Advance Refundings of Municipal Bonds

• Author: RICHARD C. GREEN,FRANCIS A. LONGSTAFF,YUHANG XING,ANDREW ANG
• Published date: 01 August 2017
• DOI: http://doi.org/10.1111/jofi.12506
• Date: 01 August 2017

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 4 •AUGUST 2017

Advance Refundings of Municipal Bonds

ANDREW ANG, RICHARD C. GREEN, FRANCIS A. LONGSTAFF,

and YUHANG XING∗

ABSTRACT

The advance refunding of debt is a widespread practice in municipal finance. In an

advance refunding, municipalities retire callable bonds early and refund them with

bonds with lower coupon rates. We find that 85% of all advance refundings occur at

a net present value loss, and that the aggregate losses over the past 20 years exceed

$15 billion. We explore why municipalities advance refund their debt at loss. Finan-

cially constrained municipalities may face pressure to advance refund since it allows

them to reduce short-term cash outflows. We find strong evidence that financial con-

straints are a major driver of advance refunding activity.

ONE OF THE MOST IMPORTANT TRENDS in municipal finance over the past decade

is the dramatic increase in the number of bonds that are advance refunded.

In an advance refunding, also known as a pre-refunding, a municipality issues

new debt to retire an existing callable bond issue that is not yet callable.1The

number of municipal bond issues advance refunded per year has increased

from roughly 300 in the late 1990s to more than 30,000 over the 2012 to 2013

period. Nearly 50% of the $300 to $400 billion of municipal bonds currently

issued each year is associated with advance refundings.

∗Andrew Ang is with Columbia Business School and NBER. Richard C. Green is with Tepper

School of Business, Carnegie Mellon University. Francis A. Longstaff is with UCLA Anderson

School and NBER. Yuhang Xing is with Jones Graduate School of Business, Rice University. We

greatly appreciate the research assistance of Aniello Bianco, Camilo Botia Chaparro, Yunfan Gu,

Karthik Nagarajan, Rafael Porsani, Jackson Pfeiffer, and Tsuyoshi Sasaki. We are grateful to

Chester Spatt for originally bringing this problem to our attention. We are grateful for the com-

ments and suggestions of Vineer Bhansali, Jennifer Carpenter, Trevor Harris, Tal Heppenstall,

Andrew Kalotay,Kemp Lewis, Dan Li, Paul Luhmann, Stefan Nagel, Matt Roggenburgh, Rich Ryf-

fel, Norman Sch¨

urhoff, Chester Spatt, Matt Whysel, and seminar participants at Carnegie Mellon,

Columbia, U.C. Davis, the Federal Reserve Bank of San Francisco, Georgetown University, Hong

Kong University, Rice University, Singapore Management University, Texas A&M, the 2013 EFA

CFEA Conference in Cambridge, the 2013 CFEA Conference in Chapel Hill, the 2013 Norwegian

Financial Research Conference, the 2014 NBER Summer Asset Pricing Workshop, and the 2016

American Finance Association Conference. We are particularly grateful for the comments and ad-

vice of Acting Editor Greg Duffee and two anonymous referees. All errors are our responsibility.We

have read the Journal of Finance Disclosure Policy and are not aware of any conflicts of interest

with interested parties (see detailed disclosure statement on the Supporting Information page in

the online version of this article).

1Section 149 of the Internal Revenue Code requires that the bond being advance refunded be

called at its earliest call date for the refunding bond to retain the tax-exempt status of the existing

debt.

DOI: 10.1111/jofi.12506

1645

1646 The Journal of Finance R

Intuitively, an advance refunding can be viewed as an early “synthetic” call

of a callable bond. The proceeds from the new debt fund a trust that de-

feases the remaining coupon payments on the existing bond up to the call

date as well as the call price for the bond. The defeasance effectively extin-

guishes the existing bond issue in much the same way that calling a bond does.

Thus, an advance refunding is economically equivalent to calling the exist-

ing bond early at a call price equal to the escrow amount needed to fund the

trust.

Viewed this way, the decision to advance refund a bond issue turns on

whether it is optimal to “call” the bond now at the escrow price or to wait

and revisit the call decision at the actual first call date. Thus, the advance

refunding decision closely parallels the early exercise decision for American

options. A key factor complicating the advance refunding decision, however,

is that the new bond can typically be issued at a lower yield than the out-

standing bond, resulting in an immediate reduction in the cash flow needed

to service the municipality’s debt. Thus, an advance refunding can provide

short-term budget relief for a municipality. This is particularly important

since municipalities generally cannot borrow to finance operating activities

and would otherwise be forced to raise taxes or lay off public workers.2A finan-

cially constrained municipality may therefore face pressure to advance refund

even when a suboptimal “early exercise” results in a net present value (NPV)

loss.

In this paper, we study the effects of advance refunding decisions using an

extensive data set of municipal bond advance refunding transactions. Specifi-

cally,the data set includes over 206,000 bond issues that were advance refunded

from 1995 to 2013, with a total par value of $582 billion. This data set, in con-

junction with term structure data from the municipal bond market, allows us

to estimate the NPV of each advance refunding in the sample, along with the

associated cash flow incentives faced by these municipalities at the time of the

transaction.

A number of surprising results emerge from this analysis. We find that, dur-

ing the 1990s, a substantial fraction of the advance refundings are optimal in

the sense that they create value for the municipality. Beginning with the re-

cession of the early 2000s, however, the NPV associated with a typical advance

refunding turns negative, and then becomes sharply negative following the

2008 financial crisis. As a result, nearly 85% of all advance refunding trans-

actions during the sample period result in a loss of value for the municipality.

On average, advance refundings result in an NPV loss of 2.66% of the total

par amount being refunded. This represents a total loss of over $15 billion

during the sample period. Furthermore, this total does not include the cost

2Almost all municipalities are required by statutes, charters, or state constitutions to balance

their operating budgets. Municipalities can only borrow for capital projects, and then there are

elaborate restrictions such as requiring approval of voters or of a state-wide board. However,

municipalities are rarely restricted from advance refunding existing debt as long as the transaction

does not extend the maturity of the original debt.

Advance Refundings of Municipal Bonds 1647

of the transaction fees paid by municipalities to bond underwriters to execute

advance refunding transactions.3

In sharp contrast, we find that more than 96% of all advance refundings

result in immediate but short-term cash flow savings to the municipality. On

average, the advance refunding of a bond issue reduces the annual cash flow

paid by a municipality to service its debt by 1.41% of the par amount of the

bond issue being refunded (141 basis points).

While it is tempting to conclude that municipalities are simply advance re-

funding their debt to relax cash flow constraints, it is important to consider

the effects of other types of financial frictions and constraints on the advance

refunding decision. First, we explore whether some advance refundings may

be motivated by an attempt to eliminate restrictive covenants in existing debt.

We find little support for this hypothesis. Second, we examine whether the re-

sults can be explained by informational frictions about the value of the implicit

American option. We show that, even under the most extreme assumption that

the implicit option is ignored altogether, more than 50% of the advance refund-

ings occur at a present value loss. Thus, the results cannot be fully explained by

informational frictions that might make it difficult for a municipality to assess

the value of the implicit option accurately.

We focus next on the question of what factors drive advance refunding activ-

ity. First, we examine how municipalities respond to the potential NPV gains

and cash flow savings associated with advance refunding. We find that advance

refunding activity increases significantly as the potential cash flow savings in-

creases. In contrast, there is little evidence that advance refunding activity

responds to the potential NPV of the transaction.

Second, we examine how advance refunding activity is affected by macroeco-

nomic and fiscal conditions. We find that advance refunding activity increases

significantly when states and local governments have current tax revenue

shortfalls or face budget deficits. This evidence supports the hypothesis that

financial constraints lead municipalities to advance refund bonds in order to

obtain short-term budgetary relief.

Third, we examine whether macroeconomic and fiscal conditions affect the

financial outcomes from advance refunding. We find that realized NPV losses

increase significantly as current tax collections decline or as unfunded public

pension liabilities increase. These results suggest that municipalities are com-

pelled to absorb greater losses to obtain short-term budgetary relief as their

fiscal condition deteriorates. This finding provides additional evidence that

financial constraints are a major driver of advance refunding activity.

Finally, we study the cross-sectional structure of advance refunding deci-

sions. Specifically, we conduct a probit analysis to examine how the potential

NPV and cash flow savings of a transaction affect the conditional probability

that a bond issue is advance refunded. We again find that the potential cash

flow savings is a major determinant of the advance refunding decision. We also

3These transaction costs are estimated to be at least 1% to 2% of the total par amount of the

transaction.