1. Introduction
This article explores the concept of swap spreads with a specific focus on the phenomenon of negative swap spreads. In the first part, we review the existing literature on the topic and present an overview of the possible explanations for this occurrence. In the second, more quantitative section, we perform a series of empirical analyses, including mean modeling, correlation analysis, regression analysis, and principal component analysis, with the objective of identifying the key macro-financial drivers behind swap spread movements.
2. The Phenomenon of Negative Swap Spread
An Interest Rate swap is a financial contract where two counterparties agree to exchange future cashflows linked to interest rates. The plain vanilla IRS includes a party that pays a fixed interest rate on the notional amount, while the floating leg is anchored to a floating rate such as the EURIBOR, SOFR, SONIA, and TONA. Swap spreads are defined as the difference between the fixed rate on a plain-vanilla fixed-for-floating Interest Rate Swap of a particular maturity and the yield on a Treasury of the same maturity. For instance, let’s assume that the 5-year swap rate is 2.8% and the 5-year Treasury yield is 3.2%, the 5-year swap spread would then be -40 basis points. A trader expecting the spread to widen (i.e., become more negative) might receive fixed in the swap and short the Treasury, profiting if the swap rate falls or the Treasury yield rises relative to each other.
Since the financial crisis of 2008, swap spreads have been in negative territory in the past years, implying that investors are willing to accept lower yield on swaps than Treasuries of the same maturity. Following the default of Lehman Brothers, the 30-year swap spread experienced a huge decline and turned negative. The 10-year spread became negative in 2015. This phenomenon has puzzled market participants since the floating rate of the IRS is anchored to the Libor rate, which reflects the credit risk of large financial institutions, whereas the Treasury yield reflects only the credit risk of the U.S. government. Even the reduction of counterparty risk priced into swaps, thanks to market innovations such as the introduction of mandatory central clearing for IRSs in 2013, could only push the swap spread towards zero, not into negative. Furthermore, spreads to U.S. Treasuries remained positive for intermediate maturities until the second half of 2015, suggesting that reduction in counterparty risk is not the main driver of negative swap spreads. This occurrence represents, at least theoretically, an arbitrage opportunity: go long on the Treasury and finance the purchase though borrowing at short term repo rate. Simultaneously, enter into a swap with the same maturity, paying fixed and receiving floating. If interest rates were the only risk factors in this trade, holding to maturity would offer pure “carry” arbitrage. Why then has this apparent pricing anomaly remained remarkably persistent over time?
According to Boyarchenko (2018), deviations of swap spreads away from zero suggest the presence of other risky factors, such as such as counterparty risk for the execution of the swap leg of the trade, ancillary costs to the trade, and limits to arbitrage—which make holding the trade to maturity infeasible. In his article, he suggests that capital increases in required leverage ratios have significantly helped to narrow spreads since 2015. Consider first the balance sheet impact of the long-Treasury leg of the Treasury-interest rate swap trade. Assuming that the trade size is $10 million, the trade increases the Treasury position on the asset side of the balance sheet by $10 million. Since the purchase is repo founded, the value of securities sold under agreements to repurchase on the liabilities side increases by $10 million. Furthermore, the dealer borrows the haircut on the repurchase agreement in short-term funding markets, increasing its short-term debt. At the trade’s inception, the fixed rate is set such that the fair value of the swap is $0, but as the floating rate changes, the market-clearing fixed rate fluctuates as well. Thus, the fair value of IRS on the balance sheet of the dealer paying the fixed rate moves, leading into either an increase in the “Derivatives with a positive fair value” line on the asset side or the “Derivatives with a negative fair value” line on the liabilities side. The gross notional amount of repo financing, initial margin required by the swap (which is reflected in an increase in receivables), repo haircut, and off-balance sheet exposure of the derivative instrument requiring dealers to hold additional equity under the Supplementary Leverage Ratio (SLR) before entering the trade.Boyarchenko’s analysis suggests that, given the balance sheet costs, these spreads must reach more negative levels to generate an adequate ROE for dealers. In fact, the ROE for a given trade is very sensitive to capital charges and has declined because of higher leverage requirements.
Of course, both demand and supply dynamics in the market for interest rate swaps and government bonds may have contributed to the negative spreads observed in recent years. According to Acquilina (2024), this phenomenon signals a pressure in government debt absorption. In cash bond markets, investors’ inability/unwillingness to absorb debt issuance or sales by other bondholders at prevailing prices exerts upward pressure on bond yields, pushing swap spreads lower.
Klingler (2019) offers a demand-driven explanation. Pension funds can balance their asset-liability duration by investing in long-term bonds or by receiving fixed in an IRS with long maturity. They prefer to hedge with IRS rather than buying Treasuries because the swap requires only modest investment to cover margins, whereas buying a government bond requires outright investment. This demand, when coupled with bank-affiliated dealers’ balance sheet constraints, results in negative swap spreads. Jermann (2020) provides instead a supply perspective. In his paper, he develops a model for pricing interest rates swaps that features limited arbitrage and where debt prices are instead exogenous. As it has become more costly for banks to hold Treasuries, dealers have smaller bond positions. With frictions, they are instead less exposed to long-term interest rate risk and therefore they require less compensation for the risk of the fixed swap rate, lowering the swap rate.
Since the financial crisis of 2008, swap spreads have been in negative territory in the past years, implying that investors are willing to accept lower yield on swaps than Treasuries of the same maturity. Following the default of Lehman Brothers, the 30-year swap spread experienced a huge decline and turned negative. The 10-year spread became negative in 2015. This phenomenon has puzzled market participants since the floating rate of the IRS is anchored to the Libor rate, which reflects the credit risk of large financial institutions, whereas the Treasury yield reflects only the credit risk of the U.S. government. Even the reduction of counterparty risk priced into swaps, thanks to market innovations such as the introduction of mandatory central clearing for IRSs in 2013, could only push the swap spread towards zero, not into negative. Furthermore, spreads to U.S. Treasuries remained positive for intermediate maturities until the second half of 2015, suggesting that reduction in counterparty risk is not the main driver of negative swap spreads. This occurrence represents, at least theoretically, an arbitrage opportunity: go long on the Treasury and finance the purchase though borrowing at short term repo rate. Simultaneously, enter into a swap with the same maturity, paying fixed and receiving floating. If interest rates were the only risk factors in this trade, holding to maturity would offer pure “carry” arbitrage. Why then has this apparent pricing anomaly remained remarkably persistent over time?
According to Boyarchenko (2018), deviations of swap spreads away from zero suggest the presence of other risky factors, such as such as counterparty risk for the execution of the swap leg of the trade, ancillary costs to the trade, and limits to arbitrage—which make holding the trade to maturity infeasible. In his article, he suggests that capital increases in required leverage ratios have significantly helped to narrow spreads since 2015. Consider first the balance sheet impact of the long-Treasury leg of the Treasury-interest rate swap trade. Assuming that the trade size is $10 million, the trade increases the Treasury position on the asset side of the balance sheet by $10 million. Since the purchase is repo founded, the value of securities sold under agreements to repurchase on the liabilities side increases by $10 million. Furthermore, the dealer borrows the haircut on the repurchase agreement in short-term funding markets, increasing its short-term debt. At the trade’s inception, the fixed rate is set such that the fair value of the swap is $0, but as the floating rate changes, the market-clearing fixed rate fluctuates as well. Thus, the fair value of IRS on the balance sheet of the dealer paying the fixed rate moves, leading into either an increase in the “Derivatives with a positive fair value” line on the asset side or the “Derivatives with a negative fair value” line on the liabilities side. The gross notional amount of repo financing, initial margin required by the swap (which is reflected in an increase in receivables), repo haircut, and off-balance sheet exposure of the derivative instrument requiring dealers to hold additional equity under the Supplementary Leverage Ratio (SLR) before entering the trade.Boyarchenko’s analysis suggests that, given the balance sheet costs, these spreads must reach more negative levels to generate an adequate ROE for dealers. In fact, the ROE for a given trade is very sensitive to capital charges and has declined because of higher leverage requirements.
Of course, both demand and supply dynamics in the market for interest rate swaps and government bonds may have contributed to the negative spreads observed in recent years. According to Acquilina (2024), this phenomenon signals a pressure in government debt absorption. In cash bond markets, investors’ inability/unwillingness to absorb debt issuance or sales by other bondholders at prevailing prices exerts upward pressure on bond yields, pushing swap spreads lower.
Klingler (2019) offers a demand-driven explanation. Pension funds can balance their asset-liability duration by investing in long-term bonds or by receiving fixed in an IRS with long maturity. They prefer to hedge with IRS rather than buying Treasuries because the swap requires only modest investment to cover margins, whereas buying a government bond requires outright investment. This demand, when coupled with bank-affiliated dealers’ balance sheet constraints, results in negative swap spreads. Jermann (2020) provides instead a supply perspective. In his paper, he develops a model for pricing interest rates swaps that features limited arbitrage and where debt prices are instead exogenous. As it has become more costly for banks to hold Treasuries, dealers have smaller bond positions. With frictions, they are instead less exposed to long-term interest rate risk and therefore they require less compensation for the risk of the fixed swap rate, lowering the swap rate.
3. Regulation of Financial Institutions in the US
The analysis of swap spreads cannot overlook the impact of banking regulations, especially in the United States, where the financial system underwent significant changes after the 2008 crisis to enhance the stability of the banking system.
The Supplementary Leverage Ratio (SLR) is a non-risk-based regulatory capital requirement introduced as part of the Basel III framework (2014) to measure a bank's leverage and ensure it has enough capital to absorb potential losses. It applies to large U.S. banks and globally systemically important banks (G-SIBs):
The Supplementary Leverage Ratio (SLR) is a non-risk-based regulatory capital requirement introduced as part of the Basel III framework (2014) to measure a bank's leverage and ensure it has enough capital to absorb potential losses. It applies to large U.S. banks and globally systemically important banks (G-SIBs):
Where Total Leverage Exposure includes all on-balance sheet assets and certain off-balance sheet exposures. U.S. banks with more than $250 billion in total consolidated assets must maintain an SLR of at least 3%. The eight largest U.S. banks (G-SIBs) must maintain an additional buffer, bringing their effective minimum SLR to 5% at the holding company level and 6% for insured depository institutions. Although we cannot precisely measure the costs SLR requirements impose, it appears that executing swap spread trades is now more expensive for dealers than in the past largely because of the capital dealers must hold against these trades, driven principally by the cash product position rather than the derivatives portion. Dealers that find spread trades to be unprofitable for their own book are also less likely to provide leverage to their clients pursuing the same trades.
The Volcker Rule, a key component of the Dodd–Frank Wall Street Reform and Consumer Protection Act (July 2015), imposed significant restrictions on proprietary trading and certain types of investments by banks, altering the IRS market dynamics.
Furthermore, there is another Basel constraint that influences the phenomenon. The risk weighted capital requirements are similar, but the accounting treatment of the positions is very different. Only the Net Present Value of the swap shows up on the GAAP balance sheet, while the full notional amount of the bond (and the repo) must be disclosed on it.
In recent years, there have been discussions about revising some of these requirements. For instance, the Trump administration expressed intentions to ease capital regulations, including potential adjustments to the SLR, to allow banks more flexibility in holding Treasuries. Such deregulatory efforts are anticipated to impact swap spreads by altering banks' capacity to participate in the swap and bond markets.
The Volcker Rule, a key component of the Dodd–Frank Wall Street Reform and Consumer Protection Act (July 2015), imposed significant restrictions on proprietary trading and certain types of investments by banks, altering the IRS market dynamics.
Furthermore, there is another Basel constraint that influences the phenomenon. The risk weighted capital requirements are similar, but the accounting treatment of the positions is very different. Only the Net Present Value of the swap shows up on the GAAP balance sheet, while the full notional amount of the bond (and the repo) must be disclosed on it.
In recent years, there have been discussions about revising some of these requirements. For instance, the Trump administration expressed intentions to ease capital regulations, including potential adjustments to the SLR, to allow banks more flexibility in holding Treasuries. Such deregulatory efforts are anticipated to impact swap spreads by altering banks' capacity to participate in the swap and bond markets.
4. Data Description
In building our empirical framework, we extracted from Bloomberg daily time series that capture some of the main macro-financial forces potentially driving the 10-year USD swap spread: the CBOE Volatility Index (VIX), spot gold prices, the U.S. Treasury 10-year constant-maturity yield (UST-10Y), the KBW Bank Index (BKX), the S&P 500 Index, the Bloomberg U.S. Corporate High Yield Total Return Index, and the 3-month SOFR-OIS spread.
The VIX represents the market’s forward-looking gauge of implied equity volatility and, as such, is a real-time indicator of risk aversion. Sudden spikes in the index tend to coincide with rising hedging demand and wider risk premia across fixed-income markets.
Gold, conversely, is the archetypal safe-haven asset. A sharp increase in gold prices usually reflects a risk-off sentiment and a flight to safety. Investors shift into Treasuries, pushing their yields lower. At the same time, swap dealers may raise fixed swap rates to reflect higher funding and counterparty risk. The result is a narrower - less negative - swap spread as Treasury yields fall and swap rates remain steady or rise.
Movements in the UST-10Y yield are mechanically linked to the fixed leg of the 10-year interest rate swap. As such, both the level and the slope of the Treasury curve are first-order determinants of the Treasury–swap differential. Shifts in the term structure impact hedging demand and valuations, feeding directly into swap spread dynamics.
The KBW Bank Index tracks the equity performance of major U.S. deposit-taking institutions, offering a real-time readout on their capitalization and funding environment. When the index rises, signalling strength in the banking sector, dealers are able to finance Treasuries more efficiently and hedge swaps at lower cost - contributing to wider (more negative) swap spreads. Conversely, declines in the BKX reflect growing concerns over capital adequacy or liquidity, which can raise repo funding costs and balance sheet constraints. In such environments, swap spreads often tighten.
The S&P 500 Index reflects the broad macroeconomic outlook and expectations for corporate profitability. Sustained rallies typically coincide with improved growth prospects and rising real interest rates, which tend to push Treasury yields higher. As swap rates adjust more gradually in benign funding environments, the result is a widening in swap spreads—that is, spreads become more negative. In contrast, equity market sell-offs often mirror deteriorating economic sentiment, triggering a flight to quality into Treasuries and a sharp decline in yields. At the same time, rising dealer funding costs and risk aversion can elevate swap rates, causing swap spreads to tighten.
The Bloomberg U.S. Corporate High Yield Index measures the total return of U.S. dollar-denominated high-yield corporate bonds. As a barometer of credit risk, movements in this index reflect changes in investor risk appetite and macroeconomic expectations. A widening in high-yield spreads typically coincides with risk-off environments, leading investors to favor safer assets such as Treasuries. This shift puts upward pressure on swap spreads by suppressing Treasury yields while swap rates remain elevated due to increased risk premiums and dealer funding costs.
The 3-month SOFR-OIS spread, calculated as the difference between the Secured Overnight Financing Rate (SOFR) and the Overnight Index Swap rate for the same maturity, is used as an indicator of short-term funding stress. An elevated SOFR-OIS spread often signals increased demand for secured funding, rising counterparty risk, or broader liquidity tensions—especially during episodes of financial instability. In such cases, the spread serves as a forward-looking signal of market dysfunction, typically coinciding with less negative swap spreads due to heightened risk aversion and reduced interbank trust.
The VIX represents the market’s forward-looking gauge of implied equity volatility and, as such, is a real-time indicator of risk aversion. Sudden spikes in the index tend to coincide with rising hedging demand and wider risk premia across fixed-income markets.
Gold, conversely, is the archetypal safe-haven asset. A sharp increase in gold prices usually reflects a risk-off sentiment and a flight to safety. Investors shift into Treasuries, pushing their yields lower. At the same time, swap dealers may raise fixed swap rates to reflect higher funding and counterparty risk. The result is a narrower - less negative - swap spread as Treasury yields fall and swap rates remain steady or rise.
Movements in the UST-10Y yield are mechanically linked to the fixed leg of the 10-year interest rate swap. As such, both the level and the slope of the Treasury curve are first-order determinants of the Treasury–swap differential. Shifts in the term structure impact hedging demand and valuations, feeding directly into swap spread dynamics.
The KBW Bank Index tracks the equity performance of major U.S. deposit-taking institutions, offering a real-time readout on their capitalization and funding environment. When the index rises, signalling strength in the banking sector, dealers are able to finance Treasuries more efficiently and hedge swaps at lower cost - contributing to wider (more negative) swap spreads. Conversely, declines in the BKX reflect growing concerns over capital adequacy or liquidity, which can raise repo funding costs and balance sheet constraints. In such environments, swap spreads often tighten.
The S&P 500 Index reflects the broad macroeconomic outlook and expectations for corporate profitability. Sustained rallies typically coincide with improved growth prospects and rising real interest rates, which tend to push Treasury yields higher. As swap rates adjust more gradually in benign funding environments, the result is a widening in swap spreads—that is, spreads become more negative. In contrast, equity market sell-offs often mirror deteriorating economic sentiment, triggering a flight to quality into Treasuries and a sharp decline in yields. At the same time, rising dealer funding costs and risk aversion can elevate swap rates, causing swap spreads to tighten.
The Bloomberg U.S. Corporate High Yield Index measures the total return of U.S. dollar-denominated high-yield corporate bonds. As a barometer of credit risk, movements in this index reflect changes in investor risk appetite and macroeconomic expectations. A widening in high-yield spreads typically coincides with risk-off environments, leading investors to favor safer assets such as Treasuries. This shift puts upward pressure on swap spreads by suppressing Treasury yields while swap rates remain elevated due to increased risk premiums and dealer funding costs.
The 3-month SOFR-OIS spread, calculated as the difference between the Secured Overnight Financing Rate (SOFR) and the Overnight Index Swap rate for the same maturity, is used as an indicator of short-term funding stress. An elevated SOFR-OIS spread often signals increased demand for secured funding, rising counterparty risk, or broader liquidity tensions—especially during episodes of financial instability. In such cases, the spread serves as a forward-looking signal of market dysfunction, typically coinciding with less negative swap spreads due to heightened risk aversion and reduced interbank trust.
5. Exploratory Data Analysis of the 10-year USD Swap Spread
The 10-year US swap spread displayed in the graph below shows a clear downward trend beginning in mid-2022, with a pronounced decline accelerating through 2023 and reaching historically low levels by late 2024. One significant event that likely contributed to this movement is the combination of aggressive Federal Reserve tightening and the liquidity stress episodes that followed. In 2022, the Fed began an intense rate hike cycle to combat surging inflation, which not only raised policy rates but also contributed to dislocations in various fixed income markets.
Swap spreads often reflect both credit and liquidity conditions. As the Fed reduced its balance sheet, Treasury market liquidity deteriorated, leading to increased demand for swaps over Treasuries in hedging strategies – putting downward pressure on the spread. Additionally, the collapse of Silicon Valley Bank and broader regional banking stress in early 2023 likely drove swap spreads lower as the market priced in systemic risks and shifted preferences toward cash instruments.
The record-low swap spreads observed in late 2024 may also reflect mounting Treasury supply due to rising deficits and a waning investor base for long-dated government debt, further pressuring the swap-Treasury differential.
Swap spreads often reflect both credit and liquidity conditions. As the Fed reduced its balance sheet, Treasury market liquidity deteriorated, leading to increased demand for swaps over Treasuries in hedging strategies – putting downward pressure on the spread. Additionally, the collapse of Silicon Valley Bank and broader regional banking stress in early 2023 likely drove swap spreads lower as the market priced in systemic risks and shifted preferences toward cash instruments.
The record-low swap spreads observed in late 2024 may also reflect mounting Treasury supply due to rising deficits and a waning investor base for long-dated government debt, further pressuring the swap-Treasury differential.
Given that the swap spread series was non-stationary, we focused our analysis on its log return series. This transformation allowed us to work with a stationary time series, enabling statistically meaningful analysis and inference.
|
|
The exploratory data analysis of the 10-year swap spread log returns reveals a slight negative drift, with a mean of -0.0007 and a standard deviation of 0.0358, reflecting moderate daily variability. The distribution is moderately right-skewed (skewness = 0.37), suggesting a higher likelihood of large positive moves. An excess kurtosis of 6.60 points to a leptokurtic distribution - one with a sharp peak and heavy tails. This is confirmed by the Jarque-Bera statistic of 2043.71 and a p-value of 0.0, strongly rejecting the null hypothesis of normality. A histogram with a kernel density overlay, compared to a Gaussian distribution with matching mean and variance, further highlights the non-normal nature of the data. The observed distribution displays a pronounced peak and fatter tails, implying a higher probability of extreme outcomes than under a normal distribution. Overall, the returns exhibit clear signs of asymmetry and excess tail risk.
6. Mean Modelling
To model the short-run dynamics of the swap spread log returns, we estimated a range of ARMA(p,q) models. Model selection was guided by the Akaike Information Criterion, aiming to identify the most parsimonious specification with the best in-sample fit. Among the 25 candidate models, the ARMA(1,1) emerged as the optimal specification, yielding the lowest AIC value of -4280.39. This result suggests that the log returns are best described by a process featuring both an autoregressive and a moving average component, each of order one.
Diagnostic checks on the ARMA(1,1) residuals confirmed the adequacy of the model. The residuals showed no significant autocorrelation and conformed to white noise behavior, supporting the specification’s validity. Overall, the ARMA(1,1) model provides a robust statistical foundation for capturing the mean behavior of the swap spread return series.
Diagnostic checks on the ARMA(1,1) residuals confirmed the adequacy of the model. The residuals showed no significant autocorrelation and conformed to white noise behavior, supporting the specification’s validity. Overall, the ARMA(1,1) model provides a robust statistical foundation for capturing the mean behavior of the swap spread return series.
The plot above displays the observed versus fitted values of log returns on the swap spread using an ARMA(1,1) model from January 2021 to May 2025. The observed data (blue) and fitted values (red) show a generally close alignment, indicating the model captures the series' dynamics reasonably well. However, the series exhibits notable volatility clustering, suggesting the presence of distinct volatility regimes. Early in 2021, volatility appears elevated with larger swings, potentially reflecting market uncertainty or economic events impacting swap spreads. This high-volatility regime persists intermittently through 2022, with sharp spikes in both positive and negative directions. By mid-2023, the series transitions into a lower-volatility regime, where fluctuations are more subdued, though occasional spikes remain. The ARMA(1,1) model, while effective in tracking the mean behavior, struggles to fully capture these volatility shifts, as seen in some discrepancies between observed and fitted values during high-volatility periods.
7. Correlation Analysis and Linear Regression
We now turn our attention to the macroeconomic variables that may be associated with the 10-year USD swap spread. To explore these relationships, we compute the correlations between the swap spread and the variables outlined in the “Data Description” section.
The correlation results indicate that none of the selected macroeconomic variables show a strong or statistically significant relationship with the 10-year USD swap spread. All correlation coefficients are close to zero, and p-values are well above conventional significance thresholds, suggesting weak linear associations.
This outcome could be driven by several factors. First, the swap spread reflects both credit and liquidity conditions in the swap and Treasury markets. Variables like the VIX or equity returns may only capture broader market sentiment and risk appetite, but not directly the technical drivers of swap spreads. For instance, while a higher VIX often reflects market stress, the 10-year swap spread may remain stable if the stress is concentrated in equities rather than in fixed income or funding markets.
Second, the 10-year maturity specifically reflects long-term expectations and term premia, which may not respond strongly to short-term fluctuations in the variables considered. For example, daily changes in gold prices or the S&P 500 might have limited impact on long-term interest rate swap dynamics.
Third, the SOFR-OIS spread - an indicator of funding stress - also shows minimal correlation. This might suggest that systemic funding tensions have a more direct effect on short-term swap spreads rather than those at the 10-year point, which are influenced more by long-term supply-demand imbalances, hedging activity, and regulatory effects.
Overall, the low correlations suggest that the drivers of the 10-year swap spread may lie outside traditional macroeconomic or market return variables, and may instead be linked to structural or technical forces within the fixed income market.
The regression results confirm what emerged from the correlation analysis: none of the explanatory variables display a statistically significant relationship with 10Y swap spread returns. The R-squared is extremely low, indicating the model explains virtually none of the variance in the dependent variable. All p-values (reported in brackets) are well above conventional significance thresholds, and the coefficients are economically small. These results suggest that short-term movements in the 10Y swap spread are not well captured by the selected macro-financial variables. This may be due to several reasons: first, swap spreads can be driven by technical factors such as market liquidity, dealer balance sheet constraints, or supply-demand imbalances in the Treasury market—elements that are not included in the model. Second, the use of daily data may obscure structural relationships that are only visible at lower frequencies. Lastly, the noise and volatility in financial markets can make it difficult to detect clear relationships over short horizons.
This outcome could be driven by several factors. First, the swap spread reflects both credit and liquidity conditions in the swap and Treasury markets. Variables like the VIX or equity returns may only capture broader market sentiment and risk appetite, but not directly the technical drivers of swap spreads. For instance, while a higher VIX often reflects market stress, the 10-year swap spread may remain stable if the stress is concentrated in equities rather than in fixed income or funding markets.
Second, the 10-year maturity specifically reflects long-term expectations and term premia, which may not respond strongly to short-term fluctuations in the variables considered. For example, daily changes in gold prices or the S&P 500 might have limited impact on long-term interest rate swap dynamics.
Third, the SOFR-OIS spread - an indicator of funding stress - also shows minimal correlation. This might suggest that systemic funding tensions have a more direct effect on short-term swap spreads rather than those at the 10-year point, which are influenced more by long-term supply-demand imbalances, hedging activity, and regulatory effects.
Overall, the low correlations suggest that the drivers of the 10-year swap spread may lie outside traditional macroeconomic or market return variables, and may instead be linked to structural or technical forces within the fixed income market.
The regression results confirm what emerged from the correlation analysis: none of the explanatory variables display a statistically significant relationship with 10Y swap spread returns. The R-squared is extremely low, indicating the model explains virtually none of the variance in the dependent variable. All p-values (reported in brackets) are well above conventional significance thresholds, and the coefficients are economically small. These results suggest that short-term movements in the 10Y swap spread are not well captured by the selected macro-financial variables. This may be due to several reasons: first, swap spreads can be driven by technical factors such as market liquidity, dealer balance sheet constraints, or supply-demand imbalances in the Treasury market—elements that are not included in the model. Second, the use of daily data may obscure structural relationships that are only visible at lower frequencies. Lastly, the noise and volatility in financial markets can make it difficult to detect clear relationships over short horizons.
8. Principal Component Analysis
Building on the findings from the linear regression - where individual macro-financial variables showed limited explanatory power for swap spread returns - we turn to Principal Component Analysis to explore whether latent factors combining multiple variables could reveal a more meaningful structure. The PCA of the explanatory variables provides insight into the underlying structure of the dataset, reducing its dimensionality while preserving key patterns.
PC1 captures a general risk sentiment component, with strong negative loadings on S&P 500 returns, HY Index returns, and the Banking Sector Index. This suggests that PC1 reflects a “risk-on/risk-off” dynamic, where equity and credit markets move together in response to changes in overall market sentiment. The VIX also loads positively on this component, consistent with its role as a fear gauge.
PC2 is dominated by a high positive loading on changes in UST 10Y yields and a strong negative loading on gold returns. This suggests that PC2 may reflect an “interest rate expectations” factor: when yields rise, gold (a non-yielding asset) tends to underperform. The banking index also loads positively, possibly due to banks benefiting from steeper yield curves or rising rates.
PC3 shows large positive and negative loadings on SOFR-OIS and VIX, respectively, highlighting a liquidity or funding stress dimension. A negative correlation between SOFR-OIS (a proxy for funding stress) and risk indicators like the VIX could point to episodes when market volatility is driven by liquidity concerns.
PC4 is most heavily influenced by gold (strongly negative) and changes in UST 10Y yields (also negative), indicating another angle of interest rate and safe-haven interaction.
Higher components (PC5–PC7) are harder to interpret economically but may represent idiosyncratic or technical variations across variables.
Overall, the PCA reveals overlapping macro-financial dimensions—such as risk sentiment, rate expectations, and liquidity, but these latent factors fail to significantly explain swap spread returns, likely due to the influence of more specific market forces beyond this macro variables.
PC1 captures a general risk sentiment component, with strong negative loadings on S&P 500 returns, HY Index returns, and the Banking Sector Index. This suggests that PC1 reflects a “risk-on/risk-off” dynamic, where equity and credit markets move together in response to changes in overall market sentiment. The VIX also loads positively on this component, consistent with its role as a fear gauge.
PC2 is dominated by a high positive loading on changes in UST 10Y yields and a strong negative loading on gold returns. This suggests that PC2 may reflect an “interest rate expectations” factor: when yields rise, gold (a non-yielding asset) tends to underperform. The banking index also loads positively, possibly due to banks benefiting from steeper yield curves or rising rates.
PC3 shows large positive and negative loadings on SOFR-OIS and VIX, respectively, highlighting a liquidity or funding stress dimension. A negative correlation between SOFR-OIS (a proxy for funding stress) and risk indicators like the VIX could point to episodes when market volatility is driven by liquidity concerns.
PC4 is most heavily influenced by gold (strongly negative) and changes in UST 10Y yields (also negative), indicating another angle of interest rate and safe-haven interaction.
Higher components (PC5–PC7) are harder to interpret economically but may represent idiosyncratic or technical variations across variables.
Overall, the PCA reveals overlapping macro-financial dimensions—such as risk sentiment, rate expectations, and liquidity, but these latent factors fail to significantly explain swap spread returns, likely due to the influence of more specific market forces beyond this macro variables.
9. Conclusion
This article investigated the dynamics of swap spreads with a particular focus on the puzzling phenomenon of negative swap spreads. The first section offered an overview of the swap market, discussed the mechanics of interest rate swaps, and explored the emergence of negative swap spreads following the 2008 financial crisis. We reviewed the main explanations from the literature, highlighting the role of post-crisis banking regulation, balance sheet constraints, demand imbalances, and limits to arbitrage. In the second part, we turned to quantitative analysis. We began with an exploratory data analysis to understand the historical behavior of swap spreads and their macro-financial correlates. We then estimated a simple mean model to establish a benchmark for prediction. Following this, we ran a linear regression using macro-financial variables to assess their explanatory power. Lastly, we performed a principal component analysis to identify underlying latent factors and assess whether these broader market forces help explain movements in swap spreads.
These insights are not only of academic interest but also increasingly relevant in today’s regulatory context. A major but under-discussed development is the potential reform of the SLR, as hinted at by Fed Vice Chair Michelle Bowman. By potentially exempting Treasuries from SLR calculations, such a reform could release trillions in balance sheet capacity for U.S. banks—making them major marginal buyers of government bonds.
If enacted, this structural shift would fundamentally change the supply/demand dynamics for Treasuries, likely affecting yield levels and curve behavior. In such a context, the analysis presented in this article offers valuable insight into macroeconomic dynamics that are set to become increasingly critical in the evolving market and regulatory landscape.
These insights are not only of academic interest but also increasingly relevant in today’s regulatory context. A major but under-discussed development is the potential reform of the SLR, as hinted at by Fed Vice Chair Michelle Bowman. By potentially exempting Treasuries from SLR calculations, such a reform could release trillions in balance sheet capacity for U.S. banks—making them major marginal buyers of government bonds.
If enacted, this structural shift would fundamentally change the supply/demand dynamics for Treasuries, likely affecting yield levels and curve behavior. In such a context, the analysis presented in this article offers valuable insight into macroeconomic dynamics that are set to become increasingly critical in the evolving market and regulatory landscape.
Written by: Edoardo Bini, Bianca Proverbio, Alberto Cocirio, Lorenzo Monteduro