limitations on state and local government

National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
LIMITATIONS
ON STATE
AND LOCAL
GOVERNMENT
BORROWING
FOR
PRIVATE PURPOSES
JUDY TEMPLE*
Abstract - This paper provides an economic explanation for the variation across
states in the use of tax-exempt state and
local government bonds to finance private investment. It is argued that, contrary to popular wisdom, private-activity
bonds do impose costs on the issuing jurisdiction. The use of these bonds is modeled as if a government official were selecting the cost-minimizing
combination
of private-activity
bonds and other incentives in support of private economic activity. Estimation undertaken using data on
industrial development bonds is consistent
with the hypothesis that the jurisdiction’s
use of these bonds depends on the relative prices of the various economic development tools.
INTRODUCTION
In the years preceding the Tax Reform Act
(TRA) of 1986, state and local governments
borrowed more money to encourage private economic activity than they borrowed
to finance traditional public projects, such
as roads, bridges, and schools. As explained by Kenyon (1991) and Zimmerman
(1991), the state-by-state volume caps im*Northern
lllmo~s Unwmty,
DeKaib. IL 601 15.
41
posed by the TRA of 1986 were a continuation of the federal government’s involvement dating from the late 1960s in
limrting these so-called private-activity
bonds. The federal government has a legitimate role in limiting the volume and use
of tax-exempt private-purpose bonds, because by issuing these bonds, state and local governments use the implicit federal
government subsidy to arbitrarily create
differing rates of returns across investments. It is often claimed (Zimmerman,
1991; Clark, 1985; Clark and Neubig,
1984) that private-activity bonds can be issued at virtually no expense to the issuing
jurisdiction, because state and local governments are not responsible for the repayment of principal and Interest. If that is the
case, then an important and yet unaddressed question arises. If jurisdictions can
issue private-activity bonds at no cost to
themselves, then what was limiting privateactrvity bond Issues in the years before
1986? Moreover, the perceived costlessness of private-activrty bonds seems inconsistent with the observation that these
bonds commonly are issued as part of a
package of fiscal incentives to potential
investors. Other incentives include tax reductions, such as property tax abatements
and investment tax credits, direct grants,
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
and subsidies. If tax abatements and other
Ionbond incentives are costly but privateactivity bonds are without cost, one would
expect (assuming some substitutibility between incentive types) that jurisdictions
would place little emphasis on nonbond incentives.
l suggest that jurisdictions act as though
there is a cost to themselves in the issuance of private-activity bonds. I provide
and estimate a model of the jurisdlctlon’s
fiscal incentive mix in the years preceding
1986 using a standard production theory
framework. The jurisdiction chooses the
cost-minimizing combination of incentives
to provide a given level of subsidy to the
private sector. Assuming substitutibility between incentive types, the relative costs of
the incentives are an important determinant of the jurisdiction’s reliance on private-activity bonds.
BACKGROUND
The exemption of state and local bond interest income from the federal income tax
allows state and local governments to borrow at rates that are lower than the rates
for corporations and even the federal government. Historically, the majority of taxexempt bonds were general obligation
bonds. Backed by the taxing power of the
jurisdiction, general obligation bonds typically are issued to finance public construction projects. Beginning in the ‘I93Os, however, state and local governments began to
Issue industrial development revenue bonds
(IDBs) to support private investment. Unlike
general obligation bonds, the payment of
revenue-bond interest and principal is paid
out of the profits or proceeds cf the project being financed. The debt service is secured by the subsidized project rather than
by the tax revenues of the juristjiction.
State and local borrowing to support private investment increased over time.’ By
1968, Congress acted to limit the large
quantity of tax-exempt bonds being Issued
to supFlort what it viewed as private purposes. The IRevenue and Expenditures Control Acl of 1968 required syate and local
government bonds to be tilxable if 1!5 percent or more of the bond roceeds were
P
_
used in a trade or buslness~ and If 2!) percent or more of the funds iused for the
debt service came from a tiade or business. Industrial development revenue bonds
issued for clsrtain activities,, however, were
thought by Congress to pr$vide benefits
that were essentially “publ’c” rather than
private. As a result of the I 968 act, tax-exernpt IDBs could continue 10 be Issued to
finance convention and sports centers; airports; pollution control equipment; local
sewerage, water, gas, and electric facilities;
and residential real estate and industrial
parks. Exempt from the 25% rule were
small IDBs (under $1 milliol;l) used to purchase or improve land or dlepreciable property.
Although the 1968 legislation limiteld certaln uses, the quantity of the allowable
tax-exempt IDBs increased greatly in the
subsequent years. The actu~al volume of
tax-exempt IDBs was uncerkain, however,
because many of these bo$ds (particularly
in the small-issue category)1 were placed
privately an’d not Included in any available
statistics. Congress included an informatoon-reporting requirement in the Tax Equllty anlj Fiscal Responsibility Act of 1982
tihat required state and locgl governments
tl3 file reports with the IRS’concerning their
t,ax-exempt private-activity bond issues. BeginninGi in ‘1983, state and, local governments ‘were required to file information
about three types of private-activity bonds:
IDBs, student loan bonds, and private exempt-entity bonds (issued primarily for prlvate nonprofit hospitals anij colleges.) In
1985, i.his reporting requirhment was extended to mortgage subsidy bonds.
THE COSTS OF STATE AND LOCAL
FISCAL lNC.ENTlVES
Junsdictlon!, offer fiscal incentives, such as
tax-exempt bonds ;Ind tax iabatements, in
I
LIMITATIONSON GOVERNMENT BORROWING
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
order to induce firms, nonprofit organizations, and individuals to locate new facilities or expand old ones within the jurisdiction. I assume that a government official
selects the mix of financing methods that
minimizes the cost of providing a given
level of support to the private sector. A
major determinant of the incentive mix is
the relative costs of the incentive tools. Tax
abatements, grants, loans, and training
programs each impose costs on taxpayers
in the year that they are granted. As a result, residents pay higher taxes to pay for
these incentives. Because of their similar
costs, I combine the nonbond tools and refer to them as tax abatements. The cost to
the jurisdiction of a dollar of tax abatements is a dollar in lost revenue. The remainder of this section examines the cost
to the jurisdiction of its private-activity
bonds, Previous researchers have noted
that an increase in the supply of tax-exempt bonds increases the cost of borrowing for all municipal issuers. Evidence reported by Peterson et al. (1981) Kormendi
and Nagle (1981). and Toder and Neubig
(1985) suggests that a $1 billion increase
in total state and local borrowing increases
the tax-exempt rate by an amount ranging
from a low of 0.37 basis points to a high
of approximately 9.0 basis points. It is
commonly assumed that a jurisdiction does
not perceive any cost associated with increasing its private-activity bond issues, because the borrowing by any one jurisdiction will have such a small impact on the
overall tax-exempt rate.
where Sr/SR > 0, Gr/SG > 0, and Sr/SW
< 0. The term G, represents the volume of
other types of bonds issued by jurisdiction
i. For the sake of simplicity, assume that
jurisdictrons issue two types of tax-exempt
long-term bonds: private-activity revenue
bonds and general obligation bonds. The
cost of i’s general obligation bonds is written: r,G = r,G(R,, G,, W,‘) Variables other
than current borrowing levels that affect
the Jurisdiction’s cost of borrowing are included in W or w’. These factors may include the level of private incomes, the unemployment rate, and the measures of the
jurisdiction’s existing obligations to bondholders. I next explain the determinants of
the jurisdiction’s cost of borrowing in more
detail.
Recent research in the areas of corporate
and municipal finance, however, makes the
assumption that borrowing costs vary directly with the level of borrowing undertaken by the individual firm or jurisdiction.
Nadeau (1988). for example, assumes that
the corporation’s cost of capital varies with
the degree of leverage. Metcalf (1989) and
Capeci (1990, 1991) assume that a jurisdiction’s cost of borrowing is affected by the
jurisdiction’s financial policy. Capeci (1990)
There are several explanations for the positive relationship between the size of the
jurisdiction’s bond issue and its associated
interest c0sts.j One is increased risk. Bondfinanced projects having lower probabilities
of default receive better credit ratings. The
better the rating, the lower the jurisdiction’s borrowing costs. For general obligation bond issues, ratings are based on the
jurisdiction’s wrllingness and capacity for
repayment of principal and interest. For
explrcitly tests this assumption for a sample
of general obligation bonds within the
state of New Jersey and finds that an increase in the amount of bonds issued of
$6.3 million increases the jurisdiction’s borrowing costs by 22 basis points.*
Consistent with this recent research, I assume that the jurisdiction’s cost of borrowing is a function of its own bond-financing
policres. To show clearly the manner In
which the jurisdiction’s borrowing costs associated with a certain type of bonds are
affected by the jurisdiction’s own borrowing, I write the tax-exempt rate r,R facing
jurisdiction i on its private-activity revenue
bonds (R) as:
r? = r,%, G,, W)
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
revenue bond Issues, the rating agencies
“see through” the is,suing jurisdiction and
base their ratings on the likely profitability
of the financed project. Greater levels of
new bond volume will be associated with
lower credit ratings as the rating agencies
express concerrl for the jurisdiction’s capacity (in the case ot general obligation
bonds) or the financed project’s capacity
(in the case of revenue bonds) for servicing
the debt.
The existence of regional segmentation in
the state and local bond market is a second explanation for the positive relationship between the level and cost of borrowing. Hendershott and Kidwell (1978) and
Kidwell et al. (1987) find that an increase
in bond issues from one state increases the
interest costs of those bonds relative to
other states’ bond issues. They suggest
that bonds marketecl regionally and bonds
marketed nationally are not perfect substitutes in investors’ portfolios. One likely
source of segmentation is the vanation
across states in the tax treatment of state
and local bond interest income. Many
states exempt from i.he state income tax
the interest income earned by residents on
bonds Issued within the state while at the
same time taxing the income earned by
the residents on out--of-state bonds. Examined recently by Lovely and Waslyenko
(1992) and Kidwell et al. (1987), this discriminatory tax policy allows governments
to borrow at a lower cost if the marginal
bondholder IS a state resident.
A third explanation involves agency costs
of state and local debt financing. Gordon
and Slemrod (1986) claim that agency
costs limit state and local borrowing. Bondholders fear that future voters or bureaucrats will act to adversely affect the value
of their claims. If agency costs are a positive function of the level of borrowing,
then jurisdictions pay higher interest rates
as the level of borrowing increases.
The discussion above presented three ex-
planations for why the interest rate associated with a particular typk of borrowing (G
or R) IS a positive functioq of the level of
that type of borrowing ujdertaken by the
jurisdiction. I also claim tqat spillover effects rnay exist within a jdrisdiction so that
r,R is a positive function of G, and that r,G
is a positive function of Ri. This may occur
for two reasons. First, the, existence of regional market segrnentati n may result in
the jurisdiction’s different 0bond types being
viewed by the bondholdel as closer substitutes .for each other than iare bonds issued
by other jurisdictions (pai/icularly those in
other states.) If so, then lbcal holders of jurisdictlon i’s G(R) bonds 3ay have to be
offered higher R(G) yields, in order to increase their holdings.
Reput.atiorl costs also are ~important. A
bond default within a juribdiction may affect the yields on all the jbrisdiction’s bond
issues and could even aff$ct borrowing
costs for other jurisdictiods within ithe region. Epple and Spatt (19156) cite evidence
that the default 01 the Wbshingtonl Public
Power Supply System (WF/PSS) on $2.5 billion of revenue bonds raised that state’s
general obligation borrov$ng costs Because the jurisdiction is tyipically not responsible for repayment qf principal and
interest in case of a revenue bond default,
a revenue bond default should not have
any impact on the jurisdidtion’s balance
sheet and, hence, its credlit worthiness.“
Potenflal general crbligatidn bondholders,
however, rnay vievv the rdvenue bond default as evidence that the: jurisdiction is run
by officials with poor finabcial judgement
or of evidence of a downward turn in regional economic conditions. Epple and
Spatt present a model of ~local government
borrowing in which reputbtion costs may
affect all jurisdictions within a state in
which a local bond default occurs.
In summary, I claim that private-activity
bonds impose costs in thq form of higher
interest charges on the is
Because the jurisdiction’s
I
LIMITATIONS ON GOVERNMENT BORROWING
are affected by its own financial policies,
an increase in private-activity bonds R, may
increase the costs (rrRand rtG) of the jurisdiction’s bond-financed projects.
THE MODEL
The choice of the fiscal incentive mix is
modeled as though the jurisdiction IS a
cost-minimizing firm choosing the optimal
combination of inputs to produce a certain
output level. It is assumed that the desired
level of subsidy activity has already been
chosen, so that the role of the government
official is to determine how to finance a
particular level of private-sector support.’
The official determines the mix of privateactivity bonds and tax abatements that
minimizes the cost of providing a given
subsidy level.
The total cost of engaging in subsidy activity is the sum of the costs of issuing tax
abatements and private-activity bonds. The
cost of tax abatements Z to the jurisdiction
is simply the revenue lost from these
abatements. Because an increase in privateactivity bond issues R, increases the cost of
the jurisdiction’s general obligation bond
issues G,, the cost to jurisdiction i of its private-activity bond issues R is (6r,G/6R,)G,R,.
The subsidy production function 5 is increasing in both Z, and in (rh - r,‘)R,. The
expression (rri - f)R, represents the interest savings available to private entities in
jurisdiction i from borrowing an amount R
at the tax-exempt rate r,R rather than at
the higher taxable rate r,,. As the jurisdiction increases its reliance on R relative to Z,
the increase in R reduces the interest savings from each dollar of R because f depends positively on R,. The cost-minimizing
incentive mix is found at a point of tangency between an isosubsidy curve and an
isocost line.6
The optimal mix of private-activity bonds
and tax abatements is expected to vary
across jurisdictions, primarily because the
relative prices of the incentives vary across
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
jurisdrctions. An increase in the price of R,
relative to Z, WIII lead to a reduction in the
share of bonds in the financing of the subsidy. As discussed above, the price of private-activity bonds depends positively on
the jurisdiction’s reliance on general obligation bonds G,. The change in relative prices
caused by an increase in G, induces the jurisdiction to substitute into the relatively
less costly tax abatements.’ Moreover, because an increase in G, increases the cost
of borrowing r,’ and, consequently, reduces the yield differential (rr, - r:), a jurisdiction relying solely on private-activity
bonds R to finance 5 would now have to
issue more R to achieve the same subsidy
level.
Another factor that may affect the cost of
private-activity bonds is the jurisdiction’s
quantity of debt outstanding. The reputation cost argument of Epple and Spatt
(1986) suggests that jurisdictions with
more debt outstanding have greater reason
to be concerned about the effect of a possible private-activity bond default on investors’ perceptions of the quality of the outstanding debt. It also is possible that the
jurisdiction may have difficulty redeeming
or refinancing Its outstanding bonds if high
levels of private-activity bond issues depress
the price of the bonds outstanding. Hence,
an increase in debt outstanding (included
in W) may increase the relative price of private-activity bond issues. The model predicts that jurisdrctions with greater outstanding debt will tend to rely more
heavrly on tax abatements in the financing
of a private-sector subsidy.
The model presented In this section illustrates that the great variation in private-activity bond issues can be attributed to one
of two sources. First, bond issues vary
across jurisdictions, because the total
amount of public subsidies to the private
sector varies. Second, the mix of bonds
and other Incentives varies across jurisdictions due to differences in the incentives’
relative prices. Holding the level of subsidy
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
constant, an Increase in the price of private-activity bonds relative to the price of
tax abatements reduces the share of bonds
in the financing of the subsidy. In the next
section, I will test two major hypotheses. I
expect that the volume of private-activity
bond issues will be negatively related to
both the jurisdiction’s reliance on general
obligation bonds and the jurisdiction’s volume of outstanding debt
EMPlRlCAL
RESULTS
The units of observalion are per capita industrial development bonds issued by state
and local governmerts by state in the
years of 1983 and 1984.” The data are
drawn from the Statistics of home
Bulletin (Clark and Neubig, 1984: Clark, 1985).
During this period, state and local governments issued approximately $70 billlon in
industrial development bonds. The Statistics
of income Bulletin disaggregates the broad
category of IDBs by Ieight purposes: small
issue and industrial park, multifamily housing, sports and convcention, airport and
dock, sewage and waste disposal, pollution
control, electric and gas, and miscellaneous
“other exempt activities.” Of the total
IDBs, bonds Issued in the small issue and
industrial park category accounted for
nearly half of all IDBs issued in 1983 and
1984. The analysis centers on the bonds in
the small issue and industrial park category. One might argue that these bonds
are issued for the most “private” purposes
of all the private-activity bonds in that they
subsidize private capital (primarily manufacturing) and real estate investment.
The quantity of IDBs issued is (equal to the
share of debt in the financing of state and
local support to the private sector times
the level of that support. Unfortunately, information concerning the total value of
state and local government assistance to
the private sector (the variable 5) is unavailable. While the reporting Iof private-activity bond issues has been recluired since
1983, littIc information is :available about
the value of tax abatemeAts or other incentives.g Without inform+tion on total
subsidy acfivity 5, .the shale of bond issues
in the financing of 5 canqot be determined. I u’re various proxiks to represent
the jurisdiction’s support if private activity.
If jurisdictions engage in dubsidy ac:tivity in
order to either lure busin sses away from
other juristlictions or to p1event other jurisdictions from doing the same to them,
then measures of interjurisdictional competition are appropriate instruments for the
level of ecl3nomic assistanbe provided by
the jurisdiction.
Two rneaslJres of interjuri$dlctional competition are used. CONTlGUbUS is the population of contiguous state; divided by the
state’s population. I preditt that the level
of subsidy activity undert;/ken by governments within a state increjases with1 CONTIGUOUS. GOVTS, the number of local
governments per 1,000 rqsidents in each
state, is a measure of intrjastate competition. I hypothesize that thie greater the
number of local governmints
in each state,
the greater the number of subsidies that
wtll be offered as jurisdictions in a state
compete against each otyer. Another approach to modeling subsiby activity involves recognizing that if kubsidies do encourage private-sector deyelopment,
measures (of private capitil Investment may
be correlated with the todal amourlt of
subsidies. I use two meas res of private inr
vestment. MANUF CAP islcapital
expenditures by manufacturing firms in eal:h state
In 1982 and 1983, and CHANGE IN
MANUF CAP is the perce$tage change in
MANUF CAP from 1977 to 1982 and from
1977 to 1983. In order tO avoid the problem of simultaneity, I use :manufacturing
expenditures data from I?82 and 1983 instead of the contemporatieous
dat,a from
1983 and 1984.‘”
I enter the subsidy proxie$ as independent
variables into the estimatihg equation. As-
I
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
LIMITATIONS ON GOVERNMENT BORROWING
suming that the subsidy proxies sufficiently
control for the differences across states in
the level of subsidy activity, the coefficients
on any other independent variables will indicate the effect of a change in that variable on private-activity bond issues, i.e., on
the share of debt in the financing of the
subsidy. Table 1 describes the variables
used in the analysis. Further information
about data sources is in the Appendix. The
independent variables included in the
regression equation are expected to have
an impact either on the level of 5 or on
the share of bonds in the financing of 5.
The first four variables (CONTIGUOUS,
GOVTS, MANUF CAP, and CHANGE IN
MANUF CAP) are subsidy-activity proxies.
Assuming that private-activity bonds are
not inferior inputs, greater levels of 5 are
associated with greater levels of bond issues. The unemployment rate and the median income level also are included as
other factors that may have an effect on
the level of 5. Their relationship with S,
TABLE 1
DEFINITIONS OF VARIABLES AND REGRESSION RESULTS*
Variable
Name
CONTIGUOUS
Mean
(standard
deviation)
Definition
U RATE
the contiguous state
population divided by the
state’s population
number of substate
governments
per 1,000
residents
per capita capital
expenditures
by
manufacturing
firms in
1982 and 1983
percent change in MANUF
CAP from 1977 to 1982
and 1977 to 1983
state unemployment
rate
MED INCOME
median
DEBT SHARE I
share of general obligation
bonds in the financing of
government
capital
expenditures
share of outstanding
state
and local bonds in the
financing of the public
capital stock
dummy variable equal to
one for states that
exempt their own bonds
from state taxation
dummy variable equal to
one for states with
appointed rather than
elected state treasurers
GOVTS
MANUF CAP
CHANGE IN
MANUF CAP
DEBT SHARE II
DISCRIM
TREASURER
individual
income
CONSTANT
Estimates for
Bond Equation
8.07
(8.73)
0.26
(3.18)
0.57
(0.71)
0.12
(1.42)
$267.12
(122.83)
1.01
(6.01)’
53.05
(57.81)
-0.40
- 3.02)’
a.22
(2.58)
58.717
0.21
(0.17)
-0.28
-0.84)
0.91
(1.W
-0.10
(-2.18)
0.20
(0.13)
-0.20
(- 1 .B2)b
0.74
(0.44
-0.15
(-0.78)
(1.327)
0.24
(0.43)
0.49
(2.32)
-9.95
(-1.60)
0.35
6.06’
R
F
‘The dependent variable is the logarithm of Per capita small issue industrial development bond issuer by state in 1983
and 1984. Because all nondummy variables are expressed in logarithms, the estimated coefficients are elasticities. The
t-statistics for the regression are shown in parentheses (%gnificance
at the 5 percent level; b10 percent). The number
of observations is 96.
47
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
however, is uncertain. For example, lower
income jurisdictions may provide hrgher
levels of 5. On the ‘other hand, successful
subsidy activity may be reflected in higher
levels of private incomes.
The next four variables (DEBT SHARE I,
DEBT SHARE II, DISCRIM, and, perhaps,
TREASURER) rnay affect the Incentive mix
To represent the reliance on general obligation bonds by state and local governments within each state and ‘:o be consistent with my focus on the ID13 debt share,
I compute DEBT SH.4RE I as the ratio of
general obligatron bonds issued to the level
of capital expenditures underiaken by the
governments. Similarly, the variable DEBT
SHARE II is the ratlo of long-term state and
local debt outstanding divided by the size
of the public capita stock. I also predict
that jurisdictions in states that offer preferential tax treatment for bonds issued
within those states (DISCRIM = 1) will finance 5 with a higher level of debt finance. TREASURER is includecl In order to
see rf jurisdictions in states with more financially sophisticabed or kno!Nledgeable
state treasurers (assumed to be those who
are appointed rather than elected) rely
more on debt finan’cing of eclonomic development assistance.
Table 1 presents coefficient estimates of
the determinants of small issue and rndustrial park IDBs. The dependent variable is
the logarithm of pelf capita small issue and
industrial park IDBs by state for the 48
contiguous states. All nondummy variables
are in logarithms. Both the use of per capita amounts and the logarithm specifications are common ways of reducing the
likelihood of having heteroskedastic errors.
Indeed, conventional tests reveal no strong
evidence of heteroskedasticity in this equation. An advantage of using a panel data
set of 2 years of cross-sectional data instead of a single cross section is that the
number of observations is increased to
96.”
The results suggest that Iinterjurisdictional
competition is an important determinant of
the volume of IDB issues. A 1 percent increase in CONTIGUOUS results in a 0.26%
increase in small issue IDlBs. Because I hypothesize that this variable has an effect
on the desired level of subsidy acuvity, it
appears that jurisdictions located in states
adjacent to high-population
states are
much more likely to issue IDBs and grant
tax abatements.”
GOVTS, another measure
of interjurisdlctional competition, also is expected to be positively correlated with the
level of subsidy activity a’nd, consequently,
the level lof bond issues. Although, the estimated coefflcrent is positive, the standard
error is large.
Industrial development revenue bond issues
are positively related to manufacturing capital expenditures and are negativelly related
to the recent past trend in these expenditures. Bond issues are unrelated to the
level of unemployment and may be positively related to the level of income. The
combined effects of MANUF CAP,
CHANGE IN MANUF CAP, U RATE, and
MED INCOME suggest that jurisdictions do
not issue IDBs (and tax abatements, by assumptior$ in order to revive falling local
economies. Instead, the federal subsidy of
IDBs (as well as the tax dollars of local residents to finance abatements) is us#ed to
support private investment in states where
the manufacturing sector is already strong
and has been strong for a while.
The remaining variables are expected to
have an effect on the fiscal incentive mix.
As predicted, the signs on the debt share
variatjles are negative and are statistically
significant. These are important findings, A
1 percent increase in the general obligation
debt share reduces the reliance on IDBs by
0.1 percent, and an increase in the outstanding debt share reduces IDB issues by
0.2 percent.‘3 The existence of a discnminatory tax. policy exemptilng interest on instate bonds but not on out-state bonds
has no effect on the volume of IDBs. AII +H
I
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
LIMITATIONSON GOVERNMENT BORROWING
though the findings of Lovely and Wasylenko (1992) suggest that this exemption
lowers interest costs, jurisdictions in states
with this policy do not issue more IDBs
than jurisdictions in states without the policy.14 Finally, jurisdictions located in states
with an appointed state treasurer issue
more IDBs. This is consistent with the hypothesis that private-activity bonds are
more likely to be issued by financial managers who have a sophisticated knowledge
of financial markets and that appointed
treasurers are more likely to have this type
of knowledge.
IDBs are more likely to be issued by jurisdictions in fiscally sophisticated, nondeclining states, where the interstate environment is competitive and where the
jurisdictions rely relatively less on bond financing of their own public capital expenditures and have relied relatively less on
bond financing in the past.
I am grateful
dews for their advice and suggestnns
I also appreciate
49
on prwous
drafts
An earlw
the edltor. and several
vewon
of this paper was
at the 1991 meetings of the M!d-ConUnent
gtonal Science Assoclatlon
’ Zimmerman
Re-
All errors are my own
(1991) prwdes
an excellent werww
htstory of and the economic
of the
ISSUESInvolved in tax-exempt
borrowing.
2 Addttlonal
ewdence of a relatlonshlp
between
bond ,ssue
we and yields or credit ratings IS found I” Kldwell et a/
(1987) and Capecl (1991)
’ An increase I” Jurlsdltilon
I’S borrowtng
has two effects:
it increases the overall average mterest rate on tax-exempt
bonds across the country, and It increases the Jurlsdlcton’s own cost of botrowlng
relative to the average tax-
exempt rate The evidence presented
here suggests that
the own cost effect (whereby a $1 mllllon muease in borrowing by Jwrdiaion
! increases Its borrowing
3.5 basis points according
costs by
to Capecl’s 1990 results) dome-
nates the general market effect (a $1 bdlmn increase m
borrowing
by jurlsdlctlon
I increases the overall mterest
rate by 0 37 to 9.0 basis pomts)
In effect, the ratlo of 1”.
risdlction i’s lnterezt rate r, to the average rate r IS a pative funalon
of the jur!sdlctlon’s
borrowmg
I make the
slmpllfying assumption that the average rate I IS unaffected by jurisdiction i’s borrowng so that an muease in
the Jurlsdlctlon’s
I argue that the use of private-activity
bonds imposes two types of cost on the
jurisdiction. First, an increase in private-activity bond issues increases the jurisdiction’s
cost of issuing its general obligation bonds
and may affect the value of the juridiction’s outstanding debt. Second, an increase in private-activity bond issues increases the interest rate on these bonds
themselves, thereby lowering the potential
cost savings to the private sector. The significance of the debt share variables in the
empirical analysis is consistent with the “as
if” cost-minimization
model. I find that
valuable
of Steve Karlson, 6111Sjos-
Paul Menchik.
referees
presented
Private-activity bonds accounted for the
majority of state and local government
bond issues in the early 1980s. and large
quantities are still being issued today. Despite their prevalence, the positive analysis
of the jurisdiction’s decision to issue private-activity bonds has not been a major
focus of the recent tax-exempt borrowing
research. I model the quantity of privateactivity bonds issued as if a government
official were choosing the cost-minimizing
combination of bonds and other fiscal incentives The amount of bonds issued in
support of private economic activity varies
across states, because both the level of
state and local support of private activity
and the chosen share of debt in the financing of this activity vary across states.
the comments
tram. Larry Martin.
a”o”ymo”s
Conclusion
to many indiwduals who prwded
a~~~~tance I especially thank Ron Fisher and John God-
borrowing
affects r, ,ione.
4 It is possible I” the WPPSS caxe that potential
llgatlon bondholders
predicted
general ob-
that state tax revenues
would be used to pay part of WPPSs’s creditors. thereby
affecting
the state’s ability to ma,nta,n debt scrwcc on !ts
general obligation
bonds. Nonetheless.
the WPPS case IS
often cited as an useful example of reputatw
’ Instead of choosing
effects
the Max of Z and R to produce a
given 5, the problem could be expressed as choosing Z
and R to produce a given user cat
be consatent
of capital
with the discussIon I” Zimmerman
This would
(1984).
where he examne~ the effects of CongresslanaI proposals
regarding tax-exempt flnancmg and accelerated depreclatlon on the cost of capital
6 The formal statement
of the off~c~al’s problem IS as fol-
lows, where C, represents the cost to the JUrlsdlctlon of
prowdlng
R, and Z
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
and SOVT’), that do not vary ov+r tune (GOVTS as con-
m,mm,ze C, =~ 'V;;G/&Q,)G,lR, + /',
structed va’res Irttle). the state-s+&
highly corwlated
subject to 5, = S(7, !r,, - r,?R,)
pendent varrabies. The problem
WAS underlaker)
marginal product,
of R and Z equates
nonconsi.ant
the ra!ro of
of R and Z and the ratlo of margmai fac-
rrrde-
11 th& the “ntithin”
t,on IS relatrvely low The regress/on reported
where r,G = r,iG,, R. W,‘) and r,’ = r,(G,. R, L’V,)
The optImaI combrnzltron
constants are
with the rernarrrng
using a constant
var~a-
I” Table 1
ter-n that does not vary
by state or by tome.
”
tor Drlces
A reieree wpressed
concern tha’l the pos,t,ve srgn on
CONTIGUOUS may merely reflecf the fact that CONTIGUOUS IS lwer
for rum states arjd the IDBs te*nd to be IS-
wed ‘now in urban areas The ppsrtw
rPlat!onshlp
be-
tween COrlTlGUOUS .rnd IDB isspes still holds, however,
even when the sample excludes the 10 most rural ‘states
(as mcaswd
The marginal product of /I (the r~umer,~tor on the left~hand
side of equatron
”
3) show!, that the effect of an ~ncreaie in
R on 5 is comprrvd
of
theproduct
of two effects
crease in R ,ncrea~e~ the subsrdy by the amount
SC& are exogenous.
genera
that an rncreasr’ in the general oblrgatron
of the I”~
the outstarldlng
,n the sham of IDBs I” the f\narwng
of the (unobswable)
subsidy
are Imperfect.
Because the subsrdy prows
pendrnt
”
rn thus paper is lrmrted to tw year!, of 1983
variables reduces the /weI of IDBs, zi stated in
Perh,tps thl! expected effect of DISCHIM on the qu.mtrty
teed
of ?Id,O, tax lk~glslatlon I” 1086 slgnlflcantly
the borrowing
behwor
margwal
rrwstor
their borrwng
al
palicy only occurs when the
IS a state resident. lurrsdictrons nuy lrmlt
to a (relatively small) quantrty that can be
sold wIthIn the state In a recent analysis of the wrration
of state and local govern-
acro!.s statvs rn the volume of long-term
ments in 1985 and 1986. Because the ~estnctions in the
lax Reform Act of 1986 were antrclpated.
Because the rnterert savings I”
states with tlxs drscrwnatory
and 1984 for two rea~o”?. Frrst, 1983 II the frrst year
that private-actrwty bond data are avarl.lble Second, the
enaCtme”t
It IS
to ><?ythat an mcredse ,r those rnde-
of IDBs IS <!mblquous
’ The analyx
debt share or
with a decrease
the text
That 1’. beyond the scope
of the current paper
debt :;hare is aswcrated
more appropriate
<obl,gatron bond i>~
Temple (1990) pro\lrdes a model rn
which R and G ace endo<)enous
density)
acrcxs statt’s in subsidy actwity. tlhen It is posuble to say
an I”-
terest sawgs lit, - r,R), but the ,ncreaw ,n R reduces the
,nterest savings becalrse $ rncreases with R
’ , aslume that the ,urisdrcl,on’s
by population
If the subsrdy vdrlablez fully conttol for the variat,on
t~on bond
there was a
gerwral oblrga
ssue~. Temple (1992) also finds thilt DISCRIM IS
Insiglllflcant
large mcrease sn bond vo “me rn 1985 as governments
REFERENCES
rushed therr bonds to market
’ Although
the U 8 government
expendrture
has d cc~mprehcnsrve tax
budget, few states rdo Benker (1986) exam-
,ne~ the status oi state tax expendrture
reporting
finds that only 17 stdtes IpublIsh tax expendrture
Mweover,
the rwthods
vary wrdely xros,
Mrnnesotd.
used rn comprlrlg
and Washington)
attempt
(1986)
403.17
Caped.
John. “Local Fiscal Pol~cres, Default Risk and Municw
to Include the rwt’!ax abatements,
I” obtarnrng
-.
,,,crnt,ve polwes ,n twstence
of avaIlable
in each jtate as used by El-
smger (1987) and by Carlton (1979) usrng data from the
Industrial Dewlo~~ment and Srte Selection Handbook
The
problems wrth !he type c’f varrable art? that Its use ~mplres
that the number of rncentrve policies IS correlated
wrth
the actual usage of the ~r~erwvfs and .hdt the existence,
of the poltcres at the stat? level is correlated
with the use
of :umrlar incentives by local governmer~ts rn each state
bclwfrl
of usmg panel data III thrs type of
analysts is tha: the inclusion of state swcrfrc constant
term5 could c<jptt,re the unobserved
subsidy not captud
colnponent
of the
by the exist,ng proxres Ths ap-
preach was not successful us,ng thrs shxt
paw
Costs ” Workrng
“Crecllt Rr\k, Credrt Ratings, and Mumapal
Even aft
ter excludrng the varrable?. such as DISCRIM, TREASURER,
Dennis
Tax ioumal
W. “Why
Do An Lconometrlc
Economic Growth,
DC
Paper 259. Waltham,
MA’
1990
Panel Study ” Natwnal
Carlton,
possible proxy fc,r 5 ,s Ihe nuniber
” An rmportant
pal Borrowing
Brander!, Univwsity.
there estimates
lc Another
Closing the
I” S’ate Budget Oversrght ” N,Hiondi T<rx Journal 63
and
these reports
but therr reports emphasrze the diffirLlties
Karen M. “Tax Expendrture Reporting
Loophole
reports
states Only three states (M!chlgan.
nue loss estimates from local government
Benker,
i4 (1991)
Yields. A
41. 56.
New Forms Locate Where They
Model ” In Int@rregronal Mwemeot
edited by W.C Wheaton
and
Washington.
U.ban lrrstltute Press, 1979
Clark, Phil. “!‘rwate-Aarvlty
Tax-Exempt Bonds. 1984. ’ St.+
fiirics or Income Bulietrn
Washington.
Suvice. II 5 Cepartmwt
of the Tie&wry,
DC
Internal Revenue
1985
--
and Thomas Neubig. “Prrvbte-Actrrrty Tax-Ext?mpt
Bonds, 1983
Stdtlstlcs of Income Bull~rrn Washington,
DC
Internal Revenue Serwce, U.5 Department
of the Trea-
sury. 1984
Eiringer, Peter K. “The Rrse of the Entrepreneurial State I”
Economic Development ” Proceedin$s of the 79th Annual
Conference
Amera
oi the Natror;al Tax louinal
(198;‘)
Epple, Dennir
~Taw institute of
34-44.
and Chester
Spatt.
“State KWr~cttons on
I
LIMITATIONS
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
BORROWlNG
ON GOVERNMENT
Lcxal Debt.” Journal of Pub/x Economrcs 29 (1986)
APPENDIX
133
221.
Gordon,
R.H. and Joel Slemrod.
of Municipal
“An EmpIrical Examlnatlon
Financial Policy ” In Studres in State and Local
Anance. edited by H Rosen Chicago
Uwersity
of Chicago
Press, 1986
Hendershott.
Patric and David Kidwell.
Relative Security Supplies.”
“The Impact of
Journal of Money. Cred!t, and
Banking
10 (1978). 337-47
Kenyon,
Daphne.
“Effects of Federal Volume Caps on State
and Local Borrowtng
” Natonai
Tax lour”al44
(1991)
81~
92.
Kidwell.
David,
Timothy
Koch, and Duane
Costs.” National Taxlouma/ 37 (1984)
Kldwell.
David,
Timothy
Koch, and Duane
in the Mumopal
Kormendi,
Roger and Thomas
CONTIGUOUS- contiguous state populatron divrded
by the state’s population.
“Issue
Effects on Regtonal
Band Market.”
Economics and Busyness 39 (1987)
Journal of
339-47
Nagle.
and Revenue Effects of Mortgage
Stock.
“The Interest Rate
Revenue Bonds ” In Effrm
oency in the Muninpal
Bond Market.
Financing for “Pwate”
Purposes, edlted by George Kaufman
Greenwich.
Lovely,
Warylenko.
Interest Income and Munupal
Tax Journal 44 (1992):
Metcalf,
The Use of Tax-Exempt
JAI Press, 1981
Mary and Michael
Gilbert
clal Behavw
“State Taxation of
Borrowing
Costs.” Natonal
37-52
E. “Federal Tax Policy and Murwpal
” Proceed!ngs of the 87sr Annual
the NatIonal Tax Journal-Tax
Flnan-
Conference
of
lnstrtute of Amenca (1989).
109-14
Nadeau,
Serge. “A Model to Measure the Effects of Taxes
on the Real and Financial Decisions of the Firm ” Natronai
Tax Journal (1988): 467-81
Peterson,
George,
John Tuccillo,
lmpacl of Mortgage
and John Weicher.
Bond Market.
“The
Revenue Bonds on Securltles, Markets.
and Housing Policy ObjectIves ” In Effioency
!n the Mumopal
The Use of Tax-Exempt Fw?anong for “Pnvate”
Purposes. edited by George Kaufman
Greenwich
JAI Press.
1981
Temple,
Judy. “The Debt/Tax
Choice I” the Flnanclng of
State and Local Capltai Expenditures
Northern
lll~nois Unwers~ty, 1992
Temple,
Judy. “Determinants
ment Borrowng
T&or,
Eric and Thomas
Lansmg
Mlchl-
Neubig.
Bonds
“Revenue
Loss Estimates
” Natronal Tax Journal 38
395-414.
Zimmerman,
Controlling
D.C.: Urban
-.
Dissertation.
Dennis.
The Private Use of Tax-Exempt Bonds
Public Subsfdy of Pnvate Actwrty
Washington,
Institute Press,1991
“Federal Tax Policy, IDBs and the Market for State
and Local Bonds,”
DISCRIM- dummy variable set equal to one If the
state exempts Interest on its own bonds from state
taxation but taxes out-state bond interest The list of
states that engage in discriminatory taxation comes
from Kidwell et a/. (1984). I combine his Group 3
(states that exempt all in-state and tax all out-state
bonds) and Group 4 (states that exempt some instate and tax all out-state bonds.)
DEBT SHARE l-state and local government Issues
of long-term new money general obllgatlon bonds
In 1983 and 1984 diwded by total state and local
nonutillty capital expenditures In 1983 and 1984.
The general obligation bond data are from the files
of the Securities Data Company Capital expenditures are found I” the Census of Governments Governmenial Fmance
DEBT SHARE It-total long-term debt outstanding
(guaranteed and nonguaranteed) in 1982 and 1983
divided by the we of the 1982 and 1983 state and
local public capital stock Debt figures are from the
Fmance. CapCensus of Governments Governmental
ital stock Information IS from the Boston Federal Reserve Bank
GOVTS-number
of local (substate) governments in
each state per 1,000 InhabItants from the 1982 and
1987 Census of Governments
Government numbers
for 1983 and 1984 were interpolated from the
1982 and 1987 figures.
1990
The Case of Tax-Exempt
(1985):
De Kalb
of State and Local Govern-
” Unpublished
gan State Unwrs~ty.
” Mlmeo
CHANGE IN MANUF CAP-percentage
change III
MANUF CAP from 1977 to 1982 and from 1977 to
1983 The percentage change was negative for
some states, so a constant term 81 was added to
each value.
“The
551-61
Size and Term Structure Segmentation
Yleld Differentials
Stock.
Borrowng
Impact of State Income Taxes on Municipal
DEPENDENT VARIABLE-per capita small issue tndustrial development bond issues by state and local
governments by state for 1983 and 1984 Bond
data are from the Stat!stics of income Bulletin (Clark
and Neubig, 1984; Clark, 1985).
National
Tax Journal 37 (1984)
41 l-20
51
MANUF CAP-per capita capital expenditures by
manufacturing firms by state in 1982 and 1983
Bufrom the 1986 Annual Survey of Manufacturers,
reau of the Census
MED INCOME-median
“effective buying Income”
in 1983 and 1984 from the annual “Survey of Buying Power” I” We5 and Marketing
Management
Magazine. Median household income levels were
National Tax Journal
Vol. 46, no. 1, (March, 1993), pp. 41-52
converted into per person amounts tly dlvidlny by
the number of people per household uiinq informatlon llsted In the 1988 Sfatntical Abstract of the
u.s
TREASURER- dummy wiable set ecjual to one for
states that have appolnted treasurers and equal to
zero for states with elected treasurer; Front the
Book
of the States,
Council of ‘State Governments,
‘1982-83.
lJ RATE--l 983 .-lnd 1!)84 state unemployment rates
as a percent,3ge of the cullan labor force. From the
U 5. Department of Commerce’s State anti Metropdoiitan Area Dara Book, 1986.