实时全球研报
Primer: swap spread frameworks: Swap spreads frameworks for relative value
研报英文原文证据摘录
Accessible version
US Rates
Primer: swap spread frameworks
Swap spreads frameworks for relative value
31 July 2026
We describe four swap spread frameworks. Our focus is UST spreads to swaps, but the
frameworks can apply across other currencies. We look at: 1) traditional regression
(commonly used spread factors), 2) PCA (factors with maximum explanatory power), 3)
term-premium (short-term factors), 4) dynamic factor model (growth + inflation + jobs
factors). The residuals can bring a dislocation that corrects, or may remain as a regime
shift, unexplained by the model’s factors. As linear models, the frameworks apply to
swaps and Treasury rate curves separately or directly to spreads with the same results.
Exhibit 1: UST swap spreads currently appear fair in most frameworks but the Macro one
Results for UST swap spreads as of 30-Jul – residuals based on the four frameworks
Framework
Traditional regressions
PCA
Term premium
Macro factors
2y
fair
5bp too tight
5y
fair
2bp too tight
14bp too tight
10y
fair
1bp too tight
16bp too tight
30y
3bp too wide
fair
1bp too tight
19bp too tight
Source: BofA Global Research
BofA GLOBAL RESEARCH
Traditional regressions: user factors
We look at some typical regression variables like USTs outstanding, level of rates,
slopes, vol, Fed holdings etc. US Swap spreads recently are explained by dealer balance
sheets, funding conditions, rate levels and vol. Regressions allow easy exploration of
spread drivers. Today, we find 2y spreads fair and 30y spread 3bp too wide (Exhibit 1).
PCA framework: statistical factors
PCA is a statistical method for extracting factors from historical data. This avoids trial
and error process of regression. The algorithm finds 2 uncorrelated factors (time series)
that maximally explain data variance in the sample period. For rates and spreads, PCA
finds “level” and “slope” factors which best re-create the spread curve across maturities
each day. Currently, spreads appear fair across the curve, with near zero residuals.
Term premium framework: market factors
This framework again uses 2 factors, but they are tradable points selected at the front
…
本摘录由系统从所标注的 PDF 证据页直接提取并保留英文原文,不做批量翻译;登录后在阅读器切换中文时才按需翻译。
打开研报阅读器