The Cost of Not Amending the MLF

Note: Additional 3-Year scenarios here.

There’s a federal budget methodology for estimating the cost of amending the MLF. And one for not doing so. These numbers should be included in the debate.

Now that the Fed’s $500 billion Municipal Liquidity Facility (MLF) appears to have accomplished its lender-of-last-resort mission for the muni market without actually having to lend much money, there’s an intense debate about what (if anything) to do with all the dry powder that’s left.

The debate so far involves important and difficult issues that would be more relevant if the MLF was just another federal loan program or central bank operation.  But it’s not – by definition, federally subsidized MLF loans would be an alternative to federally subsidized tax-exempt bonds. Since there’s no difference in principle, the actual issue would seem to be a choice between two ways to deliver $500 billion of federally subsidized debt to state & local governments: either (1) amend the MLF to be a lender-of-actual-loans, or (2) let the $500 billion be placed in the now-functioning muni market.  Either way federal taxpayers will pay the freight. So their cost should be a big factor in the choice.

The Cost of an Amended and Fully Utilized MLF

For an amended MLF, the cost obviously depends on the pricing and terms of the loans.  The minimalist baseline for that would be just enough to cause the MLF to be fully utilized as a loan facility.  Since most state & local governments are rated investment grade, the rating requirement in the current MLF can stay.  But two amendments are needed: allow loan terms of up to 30 years and lower the interest rate to US Treasuries flat.  That’s enough to encourage utilization because even the AAA muni curve is currently slightly above Treasuries, especially in longer maturities. 

A Treasury flat interest rate will cover the cash cost of funding MLF loans, but there’s still the cost of a reserve for expected credit losses.  For that, federal loan budgeting (FCRA) determines the present value of expected losses based on credit ratings.  For investment grade loans to state & local governments, the FCRA number should be pretty low.  If the MLF’s portfolio is assumed to average out with mid-range investment grade ratings (where state & local ratings are clustered anyway), the upfront FCRA cost of a fully utilized facility would be about 1.2%, or $6 billion.

Based on other federal credit programs I’m familiar with, $6 billion for a minimalist, investment grade $500 billion MLF seems like a realistic cost baseline.  It can be more if Congress wants to be generous (e.g. zero-interest loans, sub-investment grade, etc.) but those add-on options aren’t necessary for the baseline estimate. It can also be less if MLF borrowers themselves pay some or even all of the FCRA cost out of loan proceeds, like an underwriting discount.  The full charge of 1.2% would add less than 0.10% to the annual interest cost of the loan — still an attractive deal.

The Cost of Not Amending or Utilizing the MLF

If the MLF is not amended, it won’t be utilized as a loan facility.  A large part of the case against amending the MLF is that it’s already done its job as a lender-of-last-resort, thereby obviating the need for it to be a lender-of-actual-loans.  The muni market is now open for business.  If investment grade state & local governments need $500 billion of debt with terms of up to 30 years at historically low, federally subsidized interest rates, this argument goes, let them issue bonds.

Let’s assume that they do.  As shown above, an amended MLF could probably lend $500 billion in 30-year, UST flat rate loans to state & locals for about $6 billion or less.  How much will it cost federal taxpayers for the tax-exempt market to do basically the same job?

Like FCRA for federal loans, there‘s a federal budgeting methodology for the dog-that-didn’t-bark nature of the tax revenue loss caused by a tax exemption: the JCT’s  The Federal Revenue Effects of Tax-Exempt and Direct-Pay Tax Credit Bond Provisions.  I’m familiar with it from federal infrastructure programs that have some impact on tax-exempt debt volume.  The basic principle is about substitution: if an additional tax-exempt bond is issued, an equivalent taxable bond is not bought and the tax that would have been paid on the taxable bond’s interest income is forgone.  The federal cost of the $500 billion of tax-exempt debt that will be issued if the MLF is not amended is therefore the lost tax revenue from a portfolio of taxable bonds that would have been bought if it were.

As complicated as all that sounds, it’s not difficult to model – I’ve done it before for an infrastructure loan program and was able to match the JCT/CBO scoring fairly closely.  Below is a summary of assumptions and results for what I think an unbought taxable bond portfolio (and the tax on its interest income) would look like in this case.  The Excel model is here if you want to take a deeper look or change assumptions.

Despite conservative assumptions, the numbers are big. Very big. The undiscounted total of lost tax revenue for the first ten years (how CBO scores a bill) is more than $20 billion. The present value of the full 30 years of lost revenue is over $35 billion using a Treasury discount rate.

The numbers appear big, but based on prior JCT estimates for other situations involving tax-exempt bonds, they should not really be surprising. The elimination of tax-exemption for advance refunding bonds in the TCJA of 2017 involved about $60 billion of bonds annually, so about $600 billion over a ten-year period. JCT’s estimate for tax revenue increase was about $17 billion for the 10-year budget score. A bill does not need to involve tax law change to be scored for tax impact. The WRDA of 2018 was all about water, but the WIFIA loan program was reckoned (incorrectly as it turned out) to cause the issuance of about $50 billion of additional tax-exempt bonds. JCT estimated that this would cause a 10-year revenue reduction of about $2.6 billion.

Some caution is warranted. Reasonable people can disagree about tax-expenditure estimates. Even the JCT acknowledges that the static portfolio model is at best a rough indicator of what might actually happen in the very complex reality of financial markets, especially over a long time frame. It makes intuitive sense that the tax-exemption is far from costless, but the actual cost is probably unpredictable — it could be a lot less or more than the numbers above.

Still, the scale of these numbers suggests that there’s a very real difference between the cost of amending and utilizing the MLF (about $6 billion or less) and the cost of not doing so and letting state & locals issue $500 billion of additional tax-exempt debt (about $20 to $30 billion, more or less). That would seem to turn some aspects of the current debate upside down — who exactly is getting bailed out here if the MLF is amended? Or if it isn’t?

One practical suggestion — the Congressional Oversight Committee should ask JCT and CBO to do preliminary estimates of (1) the tax revenue impact of not doing an amended MLF and (2) the FCRA cost of a minimalist amended MLF. That’ll help anchor the debate in quantitative terms, however approximate. More importantly perhaps, the nature of the estimates themselves will highlight what the debate should really be about.