Parallel vs Non-Parallel Shifts
Parallel vs Non-Parallel Shifts
When yields move, they rarely move in lockstep across all maturities. Understanding the difference between uniform (parallel) and shape-changing (non-parallel) shifts reveals why duration alone is incomplete and why successful bond managers monitor curve positioning.
Key takeaways
- Parallel shifts move all yields by the same amount; duration captures nearly all the risk.
- Steepening reduces returns for long-duration portfolios relative to the yield increase alone.
- Flattening boosts returns for long-duration portfolios relative to the yield increase alone.
- Historical data shows parallel shifts explain 80–90% of yield curve variation; shape changes explain 10–20%.
- Portfolio construction should account for both parallel and non-parallel risk.
Parallel Shifts Defined
A parallel shift is a yield curve movement where every maturity's yield changes by exactly the same amount. If the 2-year yield rises 1%, the 5-year rises 1%, the 10-year rises 1%, and the 30-year rises 1%, the curve has shifted up in parallel.
In a parallel shift, the curve's shape is unchanged. If the curve was upward-sloping by 100 basis points (10-year yield 100 basis points higher than 2-year), it remains upward-sloping by 100 basis points after a 1% parallel rise.
Duration is perfectly calibrated for parallel shifts. A portfolio with 6-year average duration loses 6% when yields rise 1% in parallel. A portfolio with 5-year average duration loses 5%. The maturity composition of the portfolio—whether it's a ladder across all maturities, a bullet concentrated in 10-years, or a barbell split between short and long—does not affect the return in a parallel shift.
This is why duration is so powerful and widely used. In practice, parallel shifts account for 80–90% of daily and monthly yield curve movements. If duration explained all the risk, bond management would be trivial.
Why Parallel Shifts Dominate
Parallel shifts dominate because they reflect changes in the overall level of real interest rates and inflation expectations. When the Fed raises rates or inflation accelerates, all parts of the curve tend to move up together. When the Fed cuts rates or inflation expectations fall, the entire curve shifts down.
In early 2022, the Fed's rapid rate-hike campaign caused a broad-based rise in yields. The 2-year Treasury rose from 0.7% to 2.2% (1.5% move). The 10-year rose from 1.5% to 2.9% (1.4% move). The 30-year rose from 1.9% to 3.3% (1.4% move). These moves were roughly parallel. A portfolio manager holding a broad bond index fund (BND, AGG) with average duration 5.5 years could have predicted the loss quite accurately: 5.5 × 1.4% ≈ 7.7%. The actual loss was similar.
In contrast, when the Fed paused rate hikes in mid-2023 and the market began pricing rate cuts for later that year, yields fell across the board, but short-term yields fell less than long-term yields. This was a flattening (non-parallel). A portfolio with long duration outperformed one with short duration, beyond what the average yield change would suggest.
Steepening: Long Rates Fall, Short Rates Rise
In a steepening, the curve gets steeper. This typically means short-term yields rise (or don't fall as much) while long-term yields fall (or don't rise as much). The 2–10 year spread widens.
Consider a starting curve:
- 2-year: 4.0%, 10-year: 4.5%, spread: 0.5%
A steepening move:
- 2-year: 4.5%, 10-year: 4.3%, spread: 1.2% (widened by 0.7%)
Short-duration bonds (concentrated in 2–3 years) suffer in a steepening. Their yields rise.
Long-duration bonds (concentrated in 10–30 years) benefit. Their yields fall.
A portfolio of short-duration bonds with average duration 2.5 years would normally expect to lose 2.5 × 0.5% = 1.25% if the 2-year yield rises 0.5%. But it experiences the full loss because short rates are what went up.
A portfolio of long-duration bonds with average duration 10 years would normally expect to lose 10 × 0.3% = 3% if rates fall overall. But since it's the long-term yield falling (by 0.2%), the loss is only 10 × 0.2% = 2%, and some of that might be offset by convexity. Actually, the portfolio gains, because long-dated bonds benefit from the 0.2% yield decline.
In the most extreme steepening—like 2008–2009 or 2020—the short end is at zero or near-zero (Fed policy floor), while long rates fall dramatically. Investors holding long-duration bonds (TLT, VGLT) made spectacular returns. Investors holding money market funds (near-zero duration) made almost nothing.
Flattening: Long Rates Rise, Short Rates Fall
In a flattening, the curve becomes flatter. Short-term yields fall (or don't rise as much) while long-term yields rise (or don't fall as much). The 2–10 year spread narrows.
Consider a starting curve:
- 2-year: 4.0%, 10-year: 4.5%, spread: 0.5%
A flattening move:
- 2-year: 3.5%, 10-year: 4.2%, spread: 0.2% (narrowed by 0.3%)
Short-duration bonds benefit in a flattening. Their yields fall.
Long-duration bonds suffer. Their yields rise.
A portfolio of short-duration bonds with average duration 2.5 years would expect to gain 2.5 × 0.5% = 1.25% from the 0.5% yield decline. It gets this gain because short rates are what fell.
A portfolio of long-duration bonds with average duration 10 years would expect to lose 10 × 0.3% = 3% from the 0.3% yield increase.
Flattening cycles usually precede economic slowdowns or recessions. As growth concerns mount, investors flee short-term credit risk (seeking safety in bonds) and expect the Fed to eventually cut rates, so short-term yields fall while long-term yields fall more slowly or rise. A barbell portfolio (short and long, skipping middle maturities) or a portfolio positioned heavily in long bonds would have captured significant outperformance during the 2018–2019 flattening. A portfolio holding only short-duration bonds would have underperformed.
Real-World Example: 2022
The 2022 fixed income bear market illustrates both parallel and non-parallel dynamics.
The parallel shift: The 10-year Treasury yield rose from 1.5% to 3.88%, a 2.38% move. This alone caused broad bond losses. BND, with duration 5.5, lost approximately 5.5 × 2.38% = 13.1%. Actual loss: 13.0%. Duration worked.
The non-parallel shift: Early in 2022, the curve steepened significantly (short rates rose faster than long rates), which partially offset long-bond losses. TLT, with duration 17, should have lost 17 × 2.38% = 40.5%. Actual loss: 29.7%. The steepening saved long bonds about 10 percentage points.
But this benefit only partially offset the parallel move damage. By mid-year, the Fed paused, and the curve flattened, further hurting long-duration bonds. The combined effect—large parallel rise plus flattening—created the worst year for long-duration bonds since records began.
Measuring Curve Shifts
Professional managers decompose every daily or monthly yield curve movement into parallel and non-parallel components:
- Parallel component (PCA1): How much does the level of all yields move?
- Slope component (PCA2): How much does the curve flatten or steepen?
- Twist component (PCA3): Do some middle maturities move differently?
Using this decomposition, a manager might observe:
"The yield curve moved up 0.75% (parallel shift), then steepened by 0.25% (long yields rose only 0.5% instead of 0.75%). My portfolio with average duration 6 years and concentrated in 10-year bonds did better than a broad index with similar duration, because the steepening helped long bonds."
This is how professional managers explain performance and adjust positioning.
Strategic Implications
If you believe the Fed will soon pivot to rate cuts (typically associated with flattening or downward curve movement), positioning a portfolio toward longer maturities makes sense. Long bonds benefit from both the parallel yield decline and the flattening itself.
If you believe the Fed will stay tight and hike further (typically associated with steepening early in a cycle), short-duration bonds might outperform, because short rates rise faster than long rates, limiting losses from duration.
If you have no conviction on curve direction, holding a duration-neutral, maturity-neutral broad bond index (BND, AGG) ensures you're not exposed to curve positioning bets—you get duration risk and credit risk, but not curve tilts.
Decision Tree
Next
We've examined how the overall curve shape changes. But different parts of the curve respond differently to different economic and policy signals. Key rate duration isolates sensitivity to specific maturities, revealing where true risk lives in a portfolio.