2026-09-22
The Safe Half of the Portfolio Has Been Underwater for Six Years
By FinScope Research · Data as of 2026-09-18
Drawdown analysis is almost always pointed at the risky side of a portfolio. The bond sleeve is where you are supposed to stop worrying. So we pointed the same measurement at it, and the result inverts the usual story: in this dataset the equity side took a far deeper hit and fully repaired it, while a large part of the bond side is still below the high it set in 2020.
The FinScope universe contains 17 bond ETFs. We measured each one the same way we measure a stock: largest peak-to-trough decline on split- and dividend-adjusted closes, and calendar days from the trough back to the old high. Because the prices are dividend-adjusted, every figure below is after reinvested income — six years of coupons are already counted in.
Six of seventeen have never recovered
| Ticker | Fund | Max drawdown | Now vs peak | Peaked | Days since peak |
|---|---|---|---|---|---|
| EDV | Vanguard Extended Duration Treasury | −60.0% | −57.0% | 2020-03-09 | 2,384 |
| TLT | iShares 20+ Year Treasury Bond | −48.4% | −42.6% | 2020-08-04 | 2,236 |
| IEF | iShares 7-10 Year Treasury Bond | −23.9% | −13.5% | 2020-08-04 | 2,236 |
| LQD | iShares Investment Grade Corporate | −25.0% | −6.1% | 2021-09-22 | 1,822 |
| BND | Vanguard Total Bond Market | −18.6% | −3.9% | 2020-08-06 | 2,234 |
| AGG | iShares Core U.S. Aggregate Bond | −18.4% | −3.7% | 2020-08-04 | 2,236 |
Look at the “peaked” column. AGG, IEF and TLT set their all-time highs on the same day, 2020-08-04, and BND two days later. That is not a coincidence of three separate stories; it is one event. Long-term US yields bottomed, and every fund holding duration peaked together.
2,236 days later — six years and one month — none of them has made it back.
For AGG and BND, the core “total bond market” holdings that sit in the default position of most target-date and 60/40 portfolios, the remaining gap is modest: about −3.7% and −3.9%. But it is a gap that has persisted for six years with income reinvested. The 4% or so those funds have paid out along the way did not cover the price decline. For TLT the gap is −42.6%, and for EDV it is −57.0%, which are equity-sized losses from an asset class bought for the opposite reason.
Meanwhile the risky side already repaired itself
| Ticker | Max drawdown | Recovery time | Now vs peak |
|---|---|---|---|
| SPY | −55.2% | 1,256 days | −1.8% |
| VTI | −55.5% | 1,100 days | −2.3% |
| IWM | −58.6% | 701 days | −6.6% |
| QQQ | −83.0% | 4,517 days | −3.2% |
SPY fell 55% in the financial crisis, three times AGG’s worst decline, and took 1,256 days to get back. It is now within 2% of its high. TLT fell less than half as far and has been underwater almost twice as long.
This is the part worth sitting with. The comparison is not “bonds are worse than stocks.” It is that the two drawdowns have completely different shapes: equities took a violent, deep, mean-reverting hit, and bonds took a shallower hit that has not reverted at all, because the thing that caused it — a repricing of interest rates from near zero — has not reversed either.
Duration, not credit, is what stayed broken
Sort the same 17 funds by maturity and the pattern is almost mechanical.
| Fund | Exposure | Max drawdown | Recovered? |
|---|---|---|---|
| SGOV | 0-3 month T-bills | −0.0% | Yes, 2 days |
| BIL | 1-3 month T-bills | −0.8% | Yes, 730 days |
| SHY | 1-3 year Treasuries | −5.7% | Yes, 592 days |
| AGG | Broad US aggregate | −18.4% | No |
| IEF | 7-10 year Treasuries | −23.9% | No |
| TLT | 20+ year Treasuries | −48.4% | No |
| EDV | Extended duration | −60.0% | No |
Every step out the maturity ladder roughly doubles the damage. “Government bonds” is not one risk level; SGOV and EDV hold the same issuer’s credit and differ by a factor of about 1,500 in maximum drawdown. The default-risk-free label applies to both. The price-risk label does not.
The credit-sensitive funds make the contrast sharper, because they went the other way:
| Fund | Exposure | Max drawdown | Now vs peak |
|---|---|---|---|
| JNK | US high yield | −38.5% | −1.2% |
| HYG | US high yield | −34.2% | −1.2% |
| EMB | EM USD sovereign | −34.7% | −2.3% |
HYG and JNK took their worst losses in the 2008 credit crisis, deeper than anything the Treasury funds saw, and both are now within about 1% of their all-time highs. Credit risk showed up violently and then healed. Duration risk showed up gently and stayed.
If you hold bonds for diversification, that distinction is the whole ballgame, and it is invisible in a single “bond fund” line on a statement.
What this does and does not say
Some honest limits on the above, because this is a measurement, not a forecast:
- This is price history, not a yield argument. A bond fund that fell because rates rose now holds higher-yielding paper. Investors buying TLT today face a completely different starting yield than in 2020. Drawdown measures what happened to an existing holder; it does not price the forward return.
- “Never recovered” is a statement about a specific peak. It means the adjusted price has not reclaimed its 2020-2021 level. It does not mean the fund is impaired or that the holding was a mistake.
- Individual bonds held to maturity behave differently. A fund is perpetually rolling its maturity, so it has no maturity date at which the price is pulled back to par. That structural difference is precisely why a fund can stay underwater for six years while its underlying bonds keep maturing at 100.
- These are 17 funds, not the bond market. The universe is US-listed and skewed to large, liquid ETFs.
What to check in your own portfolio
- Find your bond sleeve’s duration, not just its name. If it is “total bond market,” the measured worst case here was about −18% and six years without a full recovery.
- Do not assume the safe sleeve recovers faster. In this dataset it has been the opposite: broad equity funds repaired a 55% drawdown while broad bond funds have not repaired an 18% one.
- Separate the two risks. Credit-sensitive funds recovered after their worst episodes; duration-heavy funds have not. If you own bonds as an equity hedge, you are mostly buying duration, which is the piece that has stayed underwater.
Each fund’s full underwater history is on its FinScope page — AGG, BND, TLT, IEF, SGOV — and you can put a bond fund and a stock fund on the same dates with the comparison tool. For how recovery is defined and why an unrecovered drawdown is not the same as a permanent loss, see the recovery guide and the limits of drawdown.
FAQ
Have bond ETFs recovered from the 2022 bond crash? Mostly not, at the long end. Six of the 17 bond ETFs in this universe — EDV, TLT, LQD, IEF, BND and AGG — have never reclaimed their prior peak. AGG, IEF and TLT all peaked on 2020-08-04 and were still below it 2,236 days later as of 2026-09-18, on a dividend-adjusted basis.
How bad was the worst bond ETF drawdown? EDV, an extended-duration Treasury fund, fell 60.0% from its peak and remains 57.0% below it. TLT fell 48.4%. Both are equity-scale losses from funds holding US government debt, driven by interest-rate duration rather than default risk.
Did stocks recover faster than bonds? In this data, yes. SPY’s worst drawdown was 55.2% and it recovered in 1,256 days, now sitting 1.8% below its high. AGG’s worst was 18.4% and it has not recovered after more than six years.
Methodology: 17 bond ETFs identified with the same name-based classifier the FinScope screener uses (classifyAsset). All figures use split- and dividend-adjusted daily closes, so reinvested income is included. Maximum drawdown is the largest peak-to-trough decline in the fund’s available history; “never recovered” means the adjusted close has not reclaimed that prior peak as of 2026-09-18. Peak dates are the all-time high within each fund’s available series. See the methodology page and the editorial policy.