Friday, February 13, 2009

Credit Markets Surprisingly Efficient At Pricing In Defaults (Or Not)

The planets must be in some millennial Friday 13th alignment, as it is finally the case that the high yield credit market is in agreement with rating agency default predictions. As we observed previously the price of the on-the-roll High Yield index (HY10) implies a specific curve of cumulative default probabilities. Based on its most recent price of 76.4, the interpolated December 31, 2009 cumulative probability for index constituents is 18.4%, which is shockingly close to the latest prediction out of S&P for speculative grade defaults.

S&P estimates high-yield-bond default rates will hit 13.9%this year, but could go as high as 18.5% if the downturn is worse than expected (which it will be). Moody's predicts a default rate around 16.4% this year. The default rates in recent downturns were 11.9% in 1991 and 10.4% in 2002, according to S&P. Such rates peaked at around 15% in 1930. In 2007, when credit flowed freely, the default rate remarkably dipped below 1%, an all-time low. Fitch Ratings estimates the biggest prior year for high-yield bond defaults, in dollar terms, was 2002, when $109.8 billion defaulted. In 2008, high-yield bond defaults topped $66.6 billion, up from $9 billion for 2006 and 2007 combined. Both rating agencies expect defaults to peak in the second half of the year, based on forecasts that the economy itself will bottom out in the first two quarters of the year.
As the chart below indicates, the HY10 price also demonstrates that based on JPM's mathematical models, by March of 2019, 90.5% of all current high yield companies will be bankrupt. So whoever is buying a basket of high yield bonds with 10 year maturities, may want to change recovery model assumptions so that only 10% of the companies in that basket will pay off their debt at maturity.

Now reader CreditTrader makes a terrific point. The chart above assumes 40% recovery rates which is the default assumption, yet wrong, as recent CDS auctions have shown: a much more realistic recovery assumption in a bankruptcy these days is 10-15%. If we rerun the numbers assuming 20% recovery rates (or half the default rate), the HY10 market price implies credit is mispricing default risk significantly! In other words to get to an 18.4% default rate at EOY 2009 using 20% recoveries, the HY10 price has to be 62, or about 14 points lower than today's price, meaning HY credit is very very rich, and all those who are buying bonds in the primary market will get burned very soon. As more defaults result in minuscule recoveries, it is inevitable that JPM will tweak their recovery default rate much lower and this will result in a huge drop in the index trading price.

Sphere: Related Content
Print this post


CreditTrader said...

Hey Tyler, great post - i have been chatting with a client about this today. People talk about how nuts and how bad an environment the spreads imply BUT if you use realistic recoveries, these rates dont make HY look cheap at all - especially in an ugly turn-down...been looking at a 5s10s steepener on IG11 recently - ovr 210bps carry DV01-neutral, covers around 5 defaults this year and 33bps further inversion - feels good to me even though I am a big bear.

Tyler Durden said...

great point. adjusted the post

Anonymous said...

What do you guys think about default rates and recovery rates for first lien loans?

Anonymous said...

How do you calculate default probabilities given a certain spread ? Is it a proprietary model by JPM ?

Tyler Durden said...

jpm has the most widespread prop model which is in the process of becoming open sourced with the transition to a cds clearinghouse. also allegedly it is full of mistakes. the math behind the model is widely available and there is any number of scholarly papers available on google discussing the relationship.