Paper: arXiv 2410.22498
Authors: Jihyun Park, Andrey Sarantsev
Abstract
Classic stochastic volatility models assume volatility is unobservable. We use the Volatility Index: S&P 500 VIX to observe it, to easier fit the model. We apply it to corporate bonds. We fit autoregression for corporate rates and for risk spreads between these rates and Treasury rates. Next, we divide residuals by VIX. Our main idea is such division makes residuals closer to the ideal case of a Gaussian white noise. This is remarkable, since these residuals and VIX come from separate market segments. Similarly, we model corporate bond returns as a linear function of rates and rate changes. Our article has two main parts: Moody’s AAA and BAA spreads; Bank of America investment-grade and high-yield rates, spreads, and returns. We analyze long-term stability of these models.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced statistical models, including autoregressive stochastic volatility frameworks with non-standard assumptions (e.g., observable VIX, dependent innovations), requiring significant mathematical formulation and theoretical proofs of stationarity and ergodicity. Empirically, it is backtest-ready with specific data sources (FRED), multiple bond market segments, code availability on GitHub, and detailed metrics like skewness and kurtosis for model validation.
Research Flowchart
flowchart TD A["Research Goal<br>Use VIX as observable volatility<br>to model corporate bonds"] --> B["Data Input<br>Moody's AAA/BAA &<br>Bank of America IG/HY"] A --> C["Data Input<br>S&P 500 VIX"] B --> D["Methodology<br>Autoregression on Rates & Spreads<br>vs Treasury"] C --> E["Methodology<br>Normalize Residuals<br>Divide by VIX"] D --> F["Computational Process<br>Analyze Residuals<br>Verify Gaussian White Noise"] E --> F F --> G["Key Findings<br>Long-term model stability<br>VIX effective for corporate bonds"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes dividing regression residuals by the VIX to obtain residuals that are closer to independent, identically distributed Gaussian noise, adding a new volatility term cVt to the bond rate model.
Source sentence
Dividing residuals of a linear regression by VIX makes them closer to the ideal case: independent identically distributed Gaussian.
Assumes 6
- VIX (Vt) is observable. model
Source sentence
Assumption (d) makes it much easier to estimate parameters of this model.
- β ∈ (0,1) ensuring mean‑reversion of VIX and bond spreads. model
Source sentence
β ∈(0, 1); this ensures mean-reversion.
- Zt may be correlated with Wt. statistical
Source sentence
As earlier, we allow Zt to be correlated to Wt.
- Wt and Zt could be non‑Gaussian. statistical
Source sentence
Also, Zt and Wt could be non-Gaussian.
Show 2 more
- Innovations W are i.i.d. (though not Gaussian). statistical
Source sentence
The innovations W can be modeled as independent identically distributed random variables.
- E[exp(u Wt)] < ∞ for small enough u. statistical
Source sentence
From the real‑world data, E[euWt] < ∞ for small enough u.
Data
- AAA10Y and BAA10Y spreads macro public
Source sentence
The data is taken from the Federal Reserve Economic Data (FRED) web site, series AAA10Y and BAA10Y, monthly end of month. The data is January 1986 – August 2024.
- VIXCLS series macro public
Source sentence
The Volatility Index is also taken from the FRED web site, series VIXCLS, January 1986 – August 2024, monthly average.
- Bank of America investment‑grade portfolio series macro public
Source sentence
Effective Yield, Option‑Adjusted Spread, and Total Returns Index Value Yt from FRED web site for December 1996 – March 2024, for investment‑grade bond portfolios: respectively, BAMLC0A0CMEY, BAMLC0A0CM, BAMLCC0A0CMTRIV.
- Bank of America high‑yield portfolio series macro public
Source sentence
high‑yield bond portfolios: respectively, BAMLH0A0HYM2EY, BAMLH0A0HYM2, BAMLHYH0A0HYM2TRIV.
Costs
Not applicable
Finds 5
- Dividing autoregression innovations by VIX yields residuals that appear i.i.d. Gaussian. statistical
Source sentence
We conclude that it is reasonable to model residuals as IID Gaussian.
- Both AAA and BAA spreads exhibit mean‑reversion (b ≠ 1). statistical
Source sentence
The rejection of b = 1 is very important: This means that Rt is indeed mean-reverting.
- Log VIX is mean‑reverting with estimated α = 0.347, β = 0.881, and its innovations have heavy tails (skewness 2, excess kurtosis 9). statistical
Source sentence
... the distribution of the innovations W is not Gaussian, but has heavier tails. Computed using (6), the skewness of W is 2, and the excess kurtosis is 9.
- The proposed model (with cVt term) fits well for each rating and is ergodic (stable in the long run). theoretical
Source sentence
The model (9) with IID Gaussian Zt and a = 0 but b ∈(0, 1) seems to fit well for each rating. As we see later, this model is ergodic: Stable in the long run.
- Adding the VIX scaling does not improve alternative regression (16) over (17) for total returns. negative
Source sentence
We see that the latter is not an improvement over the former.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.