Exceptional Proff 612

Econ Gov (2016) 17:185–209 DOI 10.1007/s10101-015-0172-y

ORIGINAL PAPER

Do fiscal constraints prevent default? Historical evidence from U.S. municipalities

John A. Dove1

Received: 21 November 2014 / Accepted: 18 August 2015 / Published online: 2 September 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract Through the nineteenth century numerous U.S. states developed extensive municipal fiscal constitutions. These generally came in the wake of financial crises and large-scale default of public debts. Although the constraints were imposed in order to minimize the likelihood that such outcomes would occur in the future, little work has been undertaken to analyze whether they were successful in achieving that goal. Therefore, this current study attempts to do so by empirically investigating how procedural safeguards and outright prohibitions on debt accumulation, along with hard budget constraints, and tax limits impacted the likelihood of default. This is done by evaluating municipal defaults that centered on the Panic of 1893. Overall, the results suggest that outright prohibitions on debt accumulation and hard budget constraints actually reduced the likelihood of municipal default across states, while tax limits and procedural safeguards increased that likelihood.

Keywords Default · Sovereign debt · Fiscal constraints · Panic of 1893 · Credible commitments

JEL Classification D78 · H73 · H74 · N21 · N41

The author would like to thank two anonymous referees as well as participants at the 2015 Economic and Business History conference in La Crosse, Wisconsin for extremely helpful comments on earlier versions of this paper.

B John A. Dove [email protected]

1 Manuel H. Johnson Center for Political Economy, Troy University, 137N Bibb Graves Hall, Troy, AL 36082, USA

123

186 J. A. Dove

1 Introduction

Through the nineteenth century U.S. state and local fiscal policy was a constant ebb and flow of public investment in the economy meant for economic development. At times this led to the rapid accumulation of public debt in order to finance many of these projects (dubbed “internal improvements”). In several instances this accumulation led to default and repudiation after macroeconomic financial crises hit. This pattern was witnessed in the aftermaths of the Panic of 1839, the Panic of 1873, and during Civil War Reconstruction.

Specifically,asaresultofthePanicof1839eightstatesandoneterritorydefaultedon their debt obligations, with five of those states repudiating all or part of those outstand- ing debts. A similar situation arose during Civil War Reconstruction when all of the former Confederate states, after having accrued large debts to finance internal improve- ment projects, also defaulted and repudiated sizeable portions of those debts.1 Finally, the Panic of 1873 culminated with nearly twenty percent of all outstanding municipal debt(againlargelyincurredfortheconstructionofrailroadsandotherinternalimprove- ment projects) being defaulted on at some point (Hempel 1964; Hillhouse 1936).

However, after each of these episodes numerous state legislatures and state con- stitutional conventions implemented various constitutional reforms in order to ensure that similar behavior was not pursued in the future. Thus, through the 1840s states passed constitutional amendments that significantly limited or prohibited those gov- ernments from issuing debt in the future, limited the ability to raise revenue through early variants of tax and expenditure limits, and limited the ability of states to invest in private companies. This situation also arose during Reconstruction, when a number of former Confederate states also imposed various fiscal restrictions.

An unintended side effect of these constraints was that through the 1850s and into the 1870s municipal governments began to accumulate large debts in order to finance these internal improvement projects, especially railroads. However, due to the Panic of 1873 many of these projects were shown to be unprofitable, which left many municipal governments unable to pay their outstanding obligations, forcing them into default and repudiation. Thus, similar restrictions as those that had been placed on state governments were finally also forced on local governments across numerous states.

All of these limits were meant to minimize or prevent similar behavior from occur- ring in the future. Both contemporary accounts of the time and more recent research would suggest that this was the purpose these fiscal constitutions were to serve (Dove 2012, 2014; Thies 2002; Hillhouse 1936; Ratchford 1941).2 However, no work has been done to formally evaluate whether these constitutional limitations were success- ful in preventing future default. Therefore, the purpose of this study is to provide a brief historical account of the development of these constitutional rules, especially those imposed on municipal governments, and to then empirically examine how these fiscal constitutions affected the likelihood that municipal governments would default in the future.

1 See Scott (1893), Ratchford (1941), McGrane (1935) for detailed discussions of both periods. 2 Anecdotal evidence from contemporary sources will be discussed in greater detail in a later section.

123

Do fiscal constraints prevent default? Historical evidence… 187

This latter work is done by evaluating the period through the Panic of 1893, with default data covering 1890 through 1905 for local and county governments. Impor- tantly, the Panic of 1893 and ensuing depression was an even more severe financial crisis that beset the country than was the Panic of 1873. However, whereas the after- math of the Panic of 1873 left nearly 20% of all outstanding municipal debt in default, in the wake of the Panic of 1893 less than 10% of all outstanding municipal debt met the same fate with no state governments defaulting (Hillhouse 1936; Hempel 1964). Therefore, this period provides an opportunity to see how, if at all, these fis- cal constitutions impacted the likelihood that a given municipal government would default.

Overall, the results suggest that outright prohibitions on the accumulation of debt significantlydecreasedthelikelihoodofdefault,whileproceduralsafeguardsincreased that likelihood, though the latter result is not robust. Further, strict tax limits also increased that likelihood whereas hard budget constraints reduced the likelihood of default.

The remainder of the paper is structured as follows: Sect. 2 discusses the relevant literature. Section 3 provides a brief historical account of the development of these fiscal constitutions at the state and local level. This section also evaluates the state and local defaults and repudiations that occurred through the nineteenth century and the possible mechanisms that may have been in place to minimize municipal default during the Panic of 1893 and ensuing depression. Section 4 describes the data and empirical specification employed in order to evaluate how these constraints affected municipal default between 1890 and 1905. Section 5 presents and discusses the results, while Sect. 6 concludes.

2 Literature review

This paper adds to several important strands of the literature. The first is the extent to which fiscal constraints might affect fiscal outcomes.3 This has been a heavily studied issue within the public finance literature. Some of this work has focused on how fiscal restrictions affect debt burdens (Alesina and Bayoumi 1996; Bohn and Inman 1996; Primo 2006; Alt and Lowry 1994; Kiewiet and Szakaly 1996), with much of this research finding a direct link between the strength of a state’s balanced budget requirements (BBRs) and the level of debt and deficits that exist. Specifically, this research finds that BBRs are relatively effective at lowering state borrowing costs and reducing debt burdens.

More recent studies have also examined the impact that the many nuances within these BBRs have had (Bohn and Inman 1996; Hou and Smith 2006). Hou and Smith (2006) develop measures of state BBRs based on both technical (i.e. no carry-over rules, debt limits etc.) and political (i.e. the governor must submit a balanced budget, the legislature must pass a balanced budget etc.) rules. Hou and Smith (2010) test these implications and find that technical rules have a larger impact than do political rules.

3 For recent literature reviews on these issues, especially as they relate to U.S. states see Rose (2010) and Krol (2007).

123

188 J. A. Dove

Additionally, analyzing different time periods, Poterba and Rueben (2001) and Lowry and Alt (2001) find that both anticipated and unanticipated deficits increase state borrowing costs, however the existence of a BBR mitigates this effect. Simi- lar evidence is also reported by Hallerberg and Wolff (2008) within the EMU, and by Johnson and Kriz (2005), the latter showing that BBRs, expenditure limits, and debt restrictions lower borrowing costs, while revenue limits increase them. Further research also indicates that BBRs may minimize political budget cycles (Streb and Torren 2013).

Finally, Hopland (2013) finds evidence that centrally imposed fiscal limits on Norwegian local governments that run chronic deficits lead to stronger operational sur- pluses in those local governments once implemented. These surpluses do not appear to be driven by central bailouts, but rather through reductions in local expenditures. Tied to this Hopland (2014) shows that these fiscal limits, once placed on Norwegian local governments, signal poor fiscal performance to voters who respond by increasing the likelihood of voting an incumbent out of office.

A related area of research concerns tax and expenditure limits (TELs). Here early work suggested that TELs had little impact on state and local fiscal outcomes (Abrams and Dougan 1986; Bails 1982, 1990; Joyce and Mullins 1991; Mullins and Joyce 1996). However, later research has in some instances found a stronger and statistically significant effect, especially when panel data is incorporated and potential endogeneity is accounted for (Bails and Tieslau 2000; Shadbegian 1996, 1998; Rueben 1997; Bae and Gais 2007). Other studies have found that tax limits increase state and local borrowing costs while expenditure limits decrease those costs (Bayoumi et al. 1995; Eichengreen 1992; Johnson and Kriz 2005; Lowry and Alt 2001; Poterba and Rueben 1999).

Similar work has attempted to pinpoint how or even if state and local TELs may limit government growth. For instance, using a principle-agent framework, Seljan (2014) finds that TELs are only effective at limiting public sector growth when limited- government preferring agents oversee them. Further, Kousser et al. (2008) show that there is no single TEL restriction that systematically limits government growth, outside of a handful of states, while Amiel et al. (2014) suggest that the overall restrictiveness of a TEL plays a crucial role in its ability to limit government growth.

This current paper also adds to the literature on the impact that hard budget con- straints have, especially within a system of federalism (Goodspeed 2002; Oates 2005; Qian and Roland 1998).4 Here much of this research focuses on the necessity of cen- tral authorities to avoid bailing out lower-level governments when fiscal imbalances emerge as a result of negative macroeconomic shocks. If accomplished, such a policy will credibly convey a central authority’s commitment to fiscal discipline and will avoid general moral hazard problems associated with soft budget constraints. Addi- tional empirical evidence indicates that in the Italian context, an important component of local government expenditures is the expectation of future transfers from the central government (Padovano 2014), while municipal bailout expectations in Sweden appear to account for nearly 20% of existing municipal debt (Pettersson-Lidbom 2010).

4 For an overview of this issue see Kornai et al. (2003) and Rodden et al. (2003).

123

Do fiscal constraints prevent default? Historical evidence… 189

3 A brief historical account

Throughout the nineteenth century state and local governments experimented heav- ily with investments in quasi-public/private development projects (known as internal improvements), usually in the form of bank, canal, and later railroad investment. Much of this initial experimentation came during the 1820s and 1830s in response to the success that the state of New York had with its heavy financial involvement in the construction of the Erie Canal. There, New York had used public funds and debt finance to support the canal’s construction.

The belief in supporting the effort (and those endeavors that other states later invested in) was that through such investment the state would be able to increase economic growth and development by promoting trade and commerce. It was also thought that once constructed the accrued debt would easily be paid off through the revenue generated by the internal improvement. Thus these investments were con- sidered to be virtually riskless. This was especially true as in the event that some of these internal improvements did not come to fruition it was also thought that under such an extreme, the obligations were ultimately the responsibility of the company that received the financial backing and not the state.

This led a number of states to pursue similar activities in their own economies, especially through bank and canal investment. Much of this aid to these internal improvements came through either explicit government guarantees of the debt issued by the private companies constructing the improvements, the outright purchase of pri- vate stock by those states, or through the issuing of state debt to be used in order to finance those projects (Pinsky 1963).5 Thus, between 1830 and 1840 total indebtedness of state governments increased by roughly $178,500,000 (Porter 1880).

Although this method of public finance was sustainable while economic activity was progressing, with the onset of economic downturn the situation could and would change. This occurred during the Panic of 1839, which began as a banking panic and rapidly spread across the country. As a result of this financial crisis, many of the internal improvement projects that had been invested in were abandoned as it was realized that they were not financially viable, while those that were completed generally did not net near the returns originally envisioned (Ratchford 1941). This led numerous publicly funded banks and internal improvement projects to fold, forcing many of those outstanding debt obligations onto the state governments that had either issued or guaranteed them. However, most states were in no position to cover those obligations. Thus, eight states and one territory defaulted on those debts, with five states eventually repudiating all or part of them.6

Comparable events also occurred during Civil War Reconstruction. Most former Confederate states, in an attempt to rebuild devastated economies, began to accrue large amounts of debt in order to fund internal improvement projects and also aid in the construction and reconstruction of railroads. However, these debt obligations

5 For an overview of state public finance policy during this period see Grinath et al. (1997). 6 The eight states were Arkansas, Illinois, Indiana, Louisiana, Maryland, Michigan, Mississippi, and Penn- sylvania. The territory was Florida. Of these Arkansas, Florida, Louisiana, Michigan, and Mississippi repudiated at least a portion of those outstanding debts.

123

190 J. A. Dove

began to mount rapidly, culminating in the eventual default and repudiation by every Confederate state on many of these obligations.7

The aftermath of these two episodes was one of constitutional change meant to significantly constrain future governments from pursuing similar behavior. This came in several forms including early versions of today’s tax and expenditure limits, proce- dural safeguards or outright prohibitions on the issuing of debt, the purchase of private stock, and the guaranteeing of private debt among several others.

For instance, in the wake of the Panic of 1839 and ensuing defaults, former governor of Indiana James Whitcomb in his annual message to the legislature of 1848 suggested that, “The constitution should be amended, [so] as to prohibit the creation of any public debt, except under restrictions as to amount, and object” (House Journal 1849). Further, former Democratic governor of New York, Silas Wright, suggested the following as to why the state saw a major push toward constitutional reform:

The recent free use of the public credit, in over-hastening loans for state works, and in lending to irresponsible corporations, the embarrassed condition of our finances, and the consequent call for direct taxation to restore public confidence, have doubtless given rise to these proposed [constitutional] amendments.” (Sen- ate Journal 1846).

Similar sentiments were echoed after the Southern defaults during Reconstruction. For example, former Chief Justice of the Alabama Supreme Court and member of the state’s constitutional convention of 1875, Samuel Davies Weakly, noted that the purpose of the convention was:

[T]he prevention of appropriations for any but public purposes, the saving of the credit of the state …by denying it use in aid of private enterprise …and the preservation of the property of a citizen from large tax levies.” As quoted in McMillan (1955, p. 234).

Although states were significantly constrained from investing in internal improve- ment projects municipal governments, beginning in the 1850s, started to invest heavily in many of those same projects, especially in railroad aid. Interestingly, although the constitutional constraints that had been imposed limited the ability of states to pursue those policies, most did not prohibit municipal governments from doing so. Thus, a significant amount of municipal aid poured into railroad investment, in much the same manner as state aid had poured into internal improvement projects previously.

As Table 1 indicates this led to a substantial increase in outstanding municipal debt, which was again sustainable so long as the economy was growing. However, with economic downturn this situation rapidly deteriorated. The downturn came with the onset of the Panic of 1873, which culminated in nearly 20% of all outstanding municipal debt entering a state of default during that period (Hillhouse 1936). The aftermath of this episode followed closely to that of the Panic of 1839, with numerous states passing constitutional amendments akin to those imposed on state governments earlier. These fiscal constitutions were largely meant to restrict municipal governments

7 See Summers (1984) for a full historical account of this period and the economic policies Southern states pursued.

123

Do fiscal constraints prevent default? Historical evidence… 191

Table 1 State and local indebtedness in dollars

Year State debt % Change Local debt % Change Combined debt % Change

1840 176,819,394a – 27,536,422b – 20,43,55,816

1850 192,527,913c 8.88 d – 19,25,27,913 −5.78789644 1860 25,74,06,940 33.70 200,000,000e – 45,74,06,940 137.5795452

1870 35,28,66,698 37.09 51,58,10,060 158 86,86,76,758 89.91333144

1880 27,47,45,772 −22.14 84,85,32,875 65 1,12,32,78,647 29.30916324 1890 21,12,10,487 −23.13 92,59,89,603 9 1,13,72,00,090 1.239357931 Data from U.S. Census Bureau 1870, 1880, and 1890 a Data interpolated from 1839 and 1841 data b Data for 1843 c Data for 1853 d No data available e Estimated

from irresponsible public investment and to also minimize potential future defaults. For instance, as an editorial in the New York Times pointed out regarding municipal safeguards in the state of New York:

The principal objects of these amendments are to limit the power of the cities of the State to create debt, and to surround the keeping and outlay of the public money with more effectual safeguards than are now provided.” (NYT Oct. 9, 1877).

Table 2 provides a list of the four main constraints that states imposed on municipal- ities by state and year of adoption, and are the limits that this paper specifically seeks to analyze.8 These include procedural safeguards on the issuing of debt (DebtGuard), outright prohibitions on the issuing of debt above a specified limit (DebtProhibit), binding tax limits based on a percentage of assessed property values (TaxLimit), and finally hard budget constraints (HardBudget). In order to employ these variables in a manner suitable for empirical analysis, each is converted into a dummy variable represented by a “1” if a state had adopted the given constraint and “0” otherwise.

The first two limits stipulated some binding cap on the amount of debt that a municipality could accumulate. For those municipalities that faced a strict prohibition, it meant that they could not increase their total indebtedness above that limit, while other municipalities could, but only after meeting some procedural hurdle. Many of these procedural safeguards required at least a majority, and in many instances a supermajority of the citizens within a given municipality to vote in the affirmative to increase the debt limit above the specified cap. Others required legislative approval, while some required both citizen and legislative approval.

Further, municipal tax limits generally tied tax rates to a certain percentage of assessed property values, while the hard budget constraints explicitly prohibited state

8 The bulk of these amendments were compiled from the NBER/University of Maryland State Constitu- tion Project (Wallis 2015). When constitutions were not available, the remaining amendments were hand collected.

123

192 J. A. Dove

T ab

le 2

M un ic ip al co ns ti tu ti on al re st ri ct io ns

by st at e an d ye ar

S ta te

D eb tP ro h ib it

D eb tG

u a rd

H a rd B u d ge t

Ta xL

im it

C en su s R eg io n

S ta te

D eb tP ro h ib it

D eb tG

u a rd

H a rd B u d ge t

Ta xL

im it

C en su s re g io n s

C on

ne ct ic ut

N or th ea st

Il li no

is 18

70 18

70 18

70 M id w es t

D el aw

ar e

N or th ea st

In di an a

18 81

18 51

M id w es t

M ai ne

18 78

N or th ea st

Io w a

18 46

18 46

M id w es t

M as sa ch us et ts

N or th ea st

K an sa s

M id w es t

N ew

H am

ps hi re

N or th ea st

M ic hi ga n

M id w es t

N ew

Je rs ey

N or th ea st

M in ne so ta

18 72

M id w es t

N ew

Y or k

18 73

18 73

N or th ea st

M is so ur i

19 00

18 75

18 75

M id w es t

P en ns yl va ni a

18 74

18 57

N or th ea st

N eb ra sk a

18 75

M id w es t

R ho

de Is la nd

N or th ea st

N or th

D ak ot a*

18 89

18 89

M id w es t

V er m on t

N or th ea st

O hi o

M id w es t

A la ba m a

18 67

S ou

th S ou

th D ak ot a*

18 89

18 89

18 89

M id w es t

A rk an sa s

18 68

18 68

S ou

th W is co ns in

18 74

M id w es t

F lo ri da

S ou

th A ri zo na

W es t

G eo rg ia

18 77

18 77

S ou

th C al if or ni a

18 79

W es t

K en tu ck y

18 91

18 91

18 91

S ou

th C ol or ad o

18 76

W es t

L ou

is ia na

18 94

18 79

18 78

S ou

th Id ah o*

18 90

18 90

18 90

W es t

M ar yl an d

18 67

S ou

th M on

ta na *

18 89

18 89

W es t

123

Do fiscal constraints prevent default? Historical evidence… 193

T ab

le 2

co nt in ue d

S ta te

D eb tP ro h ib it

D eb tG

u a rd

H a rd B u d ge t

Ta xL

im it

C en su s R eg io n

S ta te

D eb tP ro h ib it

D eb tG

u a rd

H a rd B u d ge t

Ta xL

im it

C en su s re g io n s

M is si ss ip pi

S ou th

N ev ad a

18 64

W es t

N or th

C ar ol in a

18 68

S ou th

N ew

M ex ic o

W es t

O kl ah om

a S ou

th O re go

n 18

57 18

57 W es t

S ou

th C ar ol in a

18 96

S ou

th U ta h

18 95

18 95

W es t

T en ne ss ee

S ou

th W as hi ng

to n*

18 89

18 89

W es t

T ex as

18 76

18 83

S ou

th W yo

m in g*

18 89

18 89

W es t

V ir gi ni a

19 02

18 70

S ou

th

W es t V ir gi ni a

18 72

18 63

18 72

S ou

th

* D at es

li st ed

co rr es po nd

to th e fi rs t ap pe ar an ce

in a st at e co ns ti tu ti on .H

ow ev er ,a

nu m be r of

st at es

ad op te d th es e co ns tr ai nt s in to

te rr it or ia l or di na nc es

an d ch ar te rs pr io r to

st at eh oo

d, m ea ni ng

th ey

ar e m uc h ol de r th an

th e da te li st ed

123

194 J. A. Dove

governments from assuming the debts of their respective municipal governments. This latter prohibition came as a direct result of several states, including Illinois and New York, having had various bills introduced into their respective legislative chambers which would have mandated that the state government assume any existing defaulted municipal debt. The New York Times suggested the following regarding Illinois’ legislative attempt to assume defaulted municipal debt:

The injustice of such a measure will be readily perceived. Cities and counties entirely free, or nearly so, from debt will be compelled to pay the debts of other cities and counties. This is a matter in which, of course, certain speculators are greatly interested. (NYT Jan. 3, 1875).

This issue is what the remainder of the paper intends to evaluate. Specifically, did these constraints live up to their stated goal of limiting municipal debt and default? Importantly, the Panic of 1893 and the period around that episode offer an opportunity to empirically evaluate that very question as the Panic of 1893 led to an even more severe depression than did the Panic of 1873, yet it saw fewer state and municipal financial difficulties thandidthePanics of 1839and1873.9 As noted, whereas thePanic of 1873 left nearly 20% of all outstanding municipal debt in default, the aftermath of the Panic of 1893 saw less than 10% of all outstanding debt in default, with no states defaulting at any point (Hillhouse 1936). This even given the fact that next to the Great Depression, the panic and depression of the 1890s was the deepest and most severe in U.S. history, with national unemployment rates above 11% for five years running (Romer 1986).

4 Data and model specification

In order to empirically evaluate how the above mentioned fiscal constraints may have affected the likelihood of municipal default I have taken a list of all the known munic- ipal defaults between 1890 and 1905. This information was drawn from Hillhouse (1930), who compiled those defaults between 1830 and 1930 in the United States, gathered from a number of first hand sources.10 Although this source does have some limitations in that it does not include unreported municipal defaults over the period, nor does it necessarily differentiate between a missed or delayed interest payment versus a principal payment, it is the most comprehensive list of nineteenth and early twentieth century municipal defaults in existence.11

9 For important historical analyses of the Panic of 1893 and the ensuing depression see Hoffman (1970), Steeples and Whitten (1998), and White (1982). 10 Importantly, although municipal defaults are available to 1930, after 1905 states again went through an extended period of constitutional revision. Thus, extending the analysis beyond 1905 brings in endogeneity concerns that can otherwise be largely avoided over the sample period explored. Further, in order to alleviate any endogeneity problems that may exist, those constraints that were adopted by a state through the period under analysis are dropped. For example, Louisiana passed DebtGuard in 1894. Therefore, all Louisiana DebtGuard observations are excluded from the entire analysis 1890 to 1905. 11 Per Hillhouse, “Any failure to meet an interest or principle payment on time was treated as a default” (1930, pg. iii).

123

Do fiscal constraints prevent default? Historical evidence… 195

Hillhouse (1930) included all listings of municipal defaults from the Bond Buyer Magazine, The Commercial and Financial Chronicle, state case law, statutes, and various histories. Here I look at all county and municipal defaults over the sample, but exclude any school district defaults. This was done as school districts normally fell under separate rules and requirements apart from what general municipal governments faced.

Additionally I aggregate these defaults to the state level and specifically consider the total defaults that occur within each state by year over the given sample. This aggregation is largely the result of a dearth of municipal-level data during this period. However, aggregating in this manner provides an opportunity to evaluate the proba- bility that a particular fiscal constraint may increase or decrease municipal default and thus provide an overarching understanding of how these constraints affected municipal fiscal health.

Based on this analysis then Table 3 shows all municipal defaults by state, the number of repeat municipal defaults that occurred by state (where repeat defaults are those municipal governments that defaulted again in the future after an initial default), along with the percentage of total and repeat municipal defaults by state, and finally the total and percentage of each state aggregated by Census region.12

The table shows that the state with the largest percentage of total and repeat default- ers was Kansas with almost 15 and 18% of each respective default occurring in that state. Further, the Midwest region was clearly the group of states with the largest per- centage of defaults with almost 50% of total and 52% of repeat defaulters existing in that region. Finally, the Northeastern states had the fewest defaults with 6 and 4% of total and repeat defaulters.

Next, Fig. 1 tracks both the total and repeat defaulters by year from 1890 to 1905. As shown, total municipal defaults begin to rise significantly in 1893 and peak in 1897 before generally declining thereafter. This pattern appears to track national economic trends through this period. The actual depression that emerged from the panic persisted through the 1890’s, as best indicated by unemployment rates. Based on Romer (1986) unemployment rose from 3.7 to 8.1% between 1892 and 1893, peaking at 12.4% in 1897 and falling to 5% by 1900.

Given the non-negative count nature of the default data employed and the significant number of no defaults observed in a number of states, OLS regression analysis would be inappropriate. Thus, a better model would be either a Poisson or Negative Binomial to account for this issue. Due to detected overdispersion within the results, I opt for the Negative Binomial as the obtained results are more efficient than those found with a Poisson model, while it also allows for predicted probabilities to be obtained (here the probability that a given fiscal constraint will increase or decrease municipal default).13

12 The U.S. Census Bureau groups states into specific geographic regions for statistical reporting purposes. In this paper I include the four major and broadest regions used by the Census Bureau, which are the Northeast, South, Midwest, and West regions. 13 For a more thorough discussion of these issues see Cameron and Trivedi (2013).

123

196 J. A. Dove

T ab

le 3

T ot al m un ic ip al de fa ul ts by

st at e an d ge og ra ph ic re gi on

S ta te

T ot al de fa ul ts

R ep ea t de fa ul ts

% A ll de fa ul ts

% R ep ea t R eg io n

S ta te

T ot al de fa ul ts

R ep ea t de fa ul ts

% A ll de fa ul ts

% R ep ea t R eg io n

C on

ne ct ic ut

0 0

0 0

N or th ea st

Il li no

is 31

11 5. 25

4 6. 11

1 M id w es t

D el aw

ar e

1 0

0. 16

9 0

N or th ea st

In di an a

12 1

2. 03

4 0. 55

6 M id w es t

M ai ne

1 0

0. 16 9

0 N or th ea st

Io w a

17 4

2. 88 1

2. 22 2

M id w es t

M as sa ch us et ts

1 0

0. 16 9

0 N or th ea st

K an sa s

86 32

14 .5 76

17 .7 78

M id w es t

N ew

H am

ps hi re

0 0

0 0

N or th ea st

M ic hi ga n

11 0

1. 86 4

0 M id w es t

N ew

Je rs ey

5 1

0. 84 7

0. 55 6

N or th ea st

M in ne so ta

12 1

2. 03 4

0. 55 6

M id w es t

N ew

Y or k

22 5

3. 72 9

2. 77 8

N or th ea st

M is so ur i

27 10

4. 57 6

5. 55 6

M id w es t

P en ns yl va ni a

8 1

1. 35 6

0. 55 6

N or th ea st

N eb ra sk a

28 10

4. 74 6

5. 55 6

M id w es t

R ho de

Is la nd

0 0

0 0

N or th ea st

N or th

D ak ot a

0 0

0 0

M id w es t

V er m on

t 0

0 0

0 N or th ea st

O hi o

41 15

6. 94

9 8. 33

3 M id w es t

A la ba m a

14 7

2. 37

3 3. 88

9 S ou

th S ou

th D ak ot a

14 5

2. 37

3 2. 77

8 M id w es t

A rk an sa s

0 0

0 0

S ou th

W is co ns in

10 4

1. 69 5

2. 22 2

M id w es t

F lo ri da

0 0

0 0

S ou

th A ri zo na

6 4

1. 01

7 2. 22

2 W es t

G eo rg ia

1 0

0. 16

9 0

S ou

th C al if or ni a

49 24

8. 30

5 13

.3 33

W es t

K en tu ck y

16 2

2. 71

2 1. 11

1 S ou

th C ol or ad o

23 10

3. 89

8 5. 55

6 W es t

L ou

is ia na

0 0

0 0

S ou

th Id ah o

3 1

0. 50

8 0. 55

6 W es t

M ar yl an d

0 0

0 0

S ou

th M on

ta na

0 0

0 0

W es t

M is si ss ip pi

3 0

0. 50 8

0 S ou th

N ev ad a

1 0

0. 16 9

0 W es t

N or th

C ar ol in a

13 1

2. 20

3 0. 55

6 S ou

th N ew

M ex ic o

7 4

1. 18

6 2. 22

2 W es t

123

Do fiscal constraints prevent default? Historical evidence… 197

T ab

le 3

co nt in ue d

S ta te

T ot al de fa ul ts

R ep ea t de fa ul ts

% A ll de fa ul ts

% R ep ea t

R eg io n

S ta te

T ot al de fa ul ts

R ep ea t de fa ul ts

% A ll de fa ul ts

% R ep ea t

R eg io n

O kl ah om

a 1

0 0. 16

9 0

S ou

th O re go

n 1

0 0. 16

9 0

W es t

S ou

th C ar ol in a

21 10

3. 55

9 5. 55

6 S ou

th U ta h

3 1

0. 50

8 0. 55

6 W es t

T en ne ss ee

10 4

1. 69

5 2. 22

2 S ou

th W as hi ng

to n

13 4

2. 20

3 2. 22

2 W es t

T ex as

52 13

8. 81

4 7. 22

2 S ou

th W yo

m in g

0 0

0 0

W es t

V ir gi ni a

3 0

0. 50

8 0

S ou

th

W es t V ir gi ni a

1 0

0. 16 9

0 S ou th

N or th ea st

38 7

6. 44

1 3. 88

9 M id w es t

29 0

93 49

.1 53

51 .6 67

S ou

th 13

5 37

22 .8 81

20 .5 56

W es t

10 6

48 17

.9 66

26 .6 67

123

198 J. A. Dove

Fig. 1 Total municipal defaults by year 1890–1905

The model takes the following form.14

E ( Defaultit |αi , Xit

) = αieX ′ it β (1)

where i = 1, 2, . . . , 46 and t = 1890, 1891, . . . , 1905. Parameter estimates are obtained from the above equations, while finding

∂ E(Defaultit |αi ,Xit ) ∂ Xit j

nets the marginal effects from the model. Each of the main variables of interest (DebtProhibit, DebtGuard, HardBudget,

and TaxLimit) are as discussed in the previous section. Defaultit is a count variable representing the number of municipal defaults that were reported in state i in year t. The analysis also includes a set of socioeconomic control variables, along with a set of year fixed effects to control for any unobservable differences that may arise over time.15 First I control for the number of incorporated local governments in each state. This information was taken from U.S. Census data for 1890, 1900, and 1910 with all off years interpolated. This control variable is included as, presumably, an increase in the number of total local governments will increase the likelihood that defaults will occur.

Similarly, I also include total population as another control along with the per- cent of the population that is urban. This latter variable may help to proxy for the level of economic development in a state, and may also negatively impact municipal default as a more urbanized population could require additional municipal services, which may lead to a strain on municipal finances. However, a larger urban pop- ulation might also increase the tax base available to municipal governments and

14 This model largely follows from Blundell et al. (1995), Allison and Waterman (2002), and Cameron and Trivedi (2013). For the sake of space it is presented in a much reduced form here. 15 Due to the time invariant nature of most of the main independent variables involved, state fixed effects could not be incorporated into the model.

123

Do fiscal constraints prevent default? Historical evidence… 199

thus decrease the likelihood of default. Thus, the impact of this variable should be ambiguous. Further, I include a control variable for the number of years since a state formally entered the Union. This variable should also help proxy for overall eco- nomic development, as older states are generally more integrated with the national economy relative to more newly admitted states. Additionally, I include per capita state debt to proxy for fiscal capacity within a state. However, its overall impact on municipal default may be ambiguous. A larger debt burden may leave fewer resources available to municipal governments which may increase the likelihood of default. However, it may also be the case that the provision of public goods is more central- ized in these states, which places a smaller fiscal burden on municipal governments overall.

I also include two variables to proxy for output and also personal income. The former is represented by the annual average manufacturing output per worker (in dollars) while the latter is represented by the annual average per worker manufac- turing wage. Both of these were taken from Census data for 1890, 1900, and 1910 again with off years interpolated. Two additional variables are whether a state granted home rule authority to municipal governments and also the percentage of repeat municipal defaults that occurred over a five year running average. Through the lat- ter half of the nineteenth century states began granting home rule authority to their respective municipal governments, which significantly decentralized authority and autonomy to those local governments. This also meant a more decentralized pub- lic provision of goods and services. Therefore, it would be expected that home rule authority would increase municipal fiscal burdens and thus, ceteris paribus, increase the likelihood of default. Further, it may also be the case that a current default by a municipal government may be positively correlated with an increased likelihood of future default. Table 3 would seem to indicate that this is the case. Therefore, I include a five year moving average of the defaulted municipalities over the sample. Thus, for 1890 this variable takes the percentage of municipalities that defaulted between 1885 and 1889, while for 1891 it takes the average between 1886 and 1890.16

Finally, I include several geographic variables, which are regional dummies based on the U.S. Census regions as discussed previously. This final set of variables should help alleviate at least some of the possible omitted and unobservable factors that a state fixed effect might be able to pick up but, as discussed, would be impossible to include in the analysis given the time invariant nature of many of the main indepen- dent variables of interest. Obviously, Table 3 clearly indicates regional patterns in municipal default, which would indicate that controlling for those regions would also be important. Table 4 provides the summary statistics of all of the variables discussed above.

16 This method is employed as including the lagged dependent variable would lead to coefficients that are inconsistent due to the time-invariance of the fiscal constraint variables. Therefore, including the current variable for repeat defaults should help avoid inconsistent estimates and should also help address any potential serial correlation.

123

200 J. A. Dove

Table 4 Summary statistics

Variable Observations Mean SD Min Max

Municipal defaults 768 0.740 1.574 0 18

DebtProhibit 720 0.361 0.481 0 1

DebtGuard 720 0.168 0.374 0 1

HardBudget 736 0.391 0.488 0 1

TaxLimit 752 0.245 0.430 0 1

Total incorporated municipalities (in 100s) 768 2.033 2.117 0.030 9.986

Population (in 100,000s) 768 15.079 14.785 0.423 81.913

Per capita debt (in 1,000s) 768 391.477 447.015 0 2269.695

% Urban 758 31.163 21.786 2.640 95.900

Annual average per worker manufacturing wage (in $)

768 2.434 0.838 0.148 5.392

Annual average per worker manufacturing output (in $)

768 2.951 3.568 0.630 37.553

Years since statehood 768 63.600 40.324 0 118

Homerule state (1 = yes) 768 0.078 0.269 0 1

% Repeat defaults (5-year average) 768 1.290 2.531 0 20

Northeast 768 0.229 0.421 0 1

South 768 0.292 0.455 0 1

Midewest 768 0.250 0.433 0 1

West 768 0.229 0.421 0 1

5 Results and interpretation

Overall, theresultsprovidesomeinterestingfindingsandimplications.Table5presents a very cursory inspection by presenting the simple difference-in-averages results for each of the main independent variables of interest.

Here both DebtProhibit and TaxLimit appear to increase the likelihood of default by 11 and 39% respectively and both are statistically significant. Further, HardBudget and DebtGuard indicate the opposite effect, with both reducing the likelihood of default by 3% though neither are statistically significant. These correlations would at least suggest that the introduction of these rules is not random, but rather may be indicative of their adoption in states with more fiscally problematic local governments. Along with Table 5, Table 6 shows a much more comprehensive analysis based on the discussion from the previous section.

All coefficients shown are the marginal effects from the above model. Columns 1 through 4 list all of the results for each of the main independent variables of interest run separately and without any controls, save the control for the number of incorporated municipal governments within a state. Now the results indicate that DebtProhibit (column 1) reduces the likelihood that a given municipality within a state will default by over 13% and is statistically significant.

123

Do fiscal constraints prevent default? Historical evidence… 201

Table 5 Difference in average results

Independent variable = Number of municipal defaults per year

(1) (2) (3) (4)

DebtProhibit (1 = yes) 0.106**

(0.132)

DebtGuard (1 = yes) −0.028 (0.175)

HardBudget (1 = yes) −0.027 (0.047)

TaxLimit (1 = yes) 0.386***

(0.133)

Year fixed effect Y Y Y Y

Observations 720 720 720 720

Log likelihood −717.691 −747.498 −753.736 −706.793 Standard errors in parentheses *** p < 0.01; ** p < 0.05; * p < 0.1

Next, DebtGuard (column 2) actually suggests the opposite, netting a positive result which would indicate that the presence of such a restriction actually increases the likelihood that a given municipality will default by 7%, though the result is not statistically significant. Column 3 provides the result for HardBudget which shows that this constraint reduces the likelihood of default by almost 22% with the result highly statistically significant. Finally, column 4 indicates that the presence of TaxLimit actually increases the likelihood that a municipality defaults by 15% and is statistically significant as well.

Next columns 5 through 8 present the results for each of the main independent variables run separately, but now including all of the control variables discussed above while excluding the geographic variables. Here the findings are extremely similar to those found in columns 1 through 4, with each of the sign coefficients on the main variables being identical. Specifically, DebtProhibit decreases the likelihood that a municipality defaults by almost 21%, while DebtGuard again increases that likelihood by 5%, whereas HardBudget decreases the likelihood by almost 27%, and finally TaxLimit increases the likelihood by 3%. Here again both DebtProhibit and HardBudget are statistically significant, while DebtGuard and TaxLimit are not.

Next columns 9 through 12 again show the results for each of the main indepen- dent variables of interest regressed seperately, but now including all of the control variables. Again, the findings are qualitatively similar to those previously shown, with DebtProhibit decreasing the likelihood of default by 16%, DebtGuard increasing the likelihood by 20%, HardBudget decreasing the likelihood by 23%, and TaxLimit increasing the likelihood of default by 2%. As with columns 1 through 4, DebtPro- hibit and HardBudget are statistically significant, while DebtGuard and TaxLimit are insignificant.

123

202 J. A. Dove

T ab

le 6

N eg at iv e bi no m ia l re gr es si on

es ti m at es

In de pe nd en t va ri ab le = N um

be r of

m un ic ip al de fa ul ts pe r ye ar (1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

D eb tP ro hi bi t (1

= ye s)

−0 .1 33 **

−0 .2 07 ** *

(0 .0 54 5)

(0 .0 75 6)

D eb tG ua rd

(1 = ye s)

0. 07 03

0. 05 18

(0 .0 74 0)

(0 .0 88 7)

H ar dB

ud ge t (1

= ye s)

−0 .2 16 ** *

−0 .2 65 ** *

(0 .0 53 4)

(0 .0 62 1)

T ax L im

it (1

= ye s)

0. 15 2* **

0. 02 96

(0 .0 60 7)

(0 .0 55 1)

T ot al in co rp or at ed

m un ic ip al it ie s (i n 10 0s )

0. 09 75 ** *

0. 08 58 ** *

0. 10 1* **

0. 07 86 ** *

0. 09 66 ** *

0. 08 14 ** *

0. 10 1* **

0. 08 00 ** *

(0 .0 14 6)

(0 .0 12 0)

(0 .0 14 0)

(0 .0 11 3)

(0 .0 22 5)

(0 .0 18 4)

(0 .0 21 0)

(0 .0 17 9)

P op ul at io n (i n 10 0, 00 0s )

0. 01 05 ** *

0. 00 81 7* **

0. 00 94 0* **

0. 00 71 4* *

(0 .0 03 42 )

(0 .0 02 86 )

(0 .0 02 87 )

(0 .0 02 94 )

P er

ca pi ta de bt

(i n 1, 00 0s )

−0 .0 00 34 5* **

−0 .0 00 18 6* *

−0 .0 00 22 3* *

−0 .0 00 19 6* *

(0 .0 00 12 0)

(8 .7 5e -0 5)

(9 .1 9e -0 5)

(8 .6 2e -0 5)

% U rb an

−0 .0 06 71 ** *

−0 .0 03 75 *

−0 .0 05 33 **

−0 .0 03 17

(0 .0 02 50 )

(0 .0 02 05 )

(0 .0 02 25 )

(0 .0 01 97 )

A nn ua l av er ag e pe r w or ke r m an uf ac tu ri ng

w ag e

−0 .0 48 8

0. 05 14

0. 07 25

0. 06 49

(0 .0 53 0)

(0 .0 44 5)

(0 .0 44 3)

(0 .0 39 9)

A nn ua l av er ag e pe r w or ke r m an uf ac tu ri ng

ou tp ut

(i n $)

−0 .0 12 4*

−0 .0 05 18

−0 .0 01 48

−0 .0 03 56

(0 .0 06 77 )

(0 .0 05 87 )

(0 .0 05 79 )

(0 .0 05 73 )

Y ea rs si nc e st at eh oo d

−0 .0 01 27

−0 .0 01 85

−0 .0 02 59 **

−0 .0 01 78

(0 .0 01 34 )

(0 .0 01 20 )

(0 .0 01 27 )

(0 .0 01 16 )

H om

er ul e st at e (1

= ye s)

0. 76 4* **

0. 38 5* **

0. 55 3* **

0. 38 5* **

(0 .2 78 )

(0 .1 40 )

(0 .1 72 )

(0 .1 35 )

123

Do fiscal constraints prevent default? Historical evidence… 203

T ab

le 6

co nt in ue d

In de pe nd en t va ri ab le = N um

be r of

m un ic ip al de fa ul ts pe r ye ar

(1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

% R ep ea t de fa ul ts (5 -y ea r av er ag e)

0. 06 4* **

0. 06 39 ** *

0. 06 2* **

0. 06 3* **

(0 .0 13 )

(0 .0 12 )

(0 .1 19 )

(0 .0 12 )

W es t

S ou th

M id w es t

P ea rs on

ov er di sp er si on

te st

1. 81 2

2. 77 0

3. 11 3

3. 05 4

1. 95 9

2. 04 7

1. 83 4

2. 00 8

Y ea r fi xe d ef fe ct

Y Y

Y Y

Y Y

Y Y

O bs er va ti on s

72 0

72 0

73 6

75 2

71 0

71 0

72 6

74 2

L og

li ke li ho od

−6 66 .8 90

−6 97 .0 47

−6 92 .0 14

−7 06 .7 93

−5 93 .3 74

−6 28 .2 49

−6 15 .6 79

−6 38 .4 27

(9 )

(1 0)

(1 1)

(1 2)

(1 3)

(1 4)

(1 5)

D eb tP ro hi bi t (1

= ye s)

−0 .1 64 **

−0 .1 06 *

−0 .1 53 *

−0 .1 20

(0 .0 75 5)

(0 .0 60 4)

(0 .0 78 5)

(0 .0 83 0)

D eb tG ua rd

(1 = ye s)

0. 20 3

0. 11 2

0. 31 5*

0. 33 8

(0 .1 35 )

(0 .0 84 1)

(0 .1 68 )

(0 .2 12 )

H ar dB

ud ge t (1

= ye s)

−0 .2 34 ** *

−0 .2 17 ** *

−0 .3 39 ** *

−0 .2 61 ** *

(0 .0 63 9)

(0 .0 58 0)

(0 .0 81 7)

(0 .0 82 2)

T ax L im

it (1

= ye s)

0. 02 43

0. 25 0* **

0. 26 2* **

0. 29 8* *

(0 .0 64 5)

(0 .0 79 3)

(0 .1 01 )

(0 .1 22 )

T ot al in co rp or at ed

m un ic ip al it ie s (i n 10 0s )

−0 .0 35 4

−0 .0 34 4

−0 .0 09 85

−0 .0 34 8

0. 10 3* **

0. 12 5* **

−0 .0 08 29

(0 .0 28 1)

(0 .0 25 1)

(0 .0 25 9)

(0 .0 24 9)

(0 .0 15 6)

(0 .0 26 8)

(0 .0 29 8)

P op ul at io n (i n 10 0, 00 0s )

0. 02 19 ** *

0. 01 82 ** *

0. 01 92 ** *

0. 01 70 ** *

0. 01 01 ** *

0. 02 01 ** *

(0 .0 05 51 )

(0 .0 04 34 )

(0 .0 04 58 )

(0 .0 04 25 )

(0 .0 03 62 )

(0 .0 05 54 )

123

204 J. A. Dove

T ab

le 6

co nt in ue d

In de pe nd en t va ri ab le = N um

be r of

m un ic ip al de fa ul ts pe r ye ar (9 )

(1 0)

(1 1)

(1 2)

(1 3)

(1 4)

(1 5)

P er

ca pi ta de bt

(i n 1, 00 0s )

−0 .0 00 41 3* **

−0 .0 00 24 6* *

−0 .0 00 30 8* **

−0 .0 00 28 1* **

−0 .0 00 35 0* **

−0 .0 00 35 4* *

(0 .0 00 14 5)

(0 .0 00 10 8)

(0 .0 00 11 5)

(0 .0 00 10 8)

(0 .0 00 12 6)

(0 .0 00 14 4)

% U rb an

−0 .0 04 57

0. 00 12 3

−0 .0 01 68

0. 00 04 61

−0 .0 07 26 **

−0 .0 05 39

(0 .0 03 85 )

(0 .0 03 16 )

(0 .0 03 31 )

(0 .0 03 01 )

(0 .0 02 90 )

(0 .0 04 20 )

A nn ua l av er ag e pe r w or ke r m an uf ac tu ri ng

w ag e

−0 .0 11 6

0. 06 86

0. 08 90 *

0. 10 4* *

−0 .0 37 1

−0 .0 06 96

(0 .0 58 2)

(0 .0 49 7)

(0 .0 50 9)

(0 .0 47 9)

(0 .0 59 4)

(0 .0 65 2)

A nn ua l av er ag e pe r w or ke r m an uf ac tu ri ng

ou tp ut

(i n $)

0. 00 95 0

0. 01 63 **

0. 01 84 **

0. 01 80 **

−0 .0 04 53

0. 01 51

(0 .0 08 26 )

(0 .0 07 70 )

(0 .0 07 76 )

(0 .0 07 76 )

(0 .0 06 96 )

(0 .0 09 30 )

Y ea rs si nc e st at eh oo d

0. 00 29 5

0. 00 09 44

0. 00 05 81

0. 00 18 1

−0 .0 03 18 *

0. 00 12 2

(0 .0 01 85 )

(0 .0 01 68 )

(0 .0 01 63 )

(0 .0 01 75 )

(0 .0 01 65 )

(0 .0 02 53 )

H om

er ul e st at e (1

= ye s)

0. 74 3* *

0. 17 9*

0. 36 5* **

0. 20 3*

0. 57 2* *

0. 56 5*

(0 .2 97 )

(0 .1 05 )

(0 .1 40 )

(0 .1 08 )

(0 .2 67 )

(0 .2 94 )

% R ep ea t de fa ul ts (5 -y ea r av er ag e)

0. 07 4* **

0. 07 3* **

0. 06 6* **

0. 06 9* **

0. 06 2* **

0. 07 3* **

(0 .0 16 )

(0 .0 15 )

(0 .0 14 )

(0 .0 14 )

(0 .0 14 )

(0 .0 17 )

W es t

1. 21 8* *

1. 22 2*

1. 57 9* *

1. 51 7* *

0. 98 3*

(0 .5 74 )

(0 .6 29 )

(0 .7 31 )

(0 .7 13 )

(0 .5 90 )

S ou th

0. 74 4* *

0. 91 6* *

0. 95 1* *

0. 79 7* *

0. 61 1

(0 .3 77 )

(0 .4 24 )

(0 .4 38 )

(0 .3 79 )

(0 .3 96 )

M id w es t

2. 24 8* **

2. 29 1* **

1. 98 7* **

2. 23 8* **

1. 97 2* *

(0 .8 45 )

(0 .8 31 )

(0 .7 52 )

(0 .8 37 )

(0 .8 31 )

P ea rs on

ov er di sp er si on

te st

1. 82 7

1. 88 6

1. 73 6

1. 82 5

2. 79 2

1. 87 4

1. 78 5

Y ea r fi xe d ef fe ct

Y Y

Y Y

Y Y

Y

O bs er va ti on s

71 0

71 0

72 6

74 2

67 2

66 2

66 2

L og

li ke li ho od

−5 70 .9 26

−6 03 .6 53

−5 97 .1 72

−6 15 .5 64

−6 14 .3 07

−5 40 .2 73

−5 23 .7 55

123

Do fiscal constraints prevent default? Historical evidence… 205

Finally, columns 13 through 15 list the results when all of the main independent vari- ables of interest are regressed together, with column 13 excluding all control variables, column 14 including all controls except for the geographic variables, and column 15 including all control variables. Here, DebtProhibit is negative in all specifications and statistically significant in two, DebtGuard is positive but only statistically significant in one specification, HardBudget is negative and significant in every specification, while TaxLimit is positive and also significant in every specification.

Overall then, the results appear quite robust. DebtProhibit was statistically sig- nificant in 5 of 6 specifications and suggested that such a constraint reduced the likelihood of municipal default between 11 and 21% depending on the specifica- tion. Further, DebtGuard although only statistically significant in 1 specification, suggested the opposite, actually increasing the likelihood of default between 5 and 34% depending on the specification. HardBudget was statistically significant in 6 of 6 specifications, and suggested a reduction in the likelihood of default between 22 and 34%. Finally, TaxLimit was statistically significant in 4 of 6 specifications and generally suggested an increase in the likelihood of default between 2 and 30%.

These findings shed light on some interesting implications. First, relatively more binding debt limits would appear to have a larger impact especially relative to pro- cedural safeguards. Intuitively this would make sense, given that a binding limit strictly prohibits a municipal government from increasing debt levels beyond that limit, whereas a procedural safeguard might require some voting threshold to be met before debt could be increased above a given level. Although not necessarily surpris- ing, what is more interesting is the consistently positive sign coefficient present when debt safeguards are in place.

This result might actually run consistent with the historical record, as well as more recent research on debt limits and the voting rules employed (Persson and Tabellini 2004; Lee et al. 2014).17 Further, more recent research shows that supermajority voting rules may actually lead to inefficiently high levels of debt, given that the decisive voter under such a model has greater leverage as do other coalition groups that might be needed to approve the issue. Given this, it may be the case that the decisive voter was able to extract relatively more for himself and his constituents than would otherwise be the case for a less stringent voting rule.

Additionally, these constraints may have acted to increase the overall transparency of the entire budgetary process, especially when providing voters with an active and direct role in the process. This may have translated into greater voter support and acceptance of increased public debt through time, and may have contributed to those future defaults (see Besley (2006) and Alt et al. (2002) for a more detailed overview of this issue). Further analysis of these possibilities, and the exact mechanism through which this occurred although beyond the scope of this current work, would be worthy of future research.

17 The historical and anecdotal evidence would suggest that many times these procedural safeguards were abrogated through technicalities in the language of the voting rule itself. Although beyond the scope of this current study, for a good overview of how these procedural rules were circumvented see Hillhouse (1936).

123

206 J. A. Dove

Next, the results found for constitutionally imposed tax limits are also generally consistent with the academic literature. Here again, the findings suggest that a binding tax limit on a municipal government increased the likelihood that a municipality would default. In general, tax limits make it more difficult for municipal governments to raise the funds necessary to pay down any accrued or newly added debt, and would therefore increase the likelihood of a default. This would seem to be corroborated by much of the literature in Sect. 2, where tax limits have been found to increase borrowing costs and deficits, while expenditure limits generally have a countervailing effect.

Finally, one of the more interesting results is that found for the presence of hard budget constraints. As noted, in the wake of the Panic of 1873 several states were inundated with pressures and in some instances legislation that would have required the state government to assume the outstanding debts of defaulting municipalities, with notable examples in Illinois and New York. However, these events led a number of states to impose explicit hard budget constraints on municipal governments. The results in this study indicate that the imposition of such a prohibition may have led local governments to significantly curtail expenditures and borrowing, though this conjecturewouldrequirefargreaterfutureresearchinordertobetterevaluatetheactual impact.

Overall these findings may also have some ramifications for events that are unfold- ing today, especially at the municipal level within the U.S. Several high profile and relatively recent bankruptcies in Jefferson County, Alabama, Detroit, Michigan, Stock- ton, California, and Harrisburg, Pennsylvania among others have shed light on an increasing problem within municipal public finance. Although many of these prob- lems are tied to under and unfunded pension obligations promised to public sector employees, they emerged in the aftermath of financial crisis, just as municipal defaults occurred in the wake of the panics of 1873 and 1893.

In this context it is important to realize that municipal default or bankruptcy has not occurred anywhere near the scale today as it did during the previous periods discussed. Further, many of the constraints analyzed in this current study do still exist, albeit in altered from through constitutional, legislative, and judicial change. Given that, although beyond the scope of this current paper, it may be worthwhile to analyze how these constraints may have played a role in stabilizing or destabilizing current municipal finances through the Great Recession.

One additional consideration has been the potential erosion of hard budget con- straints that are either explicit or implicit. As the above results indicate, those hard budget constraints appeared to have been the most successful at minimizing municipal default through the Panic of 1893. However, as municipal pension obligations mount this may again raise calls for state or even federal bailouts from these obligations. This very possibility became an issue in 2012 with the federally proposed Detroit Growth and Stability Act (H.R. 2012) which would have provided $500 million in long-term debt assistance to the city at the height of its fiscal crisis. Though it never passed, it does show that the potential is there for this to become a larger issue in the future.

Combined then, each of these results not only shed light on the historical signifi- cance that these fiscal constitutions had, they also provide avenues for future research regarding their current status and potential impact, especially in the wake of the recent

123

Do fiscal constraints prevent default? Historical evidence… 207

defaults experienced by several high profile cities around the United States and also internationally.

6 Conclusion

This paper has explored an important, yet under researched issue regarding municipal finance and default. Specifically, this work has addressed how municipal fiscal con- straints might impact the likelihood of a municipal default from occurring. Through the nineteenth century state and local defaults and even repudiations were a relatively common occurrence during times of financial crisis. The aftermath of these episodes saw the passage of binding fiscal constitutions on both state and local governments, meant to limit and minimize the prospect of these events again occurring in the future.

The scope of this study then has been to evaluate how successful these restrictions were in achieving their stated goals. Overall, the results would suggest that binding pro- hibitions on the accumulation of debt, rather than procedural safeguards significantly reduced the likelihood that a municipal government would default. Interestingly, but consistent with the literature procedural safeguards actually increased the likelihood. Further, municipal tax limits also increased the likelihood that a municipal govern- ment might default. Finally, hard budget constraints seemed to have the most profound effect on municipal default, significantly reducing the overall likelihood.

Given this, these results shed light on an important yet under-researched era in U.S. fiscal history and also provide a number of avenues for future researchers to pursue. Specifically, it would be important to track how and why these constraints have evolved over time and also the role that they play today or have during other moments of municipal default during the twentieth century. Along with this current work, these issues may be better able to evaluate the impact and lasting effect that these institutions have had on state and local public finance.

References

Abrams BA, Dougan WR (1986) The effect of constitutional restraints on governmental spending. Public Choice 49(2):101–116

Alesina A, Bayoumi T (1996) The costs and benefits of fiscal rules: evidence from the U.S. states. NBER working paper no. 5614 Natl Bur Econ Res, Camb

Allison PD, Waterman RP (2002) Fixed-effects negative binomial regression models. Sociol Methodol 32(1):247–265

Alt JE, Lassen DD, Skilling D (2002) Fiscal transparency, gubernatorial approval, and the scale of govern- ment: evidence from the states. State Polit Policy Q 2(3):230–250

Alt JE, Lowry RC (1994) Divided government, fiscal institutions, and budget deficits: evidence from the states. Am Polit Sci Rev 88(4):811–828

Amiel L, Deller S, Stallman J, Maher C (2014) Does the restrictiveness of state tax and expenditure limitations affect state revenues and expenditures? Int J Public Adm 37(4):237–248

Bae S, Gais T (2007) The effects of state-level tax and expenditure limitations on revenues and expenditures. The Nelson A. Rockefeller Institute of Government, Albany, New York

Bails DG (1982) A critique on the effectiveness of tax-expenditure limitations. Public Choice 38(2):129–138 Bails DG (1990) The effectiveness of tax-expenditure limitations: a re-evaluation. Am J Econ Sociol

49(2):223–238

123

208 J. A. Dove

Bails DG, Tieslau MA (2000) The impact of fiscal constitutions on state and local expenditures. Cato J 20(2):255–277

Bayoumi T, Goldstein M, Woglom G (1995) Do credit markets discipline sovereign borrowers: evidence from the U.S. states. J Money Credit Bank 27(4–1):1046–1059

Besley T (2006) Principled agents? The political economy of good government. Oxford Univ Press, Oxford Blundell R, Griffith R, Van Reenen J (1995) Dynamic count data models of technological innovation. Econ

J 105(429):333–344 Bohn H, Inman RP (1996) Balanced-budget rules and public deficits: evidence from the U.S. states.

Carnegie-Rochester Conf Ser Public Policy 45:13–76 Cameron AC, Trivedi PK (2013) Regression analysis of count data, 2nd edn. Cambridge University Press,

Cambridge Dove JA (2012) Credible commitments and constitutional constraints: state debt repudiation and default in

nineteenth century america. Const Polit Econ 23(1):66–93 Dove JA (2014) Financial markets, fiscal constraints, and municipal debt: lessons and evidence from the

panic of 1873. J Inst Econ 10(1):71–106 Eichengreen B (1992) Should the maastricht treaty be saved?. Princet Stud in Int Financ #74 Princet Univ

Press, Princet Goodspeed TJ (2002) Bailouts in a federation. Int Tax Public Financ 9(4):409–421 Grinath A III, Wallis JJ, Sylla R (1997) Debt, default, and revenue structure: the american state debt crisis

in the early 1840s. NBER work pap no. h0097 Natl Bur Econ Res, Camb Hallerberg M, Wolffand GB (2008) Fiscal institutions, fiscal policy and sovereign risk premia in EMU.

Public Choice 136(3–4):379–396 Hempel GH (1964) The postwar quality of municipal bonds. Dissertation, Univ of MI Hillhouse AM (1930) Defaulted municipal bonds, 1830–1930. Munic Financ Off Assoc, Chicago Hillhouse AM (1936) Municipal bonds: a century of experience. Prentice Hall, Upper Saddle River Hoffmann C (1970) The depression of the nineties: an economic history. Grenwood Publ, Westport Hopland AO (2013) Central government control and fiscal adjustment: Norwegian evidence. Econ Gov

14(2):185–203 Hopland AO (2014) Voter information and electoral outcomes: the Norwegian list of shame. Public Choice

161(1–2):233–255 Hou Y, Smith DL (2006) A framework for understanding state balanced budget requirement systems:

reexamining distinctive features and an operational definition. Public Budg Financ 26(3):22–45 Hou Y, Smith DL (2010) Do state balanced budget requirements matter? Testing two explanatory frame-

works. Public Choice 145(1–2):57–79 Johnson CL, Kriz KA (2005) Fiscal institutions, credit ratings, and borrowing costs. Public Budg Financ

25(1):84–103 Journal of the House of Representatives of the State of Indiana (1849) John D. Defrees, Indianapolis Journal of the Senate of the State of New York (1846) E. Mack, Albany Joyce PG, Mullins DR (1991) The changing fiscal structure of the state and local public sector: the impact

of tax and expenditure limitations. Public Adm Rev 51(3):240–253 Kiewiet DR, Szakaly K (1996) Constitutional limitations on borrowing: an analysis of state bonded indebt-

edness. J Law Econ Organ 12(1):62–97 Kousser T, McCubbins MD, Moule E (2008) For whom the TEL tolls: can state tax and expenditure imits

effectively reduce spending? State Polit Policy Q 8(4):331–361 Kornai J, Maskin E, Roland G (2003) Understanding the soft budget constraint. J Econ Lit 41(4):1095–1136 Krol R (2007) The role of fiscal and political institutions in limiting the size of state government. Cato J

27(3):431–445 Lee D, Borcherding TE, Kang Y (2014) Public spending and the paradox of supermajority rule. South Econ

J 80(3):614–632 Lowry RC, Alt JE (2001) A visible hand? Bond markets, political parties, balanced budget laws, and state

government debt. Econ Polit 13(1):49–72 McGrane RC (1935) Foreign bondholders of american state debts. Macmillan, New York McMillan MC (1955) Constitutional development in alabama 1798–1901: a study in politics, the negro and

sectionalism. The James Sprunt studies in history and political science, vol 37. University of North Carolina Press, Chapel Hill, NC

Mullins DR, Joyce PG (1996) Tax and expenditure limitations and state and local fiscal structure: an empirical assessment. Public Budg Financ 16(1):75–101

123

Do fiscal constraints prevent default? Historical evidence… 209

The New York Times (1875) Illinois: municipal and county debts The New York Times (1877) Reform in city affairs Oates WE (2005) Toward a second-generation theory of fiscal federalism. Int Tax Public Financ 12(4):349–

373 Padovano F (2014) Distribution of transfers and soft budget spending behaviors: evidence from italian

regions. Public Choice 161(1–2):11–29 Persson T, Tabellini G (2004) Constitutional rules and fiscal policy outcomes. Am Econ Rev 94(1):25–45 Pettersson-Lidbom P (2010) Dynamic commitment and the soft budget constraint: an empirical test. Am

Econ J Econ Policy 2(3):154–179 Pinsky DE (1963) State constitutional limitations on public industrial financing: an historical and economic

approach. Univ Pa Law Rev 111(3):265–327 Porter RP (1880) State debts and repudiation. Int Rev 9(1):556–604 Poterba JM, Rueben K (1999) State fiscal institutions and the U.S. municipal bond market. In: Poterba JM,

von Hagen J (eds) Fiscal institutions and performance. Univ of Chicago Press, Chicago Poterba JM, Rueben K (2001) Fiscal news, state budget rules, and tax-exempt bond yields. J Urban Econ

50(3):537–562 Primo DM (2006) Stop us before we spend again: institutional constraints on government spending. Econ

Polit 18(3):269–312 Qian Y, Roland G (1998) Federalism and the soft budget constraint. Am Econ Rev 88(5):1143–1162 Ratchford BU (1941) American state debts. Duke Univ Press, Durham Rodden J, Litvack J, Eskeland G (Eds) (2003) Decentralization and the challenge of hard budget constraints.

MIT Press, Cambridge Romer C (1986) Spurious volatility in historical unemployment data. J Polit Econ 94(1):1–37 Rose S (2010) Institutions and fiscal sustainability. Natl Tax J 63(4):807–838 Rueben KS (1997) Tax limitations and government growth: the effect of state tax and expenditure limits on

state and local government. Dissertation Ma Inst Technol Scott WA (1893) The repudiation of state debts. Thomas Y. Crowell & Co, Boston Seljan EC (2014) The limits of tax and expenditure limits: TEL implementation as a principal-agent problem.

Public Choice 159(3–4):485–501 Shadbegian RJ (1996) Do tax and expenditure limitations affect the size and growth of state government?

Contempl Econ Policy 14(1):22–35 Shadbegian RJ (1998) Do tax and expenditure limitations affect local government budgets? Evidence from

panel data. Public Financ Rev 26(2):118–136 Steeples D, Whitten DO (1998) Democracy in desperation: the depression of 1893. Greenwood Press,

Westport Streb JM, Torren G (2013) Making rules credible: divided government and political budget cycles. Public

Choice 156(3–4):703–722 Summers MW (1984) Railroads, reconstruction, and the gospel of prosperity: aid under the radical repub-

licans, 1865–1877. Princet Univ Press, Princeton Thies CF (2002) The american railroad network during the early 19th century: private versus public enter-

prise. Cato J 22(2):229–261 Wallis JJ (2015) NBER/university of maryland state constitution project. www.stateconstitutions.umd.edu.

Accessed 24 March 2015 White GT (1982) Years of transition: the united states and the problems of recovery after 1893. Univ Ala

Press, Tuscaloosa

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.