In the first two parts of our discussion of the excellent analysis of stocks and bonds by Edward McQuarrie, we showed that stocks are much riskier in the long run than many investors believe, both in absolute returns and relative to bonds. Whether stocks outperform bonds or vice versa is essentially driven by the market regime and long-term trends in interest rates, economic growth, inflation, etc. These long-term changes in market regimes also influence the volatility of stocks and bonds and the correlation between these two asset classes. And this means that in portfolio construction, we may have to abandon some long-held beliefs.
Anyway, I am increasingly amazed by your ability to write enlightening articles every day.
A new test should be created: The Klement test, a sort of Turing test.
If AI can create a blog and match the quality of “Klement on Investing”, then it means we have reached definitive AGI (Artificial General Intelligence).
Good evening sir. When you write: "if you have ever studied GARCH models, this stationarity is explicitly built into the model. Yet, the chart below shows that neither stocks nor bonds show stationary volatility", you confront GARCH volatility to rolling volatility. This is not a homogeneus comparison (garch volatility is conditional to a specific time interval and produced by a well defined parametric model, while rolling vol is an average stand deviation over a given time window). Besides, your sentence seems to claim Garch models are intrinsically stationary: but this is not nnecessarily the case are, just think about the Integrated GARCH model (although we could disvuss long about the difficulty to efficiently test the presence of a unitary root in a variance process). However, going beyond these econometric technicalities, what is more striking to me is that you seem to consider the rolling volatility a "better" estimator of the true, by default unknown, volatility being this claim hardly supported by statistical test and empirical assessment). I believe the confusion arises from tha fact that non stationarity is not the same thing as time-varying: in fact, volatility and correlation can definitely change over time without being non stationary, and this actually is the most important feature of stationary garch models. Therefore, imho, the conlusions drawn aboit thiss issue in your eccellente artiche are correct (vol/correl absolutely are not stabile over time), but they are not demonstrated by the chart you show. Sincerely, Valerio P.S. this econometric clarification, as I've already stressed, does not change my judgement about the contents of your analysis, which I fully support and appreciate
Good article, thanks for sharing. Much of what you are describing here is very much in line with Seb Page's and Robert Panariello's "When Diversification Fails" article (I think Page also wrote a book called "Beyond Diversification"). The full article is available from CFA Publications https://doi.org/10.2469/faj.v74.n3.3 Unexpected changes to
the discount rate or inflation expectations can push
the stock–bond correlation into positive territory—
especially when other conditions remain constant.
Anyway, just thought it was interesting that Page and his co-author reached some of the same conclusions as to what this may mean for asset allocation.
Apologies as my comment got cut-off, but Page's article also mentioned "Regime Shifts". Positing that regime shifts might be caused by the fact that macroeconomic fundamentals exhibiting regime shifts but also being driven by investor sentiment where panics incite indiscriminate selling.
Concluding:
"In an apocryphal story, a statistician who had his head in the oven and his feet in the freezer exclaimed, “On average, I feel great!” Similarly, as a measure of diversification, the full-sample correlation is an aver-age of extremes. Conditional correlations reveal that during market crises, diversification across risk assets almost completely disappears. Moreover, diversification seems to work remarkably well when investors do not need it—during market rallies. This undesirable
asymmetry is pervasive across markets. Our findings are not new, but we proposed a robust
approach to measure left- and right-tail correlations, and we documented the extent of the failure of
diversification on a large dataset of asset classes and risk factors. The good news is that tail risk–
aware analytics, as well as hedging and dynamic strategies, are now widely available to help inves-
True, now that you mention Sebastian's article I remember reading it at the time and in fact sitting at a conference in the audience when he talked about it. Thanks for the reminder.
Interesting and thought provoking article, thank you. I read this and then picked up this weeks Investors Chronicle which contained the following words from the Global Investment Returns Yearbook (Marsh, Dimson, Staunton).......'Since 1900 equities have outperformed bonds, bills and inflation in every market for which there is a continuous history'........ Does this clash with your article or are you saying the volatility and bond-equity correlation is misunderstood? Once again, great articles, thanks
It doesn't clash with my article at all. It just has a different focus. If you look at the time period 1900 to 2023 then yes, equities outperform bonds. But who has a 123 year investment horizon? My biggest criticism about the Global Investment Returns Yearbook is the same I have with Jeremy Siegel's Stocks for the Long Run: It focuses almost entirely on the very, very long-term outcome and completely ignores the path on how to get there. But if you are an investor, the path matters because if you have too much of a drawdown or it takes too long to recover, you will sell your investments even if eventually it will recover. Think about Amazon shares. If you bought them in the mid-1990s at their IPO and held them to this day, you would have made enormous amounts of money, but except for Jeff Bezos hardly anyone did. You would have to suffer through a 90% drawdown from 2000 to 2003 and then wait for more than a decade to make up your losses.
And this is where the advice of 'Hold stocks for the long run' is simply an illusion and in my view almost dangerous misselling. It is impossible to implement in practice.
I have written a brief summary of the Common pitfalls and fallacies in financial data analysis, applied precisely to the examples discussed.
https://themarketjourney.substack.com/p/dont-always-trust-regression-and
Anyway, I am increasingly amazed by your ability to write enlightening articles every day.
A new test should be created: The Klement test, a sort of Turing test.
If AI can create a blog and match the quality of “Klement on Investing”, then it means we have reached definitive AGI (Artificial General Intelligence).
Good evening sir. When you write: "if you have ever studied GARCH models, this stationarity is explicitly built into the model. Yet, the chart below shows that neither stocks nor bonds show stationary volatility", you confront GARCH volatility to rolling volatility. This is not a homogeneus comparison (garch volatility is conditional to a specific time interval and produced by a well defined parametric model, while rolling vol is an average stand deviation over a given time window). Besides, your sentence seems to claim Garch models are intrinsically stationary: but this is not nnecessarily the case are, just think about the Integrated GARCH model (although we could disvuss long about the difficulty to efficiently test the presence of a unitary root in a variance process). However, going beyond these econometric technicalities, what is more striking to me is that you seem to consider the rolling volatility a "better" estimator of the true, by default unknown, volatility being this claim hardly supported by statistical test and empirical assessment). I believe the confusion arises from tha fact that non stationarity is not the same thing as time-varying: in fact, volatility and correlation can definitely change over time without being non stationary, and this actually is the most important feature of stationary garch models. Therefore, imho, the conlusions drawn aboit thiss issue in your eccellente artiche are correct (vol/correl absolutely are not stabile over time), but they are not demonstrated by the chart you show. Sincerely, Valerio P.S. this econometric clarification, as I've already stressed, does not change my judgement about the contents of your analysis, which I fully support and appreciate
Thanks Valerie for the clarification. I just learned something new about GARCH models. And admittedly, I am not an econometrics specialist.
Good article, thanks for sharing. Much of what you are describing here is very much in line with Seb Page's and Robert Panariello's "When Diversification Fails" article (I think Page also wrote a book called "Beyond Diversification"). The full article is available from CFA Publications https://doi.org/10.2469/faj.v74.n3.3 Unexpected changes to
the discount rate or inflation expectations can push
the stock–bond correlation into positive territory—
especially when other conditions remain constant.
Anyway, just thought it was interesting that Page and his co-author reached some of the same conclusions as to what this may mean for asset allocation.
Apologies as my comment got cut-off, but Page's article also mentioned "Regime Shifts". Positing that regime shifts might be caused by the fact that macroeconomic fundamentals exhibiting regime shifts but also being driven by investor sentiment where panics incite indiscriminate selling.
Concluding:
"In an apocryphal story, a statistician who had his head in the oven and his feet in the freezer exclaimed, “On average, I feel great!” Similarly, as a measure of diversification, the full-sample correlation is an aver-age of extremes. Conditional correlations reveal that during market crises, diversification across risk assets almost completely disappears. Moreover, diversification seems to work remarkably well when investors do not need it—during market rallies. This undesirable
asymmetry is pervasive across markets. Our findings are not new, but we proposed a robust
approach to measure left- and right-tail correlations, and we documented the extent of the failure of
diversification on a large dataset of asset classes and risk factors. The good news is that tail risk–
aware analytics, as well as hedging and dynamic strategies, are now widely available to help inves-
tors manage the failure of diversification."
True, now that you mention Sebastian's article I remember reading it at the time and in fact sitting at a conference in the audience when he talked about it. Thanks for the reminder.
Interesting and thought provoking article, thank you. I read this and then picked up this weeks Investors Chronicle which contained the following words from the Global Investment Returns Yearbook (Marsh, Dimson, Staunton).......'Since 1900 equities have outperformed bonds, bills and inflation in every market for which there is a continuous history'........ Does this clash with your article or are you saying the volatility and bond-equity correlation is misunderstood? Once again, great articles, thanks
It doesn't clash with my article at all. It just has a different focus. If you look at the time period 1900 to 2023 then yes, equities outperform bonds. But who has a 123 year investment horizon? My biggest criticism about the Global Investment Returns Yearbook is the same I have with Jeremy Siegel's Stocks for the Long Run: It focuses almost entirely on the very, very long-term outcome and completely ignores the path on how to get there. But if you are an investor, the path matters because if you have too much of a drawdown or it takes too long to recover, you will sell your investments even if eventually it will recover. Think about Amazon shares. If you bought them in the mid-1990s at their IPO and held them to this day, you would have made enormous amounts of money, but except for Jeff Bezos hardly anyone did. You would have to suffer through a 90% drawdown from 2000 to 2003 and then wait for more than a decade to make up your losses.
And this is where the advice of 'Hold stocks for the long run' is simply an illusion and in my view almost dangerous misselling. It is impossible to implement in practice.
Agreed with your summation below. Clash was probably to strong a word, but focus is a better word. If only investing was that simple! Thanks again.