Some of the most popular content on Mindfully Investing continues to be historical returns data for various asset classes.  In a continuing effort to expand the returns datasets for different asset classes at Mindfully Investing, today’s post explores the historical returns of global central government/treasury bonds.  (If you’re looking for corporate bond returns click here.)

Beyond The United States

Most American investors are familiar with U.S. Treasury bonds and bills, which are debt instruments issued by the U.S. government.  I’ve summarized historical returns for U.S. government bonds here.  Many other central governments around the world have also issued government bonds for more than a century.  This graph from Stevens Sweet shows the relative sizes of the world’s government (dark green) and corporate (light green) bond markets as of 2015.

While the U.S. is the biggest issuer of government bonds, it still represents less than a third of the value of all government bonds issued around the world.  So, if you want to diversify the government bonds in your investment portfolio, it’s reasonable to consider countries beyond the U.S.

And it’s easy to invest in government bonds from many different countries by using low-cost index funds that track indices for:

• Global government bonds
• Developed market government bonds¹
• Emerging market government bonds¹
• Short, intermediate, and long-term global government bonds
• Global inflation-protected government bonds.

Further, many of these funds come in both “hedged” and “unhedged” forms.  In its simplest form, hedged means that the investments are tracked in U.S. dollars to eliminate currency variations from the returns.  Unhedged means the local currency is used, so that total returns are a product of both bond returns and currency fluctuations relative to your home currency.

Bond Returns Around The Globe

Although there are many types of global government bond funds, they have only existed since the mid-1990s, and many of them also include U.S. bonds as a major constituent.  So, data from these funds don’t cast much light on the long-term history of bond returns outside the U.S.

However, Credit Suisse publishes an annual report that summarizes historical government bond annualized returns (Compound Annual Growth Rate; CAGR) going back to 1900 for several groupings of 23 countries as shown in this graph.²

In inflation-adjusted (real) terms, government bonds from various developed market countries have returned 1% to 2% annualized.  The track record for global bonds is very similar to the real returns for U.S. bonds (10-Year T-Bond), which have also been around 2% annualized since 1928.   However, emerging market bonds have produced negative real returns over a comparable period.

Bond Returns by Country

The Credit Suisse study provides some additional summary analysis for individual countries.  In addition, researchers Jorda, Knoll, Kuvshinov, Schularick, and Taylor developed annual return histories for 19 developed-market countries from 1870 to 2015.  Based on their data file names, I call this the “JST Study” for short.

Given the inevitable variations that result from compiling long-term historical datasets, I think it’s useful to look at the results from both the Credit Suisse and JST studies going back to 1900 as shown in these two graphs for nominal and inflation-adjusted returns, respectively.  The JST Study (orange bars) is in local currency terms (unhedged), while the Credit Suisse Study (blue bars) is in US dollar terms (hedged).

On a nominal basis, government bond returns over more than a century ranged from about 0% to 8% annualized.  However, once inflation is accounted for, the returns ranged from -18% (the value for Germany is cutoff on the graph) to about 2.5%.  And 8 out of 19 (42%) of the countries had inflation-adjusted returns that were negative or just barely positive.

Looking further at the inflation-adjusted results, there are remarkable similarities between the two studies for Japan, Finland, Belgium, Norway, Netherlands, Australia, UK, Denmark, Switzerland, and Sweden.  This is somewhat surprising given that the two studies use different currency bases.

However, the results between the two studies were pretty different for Portugal, Italy, France, Spain, and particularly Germany.  In fact, the JST Study reported a -18% inflation-adjusted return for Germany, while the Credit Suisse Study reported a much higher -1% return.  I’ll come back to this anomaly, and some others, in a moment.

You may be interested in determining annualized bond returns between specific years for individual countries.  I previously presented historical return calculators for stocks, bonds, cash, corporate bonds, global stocks, small-cap and value stocks, gold, and real estate.  Below is a similar calculator that provides annualized bond returns (both nominal and inflation-adjusted) between any two dates from 1900 to 2015 based on the developed market countries in the JST Study.  (The Credit Suisse study does not provide their raw data, so I can’t include those data in the calculator.)  The JST Study returns are calculated on a local currency basis (unhedged) and using local inflation rates.

Data Caveats and Omissions

I already mentioned that compiling long data histories from many different countries with any consistency is undoubtedly a difficult exercise.  For this reason, you should watch out for a few issues when using the above calculator or the original JST Study dataset, if you choose to download it yourself.

Germany Data – I already noted that the JST Study produced a much lower inflation-adjusted return than the Credit Suisse study.  At least one reason for this difference is that the Credit Suisse study excluded the hyperinflationary years of 1922 and 1923 from their Germany data to arrive at a -1% real annualized return since 1900.  In contrast, the JST Study result of around a -18% real return includes all of the 1920s.

However, the JST Study has a 5-year data gap coinciding with the end of WW II.  If you assume nothing happened in this gap (no positive or negative returns), then an investor in German bonds would have suffered real losses of only about -2.4% annualized from 1943 through 1950.  This seems implausible if you consider that Hitler declared all pre-Third Reich bonds as worthless and that the international community allowed Germany to default on most of the interest that would have been due in the years after WW II.

Further, some evidence suggests that the prices of German bonds plummeted during WW II.  One study by Frey and Kucher in 2000 showed that German bond prices traded on the Swiss markets dropped by nearly 50% in 1939 and another 37% in 1945, with a few smaller dips in between.  I included these WW II price fluctuations in the calculator to help fill the data gaps for Germany.³

Other Data Gaps – Similar data gaps occur for either WW I and/or WW II for Belgium and Spain.  I didn’t try to fill these data gaps, because high inflation in these countries during these periods caused pretty drastic negative inflation-adjusted returns anyway.  So, even though the calculator assumes “nothing happened” to nominal returns during these gap years, it still generates sufficiently negative real returns to seem plausible.

Bond Data Sources – The JST Study compiled bond return data while trying to target a 10-year bond maturity.  I’ve noted before that intermediate-term bonds with a maturity of 7 to 10 years are probably the best overall representation of most bond investors.  That’s because intermediate bonds have higher returns than short-term bonds and lower interest rate risk than long-term bonds.  However, the JST Study notes that for the first half of the 20th century the only data available in every country was for longer-term bonds or even so-called “perpetuals”, which in some ways are more akin to stocks than bonds.  So, the earlier data represent long-term bonds, with only the relatively recent data coming from intermediate-term bonds.

Focus on Inflation-Adjusted Returns – In past posts about the historical returns of other asset classes, I’ve tended to focus on nominal returns for various reasons I won’t detail here.  However, in the case of global government bonds, I think it’s much more relevant to focus on inflation-adjusted returns for two reasons.  One, the JST Study (and thus my calculator) present all returns in local currency, the values of which have fluctuated wildly in some countries since 1900 due to inflation and other economic reasons.  Two, severe inflation or even hyperinflation occurred in many developed market countries around WW I and/or WW II.  For example, according to JST Study data, the inflation rate in Germany in 1923 was over 100 billion percent!  Ignoring inflation gives a distorted picture of asset performance during these unusual times.  Using Germany as an example again, the nominal annualized bond return since 1900 was nearly 6%, but the real annualized return was nearly -20%.

Although I can’t provide these sorts of detailed caveats for the Credit Suisse Study dataset, I would expect that similar data issues inevitably exist in that study as well.  I’m not disrespecting either of the two studies but rather pointing out the monumental difficulties inherent to compiling over a century’s worth of consistent bond data.

Historical Returns of Developed Market Bonds

With all those data caveats in mind, here are some additional summary statistics for developed market bonds using the JST Study and Credit Suisse datasets.  The median annualized inflation-adjusted return on government bonds excluding the U.S. since 1900 was:

• 0.6% – unhedged across all developed market countries in the JST Study.
• 1.7% – hedged across all developed market countries in the Credit Suisse Study.

Both these medians represent the central tendency of long-term results across many countries, which means that over shorter periods and in particular countries government bond returns diverged substantially from this central tendency.  This table shows some additional descriptive statistics from the JST Study for the inflation-adjusted annual returns from developed-market (excluding the U.S.) government bonds since 1900.

As you can see, in particular places at particular times over the last century government bonds could produce either frightful real losses or tremendous real gains in just one year.  But in aggregate, government bonds have produced a central tendency annual real return of about 2%, similar to the U.S. 10-Year T-Bond.

Historical Risks for Global Government Bonds

Because higher returns are usually associated with higher risks of losing money, it’s prudent to evaluate the long-term balance of both returns and risks for every investment, including global government bonds.  As measured by the standard deviation of the routine ups and downs of returns over time, volatility is the most common (but somewhat flawed) measure of investment risk.

Unfortunately, neither the Credit Suisse nor JST studies provide detailed volatility data.  However, in the past, I’ve used a relatively brief period from 2003 to 2020 to generate volatility statistics for a variety of asset classes using Portfolio Visualizer and other data sources.  I settled on this timespan because it offers volatility data for the widest range of asset classes.  These 18 years certainly don’t provide any sense of global government bond volatility in critical times like the Great Depression or world wars, but it at least provides a consistent comparison across many asset classes.

Here’s the graph plotting risk versus returns since 2003 for multiple asset classes including global government bonds (unhedged).

At least recently, global government bonds have provided both risk (volatility) and rewards (returns) that are very similar to holding U.S. 10-Year Treasury bonds.

Conclusions

Data since the beginning of the last century suggest that global government bonds offer relatively low returns as compared to stocks but with substantially lower volatility.  In many ways, global government bonds have functioned remarkably similar to U.S. government bonds.  This shallow observation suggests that diversification of government bonds across multiple developed market countries may often produce no observable benefits.

However, at the same time, we’ve seen that during periods of crisis, the government bonds of particular countries can quickly become almost worthless.  While developed market bonds have a deserved reputation as a relatively safe asset from the standpoint of routine volatility, they are nonetheless susceptible to infrequent intrinsic risks.  So, from this deeper perspective, geographic diversification of government bond holdings can act as insurance against rare catastrophic events.

These types of rare risks are sometimes called “tail risks” or “deep risk”.  Working on today’s post made it clear to me that the data from the Credit Suisse and JST studies provide a remarkable window into several types of real-life deep risk scenarios.  Consequently, I plan to dive deeper into deep risk questions in an upcoming post.

1 – If you’re curious, you can read more about how indices categorize countries into “Developed Markets” versus “Emerging Markets” here.

2 – All returns in the graph are in U.S. dollar terms (hedged), and I converted the study’s inflation-adjusted returns to nominal returns using a historical average of 3% U.S. annual inflation over this entire period.  Some values are estimated from charts in the report because Credit Suisse doesn’t publically release the raw data from the study.

3 – This is a fair warning for anyone using the calculator for German bonds.  I’m certainly no expert on bond returns during WW II, and I made these additions to the German dataset just to avoid the calculator returning results that seemed absurdly optimistic for this period.  But I wouldn’t claim that my dataset is better or even correct as compared to the original JST Study assumptions.