one unit of local currency (denoted throughout by a “$” sign).
Wow! Now that’s some deep deep risk! One “$” (Goldmark in this case) invested in German government bonds in 1900 would have been worth about one trillionth of its original value by the end of WW II. Those bonds were literally worth less than the paper they were printed on. The same 1900 investment in stocks would have lost about 91% of its value by the end of WW II. That’s awful, but the graph shows that those stocks eventually went on to grow to nearly 50 times their original value.³
The driving force behind this bond disaster was the hyperinflation that infected Weimar Germany in 1923. I read somewhere that 150 factories printed money 24 hours a day, and there was still insufficient paper money in circulation to make simple purchases like food and clothing. When Hitler came to power he declared that all pre-Third Reich bonds were worthless, although interestingly, some investors are still trying to collect on that debt today. In contrast, stock values actually rose during Germany’s hyperinflation because investors realized that shares of tangible companies were more valuable than piles of worthless paper money. But of course, the later physical devastation of WW II eventually decimated stocks too.
German bonds got crushed by hyperinflation, but what if we isolate for the devastation of WW II? Here’s the same type of inflation-adjusted graph, but starting in 1926, after hyperinflation subsided.
In this case, investing in bonds starting in 1926 and through the end of the war performed better than stocks. Bonds lost “only” 48% of their real value, while stocks lost 88% of their value. But stocks quickly boomed again due to massive economic reconstruction under the Marshall Plan, while bonds merely crept upwards.
Perhaps Germany was unique in some respects. So, let’s look at Japan over a similar period as shown in this graph.
In the case of Japan, stocks held up a little better than government bonds at their shared nadir around 1948. But again, we see that stocks retained a much greater potential future value than bonds. An investor in Japanese bonds in 1923 would today, after nearly a century of waiting, still only have 11% of their original investment value!
Japan also provides one of the few available examples of the relatively rare condition of deflation. From 1990 through 2012, the cumulative inflation rate in Japan was a mere 5%. That’s technically not deflation, but in contrast, the cumulative inflation in other countries ranged from 35% (Switzerland) to 217% (Portugal) in the same period. So, here’s a graph of stock and bond inflation-adjusted growth in Japan over this semi-deflation period.
Bernstein looks at several other relatively rare examples of deflation (mostly in South America), and in all cases, deflation was good for government bond investors and bad for stock investors similar to this example from Japan.
Although I’m not sure Bernstein would say it qualifies as deep risk, Europe encountered a low-growth/stagflation period for about 30 years from 1956 to 1986. Here’s how that played out in the two examples of France and Belgium.
For about 30 years, many European investors in both bonds and stocks saw no reward for their patience.
I could present many more examples, but I’ll confine it to one more country-specific example: Portugal from just before WW II through to the present day.
Intriguingly, investments in either Portuguese stocks or bonds in 1937 held up pretty well throughout WW II. And then stocks rocketed upward for about a quarter-century, as Portugal nurtured a flock of healthy growing industries. Unfortunately, most of those industries happened to rely on relatively cheap oil. When the 1973 oil shock hit the world, Portuguese industries were devastated much more seriously than anything caused by WW II. And even today, Portuguese stock and bond values have still not fully recovered.
The Value of Geographic Diversification
Portugal’s resilience during WW II suggests that geographic diversification might help to mitigate some deep risks. To see if that was more generally true, I plotted the same type of inflation-adjusted growth graph for both stocks and bonds but showing all countries in the JST dataset in the 40 years from 1911 to 1949. This period includes at least three pretty distinct disasters including two wars, and for many countries, severe inflation in between the wars.
I don’t expect you to follow every line in these graphs. But if you look at the final values on the right side of the plots, you can see a wide range of outcomes during this crazy period.
For stocks (top graph), 60% of these countries (most of them involved in all three disasters) managed some level of real positive stock return. If you had invested equal amounts in the stocks of each country in 1911, your geographically diversified stock portfolio would have produced an inflation-adjusted 40-year annualized return of 2.3% (Compound Annual Growth Rate; CAGR). That’s a pretty respectable outcome given the circumstances.
For bonds (bottom graph), about half the countries managed some level of positive real bond return. An equally-weighted portfolio of bonds diversified across all these countries would have produced an inflation-adjusted 40-year annualized return of negative -0.4%, or a cumulative loss of only -13%. That’s not great, but it’s far from a total wipeout.
Deep Risk Inversion
Bernstein likes to point out that the conventional principles of investing, which focus mostly on “shallow” risk, become inverted in the topsy turvy world of deep risk. His examples include:
- Routine volatility is meaningless in the world of deep risk
- “Safe” assets like bonds and cash become risky and “risky” assets like stocks become safe.
- A 10-year investing timeframe is long-term in the shallow risk world but fleeting in the deep risk world, assuming you or your offspring are around to eventually spend those investments.4
And I’d add one more inversion. In the shallow risk world, many U.S. investors assume that bond and stock returns are negatively correlated, and they expect the two assets to reliably act as a hedge against each other.5 But for most of the deep risk examples I’ve shown, and many more examples I haven’t shown, stocks and bonds plummeted simultaneously.
What to Do About Deep Risk?
Assuming that you don’t take the pragmatic view and decide to ignore deep risk entirely, what if anything can or should be done to protect against investing disasters? Answering that question is complicated because the four deep risks don’t all have the same impacts, and they have different chances of occurring in the future.
Further, I’ve considered stocks and bonds in today’s post, but many other assets might conceivably help with one or more of the four deep risks including gold, precious metals in general,
Regular readers know I like to assess risks by multiplying the probability of occurrence times the magnitude of potential investing loss. While a scarcity of data prevents such a quantitative calculation, I adapted this table from Bernstein, which presents a more qualitative assessment of deep risks. I want to emphasize that although I use some simple numbers here, these are all expressed on a relative unitless scale and meld my and Bernstein’s opinions.
Inflation – Both Bernstein and I agree that severe or prolonged inflation is the most likely deep risk and that it’s pretty easy to insure against. There’s certainly no guarantee that stocks will do well during inflation, but there are plenty of examples where stocks fared better than bonds. And by globally diversifying your stocks you may escape inflation that impacts just one or two countries. Also, based on a few historical examples, Bernstein suggests TIPs (for obvious reasons), gold, and stocks in commodity producers.
Confiscation – This is probably the most difficult deep risk to insure against. The most obvious candidate is holding assets in foreign countries, which usually involves considerable logistical hassles and costs. However, there’s little preventing most governments from enacting laws that tax foreign assets, and many countries already have such laws. For example, six OECD countries currently have a world wealth tax, and not too long ago 14 countries had this type of tax. And several progressive presidential candidates in the U.S. championed this idea in the last election cycle. Taxes and hassles aside, I find the idea of buying some European
Devastation – We’ve seen that globally diversified stocks have acted as good insurance against devastation in the past. (And the same could be said for globally diversified bonds but to a lesser extent.) Bernstein points again to foreign-held assets and
Deflation – This is the least likely deep risk, and government bonds are the best insurance against it. However, Bernstein emphasizes investors in bonds rather than stocks are sacrificing substantial potential returns, and I generally agree. But when we’re talking about scenarios where some assets drop 90% or more in value, it seems reasonable to have a small allocation (20% or less) to bonds as deflation insurance. Such a portfolio will often moderately underperform an all-stock portfolio during times of shallow risk, while potentially retaining a useful amount of capital if stocks are completely gutted by deflation. Bernstein also suggests that gold performs even better in deflationary situations than it does in inflationary situations, but I haven’t verified that myself.
Bernstein has a lot of other observations about protecting against deep risk, so if you want to think more about potential insurance strategies, I highly recommend his booklet. But in summary, to address deep risk Bernstein recommends:
- A globally diversified stock portfolio with a tilt toward value stocks and precious metals/natural resource companies, combined with TIPS, and potentially some gold and foreign
real estate .
It also occurs to me that a small allocation to cryptocurrencies could help defend against fiat currency inflation or deflation much as gold sometimes does. However, I know far too little about the risks and potential future rewards of various cryptocurrencies to write anything inciteful about this option.
Conclusions
The overall theme to deep risk and protecting against it is diversification. This is not the kind of diversification where you look for incremental risk-adjusted return benefits from things like factors, differences in routine volatility, or short-term lack of correlation. I’m talking about Diversification with a capital “D”, where you put whole eggs in completely different baskets both in terms of asset fundamentals and the countries where those assets reside (or derive value from).
Personally, I feel that my current portfolio addresses deep risk concerns adequately. I hold a geographically diversified set of stock index funds and a little cash, which I’m drawing down in retirement. Globally diversified stocks address many of the disaster scenarios we’ve seen, particularly when those disasters are isolated to one or a few countries.
As for bonds, gold, foreign
1 – For some assets and countries the data go back even further to 1870. Also, I didn’t use some more limited JST data available for a few additional countries.
2 – I could have also used the JST dataset to make comparisons with cash and
3 – All the JST stock returns are based on a basket of stocks that represent (as close as JST could determine) the entire stock market of each country. But of course, in 1900, there were no broad market index funds. So, to achieve this result, the 1900s investor would have had to invest a little money in every home-country stock that existed at the time. Conversely, someone who invested in just a handful of stocks would have experienced an outcome that could have either been considerably better than the JST data (because the investor luckily picked a few winning companies) or a complete loss (because the investor unluckily picked a handful of losers). This is the difference between aggregate probabilities and how we actually experience life.
4 – Regular readers will notice that these three observations are pretty consistent with mindful investing principles for investing timeframes as short as about 3 years. Namely, routine volatility is pretty meaningless in the shallow risk world too. Mindful investors know that bonds, at their current yields, are riskier than stocks. And investing in stocks for less than 10 years is closer to gambling than investing.
5 – In fact, U.S. bonds and stocks have more often been positively correlated between 1930 and the present day.
