Thursday, November 20, 2014

California Municipal Default Probabilities and a Reply to Lumesis

Earlier this month, we published default probability scores for 490 California cities and counties using a municipal scoring model I developed during previous research. The scores and a description of the model can be found on the California Policy Center’s website. The methodology – which relies solely on financial statement data – is further justified in this academic paper.

The accompanying CPC study identified thirteen cities that had heightened risk of default or bankruptcy. The median city in the universe had a one year default probability of 0.11%, while cities in this highly distressed category had default probabilities of 0.74% and up.

Our findings were reported in the Los Angeles Times and produced rebuttals from two of the cities on the distressed list – Compton and San Fernando.

We also received a rebuttal from an unexpected source: a municipal bond analytics provider named Lumesis. In a November 10th commentary, they compared our county default probabilities to their Geo Scores, which measure relative economic health. Finding little correlation between the two sets of results, Lumesis concluded that our model “needs improvement”.

But this conclusion begs a fundamental question: are municipal bond investors better served by socioeconomic metrics like those provided by Lumesis or by metrics that rely upon financial statement data? Even our colleagues at Lumesis appear to recognize that economic health is not conclusive, noting the “ability of bad management to create a mediocre credit from a strong economy.”

Fortunately, we have some empirical evidence at hand to assess the relative strength of fiscal and economic predictors. Back in December 1994, Orange County California filed for bankruptcy. Thanks to the magic of the MSRB’s EMMA system, I was able to locate the County’s 1993 and 1994 financial statements –as appendices to old offering documents.

For the reader’s convenience, I have posted the statements here and here. Next I input relevant numbers from the statements into my fiscal scoring tool which you can see here and in the screenshot below.

The result was a default probability of 0.85%, well above the current median and comparable to the worst performing entities in the current universe. As of June 30, 1994, the County had a negative general fund balance and had experienced declining year-on-year governmental fund revenues – two harbingers of trouble we have seen in other default and bankruptcy cases.

Would the Geo score have singled out Orange County in this way? Perhaps Lumesis can run the numbers for us and report back. Short of that, I note that according to 1990 Census figures, Orange County had the 5th highest per capita income among California’s 58 counties. So it would seem that a methodology based solely on economic health (like the Geo score) would have missed this particular calamity.

Monday, November 17, 2014

Broker Order Routing, in a High Frequency Trading World

The brokers/dealers have had their fair share of scrutiny among the recent revelations in the high-frequency trading (HFT) saga.

Among the questions being asked are whether Brokers are routing orders to whichever venues pay them most handsomely for the flow ... and potentially not to whichever venue provides best execution for their clientele.

We previously covered the discount brokerage world in which TD Ameritrade is being sued
(see for example Gerald J. Klein, on behalf of himself and all similarly situated v. TD Ameritrade et al, 14-cv-05738) for their order routing decisions.

You may recall that shortly after New York's AG filed a complaint against Broker Barclays in June 2014 for issues relating to its dark pool, Broker Barclays saw a precipitous decline (of roughly 66%) in trading within its own dark pool.   Some of that has returned, but while the tide has turned and the "true" nature of trading activity in some of the dark pools has been revealed, others like Wells Fargo have had to shut their dark pools: each venue requires a certain amount of trading activity to be relevant, or advantageous.

Brokers' Routing Decisions - Where to Send the Trades

Today we're covering a little of what we've found in the brokerage world itself: the changing nature of Brokers' routing orders to their own dark pools. We're spent some time digging through order broker routing information in their quarterly Rule 606 reports, and found some interesting changes in the regularity with which some of the large brokers are routing "non-directed" orders -- orders for which the client hasn't specified a specific execution venue.

The data are sparse, and the time periods short, but it seems like Credit Suisse (which has the largest dark pool) is generally substantially increasing its order routing to its internal dark pool, while Goldies and Broker Barclays are generally decreasing their self-routing decisions.

Monday, November 10, 2014

Minimum Authorized Denominations: A Truly MAD Restriction Now Enforced by the SEC

Last week, the SEC sanctioned thirteen brokerage firms for selling individual investors Puerto Rico General Obligation bonds in denominations less than $100,000.  The action is part of a wider SEC effort to protect municipal bond investors, which has also included pressuring issuers to post their continuing disclosures on EMMA and censuring issuers that have improperly reported their pension obligations.

While the SEC’s efforts to enforce transparency are welcome, the enforcement of minimum denominations is, in my view, counterproductive. Before I explain why, I should begin with the necessary disclosure:  I am an owner of Puerto Rico General Obligations, so I clearly have a financial interest in taking the view I am advocating.  However, it is not a very large financial interest – probably not large enough to justify the time spent writing this piece.

You see, I own an “odd lot” of Puerto Rico GOs.  In fact, my purchase of $25,000 face value Puerto Rico bonds was legal in January 2014 when I made it, and would still be legal today. The $100,000 Minimum Authorized Denomination (MAD) restriction only applies to bonds issued in 2014 (and probably thereafter).  An individual investor can still purchase smaller amounts of PR GOs that were issued prior to 2014.  Many of these bonds mature in the 2030s, thus presenting risk characteristics very similar to the March 2014 MAD bonds, which mature in July 2035.

The reason for this discontinuity is that the SEC is enforcing a restriction in the Commonwealth’s bond resolution that authorized the issuance of the 2014 securities.  That resolution includes the following:

The Bonds … shall be issuable … in the minimum denomination of $100,000, with integral multiples of $5,000 in excess thereof; provided, however, that upon receipt by the Registrar from the Secretary of written evidence from any of Fitch Ratings, Moody’s Investors Service, or Standard & Poor’s Rating Services that the Bonds have been rated “BBB-,” “Baa3,” or “BBB-,” or higher, respectively, then the minimum authorized denomination of the Bonds shall be reduced to $5,000.

Prior to 2014, Puerto Rico carried investment grade ratings from the big three credit rating agencies, so earlier resolutions apparently did not need to incorporate the $100,000 denomination floor. Thus, investors can and do continue to trade these securities in smaller pieces as anyone can see on EMMA.

Buying older PR bonds is not the only way for an individual investor to evade the new MAD requirement. Since many municipal bond funds own PR GOs, one can gain exposure to this asset by purchasing shares in any one of these funds. Not only can the investment be less than $100,000 in most cases, but the exposure to PR will only be a small fraction of the fund’s overall exposure.
But why should investors have to pay mutual fund overheads to add Puerto Rico exposure to their portfolios?  It should be possible to simply buy a small amount of PR bonds directly as part of a diverse municipal bond portfolio or multiple asset class portfolio.

The MAD restriction begins to foreclose that option to all but the wealthiest investors. Imagine, for example, an accredited investor with $5 million in investable funds.  Let’s say that, after reading all of Puerto Rico’s voluminous disclosures, she decides to take some risk on Puerto Rico, but to limit this risk to 1% of her portfolio.  If worst comes to worst, and Puerto Rico repudiates its bonds, she will still have $4.95 million – enough to survive. But, since 1% of $5,000,000 is only $50,000, that investment choice is not available for the 2014 and future bonds.

Many hedge funds and other institutional investors thought Puerto Rico’s March 2014 bonds to be a sensible investment, concluding that the 8.7% yield more than compensated them for the Commonwealth’s heightened risk. That issue was five times oversubscribed, with hedge funds being major participants.  The question is why anyone else who shares this view must be compelled to pay a hedge fund “two and 20” to express this conviction.

Besides being discriminatory and easily circumvented, the MAD restriction is just bad public policy. Investors can buy small denominations of penny stocks which frequently become worthless, so why single out PR General Obligations as something from which the “little guy” must be protected? It furthers the false narrative that municipal bonds are an especially risky asset class, when, in fact, quite the opposite is true.  No US city, county, state or territory filed for bankruptcy in the last year.  How does that compare to the corporate sector?

Municipal bonds were recently proscribed by regulators from the list of high quality liquid assets that banks could hold.  The major justification for excluding munis was their lack of liquidity. Regulatory actions that scare investors away from municipal bonds and impose floors on trade sizes simply reinforce the lack of liquidity in the market.  This, at a time when we hear that Republicans and Democrats agree on the need for greater infrastructure investment – investment that could be partially financed by municipal bonds.

Finally, the MAD restriction uses credit rating agency assessments to determine whether the bonds are “junk” and thus subject to the denomination floor. Using bond ratings to distinguish between investment grade and non-investment grade securities goes against the spirit of Dodd Frank, which sought to cleanse regulations of their reliance on credit ratings. This aspect of Dodd Frank is one of the few parts of the controversial law that seems to have elicited bipartisan support.

While it is gratifying that the SEC is trying to protect municipal bond investors, we need to be sure that the rules being enforced actually contribute to investor welfare. Puerto Rico’s MAD requirement is a restriction that is easily circumvented, that penalizes individual investors, that reduces municipal bond market liquidity and that reinforces the credit rating oligopoly.  For these reasons, it fails the test of being a rule worthy of commanding enforcement resources.