What happens to a bond when its rating changes?
Default is the dramatic outcome, and it is rare. The far more common credit event in any portfolio is a rating migration — a bond moving up or down the scale while continuing to pay perfectly well. Migration is where most credit profit and loss actually happens.
The transition matrix
Agencies publish transition matrices: a grid where each row is a starting rating, each column a rating one year later, and each cell the historical percentage of issuers that made that move.
An illustrative, rounded row for a triple-B issuer over one year:
- Upgraded: 5%
- Unchanged: 90%
- Downgraded one category: 4%
- Downgraded further or defaulted: 1%
Two properties hold across essentially every published study:
- The diagonal dominates. Most issuers stay where they are in any given year. Ratings are sticky by design.
- Stickiness falls as you descend. High ratings move rarely; low ratings move a lot, in both directions, and the downward moves increasingly end in default.
A third, subtler property is well documented: downgrade momentum. An issuer that was downgraded recently is historically more likely to be downgraded again than an issuer sitting at the same rating that arrived there without a recent move. Direction carries information beyond the level.
Why migration costs money without any default
Here is the arithmetic that makes migration matter to a holder. Spread changes move price roughly in proportion to spread duration — the sensitivity of price to a change in spread, measured in years:
Price change ≈ − spread duration × change in spread
An illustrative bond with a spread duration of 6 whose spread widens 100bp on a downgrade:
- Price change ≈ −6 × 1.00% = −6%
Nobody defaulted. Every coupon was paid on time. The bond simply moved to a rating category the market prices 100bp wider, and a holder marked to market lost 6% of the position's value. If the annual spread income was 300bp, that single migration erased two years of carry.
Multiply by the boundary effect from the last lesson: a downgrade that crosses from investment grade into high yield can move spread far more than 100bp, because the buyer base changes at the same moment.
A downgrade also changes cash flows written into the issuer's own contracts. Many bonds from issuers rated in the triple-B area, and most corporate hybrids, carry a coupon step-up: the coupon rises by a stated amount per notch below a set level, commonly 25 basis points, and steps back down on upgrade. Derivative and supply contracts often carry collateral triggers that oblige the issuer to post cash when its rating falls, which drains liquidity when funding is hardest; those collateral calls, not the mark-to-market, forced the rescue of AIG in September 2008. One notch can raise the interest bill and the cash calls at once, and the size sits in the documentation, not in the spread.
The other direction
Upgrades work symmetrically. A bond whose spread tightens 60bp with a spread duration of 6 gains roughly 3.6% in price, on top of the coupon. Credit is not a one-sided instrument; it simply has an asymmetric shape, because the upside is bounded by par and the downside runs to the recovery value.
What this does to how you hold credit
Because migration dominates default in frequency, credit portfolios are usually monitored on rating direction and spread behaviour, not only on solvency. The practical questions are: which way is this credit drifting, what would move it a notch, and what happens mechanically if it crosses the investment-grade line?
Those are questions to investigate, never conclusions. A transition matrix is a summary of what happened historically to a population of issuers. It does not tell you what happens next to any single one.
In the data
A table of current ratings cannot show a migration. The table is everything the record holds for one country: three letters and the date they were collected.
When an agency downgrades, the new letter replaces the old one; it does not appear beside it. The day the agency acted, the outlook change that preceded it, the pace of the move and the previous rating are all absent. Anything that needs the timing of a downgrade rather than the fact of one has to come from the agency's own rating history.
Try it now
Look at migration in the data:
- Confirm the limitation rather than taking it. Go through the record in the section above line by line, looking for the previous rating, the date of the change and the outlook that preceded it. Count how many of those three you find.
- Then try the price instead. The market's price of protection on the same country is below, also one annual reading per country: read the date, which is the same for every country. It dates a level and never a move. Neither table holds the day the market repriced.
- Now price a downgrade across the investment-grade line. The lookup below gives the spread typically charged to the Baa3 bucket and to Ba1, one notch lower, as fractions. Take the gap between the two in basis points as the spread change, and estimate the price impact using price change ≈ − spread duration × spread change, assuming a spread duration of 5. That figure is the cost of a one-notch downgrade to a holder. Then set it beside the lesson's point about the boundary: the lookup is a smooth table, and a real crossing can move far more because the buyer base changes.