Stock-Bond Correlation
Jan 10, 2024
The Quanted Round-up is a curated summary that covers relevant research on key topics in quantitative financial decision-making.
Highlights
This edition explores the fundamental drivers of stock-bond correlation, emphasising the role of macroeconomic factors and market regimes in shaping their dynamics. Through empirical and theoretical analysis, the research provides a sophisticated understanding of how these relationships impact asset allocation, risk management, and broader portfolio strategies.
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations
Garth Flannery & Daniel Bergstresser
This paper builds on a framework that uses macroeconomic drivers to explain long-term variation in the correlation between stocks and bonds. The existing work focuses on the relative volatility of growth and inflation and the correlation between them and explains about 70 percent of the variation in rolling 10-year stock- bond correlation. We focus on forecasting short-term variation in stock-bond correlation with measures that capture the extent to which individual forecasters’ predictions about those markets have the same sign or opposing signs. Our framework enhances stock-bond correlation forecasting at tactical horizons, which we define here as the next three months.
Empirical evidence on the stock-bond correlation
Roderick Molenaar, Edouard Senechal, Laurens Swinkels & Zhenping Wang
The correlation between stock and bond returns is a cornerstone of asset allocation decisions. History reveals abrupt regime shifts in correlation after long periods of relative stability. We investigate the drivers of the correlation between stocks and bonds and find that inflation, real rates, and government creditworthiness are important explanatory variables. We examine the implications of a shift in the stock-bond correlation and find that increases are associated with higher multi-asset portfolio risk and higher bond risk premia.
US Stock-Bond Correlation: What are the Macroeconomic Drivers?
Junying Shen & Noah Weisberger
US stock-bond correlation, which plays an important role in institutional portfolio construction, has been persistently negative for the last 20y. This negative correlation allows stocks and bonds to serve as a hedge for each other, enabling CIOs to increase stock allocations while still satisfying a portfolio risk budget. However, stock-bond correlation is not immutable. In fact, it was consistently positive for more than 30y prior to 2000. A return to positively correlated stock and bond returns may require CIOs to rethink their asset allocation.
Stock-Bond Correlation: Theory & Empirical Results AI and Bond Values: How Large Language Models Predict Default Signal Results
Lorenzo Portelli & Thierry Roncalli
Stock-bond correlation is an important component of portfolio allocation. It is widely used by institutional investors to determine strategic asset allocation, and is carefully monitored by multi-asset fund managers to implement tactical asset allocation. Over the past 20 years, the correlation between stock and bond returns in the US has been negative, while it was largely positive prior to the dot-com crisis. Investors currently believe that a negative stock-bond correlation is more beneficial than a positive stock-bond correlation because it reduces the risk of a balanced portfolio and limits drawdowns during periods of equity market distress.
In this study, we provide an overview of stock-bond correlation modeling. In the first part, we present several theoretical models related to the comovement of stock and bond returns. We distinguish between performance and hedging assets and show that negative correlation implies a negative bond risk premium due to the covariance risk premium component. In contrast, the payoff approach can explain that bonds can be both performance and hedging assets. In addition, a good understanding of the stock-bond correlation requires an assessment of the relationship between the aggregate stock-bond correlation at the portfolio level and the individual stock-bond correlation at the asset level. Macroeconomic models are also useful in interpreting the sign of the stock-bond correlation. They can be divided into three categories: inflation-centric, real-centric, and inflation-growth based.
The second part presents the empirical results. We find that the joint dynamics of stock and bond returns differ across countries. The negative stock-bond correlation is mainly associated with the North American market and the European market before the European debt crisis. When sovereign credit risk is a concern, we generally observe a positive stock-bond correlation. However, even in the US, we cannot speak of a unique stock-bond correlation, as the level depends strongly on the composition of the equity portfolio. We also confirm the influence of the inflation factor, but the results for the growth factor are not robust. Finally, we show that the stock-bond correlation is mainly explained by the extreme market regimes, since the stock-bond correlation can be assumed to be zero in normal market regimes.
AI and Bond Values: How Large Language Models Predict Default Signals
Moazzam Khoja
This paper investigates the potential of Large Language Models (LLMs) to interpret textual data from company earnings calls and assess default likelihood. Using ChatGPT to analyse earnings call transcripts, I find that LLM-derived default likelihoods enhance the predictability of corporate bond credit spreads, independent of stock prices and liquidity measures. By comparing actual credit spreads with counterfactual spreads predicted using LLM-based default likelihoods, I show that LLMs reduce serial correlation in credit spreads, indicating less under-reaction to information. This research highlights LLMs' role in improving market efficiency by providing a consistent method for processing textual data.
Where Siegel Went Awry: Outdated Sources & Incomplete Data
Edward F. McQuarrie
When Jeremy Siegel published his Stocks for the Long Run thesis, little information was available on stocks before 1871 or bonds before 1926. But today, digital archives have made it possible to compute real total return on stock and bond indexes back to 1793. This paper presents that new market history and compares it to Siegel’s narrative. The new historical record shows that over multi-decade periods, sometimes stocks outperformed bonds, sometimes bonds outperformed stocks, and sometimes they performed about the same. More generally, the pattern of asset returns in the modern era, as seen in the Ibbotson SBBI and other datasets that begin in 1926, emerges as distinctly different from what came before. Contrary to Siegel, the pattern of asset returns seen in the 20th century does not generalize to the 19th century. A regime perspective is introduced to make sense of the augmented historical record. It argues that both common stocks and long bonds are risk assets, capable of outperforming or underperforming over any human time horizon. [A version has now published in the Financial Analysts Journal, see revision notes in paper.]
References
AI and Bond Values: How Large Language Models Predict Default Signals. October 2024. Khoja, M. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4965227
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations. December 2023. Flannery, G. and Bergstresser, D. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4672744
Empirical evidence on the stock-bond correlation. July 2023. Molenaar, R.; Senechal, E.; Swinkels, L. and Wang, Z. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4514947
Stock-Bond Correlation: Theory & Empirical Results. May 2024. Portelli, L. and Roncalli, T. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4823094
US Stock-Bond Correlation: What are the Macroeconomic Drivers? May 2021. Shen, J. and Weisberger, N. PGIM IAS - May 2021. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3855610
Where Siegel Went Awry: Outdated Sources & Incomplete Data. March 2021. McQuarrie, E.F. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3805927
The Quanted Round-up is a curated summary that covers relevant research on key topics in quantitative financial decision-making.
Highlights
This edition explores the fundamental drivers of stock-bond correlation, emphasising the role of macroeconomic factors and market regimes in shaping their dynamics. Through empirical and theoretical analysis, the research provides a sophisticated understanding of how these relationships impact asset allocation, risk management, and broader portfolio strategies.
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations
Garth Flannery & Daniel Bergstresser
This paper builds on a framework that uses macroeconomic drivers to explain long-term variation in the correlation between stocks and bonds. The existing work focuses on the relative volatility of growth and inflation and the correlation between them and explains about 70 percent of the variation in rolling 10-year stock- bond correlation. We focus on forecasting short-term variation in stock-bond correlation with measures that capture the extent to which individual forecasters’ predictions about those markets have the same sign or opposing signs. Our framework enhances stock-bond correlation forecasting at tactical horizons, which we define here as the next three months.
Empirical evidence on the stock-bond correlation
Roderick Molenaar, Edouard Senechal, Laurens Swinkels & Zhenping Wang
The correlation between stock and bond returns is a cornerstone of asset allocation decisions. History reveals abrupt regime shifts in correlation after long periods of relative stability. We investigate the drivers of the correlation between stocks and bonds and find that inflation, real rates, and government creditworthiness are important explanatory variables. We examine the implications of a shift in the stock-bond correlation and find that increases are associated with higher multi-asset portfolio risk and higher bond risk premia.
US Stock-Bond Correlation: What are the Macroeconomic Drivers?
Junying Shen & Noah Weisberger
US stock-bond correlation, which plays an important role in institutional portfolio construction, has been persistently negative for the last 20y. This negative correlation allows stocks and bonds to serve as a hedge for each other, enabling CIOs to increase stock allocations while still satisfying a portfolio risk budget. However, stock-bond correlation is not immutable. In fact, it was consistently positive for more than 30y prior to 2000. A return to positively correlated stock and bond returns may require CIOs to rethink their asset allocation.
Stock-Bond Correlation: Theory & Empirical Results AI and Bond Values: How Large Language Models Predict Default Signal Results
Lorenzo Portelli & Thierry Roncalli
Stock-bond correlation is an important component of portfolio allocation. It is widely used by institutional investors to determine strategic asset allocation, and is carefully monitored by multi-asset fund managers to implement tactical asset allocation. Over the past 20 years, the correlation between stock and bond returns in the US has been negative, while it was largely positive prior to the dot-com crisis. Investors currently believe that a negative stock-bond correlation is more beneficial than a positive stock-bond correlation because it reduces the risk of a balanced portfolio and limits drawdowns during periods of equity market distress.
In this study, we provide an overview of stock-bond correlation modeling. In the first part, we present several theoretical models related to the comovement of stock and bond returns. We distinguish between performance and hedging assets and show that negative correlation implies a negative bond risk premium due to the covariance risk premium component. In contrast, the payoff approach can explain that bonds can be both performance and hedging assets. In addition, a good understanding of the stock-bond correlation requires an assessment of the relationship between the aggregate stock-bond correlation at the portfolio level and the individual stock-bond correlation at the asset level. Macroeconomic models are also useful in interpreting the sign of the stock-bond correlation. They can be divided into three categories: inflation-centric, real-centric, and inflation-growth based.
The second part presents the empirical results. We find that the joint dynamics of stock and bond returns differ across countries. The negative stock-bond correlation is mainly associated with the North American market and the European market before the European debt crisis. When sovereign credit risk is a concern, we generally observe a positive stock-bond correlation. However, even in the US, we cannot speak of a unique stock-bond correlation, as the level depends strongly on the composition of the equity portfolio. We also confirm the influence of the inflation factor, but the results for the growth factor are not robust. Finally, we show that the stock-bond correlation is mainly explained by the extreme market regimes, since the stock-bond correlation can be assumed to be zero in normal market regimes.
AI and Bond Values: How Large Language Models Predict Default Signals
Moazzam Khoja
This paper investigates the potential of Large Language Models (LLMs) to interpret textual data from company earnings calls and assess default likelihood. Using ChatGPT to analyse earnings call transcripts, I find that LLM-derived default likelihoods enhance the predictability of corporate bond credit spreads, independent of stock prices and liquidity measures. By comparing actual credit spreads with counterfactual spreads predicted using LLM-based default likelihoods, I show that LLMs reduce serial correlation in credit spreads, indicating less under-reaction to information. This research highlights LLMs' role in improving market efficiency by providing a consistent method for processing textual data.
Where Siegel Went Awry: Outdated Sources & Incomplete Data
Edward F. McQuarrie
When Jeremy Siegel published his Stocks for the Long Run thesis, little information was available on stocks before 1871 or bonds before 1926. But today, digital archives have made it possible to compute real total return on stock and bond indexes back to 1793. This paper presents that new market history and compares it to Siegel’s narrative. The new historical record shows that over multi-decade periods, sometimes stocks outperformed bonds, sometimes bonds outperformed stocks, and sometimes they performed about the same. More generally, the pattern of asset returns in the modern era, as seen in the Ibbotson SBBI and other datasets that begin in 1926, emerges as distinctly different from what came before. Contrary to Siegel, the pattern of asset returns seen in the 20th century does not generalize to the 19th century. A regime perspective is introduced to make sense of the augmented historical record. It argues that both common stocks and long bonds are risk assets, capable of outperforming or underperforming over any human time horizon. [A version has now published in the Financial Analysts Journal, see revision notes in paper.]
References
AI and Bond Values: How Large Language Models Predict Default Signals. October 2024. Khoja, M. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4965227
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations. December 2023. Flannery, G. and Bergstresser, D. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4672744
Empirical evidence on the stock-bond correlation. July 2023. Molenaar, R.; Senechal, E.; Swinkels, L. and Wang, Z. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4514947
Stock-Bond Correlation: Theory & Empirical Results. May 2024. Portelli, L. and Roncalli, T. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4823094
US Stock-Bond Correlation: What are the Macroeconomic Drivers? May 2021. Shen, J. and Weisberger, N. PGIM IAS - May 2021. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3855610
Where Siegel Went Awry: Outdated Sources & Incomplete Data. March 2021. McQuarrie, E.F. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3805927
The Quanted Round-up is a curated summary that covers relevant research on key topics in quantitative financial decision-making.
Highlights
This edition explores the fundamental drivers of stock-bond correlation, emphasising the role of macroeconomic factors and market regimes in shaping their dynamics. Through empirical and theoretical analysis, the research provides a sophisticated understanding of how these relationships impact asset allocation, risk management, and broader portfolio strategies.
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations
Garth Flannery & Daniel Bergstresser
This paper builds on a framework that uses macroeconomic drivers to explain long-term variation in the correlation between stocks and bonds. The existing work focuses on the relative volatility of growth and inflation and the correlation between them and explains about 70 percent of the variation in rolling 10-year stock- bond correlation. We focus on forecasting short-term variation in stock-bond correlation with measures that capture the extent to which individual forecasters’ predictions about those markets have the same sign or opposing signs. Our framework enhances stock-bond correlation forecasting at tactical horizons, which we define here as the next three months.
Empirical evidence on the stock-bond correlation
Roderick Molenaar, Edouard Senechal, Laurens Swinkels & Zhenping Wang
The correlation between stock and bond returns is a cornerstone of asset allocation decisions. History reveals abrupt regime shifts in correlation after long periods of relative stability. We investigate the drivers of the correlation between stocks and bonds and find that inflation, real rates, and government creditworthiness are important explanatory variables. We examine the implications of a shift in the stock-bond correlation and find that increases are associated with higher multi-asset portfolio risk and higher bond risk premia.
US Stock-Bond Correlation: What are the Macroeconomic Drivers?
Junying Shen & Noah Weisberger
US stock-bond correlation, which plays an important role in institutional portfolio construction, has been persistently negative for the last 20y. This negative correlation allows stocks and bonds to serve as a hedge for each other, enabling CIOs to increase stock allocations while still satisfying a portfolio risk budget. However, stock-bond correlation is not immutable. In fact, it was consistently positive for more than 30y prior to 2000. A return to positively correlated stock and bond returns may require CIOs to rethink their asset allocation.
Stock-Bond Correlation: Theory & Empirical Results AI and Bond Values: How Large Language Models Predict Default Signal Results
Lorenzo Portelli & Thierry Roncalli
Stock-bond correlation is an important component of portfolio allocation. It is widely used by institutional investors to determine strategic asset allocation, and is carefully monitored by multi-asset fund managers to implement tactical asset allocation. Over the past 20 years, the correlation between stock and bond returns in the US has been negative, while it was largely positive prior to the dot-com crisis. Investors currently believe that a negative stock-bond correlation is more beneficial than a positive stock-bond correlation because it reduces the risk of a balanced portfolio and limits drawdowns during periods of equity market distress.
In this study, we provide an overview of stock-bond correlation modeling. In the first part, we present several theoretical models related to the comovement of stock and bond returns. We distinguish between performance and hedging assets and show that negative correlation implies a negative bond risk premium due to the covariance risk premium component. In contrast, the payoff approach can explain that bonds can be both performance and hedging assets. In addition, a good understanding of the stock-bond correlation requires an assessment of the relationship between the aggregate stock-bond correlation at the portfolio level and the individual stock-bond correlation at the asset level. Macroeconomic models are also useful in interpreting the sign of the stock-bond correlation. They can be divided into three categories: inflation-centric, real-centric, and inflation-growth based.
The second part presents the empirical results. We find that the joint dynamics of stock and bond returns differ across countries. The negative stock-bond correlation is mainly associated with the North American market and the European market before the European debt crisis. When sovereign credit risk is a concern, we generally observe a positive stock-bond correlation. However, even in the US, we cannot speak of a unique stock-bond correlation, as the level depends strongly on the composition of the equity portfolio. We also confirm the influence of the inflation factor, but the results for the growth factor are not robust. Finally, we show that the stock-bond correlation is mainly explained by the extreme market regimes, since the stock-bond correlation can be assumed to be zero in normal market regimes.
AI and Bond Values: How Large Language Models Predict Default Signals
Moazzam Khoja
This paper investigates the potential of Large Language Models (LLMs) to interpret textual data from company earnings calls and assess default likelihood. Using ChatGPT to analyse earnings call transcripts, I find that LLM-derived default likelihoods enhance the predictability of corporate bond credit spreads, independent of stock prices and liquidity measures. By comparing actual credit spreads with counterfactual spreads predicted using LLM-based default likelihoods, I show that LLMs reduce serial correlation in credit spreads, indicating less under-reaction to information. This research highlights LLMs' role in improving market efficiency by providing a consistent method for processing textual data.
Where Siegel Went Awry: Outdated Sources & Incomplete Data
Edward F. McQuarrie
When Jeremy Siegel published his Stocks for the Long Run thesis, little information was available on stocks before 1871 or bonds before 1926. But today, digital archives have made it possible to compute real total return on stock and bond indexes back to 1793. This paper presents that new market history and compares it to Siegel’s narrative. The new historical record shows that over multi-decade periods, sometimes stocks outperformed bonds, sometimes bonds outperformed stocks, and sometimes they performed about the same. More generally, the pattern of asset returns in the modern era, as seen in the Ibbotson SBBI and other datasets that begin in 1926, emerges as distinctly different from what came before. Contrary to Siegel, the pattern of asset returns seen in the 20th century does not generalize to the 19th century. A regime perspective is introduced to make sense of the augmented historical record. It argues that both common stocks and long bonds are risk assets, capable of outperforming or underperforming over any human time horizon. [A version has now published in the Financial Analysts Journal, see revision notes in paper.]
References
AI and Bond Values: How Large Language Models Predict Default Signals. October 2024. Khoja, M. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4965227
A Changing Stock-Bond Correlation: Explaining Short-term Fluctuations. December 2023. Flannery, G. and Bergstresser, D. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4672744
Empirical evidence on the stock-bond correlation. July 2023. Molenaar, R.; Senechal, E.; Swinkels, L. and Wang, Z. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4514947
Stock-Bond Correlation: Theory & Empirical Results. May 2024. Portelli, L. and Roncalli, T. Available at SSRN: http://dx.doi.org/10.2139/ssrn.4823094
US Stock-Bond Correlation: What are the Macroeconomic Drivers? May 2021. Shen, J. and Weisberger, N. PGIM IAS - May 2021. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3855610
Where Siegel Went Awry: Outdated Sources & Incomplete Data. March 2021. McQuarrie, E.F. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3805927
