There’s been plenty of hurly-burly lately over forecasts of default rates in the municipal bond market.
But what of recovery amounts, should muni bonds default?
Via Cate Long’s Muniland blog, a paper by RockFleet financial services (RF) argues that forecasts of “hundreds of billions” of dollars worth of defaults are both wrong and less relevant to investors than the likely extent of unrecoverable losses. Haircuts, and all that.
The seven-page paper is worth reading in full, subject to some caveats mentioned below, but here’s the top line:
We have concluded that investors may suffer unrecoverable losses in the range of $10-20 billion over the next ten years.
Hmmm. That doesn’t sound like hundreds of billions.
Fortunately, RF’s methodology is clear and provides a snappy overview to the structure of the muni bond market. It starts with the aggregate amount of outstanding municipal debt and, through a process of elimination, estimates unrecoverable losses.
Here’s the process in chart form:
The pie chart on the left depicts the $4trn (approx.) of outstanding muni debt. The one in the middle shows the $1.695trn of bonds that could enter some form of recovery process. The pie chart on the right breaks down the $1.028trn of bonds where only a “partial recovery” could be expected if default were to occur. Of this $1.028trn, only $16bn is forecast to be lost to investors after 10 years.
To get to the pie chart in the middle from the one on the right, RF first assumes the $468bn of prerefunded and escrowed to maturity bonds (“Refunded and ETM”) are safe. It then assumes that holders of the $823bn of revenue bonds covered by monoline insurers (“Rev/COPs – insured”) would be repaid in full.
Lastly, it assumes that all holders of state and local general obligation bonds would, based on historical precedent and the priority of debt repayments in budgets, eventually be repaid in full. Thus another $1.1trn (“State and local GO”) are not classified as unrecoverable.
We’re now looking at the middle pie chart, which shows all muni bonds that RF thinks could enter a recovery process. Using Fitch data, it eliminates $662bn of uninsured revenue bonds where repayment could be delayed but full debt service is assumed to resume within five years. These bonds include those based on state sales tax, public college tuition revenue, local utilities revenue and leases from airports.
This leaves just over $1trn worth of bonds where if default were to occur, the recovery rates would be less than 100 per cent. Using Fitch data again, it says there are roughly three groups of bonds, comprising 90, 70 and 40 per cent recovery rates. The last group includes those bonds with no liens on assets and the worst recovery prospects, for example those based on Indian tribal gaming revenue or student loan revenue(!).
But of this $1trn, RF expects only $69bn to default. Why?
Correctly, it does not rely merely on historical default data for muni bonds. “Given current conditions, we thought 0% [default rates on AAA, AA or A muni bonds] might be a little low, even if historically accurate,” write the authors.
Indeed.
Instead it uses two sets of default data. First, 2009 corporate default rates. Second, 2007 municipal-to-corporate default ratios. It argues that the lagging nature of state and (especially) local government finances make a comparison with crisis-era corporate defaults relevant. The two different years are chosen to provide the most conservative (i.e. highest) estimate possible.
The result:
The municipal-to-corporate ratios from the 2007 Fitch Report (55.81%, 92.11%, and 91.94%) were applied to Moody’s 2009 corporate rates from the 2010 Moody’s Report to determine the default rates for the analysis — 0.28%, 0.50%, and 4.46% for municipal Default Classes 1, 2, and 3, respectively [i.e. AAA, AA and BB+], as shown in the charts.
Of the $1 trillion (or more precisely, $1.028 trillion), $960 billion is not expected to default over the next ten years.
It then maps these default rates against the recovery rates outlined above and estimates that:
To the remaining $69 billion that may potentially default over the next ten years, the expected recovery rates were applied. It is assumed that $53 billion will be recovered when the issuer resumes debt service payments or through the sale of the entities’ assets.
Cate salutes the report but would have liked some estimates on how derivatives affected default and recovery rates. We agree (on both counts) and would also add three caveats related to the methodology (doubtless there are others):
1. The total faith shown in monolines and GO bonds could be considered optimistic.
2. Even with a “conservative” use of corporate default data, it’s worth remembering that these were far less shocking than many expected in the early stages of the crisis. And that corporate and muni bonds behave differently.
3. Historical data still, inevitably, influence the final estimates.
There are also the risks that can’t be assumed in an exercise such as this one. John Carney outlines some of them in a post on Monday: political risk, a lack of transparency, reflexivity, arbitrage leading to bubbles, etc. He concludes that ”investors need to be aware that the performance of muni bonds is subject to unpredictable risks that can make accurate forecasting impossible.”
Nevertheless, with pinches of salt at the ready, this remains an interesting paper and provides a decent counterpoint to the analysis of Meredith Whitney she who must not be named.
Related links:
Muniland: a muni bonds blog – Reuters
Black Swans and Muni Bonds – CNBC
Meredith Whitney and the muni fifth dimension – FT Alphaville
Supply and demand in the muni market – Self-evident