Rating Migration: a Worked Example

A business rarely goes from sound to defaulted in one review. It slips: a covenant gets tight, a customer is lost, a year's results disappoint, and at each step its banker moves the risk grade down a notch. Across a book of a few hundred borrowers, a handful may default in a year, too few to read a trend from. Dozens will change grade. Migration is therefore the statistic that tells a credit portfolio manager the direction of the book while there is still time to act on it.

Capital rules make it a standing report. The UK's retained Capital Requirements Regulation and Australia's APS 113 both require banks that use internal ratings for capital to report migration across grades to the board and senior management, alongside the risk profile by grade and a comparison of realised default rates with expectations. APS 113 also requires those banks to keep the data on rating migrations behind each grade.

The usual method follows a cohort. Take every borrower rated at the start of the year, note its grade at the end, and count. Each row is a starting grade; each column is where those borrowers finished, with a column for default. The diagonal holds the borrowers that stayed put. Below the diagonal sit upgrades, above it downgrades, and the further from the diagonal, the more notches the move covered.

Several choices change what the matrix says. Counting borrowers treats a small file like a large one; weighting by exposure shows where the money moved. Borrowers that repaid or left during the year drop out of the cohort, and a book whose best names refinance elsewhere can look as though its stayers deteriorated. And the bank's rating philosophy matters: a grade built to reflect current conditions moves with the economy, while one built to hold through a cycle moves less, mainly on changes specific to the borrower or its industry. APS 113 expects rating systems to be designed so that borrower and industry changes drive migration, with the business cycle also a possible driver.

A bank's book of 150 business borrowers, constructed for this page like every figure in it, is graded in four bands from A, the strongest, to D, the weakest. A year later the cohort looks like this:

The average barely moved, partly because the five defaults left the rated book and took their bad grades with them. The counts tell a different story: nearly two downgrades for every upgrade, a weak band that grew from 20 to 22 borrowers while sending four into default, and nine borrowers newly in D. The bank's own forecasts tell a third:

Band D defaulted at more than three times its forecast. One year is a small sample, and the probabilities are the example's, not a market calibration, but a portfolio manager would ask whether the bad year was the economy, a sector, or a grade scale that no longer separates its weak borrowers from its failing ones.

The default rate counts the files that have already failed; migration counts the ones heading that way. The two are linked, since today's downgrades into the weakest band are next year's defaults, but they answer different questions. A falling default rate alongside rising net downgrades is a book that has been lucky so far. A rising default rate with stable migration is often a few concentrated failures rather than a general slide.

Migration also differs from a change in probability of default estimates. A migration is a borrower moving between grades; a recalibration is the bank changing the PD attached to a grade. Both raise expected loss, but the first is information about the borrowers and the second about the bank's model, and a reader of a bank's results should know which one drove the number.

DEFINED IN ONE PARAGRAPH → · EXPECTED LOSS, WORKED → · RISK GRADE, DEFINED →

Each downgrade starts as one banker's call on one file, and it carries consequences beyond that file. Under IFRS 9 and its NZ and Australian equivalents, an internal downgrade is one of the listed signs that credit risk may have increased significantly, so a move into a weak band often moves the loan into stage 2 and a lifetime allowance. Where the bank is approved to use internal ratings for capital, the higher probability of default raises the capital the loan uses. Under the standardised approach internal grades do not set the risk weight directly, so migration shows up first in provisions, pricing and appetite.

At portfolio level, migration is the trigger for action rather than a report to file. A manager who sees one sector producing most of the downgrades can tighten new lending there, ask for earlier reviews and early warning checks across that sector, or bring forward reviews of the files just above the weak band. The question the example's matrix poses is not whether the average is acceptable. It is which of the 22 files in band D the bank still wants to hold in a year.

HOW BANKS MANAGE CREDIT RISK → · THE EXAM QUESTION ON IT → · RISK-WEIGHTED ASSETS, EXPLAINED →