AgPa #5: The Return on Everything

The Rate of Return on Everything, 1870–2015 (2019)
Òscar Jordà, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, Alan M. Taylor
The Quarterly Journal of Economics 134(3), 1225-1298, URL

This week’s AGNOSTIC Paper is somewhat related to the global market portfolio but has a different focus. The authors tackle a simple, yet difficult question: what is the rate of return in an economy? To answer that, they go far into history. They estimate total returns of equity, housing, bonds, and bills in 16 advanced economies for the period from 1870 to 2015. As you can imagine, this involves some heavy data work and the authors received a lot of credit for constructing this unique dataset.

Everything that follows is only my summary of the original paper. So unless indicated otherwise, all tables and charts belong to the authors of the paper and I am just quoting them. The authors deserve full credit for creating this material, so please always cite the original source.

Setup and Idea

The idea of the paper is simple. What is the aggregate rate of return in an economy? This is an important question for several reasons. First, investors use historic returns to form expectations about the future. Although there are a lot of issues with that (see below), most people do it anyway.1”Most people do it” is of course no valid excuse and I will refer to some alternatives by Cliff Asness and Aswath Damodaran below. Second, the rate of return is at the heart of some fierce discussions among economists. For example, Thomas Piketty discusses in his popular book “Capital in the Twenty-First Century” how the fact that the rate of return is larger than the economic growth rate (r versus g) is driving wealth-inequality.

As a consequence, the paper combines a lot of different but related topics. I will mainly focus on the empirical analyses of asset prices over this unique 145 year sample period and not comment too much on the rest. That said, the other sections of the paper are definitely also worth reading. So please keep in mind that this is not a comprehensive summary but rather a focused cherry-picking from someone who is mostly interested in finance and investing.

Data and Methodology

The authors estimate nominal and real returns of bills, bonds, equities, and housing for the period from 1870 to 2015 in 16 advanced economies. Those asset classes are of course not very granular but I guess it is hardly possible to cover such a long period with more details. In addition to that, public equities and housing make up a large fraction of household wealth. Therefore, they are a reasonable proxy for risky returns. The authors also validate this assumption with data on national wealth holdings. The following table summarizes the structure of the sample.

Table I of Jordà et al. (2019).

The authors define Bills as short-term, fixed-income government securities. The yield on those securities approximates the risk-free return of the respective economy. Bonds are longer-term, fixed-income government securities with a target maturity of about 10 years. The data mainly comes from previous research of the same authors and exchanges.2The earlier research is also peer-reviewed, so the data should be reliable.

Data on Equity returns comes from various sources like statistical offices, central banks, stock exchanges, and financial history journals. Whenever possible, the authors approximate equity returns with market-capitalization-weighted indices. This is reasonable since such indices are the closest proxies for the equity-market-portfolio. To construct the return on Housing, the authors combine data on rents and house prices from various sources. The long history of housing returns is a major contribution of the paper, as this hasn’t been done at that scale before.

Together, the value-weighted composite of Equities and Housing represents the risky return of the respective economy. In general, the authors estimate total returns for all asset classes. Total returns capture both price appreciation and running yields such as coupons, rents, or dividends.

Given the complexity of the task, there are of course a lot more details about the data and methodology. In fact, the authors spend about 25 pages to discuss their data and how it relates to existing literature. In addition to that, there is a very comprehensive online appendix that includes even more information. Since most of this is fairly technical, I will not cover it at this point. The authors are well-respected in the academic community and transparently explain what they are doing. Furthermore, the paper is published in The Quarterly Journal of Economics, one of the Top 3 journals in the field of economics. Overall, I am therefore pretty confident that the following results are based on sound methodology and data.

Important Results and Takeaways

Real returns on risky assets were 7-8% from 1870 to 2015

The following table shows the average annual return for each of the four asset classes between 1870 and 2015 (Panel A) and from 1950 to 2015 (Panel B). To avoid (positive) biases, the authors include all periods of war. However, they exclude periods of hyperinflation because of data issues but mention that those are negligible.

Table II of Jordà et al. (2019).

Among the most surprising results is certainly the risk-adjusted outperformance of Housing compared to Equity. The average return per year is almost identical3Nominal: 11% vs. 10.65%, Real: 7.06% vs. 6.88%, but Housing had less than half the volatility (standard deviation) of Equity.4Nominal: 10.64% vs. 22.55%, Real: 9.93% vs. 21.79%

To see the impact of that, consider the geometric mean. Due to lower volatility, Housing outperforms Equity by about 2 percentage points per year.5Nominal: 10.53% vs. 8.49%, Real: 6.62% vs. 4.66% For a patient buy-and-hold investor, 2 percentage points a year make a huge difference. A dollar invested at 10.53% becomes $2,016,623 after 145 years whereas a dollar invested at 8.49% becomes “only” $135,365. That’s the power of exponential growth.

In my opinion, the statistics in the table are just amazing. Risky assets (Equity and Housing) generated astonishing double-digit nominal returns over those 145 years. Thus, they more than compensated inflation and real returns are clearly positive.6Applying the basic formula for t-statistics to the numbers in the table also supports that statistically. That’s a remarkable creation of wealth and the statistics are even better for the post-war period from 1950 to 2015. Depending on the period, real returns on risky assets were between 6.88% and 8.3% per year.

Can we expect the same high returns for the next 145 years? Unfortunately, nobody knows and we cannot answer this question with historic data. However, the table tells us that capitalism worked just great over the last 145+ years. The system is certainly not perfect and not everybody got her fair share. But in aggregate, the creation of wealth is just amazing. I firmly believe that productive assets will continue to produce decent real returns in the future. Businesses and real estate serve enduring human needs and as long as we preserve some form of market economy, their value should increase over time. In summary, I have no idea about the specific number but I strongly believe that risky returns will continue to be positive in the future.

Returns on risky assets were substantial but volatile

Let’s take a closer look at the returns of the two risky assets. The following chart shows real returns of Equity and Housing over time. To make the chart more readable, the authors show 10-year moving averages. This of course introduces some form of look-ahead bias7For example, the point for 1900 is the average annual return between 1895 and 1905. Obviously, the returns for 1901 to 1905 were not known in 1900. Thus, this is not a proper simulation, but an ex-post analysis., but highlights some long-term cycles. The shaded gray areas are the two world-wars.

Figure VII of Jordà et al. (2019).

The chart also reveals the much higher volatility of Equity compared to Housing. Real returns on Housing (green line) were mostly between 4% and 8% per year. In contrast, annual returns on Equity fluctuated widely between -4% and +13%. In my opinion, there are no clear long-term trends. Instead, the authors mention that risky assets experienced some severe boom-and-bust cycles. For example, annual returns on equities crashed during WW1 and the great depression around 1930. The two “more recent” stock market crashes around 2000 and 2008 are also observable in the chart. Some patterns are similar for Housing, but the extent of fluctuations is much smaller.

Overall, returns fluctuate tremendously over time and the high average returns mentioned before are by far not stable. In addition to that, we have so far only looked at the average return of all 16 economies. But unsurprisingly, the returns differ quite a lot across countries.

Table VII of Jordà et al. (2019).

I will of course not comment on all numbers in the table. Instead, I want to highlight one particular issue that was surprising to me. By far the best performing country was Finland.8Other nordic countries like Sweden or Denmark also stand out positively. For the most recent period (Post-1980), the Finish equity market returned astonishing 16.32% per year on average. And those are already real returns after inflation. While we are all used to the outperformance of the US stock market, the last decade was not normal by historic standards. The US equity market performed better than average in all depicted periods, but not as strong as over the last 14 years.

Table VIII of Jordà et al. (2019).

In the final part of their discussion on risky returns, the authors further dig into the composition of Equity– and Housing returns. They decompose the total return into a running yield (Dividend income for Equity, Rental income for Housing) and capital gains. For both asset classes, the yield-component is very important. For example, 5.5% of the 6.92% annual return on housing comes from rental income. This corresponds to almost 80% of the total return. Capital gains became more important in “more recent” years as the numbers for the Post-1950 period show.

It is also fascinating that the running yields are fairly stable over time. Basically all of the volatility is driven by price fluctuations, i.e. the standard deviation of capital gains. I haven’t looked at the data, but I would guess that this pattern is still valid today.

Realized risk premiums fluctuate widely across time and countries

As mentioned before, the paper is very comprehensive and the authors also provide detailed analyses of historical safe returns and the ongoing inequality discussion about r versus g. To maintain my finance cherry-picking, I will skip those parts and focus on the section about realized risk premiums in this final part of the post.

The following chart shows 10-year moving averages for real risk premiums. The risk premium is simply the difference between risky- and safe returns. This is an important concept because it tells us something about the risk-preferences of investors. The more risk averse they are, the higher the risk premium and vice versa. Also note that the risk premium should be positive. In the end, almost everybody wants to have some compensation for risk and demands a positive risk premium. Typically, the risk premium spikes during crises and contracts during booms.

As illustrated in the two charts, there are of course two components. For example, the risk premium spiked tremendously during WW2 but this was mainly driven by a collapse of safe returns. Since both risky- and safe returns vary heavily over time, the risk premium is also very unstable. However, it fluctuated within a more narrow range around 1.5-4.5% in more recent times (1980-2015).

Figure XIII of Jordà et al. (2019).

Similar to the previous section on risky returns, this is the global average for all 16 economies. But again, it is not surprising that the risk premium also fluctuates heavily across countries. The table reveals a similar picture as the chart above. For all three sample periods, real risky returns differ among countries but were consistently positive across time. In contrast, safe returns were actually negative in most countries (on a real basis after inflation) for the period from 1950 to 1980. On the other hand, they were again positive for the period Post-1980 and for some countries almost comparable to their risky returns (e.g. 5.21% vs 4.14% for Italy).

Table XI of Jordà et al. (2019).


One final but important comment on those statistics. All results are based on historic realized returns. In most applications, however, we need forward-looking expected returns. Past returns are a good starting point for that, but they are not sufficient. For example, the S&P 500 returned about 13% per year over the last 10 years. Using this historic average as expected return for the future is the same as assuming a dice can only return 3s after you observed a series of ten 3s in a row. It is psychologically tempting but still a bad idea.

Don’t get me wrong. It is great to have this long history of risky returns and there is a lot to learn from it. But to get estimates for the future, other approaches may be better. For example, Cliff Asness presents a framework to form expectations about future returns while considering the impact of valuation changes. Similarly, Aswath Damodaran advocated for years that historic averages are a poor measure for risk premiums and recommends a forward-looking approach based on earnings estimates.

Conclusions and Further Ideas

Why did I choose this paper for AGNOSTIC Papers? To be honest, I don’t think that the specific numbers are that useful for practical applications. The authors capture very long-term averages and cycles. It is very hard too trade on such insights. Even if your forecast for the next 40 years is correct, you need to wait 40 years to check it. So the numbers and statistics themselves were not the primary reason.

I picked the paper because on a higher level, I think the results are just amazing. My personal bottom line of the paper is simple: capitalism worked exceptionally well over the last 140+ years. Whether the real return was 7, 8, or 9% doesn’t really matter. What matters is that it was consistently positive over the very long-term and that the world has produced an insane amount of wealth. Yes, wealth- and income-distributions are very unequal and this becomes more and more a problem. Yes, there are severe crises from which some people don’t recover. Yes, there is a lot of volatility in asset prices which is psychologically challenging. But overall, capitalism is probably the best system we have and we should appreciate the power of markets.

Looking at the very long history of financial markets has probably rarely been as important as today. We currently have something like the perfect storm and there is a lot of pessimism. War in Ukraine, high inflation rates across the globe, a pandemic that is still not under control, climate change remains a severe threat to humanity, and I am sure you will find a lot more negative issues. With this post, I want to take the other side of this. As long as we maintain some form of market-system, I am optimistic for the long-term. I don’t know how many wars and what other bad stuff happened between 1870 and 2015, but thanks to the authors we know that an investment in global equities and housing still turned out very well. I think this is important to keep in mind whenever you read something negative about Ukraine, inflation, or interest rates.9Of course, I don’t want to downplay any of those issues and this is just the financial perspective on the world. For people suffering from war, a paper with impressive statistics is not really helpful and there are more important things in life than some rate of return.



This content is for educational and informational purposes only and no substitute for professional or financial advice. The use of any information on this website is solely on your own risk and I do not take responsibility or liability for any damages that may occur. The views expressed on this website are solely my own and do not necessarily reflect the views of any organisation I am associated with. Income- or benefit-generating links are marked with a star (*). All content that is not my intellectual property is marked as such. If you own the intellectual property displayed on this website and do not agree with my use of it, please send me an e-mail and I will remedy the situation immediately. Please also read the Disclaimer.

Endnotes

Endnotes
1 ”Most people do it” is of course no valid excuse and I will refer to some alternatives by Cliff Asness and Aswath Damodaran below.
2 The earlier research is also peer-reviewed, so the data should be reliable.
3 Nominal: 11% vs. 10.65%, Real: 7.06% vs. 6.88%
4 Nominal: 10.64% vs. 22.55%, Real: 9.93% vs. 21.79%
5 Nominal: 10.53% vs. 8.49%, Real: 6.62% vs. 4.66%
6 Applying the basic formula for t-statistics to the numbers in the table also supports that statistically.
7 For example, the point for 1900 is the average annual return between 1895 and 1905. Obviously, the returns for 1901 to 1905 were not known in 1900. Thus, this is not a proper simulation, but an ex-post analysis.
8 Other nordic countries like Sweden or Denmark also stand out positively.
9 Of course, I don’t want to downplay any of those issues and this is just the financial perspective on the world. For people suffering from war, a paper with impressive statistics is not really helpful and there are more important things in life than some rate of return.