“I think the whole idea is to look for good ideas, be innovative. I was interested in things even if I didn’t have much to do with them. When I had a job to do, an assignment, I did things with it. I built a fire under it.” - Joseph Siegel, aka Mr. CUSIP
Besides delivering inspiring quotes, Joe Siegel provided a massive service to the securities industry by chairing the Committee on Uniform Securities Imprinting and Processing (CUSIP) back in 1964. At the time securities dealers had to literally shut down, often for days, just to allow what we now call the “back office” to dig out from under the hundreds of thousands of non-standardized stock and bond certificates sloshing around during the early 60s bull market. Mr. Siegel fought
tremendous institutional inertia by saying dealers needed to change the way things had been done for decades. But over the years, he won over the naysayers. Now CUSIPs are an indispensable tool for dealers and investors to quickly identify and evaluate securities – and that was the purpose of CUSIPs.
In public finance there’s a similar institutional inertia towards managing primary market public finance data the old way. Case in point, a recent industry survey showed 84% of people still pull data from official statements and manually keys in numbers. After thousands of hours of our own research we’ve discovered that that method is now officially obsolete. It’s a string between 2 cans in a Skype world.
Mr. CUSIP, thank you. The CUSIP technology you helped develop and more importantly, compel the industry to adopt, can also now be used to dramatically increase the efficiency and productivity of public finance departments across the country. This will lead to greater accuracy and more time for the industry to help tackle the tough problems public finance issuers face.
Our resolution is to light some fires in 2012. And if you don’t see them directly we're confident you’ll at least smell some smoke.
What fires will you be starting?
With Steve Jobs’ recent passing and new biography on the bookshelves it’s been nearly impossible to avoid learning something about the guy who completely re-engineered some pretty major markets: portable music player, cell phones, and now the tablet computer. The thing that has intrigued me as we wash in the continuous stream of Jobs exposure is how he seemed to successfully embrace a number of contradictions. A Zen student/philosopher but also tyrannical to the point of one former colleague suggesting he would’ve made an excellent King of France. He was solidly in the billionaires club yet lived in a very modest home that he often left unlocked.
But the most interesting contradiction to me was his ability to wrap simplicity and complexity all into one design. The ipod operationally is phenomenally simple for the user – effectively 5 buttons and a touch-sensitive wheel on top. That said the guts of the ipod address some of the toughest challenges in engineering including data storage capacity, long battery life, and robustness of the sensitive internals in the face of misuse/abuse by music lovers.
As software designers we struggle with this same issue. We love to put great features into our tools but too many features, if not well organized and laid out, can quickly lead to analysis paralysis and an unhappy user.
In your public finance plans and client presentations, you may have the hard work and complexity done…but have you got the simplicity? How much Apple is in there?
The income tax has made more liars out of the American people than golf has. - Will Rogers
The wages of sin are death, but by the time taxes are taken out, it's just sort of a tired feeling. - Paula Poundstone
To put it mildly, tax talk is hot these days and of course with it comes uncertainty as to the value of the income exemption on municipal bonds. The administration has floated the so-called "Buffet Rule" which would enforce a minimum effective tax rate for wealthy individuals along with the draft Debt Reduction Act of 2011 that would involve automatic spending cuts, including ones on tax preferences on interest. Needless to say, many muni market participants are not fans of the additional uncertainty of the value of tax-exemption.
But really how new is this? The chart below shows historical top Federal marginal tax rates from the inception of the income tax in 1913 to today.
Source: National Taxpayer's Union
A few interesting things to note. First the highest federal marginal tax rate was often well north of 50% from 1932 through 1986. For most of that period this rate kicked in for those earning more than $200k to $400k, though during the late 30s and early 40s it was reserveded for those Rockefellers of the day taking down more than $5 million. Next, note the creation of tax-exempt VRDBs coincided fairly closely with a big drop in the top Federal Marginal Tax Rate (FMTR) with the Tax Reform Act of 1986.
How has this changed the value of tax-exempt bonds? This is at best difficult to say as the municipal market has changed so dramatically (along with other financial markets) during the period after the Tax Reform Act of 1986. Since then, the top Federal Marginal Tax Rate (FMTR) in relative terms simply hasn't changed much.
Once you remove the effect of rate levels themselves we find that the very approximate relationship between tax rates and tax-exempt / taxable ratios in the variable rate market are
Tax-Exempt / Taxable Ratio = (1 - TopFMTR) + 2%
Depending on the data set you use and how you control for rates you can get that 2% to move around quite a bit but we find this a reasonable rule of thumb when rates are in a normal range. Of course they haven't been anywhere near normal for many years since Fed Funds has been effectively zero.
Today the true tax risk could go in either direction. The threat of higher marginal tax rates on high earners would bode well for the value of tax-exemption, but proposals pushing a lower FMTR with fewer loopholes would likely hurt.
Tax risk is certainly real, now how much SIFMA/LIBOR swap do you need to hedge it?
The Committee on Uniform Security Identification Procedures (CUSIP®) came up with its alphanumeric protocol in the late 1960s to streamline the dizzying amount of paperwork piling up in the back offices of securities dealers. Despite a few changes since then, the same technology is used today.
CUSIPs are a key ingredient in performing a public finance debt profile in an automated way using electronic data sources. Knowing how and when CUSIPs are issued and general CUSIP practices is a vital part of understanding what might be going on with the raw data when numbers fail to foot.
In the tax-exempt market, new CUSIPs will generally be created with:
- a primary market, new public bond offering
- partial refunding - refunded portions are assigned new CUSIPs
- new securities issued against a trust such as a tender option bond program
- bonds partially remarketed
- bonds insured in the secondary market
The CUSIP rule of thumb is that if the character/terms of an entire CUSIP change (refunded or remarketed) no new CUSIP will be issued. If a portion of a CUSIP changes it gets more complicated as described below.
Refundings
In the case of a full refunding of a bond no new CUSIP will be assigned to the refunded bond – that’s the simple case. In the case of a partial refunding, things get hairier. Usually in the event of a partial refunding the original CUSIP will be replaced by two new CUSIPs, one each for the refunded and unrefunded portions. The exception to this rule (NYC is notorious for this given its many partials) is the CUSIP of the unrefunded portion may be re-used even after multiple partial refunding. In rare cases the original CUSIP won’t change and will simply serve the function of the unrefunded CUSIP.
Remarketing
If a bond is fully remarketed the CUSIP won’t change. If portion of a CUSIP is remarketed, the unremarketed piece keeps the same CUSIP and a new CUSIP is assigned to the remarketed portion. The new portion is likely to have a new interest rate mode.
Secondary Market Insurance
Whether full or partial, a new CUSIP will be issued in the case of secondary market insurance. In general, these CUSIPs will not be relevant when performing an issuer balance sheet analysis and should be tossed out. If you don’t, the same debt will be double-counted.
For information on how to use only original CUSIPs for a complete balance sheet analysis, see
here or give us a ring.
“It is exceedingly difficult to make predictions, particularly about the future.”
- Niels Bohr (Danish physicist)
Forecasting is a difficult though necessary evil in finance. So what rates should be used for forecasting? We strongly believe that when forecasting market variables, entire distributions should be created and not just single point estimates. Nevertheless, what should be the “center” of the distribution or our true expectation? We see too many bankers relying on forward rates as if they were the holy grail of market forecasts when in fact, forward rates are appropriate to use in only ONE specific case.
Perhaps the greatest breakthrough in financial engineering is the concept of pricing assets in a “risk-neutral” world. Skipping the unnecessary details, this idea essentially allows us to conclude that in complete and arbitrage-free environments only, we can assume that all securities will earn the risk free rate of return. This dramatically simplifies asset valuation in arbitrage-free markets.
What types of environments are arbitrage-free? The closest thing we get in public finance are the swap markets. That’s it. For everyone else (investment bankers, financial advisors, issuers and even investors) forward rates are interesting but simply inappropriate for use in forecasting. It’d be nice to have a simple calculation to do see what the future is most accurately expected to be, but the Real world and the Risk-neutral one are fundamentally different and should not be confused.
"More than any other time in history, mankind faces a crossroads. One path leads to despair and utter hopelessness. The other, to total extinction. Let us pray we have the wisdom to choose correctly." - Woody Allen
Milton Friedman argued in this 2003 WSJ op ed piece that though the Fed's performance for the first 70 years of its existence left a healthy dose of room for improvement, over the years since about 1986 the Fed's thermostat was honing in on a Goldilocks temperature: policy was getting just about right. I'm not sure how much I agree. I'm a believer in the nature of human history being at its core (unfortunately) repetitive, particularly when it comes to economic cycles.
As we sit in the longest run of monetary expansion in this country's history, thought it’d be interesting to offer a long view of long rates using everyone’s favorite benchmark, the 10 year Treasury Note.
The chart shows UST Note yields starting in January, 1962 up through June 28, 2011. Back in the early 60s when the gold standard imposed some discipline we had yields that are very close to where they are today, though with far less volatility. From there, if you look closely you can see the yield dips associated with these events:
Will a technical or other default on US Treasuries be the next event? Which way will yields go? How does a flight to quality work when the canonical “quality” asset is defaulting? I'm hopeful we won't find out though I'd say the chances are about 50/50 at this point.
Over the next few weeks we’ll put up some posts looking at yield curves more quantitatively and ultimately, how they can inform relative value and borrowing strategies for tax-exempt issuers.
"Those who trust to chance must abide by the results of chance."
- Calvin Coolidge
One of the early lessons of modern finance that I was taught, like many others, in portfolio management class is that correlation is risk in a large diversified portfolio. This lesson for many has morphed into myth in light of the financial crisis. Based upon fairly low correlation calculations in normal market environments, investors thought they held diversified positions in say domestic equities, real estate, or even credit default swaps. However in mathematical terms correlations are time inhomogeneous – that is, they aren’t stable over time. And worse, recent experience showed us that for many of the aforementioned markets in times of stress, when you need diversification most, correlations move to 1; now your subtly engineered, diversified portfolio is one big boulder heading south.
What does this have to do with municipal issuers? Municipal and
tax-exempt borrowers often focus their energy predominantly on cash flow and cost of capital when looking at debt management. That’s not to say mark to market changes are ignored, particularly in the case of liability based hedges which can lead to collateral calls and liquidity pressure, but this is usually secondary to understanding the impact on debt service and budgets.
From a cash flow perspective, the two primary types of cash flow risk factors in which munis dabble are interest rate (say LIBOR) and tax-exempt/taxable ratio (SIFMA/LIBOR) risks. Looking at these two factors, through up and down market cycles including the crisis, we find that the correlation is negative and has stayed negative. Ratio risk performs poorly when rates are low but very well when rates are higher. Market microstructure reasons abound to explain the phenomenon which should give financial managers comfort.
Interesting thing about tax-exempt variable rate demand bonds (VRDBs) is they contain both risks within the structure (no, LIBOR swaps do
not add tax risk – it exists in the VRDBs whether hedged or not!). With many issuers struggling against tight budgets and looking for ways to lower debt service, at least in the near term, how do you determine the optimal amount of each factor? That depends upon your expected view of rates, tax-exempt/taxable ratios, the volatility of each, and the correlation between the two. A good tax-exempt risk structuring tool can get you there.
"Those who trust to chance must abide by the results of chance." - Calvin Coolidge
"The problem is not their estimates, it's the range of potential error in those estimates." - Alan Greenspan
Despite all behavioral finance has revealed about over-confidence and specifically, the overuse of historic averages in finance, one pervasive legacy of the ubiquitous spreadsheet is a stubborn reliance on static numeric assumptions for market risk factors - “Let’s just use 3% for SIFMA to run that debt service schedule and do the analysis.” It’s such a common modus operandi that it happens often without a second thought. The reality is that keeping variability in the analysis of market variables (where it belongs) is more conceptually challenging, but infinitely better at providing meaningful information to issuers; the right tools can offer critical intuition.
This post describes four risks a skilled modeler with well-designed Monte Carlo tools can explore when looking at tax-exempt variable rate debt (VRDBs).
1. Interest rates
Perhaps less so today, but the primary risk factor people evaluate
in VRDBs is what might happen to the economy and its impact on the demand for short-term money. Of course, what happens at
the front of the yield curve is almost entirely driven by those dwellers of the Temple at the Fed. When evaluating tax-exempt VRDBs it makes the most modeling sense (for reasons explained in far more detail in this paper) to first capture some primary benchmark of overall interest rates. Usually, we start with LIBOR.
Note that for each period in the analysis, we’re creating an entire distribution of rates from the simulations as shown in the left LIBOR graph above. We use a straightforward but powerful interest rate model described in detail here.
2. Tax-exempt / Taxable Ratio (SIFMA / LIBOR)
If US tax law changes in a way that diminishes the benefit of tax-exempt income to investors, issuers will immediately face higher costs of borrowing. For this reason, the risk that tax-exempt yields trade closer to taxable ones is actually a component of the variable rate bonds themselves. NOTA BENE: this has nothing to do with whether or not there’s a swap hedging the interest rate risk in the bonds (the press has a hard time with this one).
The essential feature to capture of this basis risk is its inverse relationship with rates. When rates are high, ratios are low and vice versa. This has been a persistent feature of these markets through boom and bust and ignoring it, frankly can lead to expensive mistakes detailed here.
This means the Monte Carlo model needs to have a correlated, multi-factor component. Many are familiar with the most famous distribution (Gaussian or Normal) in one dimension: in this case we need 2 per the image below.
3. Credit support costs
Liquidity support behind VRDBs is no longer a slam dunk. It’s now a much more significant component of cost. In fact, with SIFMA where it is currently, support costs today are generally greater than the interest rate itself! That said these costs are fixed only through the next renewal date. What assumption does an issuer make after that?
A well-designed Monte Carlo simulator allows the user to model the potential changes in credit support costs and add them to the interest rate and basis risks described above.
4. Trading spreads
Much of the joy in public finance lies in the 50,000+ issuer community and its attendant variety. VRDBs are issued by different credits in different states with different tax regimes backed by different banks. As much as SIFMA offers the benchmark against which most variable rate programs are compared, there will undoubtedly be noise around that benchmark, and sometimes that noise is worth trying to understand.
A flexible Monte Carlo rig should provide the user with the ability to model trading spreads, both in expected level and volatility.
5. Converting the VRDBs to Fixed
One plausible “worst case” (one of the three key questions every CFO must ask) might occur if there’s no l the providers with reasonable prices and terms go hiding. In that case, the multi-modal features of the bonds may kick in and the issuer faces a fixed rate remarketing. But what might the market look like then?
With a powerful, generalized simulation framework, the analyst can simulate a long term rate factor, either as part of a complete yield curve simulation or as a separate factor correlated to short term rates. The payments on the bonds then “flip” from the floating index to a (likely higher) fixed rate at the expiration of the letter/line of credit. It’s important to note that the fixed rates are simulated as well, so it’s not just a single fixed rate assumption on the remarketing date, but an entire range of possibilities. This is consistent with the uncertainty associated with that future unknown rate environment.
There you have it. If you need help setting up a model to capture these risks, let us know.
“An Expert is a person who avoids small errors while sweeping on to the grand fallacy.” - Steven Weinberg
Lots of people are looking for a “muni expert” to tell them either all will be well in muni-land or if in fact the sky is falling. Ms Whitney and others, usually with less experience in the municipal market, have tended towards the latter, while the rating agencies and an increasing number of hedge funds and smart money seem to be saying “buying opportunity.” If there were such a thing as a muni expert and I'm not playing one here or on TV, here are 5 things she would tell you:
1. Municipal Defaults are Rare; Predicting Rare Events is Hard
This should be obvious but some people seem to forget it. Extreme events are by definition rare and as a result, not usually well understood. If they were common and well understood…they
would no longer be called “extreme.” The level of structural gaps in state and local budgets is currently extreme. The usual strategy of “weather the slowdown and wait for the economy to bounce back”, isn’t working this time around. Hope as a strategy is increasingly ineffective. That said, anyone that says with confidence that because of this reality, municipalities will suffer 100x historical default experience doesn’t understand munis. A number of the reasons follow.
2. Municipal credit history IS relevant
Yes, the disclaimer in any registered investment that says, “Past performance is not an indicator of future results” is duly noted. However, the fundamental characteristics of states and municipalities as going concerns who usually feel very strongly they must pay their bills are relevant to the current analysis and any predictions about defaults. These are fundamental features and characteristics of states and local governments and that history is extremely relevant.
3. Municipal Bonds are not Mortgage Backed Securities
The problems that led to investor flight from the RMBS and asset-backed markets generally are numerous and I won’t even attempt to list them here. They’ve been thoroughly described by others, in some cases very entertainingly. To compare that market to the municipal bond market is simply inappropriate and shows a deep lack of muni ‘expertness’.
4. Underfunded pensions don’t necessarily mean bondholders will lose
Unfunded pension liabilities are a large and real structural problem. In the very cruel budget calculus that’s going on at the Federal level on down (sorry Calculus, I know that’s deeply insulting to you…), winners and losers will be chosen by elected officials working with budget managers. But if it’s a choice between a permanent scar in the capital markets due to a bond default and deeply unhappy pension beneficiaries (current and former municipal employees), large states and municipalities will generally cut the pensions. This is an opinion, but it’s an educated one. There will be outliers – budgets haven’t been this difficult in decades, but this will be the rule.
5. Market Discipline is Real and Big Issuers Get it
Right or wrong, at the highest levels state and local government finance officials listen to their public finance investment bankers and advisors. What they hear is that going down the road of default could close off access to capital markets for years not to mention the millions in legal costs. It is an absolute last resort and to be avoided if there is any reasonable, practical alternative: slashing services, raising taxes, cutting pensions, whatever. These things have only begun to play out. In the majority of cases they will happen well before a real threat of default.
With all the Meredith Whitney press and its derivatives (pun
intended), I wondered what public finance bankers, advisors, issuers, and investors do to keep track of the impending flood of material event notices that seemingly half the world thinks are coming down the pike. After all, you've got these required notifications per SEC rule 15c2-12 on munis:
- Principal/Interest payment delinquency
- Non-payment related default
- Unscheduled draw on debt service reserve reflecting financial difficulties
- Unscheduled draw on credit enhancement reflecting financial difficulties
- Substitution of credit or liquidity provider, or its failure to perform
- Adverse tax opinion or event affecting the tax-exempt status of the security
- Modification to the rights of security holders
- Bond call
- Defeasance
- Release, substitution or sale of property securing repayment of the security
- Ratings Change
- Annual Financial Information & Operating Data
- Annual Financial Statements or CAFR
- Failure to provide annual financial information as required
These are all posted on EMMA, though it's pricey as a subscription service for $45k/year. Do people have an automated system for tracking events for the issuers/investments you care about?
We think one should exist...