FINANCIAL VULNERABILITY: HOW EXPOSURE TO THE WORLD ECONOMY INFLUENCES DEBT 

SUSTAINABILITY AND RESPECT FOR HUMAN RIGHTS 

 

By  

 

Daniel Scott Hansen 

 

 

 

 

 

 

 

 

 

 

 

 

A DISSERTATION 

 

Submitted to 

Michigan State University 

in partial fulfillment of the requirements 

for the degree of 

 

Political Science – Doctor of Philosophy 

 

2018 

 

 

FINANCIAL VULNERABILITY: HOW EXPOSURE TO THE WORLD ECONOMY INFLUENCES DEBT 

SUSTAINABILITY AND RESPECT FOR HUMAN RIGHTS 

ABSTRACT 

 

 

By  

 

Daniel Scott Hansen 

 

A defining feature of the post-World War II global order has been the continued 

integration of national economies, and much scholarship has investigated the general impact 

which this integration has had on national outcomes ranging from development (both 

economic and political) to security (whether it impacts the likelihood of war). In this 

dissertation, I examine not simply integration itself but the manner of integration which has an 

important influence on world affairs. Specifically, I investigate what conditions render a country 

vulnerable to international financial markets, and what impact this financial vulnerability has on 

important national outcomes such as debt sustainability, dealt with in chapters one and two, 

and human rights practices, in chapter three. In the first chapter, I examine a long-held theory 

of financial market discipline which argues that market exposure induces prudent fiscal policies 

by governments. To date, there has not been a clear and systematic demonstration that 

markets actually have this effect, and so here I theorize the most likely market conditions which 

may give rise to fiscal discipline. While I demonstrate that there is little evidence in favor of the 

market discipline hypothesis, I highlight alternate pathways of political competition which do 

lead to improvements in fiscal budgeting practices. 

In the second chapter, I examine another widely held theory, that democratic institutions foster 

the credibility of debt repayment, the so called “democratic advantage”. Focusing on 

developing countries, I show that democracies under financial stress – countries with high 

levels of debt and low levels of foreign exchange reserves – actually have worse sovereign 

credit ratings than their autocratic counterparts. I argue that this occurs because democratic 

institutions can inhibit or slow a policy response when there is a pressing need for economic 

adjustment. Further illustrating the causal mechanism, I show that high debt levels reduce 

credit ratings, but only in more democratic countries. I therefore show that when debt 

commitments are most difficult uphold, in the developing world autocracies actually fare better 

than democracies, offering an important theoretical advance in how we understand democratic 

institutions relate to economic stability. 

In the last chapter, I examine an important dilemma in economic statecraft – that the use of 

human rights sanctions has the perverse effect of worsening a countries’ human rights 

practices. Drawing on a similar conception of financial vulnerability, I argue that countries with 

high levels of financial vulnerability will more closely approximate the ability of the ruling elite 

to suffer financial hardship as a consequence of the sanctions, in ways unexplained simply by 

economic wealth alone. As a consequence, I show that, contrary to the literature, financially 

vulnerable countries actually improve their human rights practices when sanctions are 

imposed. This therefore demonstrates that sanctions can have a positive impact even in 

difficult cases which aim to coerce national use of repressive strategies. It further highlights the 

types of sanctions tools which are most like to be efficacious given a countries exposure to the 

world economy. 

 

 

To my wife Nicole, who has been a constant source of joy, the bedrock of our family, and has 
sacrificed much to help me realize this goal. To my kids, Audrey, Olivia, and Scott, who always 
help me to smile regardless of the stresses of life. And to my parents, Scott and Antonella, and 

to my in-laws Mike and Barbara.  

 

iv 

ACKNOWLEDGEMENTS 

 
 

I am deeply grateful to Cristina Bodea, whose insights and guidance tremendously helped me to 

develop the skills necessary to complete this work. Her generosity and attentiveness were 

invaluable and I am forever indebted to her for her mentorship. I also could not ever have 

replaced the unique theoretical and methodological insights which Michael Colaresi has shared 

with me over the years. It was a privilege to have had this experience and to have worked for 

him at International Interactions. I am deeply grateful to Eric Chang, who taught me some of 

the most transformative academic lessons, and for his genuine nature and thoughtfulness. I am 

also very grateful to Ben Appel, who provided some of the intellectual foundation which has 

informed much about how I think about international relations broadly. I have been honored to 

have learned from such tremendous scholars. 

I also would like the thank Rob Franzese, Bill Jacoby, Tom Hammond, Shahryar Minhas, Gerald 

Schneider, Eric Juenke, and Johannes Karreth for their notable influences on my development 

as a scholar. I could never thank enough my friends and colleagues Tolgahan Dilgin, Peter 

Penar, and Fang-Yu Chen, who helped to make graduate school some of the best memories of 

my life. I enjoyed great years at Michigan State with Jacob Snider, Miles Armaly, Marty Jordan, 

William Isaac, Matt Zalewski, Erica Maylee (now Loding), Tara Iseneker, Nate Smith, Alex 

Mardon, Zuhaib Mahmood, Maya Hadar, Thomas Bentley, Michael Giles, Emma Slolina, Lora 

DiBlasi, Jamil Scott, Jessica Schoenherr, Elizabeth Lane, Erika Rosebrook, Hyunwoo Kim, Daniel 

Fram, Bob Lupton, and Adam Enders. I am further grateful to Jeremy Cantu, Waldo Gonzalez, 

Russel E. Banks, and Cody Shoemaker, for their incredible and sustaining friendship.  

 

 

v 

 

TABLE OF CONTENTS 

 
 
LIST OF TABLES ............................................................................................................................................ vii 
 

LIST OF FIGURES ......................................................................................................................................... viii 
 

CHAPTER 1: The Effectiveness of Fiscal Institutions: International Financial Flogging or Domestic 
Constraint? .................................................................................................................................................... 1 
Literature:  The Market Discipline Hypothesis.......................................................................................... 8 
Literature: The Financial Flogging Hypothesis ........................................................................................ 10 
When the Market Can Flog ..................................................................................................................... 13 
Why Fiscal Institutions are Effective: Domestic Political Competition ................................................... 19 
Data ......................................................................................................................................................... 25 
Research Design ...................................................................................................................................... 29 
Empirical Results ..................................................................................................................................... 31 
Endogeneity of Fiscal Institutions ........................................................................................................... 37 
Further Robustness ................................................................................................................................. 41 
Conclusion ............................................................................................................................................... 44 
APPENDIX ................................................................................................................................................ 46 
BIBLIOGRAPHY ........................................................................................................................................ 59 

 

CHAPTER 2:  The Democratic (Dis)Advantage: How Financial Vulnerability Conditions the Effect of 
Democracy on Sovereign Credit Ratings in Developing Countries ............................................................. 65 
The Democratic Advantage ..................................................................................................................... 71 
The Democratic Disadvantage ................................................................................................................ 74 
The Influence of Financial Vulnerability .................................................................................................. 77 
Research Design ...................................................................................................................................... 82 
Empirical Results ..................................................................................................................................... 89 
Robustness .............................................................................................................................................. 96 
Conclusion ............................................................................................................................................... 98 
APPENDIX ................................................................................................................................................ 99 
BIBLIOGRAPHY ...................................................................................................................................... 113 

 

CHAPTER 3: Dollars and Debt: How Financial Vulnerability Improves the Effectiveness of Economic 
Sanctions ................................................................................................................................................... 118 
The Influence of Economic Sanctions ................................................................................................... 124 
Sanctions and the Influence of Financial Vulnerability ......................................................................... 127 
Financial Vulnerability Conditions ........................................................................................................ 131 
Research Design and Data .................................................................................................................... 137 
Results ................................................................................................................................................... 140 
Conclusion ............................................................................................................................................. 146 
BIBLIOGRAPHY ...................................................................................................................................... 148 

 
 

 

vi 

 

LIST OF TABLES 

 
 

Table 1.1: Fiscal Institutions Models and Robustness Specifications ........................................... 35 
 
Table 1.A.1: GMM Robustness Specifications .............................................................................. 47 
 
Table 1.A.2: Triple Interaction Models ......................................................................................... 48 
 
Table 1.A.3: Fiscal Institution Adoption (Selection) Models ......................................................... 49 
 
Table 1.A.4: Endogeneity Robustness Models .............................................................................. 50 
 
Table 1.A.5: Robustness to Splitting Sample ................................................................................ 54 
 
Table 1.A.6: Jackknife Panel Regressions ...................................................................................... 55 
 
Table 1.A.7: Expanded Control Set for Main GMM Model ........................................................... 56 
 
Table 2.1: Heckman Selection Models (Outcome Equations) ...................................................... 92 
 
Table 2.2: Endogeneity Models – GMM and 2SLS ........................................................................ 95 
 
Table 2.A.1: Descriptive Statistics for Main Regression Sample................................................. 100 
 
Table 2.A.2: Countries in Regression Sample ............................................................................. 100 
 
Table 2.A.3: PCSE and Random Effects Models .......................................................................... 101 
 
Table 2.A.4: Polity2 Robustness Models ..................................................................................... 102 
 
Table 2.A.5: Heckman Selection Equations (corresponding to Table 2.1) ................................. 104 
 
Table 2.A.6: Additional Controls ................................................................................................. 105 
 
Table 2.A.7: Alternate Exclusion Variable for Heckman ............................................................. 106 
 
Table 2.A.8: Robustness Error Correction Models...................................................................... 107 
 
Table 2.A.9: Seemingly Unrelated Regression Robustness Model ............................................. 112 
 
Table 3.1: Effect of Sanctions on Government Repression (FVI Subcomponents) .................... 142 
 
Table 3.2: FVI Index and the Impact of Government Repression ............................................... 143 

 

 

vii 

 

LIST OF FIGURES 

 
 

Figure 1.1: Marginal Effects of Fiscal Institutions ......................................................................... 36 
 
Figure 1.2: Financial Flogging Models ........................................................................................... 36 
 
Figure 1.3: Political Competition: Rule Basis (MTF – No MTF) ..................................................... 37 
 
Figure 1.A.1: Endogeneity Robustness: RBasis (MTF – No MTF) .................................................. 51 
 
Figure 1.A.2: Marginal Effect of Rule Basis (MTF – No MTF) ........................................................ 52 
 
Figure 1.A.3: ME Plots from Table 1.A.5 Region A ........................................................................ 53 
 
Figure 1.A.4: Lambda Triple Endogeneity Models ........................................................................ 57 
 
Figure 1.A.5: Out of Sample Forecasts .......................................................................................... 58 
 
Figure 2.1: Marginal Effects of Democracy & Debt ...................................................................... 93 
 
Figure 2.2: Marginal Effects of Democracy & Reserves/Debt ...................................................... 93 
 
Figure 2.A.1: Endogeneity Marginal Effects (Electoral Democracy) ........................................... 103 
 
Figure 2.A.2: ECM Robustness .................................................................................................... 108 
 
Figure 2.A.3: (log) Debt / GDP – Full Model (Table 1, M4) ......................................................... 109 
 
Figure 2.A.4: Developing Country Debt & Reserves, By Regime Type) ...................................... 110 
 
Figure 2.A.5: Debt and Reserves Correlation, Developing Countries) ........................................ 111 
 
Figure 3.1: Marginal Effect of HR Sanctions ............................................................................... 145 
 

 

 

viii 

 

CHAPTER 1: The Effectiveness of Fiscal Institutions: International Financial Flogging or Domestic 

 

Constraint? 

Many have argued that financial markets are able to exert pressure on governments to 

maintain sustainable budget balances (i.e. smaller deficits), in what has been known as the 

market discipline hypothesis. Since investors charge higher interest rates on debt for countries 

who have accumulated large debt burdens, scholars have suggested that governments will feel 

compelled by these rising interest rate costs to reduce deficit spending. Recently, it has been 

argued that two important fiscal institutions – fiscal rules (such as a balanced budget 

requirement law) and fiscal transparency (how observable the budgeting process is) – are 

crucial to enable financial markets to exert this disciplining pressure on governments. This idea, 

that financial markets can effectively pressure governments in the presence of these fiscal 

institutions, is what I call the Financial Flogging Hypothesis. Despite the prominence of this 

market discipline hypothesis, however, studies of fiscal rules or fiscal transparency have not 

actually demonstrated that financial markets play a causal role in the effectiveness of these 

fiscal institutions. Instead of showing that market prices augment the effectiveness of fiscal 

institutions at reducing deficit levels, studies have tended to show the reverse: how fiscal 

institutions influence debt prices (Kelemen and Teo 2014; Lowry 2001; Iara and Wolff 2011; 

Afonso and Guimarães 2013). As a result, the emphasis has been only on one-half of this 

theorized relationship, and therefore it is not clear whether fiscal rules and fiscal transparency 

are effective because of financial markets, as the financial flogging hypothesis suggests, or due 

to other mechanisms.  

 

1 

 

In this paper, I redress the empirical shortcoming of the literature by specifying direct tests of 

the financial flogging hypothesis. In addition, I suggest that an alternative causal mechanism – 

political competition and public pressure – is a more likely mechanism for the effectiveness of 

fiscal institutions in reducing fiscal deficits. As a result, I both test the financial flogging 

hypothesis in a more systematic way than has previously been done, but also test this 

hypothesis against an alternative causal pathway of domestic politics. And while these are not 

mutually exclusive possibilities – markets or political competition – the framework that I 

articulate here allows us to begin to understand more concretely the existing evidence for each 

mechanism.  Contrary to expectations from the literature, I show that there is little support for 

the financial flogging hypothesis: markets do not seem to have a significant impact on fiscal 

spending. However, there is robust support for the role of various domestic political actors.  

Theoretically, the literature on market discipline does not fully specify a causal chain explaining 

why we should expect financial markets to effectively flog governments into better fiscal 

performance. While they argue that costs should pressure governments, it is never clear why 

such market-based costs should translate into meaningful political action. Governments which 

are deficit spending at unsustainable levels are likely doing so because they see a rational 

political benefit to the policy. This may arise, as an example, if political leaders use fiscal policy 

to support their constituents but also believe that they are insulated from the risk of blame for 

such deficit spending. In order to compel fiscal discipline, therefore, the economic costs faced 

by governments running fiscal deficits need to be large enough to generate substantial political 

pressures. Small changes to debt costs would likely be insufficient to coerce political leaders to 

enact unpopular deficit control measures, such as spending cuts or tax increases. Since it has 

 

2 

 

been shown that higher debt only leads to small increases in interest rate costs (Mosley 2003; 

Lowry and Alt 2001), and that deficit spending is not electorally punished in the developing 

world (Brender and Drazen 2008), it is not clear exactly how financial markets supposedly 

generate strong enough political pressure to impact fiscal policies. 

I argue that, rather than financial markets, a more likely mechanism by which fiscal rules and 

fiscal transparency generate political pressure is domestic political competition (competitive 

opposition parties and veto actors) and public pressure (by active civil society organizations and 

in democracies). First, fiscal transparency enables the public to more easily monitor the fiscal 

performance of the government, thereby subjecting the executive to greater scrutiny of its 

budget. In a situation of high transparency, the political opposition is then able to use a fiscal 

rule, such as a balanced budget law, as a means of gaining political leverage against the 

incumbent. Where the government is not living up to its fiscal commitments, opposition parties 

stand to benefit by using fiscal rules as a rhetorical weapon against the regime, alleging 

recklessness and endangerment of the economy.  In this way, in politically competitive 

environments, these fiscal institutions may lead to smaller fiscal deficits because incumbents 

may fear the political backlash associated with policies of excessive spending. This implies, 

then, both that political opponents will fight back against government fiscal policies, but also 

that the public cares enough to facilitate the political backlash. Empirically, I will show both that 

political competitors (elites) matter, and that a vibrant civil society is another manifestation of 

political power which augments the effectiveness of fiscal institutions.  

This is a novel contribution, first because much of the existing market discipline literature has 

neglected to strongly consider the role of political competition as a causal mechanism. For 

 

3 

 

instance, Kelemen and Teo (2014) suggest that the literature of fiscal institutions has explored 

only judicial mechanisms or financial markets, with no mention of political competition. By 

examining the financial flogging hypothesis systematically alongside these political competition 

channels, it helps to highlight a deficiency in the market discipline research while at the same 

time advancing our understanding of how these fiscal institutions can succeed via political 

competition. In short, we gain a clearer understanding of what causal pathways are actually at 

play. Furthermore, while an existing literature has examined the effectiveness of certain fiscal 

institutions due to political competition (Hallerberg, Strauch, and von Hagen 2009; Hallerberg 

and Yläoutinen 2010; Hallerberg and Wolff 2008), such studies have almost exclusively focused 

on the European Union or the EMU and have emphasized institutional features often particular 

to parliamentary democracies1. Consequently, it is a very open question how fiscal institutions 

function in contexts without an overarching supranational body (EU), a currency union, or in 

diverse political regimes. By contrast, I examine fiscal institutions - fiscal rules and fiscal 

transparency – which are not particular to any regime type. More importantly, I emphasize 

aspects of political contestation which apply globally and help to illuminate the specific features 

of domestic politics which make such institutions binding constraints. 

As an example of the importance of fiscal rules and political competition, the US “debt-ceiling 

negotiations” of late 20112 are illustrative of this logic. In this case, the US has a “debt ceiling 

rule”, which caps the legally allowed amount of borrowing the US government is able to issue. 

                                                           

1 See, for instance, Hallerberg’s many excellent co-authored works and emphasis on the “contracting 
approach” and the strength of spending ministers. 
2 This experience of political brinksmanship led the credit rating agency Standard and Poor’s to downgrade the 
rating quality of US debt.  

 

4 

 

Every few years the US Congress must increase this debt ceiling or the US government would no 

longer be able to issue new debt, which would then trigger a debt default. In 2011, the political 

opposition in the US House of Representatives (the Republican Party led by Representative 

John Boehner) insisted that President Obama agree to drastic spending cuts, or they would not 

pass an increase to the US debt-ceiling limit. While failing to achieve a comprehensive debt 

reduction agreement, in the end a compromise known as sequestration was passed which has 

improved the US fiscal position for the remainder of President Obama’s full term in office3. 

What we see from this example is that an improvement in the US fiscal position was achieved 

as a consequence of the opposition party leveraging the existence of a form of fiscal rule (the 

debt ceiling law) in order to extract fiscal concessions from the president. At the same time, the 

financial flogging hypothesis would predict that political actors were motivated by the cost of 

debt. In the US episode, however, it is clear that interest rate costs of US issued debt (both in 

terms of 10 year and two-year nominal treasury bonds)4 were declining from 2010 throughout 

these debt-ceiling negotiations. So while this highlights the role of fiscal rules and political 

competition, it also shows that the costliness of US debt was not a driving force of the 

subsequent improvement in the US fiscal budget, contrary to financial flogging expectations.  

To directly test the financial flogging hypothesis, I propose a series of hypotheses regarding 

where we would be most likely to observe financial markets influencing government deficits. 

Given the presence of fiscal institutions – fiscal rules and transparency – I posit that we are 

likeliest to observe smaller deficits for countries who are most financially vulnerable to the risk 

                                                           

3 See the Congressional Budget Office webpage for historical budget data – discretionary outlays. 
https://www.cbo.gov/about/products/budget-economic-data#2 
4 This is according to data publicly available at www.treasury.gov 

 

5 

 

of economic shocks. For some countries, large deficits do not just mean more expensive debt 

financing but indicate a significant escalation in the risk of a negative economic shock. For these 

countries, deficit spending is arguably much more deleterious than for countries with a more 

stable financial position. Therefore, if the financial flogging hypothesis is correct, then we 

should observe that fiscal deficits are smaller for financially vulnerable countries who have 

adopted fiscal institutions. Comparatively, I argue that fiscal institutions are effective not 

primarily due financial markets, but due to domestic political competition and public pressure. 

In this case, we should expect that fiscal institutions lead to smaller deficits where political 

opponents have the most leverage against the executive, and therefore have the greatest 

means of restraining fiscal deficits. I suggest that countries with divided government and more 

politically competitive environments represent such cases of increased political leverage. 

Specifically, these are countries with more veto players, in democracies, and countries with a 

higher opposition party vote share in elections. Additionally, I argue that countries with higher 

participation levels from civil society organizations are another manifestation of political 

pressure on the government which enhances the effectiveness of these fiscal institutions. As a 

final note, these competing hypotheses certainly are not mutually exclusive, and reality could 

lie somewhere in between. Nonetheless, a fruitful starting point for such an investigation 

begins with a clear exposition of the conditions under which both mechanisms could present 

themselves. If one set of hypotheses finds clear support while the other does not, this would be 

evidence which helps us begin to distinguish whether both theories have support or whether it 

causes us to reflect more deeply on the absence of evidence for a long-held theory such as 

market discipline. 

 

6 

 

Using global data on 69 countries from 1990 to 20085, I test these hypotheses and find that 

fiscal institutions do lead to smaller fiscal deficits, but do not appear consistent with the 

financial flogging hypothesis. Rather than finding significant results where international 

financial pressures are the greatest, we see a significant effect of these institutions in settings 

of greater political contestation. As a result, we have the clearest test of the financial flogging 

mechanism to date, and we fail to find support for this long-held theory. Moreover, we can take 

these findings as suggestive of the importance of political competitors underlying the 

effectiveness of these fiscal institutions. I further show that these results are robust to 

controlling for possible selection bias among countries who choose to adopt these fiscal 

institutions. Lastly, I also demonstrate that these institutions are effective even excluding 

Eurozone countries, who arguably have stronger incentives than most to maintain fiscal 

discipline. 

These findings have noteworthy implications for financial stability and helps inform the design 

of international economic policy. Understanding the causal mechanisms of fiscal rules is 

important, because a mistaken understanding of why such institutions seem to be effective can 

lead financial markets to price the riskiness of debt incorrectly. Consequently, a better causal 

understanding can mitigate the negative economic consequences of a financial shock because 

markets could hedge against such risk. A misplaced understanding of the role of financial 

markets could also lead to the adoption of policies which will do little to address fiscal 

shortcomings and could plausibly be harmful. For instance, it could incorrectly inform an 

emphasis on capital account liberalization where it may not be optimal. While sometimes 

                                                           

5 The data ends in 2008 due to data constraints in the fiscal transparency (MTF) variable. 

 

7 

 

positive, recent literature has found that for some countries liberalization of the capital account 

can increase risk to the banking sector (Klomp and de Haan 2009; Vasicek et al 2017; Hamdaoui 

et al 2016; Qin and Luo 2014). As a result of these issues, this paper advances a research 

agenda which informs our understanding of risk and development strategies with respect to 

fiscal stability. 

Literature:  The Market Discipline Hypothesis 

Much work has been done examining the role for fiscal rules or fiscal transparency as possible 

solutions to the problem of excessively spending governments, with the general finding that 

these institutions are an effective means of promoting more stable fiscal policies. The key issue 

in the present context is why these fiscal institutions are effective.  

The conventional mechanisms are suggested to be either judicial enforcement or financial 

markets (Kelemen and Teo 2014). Here I overview the literature arguing in favor of the market 

discipline hypothesis, after which I discuss the newer financial flogging hypothesis and its 

emphasis on fiscal institutions. Most do not explicitly test whether markets influence fiscal 

policies, and the few who do either fail to find supportive evidence or present mixed results.  

As noted, the 'financial flogging hypothesis' is a term I employ to distinguish the Kelemen and 

Teo (2014) argument about market-based discipline from its antecedent, the simpler 'market 

discipline hypothesis' (Bayoumi, Goldstein, and Woglom 1995). The market discipline 

hypothesis argues that international lenders (the financial markets) exert meaningful pressure 

on governments to restrain spending because they charge higher prices for government debt 

the larger the debt becomes. In this literature, scholars most commonly show the effects of 

debt or fiscal rules on two main measures. The first is sovereign bond rates, which are interest 

 

8 

 

rate costs of debt (higher interest rates reflecting greater risk). The second is a sovereign credit 

rating6, which is a 21-point rating scale which assesses the risk of a national default on debt 

obligations, and are given by agencies such as Moody’s Analytics or Standard & Poor’s (the 

highest rating of which would be the “AAA” rating). These studies assess these financial market 

costs with the direct implication that higher sovereign financing costs will be onerous for 

governments and therefore can compel a change in behavior.  

While there are a few exceptions, most of this literature does not actually demonstrate that 

markets have changed the course of fiscal policy due to market costs.  Bayoumi et al (1995a) is 

an early example, examining US State governments and estimating how much more expensive 

sovereign debt becomes given an increase in fiscal deficits (they find an upper range estimate 

of 35 basis points). Kumar and Baldacci (2010) and Mosley (2003) undertake similar analysis to 

estimate the effect of deficits on bond prices and find comparable results - higher deficits do 

increase bond prices, although this effect is small in magnitude. Hallerberg and von Hagen 

(1999) also directly examine the effect of debt costs on fiscal deficits, finding that debt-service 

costs are not significantly related to deficits, contrary to market discipline expectations. 

Similarly, research has examined how credit rating downgrades influence fiscal behavior. If 

countries responded to a rating downgrade (signaling an increase in default risk) by improving 

their fiscal balances, then it would seem to validate the theory of market discipline. Duygun et 

al (2016) come close, although they do not show that downgrades influence fiscal policy; 

rather, only after rating upgrades do countries improve their fiscal behavior. As such, it is not 

                                                           

6 “Sovereign in this case indicates a ‘national’ government credit rating, but credit ratings can be applied to 
other debt-issuing governments as well. For instance, since US state governments can themselves issue debt, 
ratings agencies have long issued credit ratings for state government bonds as well. 

 

9 

 

clear that governments respond to financial market pressure or that they feel coerced by 

markets to adjust their spending preferences, according to market discipline expectations7. 

Hanusch and Vaaler (2015) show that ratings downgrades reduce the likelihood of incumbent 

reelection, which is noteworthy, but fail to find that downgrades or negative outlooks have an 

effect on government fiscal spending. Specifically, when restricting the analysis to countries 

with poorer ratings, they show not that countries threatened with a negative outlook spent 

less8, but that countries with a positive outlook did. Soudis (2014) also examines whether 

ratings downgrades influence government policy, showing that ratings downgrades do not lead 

to the adoption of neo-liberal market-friendly reforms. This is noteworthy because, presuming 

market discipline, we would expect downgraded countries to “course correct” with policies that 

financial markets would find favorable, although Soudis did not find it to be the case. We can 

therefore see that there is a relative dearth of studies who actually observe an influence of 

financial market pressure on government spending decisions. 

Literature: The Financial Flogging Hypothesis  

Kelemen and Teo (2014) depart from the previous market discipline literature by essentially 

agreeing that markets, on their own, are not able to discipline governments effectively. If they 

did, they suggest, then bond yields would rise approximately linearly as an increasing function 

of debt. As a result, governments often do not face much pressure from markets, and there is 

                                                           

7 Caution seems warranted here since reverse causality could also be at play – that, upgrades don’t cause 
better behavior but merely that ratings agencies sense a positive improvement in budgeting practices and 
therefore upgrade because they already sense a shift in discipline. For instance, as Duygun et al do not include 
fiscal institutions, it could be that countries have adopted a balanced budget law, receive a rating upgrade, 
and then subsequently the budget law is what is successful in suppressing deficit spending rather than the 
rating upgrade. 
8 In fact, while the variable is not significant, the coefficient is actually negative – suggesting that negative 
outlook countries still deficit spend in election years. 

 

10 

 

little to prevent them from reaching a fiscal crisis. And at such a point, I argue, markets have 

clearly failed to fiscally “discipline”. Yet Kelemen & Teo argue that financial markets do 

discipline governments, but are only able to do so when there are fiscal rules (such as a 

balanced budget law, which can act as a focal point to coordinate financial market punishment) 

and fiscal transparency is high (such that markets can independently and clearly discern the 

fiscal performance of governments). This argument, that financial markets can impose fiscal 

discipline only when there are these combined fiscal institutions, is what I call the financial 

flogging hypothesis to distinguish this innovation from the earlier strand of market discipline 

research.  

Since Kelemen and Teo emphasize the role of fiscal institutions, they follow much of the earlier 

market discipline literature by showing that for US states, strict fiscal rules and high fiscal 

transparency increases the states’ credit ratings. However, it is not clear in this case that 

financial markets really impose more severe debt costs given the presence of fiscal institutions. 

Like much of the literature, they do not actually establish that government’s spending policies 

vary at all contingent on such ratings differences9. Earlier, Alt and Lowry (2001) made a similar 

argument by suggesting that a specific type of fiscal rule10 could enable markets to "extract 

information from noisy signals," arguing that bond markets will be able to more effectively 

discipline governments in such cases. Like Kelemen and Teo, they show that for the US states 

fiscal rules reduce interest rates on debt, but this effect is small in nature11 and they do not 

                                                           

9 Additionally, since virtually all US states have some form of fiscal rule, their study is not able to demonstrate 
rating differentials between units with and without such fiscal institutions. 
10 This is a “no carry-over” rule, prohibiting deficits from rolling into the next annual budget cycle. 
11 They find a reduction of interest rate costs of 7.5 basis points. 

 

11 

 

explore any impact on government fiscal policy. This finding that fiscal rules can reduce debt-

financing costs is similar to Afonso and Guimarães (2013), who show that expenditure rules in 

the EU also reduce interest rates on government debt. And in another study Iara and Wolff 

(2011) demonstrate that fiscal rules with a stricter legal basis also reduce bond yields on debt.  

Taken collectively, then, we see from both the earlier market discipline literature and the 

specific literature on fiscal rules that much of this research shows what causes changes in debt 

costs (or credit ratings) without showing that debt costs actually influence government fiscal 

behavior. We see that fiscal rules reduce debt costs, but generally not that debt costs then 

influence deficits. This is still an interesting and noteworthy finding, but it is possible that 

financial markets charge less for debt from countries with fiscal rules simply because they 

perceive that such rules are effective debt management devices. It does support the idea that 

these fiscal institutions are effective, but it does not obviously suggest that markets themselves 

are causally related to governments spending decisions.  

Moreover, in much of the literature above, the demonstrated effect on debt costs is only 

marginal, which is problematic for the expectation of market discipline or financial flogging. The 

key reason why we shouldn’t expect marginal increases in debt costs to influence government 

fiscal policy is because countries who are running substantial deficits annually are likely doing 

so because they perceive a rational political incentive to doing so. The literature explaining why 

countries incur fiscal deficits is very large and links deficits to electoral incentives (Drazen 2001; 

Brender and Drazen 2005; Shi and Svensson 2006), partisan preferences (Eslava 2006), because 

divided governments are trying to force other parties to incur a larger share of the fiscal 

adjustment costs (Franzese 2002; Alesina and Drazen 1991; Roubini and Sachs 1989), or 

 

12 

 

because coalition governments obscure who is to blame for deficits (Perotti and Kontpoulos 

1998). In order to force governments to change course, it would therefore seem that the 

economic costs associated with deficit spending would need to be large in order to generate 

sufficient political pressure to effectuate a change. In the next section, I expand upon this idea 

to generate predictions about when financial markets may be able to successfully flog 

governments into fiscal discipline. 

 When the Market Can Flog  

Financial markets are most likely to be able to flog governments when small or average 

increases in fiscal deficits could trigger significant economic costs. In these cases of heightened 

vulnerability, politicians are likely to feel significant pressure to change fiscal spending to hedge 

against the risk of such economic shocks, because excessive deficits in such a context would 

become politically and economically self-defeating. The political benefit of deficits – spurring 

the economy or giving more goods to supporters – is dubious when that very practice could 

trigger a crisis which could eliminate such gains. Even when numerous veto-actors make a fiscal 

adjustment a difficult bargain to achieve, large enough economic costs could compel 

government actors to calculate the risk of inaction (crisis) to be larger than action (adopting 

unpopular deficit control measures).  

In this paper, I focus on several cases for which I argue we would be more likely to observe the 

influence of financial markets on fiscal discipline. These are when the country: (a) has fixed 

exchange rates, (b) has a credit rating near the threshold of being downgraded to “speculative 

quality”, (c) the country has a speculative quality credit rating, and finally (d) if the country is an 

emerging market country. While deficits are not always, ceteris paribus, a negative from the 

 

13 

 

perspective of the international investor, all of these cases represent instances of increased 

sensitivity to declines in the fiscal budget. Ahlquist (2006) notes that deficits are related to all 3 

of the major forms of investor risk: inflation, currency, and default risk. Mosley (2003) discusses 

debt issuance in the context of a signaling game, such that large or persistent deficits (or, high 

debt levels) are taken as a signal of a governments unwillingness to repay their debts. 

Unsurprisingly, Montes, de Oliveira, and de Mendonça (2016) find strong evidence along these 

lines that deficits have a consistent negative impact on sovereign credit ratings. Moreover, the 

financial flogging conditions I hypothesize here make large deficits significantly more likely to 

be viewed as a negative risk indicator to investors. 

Countries maintaining a fixed exchange rate face the ever-present risk that, if international 

holders of their national currency think that the government will not be able to maintain the 

currencies fixed rate, then international investors will rapidly sell the currency, forcing the 

government to abandon the fixed rate. Such an episode would involve significant economic 

fallout, since a currency crash “puts upward pressure on inflation, increases countries’ foreign 

debt burden, and erodes economic growth” (O’Mahony and Nyblade 2014). Because the 

economic costs under fixed exchange rates can be so severe, causing governments to collapse 

(Bernhard and LeBlang 2008), governments should have strong incentives to guard against the 

risk that the currency would be devalued.   

The idea that deficit spending is related to a countries ability to maintain its fixed exchange rate 

goes back to the canonical currency crash model of Paul Krugman, who argues that 

governments running sustained fiscal deficits make fixed rates unmanageable, leading to such 

currency crashes (Krugman et al 1999). To be clear, this argument that deficits make countries 

 

14 

 

more vulnerable was the very reason that it was initially supposed that fixed rates would make 

governments more disciplined. The reason is that excessive deficits would constantly erode a 

countries base of foreign exchange reserves, materially depleting a countries ability to maintain 

its peg. This is notable because evidence in favor would be an example of market discipline, as I 

posit, although the early evidence had been mixed (Tornell and Velasco 2000). The alternate 

and less conventional perspective is that flexible exchange rates would deter fiscal excesses 

because such policies would immediately filter through to the exchange rate and generate 

inflationary costs. It is thus argued that in the short run more discipline arises from flexible 

rates. A key pitfall exists, however, because much like the rest of the market discipline 

literature this contention only holds if one considers that politicians respond smoothly to 

increasing costs. This is an unlikely position because if this were true, then there would have 

been no need to insulate central banks from political pressures with greater independence 

(Cukierman et al 2012). Further, it is generally recognized that the mechanism by which fiscal 

spending influences exchange rates, via the purchasing power parity condition, is a long-run 

phenomena which is unlikely to react very quickly to fiscal changes in the short-run (Feenstra 

and Taylor 2018). As a consequence, politicians might easily exploit fiscal policy for short-run 

advantages under flexible exchange rates, even considering the impact of fiscal policy and 

inflation on exchange rates. 

Regardless, the literature broadly supports the key contention for this flogging mechanism – 

that deficits under a fixed rate could be particularly harmful. Fixed exchange rates, to start, are 

more economically harmful in the event of economic shocks, since flexible exchange rates are 

“shock absorbers” against terms of trade or other external shocks (Broda 2004). A number of 

 

15 

 

other studies have confirmed the risks of deficits under fixed rates, by increasing the probability 

of a currency crash (Marini and Piersanti 2003), and by substantially increasing the risk of a 

crash when other economic shocks present themselves (Fratzscher 2011). In that sense, even if 

deficits do not cause a currency crash, it is significant crisis escalator. Moreover, even when 

deficits do not actually materialize, Gumus (2015) argues that simple expectations of deficits 

can be sufficient to spark a speculative currency attack. O’Mahony and Nyblade (2014) argue 

that the risks posed by fiscal deficits under fixed rates is exactly why governments only boost 

spending around elections when the government is sufficiently insulated from the risk of 

currency crashes. We therefore see from this literature that fiscal deficits are an inherent 

source of risk for countries who maintain a fixed exchange rate. 

In the context of the financial flogging hypothesis, which emphasizes that markets should be 

more effective with fiscal institutions, we would then expect that countries with fixed exchange 

rates will have smaller deficits when they have fiscal rules and fiscal transparency. This leads us 

to the first hypothesis: 

Flogging H1: Countries with fixed exchange rates will have smaller deficits when there are fiscal 

rules and fiscal transparency. 

As previously noted, sovereign (country) credit ratings reflect the riskiness of government 

issued debt, spanning 21 different ratings from a high of “AAA” or “AA+” down to junk status of 

“CC” and “C”12. These ratings scales are comprised by two broad categories: investment grade 

and speculative grade securities, the cut-point threshold of which occurs roughly midway down 

                                                           

12 This rating scale is for Standard & Poor’s. The Moody’s Analytics analog would be: “Aaa” and “Aa1” as the 
two highest, with “Ca” and “C” the lowest. 

 

16 

 

the rating scale at the transition of BBB- to BB+. Once a countries’ credit rating crosses the 

threshold from Investment Grade to Speculative Grade, economic costs and risk begin to 

escalate. For instance, interest rate costs on government debt begin to increase much more 

dramatically as credit ratings decline within the speculative category (Gaillard 2011; Hanusch 

and Vaaler 2015). As a result, deficit financing at the speculative grade can much more quickly 

increase a countries debt burden than is the case for countries with an investment grade rating.  

Being near the “threshold” of the speculative grade – the point at which a country is one or two 

downgrades away from losing its investment grade status - could arguably be an even greater 

source of fiscal discipline. The reason this could plausibly induce fiscal discipline by the 

government is that many large-scale institutional investors as a rule are only allowed to hold 

investment grade securities. A rating downgrade below the threshold can therefore have large 

negative consequences, because it can cause substantial capital flight, reducing the pool of 

investors and further increasing the cost of raising capital (Kaminsky and Schmukler 2002). 

Beyond governments, ratings downgrades can also increase the cost of business finance, rather 

than strictly government, because government bond ratings often serve as a benchmark for 

bond risk within the country. As a result, routine business operations can become more costly 

even for businesses without sovereign debt holdings, which is why ratings downgrades can 

dampen stock markets and increase volatility (Hooper, Hume and Kim (2008); Kabadayi 2013). 

Since this cost isn’t isolated to governments, this likely is a powerful motivator for business 

elites to lobby to ensure the sovereign credit rating remains investment grade. As a country 

nears the threshold of speculative grade quality, policymakers could arguably feel increasing 

pressure to reduce deficit spending to try and prevent crossing the threshold. 

 

17 

 

We therefore state the next two flogging hypotheses from these scope conditions: 

Flogging H2: Countries with a speculative credit rating should have smaller deficits when there 

are fiscal rules and fiscal transparency. 

Flogging H3: Countries near the threshold of a speculative credit rating should have smaller 

deficits when there are fiscal rules and fiscal transparency. 

Additional research has shown that sovereign credit ratings have a larger financial sector 

impact on emerging market economies. Because emerging markets are less developed and 

generally have more opaque institutions, there tends to be more uncertainty regarding 

investment risk, and is why Christopher, Kim, and Wu (2012) and Kaminsky and Schmukler 

(2002) both argue that emerging markets face more severe consequences following ratings 

downgrades. Kabadayi (2013) validates this finding, demonstrating that negative credit ratings 

for Turkey depressed the Turkish stock market at select periods. Additionally, Mosley (2003) 

argues they are more sensitive to financial markets, and Gaillard (2011) shows they face much 

higher interest costs for issued debt. Thus, they may be especially likely to refrain from 

significant deficit spending, because the inherent costs and risks involved are so much more 

pronounced. This leads us to the final financial flogging hypothesis. 

Flogging H4: Emerging market countries should have smaller deficits when there are fiscal rules 

and fiscal transparency. 

It should be emphasized that these scope conditions represent a best case for financial markets 

to be able to pressure governments into better fiscal performance. If markets do pressure 

governments, I argue we would be likely to observe it in these cases. As a consequence of the 

analysis, I suggest that we have a clearer understanding of whether or how markets can flog 

 

18 

 

governments, as well as in understanding why fiscal rules and fiscal transparency are effective 

institutions. In the next section, I turn to discuss the alternate causal pathway. Rather than 

these fiscal institutions (rules and transparency) enabling financial markets to punish 

governments, these institutions enable domestic political opponents to pressure the regime 

into better fiscal performance.  

Why Fiscal Institutions are Effective: Domestic Political Competition 

Kelemen and Teo (2014) criticize the literature on fiscal rules for presuming that judicial 

enforcement is the causal mechanism underpinning the success of fiscal rules. Because courts 

almost never have rendered a judgment against a US state governor (for state government 

bond issuance), they argue that courts are not a viable mechanism to enforce fiscal discipline. 

Instead, Kelemen and Teo argue that, with the aid of fiscal institutions, financial markets are 

the real means of fiscal discipline. 

I argue, however, that the discussion of courts versus financial markets as enforcers of fiscal 

rules misses the likeliest source of fiscal discipline – political competitors. The reason why 

political competition matters for the success of fiscal institutions is due to several reasons. 

Existing research on fiscal institutions within the EU and EMU has indicated that fiscal “rules” 

are more likely to be effective in environments with higher political contestation13. For 

instance, Hallerberg et al (2009) suggest that “in countries with high ideological distances and 

high office-seeking competition, the fiscal target and contract indices should be most 

                                                           

13 Fiscal “rules”, in the sense of a numerical fiscal target, are closest to Hallerberg et al’s (2009) discussion of 
the “contracting approach”. It is this approach that the authors identify as working best in competitive 
environments. While distinct in operationalization, conceptually their discussion maps very well onto a global 
framework examining numerical rules. 

 

19 

 

significant”. In Hallerberg and Wolff (2008), they elaborate by stating that “in coalition 

governments a contract approach… helps to restrict profligate fiscal policy.” While their 

operationalization of fiscal rules is different and specific to EU countries, the essential reasoning 

is that if a country has a single-party government (low competition in a parliamentary system), 

then such a government may easily remove a “contracting” approach, such as spending 

agreements or fiscal laws passed. Thus, when opposition parties are stronger or even share 

power in government, then this is a means by which the government may be bound to a fiscal 

rule.  

While I agree with this reasoning, I argue that there are broader political connections between 

competition and fiscal institutions. For instance, it is not merely that shared power prevents the 

government from unilaterally removing unfavorable spending constraints, although that is part 

of the logic. Even when not strictly in power, political opponents to the government can use 

these institutions against the regime in the event of deficit spending, generating elite-driven 

blowback against spending policies. In a case where the government is spending recklessly such 

that fiscal deficits are exceeding budget targets, political opponents have strong incentives to 

utilize these fiscal rules as a rhetorical weapon against the regime, even if they cannot directly 

impact the budget process. Governments would be vulnerable to accusations of capriciousness, 

economic endangerment and possibly corruption. Thus, I argue that in the presence of fiscal 

institutions, the executive would be more likely to choose a path which minimizes such political 

blowback against the government when the political opposition is sufficiently strong. This may 

mean direct cases of shared government, or it may be present in cases where opposition 

parties and groups are large enough to pose a threat even from outside the government. 

 

20 

 

Beyond the direct blowback which the executive would likely experience from political elites, I 

also argue that fiscal rules more readily enable countervailing pressure from the public, in 

settings where the executive is more accountable. On the one hand, this may seem surprising 

because democratic accountability is also what supposedly leads government leaders to feel 

pressured to use electoral fiscal manipulation (Drazen 2000). However, I argue that the 

presence of fiscal institutions, such as a balanced budget law, may incentivize the public to 

mobilize against the executive in anticipation of fiscal abuses because rules clarify to political 

constituents what fiscal failure looks like. In this sense, political observers may not have a clear 

idea exactly how bad the fiscal outlook is or even whether the fiscal trajectory is unsustainable 

or not. Much like Lowry and Alt (2001) argued that balanced budget laws enable financial 

markets to "extract information from noisy signals," here too laws can help voters extract 

meaningful information from noisy signals. The cognitive and conceptual costs for citizens to 

assess the viability of fiscal policy may be prohibitively high and consequently, their ability to 

assess the ex-ante need for incumbent punishment may be quite low. But a clear balanced 

budget law makes it theoretically much easier for the public to determine if the government is 

employing fiscal policy in a socially harmful manner. As a result, a fiscal rule can animate 

political constituencies such that the executive feels constrained in their freedom to use fiscal 

policy as they please.  

A direct and observable implication of this is that fiscal institutions should be more effective in 

countries with numerous and active civil society organizations (CSOs). If the public is animated 

by fiscal excesses when it may violate a national fiscal law, as I have argued, then such laws 

should be more effective where large numbers of social and professional groups are mobilized 

 

21 

 

to lobby for policy changes. Civil society has been key, for instance, in the funding and 

development of the Open Budget Index at the International Budget Partnership14, which is 

seeking to bolster dialogue between governments and the public regarding spending issues15. 

Higher participation of CSO’s thus raises the ability of public mobilization in the event that the 

government is threatening economic stability with excessive deficit spending. CSO’s also serve 

as “watchdogs” of government policy that other participants in society are unable to perform. 

For instance, while the literature argues that fiscal transparency is good because it enables 

monitoring of the government, a crucial question is – who will do the monitoring? Political 

opponents (elites) may be able to monitor, but CSOs are another public mechanism by which 

resources can be brought to bear to actually take advantage of the fiscal transparency. Thus, 

CSO’s are a key means by which monitoring can take place, information is disseminated 

regarding country compliance, and the public may be mobilized.  This leads me to state the first 

political hypothesis of the paper: 

Politics H1: Fiscal Rules and Fiscal Transparency should lead to smaller deficits at higher levels 

of civil society participation. 

Another important manifestation of political competition are instances of shared or divided 

government. While early literature emphasized that this can create a common-pool problem 

for fiscal policy (Kontopoulos and Perotti 1999), so long as a fiscal rule is a clearly defined 

numerical target, and the system is transparent enough such that the public can actually 

interpret this information, then even in the presence of a common-pool it can force a political 

                                                           

14 See http://www.internationalbudget.org/why-budget-work/role-of-civil-society-budget-work/ 
15 http://www.fiscaltransparency.net/giftaward/ 

 

22 

 

discussion on fiscal policy which otherwise would be easy to ignore. After all, if the government 

simply tried to renege on the rule, then the political opposition could use their power in 

government to force concessions on the issue. This was the motivation of Hallerberg et al 

(2009) and Hallerberg and von Hagen (1997) in their emphasis on features of shared or 

coalition governments, number of parties, and the ideological dispersion of involved parties 

(Franzese 2002). Of course, these issues may apply more broadly than to just parliamentary 

systems, since a divided congress can act as a restraint on presidential fiscal policy. These 

reasons are why veto-players are likely to be such a fundamentally important source of political 

pressure on the regime to abide by its commitments to fiscally sustainable policies. Veto-actors, 

similar to the argument advanced by Keefer and Stasavage (2003), make it more difficult for the 

executive to renege on budget commitments - both because they must also agree to reneging, 

but also because they face important political incentives not to do so. If the executive wishes to 

renege, it creates an opportunity for other political actors to attempt to assume the role of the 

responsible financial steward and attack the executive as endangering the countries' well-

being. I therefore state the next political hypothesis: 

Politics H2: Fiscal Rules and Fiscal Transparency should lead to smaller deficits as the number of 

veto-actors increases. 

Additionally, fiscal institutions should be more effective in democracies partially because 

democracies will have higher and more routine levels of political contestation. They also tend 

to have other critical institutions such as media freedom and greater political liberties which 

also may facilitate popular political pressure against the governments’ spending policies. In this 

view, democracies should augment the effectiveness of fiscal institutions not just because of 

 

23 

 

the electoral mechanism but because of the broader impact of a free society able to actively 

play a role in political outcomes. Consequently, a proper measure of democracy would not be a 

dichotomous measure, but one that is continuous and thus tracks these multiple democratic 

factors.  

Opposition vote share in the legislature is also an important manifestation of political 

contestation which may causally link fiscal institutions to fiscal discipline. While related to 

democracies insofar as democracies will tend to have higher levels of opposition vote shares, it 

nonetheless captures a distinct aspect of political competition because even in countries which 

lack free and fair elections there can still be meaningful levels of political contestation. If 

incumbents are motivated by the political backlash associated with deficit spending, this 

pressure could still be felt in nondemocracies with meaningful levels of opposition party 

participation in politics. Additionally, as previously argued, a strong opposition need not 

necessarily be in government in order to exert pressure on the executive, and as such 

opposition vote share is able to capture the strength of political opponents without mandating 

a legal role in government. Thus, opposition vote share may capture a meaningful and distinct 

aspect of political fiscal constraints. I thus state the last two political hypotheses: 

Politics H3: Fiscal Rules and Fiscal Transparency should lead to smaller deficits as the level of  

democracy increases. 

Politics H4: Fiscal Rules and Fiscal Transparency should lead to smaller deficits as the opposition 

vote share increases. 

Financial markets may exert pressure on governments, but it does not seem that markets can 

so clearly intervene to impede the incumbents’ policies as political competitors can. Fiscal rules 

 

24 

 

raise the costs of fiscal imprudence in the presence of political competition in more direct ways 

than with financial markets, and it is for this reason I argue that while the financial flogging 

hypothesis is plausible, it seems less likely as a causal mechanism of fiscal institutions.  

Having laid out the flogging and competition hypotheses, in the next section I will test these 

hypotheses against each other in order to ascertain which causal pathway enjoys more 

empirical support.  We will see that fiscal institutions (Fiscal Rules * Fiscal Transparency) do 

significantly reduce fiscal deficits, but that financial markets do not significantly impact fiscal 

balances. Instead, fiscal institutions are effective in environments of increased political 

competition.  

Data  

The main dependent variable for all of the models estimated is intended to capture if 

governments are improving their fiscal performance. As a consequence, central government 

fiscal balances (% of GDP) was chosen as the best representation of this concept. This variable 

comes from Bodea and Higashijima (2015), who updated the IMF’s data for central government 

fiscal balances by using the IMF Article IV Consultation Staff Reports. In the regressions, a 

positive value for the independent variables thus means that deficits are smaller for these 

countries (or a surplus is present), so a positive value corresponds with improved fiscal 

performance conceptually. 

To capture the essential concept of fiscal transparency, I employ the data from Vlaicu et al. 

(2014). Other data sources providing nuanced estimates of fiscal transparency have been fairly 

limited to either the US states, EU countries or a subset of OECD countries. Vlaicu et al. provide 

for a comprehensive coverage of countries using what are called Medium Term Budgeting 

 

25 

 

Frameworks (MTFs) and they examine the impact of these frameworks on fiscal performance. 

MTF's work in two ways, both by taking a longer-term budgeting perspective and by improving 

the transparency of the budget process. Since they take a medium-term budget perspective 

(typically between 3 to 5 years), it enables more prudent planning. But importantly, the process 

also involves forecasting economic variables, establishing budget objectives, often involve 

public forums for discussion, and publishing these forecasts. As a consequence, the budgeting 

process is more transparent16. The time sample for the regressions begin in 1990, as this 

denotes the beginning of the IMF fiscal rules data, and ends in 2008 as this is when the MTF 

database ends. 

In the IMF's 2007 manual on Fiscal Transparency, MTFs are said to ensure, when rigorously 

applied, "a transparent basis for the accountability of the executive branch." Additionally, 

Hameed (2005) specifically examined MTFs as a fiscal transparency index, and demonstrated 

that MTFs have a very high correlation coefficient of 0.84 with the IMF's ROSC scores, which are 

arguably the "purest" of fiscal transparency variables (but is only available as a cross-section). 

He further compared MTFs against other indices of fiscal transparency and argues that it 

performed best among the available measures, and argued that MTFs may "indicate a major 

step toward policy accountability" and can make it easier to assess deviations from planned 

expenditures. Keefer and Vlaicu (2008) built on this concept to argue that MTFs should be 

expected to reduce the clientelism and corruption of states, since this heightened scrutiny of 

public finances may make it more difficult for politicians to get away with graft. Using MTF's as 

                                                           

16 MTFs vary in their sophistication. For the purposes of transparency, however, even if the forecasting in the 
MTF is not of high quality, the public nature of MTFs is what enables it to be a fiscal transparency concept. 

 

26 

 

a proxy for fiscal transparency is therefore well-established in practice and provides good 

leverage because then we are able to analyze the impact of fiscal transparency for a notably 

greater time sample than other available measures. 

Data on Fiscal Rules comes from the IMF's Fiscal Rules Index, which codes the existence of a 

variety of types of fiscal rules for a sample of up to 78 countries. "Legal Basis” or “Rule Basis” 

captures the national level at which the fiscal rule is legally established. This is coded 1 through 

4, with 1 representing a "political commitment", 2 is a coalition agreement, 3 is statutory based, 

and 4 is a constitutional law. For all estimations, the Legal Basis is used in the interaction with 

fiscal transparency because of the expectation that a higher legal commitment to a fiscal rule 

should mean that it would be increasingly difficult to circumvent, and should be a more visible 

rule to political elites. While this has not been done previously in the literature, this expectation 

is theoretically well-grounded. For instance, Alt, Lassen, and Wehner (2012) and Alt and Lassen 

(2006) have established fiscal transparency is an important complement with fiscal rules (see 

also IMF (2007)), since without transparency fiscal rules may induce the use of “fiscal gimmicks” 

in order to circumvent the rule. While these dual institutions are widely viewed as important 

complements by the IMF, data limitations for fiscal transparency variables have limited the 

ability of researchers to explore the interplay of fiscal rules and transparency. Theoretically, 

therefore, the proper specification is the interaction of transparency and fiscal rules. 

The data on country credit ratings comes from Bodea and Hicks (2017), who have updated and 

expanded the data from Beaulieu, Cox, and Saiegh (2012b). Because other main independent 

variables for these models only begin coverage in 1985, I decided to limit the fiscal balance and 

sovereign ratings data to begin in 1985 as well. The sovereign credit ratings are used to 

 

27 

 

generate the variable "Threshold", which is a dummy variable capturing whether the sovereign 

rating is either (Baa2/BBB) or (Baa3/BBB-) from Moody’s Analytics, which would place a country 

2 downgrades away from crossing the threshold to the Speculative Grade investment quality. 

This would seem to theoretically capture the range around which market participants would 

exhibit greater fear of a downgrade (and therefore stronger incentive to lobby political leaders), 

since only one or two downgrades takes them out of the Investment Quality grade range. I then 

also create a dummy variable indicating whether a country is below the threshold and has a 

speculative grade credit rating. For both variables, Threshold and Speculative, there is a 

concern regarding simultaneity bias between these variables and fiscal balances, since large 

fiscal deficits should make it more likely a country will have poorer credit ratings. In the models, 

I therefore use the lags of Threshold and Speculative in order to mitigate this concern. 

The variable for Civil Society Participation comes from the VDEM version 7 database, 2.12. It 

specifically measures both the number of CSO’s and their regularity of involvement in public 

activities (or if they are suppressed), whether they are consulted in the process of policymaking, 

their involvement in candidate selection as well as female involvement and participation. 

Included can be a wide variety of interest groups, NGOs, labor unions, social movements or 

other professional organizations. The variable, along with many of the VDEM items, was 

estimated with Bayesian factor analysis, and is a continuous variable ranging from 0 to 1. The 

other domestic politics variables, including Opposition Vote Share and veto-players (Checks & 

Balances) come from the World Banks' Database of Political Institutions. Opposition vote share 

is the vote share of all opposition parties for the legislature. As a result, there are instances 

when vote share is larger than 50%, which is an indication of either divided or shared 

 

28 

 

government. For the Checks variable, low values indicate that there is no meaningful constraint 

on the executive. Legislatures are not elected and only the executive dictates policy. However, 

this increases as other actors gain influence. For instance, when the legislature is elected, and 

even moreso when the opposition party gains a majority in the legislature. Other conditions 

also increase the number of checks, such as in parliamentary systems it increases with 

additional parties in a government coalition. The democracy variable is the Polity2 score, scaled 

to range between 0 and 20. 

"Fixed" comes from Reinhart and Rogoff (2002), which is a dummy variable taking a 1 when the 

country is on a de facto fixed exchange-rate. The variable for Judicial Independence, utilized 

only as a control for the polity model, comes from Linzer and Staton (2011). The economic 

control variables are taken from the World Bank WDI database. In the models, economic 

growth, Debt/GDP and inflation are lagged one-year to reduce the concern for endogeneity 

with fiscal balances. The Inflation variable is GDP Deflator Inflation, and is logged to manage the 

severe skewness of the variable. To deal with negative values, the absolute value of inflation 

was first taken, after which the log of (1+ inflation) was performed. I then added a negative sign 

for negative observations of the original inflation variable. Additionally, the full list of countries 

in the main regression sample is also available in the appendix, in Table 1.A.6. 

Research Design 

Aside from the main theoretical variables, the model specification and control variables are 

largely inspired by Bodea and Higashijima (2015) and Vlaicu et al (2014) regarding the 

determinants of central government fiscal balances. The main estimation models are twostep 

System-GMM models with Windmeijer corrected robust standard errors, using orthogonal 

 

29 

 

deviations for unbalanced data with missing values as recommended by Roodman (2009). 

System GMM is very common when estimating fiscal balance equations, because most such 

studies have "Large-N, Small-T" samples, in which there is serious concern for dynamic panel 

bias with fixed-effects. System GMM avoids the dynamic panel bias by estimating the 

parameters in both first-differences (thereby expunging the fixed effects) and levels, and 

instrumenting the lagged-DV in differences and levels. It is also preferable to other dynamic 

panel data estimators such as the Anderson-Hsiao (A-H) or the difference GMM estimator, as 

both the fiscal balance data and the main fiscal institutions variables are extremely persistent 

and thus suffer from significant inefficiency in an A-H or D-GMM framework17, which relies on 

the difference transformation of the data alone.  

In the GMM models, I report the Hansen test of overidentifying restrictions as well as the 

Arellano and Bond (1991) test for AR(2) serial correlation, both of which are necessary to infer 

the consistency of the estimation and exogeneity of the instruments. In all GMM models, these 

statistics were insignificant, indicating that we fail to reject the null of exogeneity. Additionally, 

I estimated the baseline GMM model using pooled-OLS for the purposes of estimating a GMM 

credible bracketing range for the lagged dependent variable. Estimating the pooled-OLS model 

both without, and then with country fixed effects gives upper and lower bound estimates on 

the bias of the lagged-DV, and therefore GMM estimates where the lagged-DV fall outside this 

range “hint at specification error” (Roodman 2009a). Doing so yields a range of 0.68 to 0.81 on 

the lagged-DV. All of the GMM models estimated with the same fiscal balance dependent 

                                                           

17 In fact, this is what led to the development of the “system” GMM estimator, in order to improve the 
efficiency of dynamic models with highly persistent data (Wooldridge 2010; Roodman 2009a). 

 

30 

 

variable return a lagged-DV coefficient within the GMM credible range. Finally, I also use a 

collapsed instrument set to guard against the risk of overfitting bias with instrument 

proliferation (Roodman 2009b). However, in appendix Table 1.A.1, I also present the main 

GMM model but demonstrate that the results are consistent across various specifications, 

including instrumenting lags of 2, 4 or 6 periods and varying other estimation choices. 

Lastly, it is worth addressing the obvious fact that the mechanisms of political competition 

should all correlate with each other and could be considered alternate manifestations of 

democracy. It is thus crucial to include democracy in each model when analyzing the effects of 

opposition vote share, CSOs and veto players, because only then can we clearly distinguish that 

the mechanism is not simply a proxy for democracy. Thus, all models control for both 

democracy (polity) and veto players (checks and balances). Additionally, when investigating the 

role of democracy, I also control for judicial independence in order to isolate the other 

elements of democracy that theoretically are important. 

Empirical Results 

The key independent variables for all of the analysis, mapped from the hypotheses stated 

earlier in the paper, are as follows: 

Fiscal Institutions : Rule Basis * MTF 

Financial Flogging: Rule Basis * MTF * [Speculative Threshold / Speculative Grade / Fixed 

Exchange Rate / Emerging Market Economy] 

Domestic Politics: Rule Basis * MTF * [Opposition Vote Share / Civil Society Participation / Veto 

Players / Polity] 

 

 

31 

 

All of the key theoretical variables are therefore either two-way or triple interaction models. 

For interpretation, I follow Brambor, Clark, and Golder (2006) who argue that judging 

interactions only from their statistical significance (or lack thereof) can be very misleading. As a 

consequence, the interpretation comes mainly from the marginal effects plots of the 

interactions. All reported confidence intervals in the main marginal effects plots are 95% 

confidence intervals.  

A key contention of the paper is that fiscal rules and fiscal transparency will interact to produce 

higher budget balances (i.e. smaller deficits). In Table 1.1, I report the main GMM results for the 

baseline specification. Model (1) is the baseline model and was used to produce the marginal 

effects plots in Figure 1.1. We see from Table 1.1 that the Rule Basis * MTF interaction is 

significant and positive in line with theoretical expectations.  

We also see in Figure 1.1 that both MTF and fiscal rules improve budget balances at higher 

values of the other. Rule Legal Basis increases budget balances but only when MTF=1, and the 

difference between these coefficients in Figure 1.1 (A) is significant at p<0.05. Conversely, MTF 

increases balances, but only significantly once the legal basis is equal to a value of at least 2, 

which represents rules governed by a coalition agreement. Additionally, the estimated effect 

appears to be large, since MTF is estimated to improve budgets by about 1% of GDP when the 

fiscal rule is constitutionally based. For a country with a starting Debt/GDP ratio of 100, over 15 

years this would lead to a ceteris paribus difference of approximately 20% Debt/GDP18. These 

findings are a novel contribution to the literature because it shows that neither fiscal institution 

                                                           

18 This assumes a baseline level of deficit spending of about 2.5% of GDP, which is roughly average for a 
country with such a debt level. This is according to data from the IMF Public Debt Database.   

 

32 

 

is effective without the other, contrary to the earlier fiscal rules literature (Kelemen and Teo 

2014; Alt 2001; Bohn and Inman 1996).  

The output results for the triple interactions are reported in Appendix Table 1.A.2, since the 

interpretation for these models hinges entirely upon the marginal effects plots. Figure 1.2 (C) is 

therefore the marginal effect plot corresponding to model 9 in the appendix. In Figure 1.2, we 

see the marginal effect of each financial flogging condition across the different values of the 

fiscal institutions. Rules Basis is on the x-axis, and the point estimate and confidence intervals 

estimate the difference of the effect when MTF=1  vs when MTF=0. We should therefore expect 

that the effect of the financial market condition should lead to significant and positive 

differences as the Rule Basis increases. However, in each of the figure 1.2 subplots C through F, 

there does not appear to be a statistically significant effect. In all of the models, the effect of 

the financial condition Speculative Threshold (lag), Speculative Grade (lag), Fixed Exchange Rate 

and Emerging Market all do not seem to significantly vary in their impact on fiscal balances. In 

appendix Figure 1.A.2, I examine the marginal effect of fiscal institutions with the flogging 

condition as the moderating variable, but this examination also reveals no support for the 

flogging hypotheses. 

I turn now to a discussion of the results for the political competition models. Plot (H) of Figure 

1.3 shows the marginal effect of Rule Basis across the range of opposition vote share in the 

sample, and reports the difference of the marginal effect of Rules Basis when MTF=1 vs MTF=0. 

Doing so enables us to more clearly see that the effect between these estimates is significantly 

different from each other. As expected, we see that the effect of Rule Basis is increasing along 

with the opposition vote share and this effect is significantly different from 0. However, when 

 

33 

 

MTF=0, Rule Basis has no effect. This underscores the importance of having both fiscal 

institutions for there to be fiscal discipline. Importantly, we in plot (H) that opposition vote 

share becomes statistically significant at roughly 30% vote share, controlling for the countries 

level of democracy. Thus, we have isolated the effect of opposition vote share aside from the 

democracy level (this is also the case with plots (I) and (J)). It is further noteworthy, since there 

are non-democracies in the sample who have an opposition vote share greater than 30%, such 

as Zambia, Botswana, or Burkina Faso (and more). 

Plot (I) shows that fiscal institutions are indeed more effective at reducing deficits when the 

level of civil society participation is highest, and this effect is statistically significant and covers a 

good portion of the variation. Therefore, this shows the effect of civil society engagement apart 

from a countries’ level of democracy. We find further validation with the other political 

variables as well.  In plot (J), fiscal institutions significantly increase budget balances once the 

number of Checks & Balances is at least 3. And Plot (K) confirms that fiscal institutions, 

controlling for a countries Checks & Balances and level of judicial independence, have a 

significantly positive effect when a country has a democratic Polity score of 6 or greater. This is 

suggestive that there is in fact something about democracy which successfully promotes fiscal 

discipline that is aside from the rule of law or institutional constraints such as veto actors. 

Taken as a whole, we therefore see from these marginal effects plots that the financial flogging 

hypothesis fails to find empirical support in any of the models. At the same time, domestic 

political competition seems to be an important conditioning factor for whether fiscal 

institutions are able reduce deficit spending. Both the role of political elites matters in addition 

to an active civil society for these fiscal institutions to be effective. 

 

34 

 

 

 

35 

(1)(2)(3)(4)(5)(6)(7)(8)Robust to:BaseRegion DummiesExcluded EurozoneRbasis*MTF [lagged 3 Yrs]DV: Forward 5yr Fiscal AveOLS-DKFixed EffectsEndogeneityRule Basis-0.088-0.083-0.175*-0.178*-0.323*-0.07-0.038-0.177**(0.076)(0.082)(0.092)(0.093)(0.171)(0.098)(0.118)(0.086)MTF-0.165-0.157-0.3930.064-0.004-0.06-0.331-0.517**(0.222)(0.214)(0.371)(0.242)(0.475)(0.213)(0.425)(0.245)Legal Basis * MTF0.274***0.233**0.353**0.310**0.697**0.240**0.257*0.335***(0.081)(0.089)(0.133)(0.122)(0.224)(0.097)(0.134)(0.096)Fiscal Balance (lag)0.719***0.712***0.693***0.687***0.295**0.803***0.678***0.629***(0.048)(0.048)(0.056)(0.060)(0.094)(0.048)(0.041)(0.052)Polity-0.037-0.04-0.055-0.054-0.038-0.0480.034-0.089*(0.034)(0.034)(0.043)(0.047)(0.070)(0.043)(0.047)(0.047)Checks and Balances-0.042-0.033-0.036-0.012-0.009-0.03-0.08-0.052(0.035)(0.032)(0.057)(0.032)(0.068)(0.054)(0.052)(0.044)Executive Election-0.400**-0.407**-0.194-0.529**0.114-0.488**-0.438**-0.358**(0.133)(0.134)(0.186)(0.178)(0.122)(0.186)(0.158)(0.122)(log) GDP-pc0.1630.410.2620.1740.2380.193*2.024*0.072(0.149)(0.272)(0.202)(0.166)(0.298)(0.104)(1.138)(0.206)Trade (% GDP)0.004*0.005**0.006**0.005*0.017***0.004-0.0060.007(0.002)(0.002)(0.002)(0.003)(0.003)(0.003)(0.013)(0.004)Growth (lag)0.055**0.058**0.0640.0480.0430.0460.0620.059**(0.028)(0.029)(0.041)(0.040)(0.047)(0.027)(0.043)(0.027)(log) Inflation (lagged)0.192***0.176***0.186**0.153**-0.1050.210**0.214**0.214***(0.044)(0.041)(0.082)(0.069)(0.162)(0.077)(0.078)(0.055)(log) Debt/GDP (lagged)0.1980.368*0.0630.155-0.1860.2050.864**0.083(0.181)(0.192)(0.223)(0.188)(0.346)(0.223)(0.338)(0.222)Fixed Exch. Rate0.0420.0040.221-0.163-0.5330.0620.5560.328(0.216)(0.228)(0.387)(0.256)(0.486)(0.220)(0.396)(0.328)Internal Conflict-0.131***-0.120***-0.094*-0.163***-0.096-0.125**-0.054-0.151***(0.034)(0.032)(0.049)(0.049)(0.110)(0.053)(0.083)(0.041)Resource Rich Dummy0.843**0.703*0.938*0.948**2.028**1.063**.1.042**(0.357)(0.370)(0.494)(0.420)(0.900)(0.394).(0.451)Lambda (IMR)-0.308**-0.13Constant-2.363**-6.024**-2.301-2.259-3.631-2.390**-22.636**0.967(1.028)(2.479)(1.637)(1.524)(2.693)(1.013)(10.823)(1.883)Observations957957714921873957957924# of Countries6565646563656564Ave Years/Country14.72314.72311.15614.16913.85714.72314.438Hansen p-value0.7890.7990.5140.4650.2440.871AR(2) p-value0.6550.6420.6720.590.9930.471# of Instruments (GMM)566156575657Table 1.1: Fiscal Institutions Models and Robustness Specifications* p<0.05 ** 0.01 *** 0.001; All models except for model 5 use the DV "Fiscal Balance (% GDP)". Model 5 employs a 5 year average of the *future* fiscal balance as the DV. All models except for 6 and 7 are System GMM models with Windmeijer corrected robust standard errors. Models 6 and 7 are OLS models with Driscoll-Kraay standard errors, with model 7 including country fixed effects. Model 8 is the same as model 1, except that it controls for the (endogenous) probability of country adoption of BOTH fiscal institutions. All models include year fixed effects but not reported. Model 2 includes regional dummies and a dummy for "Developing" countries, but are omitted due to space. 

                                                                                                                                                                            

 

36 

 

 

Endogeneity of Fiscal Institutions 

Countries who adopt fiscal rules may inherently possess a different inclination towards 

budgeting discipline than states who do not, and as a consequence there could be a positive 

selection bias when estimating the effect of these fiscal institutions. In order to address this 

endogeneity risk, I follow a similar approach to that of DeMeritt (2012) or Bernhard and 

LeBlang (2008)19 by first estimating a random-effects probit20 model predicting the joint 

                                                           

19 While they do not study fiscal balances explicitly, their approach to dealing with endogeneity is relevant and 
applicable for the endogeneity concern in this paper. These are both similar in spirit to my approach here, 
although more closely matches DeMeritt (2012) in that I also calculate the IMR as a control function. Similar to 
Bernard and LeBlang (2008), I also tried calculating predicted probabilities, and the main inferences were 
unchanged. 
20 A pooled probit was also estimated, but I use the random effects probit because it yielded better out of 
sample estimates. Inferences were the same with both approaches, however. 

 

37 

 

adoption of the fiscal institutions, and then calculate the Inverse Mills Ratio21 to model the 

endogenous nature of fiscal institutions adoption . I then use IMR as a control in the original 

GMM equations models, which use two-step corrected standard errors. As a consequence, I am 

able to say that, controlling for a countries inherent propensity to adopt these fiscal 

institutions, the institutions themselves have an independent effect on a countries fiscal 

balances. 

The reason why it is necessary to model the adoption of both fiscal rules and fiscal transparency 

is because I have argued, and have shown empirically in Figure 1.1, that having one type of 

institution without the other is insufficient to impact country fiscal balances. For example, a 

country could adopt a fiscal rule knowing that monitoring will be difficult because transparency 

is low. In such a case, controlling for the probability of fiscal rule adoption without fiscal 

transparency could be meaningless because they are insincere rule adopters and would not 

actually be perceiving a fiscal constraint with its adoption. While this is the main strategy, in the 

appendix I also report models where instead of predicting the endogeneity of joint adoption, I 

separately predict the probabilities of MTF adoption, Rule Basis adoption, in addition to joint 

adoption. I then predict separate IMR ratios from each model and include all of these in the 

main GMM estimations. Inferences remain virtually unchanged based on this specification, and 

I report the triple interactions from these “triple endogeneity” control models in appendix 

Figure A5. 

                                                           

21 This is the ratio of the Standard normal PDF / 1 – standard normal CDF, pertaining to fiscal institution 
adoption. 

 

38 

 

For the model specification, I include all of the political competition variables so that we can be 

sure that political factors influencing fiscal institution adoption are not biasing the estimates of 

these political factors in the second stage fiscal balances regressions. As an example, I control 

for the role of Opposition Vote share in adopting fiscal institutions, such that we can say in the 

main fiscal balance model that Opposition Vote share independently influences the 

effectiveness of the institutions even apart from their adoption. I therefore include Civil Society 

Participation, Polity, Checks, and Opposition Vote Share in the model. In order to facilitate 

identification of the parameters in the second stage outcome models, in these adoption models 

I include as an excludable measure the number of fiscal institutions adopted globally in a given 

year. Theoretically, prominence of global adoption of fiscal institutions may influence a 

countries’ decision process to adopt fiscal institutions themselves, but the number of global 

fiscal rules should not be related to a countries domestic fiscal balance. It is noteworthy that, in 

appendix table 1.A.3, we see that “world rules” is a significant predictor of rule adoption22. In 

addition, I also include two measures which significantly predict fiscal institutions adoption, but 

which are insignificant in the fiscal balances model23. These are the (log) population and a 

dummy variable “Ever Default” indicating whether a country has ever defaulted on its debt 

obligations. A joint test of these variables in the probit adoption model shows they are jointly 

significant at p<0.001.  

                                                           

22 Alternatively, I also used in separate tests (though not reported) “Government Effectiveness” as an 
excludable variable, since this likely could correlate with the adoption of fiscal institutions but is unlikely to 
affect fiscal balances. The correlation coefficient of Gov’t Effectiveness with fiscal balances is a low 0.1 and 
very significantly predicts fiscal institutions adoptions. This approach yielded the same conclusions as reported 
in the paper 
23 In a fixed effects model in the outcome fiscal balances equation, these measures were jointly insignificant at 
p > F = 0.6. 

 

39 

 

I also include a control for EuroZone membership, since these countries likely have different 

structures surrounding their fiscal procedures. While the economic control variables are the 

same as those in the main GMM model, I omitted the dummy variable for resource wealth, 

internal conflict variable, openness and executive election24. This choice was driven by the fact 

that the Z-statistics for all of these were below 0.7, the lowest being unemployment with 0.38. 

An F-test of this group of variables yielded a χ2 p-value of 0.8. These variables therefore seem 

reasonably excludable, and omitting them returned a slightly higher Area under the ROC curve, 

suggesting an improvement in the models ability to predict institutional adoption. Finally, for 

this to successfully control for the endogenous nature of fiscal institutions, it is important that 

the adoption model has a reasonably strong fit of the data. The Area under the ROC curve for 

the model is a very high 0.89, giving us little cause for concern that it is poorly modeled here. 

Appendix Table 1.A.3 shows the random effects probit fiscal institutions adoption models. In 

Model (8) of Table 1.1, I report the baseline GMM regression results including the IMR 

reflecting joint adoption, therefore controlling for a countries propensity to jointly adopt these 

fiscal institutions. In appendix Table 1.A.3, all of the triple interaction models are shown 

controlling for the adoption of fiscal institutions. In almost all models, the IMR is significant, 

suggesting that it is indeed necessary to control for this endogeneity. However, it appears to 

have virtually no impact on the inferences made for both the flogging and political competition 

                                                           

24 I also followed a different route from the procedures guiding model specification I describe here. I also tried 
more explicitly following the recommended approach to identification for a Heckman selection model, even 
though a Heckman method is slightly different than the nature of my endogeneity problem here. In this case, I 
used all of the same covariates from the fiscal balances equation as predictors in the probit adoption model, 
while adding the two excludable variables (log)population and Ever Default. Using this simpler approach yields 
the same inferences for the influence of the fiscal institutions, but I choose to report the method above 
because I believe it gives a better specified model for the probit fiscal rule adoption equation. 

 

40 

 

hypotheses. In just the two-way interaction of the fiscal institutions (model 8), the coefficient of 

the interaction is actually larger than the previous estimations. The results hold as well for all of 

the triple interaction models, and Figure 1.A.2 reports all of the political competition 

interactions calculated from the endogeneity control models. As noted, these appear 

unchanged or even stronger relationships than in the initial estimations. 

As a consequence, it seems that our inferences regarding the impact of fiscal institutions 

appear robust to consideration of the endogenous country-adoption of these institutions. Fiscal 

rules and fiscal transparency do improve fiscal balances, but only in countries which have 

strong enough political competition to constrain the executive and a civil society capable of 

pressuring the executive. At the same time, despite a more clearly specified theoretical 

framework regarding the potential influence of financial markets, the financial flogging 

hypothesis is not supported by the evidence investigated here. It may yet be possible to 

uncover such influence, but doing so will likely require additional theorizing regarding the 

nuances which make politicians feel vulnerable to the conditions of financial markets along the 

lines I have done here. 

Further Robustness 

In addition to addressing the concern regarding endogenous fiscal institutions, I also examined 

the sensitivity of the results to a variety of other specifications, the results of which are 

reported in Table 1.1. Out of concern that the results may be driven by sub-regions, in Model 

(2) I include regional dummies as well as a dummy indicator for developing countries. These 

were not significant and are omitted from the table to conserve space. Out of concern that 

Eurozone countries may be unduly influencing the regression results, in Model (3) I estimated 

 

41 

 

the model but excluding such country-members. Next, I estimated the GMM model using 3 year 

lags of the fiscal institutions (reported in Model (4)), and the (L3)Rule Legal Basis * (L3)MTF 

interaction remained positive and significant. Alternatively, I also estimated the model but with 

an alternative dependent variable. In model (5), instead of current year fiscal balances I used 

the forward 5 year average25 of the fiscal balance as the DV. This is a useful measure because it 

gives a better sense of the long-run change in fiscal balances than simply regressing on the 

current time periods value. Here, we see that the coefficient on the Rule Basis * MTF 

interaction is 44% larger and remains significant. This larger effect makes sense, because 

regressing only on the current period does not account for the dynamic response of the fiscal 

balance to an increase in fiscal institutions.  

Additional robustness26 procedures include using a logarithmic form of the fiscal balance as the 

dependent variable, as well as switching to using pooled-OLS with Driscoll-Kraay standard 

errors. The value of using OLS with DK standard errors is that these are robust to 

contemporaneous correlation, which the GMM standard errors do not account for. In the GMM 

estimations, I included year fixed effects in all of the models but are omitted from the tables. As 

Roodman (2009a) suggests, this is a way of mitigating the concern that contemporaneous 

shocks are biasing the estimates. Nonetheless, it is fruitful to examine the model performance 

with standard errors actually suited to the task. In model (6) we see the results hold. In model 

(7), I report an OLS-DK model but include country fixed effects, to ensure the results are not 

                                                           

25 Therefore, it excludes the current year fiscal balance in time ‘t’.  
26 Another concern may be whether there are issues of misattribution arising from countries abandoning a 
fiscal rule. However, only 3 countries in this sample actually abandoned their fiscal institutions, and the results 
hold when excluding the period after the withdrawal. 

 

42 

 

driven primarily by between country variance. In another appendix table 1.A.5, I demonstrate 

the triple interactions for the political competition variables but split according to different 

samples. The include restricting the sample to non-European countries, to developing countries 

only, excluding dictatorships and lastly excluding the highest level of democracy (9 or 10 polity 

score). In these cases we see supportive evidence for the original hypotheses. As a means of 

assessing robustness to outlier countries, I also report estimation results where the main GMM 

model is run but excluding one country from each estimation. We see in table A(5) that the 

results from the Rule Basis * MTF interaction are significant and highly consistent across all of 

these regressions. Because the table lists the country omitted for the relevant coefficient, it 

also serves as a default list of countries which comprise the main regression sample.  

Finally, another concern is that the significant results for the fiscal institutions may be resultant 

from overfitting of the data. To investigate this possibility, I performed a 10-fold cross-

validation procedure, where one-tenth of the data (the test set) is randomly held-out of the 

regression sample, and then predicted values of fiscal balances are used to assess model fit in 

the held-out sample. I performed this technique both with and without the fiscal institutions 

variables, in order to assess which model has lower out-of-sample forecast error. Because the 

model is linear, I use the RMSE to assess forecast error in the held out sample. In this case, 

because the cross-validation procedure is reducing the sample for the main estimation and 

because GMM models have limiting properties in small samples (Roodman 2009a), I used a 

pooled-OLS model for this procedure, although results were similar when using the GMM 

model. The average forecast RMSE for the estimation without fiscal institutions was 2.98, 

whereas including the fiscal institutions reduced the average RMSE to 2.78, indicating that 

 

43 

 

including the fiscal institutions has improved the ability to predict out-of-sample. To illustrate 

this same process visually, I also estimated the model with and without the fiscal institutions on 

a restricted time sample, up until 2003. I then predicted the (Y-hat) fiscal balances for the 

remaining period after 2003, and show these predictions for a small set of countries in 

appendix Figure 1.A.5. We can then visually see what the cross-validation procedure indicates, 

which is that including the fiscal institutions in the model improves the ability to predict out of 

sample, and therefore overfitting bias does not seem to be an issue in this case. 

Conclusion 

It has long been argued that financial markets are able to compel political leaders to restrain 

deficit spending, yet there has been fairly limited investigation of this exact hypothesis. In this 

paper I have laid out a clearer theoretical framework from which we might expect that financial 

markets do compel fiscal discipline. Despite this, we fail to find evidence of a significant role for 

markets. Instead of the usual mechanisms – either markets or the rule of law – I have argued 

that an important causal mechanism missing from this discussion is the role of domestic 

political competition. Because political opponents can use fiscal rules as leverage against the 

executive, in the presence of such fiscal institutions the executive is actually compelled to 

reduce deficit spending. Furthermore, we have evidence that a vibrant civil society is able to 

function as an additional form of political-fiscal pressure. These findings are noteworthy 

because it indicates that these fiscal rules could be effective constraining devices even for 

countries who lack robust and independent judiciaries. This paper thus highlights how and why 

such institutions may be effective and in what environments globally, which may help 

policymakers in the design of such policies.  It also furthers a research agenda pursuing the 

 

44 

 

understanding of how and under what conditions political leaders’ policies are shaped by 

international economic conditions. While I found that there is a lack of such evidence with 

respect to fiscal policy, this paper contributes to a more nuanced research program seeking to 

uncover such potential influences. It may yet be that there are international financial market 

influences, but in an effort to uncover them we first need a clearer framework of analysis to 

detect and untangle them. At the same time, it seems clear that the evidence thus far is much 

more supportive of the role of domestic politics than it is for financial markets. The paper thus 

is an important advance in a long line of research contending for this influence. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

45 

 

 

APPENDIX 

46 

 

Appendix  

 

 

 

47 

(1)(2)(3)(4)(5)(6)(7)Base Model-----No OrthogonalInstrumenting Lagscollapsed2 years2 years4 years6 years4 yearscollapsedlevels or diff equation?bothbothDifferenceDifferenceDifferenceBothBothRule Basis-0.132*-0.128*-0.120-0.118-0.102-0.121-0.0960(0.0790)(0.0774)(0.0794)(0.0908)(0.0980)(0.0978)(0.0864)MTF-0.366*-0.394-0.370-0.308-0.173-0.299-0.324(0.214)(0.297)(0.290)(0.262)(0.284)(0.335)(0.240)Rule Basis * MTF0.297***0.243**0.252**0.202**0.226**0.202*0.246**(0.0843)(0.107)(0.0967)(0.0975)(0.108)(0.109)(0.107)Fiscal Balance (t-1)0.697***0.746***0.797***0.781***0.724***0.728***0.733***(0.0449)(0.0599)(0.0751)(0.0622)(0.0635)(0.0733)(0.0643)Polity-0.0437-0.0698-0.0672-0.0840*-0.0465-0.0677-0.0390(0.0355)(0.0496)(0.0452)(0.0438)(0.0496)(0.0593)(0.0377)Checks and Balances-0.0422-0.0410-0.0503-0.0564-0.0661-0.0178-0.0393(0.0375)(0.0474)(0.0491)(0.0557)(0.0584)(0.0499)(0.0406)Executive Election-0.367**-0.406**-0.435**-0.386**-0.465**-0.405**-0.376**(0.141)(0.175)(0.169)(0.159)(0.148)(0.168)(0.136)(log) GDP-pc0.244*0.256*0.274*0.3790.3840.423*0.171(0.143)(0.155)(0.151)(0.296)(0.273)(0.232)(0.164)Trade (% GDP)0.00460*0.002200.001740.001930.003890.003810.00444*(0.00252)(0.00235)(0.00219)(0.00290)(0.00262)(0.00315)(0.00258)(lag) Growth0.0506*0.0506*0.03820.03250.0621*0.0449-0.0171(0.0265)(0.0285)(0.0333)(0.0331)(0.0349)(0.0358)(0.0335)(lag) (log) Inflation0.211***0.245***0.196**0.234**0.286**0.232**0.169***(0.0466)(0.0651)(0.0627)(0.0949)(0.0924)(0.0895)(0.0494)(lag) (log) Debt/GDP0.1680.1620.2720.1970.2880.3540.210(0.188)(0.168)(0.207)(0.366)(0.411)(0.318)(0.177)Fixed Exch. Rate-0.0269-0.0781-0.0944-0.05560.01490.08360.0127(0.230)(0.231)(0.196)(0.262)(0.260)(0.290)(0.213)Internal Conflict-0.128***-0.139**-0.115**-0.0981-0.0618-0.0726-0.144***(0.0357)(0.0608)(0.0511)(0.100)(0.0892)(0.0941)(0.0381)Resource Rich Dummy0.815**0.835**0.707**0.724*1.090**1.135**0.965**(0.374)(0.385)(0.348)(0.415)(0.494)(0.456)(0.346)Constant-3.334**-2.628**-3.191**-3.885-5.718-5.559-2.484**(1.020)(1.254)(1.338)(4.695)(4.219)(3.836)(1.123)Observations957957957957957957957# of Countries65656565656565Ave # years/country14.7214.7214.7214.7214.7214.7214.72Hansen p-value0.6900.4150.1280.9521.0001.0000.918AR(2) p-value0.6570.6850.6670.6760.6450.6580.637# of instruments5468498512010454Table 1.A.1: GMM Robustness SpecificationsDependent Variable is Fiscal Balance (%GDP). All Models estimated by twostep System GMM with Windmeijer corrected standard errors, but vary other estimation criteria in each column to demonstrate robustness to alternate specification choices. Column 1 is identical to estimation reported in table 1. All models use orthogonal deviations rather than the difference transformation, with the exception of column 7. 

 

48 

 

(9)(10)(11)(12)(13)(14)(15)(16)Theory:FloggingFloggingFloggingFloggingPoliticsPoliticsPoliticsPoliticsVariable for interaction:Speculative ThresholdSpeculative GradeFixed EREmerging MktOppVoteCivil SocietyChecksPolityRule Basis0.039-0.086-0.0170.019-0.031-0.486-0.005-0.697(0.258)(0.177)(0.080)(0.081)(0.158)(0.376)(0.175)(0.425)MTF0.2250.106-0.177-0.0550.2064.303**0.3162.662(0.831)(1.069)(0.271)(0.237)(0.714)(2.067)(0.575)(1.695)Legal Basis * MTF0.1690.4290.234**0.221**-0.009-1.739**-0.011-0.363(0.232)(0.334)(0.083)(0.081)(0.291)(0.793)(0.234)(0.647)[variable]-0.268-1.2360.1880.632*-0.002-0.473-0.045-0.043(2.956)(2.703)(0.291)(0.325)(0.007)(1.497)(0.051)(0.033)Rule Basis * [variable]-0.720.405-0.29-0.21-0.0020.457-0.020.032(1.932)(0.531)(0.222)(0.131)(0.003)(0.454)(0.031)(0.023)MTF * [variable]-1.2572.16-0.0220.025-0.01-5.101**-0.128-0.148*(2.716)(6.947)(0.372)(0.301)(0.016)(2.365)(0.134)(0.088)Rbasis * MTF * [variable]1.381-1.2060.202-0.0290.0072.270**0.0720.034(1.360)(2.446)(0.262)(0.141)(0.007)(0.915)(0.051)(0.034)Fiscal Balance (lag)0.778***0.777***0.717***0.719***0.710***0.699***0.719***0.725***(0.068)(0.066)(0.048)(0.047)(0.050)(0.049)(0.048)(0.051)Polity-0.076-0.13-0.035-0.053-0.039-0.039-0.039(0.140)(0.184)(0.035)(0.036)(0.034)(0.037)(0.037)Checks and Balances-0.099-0.124-0.039-0.039-0.041-0.036-0.031(0.076)(0.094)(0.035)(0.037)(0.039)(0.039)(0.036)Executive Election-0.394**-0.380**-0.410**-0.410**-0.399**-0.376**-0.396**-0.389**(0.163)(0.175)(0.135)(0.133)(0.134)(0.124)(0.133)(0.135)(log) GDP-pc-0.604-0.9270.1630.2040.2050.2160.1830.265(1.160)(1.270)(0.154)(0.163)(0.147)(0.172)(0.153)(0.208)Trade (% GDP)0.002-0.0010.005**0.004*0.005*0.0050.005**0.007**(0.005)(0.007)(0.002)(0.002)(0.003)(0.004)(0.003)(0.003)Growth (lag)0.0350.0160.054**0.051*0.058**0.050*0.055**0.054*(0.116)(0.098)(0.026)(0.028)(0.028)(0.029)(0.028)(0.029)(log) Debt/GDP (lagged)-0.107-0.1410.1820.2530.1910.1770.1830.252(1.029)(0.736)(0.177)(0.187)(0.187)(0.200)(0.179)(0.176)(log) Inflation (lagged)-0.114-0.1410.199***0.159***0.207***0.183***0.197***0.208***(0.472)(0.327)(0.047)(0.042)(0.046)(0.047)(0.044)(0.046)Fixed Exch. Rate0.340.5780.1280.020.0090.032-0.02(0.487)(0.605)(0.205)(0.227)(0.246)(0.216)(0.225)Internal Conflict-0.293-0.352-0.133***-0.140***-0.126***-0.128***-0.128***-0.104**(0.322)(0.327)(0.037)(0.030)(0.035)(0.038)(0.034)(0.037)Resource Rich Dummy0.6811.0220.872**0.812**0.896**0.912**0.859**0.864**(0.796)(0.861)(0.356)(0.374)(0.366)(0.387)(0.351)(0.330)Judicial Independence-0.262(1.031)Observations703703957957957938957957# of Countries5050656565646565Ave Years/Country14.0614.0614.72314.72314.72314.65614.72314.723Hansen p-value0.9710.9770.7790.8290.8120.8620.8030.783AR(2) p-value0.240.2030.6560.6330.6540.450.6630.658# of Instruments (GMM)6060596060605960Table 1.A.2: Triple Interaction ModelsStandard Errors in Parantheses. * p<0.1 ** p<0.05 *** 0.001;  DV = Fiscal Balance (% GDP). All models are System GMM models with twostep Windmeijer corrected robust standard errors. Year fixed effects estimated in all models, not reported  

 

 

 

 

 

49 

(1)(2)(3)DV = MTFRbasisJoint Adopt(lag) Fiscal Balance0.240*-0.0110.053(0.143)(0.019)(0.040)World # of Rules0.285*0.062**0.078**(0.155)(0.026)(0.036)Polity0.496-0.0130.121**(0.386)(0.037)(0.044)Checks and Balances0.0010.017-0.064(0.175)(0.057)(0.084)Opposition Vote Share-0.0080.0090.018*(0.020)(0.007)(0.010)Civil Society Participation0.011-0.019*-0.011(0.060)(0.011)(0.017)(log) Population1.866*0.410***0.542**(1.018)(0.116)(0.168)EuroZone3.7070.786*1.174(3.064)(0.455)(0.746)(lag) Debt/GDP-2.963-0.308-0.658**(1.829)(0.264)(0.322)Ever Default-6.844**-0.074-0.844(3.228)(0.533)(0.585)(log) GDP-pc1.4800.490**0.332(1.225)(0.198)(0.295)Growth-0.0670.030**0.045(0.073)(0.014)(0.032)(log) Inflation-0.323-0.173**-0.366**(0.373)(0.072)(0.113)Fixed Exch. Rate2.939-0.871**-0.977*(2.777)(0.335)(0.502)Time Trend0.667*0.0140.057(0.384)(0.037)(0.083)Constant-1379.561*2.287**-127.690(783.886)(0.721)(166.793)lnsig2u4.182***1.216**(0.368)(0.407)Observations102012931020Table 1.A.3: Fiscal Institution Adoption (Selection) ModelsDependent Variable changes per model. Model 1 is a binary MTF indicator, model 2 is the legal basis of a fiscal rule (range from 0 to 5), and 3 is a binary indicator of joint adoption. Models 1 and 3 use a random effects probit, with model 2 being a random effects ordinal probit. Clustered Standard Errors are in parantheses.  

 

 

50 

(9)(10)(11)(12)(13)(14)(15)(16)Theory:FloggingFloggingFloggingFloggingPoliticsPoliticsPoliticsPoliticsVariable for interaction:Speculative ThresholdSpeculative GradeFixed EREmerging MktOppVoteCivil SocietyChecksPolityRule Basis-0.188-0.275-0.0790.045-0.134-0.898**-0.078-1.200**(0.163)(0.183)(0.102)(0.093)(0.217)(0.382)(0.219)(0.445)MTF-0.881-0.898-0.486-0.372-0.2053.487-0.1221.88(0.570)(1.201)(0.321)(0.274)(0.600)(2.188)(0.520)(1.468)Rule Legal Basis * MTF0.2510.588*0.262**0.240**-0.047-1.528*0.042-1.122(0.171)(0.330)(0.119)(0.091)(0.323)(0.888)(0.225)(0.723)[variable]-1.603-0.7720.530.718*-0.007-0.984-0.063-0.113**(2.655)(2.681)(0.393)(0.385)(0.008)(1.488)(0.072)(0.044)Rule Basis * [variable]-1.050.419-0.448-0.402**-0.0010.877*-0.0250.056**(1.353)(0.592)(0.333)(0.161)(0.004)(0.491)(0.041)(0.024)MTF * [variable]-0.412.113-0.1340.041-0.008-4.434*-0.114-0.124(3.633)(6.206)(0.403)(0.355)(0.015)(2.562)(0.135)(0.078)Rbasis * MTF * [variable]2.452-1.3260.386-0.0010.012.061**0.0760.074**(2.220)(2.271)(0.378)(0.174)(0.007)(1.032)(0.050)(0.037)Fiscal Balance (lag)0.758***0.713***0.626***0.622***0.614***0.640***0.625***0.624***(0.078)(0.077)(0.050)(0.052)(0.053)(0.049)(0.052)(0.053)Polity-0.199-0.218-0.090*-0.119**-0.094**-0.091**-0.091*(0.265)(0.245)(0.046)(0.048)(0.048)(0.044)(0.050)Checks and Balances-0.197*-0.145-0.048-0.047-0.059-0.048-0.053(0.105)(0.139)(0.045)(0.047)(0.042)(0.044)(0.043)Executive Election-0.313*-0.421**-0.373**-0.381**-0.355**-0.358**-0.352**-0.353**(0.163)(0.160)(0.126)(0.121)(0.123)(0.117)(0.120)(0.123)(log) GDP-pc-1.707-1.3890.0490.0690.1130.0760.090.092(1.817)(1.921)(0.212)(0.223)(0.197)(0.204)(0.214)(0.243)Trade (% GDP)0.0060.012**0.007*0.008*0.0070.008*0.0080.009**(0.008)(0.006)(0.004)(0.004)(0.005)(0.005)(0.005)(0.004)Growth (lag)-0.0420.0080.057**0.054**0.061**0.051*0.058**0.061**(0.138)(0.112)(0.026)(0.028)(0.027)(0.029)(0.026)(0.029)(log) Debt/GDP (lagged)-0.2940.3570.0620.1470.0870.1760.060.117(0.832)(0.676)(0.216)(0.233)(0.234)(0.215)(0.222)(0.230)(log) Inflation (lagged)-0.1150.1640.218***0.197***0.249***0.221***0.220***0.254***(0.362)(0.193)(0.057)(0.054)(0.063)(0.054)(0.055)(0.058)Fixed Exch. Rate0.9461.3990.450.3850.2380.2930.315(1.196)(1.407)(0.326)(0.349)(0.325)(0.331)(0.317)Internal Conflict-0.519-0.335-0.157***-0.165***-0.143***-0.151***-0.148***-0.128**(0.436)(0.373)(0.045)(0.037)(0.040)(0.042)(0.041)(0.043)Resource Rich Dummy0.4731.3481.066**1.077**1.079**1.117**1.064**1.179**(0.711)(0.879)(0.452)(0.490)(0.449)(0.441)(0.436)(0.440)Lambda (IMR)-1.283-1.78-0.328**-0.365**-0.368**-0.343**-0.317**-0.379**(1.470)(1.429)(0.132)(0.135)(0.146)(0.131)(0.130)(0.126)Judicial Independence-0.133(1.030)Observations675675924924924924924924# of Countries4949646464646464Ave Years/Country13.77613.77614.43814.43814.43814.43814.43814.438Hansen p-value0.9930.9940.860.9070.8990.8920.8890.867AR(2) p-value0.5290.6150.4640.4410.4690.4720.4780.455# of Instruments (GMM)6161606161616061Table 1.A.4: Endogeneity Robustness ModelsStandard Errors in Parantheses. * p<0.1 ** p<0.05 *** 0.001;  DV = Fiscal Balance (% GDP). All models are System GMM models with twostep Windmeijer corrected robust standard errors. Year fixed effects estimated in all models, not  

 

 

 
 
 

 

 

 

 

 

51 

 

 

Note: Above, we see the marginal effect of fiscal rules as the financial flogging condition varies. As we 

see, the coefficient values do not seem to significantly change when the flogging condition =1 vs when it 

is 0. Further, the prediction should be that fiscal institutions have a more positive effect when the 

financial flogging condition==1, but in some of these the value declines. The standard errors also 

significantly overlap. Also, while in the Threshold and speculative grade plots it does not appear as 

though fiscal institutions are effective, an examination of the two-way interaction plots for these models 

mirror those reported in Figure 1.1: We find that MTF significantly increases fiscal balances when the 

Rule Basis equals 3 or 4, and Rule Basis significantly improves balances but only when MTF is equal to 1. 

 

52 

 

 

 

 

 

Here, we can see the marginal effects plots from the table 1.A.5 below, which corresponds to 

sub-table region A. In this case, we see the triple interactions when countries with a polity score 

less than 0 are excluded, in order to see if key clusters of countries are driving the results. We 

see instead that we still have strong support for the hypotheses. In the above plots, the 

standard errors are larger due to the reduced sample and some of the coefficients for the 

interaction terms are not significant, but the marginal effects plots reveal the same significant 

relationships as explored in the main text of the paper. 

53 

 

 

 

 

 

 

 

 

54 

Sample Restriction:(1)(2)(3)(4)(5)(6)(7)(8)Rule Basis * MTF * Checks0.03540.125(0.0658)(0.128)Rule Basis * MTF * Polity0.157*0.179***(0.0846)(0.0444)Rule Basis * MTF * Civil Society Org's1.2405.405***(1.132)(1.130)Rule Basis * MTF * Opp Vote Share0.0135**0.0225*(0.00587)(0.0116)Observations826826826826557557538557# of Countries6060606040403940Ave # years/country13.7713.7713.7713.7713.9313.9313.7913.93Sample Restriction:(9)(10)(11)(12)(13)(14)(15)(16)Rule Basis * MTF * Checks0.1440.314*(0.158)(0.162)Rule Basis * MTF * Polity0.223***0.127*(0.0511)(0.0640)Rule Basis * MTF * Civil Society Org's5.392***4.657***(1.452)(1.285)Rule Basis * MTF * Opp Vote Share0.0306**0.0138(0.0117)(0.0132)Observations451451451451392392373392# of Countries3636363637373637Ave # years/country12.5312.5312.5312.5310.5910.5910.3610.59Dependent Variable is Fiscal Balance (% GDP). All models include full set of controls from Table 1 but omitted for space and ease of presentation. Section headers denote the sample manipulation. All models are country fixed effects models with clustered standard errors.Table 1.A.5: Robustness to Splitting Sample(A)  Excluding Polity < 0(B)   Excluding Europe(C )   Developing Countries Only(D)   Excluding Polity level 9 or 10 

 

55 

Country OmittedRule Basis * MTFp-valueCountry OmittedRule Basis * MTFp-valueCountry OmittedRule Basis * MTFp-valueUS0.363***(0.003)Hungary0.381***(0.002)Burkina Faso0.381***(0.003)Canada0.375***(0.002)Solvak Republic0.366***(0.004)Liberia0.344***(0.004)Jamaica0.378***(0.002)Italy0.388***(0.002)Togo0.373***(0.002)Mexico0.386***(0.002)Croatia0.385***(0.002)Cameroon0.414***(0.000)Costa Rica0.397***(0.001)Slovenia0.374***(0.003)Nigeria0.368***(0.004)Panama0.336***(0.010)Greece0.375***(0.003)Gabon0.335***(0.004)Columbia0.399***(0.001)Cyprus0.392***(0.002)Congo, Republic0.374***(0.003)Ecuador0.355***(0.003)Bulgaria0.366***(0.003)Kenya0.373***(0.002)Peru0.370***(0.002)Romania0.376***(0.002)Namibia0.384***(0.002)Brazil0.386***(0.002)Russia0.368***(0.002)Botswana0.356***(0.004)Chile0.377***(0.002)Latvia0.381***(0.002)Israel0.402***(0.002)Argentina0.417***(0.001)Lithuania0.381***(0.002)Mongolia0.339***(0.005)UK0.373***(0.002)Armenia0.375***(0.002)Japan0.338***(0.006)Ireland0.366***(0.008)Finland0.386***(0.002)India0.361***(0.003)Netherlands0.375***(0.002)Sweden0.363***(0.003)Pakistan0.377***(0.002)Belgium0.379***(0.002)Norway0.373***(0.002)Sri Lanka0.390***(0.001)France0.399***(0.001)Denmark0.372***(0.002)Malaysia0.369***(0.005)Switzerland0.390***(0.002)Giunea-Bissau0.389***(0.001)Singapore0.369***(0.004)Spain0.407***(0.001)Mali0.381***(0.002)Indonesia0.361***(0.003)Portugal0.367***(0.003)Senegal0.357***(0.002)Australia0.369***(0.003)Poland0.372***(0.004)Niger0.365***(0.004)New Zealand0.364***(0.003)Austria0.404***(0.001)Ivory Coast0.362***(0.003)Results from panel jackknife regressions, exluding one country per regression. Estimated from model (1), estimates are system GMM. While not reported, GMM diagnostic statistics in all models suggested instruments used are valid (all hansen and AR(2) statistics were insignificant).Table 1.A.6: Jackknife Panel Regressions 

 

 

 

56 

(1)Base + controlsRule Basis-0.139(0.087)MTF-0.063(0.262)Rule Basis * MTF0.298**(0.109)(lag) Fiscal Balance0.753***(0.060)Polity-0.066**(0.027)Checks and Balances-0.004(0.036)Executive Election-0.408**(0.148)(log) GDP-pc0.186(0.134)CBI Garriga (weighted)0.011(0.453)(log) Population-0.156(0.105)(log) IMF Credits0.022(0.020)(lag) Unemployment (%LF)0.011(0.026)Capital Account Openness0.406(0.375)Current Account Balance0.056**(0.020)Trade (% GDP)-0.002(0.003)(lag) Growth0.108***(0.025)(lag) (log) Inflation0.177**(0.083)(lag) (log) Debt/GDP0.221(0.170)Fixed Exch. Rate-0.035(0.240)Internal Conflict-0.174***(0.038)Resource Rich Dummy0.674**(0.270)Constant0.324(2.637)Observations781hansenp0.781ar2p0.390# of Countries63.000Years/country12.397Model estimation the same as reported in Table 1Table 1.A.7: Expanded Control Set for Main GMM Models 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

57 

 

 

 

 

 

58 

 

 

BIBLIOGRAPHY 

59 

 

BIBLIOGRAPHY 

 

 

Afonso, Antonio, and Ana Sofia Guimarães. 2013. “The Relevance of Fiscal Rules for Fiscal and 

Yield Developments.” Rochester, NY: Department of Economics, ISEG-UL, Working Paper 
no 05/2014/DE/UECE. 

Alesina, Alberto, and Tamim Bayoumi. 1996. “The Costs and Benefits of Fiscal Rules: Evidence 

from U.S. States.” Working Paper 5614. National Bureau of Economic Research. 
http://www.nber.org/papers/w5614. 

Alesina, Alberto, and Allan Drazen. 1991. “Why Are Stabilizations Delayed?” The American 

Economic Review 81 (5): 1170–88. 

Alt, James E., and David Dreyer Lassen. 2006. “Transparency, Political Polarization, and Political 
Budget Cycles in OECD Countries.” American Journal of Political Science 50 (3): 530–50. 
doi:10.1111/j.1540-5907.2006.00200.x. 

Alt, James E., David Dreyer Lassen, and Joachim Wehner. 2012. “Moral Hazard in an Economic 

Union: Politics, Economics, and Fiscal Gimmicks in Europe.” SSRN Scholarly Paper ID 
2106611. Rochester, NY: Social Science Research Network.  

Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Specification for Panel Data: Monte 

Carlo Evidence and an Application to Employment Equations.” Review of Economic 
Studies 58 (2): 277–97. 

Bayoumi, Tamim, Morris Goldstein, and Geoffrey Woglom. 1995. “Do Credit Markets Discipline 
Sovereign Borrowers? Evidence from U.S. States.” Journal of Money, Credit and Banking 
27 (4): 1046–59. doi:10.2307/2077788. 

Ahlquist, John S. 2006. “Economic Policy, Institutions, and Capital Flows: Portfolio and Direct 

Investment Flows in Developing Countries.” International Studies Quarterly 50 (3): 681–
704. 

Beaulieu, Emily, Gary W. Cox, and Sebastian Saiegh. 2012a. “Sovereign Debt and Regime Type: 

Reconsidering the Democratic Advantage.” International Organization 66 (04): 709–738. 
doi:10.1017/S0020818312000288. 

———. 2012b. “Sovereign Debt and Regime Type: Reconsidering the Democratic Advantage.” 

International Organization 66 (04): 709–738. doi:10.1017/S0020818312000288. 

 

60 

 

Bodea, Cristina and Hicks, Raymond. 2017. “Sovereign Credit Ratings and Central Banks: Do 

Analysts Pay Attention to Institutions? Available at 
SSRN: https://ssrn.com/abstract=2916334 

Bodea, Cristina, and Masaaki Higashijima. 2015. “Central Bank Independence and Fiscal Policy: 

Can the Central Bank Restrain Deficit Spending?” British Journal of Political Science 
FirstView (July): 1–24. doi:10.1017/S0007123415000058. 

Bohn, Henning, and Robert P. Inman. 1996. “Balanced Budget Rules and Public Deficits: 

Evidence from the U.S. States.” Working Paper 5533. National Bureau of Economic 
Research. http://www.nber.org/papers/w5533. 

Brender, Adi, and Allan Drazen. 2005. “Political Budget Cycles in New versus Established 

Democracies.” Journal of Monetary Economics 52 (7): 1271–95. 

Bernhard, William, and David Leblang. “Cabinet Collapses and Currency Crashes.” Political    

Research Quarterly 61, no. 3 (September 1, 2008): 517–31. 
doi:10.1177/1065912908320665. 

Brambor, Thomas, William Roberts Clark, and Matt Golder. 2006. “Understanding Interaction 

Models: Improving Empirical Analyses.” Political Analysis 14 (1): 63–82. 
doi:10.1093/pan/mpi014. 

Christopher, Rachel, Suk-Joong Kim, and Eliza Wu. 2012. “Do Sovereign Credit Ratings Influence 
Regional Stock and Bond Market Interdependencies in Emerging Countries?” Journal of 
International Financial Markets, Institutions and Money 22 (4): 1070–89. 
doi:10.1016/j.intfin.2012.01.003. 

Clark, William Roberts, and Mark Hallerberg. 2000. “Mobile Capital, Domestic Institutions, and 
Electorally Induced Monetary and Fiscal Policy.” The American Political Science Review 
94 (2): 323–46. doi:10.2307/2586015. 

Debrun, Xavier, Laurent Moulin, Alessandro Turrini, Joaquim Ayuso-i-Casals, Manmohan S. 

Kumar, Allan Drazen, and Clemens Fuest. “Tied to the Mast? National Fiscal Rules in the 
European Union [with Discussion].” Economic Policy 23, no. 54 (2008): 297–362. 

DeMeritt, Jacqueline H. R. 2012. “International Organizations and Government Killing: Does 

Naming and Shaming Save Lives?” International Interactions 38 (5): 597–621. 
doi:10.1080/03050629.2012.726180. 

Drazen, Allan. 2001. “The Political Business Cycle after 25 Years.” In NBER Macroeconomics 

Annual 2000, Volume 15, 75–138. MIT PRess. 
http://www.nber.org/chapters/c11055.pdf. 

 

61 

 

Franzese, Robert J. 2002. Macroeconomic Policies of Developed Democracies. Cambridge 

University Press.  

Fratzscher, Marcel, Arnaud Mehl, and Isabel Vansteenkiste. 2011. “130 Years of Fiscal 

Vulnerabilities and Currency Crashes in Advanced Economies.” IMF Economic Review 59 
(4): 683–716. doi:10.1057/imfer.2011.23. 

Gaillard, Norbert. 2011. A Century of Sovereign Ratings. 2012 edition. London: Springer. 

Gumus, Inci. 2015. “Fiscal Uncertainty and Currency Crises.” Review of Development Economics 

19 (4): 957–70. doi:10.1111/rode.12183. 

Hallerberg, Mark, and Jurgen von Hagen. 1997. “Electoral Institutions, Cabinet Negotiations, 
and Budget Deficits in the European Union.” Working Paper 6341. National Bureau of 
Economic Research. http://www.nber.org/papers/w6341. 

Hallerberg, Mark, Rolf Strauch, and Jürgen von Hagen. 2009. Fiscal Governance in Europe. 

Cambridge University Press. 

Hallerberg, Mark, and Guntram B. Wolff. 2008. “Fiscal Institutions, Fiscal Policy and Sovereign 

Risk Premia in EMU.” Public Choice 136 (3–4): 379–96. doi:10.1007/s11127-008-9301-2. 

Hallerberg, Mark, and Sami Yläoutinen. 2010. “Political Power, Fiscal Institutions and Budgetary 

Outcomes in Central and Eastern Europe.” Journal of Public Policy. April.  

Hamdaoui, Mekki, Abir Zouari, and Samir Maktouf. 2016. “The Effect of Financial Liberalization 
on Banking Sector Stability.” International Review of Applied Economics 30 (5): 644–67.  

Hameed, Farhan. 2005. “Fiscal Transparency and Economic Outcomes.” IMF Working Paper 

05/225. 

Iara, Anna, and Guntram Wolff. 2010. “Rules and Risk in the Euro Area.” Economic Papers: 

European Commission. 

Kabadayi, Burhan. 2013. “The Effects of Sovereign Ratings Changes on Turkey’s Stock Market.” 

Journal of Applied Finance and Banking 3 (6): 87–96. 

Kaminsky, Graciela, and Sergio L. Schmukler. 2002a. “Emerging Market Instability: Do Sovereign 

Ratings Affect Country Risk and Stock Returns?” The World Bank Economic Review 16 
(2): 171–95. 

———. 2002b. “Emerging Market Instability: Do Sovereign Ratings Affect Country Risk and 

Stock Returns?” The World Bank Economic Review 16 (2): 171–95. 

 

62 

 

Keefer, Philip, and David Stasavage. 2003. “The Limits of Delegation: Veto Players, Central Bank 

Independence, and the Credibility of Monetary Policy.” American Political Science 
Review 97 (3): 407–23. 

Keefer, Philip, and Razvan Vlaicu. 2008. “Democracy, Credibility, and Clientelism.” Journal of 

Law, Economics, and Organization 24 (2): 371–406. doi:10.1093/jleo/ewm054. 

Kelemen, R. Daniel, and Terence K. Teo. 2014. “Law, Focal Points, and Fiscal Discipline in the 
United States and the European Union.” American Political Science Review 108 (02): 
355–70. doi:10.1017/S0003055414000100. 

Klomp, Jeroen and de Haan, Jakob, (2009), “Central bank independence and financial 

instability” Journal of Financial Stability, 5 (4): 321-338. 

Kontopoulos, Yianos, and Roberto Perotti. 1999. “Government Fragmentation and Fiscal Policy 

Outcomes: Evidence from OECD Countries.” In Fiscal Institutions and Fiscal Performance, 
81–102. University of Chicago Press. http://www.nber.org/chapters/c8024.pdf. 

Kumar, Manmohan S., and Emanuele Baldacci. 2010. “Fiscal Deficits, Public Debt, and Sovereign 

Bond Yields.” IMF Working Paper 10/184. International Monetary Fund. 
https://ideas.repec.org/p/imf/imfwpa/10-184.html. 

Linzer, Drew A., and Jeffrey K. Staton. 2011. “A Measurement Model for Synthesizing Multiple 

Comparative Indicators: The Case of Judicial Independence.” In Presentation at the 2011 
Annual Meeting of the American Political Science Association, September, 1–4.  

Lowry, Robert C., and James Alt. 2001. “A Visible Hand? Bond Markets, Political Parties, 

Balanced Budget Laws, and State Government Debt.” Economics & Politics 13 (1): 49–72. 
doi:10.1111/1468-0343.00083. 

Montes, Gabriel Caldas, Diego S. P. de Oliveira, and Helder Ferreira de Mendonça. 2016. 
“Sovereign Credit Ratings in Developing Economies: New Empirical Assessment.” 
International Journal of Finance & Economics 21 (4): 382–97. doi:10.1002/ijfe.1551. 

Mosley, Layna. 2003. Global Capital and National Governments. New York: Cambridge 

University Press. 

Nyblade, Benjamin, and Angela O’Mahony. 2014. “Playing with Fire: Pre-Electoral Fiscal 

Manipulation and the Risk of a Speculative Attack.” International Studies Quarterly 58 
(4): 828–38. doi:10.1111/isqu.12168. 

O’Mahony, Angela. 2011. “Engineering Good Times: Fiscal Manipulation in a Global Economy.” 

British Journal of Political Science 41 (2): 315–40. 

 

63 

 

Persson, Torsten, and Guido Tabellini. 2005. The Economic Effects of Constitutions. Cambridge, 

Mass.: The MIT Press. 

Poterba, James M., and Kim S. Rueben. 1999. Fiscal Rules and State Borrowing Costs: Evidence 

from California and Other States. San Francisco, Calif: Public Policy Institute of California. 

Qin, Xiao, and Chengying Luo. 2014. “Capital Account Openness and Early Warning System for 

Banking Crises in G20 Countries.” Economic Modelling 39 (April): 190–94.. 

Reinhart, Carmen M., and Kenneth S. Rogoff. 2002. “The Modern History of Exchange Rate 

Arrangements: A Reinterpretation.” Working Paper 8963. National Bureau of Economic 
Research. http://www.nber.org/papers/w8963. 

Roodman, David. 2009a. "How to do xtabond2: An introduction to difference and system GMM 

in Stata," Stata Journal, StataCorp LP, vol. 9(1), pages 86-136, March. 

Roodman, David. 2009b. “A Note on the Theme of Too Many Instruments*.” Oxford Bulletin of 

Economics and Statistics 71 (1): 135–58. doi:10.1111/j.1468-0084.2008.00542.x. 

Roubini, Nouriel, and Jeffrey D. Sachs. 1989. “Government Spending and Budget Deficits in the 

Industrial Economies.” SSRN Scholarly Paper ID 459425. Rochester, NY: Social Science 
Research Network. http://papers.ssrn.com/abstract=459425. 

Shi, Min, and Jakob Svensson. 2006. “Political Budget Cycles: Do They Differ across Countries 

and Why?” Journal of Public Economics 90 (8): 1367–89. 

Soudis, Dimitrios. 2014. “Credit Rating Agencies and the IPE: Not as Influential as Thought?” 

Review of International Political Economy 0 (0): 1–25. 
doi:10.1080/09692290.2014.957234. 

Tsebelis, George. 2002. Veto Players: How Political Institutions Work. Princeton University 

Press.  

Vaaler, Paul Martin, and Marek Hanusch. 2015. “Credit Ratings and Fiscal Responsibility.” 

97556. The World Bank. 
http://documents.worldbank.org/curated/en/2015/06/24677886/credit-rating-fiscal-
responsibility. 

Vlaicu, Razvan, Marijn Verhoeven, Francesco Grigoli, and Zachary Mills. 2014. “Multiyear 

Budgets and Fiscal Performance: Panel Data Evidence.” Journal of Public Economics 111 
(March): 79–95. doi:10.1016/j.jpubeco.2013.12.011. 

 

 

64 

 

CHAPTER 2:  The Democratic (Dis)Advantage: How Financial Vulnerability Conditions the Effect 

of Democracy on Sovereign Credit Ratings in Developing Countries 

 

The question regarding whether democratic institutions can serve as effective credible 

commitment devices has received considerable attention and has been examined in diverse 

ways. In one context, North and Weingast (1989) and  Schultz and Weingast (2003) have argued 

that because democracies have greater accountability and the public can punish its leaders, 

democracies can credibly commit to repay sovereign (national) debt. As a consequence, this 

reasoning goes, democracies should be more creditworthy and have an advantage in financial 

markets. First, they should enjoy better sovereign credit ratings such as the highest “AAA” 

rating, which indicate a countries’ ability and willingness to repay its debt commitments. And 

secondly, they should also have greater ability to issue more of and pay less interest for their 

debt. While there was some early debate over this argument (Archer, Biglaiser, and DeRouen 

2007; Saiegh 2005), DiGiuseppe and Shea (2015) note that most scholars seem to have 

converged toward agreement that democracies do enjoy this “democratic advantage” of 

sovereign credit ratings. 

Contrasting the emphasis of the literature, I introduce a novel argument that democracies do 

not have an inherent credit market advantage. Instead, the advantage of democracies is 

conditional on the countries’ level of financial vulnerability, such that when democracies are 

highly vulnerable they have worse credit ratings, but enjoy ratings advantages when their 

economic vulnerability is low. Because democratic institutions enable and facilitate the ability 

of the public to mobilize against the use of unpopular economic adjustment measures, the 

adjustment process can be more challenging for such governments. As a result, financial 

 

65 

 

vulnerability interacts with democratic institutions in specific ways which lead international 

creditors to be more suspicious of their ability to adapt to challenging macroeconomic 

circumstances. Specifically, I argue that two important phenomena jointly determine a 

countries’ level of financial vulnerability: a countries’ overall debt burden (which increases 

financial risk) and its level of foreign exchange reserves (which, as a form of international 

“financial insurance”, decreases risk). These conditions jointly determine financial vulnerability 

because high debts can be a signal of weakening resolve to repay debts, signifies that in the 

event of an unrelated shock high debts can magnify crises, and because large debts leave little 

space for stabilizing counter-cyclical policy during such events. However, a large pool of foreign 

exchange reserves enables domestic financing of foreign loans, currency convertibility, and 

even provides an additional mechanism to provide counter-cyclical policy. In these ways, both 

of these issues combine such that high debt and low foreign reserves represent a perfect storm 

of economic vulnerability for a country. Risk is higher when debt is high and, in the event of a 

crisis, a low level of foreign reserves means a country has fewer policy options to manage the 

crisis. 

Thus, I advance several related arguments. First, democracies should have worse credit ratings 

at high levels of debt, holding foreign reserves constant. Debt levels are the clearest indicator 

that a country is in need of fiscal rebalancing, and so democracies at high debt should have 

poorer credit ratings. Second, while it is an accepted finding that higher debt predicts lower 

ratings (Gaillard 2011), this effect should be more strongly negative for democracies than 

autocracies. This is because large debt burdens are viewed as less risky for future debt 

repayment when financial actors believe that the debt burden can and will be easily resolved. 

 

66 

 

On the other hand, if financial actors believe that substantial debts are unlikely to be reversed, 

then this should be reflected in the form of lower credit ratings. As a result, because I argue 

that democratic institutions make economic adjustment more difficult, financial actors will view 

large debts for democracies more negatively. And lastly, democracies should have worse credit 

ratings at their highest vulnerability, which is when the ratio of foreign reserves relative to 

external debt is low. At the same time, precisely because foreign reserve assets can attenuate 

the risk of external debt, democracies should enjoy higher ratings when their reserves/external 

debt ratio is sufficiently high (that is, when they are financially secure, instead of vulnerable). As 

a consequence of this analysis, we gain important insights into the precise mechanisms which 

enable democracies to enjoy a credit market advantage. 

On the first dimension, when countries have a high level of debt, this increases the risk of debt 

default and can also increase the risk of other financial crises (Cantor and Packer 1996; 

Steinberg, Koesel, and Thompson 2015). As a result, higher levels of debt can increase the 

pressure on governments to undertake financial adjustment policies to mitigate the risk posed 

by such debts. However, while proponents of the democratic advantage hypothesis emphasize 

that electoral punishment (accountability mechanisms) and checks and balances (veto players) 

enable credible commitments by democracies to repay, these same features of democracy 

should also make policy adjustment more difficult when there is a pressing need to do so. 

Because democracies are more accountable, incumbents should be more hesitant to embrace 

painful adjustment policies such as tax increases and spending and benefit cuts. It is also 

because democracies have more diffuse policymaking that such adjustment can be more 

difficult in democracies (Alesina and Drazen 1991).  

 

67 

 

Autocracies, by contrast, should have an easier time passing economic adjustment measures 

because they do not need the approval of other government members, enabling them to shift 

the adjustment burden onto those whom the regime does not rely on for support. And very 

importantly, democratic institutions are also what arguably compels fiscal mismanagement in 

the first place, in the form of fiscal budget cycles (Shi & Svensson 2006; O’Mahoney 2011) or 

excessive debt accumulation due distributive politics or electoral design (Hallerberg and Hagen 

1997; Persson and Tabellini 2005). Autocracies, then, have lower adjustment costs in order to 

ensure their ability to repay debt obligations, and may not face the same interest group 

pressures for deficit spending.  

A recent Bloomberg article27 titled “The Bond Market Prefers Dictators to Democracies,” helps 

to illuminate this argument with a comparison of recent debt payment cases. While specifically 

showing that financial returns from autocratic government debt has been more favorable in 

recent years, the article suggests: 

 

On April 12, Venezuela’s creditors reaped large returns when President Nicolas Maduro 
made good on $2.5 billion in debt payments even as he struggles to come up with 
enough money for food imports. Two days earlier, El Salvador, a democracy for the last 
quarter-century, defaulted as a feud between the president and an opposition party -- a 
"high-stakes game of chicken," as Nomura Holdings Inc. strategist Benito Berber called it 
-- left the government unable to make a $29 million payment to a local pension fund.” 

 

Venezuela has in fact defaulted since this example was made; however, it nonetheless 

demonstrates the extremes to which autocrats may go in order to maintain debt repayments, 

while contrasting a potentially lower threshold for democracies. Similar issues to are found by 

                                                           

27 https://www.bloomberg.com/news/articles/2017-04-20/autocracies-beating-democracies-in-emerging-
market-bond-world 

 

68 

 

examining public statements given by the ratings agencies, such as a Moody’s Analytics 

explanation for a negative rating outlook for Honduras in 201328, saying it “reflects Moody's 

concerns about (1) the government's fiscal deterioration in 2012 and the low likelihood of 

significant fiscal consolidation in 2013, given that it is an election year…” (emphasis added). In 

March of 2017, Fitch Ratings issued a statement concerning the increasing debt burden of 

Ecuador, noting that a potential electoral winner would face legislative challenges in passing 

pro-IMF reforms, given the strength of the opposition party29. And in 2016, Fitch discussed the 

negative rating outlook for Costa Rica30, stating that despite a strong macroeconomic position, 

“The Negative Outlook reflects adverse public debt dynamics, driven by large fiscal deficits, and 

legislative gridlock preventing progress on reforms to correct fiscal imbalances in a timely 

manner.” (emphasis added). We therefore see several direct examples for which these key 

features of democracy – elections, partisan obstacles and legislative gridlock - which may 

actually be a source of disadvantage. As a consequence, two expectations emerge from this 

consideration of economic adjustment costs. First, that democracies will have worse credit 

ratings than autocracies when debt levels are high. Secondly, high debt levels should reduce 

credit ratings moreso in democracies than in autocracies. 

Yet as noted, an important means by which countries seek to insulate themselves from financial 

risk is by accumulating a larger pool of foreign exchange reserves (Obstfeld, Shambaugh, and 

Taylor 2010; Nyblade and O’Mahony 2014). A 2004 Moody’s Analytics statement31 on the 

                                                           

28 https://www.moodys.com/research/Moodys-changes-Honduras-rating-outlook-to-negative--PR_267052 
29 https://www.fitchratings.com/site/pr/1021287 
30 http://en.centralamericadata.com/en/article/home/Costa_Rica_Fitch_Maintains_Sovereign_Rating 
31 https://www.moodys.com/research/MOODYS-REPORTS-THAILANDS-Baa1-RATINGS-AND-STABLE-OUTLOOK-
REFLECT-IMPROVED--PR_81226 

 

69 

 

positive ratings outlook for Thailand exemplifies this argument, citing “a high level of 

international reserves” among key reasons that “provide considerable leeway for the 

authorities to manage external shocks.” Foreign exchange reserves can be used in emergencies 

or abrupt financial shocks, such that domestic institutions are able to draw upon the reserves 

for emergency financing. As such, while rising debt increases both the risk and costliness of 

crises, a higher volume of foreign currency reserves provides a form of financial insurance 

which mitigates the very weakness posed by democratic institutions. In such a case, this leaves 

only the positive effects of democracies to have an influence on credit ratings. Such positive 

effects are, for a few examples: economic transparency (Hollyer, Rosendorff, and Vreeland 

2014), reducing corruption (Kolstad and Wiig 2016), stronger rule of law (Biglaiser and Staats 

2012), and a greater ratio of public vs private goods provisions (Bueno de Mesquita and Downs 

2005; DiGiuseppe and Shea 2015).  Therefore, in keeping with the theory that democracies will 

have a disadvantage when they are most financially vulnerable (considering both debt and 

reserves), I argue that democracies should have lower credit ratings when their relative level of 

foreign reserves-to-debt is low, but that at a sufficiently high level of reserves-to-debt 

democracies should regain the advantage. 

I utilize credit ratings data from Moody’s Analytics and Standard & Poor’s on 54 developing 

countries from 1977 to 2012 to investigate these financial vulnerability hypotheses. Even after 

controlling for the selection bias of which countries receive credit ratings (Beaulieu, Cox, and 

Saiegh 2012), and possible simultaneity bias, I show robust support for these predictions. When 

the Reserves/External Debt ratio is low – indicating a countries’ highest level of financial 

vulnerability – democracies have significantly poorer credit ratings. However, at higher levels of 

 

70 

 

Reserves/ External Debt, democracies have a ratings advantage. Importantly, the analysis also 

reveals that high debt levels significantly reduce ratings, but only for more democratic 

countries, which confirms a key contention of the paper that democracies are more likely to 

struggle with debt consolidation because of the increased political adjustment costs they face. I 

also show that this result holds using several indicators of democracy. 

This paper thus makes an important theoretical contribution to the literature by helping to 

clarify exactly what it is about democracies that leads to better credit ratings. The democratic 

advantage literature has argued that democracies are able to credibly commit to debt 

repayment, and hence they enjoy ratings advantages. This paper calls this basic line of 

reasoning into question. Democracies do have some advantages in finance markets, but it does 

not seem to be because of a ceteris paribus credible commitment to repay debt holdings. If 

democracies were, by default, more credible debt issuers than autocracies, then it should not 

matter whether the countries’ level of debt is high or low, or whether a country has large 

foreign exchange reserves. Instead, this analysis reveals an inherent institutional weakness of 

democracies, but implicitly highlights other pathways by which democracies enjoy credit rating 

advantages. 

The Democratic Advantage 

The literature which examines the “democratic advantage” of country credit ratings is a smaller 

subset of a broader literature which examines the ability of democracies to deliver more 

optimal economic policies. Here, I overview the specific literature of the democratic advantage, 

and will draw upon the literature on economic crises to illustrate that the theoretical structure 

predicting the democratic advantage is much more logically ambiguous than is generally given 

 

71 

 

credit. In the next section, I will argue that considering the democratic advantage through the 

lens of financial vulnerability gives us important insights into when democracies are at a 

disadvantage, and in what cases they may regain preferential credit ratings. 

The argument surrounding democratic institutions and preferential financial market access has 

its origins from North and Weingast (1989), who argued that the English Crown was only able to 

commit to stable fiscal policies when the parliament was granted constitutional authority to 

oversee royal spending decisions (Dincecco 2009). From there, Schultz and Weingast (2003) 

expanded on the idea by arguing that, because democratic leaders are accountable to their 

public (who can punish the sovereign lender via elections), they are able to credibly commit to 

debt repayment. As a consequence, investors would be more likely to lend to democracies 

because of this de facto contract enforceability, which therefore increases their financial 

market access. As a result, they argued, democracies are better able to endure and prevail in 

long running wars. 

Most scholars of the democratic advantage have agreed with this essential logic, that 

democracies are able to credibly commit to debt repayment because of electoral accountability 

and because of the diffusion of decision-making authority (veto player arguments). These 

essential features of democracy are also argued to provide a range of other beneficial and pro-

growth outcomes, such as a higher provision of public vs private goods (Bueno de Mesquita et 

al. 2003; DiGiuseppe and Shea 2015), reduction of corruption (Schultz and Weingast 2003; 

Kolstad and Wiig 2016), and greater rule of law and property rights protections (Biglaiser and 

Staats 2012).  

 

72 

 

While recent scholarship on the democratic advantage has been very supportive, some of the 

earlier literature demonstrated some contrary findings. For instance, Saiegh (2005) examined 

the democratic advantage in developing countries, arguing that the essential logic of the 

democratic advantage holds only for developed countries. Specifically, he argued that because 

developing countries have higher levels of foreign ownership over government debt, the 

electoral punishment mechanism does not apply. He showed that, contrary to expectations 

regarding the democratic advantage hypothesis, democracies are more likely to default on their 

debt obligations, and do not differ in the interest paid on debt. This reasoning has been 

disputed, however, as it has been argued that defaulting even on foreign debt can have ripple 

effects on the domestic economy (DiGiuseppe and Shea 2015). Archer, Biglaiser, and DeRouen 

(2007) also examined developing countries, showing that sovereign credit ratings did not differ 

notably between regime types, and suggested from interviews with investors that regime type 

is not an important consideration. 

More recently, Biglaiser and Staats (2012) have argued that the earlier literature ignored the 

important influence of judicial independence and the rule of law as an indicator of political 

stability. Again examining developing countries, they show that measures of judicial strength 

and the rule of law significantly predict higher credit ratings, augmenting the analysis with 

compelling direct quotations from the ratings agencies. Beaulieu, Cox, and Saiegh (2012) 

arguably provide the most decisive analysis on the subject, attempting to reconcile the 

previously inconsistent findings. They argue that credit access is an important aspect of the 

democratic advantage, but that no one had yet examined that implication sufficiently. They find 

 

73 

 

that democracies were more likely to secure credit ratings, and that once controlling for this 

selection effect democracies again appear to have significantly higher ratings.  

These recent papers are likely what inspired DiGiuseppe and Shea (2015) to suggest that, 

despite a few democratic outliers such as Greece and Weimar Germany, “scholars typically 

agree that democracies are more creditworthy than nondemocracies.” Even while they argue 

that one of the supposed mechanisms of the democratic advantage is false – that of electoral 

punishment – they yet argue that the democratic advantage exists because democracies are 

more likely to adopt pro-growth policies which maximize the use of public rather than private 

goods. 

The logic for the democratic advantage thus has rested upon the notion that democracies are 

better able to credibly commit to debt repayment. However, I argue that this expectation is 

ambiguous in light of other consequences of democratic politics, and in some cases should lead 

us to expect that democracies are actually worse off. In the next section, I overview relevant 

segments from the literature on economic crises and debt accumulation, and show how these 

literatures lead us to opposite expectations. Afterword, I will elaborate how democracies may 

have different effects conditioned by financial vulnerability. 

The Democratic Disadvantage 

While all of these above reasons regarding why democracies may have a credit advantage are 

sound and plausible reasons, I suggest that the broad theoretical expectations are not as clear 

as it would seem. The reason is that, while it may be true that democracies may be more likely 

to adopt growth maximizing policies, others argue that democracies may be unable to adopt 

optimal policies (via economic adjustment) because of intensive interest group pressures. It is 

 

74 

 

this reasoning why Bearce and Hallerberg (2011)  argue that democracies have less stable 

exchange rates, and why Steinberg, Koesel, and Thompson (2015) argue that democracies are 

more likely to experience currency crashes. When the need for economic adjustment presents 

itself, democracies have trouble facilitating this adjustment to maintain the exchange-rate. 

Along these same lines, Oatley (2004) argues that democracies are unable to promptly adjust to 

episodes of high-inflation, because they “cannot overcome societal opposition and distributive 

conflict”.   

Societal pressure and electoral incentives are also what drive democracies to use electoral 

spending cycles as a reelection effort (Shi and Svensson 2006; Drazen 2001; Nyblade and 

O’Mahony 2014). Block and Vaaler (2004), for instance, find that elections in developing 

countries increase the cost of debt finance as well as the probability of a credit rating 

downgrade. Yet not only do elections generate negative fiscal pressures, as the literature 

generally argues, but Dreher (2003) specifically shows that democracies are also more likely to 

be non-compliant with IMF programs prior to elections. Such a finding was similar to earlier 

research on middle-income countries by Haggard and Kaufman (1992), who showed that 

“virtually all high inflation was brought down only under the auspices of authoritarian regimes” 

and that such regimes “are more likely to stabilize when inflation and social conflict are high” 

(Haggard and Webb 1993). These findings thus also illustrate the unique complications of 

economic adjustment which democracies may face. So while one may suggest that electoral 

punishment incentives in democracies create a credible commitment to repay debt, it is not 

clear that this very same mechanism does not also make crisis (or default) more likely.  

 

75 

 

Further, Leblang and Satyanath (2006) also argue that democracies and divided government are 

significant predictors of currency crashes, although by slightly different logic. While 

democracies may provide greater economic transparency (Hollyer, Rosendorff, and Vreeland 

2014), Leblang and Satyanath argue that divided government in democracies subverts the good 

accomplished by transparency. Their reasoning is that, because there are multiple veto players, 

currency speculators are unable to settle on a focal point and therefore are unable to coalesce 

around an expected ex-post currency rate. As a consequence of this unpredictability, 

speculative runs become more likely.    

The literature on debt accumulation also provides some key insights in this regard. Here again, 

the second posed mechanism for the credibility of democratic commitments – multiple veto 

players – is likely also a source of institutional weakness. The reason is that, if the country is in 

need of a debt consolidation, so as to avoid a debt default, such economic adjustment policies 

will be difficult to achieve when there are multiple veto players (Roubini and Sachs 1989). 

Alesina and Drazen (1991) formalize this expectation, that multiple government actors have 

incentives to delay adjustment in the hopes that other government actors will shoulder more of 

the adjustment costs. Franzese (2002) confirms this finding, showing that additional veto 

players cause significant enlargements to the debt burden as the level of debt increases. 

Additionally, Arpac, Bird, and Mandilaras (2008) also show that a higher number of veto-actors 

increases the probability of non-compliance with IMF programs for this very reason. This 

concern over the adjustment burden – as a function of divided government (veto players) - also 

resonates well with recent experience. For instance, it was the political brinksmanship caused 

by divided government in the United States in 2011 which led Standard and Poor’s to 

 

76 

 

downgrade its rating of US debt for the first time in its history32. S&P stated “the downgrade 

reflects our view that the effectiveness, stability, and predictability of American policymaking 

and political institutions have weakened.” They go on to emphasize “the difficulties in 

bridging the gulf between the political parties over fiscal policy” and “our perception of greater 

policymaking uncertainty”. This example thus clearly illustrates how, even in an advanced 

economy such as the United States, veto-players can make economic adjustment more difficult 

to achieve and that this may directly impact a countries sovereign credit rating.  

In this sense, even if democracies adopt pro-growth policies, as democratic advantage 

proponents have argued, it may still be the case that democracies spend at unsustainable rates, 

either deliberately (via electoral spending cycles) or as a consequence of unexpected economic 

shocks (via fiscal stabilizers during recessions). In any case, it is not clear from democratic 

advantage proponents why the potential advantages of democracies should ceteris paribus 

dominate these contrasting influences that democratic institutions have on policy outcomes. 

And, as I argue in the next section, it seems we have clear reason to expect democracies to be 

worse off given the need for economic adjustment. 

The Influence of Financial Vulnerability 

I propose that there may be a way to reconcile the discordant predictions of the democratic 

advantage. To restate: there are two democratic features which are presumed to enable 

democracies to credibly commit to debt repayment: electoral punishment and executive 

constraints via veto players. I argue that both of these institutions clearly suggest that at times 

of financial vulnerability, democracies will actually be worse off than their autocratic 

                                                           

32 http://blogs.wsj.com/marketbeat/2011/08/05/sp-downgrades-u-s-debt-rating-press-release/ 

 

77 

 

counterparts. First, if a country is financially vulnerable by some means, it is suggestive that the 

country is in need of economic adjustment policies to lower risk and forestall potential crises. 

Presumably, this is implied by suggesting that democracies have a credible commitment to 

repay. Such a claim is suggestive that the country would be more likely to undertake policies 

commensurate with the goal of debt repayment. However, given an obvious need for countries 

to adopt adjustment policies, democracies should clearly have a more difficult time 

accomplishing this task. 

Both of the “credible commitment” mechanisms are at play in this case. In the first instance, 

veto actors make policy adjustment more difficult (Franzese 2002; Oatley 2004). When the 

executive needs no other actor to sign off on policy changes, the process is achieved much 

more readily, and so autocracies have the advantage. Secondly, DiGiuseppe and Shea (2015) 

argue that the smaller size of the winning coalition for autocracies means that they have 

stronger incentives to maintain the stability of their coalition (by not defaulting). Similarly, I 

argue it is precisely because autocracies have smaller winning coalitions (WCs) that the winning 

coalition is better insulated from the risk of a painful economic adjustment program. Since 

autocrats have smaller WCs, it should be easier for them to shift the adjustment burden onto 

other segments of society without fear of meaningful political retribution. As a result, it should 

be easier for autocrats to both enact economic adjustment measures while also protecting their 

supporters. Sensing these inherent institutional weaknesses as risk accumulates, credit rating 

agencies and international bond investors should therefore give lower ratings and charge 

higher interest rates for financially vulnerable democracies. It also suggests that given an 

equally heavy debt burden, credit rating analysts should perceive the large debt burden for a 

 

78 

 

democracy to be more threatening, because they have lower expectation that it will be 

subsequently adjusted downward. 

But simply because democracies should have adjustment problems does not indicate that they 

will in all cases be disadvantaged, though. To be clear, all of the argued advantages of 

democracies in the form of better governance (or lower corruption), political and economic 

transparency, political stability and judicial strength are all sound reasons why investors and 

credit rating agencies would assess lower levels of repayment risk (North and Weingast 1989; 

Beaulieu, Cox, and Saiegh 2012; DiGiuseppe and Shea 2015). At low levels of financial 

vulnerability, the adjustment risks posed by democracies simply don’t exist. For democracies 

who do not stand in need of adjustment policies, the risk of “delayed stabilization” simply isn’t 

a concern for investors, and therefore in such cases all of the positives associated with 

democracies should dominate.  

I suggest that two fundamental measures of financial vulnerability successfully capture these 

concepts. First, a country’s level of accumulated debt is a well-established predictor of country 

risk of non-repayment or other financial crises (Cantor and Packer 1996; Leblang and Satyanath 

2006; Oatley 2004). Ahlquist (2006) and Mosley (2003) discuss this issue in the context of a 

signaling game, such that large accumulated debts are a signal to investors of a country’s 

lacking intent to repay their debt burden. Ahlquist (2006) elaborates on the connection 

between large debts and investor risk perception, arguing that it is an important risk indicator 

because it is connected with all of the main dimensions of investor risk. He states that portfolio 

(bond) investors “face 3 kinds of risk: inflation, currency, and default. Deficits [and thus debt] 

are potentially linked to all three”. All three of these phenomena are interconnected, since debt 

 

79 

 

(and thus deficits) can influence the local inflation rate, which causes currency volatility and 

depreciation. Further, given a depreciating currency, this would also exacerbate a debt burden 

if there are substantial holdings of external debt, because this drives up the cost of repayment 

on foreign-denominated debt issues. Thus, not only do high debt levels increase the risk of a 

crisis, but it also indicates that if a crisis does occur, the adjustment burden will be much larger. 

Corneli and Tarantino (2016) argue that “debt increases both the likelihood that the sovereign 

is hit by a liquidity shock and the probability that the due repayment to lenders falls short…”. 

Finally, when governments default on their debt and investors agree to a “debt restructuring”, 

the volume of total debt owed is also likely related to how much investors agree to reduce the 

debt principal.  

A countries debt level is therefore an established indicator of financial vulnerability which we 

would expect to mediate the effect of regime type on country credit ratings. This leads me to 

state the first two hypotheses of the paper: 

H1: Democracies will have worse (better) credit ratings than autocracies at higher 

(lower) levels of Debt/GDP. 

H2: Higher levels of Debt/GDP will have a larger negative effect on credit ratings for 

democracies compared to autocracies. 

On the other hand, when democracies are not financially vulnerable, I expect that democracies 

will enjoy better credit ratings because of the other benefits which flow from democratic 

institutions. We should therefore expect that, at low levels of Debt/GDP, democracies should 

enjoy better credit ratings, as is described in the two conditions embedded within H1. Beyond 

this, it has also been long established that a countries’ level of foreign exchange reserves is a 

 

80 

 

key means by which countries seek to mitigate the risks posed by large debt burdens and 

exchange rate risk. For instance, Corneli and Tarantino (2016) argue that “although raising debt 

increases the sovereign exposure to liquidity and productivity crises, the simultaneous 

accumulation of reserves can mitigate the negative effects of such crises.” It is this reason why 

Nyblade and O’Mahony (2014) argued that governments were only likely to boost deficits 

around elections when they had sufficiently high foreign reserves (thus mitigating the risk of 

economic blowback). Further,  Dominguez, Hashimoto, and Ito (2012) demonstrate that 

countries with large accumulated foreign reserves prior to the 2008 financial crisis also had 

higher post-crisis GDP growth rates. And more recently, Cheng (2015) has articulated that 

foreign reserves can boost investor confidence in the currency’s valuation. He expands on this 

logic by also demonstrating that large reserves, in the face of an economic shock, can also 

foster expansionary fiscal policy because government’s can sell the foreign assets for this 

purpose. Reserves therefore can act as stop-gap liquidity measures in the event of a shock, 

allowing the government to maintain domestic liquidity for foreign-denominated debts, as well 

as allowing the government to defend the currency against debilitating sharp depreciations. 

These reasons are why reserves accumulation is often discussed as a form of national insurance 

(Rodrik 2006; Obstfeld, Shambaugh, and Taylor 2010; Jeanne 2007). Recalling that Oatley 

(2004) and Leblang and Satyanath (2006) both found that democracies were more likely to face 

currency crises, foreign exchange reserves thus seem an ideal means to assess whether 

democracies are insulated from financial risks. This leads me to the third hypothesis: 

H3: Democracies will have better credit ratings at higher levels of Foreign 

Reserves/GDP. 

 

81 

 

Importantly, both of these financial vulnerability conditions – debt levels and foreign reserves 

holdings - are important considerations for the overall level of financial risk. But the clearest 

case for when democracies should be at a disadvantage to autocracies is both when there is a 

pressing need for economic adjustment, such as when debt is high, and when the country is 

insufficiently protected with foreign reserves. In this case, democracies both need to adjust 

their policies and are vulnerable to economic shocks because of low reserves. As a result, 

considering the impact of one without the other may present an incomplete picture. Therefore, 

the best test of the argument that financial vulnerability mediates the effect of democracies on 

country credit ratings utilizes a ratio of reserves to the debt level (rather than to GDP). This 

captures the purest indicator of the concept of “financial vulnerability” as laid out in the paper, 

and therefore will form the main component of the empirical analysis33. The final hypothesis of 

the paper is thus: 

H4: Democracies will have worse (better) credit ratings at low (high) levels of the 

reserves-to-debt ratio. 

Research Design 

Beaulieu, Cox, and Saiegh (2012) argue that democracies are more likely to secure credit access 

by gaining credit ratings, which they argue is not only an important advantage for democracies 

but is also a source of selection bias when this is not modeled. For instance, they suggest that 

rated autocracies are likely very different than rated democracies, and they show that rated 

                                                           

33 However, though I argue that this captures the single best metric of financial vulnerability, because 
Reserves/Debt contains both reserves and debt, using this measure exclusively prevents the investigation of 
the impact of debt alone, which forms the basis for hypotheses 1 and 2. Therefore, I present both measures 
while relying most upon the Reserves/Debt measure in the analysis. 

 

82 

 

autocracies are more likely to have large amounts of resource wealth, potentially be geo-

politically important, or possess stronger economic fundamentals. Thus, neglecting to model 

the determinants of entry into bond markets (or, securing credit ratings) is plausibly biasing the 

estimates of democracy in a single-equation analysis. As a consequence, the main model used 

in this paper is the Heckman selection model with clustered standard errors, following the 

strategy taken by Beaulieu et al (2012). I further follow their approach to satisfy the exclusion 

restriction by including the log of US-Imports in the selection equation but omitting this variable 

in the outcome equation. Beaulieu et al argue that this proxies for a countries’ “friendliness 

with major western powers, which should affect its decision to seek a rating but not the rating 

it receives, given that we already account for total trade”.  

While formally testing this excludability is not strictly possible, an observed significant effect in 

the outcome equation would call its suitability into question. Estimating Model 1 of Table 2.1 

including the (log) US-Imports in the outcome equation, it returns a very low p-value of 0.92. 

Additionally, I extend this approach to excludability by including a more explicit measure of 

‘western friendliness’ and thus the decision to enter sovereign debt markets. This is the UN 

Ideal Point distance estimate of state preferences produced by Bailey, Strezhnev, and Voeten 

(2017). Their measure estimates the distance between two countries voting ideal-points, and so 

for the present context I isolate the ideal distance between each country and the US. For the 

Moody Rating models, both this variable and the (log) US Imports variable are significant in the 

selection equation34. In the robustness section, I discuss an alternate exclusion restriction, 

finding that the results are unchanged across these specifications. 

                                                           

34 This variable was insignificant in the outcome equation when it was tested. 

 

83 

 

The sample covers the years 1977 to 2012, and is restricted to developing countries because 

the literature emphasizes that both levels of debt and foreign exchanges reserves are 

significantly more relevant for economic risk in these countries (Reinhart and Rogoff 2009; 

Jeanne 2007; Obstfeld, Shambaugh, and Taylor 2010; Rodrik 2007). For instance, interest in the 

determination of foreign exchange reserves is driven largely by the observation that, while 

reserves holdings have been very steady in the advanced economies, reserves have more than 

quintupled in poorer countries (Obstfeld et al 2010). This illustrates the fact that countries in 

the developing world are more susceptible to a variety of external crises and demonstrates that 

the use of reserves may have different policy inputs in the developed world. Further, Mosley 

(2003) argues that non-developed countries face a variety of additional financial pressures, 

including capital scarcity, exposure to investment flow volatility, currency and exchange rate 

risk, and especially the potential for sovereign debt default (which is essentially “absent” in 

developed countries).  Equally important to the concept of the democratic advantage is that 

Mosley then argues that investors care much more about political conditions in developing 

countries, which therefore also suggests that the influence of democratic institutions should be 

especially pronounced in the developing world. As a consequence of these factors, it is precisely 

in the developing world where we would expect the interplay between democratic institutions 

and financial vulnerability to be most apparent in their impact on credit ratings. 

There are two essential dependent variables used in the analysis. These are the country credit 

ratings given by Moody’s Analytics (the first DV) and Standard & Poor’s (the second). These 

ratings are a 17 point rating scale, which peaks at S&P’s “AAA” or Moody’s “Aaa” rating, 

indicating a country is extremely creditworthy and likely to repay its debts. Conversely, the 

 

84 

 

ratings fall to 0 for junk bond status for countries likely to default, which are ratings lower than 

CCC+ and Caa1, respectively.  

Table 2.A.1 of the appendix includes the full list of variables and the associated descriptive 

statistics for developing countries, along with the source for each variable. These are mostly 

standard control variables in the literature, as can be found in Beaulieu et al (2012) or Bodea 

and Hicks (2017). Table 2.A.2 in the appendix reports the full list of countries which appear in 

the main regression sample for Table 2.1, Model 5. Beaulieu et al (2012) also raise the concern 

over the endogeneity between democracy and GDP per capita, and so I follow their 

recommendation to use GDP as a means of addressing this issue35.  

To investigate the impact of democracy on credit ratings, I first employ the variable “Electoral 

Democracy”, from the publicly available Varieties of Democracy version 6 database. This 

variable aggregates the combination of factors considered essential for democracy: Freedom of 

Association, clean elections, freedom of expression, elected executive, and suffrage. This 

democracy measure is especially useful in this context because it incorporates the numerous 

features of democracy which can combine to make economic adjustment a difficult policy 

process. For instance, even if there are relatively clean elections and the executive is elected, an 

executive who intensely stifles freedom of association and expression likely would not face the 

same sorts of public pressure in the face of unpopular adjustment programs. Thus, combining 

these factors is ideal for the democratic advantage hypothesis.  

Beyond this, the VDEM Electoral Democracy index has numerous advantages compared with 

other democracy indicators. For instance, it was coded with a much larger number of coders, 

                                                           

35 I use the log form of GDP. 

 

85 

 

with broader source materials, and as it is a continuous scale between 0 and 1 it better 

discriminates different types of democracy and has greater variation than Polity, which is 

limited to its 21point scale. Perhaps most importantly, because the development team used 

Bayesian IRT modelling in the variable construction, they also account for measurement error 

and coder bias in the index (for a deeper discussion, see: Coppedge et al 2016 Methodology36). 

At the same time, I also report the Cheibub, Gandhi, and Vreeland (2010) “D-D” Democracy 

index in the main tables, which is the measure used by Beaulieu et al (2012). In doing so, we 

can see that the relationship is robust to both specifications and the results are therefore not  

contingent on the choice of democracy variable. 

The Debt/GDP ratio comes from the IMF’s Historical Public Debt Database. The data on foreign 

exchange reserves, for which I use two variables, comes from the World Bank WDI database, 

and both are standard in the literature (Rodrik 2006). The main variable, Foreign 

Reserves/External Debt, comes directly from the WDI. The second variable is the countries’ 

“total reserves minus gold”, which I then divide by GDP (also from the WDI) and I use this to 

create the ratio of a countries “foreign reserves / GDP”. Lastly, I take the natural logarithm of 

the reserves and debt variables because they all are highly skewed.  

Importantly, while in the main models the possibility of selection bias is controlled for by 

employing Heckman models, there further exists the possibility that reverse causality could bias 

the estimates. For instance, it is well established that debt levels influence ratings, but the level 

of ratings also influence the cost of debt, which therefore can also influence the overall debt 

level. Another possibility is that financial transparency may have a different impact on bond 

                                                           

36 https://www.v-dem.net/en/data/data-version-6-2/ 

 

86 

 

prices at higher debt levels37 (Copelovitch, Gandrud and Hallerberg forthcoming), and since 

democracies are more transparent this could potentially influence the results. I would argue, 

however, that even if transparency has a larger impact at high debt levels, this would be a 

positive bias for democracies and therefore a finding that democracies are worse at high debt 

levels would still validate my key hypothesis. Even so, the causal chain among these variables is 

not clearly established, and so it remains important to account for this potential endogeneity 

bias. In Table 2.2 I employ both System GMM and Two-Stage Least Squares models, using 

internal instruments to mitigate this concern.  The system GMM model is useful for several 

reasons. First, when the time sample is short but the panel is much larger, this can give rise to 

dynamic panel bias with a lagged dependent variable. As such, because the System GMM 

model estimates the parameters in levels and differences, the estimator expunges the fixed 

effect and instruments the lagged DV, such that there is no bias due to country fixed effects. 

System GMM, rather than Difference GMM or the Anderson-Hsiao estimator, is also useful for 

cases with a persistent dependent variable, making it well-suited for analyzing country credit 

ratings.  

Beyond this, the model also readily allows for the use of “internal” instruments for endogenous 

variables. In this case, in the GMM models 8:11, I treat the interaction estimates (democracy 

and reserves/debt ratio), the debt level, and FDI flows as endogenous, limiting the instruments 

to second-order lags in an effort to minimize the instrument count. In doing so, the GMM 

                                                           

37 In this study, CGH find that at high debt levels, transparency reduces the costliness of debt and has a larger 
substantive impact compared with small debt levels. It should also be noted that I do not control for 
transparency in the regression models because, as a key feature and benefit of democracies, controlling for 
this would bias the estimate of democracy by partialling out one of its key operative mechanisms (Wooldridge 
2010). 

 

87 

 

diagnostic statistics are all insignificant, suggesting that the instruments appear to be valid and 

exogenous. At the same time, the Hansen p-value is a value of one for these models, suggesting 

that an alternative approach reducing the instrument set even further could be helpful. In this 

case, the 2SLS framework allows for a more precise selection of instrumenting variables, in 

addition to illustrating robustness to using an alternative instrumenting framework to the 

System GMM. Doing so for models 12:15 reduces the excluded instrument38 set to 7, and also 

returns insignificant Sargan statistic p-values39, though now less than one. Additionally, tests for 

weak instrument bias were also insignificant40. Consequently, all models satisfy the exogeneity 

requirements with respect to the GMM or the 2SLS models. In addition to these standard tests, 

I also performed 5-fold cross validation among the core models I present in the main text and 

the appendix, wherein each model randomly held out a subset of data to examine how well the 

model performs in prediction out of sample. In this way, we can get better and unbiased sense 

of whether added model complexity is yielding a performance benefit or if it is somehow 

weakening the out of sample performance of the models. I compared Pooled-OLS, random 

effects, PCSE, 2SLS and the System GMM model, exactly as reported in the corresponding 

tables. Among these, the System GMM model yielded the lowest out of sample prediction error 

(conversely, the highest out of sample prediction accuracy), suggesting that the estimation 

method is valuable in addressing the potential issues of bias. Regardless, in all cases we see that 

                                                           

38 I only use a democracy lag of 1 period, since democracy itself is not an endogeneity concern. For Reserves 
and Debt, I used up to 3 year lags to instrument, as well as their first-order differences. 
39 The Hansen statistic, preferred after GMM (Roodman 2009), is not available after 2SLS. Therefore, I use the 
Sargan statistic. 
40 tests for instrument strength were conducted and the F-tests of the excluded instrument sets were very 
large, suggesting that weak instrument bias is not a concern here. Precise values for these tests are available 
upon request. 

 

88 

 

the key theoretical interactions retain statistical significance and are supportive of the main 

hypotheses. 

All of the hypotheses for the paper are conditional hypotheses. As such, the proper tests of the 

theory examine the interaction effects of democracy, Debt/GDP, and Reserves/ External Debt, 

which will require marginal effects plots for interpretation. Specifically, the test of the 

democratic (dis)advantage hypotheses will be the marginal effect of democracy across the 

range of debt and foreign reserves levels. All marginal effects plots use 95% confidence 

intervals, and the range on the x-axis for all graphs is limited to the range of the conditioning 

variable in the sample.  

Empirical Results  

Table 2.1 shows the estimation results from the 2nd stage outcome equations of the Heckman 

selection models. The corresponding selection equations are reported in Table 2.A.5 of the 

appendix. To show that the observed relationships are not driven by overparameterization, 

Models 1 through 3 are estimated with only a basic set of economic control variables, while 

models 4:7 are run with a full set of controls informed by the literature. Model 1 tests the first 

two hypotheses of the paper, that democracies experience worse credit ratings as debt 

increases (H1) and that high debt levels reduce credit ratings more strongly for democracies 

(H2). Model 2 then tests whether that democracies have an advantage as Reserves / GDP 

increase. In these cases, the interactions are highly significant (at p< 0.001) and in the expected 

direction. Figure 2.1 subplot (A) plots the marginal effect of Electoral Democracy across all 

observed values of Debt/GDP and we see that at low debt levels democracies enjoy significantly 

higher credit ratings (about 2.5 ratings higher). However, at higher debt levels, democracies 

 

89 

 

have significantly worse credit ratings by roughly the same degree. We therefore find support 

for this key contention. In Figure 2.1 subplot (B) we observe the marginal effect corresponding 

to H2, the effect of increasing Debt across all observed values of electoral democracy. We find 

strong support for the main argument of the paper, finding that higher debt does decrease 

credit ratings but this effect occurs much more strongly for democratic countries than more 

autocratic ones. Finally, Model 2 tests the interaction of democracy and Reserves/GDP, and see 

here as well that the interaction is highly significant and positive, as expected. While not 

reported, inspection of the marginal effects plot displays a pattern consistent with 

expectations. At low levels of reserves, democracy has a negative effect on ratings, while at a 

higher level democracies enjoy significant ratings advantages. 

Models 3 through 7 test hypothesis 3, showing the interaction of the democracy variables and 

Reserves/ External Debt (the unified risk measure), rather than both as a comparison of GDP. 

Because these models take into account both dimensions of financial risk, these models give us 

the clearest test of how financial vulnerability conditions the impact of democracy on credit 

ratings. For ease of visualization, the models employing the Reserves/Debt interaction are 

shaded in darker grey, matching the column header with the independent variable listed in the 

table. We see in all of these models that the interaction is significant and in the expected 

direction. In the fully specified models, Electoral Democracy is used in models 4 and 6, while D-

D Democracy is used in 5 and 7. The last two models, 6 and 7, use Standard & Poor’s credit 

ratings as the dependent variable. 

Figure 2.2 shows the marginal effects plots both electoral democracy (subplot A) and D-D 

Democracy (subplot B). We see strikingly similar relationships, although it is clear that electoral 

 

90 

 

democracy predicts a much larger ratings difference at various levels of Reserves/Debt. For 

instance, at low Reserves/Debt electoral democracy predicts about 3 ratings lower, compared 

with about 1.5 ratings lower for D-D Democracy. In order to get a sense of where democracies 

typically fall in this relationship, the vertical dashed blue line indicates the democracy mean 

value of Reserves/Debt. This shows us, therefore, that the average developing democracy 

appears to be at a disadvantage compared with their authoritarian counterparts. 

Lastly, for the selection equation in appendix 2.A.5, the US imports and UN ideal variables are 

significant and in the expected direction. The negative coefficient for the UN Ideal variable 

indicates that countries whose ideal point of state preferences is very distant from the US were 

less likely to secure a credit rating. Similarly, it shows that countries who import more goods 

from the US were also more likely to receive a credit rating. It is worth noting, however, that 

these selection variables are insignificant in the S&P models. 

As previously noted, aside from selection bias influencing the results, another endogeneity 

concern is that country ratings may affect debt levels and possibly investment. Table 2.2 

reports the models where these variables were treated endogenously, using internal 

instruments. None of the GMM or 2SLS diagnostic statistics are significant, which would be 

evidence that the instruments used fail to satisfy the exogeneity requirements. In all models 

the theoretical interactions are significant and in the expected direction. Appendix Figure 2.A.1 

plots the marginal effects of the instrumented relationships for the Electoral Democracy 

variable on both Moody’s and S&P, and we largely observe the same relationship as that found 

in the marginals of Figure 2.2. The results are thus robust to multiple endogeneity risks. 

 

 

91 

 

   

 

92 

(1)(2)(3)(4)(5)(6)(7)Interacted variable:Debt/GDPReserves/GDPRating Agency:Moody'sMoody'sMoody'sMoody'sMoody'sS&PS&PLagged DV (Rating t-1)0.890***0.893***0.835***0.830***0.842***0.785***0.787***(0.026)(0.024)(0.025)(0.041)(0.042)(0.030)(0.035)(log) Debt/GDP0.479**-0.155-0.0820.017-0.0350.202**0.113(0.231)(0.146)(0.114)(0.123)(0.127)(0.099)(0.102)Electoral Democracy4.997***-2.953**-1.980**-4.487***-4.570***(1.285)(0.914)(1.001)(0.782)(1.173)Electoral Dem * Debt/GDP-1.402***(0.392)Electoral Dem * Reserves/GDP1.175***(0.305)Electoral Dem * Reserves/Debt0.611**1.148***1.278***(0.256)(0.253)(0.344)D-D Democracy * Reserves/Debt0.478**0.516**(0.147)(0.188)D-D Democracy-1.848***-1.836**(0.435)(0.686)(log) Reserves/Ext Debt0.164-0.332**0.003-0.2620.218*(0.168)(0.136)(0.099)(0.166)(0.114)(log) Reserves/GDP0.150-0.398*(0.146)(0.219)Fiscal Balance (% GDP)0.0190.0210.046**0.0430.0390.037**0.036*(0.020)(0.019)(0.021)(0.032)(0.031)(0.017)(0.022)Growth0.092***0.089***0.089***0.085***0.082***0.133***0.125***(0.019)(0.020)(0.020)(0.020)(0.018)(0.017)(0.017)Trade Openness0.0010.0020.0020.0010.001-0.000-0.000(0.002)(0.001)(0.001)(0.001)(0.001)(0.002)(0.002)(log) GDP0.0520.0700.102**0.331**0.2190.400***0.326**(0.062)(0.048)(0.044)(0.122)(0.145)(0.119)(0.103)Time-0.018-0.015-0.010-0.054**-0.053*-0.031*-0.023(0.013)(0.011)(0.019)(0.026)(0.028)(0.018)(0.018)Central Bank Independence0.1730.1550.090-0.055(0.260)(0.243)(0.299)(0.296)(log) Population-0.285**-0.210*-0.318**-0.252**(0.119)(0.123)(0.121)(0.112)Capital Acct Openness-0.0700.023-0.0800.072(0.229)(0.201)(0.298)(0.298)(log) CPI Inflation-0.175*-0.170*-0.147-0.102(0.105)(0.103)(0.107)(0.089)(log) Oil Rents-0.079-0.120-0.045-0.056(0.087)(0.085)(0.053)(0.061)(log) FDI-0.043-0.0500.1120.083(0.114)(0.124)(0.084)(0.079)Ever Default-0.067-0.132-0.413-0.391(0.284)(0.307)(0.328)(0.354)Constant33.42429.29618.503107.207**104.841*59.728*41.424(26.953)(21.958)(37.909)(52.795)(57.859)(35.254)(35.924)Observations1818181816491193119911911196Rho p-value0.2980.1700.4080.8520.8920.9550.531Table 2.1: Heckman Selection Models (Outcome Equations)Clustered standard errors in parentheses; * p<0.1 ** p<0.05 *** p<0.001; Models are Heckman selection outcome equations. DV is country credit rating. Corresponding Selection equations reported in Appendix. Reserves/Debt   

 

 

93 

 

As a final point, I had previously mentioned that the average developing democracy seems to 

have an insufficient level of Reserves/Debt in order to enjoy the democratic advantage. While a 

full exploration of this is beyond the scope of the paper, it is worth considering whether 

democracies properly utilize the benefits of foreign reserves. In other words, it may be that 

democracies are also predisposed to underutilize foreign reserves as a function of economic 

insurance. The reason is that reserves are not costless insurance. As Rodrik (2006) discusses, 

the opportunity costs to holding reserves is high because reserves are low interest bearing 

assets and consequently only yield very small returns on investment. It may be that 

democracies are less willing to bear the costs of reserves. Further, Jäger (2016) shows that since 

the middle-1990’s developing democracies have accumulated fewer foreign reserves than 

autocracies, and shows that democracies opportunistically deplete reserves around elections 

(therefore, over time, depleting the stock of reserves). As a result, even though I demonstrate 

that democracies have ratings advantages as the level of reserves increases, it is at the same 

time not immediately clear that democracies will be readily able to draw upon this financial 

insurance in order to maintain preferential ratings and debt-finance. This is an issue which will 

require deeper exploration in the future. 

 

94 

 

 

 

95 

(8)(9)(10)(11)(12)(13)(14)(15)Lagged DV (Rating t-1)0.836***0.859***0.799***0.789***0.906***0.909***0.876***0.864***(0.037)(0.050)(0.038)(0.042)(0.014)(0.018)(0.017)(0.021)Electoral Democracy-3.461***-2.767**-2.874***-1.156*(0.560)(1.061)(0.768)(0.700)(log) Reserves/Ext Debt-0.1440.103-0.05610.279**-0.326**-0.106-0.05040.154**(0.089)(0.105)(0.140)(0.102)(0.131)(0.088)(0.107)(0.059)Electoral Dem * (log) Reserves/Debt0.744***0.720**0.711***0.325**(0.177)(0.268)(0.181)(0.162)D-D Democracy * (log) Reserves/Debt0.244*0.298**0.383***0.206*(0.136)(0.136)(0.092)(0.106)D-D Democracy-1.327**-1.233**-1.636***-0.766*(0.415)(0.490)(0.370)(0.424)(log) Debt/GDP0.1070.1050.01610.037-0.0598-0.0530.0290.124*(0.071)(0.086)(0.102)(0.101)(0.060)(0.088)(0.063)(0.075)Fiscal Balance (% GDP)0.0371**0.0346**0.0269**0.0188**0.01210.006560.0287**0.0201*(0.014)(0.017)(0.008)(0.008)(0.012)(0.014)(0.009)(0.011)Central Bank Independence0.0040.194-0.06-0.143-0.164-0.1-0.069-0.202(0.297)(0.323)(0.241)(0.222)(0.191)(0.206)(0.147)(0.179)(log) Population-0.337***-0.247**-0.250**-0.247**-0.184**-0.180**-0.170**-0.222**(0.092)(0.082)(0.077)(0.085)(0.066)(0.076)(0.059)(0.070)Growth0.0514***0.0619***0.0688***0.0904***0.0609***0.0708***0.0900***0.107***(0.009)(0.009)(0.010)(0.012)(0.013)(0.017)(0.013)(0.016)(log) GDP0.482***0.328***0.414***0.358***0.267***0.213***0.278***0.286***(0.085)(0.085)(0.076)(0.080)(0.049)(0.061)(0.050)(0.065)Capital Acct Openness0.1260.106-0.02970.05190.1870.2510.06840.122(0.171)(0.180)(0.147)(0.179)(0.169)(0.193)(0.140)(0.177)(log) CPI Inflation-0.138*-0.122-0.146*-0.115-0.147**-0.121*-0.144**-0.120*(0.081)(0.088)(0.087)(0.097)(0.061)(0.062)(0.060)(0.069)(log) FDI-0.014-0.0870.138*0.114-0.033-0.0890.007-0.016(0.156)(0.177)(0.079)(0.087)(0.083)(0.092)(0.045)(0.056)(log) Oil Rents-0.094-0.108-0.0819*-0.051-0.008-0.019-0.0290.018(0.070)(0.078)(0.045)(0.057)(0.033)(0.036)(0.038)(0.043)Trade Openness0.0010.0010.003*0.002-0.001-0.0010.0010(0.001)(0.002)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)Ever Default0.049-0.04190.0432-0.0459-0.0619-0.134-0.0978-0.116(0.181)(0.246)(0.131)(0.204)(0.114)(0.172)(0.130)(0.220)Constant-3.124***-2.152***-2.369***-2.377***-1.22-0.749-3.404***-3.697***(0.833)(0.998)(0.716)(0.657)(0.798)(1.072)(0.628)(0.753)Observations675538658493631493612447# of Countries5046595350445952Hansen p-value11110.3310.3180.4050.714AR(2) p-value (for GMM)0.5980.5320.2510.218....* p<0.1 ** p<0.05 *** p<0.001; Models8:11 are One-Step System GMM models with robust standard errors, instrumenting for specific variables (See text). Models 12:15 are 2SLS models instrumenting for the interaction terms. Dependent variable is country credit rating (See column header for specific rating agency). For the 2SLS models, the Hansen p-value actually corresponds to the Sargan p-value for assessing instrument exogeneity.Table 2.2: Endogeneity Models - GMM and 2SLS2SLSSystem GMMMoody'sMoody'sS&PS&P 

Robustness 

One might be concerned to know the differences of debt and reserves between regime types, 

and whether there are notable differences. Figures 2.A.4 and 2.A.5 in the appendix show that 

democracies and autocracies have very similar descriptive profiles in terms of debt and reserves 

levels. It also demonstrates significant variation in terms of countries with high debt and low 

reserves, and vice-versa, and thus the paper estimates meaningful real-world cases. 

Additionally, in appendix Table 2.A.9 I present a seemingly unrelated regression model, which 

demonstrates that democracies and autocracies do not differ in terms of how much debt or 

reserves they accumulate and that these determinants do not mitigate the main hypotheses – 

even controlling for potential correlation between these equations, the main results hold. 

Additionally, it is noteworthy that in most models the (ρ) Rho parameter, which tests the 

necessity of modeling the selection process, is insignificant. This is suggestive that modeling 

selection in this way may be introducing unnecessary inefficiency in the outcome equation.It 

would be fruitful to examine the estimated relationship with a model that ignores the potential 

endogeneity. Similar to Beaulieu et al (2012), in appendix table 2.A.3 I estimate the models as 

OLS with panel-corrected standard errors, as well as random effects models with country 

clustered standard errors. In this case, we find very consistent results and the main hypotheses 

hold. Turning next to appendix table 2.A.4, I estimate the interaction models using Polity2, and I 

separately estimated a quantile regression out of concern that the results are driven only 

among the lowest rated of countries41. In both cases the results hold. Table 2.A.6 shows 

                                                           

41 I estimated a “median regression” in this case. Alternatively, I reestimated the main regression models 
excluding countries within 2 ratings of being “junk status”, and the results held. 

 

96 

 

condensed Heckman results, but with the inclusion of additional control variables – Including a 

measure of the amount of IMF Credit a country has outstanding, as well as controlling for 

whether a country employs a fixed exchange rate regime (which likely would influence its level 

of reserve holdings). IMF Credit is significant in both the selection and outcome equations, 

while the fixed exchange rate measure is insignificant in all cases. The main results, however, 

remain the same across these models. 

Out of concern that the results are unduly driven by Asian countries accumulating large 

amounts of reserves, I estimated Models 1 and 4 from Table 2.1 but excluded the Asian region, 

and the results appeared virtually identical. Additionally, noting that the UN Ideal and US 

Imports exclusion variables were not significant in the S&P model, I also estimated a model 

replacing the UN Ideal measure with the (log) of the M2 monetary base (% GDP), in table 2.A.7, 

serving as an alternate exclusion restriction. This likely would influence a countries decision to 

seek a credit rating, and therefore be selected into the ratings markets, but its potential effect 

on the outcome is not obviously positive or negative. For instance, it could indicate greater 

financial ability (a positive), but would also be higher in hyperinflationary cases or where the 

central bank is providing too much credit to the economy.  In this case, the variable is highly 

significant in all selection equations, and was insignificant in all outcome equations. The 

inferences of the analysis hold in this case as well. Figure 2.A.3 in the appendix reports the 

marginal effect plot of debt across democracy scores analogous to Figure 2.1 (B), but instead 

estimated from a full model with controls comparable to Table 2.1, model 4, showing similar 

results. Lastly, the lagged dependent variables for the ratings variables indicate the data is very 

persistent. One might be concerned with the presence of a unit root in such a context, and 

 

97 

 

consequently I estimate an Error-Correction Model to examine robustness to this specification. 

Reported in table 2.A.8, we see for the levels estimations that the significance holds, and Figure 

2.A.2 shows the estimated marginal effects from the levels-ECM model. 

Conclusion 

This paper goes a long ways to advancing our understanding of the role that democratic 

institutions play to enable credible commitments, and it significantly revises our understanding 

of whether democracies enjoy preferential treatment in financial markets. I demonstrate that 

democracies, contrary to the literature on the democratic advantage, tend to have a credit 

rating disadvantage compared with autocracies.  Specifically, financially vulnerable democracies 

are significantly more likely to have negative ratings, while financially secure democracies do 

enjoy financial market advantages. This paper also shows the powerful role which foreign 

reserves can play in attenuating the risks posed by debts and economic risk, and helps to 

highlight the importance for such reserves in democratic polities.

 

98 

 

 

 

APPENDIX 

 

99 

 

 

Appendix 

 

100 

AlbaniaDominican RepublicaLithuaniaPhilippinesArgentinaEcuadorMalaysiaRomaniaArmeniaEgyptMauritiusRussiaAzerbaijanEl SalvadorMexicoSenegalBangladeshFijiMoldovaSouth AfricaBelarusGeorgiaMongoliaSri LankaBoliviaGuatemalaMoroccothailandBotswanaHondurasNamibiaTunisiaBrazilIndiaNicaraguaTurkeyBulgariaIndonesiaPakistanUkraineChileJamaicaPanamaUruguayChinaJordanPapua New GuineaVenezuelaColombiaKazakhstanParaguayCosta RicaLebanonPeruTable 2.A.2: Countries in Regression SampleVariableObsMeanStd. Dev.MinMaxSource:Moody's6755.563.61016Bodea & Hicks (2017)S&P5725.422.99016Bodea & Hicks (2017)Electoral Democracy6750.540.220.100.93Varieties of Democracy v. 6D-D Democracy5290.650.4801Cheibub et al (2010)(log) Reserves (% External Debt)6753.580.940.787.75World Bank WDI(log) Reserves (% GDP)6752.630.700.384.62World Bank WDI(log) Debt (% GDP)6753.740.611.795.47IMF Public Debt DatabaseFiscal Balance (% GDP)675-1.553.76-14.6019.99Bodea & Higashijima (2015)Central Bank Independence6750.550.210.130.90Garriga (2016)(log) Population67517.021.7413.6121.02World Bank WDIGDP Growth (%)6754.254.18-14.8025.05World Bank WDI(log) GDP67524.891.7020.8829.78World Bank WDICapital Account Openness6750.520.3501Chinn-Ito Index(log) CPI Inflation6752.031.07-0.887.99World Bank WDI(log) FDI6751.290.75-2.413.47World Bank WDI(log) Oil Resource Wealth6750.901.0304.07World Bank WDITrade Openness67577.6339.6713.75220.41World Bank WDIEver Default6750.820.3901Bodea & Hicks (2017)UN Ideal Scores3393.050.611.534.55Bailey et al (2009)(log) M2 (% GDP)6713.890.561.225.49World Bank WDIIMF Credit6402.05E+094.09E+0902.89E+10World Bank WDI(log) US Imports5657.591.763.3512.02World Bank WDIFixed Exchange Rate5710.280.4501Reinhart and Rogoff (2002)Table 2.A.1: Descriptive Statistics for Main Regression Sample 

 

 

 

101 

(1)(2)(3)(4)(5)(6)(7)(8)Model:Rating Agency:Lagged DV (Rating t-1)0.873***0.872***0.854***0.823***0.873***0.872***0.854***0.823***(0.024)(0.028)(0.027)(0.036)(0.022)(0.029)(0.024)(0.031)(log) Reserves/Ext Debt-0.195*0.051-0.0170.202**-0.195*0.051-0.0170.202**(0.112)(0.084)(0.109)(0.076)(0.102)(0.079)(0.105)(0.081)Electoral Democracy-2.839***-1.438**-2.839***-1.438*(0.819)(0.718)(0.581)(0.784)Electoral Dem * Reserves/Debt0.689***0.390**0.689***0.390**(0.190)(0.173)(0.140)(0.188)D-D Dem * Reserves/Debt0.337***0.229**0.337***0.229**(0.095)(0.094)(0.082)(0.112)Democracy-1.416***-0.895**-1.416***-0.895*(0.394)(0.370)(0.304)(0.462)(log) Debt/GDP-0.07-0.041-0.0090.035-0.07-0.041-0.0090.035(0.067)(0.076)(0.062)(0.090)(0.058)(0.076)(0.061)(0.079)Fiscal Balance (% GDP)0.0150.0150.024**0.018**0.0150.0150.024**0.018*(0.011)(0.013)(0.008)(0.008)(0.013)(0.018)(0.008)(0.009)Central Bank Independence-0.0090.0510.013-0.076-0.0090.0510.013-0.076(0.213)(0.282)(0.145)(0.187)(0.204)(0.209)(0.178)(0.210)(log) Population-0.267***-0.225***-0.202***-0.206**-0.267***-0.225**-0.202**-0.206**(0.061)(0.066)(0.051)(0.068)(0.074)(0.080)(0.067)(0.079)Growth0.056***0.066***0.088***0.105***0.056***0.066***0.088***0.105***(0.013)(0.016)(0.012)(0.014)(0.012)(0.013)(0.012)(0.014)(log) GDP0.370***0.277***0.314***0.302***0.370***0.277***0.314***0.302***(0.059)(0.067)(0.049)(0.057)(0.075)(0.082)(0.063)(0.072)Capital Acct Openness0.1180.1350.0330.1340.1180.1350.0330.134(0.121)(0.149)(0.100)(0.147)(0.140)(0.163)(0.138)(0.188)(log) FDI-0.009-0.0380.0280.023-0.009-0.0380.0280.023(0.056)(0.063)(0.053)(0.054)(0.076)(0.090)(0.048)(0.065)(log) CPI Inflation-0.172***-0.155***-0.159***-0.132**-0.172**-0.155**-0.159**-0.132*(0.040)(0.042)(0.043)(0.048)(0.075)(0.077)(0.068)(0.075)(log) Oil Rents-0.049-0.054-0.051-0.023-0.049-0.054-0.051-0.023(0.054)(0.066)(0.040)(0.049)(0.050)(0.064)(0.039)(0.055)Trade Openness0.0010.0010.0020.0010.0010.0010.002*0.001(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)Ever Default0.0390.001-0.081-0.1420.0390.001-0.081-0.142(0.094)(0.128)(0.091)(0.127)(0.130)(0.199)(0.135)(0.220)Time-0.032**-0.040**-0.019**-0.030**-0.032**-0.040**-0.019**-0.030**(0.010)(0.015)(0.009)(0.013)(0.010)(0.014)(0.007)(0.012)Constant60.498**77.644**35.396**56.789**60.498**77.644**35.396**56.789**(19.961)(28.849)(17.698)(25.976)(19.106)(27.270)(14.299)(22.605)Observations675538658493675538658493# of Countries5046595350465953Ave # years/country13.511.69611.1539.30213.511.69611.1539.302Table 2.A.3: PCSE and Random Effects ModelsCountry clustered standard errors in parentheses; * p< 0.1 ** p< 0.05 *** p< 0.001; Models 1:4 are OLS with PCSEs; Models 5:8 are random effects. PCSEREMoody'sS&PMoody'sS&P 

 

 

 

102 

(1)(2)(3)(4)Model:Rating Agency:Moody'sS&PMoody'sS&PLagged DV (Rating t-1)0.844***0.798***0.850***0.797***(0.045)(0.034)(0.039)(0.041)(log) Reserves/Ext Debt0.0850.273**0.1000.276**(0.090)(0.094)(0.092)(0.088)Polity 2-0.147***-0.129**-0.113***-0.063*(0.037)(0.047)(0.024)(0.035)Polity2 Dem * Reserves/Debt0.037**0.038**0.026***0.019**(0.012)(0.013)(0.007)(0.009)(log) Debt/GDP-0.0080.182*0.0760.008(0.119)(0.100)(0.072)(0.093)Fiscal Balance (% GDP)0.040.033*0.034**0.024**(0.032)(0.018)(0.014)(0.008)Central Bank Independence0.213-0.010.033-0.219(0.251)(0.316)(0.261)(0.217)(log) Population-0.23-0.295**-0.255**-0.242**(0.140)(0.113)(0.097)(0.079)Growth0.088***0.133***0.055***0.072***(0.020)(0.017)(0.008)(0.010)(log) GDP0.2530.358***0.394***0.386***(0.159)(0.107)(0.098)(0.083)Capital Acct Openness-0.067-0.0890.132-0.045(0.253)(0.322)(0.169)(0.148)(log) CPI Inflation-0.178*-0.125-0.152*-0.151(0.107)(0.112)(0.087)(0.093)(log) Oil Rents-0.088-0.043-0.093-0.059(0.085)(0.055)(0.069)(0.050)(log) FDI-0.050.0920.0260.132(0.116)(0.085)(0.165)(0.086)Trade Openness0.001-0.0010.0010.003(0.002)(0.002)(0.001)(0.002)Ever Default-0.091-0.351-0.0180.063(0.320)(0.366)(0.191)(0.150)Observations11911189673656# of Countries in outcome50595059# years/Country13.4611.119Hansen p-value1.0001.000AR(2) p-value0.5710.267Cluster robust standard errors in parentheses; * 0.1 ** 0.05 *** 0.001; Models 1 and 2 are Heckman Selections Models (outcome equations); Models 3 and 4 are system GMM, with the same estimation procedures as reported in Table 2System GMMHeckmanTable 2.A.4: Polity2 Robustness Models 

 

 

 

 

103 

 

 

 

104 

(1b)(2b)(3b)(4b)(5b)(6b)(7b)Interacted variable:Debt/GDPReserves /GDPReserves /DebtReserves /DebtReserves /DebtReserves /DebtReserves /DebtRating Agency:Moody'sMoody'sMoody'sMoody'sMoody'sS&PS&PElectoral Democracy1.99-2.0732.4660.5480.622(1.505)(2.324)(2.271)(2.223)(2.470)Electoral Dem * Debt/GDP-0.396(0.390)Electoral Dem * Reserves/GDP1.158(0.880)Electoral Dem * Reserves/Debt-0.3120.2020.385(0.674)(0.624)(0.686)D-D Dem * Reserves/Debt-0.420*0.298(0.250)(0.237)Democracy1.898**-0.023(0.798)(0.892)(log) Debt/GDP0.077-0.0620.1580.2480.236-0.262-0.213(0.231)(0.138)(0.239)(0.262)(0.267)(0.222)(0.222)(log) Reserves/GDP0.166-0.35(0.204)(0.521)(log) Reserves/Ext Debt0.632-0.1940.034-0.0850.036(0.414)(0.375)(0.217)(0.348)(0.137)Fiscal Balance (% GDP)-0.029*-0.029*-0.0120-0.0010.0470.033(0.016)(0.017)(0.028)(0.027)(0.026)(0.032)(0.034)Growth-0.005-0.005-0.017-0.074***-0.074***-0.101***-0.084***(0.012)(0.013)(0.022)(0.017)(0.017)(0.017)(0.017)(log) GDP-0.0270.0210.251*1.658***1.726***1.241***1.491***(0.211)(0.185)(0.151)(0.327)(0.369)(0.319)(0.320)(log) US-Imports0.649***0.640***0.638***0.309*0.370*0.2060.151(0.137)(0.135)(0.132)(0.168)(0.191)(0.176)(0.166)UN voting Ideal-p dist-0.595***-0.572***-0.539**-1.067**-1.056**-0.1180.091(0.167)(0.164)(0.274)(0.334)(0.332)(0.201)(0.200)Time0.116***0.120***0.186***0.157***0.163***0.187***0.189***(0.019)(0.018)(0.024)(0.026)(0.030)(0.027)(0.028)Central Bank Independence-1.177-1.388*-0.032-0.131(0.743)(0.761)(0.812)(0.752)(log) Population-0.687**-0.838**-0.364-0.726**(0.251)(0.259)(0.244)(0.238)Capital Acct Openness0.6320.6570.3980.202(0.438)(0.509)(0.402)(0.441)(log) CPI Inflation-0.379***-0.382***-0.457***-0.438***(0.090)(0.092)(0.104)(0.103)(log) Oil Rents-0.123-0.0620.0990.118(0.160)(0.165)(0.184)(0.189)(log) FDI0.1250.0830.2220.224(0.132)(0.126)(0.163)(0.153)Trade Openness0.019**0.020***0.010*0.007(0.006)(0.006)(0.006)(0.005)Constant-235.273***-241.135***-383.787***-341.761***-353.480***-398.403***-402.617***(40.702)(38.159)(49.254)(51.646)(58.714)(53.621)(56.486)Observations1818181816491193119911911196Rho p-value0.2980.170.4080.8520.8920.9550.531Dependent Variable is a dummy variable indicating that a country has received a credit rating. Columns correspond to the same numbers in Table 1 (thus, model 1 in Table 1 corresponds to Model 1b here). Table 2.A.5: Heckman Selection Equations (corresponding to Table 2.1) 

 

 

 

 

 

 

 

 

 

 

 

105 

(1)(2)(3)(4)(5)(6)Lagged DV (Rating t-1)0.852***0.866***0.782***0.783***0.819***0.828***(0.025)(0.026)(0.030)(0.034)(0.048)(0.051)(log) Reserves/Ext Debt-0.273*0.101-0.2720.260**-0.2300.076(0.145)(0.092)(0.196)(0.121)(0.178)(0.111)Electoral Democracy-4.330***-4.562***-3.955***(0.764)(1.265)(0.870)Electoral Dem * Reserves/Debt1.164***1.314***0.934***(0.217)(0.397)(0.249)D-D Dem * Reserves/Debt0.457***0.463**0.325*(0.110)(0.225)(0.167)Democracy-1.660***-1.629**-1.388**(0.363)(0.732)(0.516)IMF Credit (curr USD$)-0.000**-0.000*-0.000-0.000(0.000)(0.000)(0.000)(0.000)Fixed Exchange Rate-0.059-0.075(0.173)(0.164)Constant56.04659.90353.30818.35445.77543.592(40.554)(47.949)(66.937)(56.886)(41.344)(46.238)Selection Models (condensed)(log) US-Imports0.2130.2850.1360.1360.387**0.506**(0.187)(0.200)(0.199)(0.198)(0.191)(0.199)UN voting Ideal-p dist-1.107***-1.132***-0.1620.111-1.143**-1.179**(0.324)(0.307)(0.260)(0.264)(0.420)(0.411)IMF Credit (curr USD$)0.000*0.000**0.0000.000(0.000)(0.000)(0.000)(0.000)Fixed Exchange Rate0.4860.585(0.381)(0.375)Observations1053105910511056821825Rho p-value0.0160.0010.9380.5000.8250.737* p< 0.1 ** p<0.05 *** p< 0.001; All Models are Heckman Selection Models. Cluster Robust SE's in parentheses; Extra control variables Fixed exchange rate and IMF Credit are employed.[Other control varibles omitted]Table 2.A.6: Additional ControlsMoody'sS&PMoody's 

 

 

 

 

 

 

 

 

 

 

 

 

 

106 

(1)(2)(3)(4)Lagged DV (Rating t-1)0.868***0.872***0.844***0.824***(0.024)(0.029)(0.027)(0.030)(log) Reserves/Ext Debt-0.258**0.048-0.0980.198**(0.091)(0.078)(0.111)(0.081)Electoral Democracy-3.511***-2.369**(0.575)(0.775)Electoral Dem * Reserves/Debt0.848***0.649**(0.142)(0.198)D-D Dem * Reserves/Debt0.334***0.233**(0.081)(0.109)D-D Democracy-1.398***-0.899**(0.310)(0.454)(log) Debt/GDP-0.029-0.0400.0450.031(0.067)(0.075)(0.073)(0.078)Constant67.022**74.948**40.194**53.345**(25.000)(27.238)(17.120)(22.614)Selection Equations:(log) US-Imports0.214*0.2010.1630.172(0.128)(0.135)(0.149)(0.150)(log) M2/GDP0.404**0.419**0.410**0.500***(0.163)(0.162)(0.138)(0.148)Constant-130.128***-140.084***-234.021***-238.106***(33.115)(38.528)(38.983)(41.809)Observations1616158016051568Rho p-value0.4730.4390.1320.461Table 2.A.7: Alternate Exclusion Variable for HeckmanMoody'sS&P* p< 0.1 ** p< 0.05 *** p< 0.001; Cluster Robust SE's in parentheses.  All Models are Heckman Selection Models, run with full set of controls in Table 1 but omitted here to conserve space. (log) M2/GDP serves as the exclusion variable, replacing "UN Ideal Score" in Table 1.  

 

 

 

 

 

107 

lagdifferencedlagdifferencedlagdifferencedlagdifferencedLagged DV (Rating t-1)-0.117***-0.119***-0.121***-0.135***(0.022)(0.029)(0.018)(0.024)(log) Reserves/Ext Debt-0.194*0.0220.0440.331**0.037-0.1150.237***0.26(0.115)(0.205)(0.082)(0.135)(0.091)(0.327)(0.064)(0.164)Electoral Democracy-2.740***-2.501-1.220*-2.969(0.622)(1.852)(0.684)(2.139)Electoral Dem * Reserves/Debt0.657***0.6120.337**0.753(0.153)(0.438)(0.167)(0.571)D-D Democracy * Reserves/Debt0.300***0.218**0.160*0.171(0.085)(0.108)(0.091)(0.157)D-D Democracy-1.290***-0.943**-0.611*-0.874(0.315)(0.378)(0.367)(0.670)(log) Debt/GDP-0.064-0.407-0.022-0.4740.019-0.3210.128**-0.383(0.051)(0.376)(0.078)(0.390)(0.056)(0.254)(0.062)(0.266)Fiscal Balance (% GDP)0.012-0.0020.017-0.0050.016**0.010.0130(0.012)(0.018)(0.017)(0.022)(0.008)(0.009)(0.009)(0.008)Central Bank Independence-0.1130.051-0.0260.249-0.13-0.744-0.136-0.652(0.180)(0.717)(0.181)(0.802)(0.141)(0.679)(0.139)(0.713)(log) Population-0.226***-2.642-0.185**-5.669*-0.164***-4.632*-0.234***-4.576(0.060)(3.196)(0.064)(3.199)(0.047)(2.633)(0.046)(3.471)Growth0.0160.032**0.0160.036**0.052***0.066***0.062***0.060***(0.017)(0.013)(0.019)(0.014)(0.011)(0.012)(0.014)(0.012)(log) GDP0.334***1.338**0.248***1.554**0.266***0.855*0.279***1.623**(0.059)(0.520)(0.069)(0.666)(0.049)(0.438)(0.056)(0.621)Capital Acct Openness0.1610.3890.1160.329-0.0251.412**-0.1171.429**(0.148)(0.425)(0.165)(0.473)(0.100)(0.649)(0.116)(0.585)(log) CPI Inflation-0.197**-0.084-0.201**-0.051-0.168**-0.151**-0.144**-0.154*(0.065)(0.080)(0.071)(0.091)(0.058)(0.073)(0.068)(0.086)(log) Oil Rents-0.021-0.269-0.0410.018-0.0420.241-0.0291.059***(0.035)(0.411)(0.045)(0.478)(0.034)(0.201)(0.041)(0.287)Trade Openness0.001-0.0070.001-0.0090.002**-0.0060.002-0.001(0.001)(0.006)(0.001)(0.006)(0.001)(0.006)(0.001)(0.007)Ever Default0.0430.2870-0.3620.083-1.5810.092-4.169***(0.095)(0.436)(0.155)(0.434)(0.093)(1.326)(0.112)(1.229)Time-0.029**-0.046***-0.021**-0.047***(0.009)(0.013)(0.008)(0.011)Constant55.988**90.619***37.754**91.138***(17.678)(24.960)(16.165)(20.855)Observations# of CountriesAve Years/CountryTable 2.A.8: Robustness Error Correction Models(1)(2)(3)(4)Models are Error Correction Models - each model has output from a lagged-variable and differenced-variable specification. Clustered standard errors in parentheses; * p<0.1 ** p<0.05 *** p<0.001.Moody'sS&P6645013.285294611.56435910.898480539.057 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

108 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

109 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

110 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

111 

 

 

 

 

 

 

112 

(1)(2)(3)DV =Moody Rating(log) Debt / GDP(log) Reserves / GDPLagged- DV0.897***0.899***0.875***(0.014)(0.010)(0.017)VDEM Democracy1.972*0.076-0.037(0.877)(0.095)(0.229)(log) Debt/GDP0.219(log) Reserves / GDP0.023(log) Debt/GDP0.038(0.142)(0.021)(0.037)VDEM Democracy * Debt/ GDP (log)-0.598**  Democracy * Reserves / GDP (log)-0.032 Democracy * Debt / GDP (log)-0.004(0.230)(0.035)(0.060)Fiscal Balance (% GDP)0.019-0.012***0.014***(0.010)(0.002)(0.003)Central Bank Independence-0.034-0.129***0.074(0.193)(0.030)(0.049)(log) Population-0.227***0.0050.055***(0.056)(0.009)(0.014)Growth0.066***-0.021***-0.012***(0.009)(0.001)(0.002)(log) GDP0.309***-0.004-0.026*(0.057)(0.008)(0.013)Capital Acct Openness0.1360.0010.019(0.118)(0.019)(0.031)(log) CPI Inflation-0.166***-0.015*0.005(0.038)(0.006)(0.009)(log) Oil Rents-0.0310.007-0.012(0.041)(0.006)(0.011)(log) FDI-0.009-0.0050.045**(0.055)(0.009)(0.014)Trade Openness-0.000-0.0000.001**(0.001)(0.000)(0.000)Ever Default-0.1250.003-0.052*(0.098)(0.016)(0.026)Time Trend-0.017*-0.0010.003(0.007)(0.001)(0.002)Constant31.270*1.682-5.369(13.571)(2.017)(3.334)Model is SUR Model. Models 1 through 3 estimated with same control variables, but different DV and different interaction variable with Electoral Democracy. * p<0.05 ** p<0.01 *** p<0.001.Table 2.A.9: Seeming Unrelated Regression Robustness Model 

 

 

 

BIBLIOGRAPHY 

113 

 

 
 

Bibliography 

Ahlquist, John S. 2006. “Economic Policy, Institutions, and Capital Flows: Portfolio and Direct 

Investment Flows in Developing Countries.” International Studies Quarterly 50 (3): 681–
704. 

Alesina, Alberto, and Allan Drazen. 1991. “Why Are Stabilizations Delayed?” The American 

Economic Review 81 (5): 1170–88. 

Archer, Candace C., Glen Biglaiser, and Karl DeRouen. 2007. “Sovereign Bonds and the 

‘Democratic Advantage’: Does Regime Type Affect Credit Rating Agency Ratings in the 
Developing World?” International Organization 61 (2): 341–65. 
doi:10.1017/S0020818307070129. 

Arpac, Ozlem, Graham Bird, and Alex Mandilaras. 2008. “Stop Interrupting: An Empirical 

Analysis of the Implementation of IMF Programs.” World Development 36 (9): 1493–
1513. doi:10.1016/j.worlddev.2007.09.001. 

Bailey, Michael A., Anton Strezhnev, and Erik Voeten. 2017. “Estimating Dynamic State 

Preferences from United Nations Voting Data.” Journal of Conflict Resolution 61 (2): 
430–56. doi:10.1177/0022002715595700. 

Bearce, David H., and Mark Hallerberg. 2011. “Democracy and De Facto Exchange Rate 

Regimes.” Economics & Politics 23 (2): 172–94. doi:10.1111/j.1468-0343.2011.00381.x. 

Beaulieu, Emily, Gary W. Cox, and Sebastian Saiegh. 2012. “Sovereign Debt and Regime Type: 

Reconsidering the Democratic Advantage.” International Organization 66 (04): 709–738. 
doi:10.1017/S0020818312000288. 

Biglaiser, Glen, and Joseph L. Staats. 2012. “Finding the ‘Democratic Advantage’ in Sovereign 

Bond Ratings: The Importance of Strong Courts, Property Rights Protection, and the Rule 
of Law.” International Organization 66 (3): 515–35. doi:10.1017/S0020818312000185. 

Block, Steven A., and Paul M. Vaaler. 2004. “The Price of Democracy: Sovereign Risk Ratings, 

Bond Spreads and Political Business Cycles in Developing Countries.” Journal of 
International Money and Finance 23 (6): 917–46. doi:10.1016/j.jimonfin.2004.05.001. 

Bodea, Cristina, and Raymond Hicks. 2017. “Sovereign Credit Ratings and Central Banks: Do 
Analysts Pay Attention to Institutions?” SSRN Scholarly Paper ID 2916334. Rochester, 
NY: Social Science Research Network. https://papers.ssrn.com/abstract=2916334. 

 

114 

 

Bueno de Mesquita, Bruce, and George W. Downs. 2005. “Development and Democracy.” 

Foreign Aff. 84: 77. 

Bueno de Mesquita, Bruce, Randolph M. Siverson, James D. Morrow, and Alastair Smith. 2003. 

The Logic of Political Survival. The MIT press.  

Cantor, Richard, and Frank Packer. 1996. “Determinants and Impact of Sovereign Credit 

Ratings.” SSRN Scholarly Paper ID 1028774. Rochester, NY: Social Science Research 
Network. https://papers.ssrn.com/abstract=1028774. 

Cheibub, José Antonio, Jennifer Gandhi, and James Raymond Vreeland. 2010. “Democracy and 
Dictatorship Revisited.” Public Choice 143 (1–2): 67–101. doi:10.1007/s11127-009-9491-
2. 

Cheng, Gong. 2015. “Balance Sheet Effects, Foreign Reserves and Public Policies.” Journal of 

International Money and Finance 59 (December): 146–65. 
doi:10.1016/j.jimonfin.2015.08.005. 

Copelovitch, Mark S., Christopher Gandrud, and Mark Hallerberg. 2015 forthcoming. “Financial 

Regulatory Transparency and Sovereign Borrowing Costs.” SSRN Scholarly Paper ID 
2701852. Rochester, NY: Social Science Research Network. 
https://papers.ssrn.com/abstract=2701852. 

Corneli, Flavia, and Emanuele Tarantino. 2016. “Sovereign Debt and Reserves with Liquidity and 

Productivity Crises.” Journal of International Money and Finance 65 (July): 166–94. 
doi:10.1016/j.jimonfin.2016.02.016. 

DiGiuseppe, Matthew, and Patrick E. Shea. 2015. “Sovereign Credit and the Fate of Leaders: 
Reassessing the ‘Democratic Advantage.’” International Studies Quarterly 59 (3): 557–
70. doi:10.1111/isqu.12181. 

Dincecco, Mark. 2009. “Fiscal Centralization, Limited Government, and Public Revenues in 

Europe, 1650–1913.” The Journal of Economic History 69 (1): 48–103. 
doi:10.1017/S0022050709000345. 

Dominguez, Kathryn M. E., Yuko Hashimoto, and Takatoshi Ito. 2012. “International Reserves 

and the Global Financial Crisis.” Journal of International Economics, NBER Global, 88 (2): 
388–406. doi:10.1016/j.jinteco.2012.03.003. 

Drazen, Allan. 2001. “The Political Business Cycle after 25 Years.” In NBER Macroeconomics 

Annual 2000, Volume 15, 75–138. MIT PRess. 
http://www.nber.org/chapters/c11055.pdf. 

 

115 

 

Dreher, A. 2003. “The Influence of Elections on IMF Programme Interruptions.” Journal of 

Development Studies 39 (6): 101–20. doi:10.1080/00220380312331293597. 

Franzese, Robert J. 2002. Macroeconomic Policies of Developed Democracies. Cambridge 

University Press.  

Gaillard, Norbert. 2011. A Century of Sovereign Ratings. 2012 edition. London: Springer. 

Haggard, Stephan, and Kaufman. 1990. “The Political Economy of Inflation and Stabilization in 

Middle-Income Countries.” WPS444. The World Bank.  

Haggard, Stephan, and Webb, Steven. 1993. “What Do We Know about the Political Economy of 

Economic Policy Reform?” The World Bank Research Observer 8 (2): 143–68. 

Hallerberg, Mark, and Jurgen von Hagen. 1997. “Electoral Institutions, Cabinet Negotiations, 
and Budget Deficits in the European Union.” Working Paper 6341. National Bureau of 
Economic Research. http://www.nber.org/papers/w6341. 

Hollyer, James R., B. Peter Rosendorff, and James Raymond Vreeland. 2014. “Measuring 

Transparency.” Political Analysis, March, mpu001. doi:10.1093/pan/mpu001. 

Jäger, Kai. 2016. “The Role of Regime Type in the Political Economy of Foreign Reserve 
Accumulation.” European Journal of Political Economy 44 (September): 79–96. 
doi:10.1016/j.ejpoleco.2016.05.001. 

Jeanne, Olivier. 2007. “International Reserves in Emerging Market Countries: Too Much of a 

Good Thing?” Brookings Papers on Economic Activity 2007 (1): 1–79. 
doi:10.1353/eca.2007.0013. 

Kolstad, Ivar, and Arne Wiig. 2016. “Does Democracy Reduce Corruption?” Democratization 23 

(7): 1198–1215. doi:10.1080/13510347.2015.1071797. 

Leblang, David, and Shanker Satyanath. 2006. “Institutions, Expectations, and Currency Crises.” 

International Organization 60 (1): 245–62. 

Mosley, Layna. 2003. Global Capital and National Governments. New York: Cambridge 

University Press. 

North, Douglass C., and Barry R. Weingast. 1989. “Constitutions and Commitment: The 

Evolution of Institutions Governing Public Choice in Seventeenth-Century England.” The 
Journal of Economic History 49 (4): 803–32. 

 

116 

 

Nyblade, Benjamin, and Angela O’Mahony. 2014. “Playing with Fire: Pre-Electoral Fiscal 

Manipulation and the Risk of a Speculative Attack.” International Studies Quarterly 58 
(4): 828–38. doi:10.1111/isqu.12168. 

Oatley, Thomas. 2004. “Why Is Stabilization Sometimes Delayed?: Reevaluating the Regime-

Type Hypothesis.” Comparative Political Studies 37 (3): 286–312. 
doi:10.1177/0010414003262072. 

Obstfeld, Maurice, Jay C. Shambaugh, and Alan M. Taylor. 2010. “Financial Stability, the 

Trilemma, and International Reserves.” American Economic Journal: Macroeconomics 2 
(2): 57–94. doi:10.1257/mac.2.2.57. 

Persson, Torsten, and Guido Tabellini. 2005. The Economic Effects of Constitutions. Cambridge, 

Mass.: The MIT Press. 

Reinhart, Carmen M., and Kenneth S. Rogoff. 2009. This Time Is Different: Eight Centuries of 

Financial Folly. Princeton, New Jersey: Princeton University Press. 

Rodrik, Dani. 2006. “The Social Cost of Foreign Exchange Reserves.” International Economic 

Journal 20 (3): 253–66. doi:10.1080/10168730600879331. 

Roubini, Nouriel, and Jeffrey D. Sachs. 1989. “Government Spending and Budget Deficits in the 

Industrial Economies.” SSRN Scholarly Paper ID 459425. Rochester, NY: Social Science 
Research Network. http://papers.ssrn.com/abstract=459425. 

Saiegh, Sebastian M. 2005. “Do Countries Have a ‘Democratic Advantage’?: Political 

Institutions, Multilateral Agencies, and Sovereign Borrowing.” Comparative Political 
Studies 38 (4): 366–87. doi:10.1177/0010414004273204. 

Schultz, Kenneth A., and Barry R. Weingast. 2003. “The Democratic Advantage: Institutional 

Foundations of Financial Power in International Competition.” International 
Organization 57 (1): 3–42. doi:10.1017/S0020818303571065. 

Shi, Min, and Jakob Svensson. 2006. “Political Budget Cycles: Do They Differ across Countries 

and Why?” Journal of Public Economics 90 (8): 1367–1389. 

Steinberg, David A., Karrie J. Koesel, and Nicolas W. Thompson. 2015. “Political Regimes and 

Currency Crises.” Economics & Politics 27 (3): 337–61. doi:10.1111/ecpo.12060. 

Wooldridge, Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data. Second 

edition edition. Cambridge, Mass: The MIT Press. 

 

 

117 

 

 

CHAPTER 3: Dollars and Debt: How Financial Vulnerability Improves the Effectiveness of 

Economic Sanctions 

The literature on the imposition of economic sanctions has generally held that the use of 

sanctions is ineffective in achieving its goals (Wood 2008; Drury and Li 2006; Li and Drury 2004) 

or is especially harmful. In particular, many have argued that sanctions can worsen human 

rights practices (Wood 2008; Peksen 2009; Carneiro and Apolinário Jr 2015) by increasing 

threats to the regime, in addition to destabilizing the economy with banking crises (Hatipoglu 

and Peksen 2016), fomenting the overthrow of leaders (Marinov 2005; Escriba-Folch and 

Wright 2010), and eroding political liberties, press freedoms and the level of democracy 

(Peksen 2010; Peksen and Drury 2010, 2009). Recently, Cerneiro and Apolinário Jr (2015) 

confirmed that even targeted sanctions, which are sometimes thought to be less harmful 

overall than broad-based sanctions, are just as harmful to human rights practices. The literature 

therefore collectively has found that human rights sanctions are detrimental to the precise 

goals they are sometimes designed to achieve, calling into serious question their use as foreign 

policy tools.  

In this paper, I argue that sanctions designed to coerce governments into reducing repression 

can succeed when they are imposed on countries who are financially vulnerable to the 

international economy and thus are at risk of severe losses in the event of an economic shock. 

While previous literature (Peksen and Drury 2009; Peksen 2010) has argued that regime leaders 

are well equipped to maintain the loyalty of regime elites during such cases, I argue that clear 

financial indicators signal to leaders that they will be unable to maintain the political solvency 

of their rule. For instance, in the event that economic sanctions are imposed, they can observe 

 

118 

 

the financial resources of the country and their political allies broadly as a means of 

understanding the likelihood of threats against their government. If a leader senses they face 

little direct economic exposure and that the country has sufficient domestic capital to deal with 

the onset of sanctions, this enables the regime to safely employ repression against threats of 

popular mobilization and protests which arise in such cases (Grauvogel, Licht, and von Soest 

2017). On the other hand, if they are highly exposed to financial losses and have little domestic 

capital, increased repression would run substantial risks of causing a currency attack (Peksen 

and Son 2015), depreciation (Dreger et al. 2016) or banking crises (Hatipoglu and Peksen 2016), 

all of which lead to significant financial losses for elites and even sharper recessions. Given that 

much research has shown that poor economic circumstances can lead to instability (violent or 

otherwise) for the ruling regime (Svolik 2012;  

Svolik 2008; Houle 2009; Reuter and Gandhi 2011; Tang, Huhe, and Zhou 2017), political leaders 

would be highly unlikely to undertake increased repression when they observe it would likely 

only generate a political backlash which undermines their rule. Moreover, under such 

circumstances they would have increased incentives to avoid specific actions which could 

further destabilize the economy, and as a result should be expected to improve human rights 

practices, contrary to the argument from the literature. 

I hypothesize that there are specific financial vulnerability channels which enhance the 

likelihood of sanctions’ success due to the means by which these channels penetrate the ruling 

elite’s wealth and erode the resolve of the regime to withstand the imposition of sanctions. 

These are (a) a country’s level of external debt, (b) it’s level of foreign liabilities (foreign 

financial investment) (c) it’s ratio of short-term debt to total debt (d) current account deficits, 

 

119 

 

which signal a net inflow of foreign finance over exports, and (e) a countries level of foreign 

exchange reserves, which are considered a form of financial insurance against the possibly of 

abrupt stops to financial capital flows (Oatley 2010; Rodrik 2006). Each of these conditions carry 

important implications for how leaders make foreign policy decisions broadly, including political 

calculations regarding the risks and benefits of both internal and international conflict (Shea 

and Poast 2017; Schultz and Weingast 2003; DiGiuseppe 2015; Clay and Digiuseppe 2017; 

DiGiuseppe and Shea 2016). For instance, an early illustration regarding the role of foreign 

exchange reserves is evident with an examination of the lead up to WWI in Germany. In the 

Agadir crisis of 1911, Germany, the aggressor and chief antagonist of this episode, deliberately 

provoked France in an attempt to gain influence over Morocco (Ahamed 2009; Mortimer 1967). 

However, soon after the incident began, Germany encountered a financial panic and was 

suffering considerable losses to their stock of Gold Reserves, jeopardizing their ability to remain 

on the Gold Standard. As a consequence, Germany backed down against France, enduring the 

public embarrassment but stabilizing their financial conditions. Subsequently, bankers at the 

Reichsbank were instructed to amass a substantial quantity of reserves, such that the country 

could sustain itself during a continental war. By the time WWI began, Germany held the largest 

stock of reserves among the major European powers42 (Ahamed 2009). This case demonstrates 

that even amongst major world powers, domestic economic circumstances are a strong input to 

policy-making.  

                                                           

42 According to Ahamed (2009), it was this that helped Germany from suffering much gold losses once WWI 
began, which contrasted sharply with Britain who had fewer reserves and faced larger outflows of liquidity 
due to the credibility problems it fostered. 

 

120 

 

Each of the financial vulnerability conditions I posit are key indicators of financial stress with the 

ability to pressure policymakers in precisely the same way as evidenced by the Germany case. 

Importantly, foreign reserves are also seen as an important stabilizer when other economic 

risks are mounting. For instance, as the level of foreign debt for a country rises, domestic actors 

have greater needs for foreign financing and must also be able to repay their international 

debts, without which companies (even aside from borrowing governments) could be unable to 

borrow from other lenders in the event of a default43 (Corneli and Tarantino 2016). Foreign 

reserves enable continued lending even in the event of a liquidity crisis where external 

financing is scarce, and safeguard countries in the event of sudden outflows of capital (Jung and 

Pyun 2016). Separately, high levels of government borrowing also raise the risk of a debt 

default, but also raise the interest costs of corporate finance because financial risk is linked 

within countries (Gaillard 2011; Kabadayi 2013). Further, in addition the overall levels of debt, 

the term structure of debt finance is also an important indicator of systemic risk. When 

businesses finance much of their debt with short-term financing, for instance, much lending is 

made and accepted based on the presumption that loans will be “rolled over” and refinanced 

rather than repaid in full. Consequently, financial crises or sudden disruptions to financial 

markets leave borrowers without the ability to cover their short term loans (Du and Palia 2018), 

or at least face sharply increased financing costs.  Finally, a countries level of external liabilities 

as well as current account deficits (which show on net more financial inflows) both demonstrate 

to what degree a countries’ economy is dependent on international finance.  Severe disruptions 

to financial flows to countries with high levels of external liabilities would thus be exposed to 

                                                           

43 Or pay substantial rates of interest 

 

121 

 

dramatic financial frictions where, even if replacement capital is locatable it would be on more 

expensive financing terms.  

A consequence of this analysis is to highlight and distinguish the manner in which politicians 

respond to pressure via the international economy. Studies on economic growth or trade are 

undoubtedly an important component of the discussion, and yet it is difficult to discern exactly 

how such issues influence domestic actors and for what purposes. A drop in trade may hurt 

businesses, but disruptions to financial markets potentially impact a much broader range of 

domestic actors regardless of industry, and recent evidence shows that financial crises may lead 

to subsequent reductions in income inequality because it “hits the top of the income ladder” 

(Agnello and Sousa 2012). Elites would also financially suffer more clearly in the event of a 

shock which sparks a currency devaluation or currency attack (Peksen and Son 2015), because 

counties with high levels of external liabilities indicate that debt repayment would be much 

costlier in such cases. Conversely, elites with most of their financial assets abroad would 

actually be insulated from some of these risks. Thus, I argue that the financial vulnerability 

conditions which I discuss likely more closely track elite interests than in other 

conceptualizations. Consequently, leaders under human rights sanctions should be much less 

willingly to prolong the crisis with even worse repressive tactics. Increased repression may only 

induce a steeper recession and lengthen the duration of the sanctions, which would increase 

the likelihood that the regime is overthrown (Svolik 2012; Reuter and Gandhi 2011). The logical 

consequence of this scenario is that political leaders will be more likely to acquiesce and 

improve human rights conditions when sanctions are imposed, precisely because they perceive 

this financial vulnerability as a threat to the stability of the ruling elite. 

 

122 

 

Utilizing data covering 1960 to 2005, which represents the broadest possible sample given the 

data involved, I utilize Bayesian fixed effects and hierarchical models showing strong results for 

the main hypotheses. Whereas most previous studies of economic sanctions utilize PTS or CIRI 

human rights data to study the influence of sanctions on human rights, in this paper I utilize the 

‘Physical Violence Index’ from the Varieties of Democracy database in order to study the 

relation between sanctions and human rights. This is a worthwhile advance because VDEM was 

coded with a greater number of experts, covers a substantially longer time span, and offers 

much greater variability due to its estimation via Bayesian factor analysis. While the main 

analysis is on the influence of the vulnerability index (FVI), I also examine the influence of the 

subcomponents as well as robustness to utilizing the Fariss measure of respect for human rights 

(Fariss 2018a). 

The paper thus makes a strong contribution to the literature on economic sanctions by 

demonstrating how sanctions can, in fact, have their desired effect of coercing positive changes 

in human rights practices and reducing the level of repression. Recently, Peterson (2014) has 

shown that sanctions can have positive effects on external actors’ human rights practices, but 

here I demonstrate this positive influence even on intended target states for a broad class of 

governments. Further, by demonstrating these precise financial channels of sanctions’ 

influence, it has the potential to update our understanding of which precise tools of economic 

statecraft can maximize the ability of western governments to gain leverage over and positively 

influence harmful global actors. It also demonstrates how allowing, rather than constraining, 

the flows of international investment to begrudged nations can actually be beneficial in the 

long-run and maximize the potential for coercive foreign policy. In this way, this study offers 

 

123 

 

insights into the importance of meaningful international rewards for cases in which foreign 

actors demonstrate a positive response to international sanctions demands.  

The Influence of Economic Sanctions 

The contemporary literature on economic sanctions has to some degree debated how targeted 

governments respond to perceived threats to the regime. This is an important point of 

emphasis since I argue that financial vulnerability conditions how governments respond to 

perceived threats to the regime. After a brief overview of this research, in the next section I will 

elaborate on how my theory of financial vulnerability relates to government incentives’ to 

acquiesce or repress in the face of sanctions. 

Kaempfer , Lowenberg , and Mertens (2004) and Marinov (2005) are notable cases  by arguing 

that sanctions may be successful when it successfully threatens the stability of the governing 

coalition, suggesting that it is exactly when target leaders sense that their survival is at stake 

when they are likely to compromise and offer concessions. In what Peksen (2009) refers to as 

the naïve economic model of sanctions, human rights practices improve because it reduces the 

repressive capacity of the state with a direct loss of such financial resources, but importantly 

also reduces the loyalty premium to elites because the regime has less resources to distribute 

rents and loyalty benefits (Wood 2008; Peksen 2009). As a consequence, the target state should 

perceive their vulnerability (Mirinov 2005) and improve their human rights practices. 

This logic has not received much support from the literature, however, both on theoretical 

grounds and with more precise empirical testing. Wood (2008), for instance, argues that by 

almost the same logic that one might argue perceived vulnerability leads to improved human 

rights conditions, perceived vulnerability should lead to an erosion of human rights standards. 

 

124 

 

Because sanctions impose a loss of financial resources on the regime, this increases the chances 

of defection from the regime, which then increases the incentives to repress in an effort to 

forestall broader defections and stifle domestic protest. Peksen (2009) further argues a 

skeptical view, emphasizing that targeted regimes are able to consolidate their control over the 

distribution of resources in the wake of sanctions imposition, deepening their ties with 

supporters and alienating opponents. Based on this reasoning, Peksen and Drury (2009, 2010) 

argue that economic sanctions reduce a countries’ level of democracy and political rights, 

noting that “economic coercion barely hams the economic and political capacity of the targeted 

regimes.” Additionally, they argue that sanctions do not harm the political elites but worsen 

perceived popular grievances (due to unemployment and health outcomes), and that it is 

therefore necessary for the regime to repress in order to stifle popular mobilization and the 

potential for any incipient political opposition. In other work, Peksen (2010) has shown how 

sanctions also reduce media openness due to these changes in political power. 

More recently, Carneiro and Apolinário Jr (2016) have drawn upon an emphasis for the use of 

limited or targeted economic sanctions in an effort to more precisely coerce governments 

without all of the attendant consequences to human security. They find, however, that even 

targeted sanctions are likely to worsen human rights practices, thus leaving policymakers with 

the original dilemma regarding how to use foreign policy to coerce foreign actors. A 

consequence of this collective body of research is that, in general, the expectation is that 

sanctions do not sufficiently threaten the governments ability to maintain support while 

contributing to a broader deterioration in the governments’ security environment. A recent 

exception to this is Peksen (2017), who argues that differences in autocratic regime types 

 

125 

 

involve different responses to the threat of sanctions. Because personalist dictators tend to rely 

more on foreign aid and resource rents, have poorer economic fundamentals, and lack strong 

institutional governance structures which enable the efficient collection of tax revenues, he 

argues that personalist regimes lack the resources necessary to survive sanctions episodes. As a 

consequence, they are more likely than single-party authoritarian governments to acquiesce in 

the face of sanctions, although they do not differ compared with democracies. As I will discuss 

in the following section, this has implications for how we conceive of the role of governments 

with financially vulnerable economies in terms of their response to sanctions episodes. 

Finally, Peterson (2014) represents a notable advance in the literature, in that he examines the 

effectiveness of economic sanctions against external actors. Thus, he argues and provides 

compelling evidence that sanctions lead to improvements in human rights practices against 

other countries who qualify as potential sanctions targets (such as countries with similar 

repression histories). Others emphasize how international economic issues such as trade 

linkages are important factors for sanctions success. For instance, McLean and Whang (2010) 

argue that, since globalization makes the success of sanctions more difficult (Elliott 1998), 

cooperation from target states major trade partners is important for the success of sanctions. 

In related work, Early (2009, 2011) and Early and Spice (2015) examine what determines 

whether countries utilize economic ties to sanctioned governments as a means of “busting” the 

sanctions. From these works we gain important insights into how the global system responds to 

sanctions events and how these international dimensions influence the efficacy of sanctions as 

a policy tool. For our present purposes, it is notable that while trade is thus an important 

dimension to the analysis of economic sanctions, considerations of financial vulnerability 

 

126 

 

arguably have much clearer implications for the success of sanctions in general. I turn to a 

discussion of these issues in the next section. 

Sanctions and the Influence of Financial Vulnerability 

In the previous section, we see that whether or not sanctions actually lead governments to 

increased repression hinges on consideration of whether the government actually perceives a 

viable threat to their regime. Kaempfer et al (2004) and Marinov (2005) argue for cases where 

the regime does feel threatened, since the loss of resources threatens their ability to maintain 

the loyalty to the regime. Most compellingly, Peksen and Drury (2009, 2010) argue that most 

governments feel little in the way of this type of pressure, because they are able to utilize 

increased monopoly power of economic resources as political levers of loyalty. This takes place 

amidst a broader deterioration in the security climate, however, and thus the government has 

increased incentive to repress all the while protecting the elites on whom they rely on for 

support. The key matter here is in understanding whether the government can sufficiently 

insulate their supporters from the deterioration to the macroeconomy, which involves typical 

recessions of around 3.3% of GDP and an inflation rate of 37% (Peksen 2017), with lasting 

negative growth effects up to a decade in length (Neuenkirch and Neumeier 2015). Indeed, 

Peksen (2017) argues that personalist regimes fit exactly the case in which they lack the ability 

to insulate regime supporters from the associated economic costs, and thus sanctions against 

personalist governments are more likely to succeed. 

I argue that conditions of international financial vulnerability suggest that governments will be 

less able to protect domestic elites from the harmful economic fallout of sanctions, and thus 

will be more likely to respond positively to being sanctioned. Wood (2008) suggests that 

 

127 

 

sanctioned governments redistribute resources “in the hope of weathering sanctions”, and 

Peksen (2017) that personalist regimes don’t have the “capacity to weather” such costs. I argue 

that in financially vulnerable countries, leaders have an acute understanding that they will not 

be able to successfully protect domestic elites and weather the economic storm of sanctions. 

Given that they understand the economic costs to wealthy elites will be particularly severe, 

they have little incentive to engage in repression because the onset of repression or related 

political violence is likely to only exacerbate economic conditions (Blomberg, Hess, and 

Orphanides 2004) by simply extending the duration of sanctions, by increasing the intensity of 

imposed sanctions, or generating other economic costs such as capital flight. In such a case, the 

leader essentially is faced with a high elasticity of loyalty from the winning coalition. An 

increase in repression, given the additional economic costs it entails for them, could provoke a 

large backlash in regime loyalty and materially threaten the leaders’ survival. Thus, when a 

country is financially vulnerable given its exposure to the world economy, leaders are much 

more likely to respond to sanctions by improving their respect for human rights. 

Peksen and Drury (2009, 2010) discuss how leaders refrain from improving human rights 

practices at least in part because they do not want to be perceived by domestic actors as weak. 

Indeed, this is sensible when the leader and his winning coalition is secure. Peksen (2017) notes 

that personalist regimes would of course desire to repress, but when they sense that defying 

sanctions will only erode their political support base, they acquiesce to the sanctions demands. 

Given the perception of instability, therefore, leaders would rather appear weak but able to 

preserve the wealth of their winning coalition, which necessitates acquiescence. This was 

demonstrated by Germany in the Agadir crisis, mentioned earlier, suggesting that such 

 

128 

 

concerns are indeed secondary to the most important goal of maintaining the loyalty from 

regimes elites.  

The literature explicitly examining whether political leaders are punished for poor economic 

performance further reinforces the importance that financial vulnerability has on the political 

calculations of leaders. If leaders were rarely ever punished during times of economic stress, 

then arguably it would not matter whether the country was “financially vulnerable” or not, as it 

would remove the key source of political pressure which I argue applies. However, the 

literature generally finds that such political pressure does exist. Speaking broadly to this issue, 

(Svolik 2008) shows that there are important long and short run influences of growth on 

political stability. Economic development is an extremely strong predictor that democracies 

consolidate, which we can take as evidence that long-run growth reduces the political tensions 

which can foster political breakdowns; but he also argues that the timing of reversals to 

autocracy is strongly predicted by the association of economic recessions. Elsewhere, Svolik 

(2012) finds that economic development also reduces the likelihood of coups.  

Reuter and Gandhi (2011) also find strong evidence which is consistent with these expectations 

regarding the importance of economic conditions for the viability of the political regime. For 

instance, they argue that poor economic growth leads to increased episodes of elite defection 

from the ruling regime in hegemonic party systems. They argue that poor economic 

performance deprives the regime of resources to continually provide benefits and maintain 

cooptation policies, but it also provides political opponents with windows of opportunity to 

argue that they are better able to deliver stable growth policies than the government. 

Interestingly, they reference important case examples which reveal some of the key financial 

 

129 

 

vulnerability mechanisms I posit, such as in Turkey in the 1950’s where a massive build-up in 

foreign debt along with a decline in the countries gold and foreign exchange reserves was 

putting a significant strain on the economy.  

Tang, Huhe, and Zhou (2017) undertake a thematically very similar study by showing that 

“economic crises” lead to democratic transitions only when the government plays a large and 

active role in the economy, which is thus evocative of the Reuter and Gandhi (2011) study since 

this is arguably precisely the role played by hegemonic parties’. They specifically state that 

“authoritarian rule built on material incentives is likely to collapse when dictators’ can no 

longer provide sufficient resources to sustain pollical support as a result of the crisis.” 

Contrasting Tang et al (2017), in the context of this study on economic sanctions I do not argue 

that the history of government involvement in the economy is explicitly necessary, since 

financially vulnerable countries will still face the same incentives to cooperate regardless. 

However, their study does highlight the key issue involved which is that failure to respond 

appropriately during times of economic stress can signal the incipient failure of the regime 

itself.  

We can therefore distill from these literatures some important points. First, the key to 

understanding whether human rights sanctions only lead to increased repression is contingent 

on the leaders’ understanding and ability to insulate their supporters from the economic 

distress which the sanctions creates. This is especially important since repression itself is likely 

to exacerbate such economic conditions. Further, since the literature has found convincing 

evidence that economic weakness in the form of either slow growth, recessions or general 

economic crises causes political instability for the regime, it suggests that political leaders under 

 

130 

 

economic sanctions would be highly sensitive to domestic economic conditions and their 

perceived ability to weather such financial storms. As a consequence, sanctioned’ leaders 

should theoretically pay close attention to how vulnerable their economy is to such economic 

shocks and, when the level of financial vulnerability is high, should be much more willing to 

acquiesce to demands over human rights’ practices because failure to do so may only hasten 

their political demise.  

Financial Vulnerability Conditions 

My above discussion therefore illustrates how sanctioned regimes have different incentives to 

repress depending on their level of financial vulnerability. When their elites are well insulated 

from economic risk, governments retain the desire to repress due to the imposition of 

sanctions, much as Peksen and Drury have argued. However, when they are financially 

vulnerable, they understand that resisting sanctions and using repression are perilous, self-

defeating exercises. I argue that there are several conditions which demonstrate how elites are 

insulated or exposed to financial risk during an episode of economic sanctions. First, a key 

measure of financial vulnerability is a countries level of external liabilities. This measure is an 

indication of what percent of the economy is indebted to foreign actors, and thus demonstrates 

to what degree the economy will be harmed by a negative economic shock such as the 

imposition of sanctions. It represents, in other words, the degree to which the domestic 

economy is reliant on foreign sources of financial capital. An abrupt economic shock would thus 

be very harmful, because being literally cut off from financial flows due to sanctions implies a 

dramatic increase in the cost of business finance (Alfaro, Chari, and Kanczuk 2017; Alfaro, 

 

131 

 

Kalemli-Ozcan, and Volosovych 2007), and for highly indebted countries this directly indicates 

increased risk of defaults and debt crises (Catão and Milesi-Ferretti 2014).  

It is also important because, in the event that economic sanctions lead to currency 

depreciations (Dreger et al. 2016), it also implies that the cost of foreign indebtedness will rise 

in the amount of the currency depreciation. Peksen and Son (2015) find that sanctions lead to 

the onset of currency crises, which further demonstrates the financial risks to domestic actors 

in the wake of a sanctions episode, especially, I argue, when the level of external liabilities is 

high. For instance, a high level of liabilities indicates foreign investors willingness to take on the 

risks involved in doing business in/with the target country (at least prior to the onset of 

sanctions). A large component of this risk is currency risk, and large depreciations of the 

national currency would mean that countries with high levels of external liabilities suddenly are 

dramatically more exposed to financial risk and liquidity crises because not only is risk itself 

higher because of political turmoil, but the depreciated currency makes it even costlier to repay 

international debts. 

Additionally, Trade flows and trade openness do not clearly capture this specific dynamic of 

economic loss, because trade flows can more easily be offset or replaced by purchases or 

transactions with businesses from other countries (McLean and Whang 2010). External 

liabilities, as here discussed, involve a contracting type of relationship which raise the level of 

frictions involved given a shock to these economic relationships. Catão and Milesi-Ferretti 

(2014)  highlight this vulnerability channel, demonstrating that a countries’ level of external 

liabilities is a very strong predictor of economic crises. While it was not part of their study, it 

stands to reason that a high level of liabilities will exacerbate the costs of a different economic 

 

132 

 

shock (sanctions), as discussed above. Further, the exceptional volume of these liabilities 

represent risk to the wealthiest actors in society, since typical individuals themselves do not 

take out foreign-denominated loans, receive equity stakes, or incur foreign exchange risk. I 

therefore argue that this measure more clearly taps into the financial incentives of the ruling 

class than other considerations of the international economy, and leads to the first hypothesis 

of the paper: 

H1: Human Rights Sanctions will lead to lower levels of repression as the level of 

external liabilities increases 

A related measure of vulnerability, similar to external liabilities, is a countries’ current account 

deficit, which is a net approximation of foreign financial flows. While overall financial exposure 

is captured by the stock of external liabilities, the current account balance (deficit) can capture 

more immediate movements in exposure and thus help to identify how political actors would 

relate to the political risk of economic shocks. Beyond this, it also reveals important signals 

about the exposure to economic shocks. Teixeira et al (2018), for instance, find that during 

periods of economic crisis current account deficits lead to lower sovereign credit ratings, 

arguably because its indication of reliance on external creditors implies long-run 

unsustainability. Additionally, Davis et al (2016) show that current account balances interact 

with levels of externally financed debt to significantly increase the probability of banking crises. 

This also resonates with recent experience in the US, since the current account balance has 

been much discussed for its potential role in the build-up of systemic risk prior to the US 

 

133 

 

financial crisis of 200844. Current Account Deficits therefore track a distinct dimension of 

financial vulnerability and this leads to the second hypothesis of the paper: 

H2: Human Rights Sanctions will lead to lower levels of repression as current account 

deficits increase 

Another important component of financial vulnerability considers a countries level of foreign 

exchange reserves, which is considered to be a form of financial insurance against the potential 

for negative international shocks (Rodrik 2006; Obstfeld, Shambaugh, and Taylor 2010; Jeanne 

2007). Foreign reserves can be used in order to guard against currency exchange risk, and can 

act as emergency financing when international shocks stop the flow of funds and domestic 

banks face increased risk of default (Corneli and Tarantino 2016). Dominguez, Hashimoto, and 

Ito (2012), for instance, found that countries with substantial accumulated foreign reserves 

prior to the 2008 financial crisis experienced higher ensuing growth and recovery rates. 

Because foreign reserves are so crucial to stabilize economies against sudden shocks, it thus 

represents a clear and important indicator of how vulnerable a country would be to the 

imposition of economic sanctions. It is worth noting that, in this case, a large level of reserves 

insulates a country from economic risk, whereas the other vulnerability conditions exacerbate 

risk. It is therefore the only hypothesis which predicts an increase in repression due to 

sanctions, rather than a decrease. Thus, we can state the third hypothesis: 

H3: Human Rights Sanctions will lead to higher levels of repression as the level of 

foreign exchange reserves increases 

                                                           

44See, for instance, a prominent speech by Fed chairman Ben Bernanke 
https://www.federalreserve.gov/boarddocs/speeches/2005/200503102/ 

 

134 

 

Another measure of financial vulnerability is a countries level of external debt. This is important 

for a similar reason regarding why the level of liabilities can indicate how dramatic an economic 

shock may be: currency risk. If a country has a large volume of external debt, then a sharp 

depreciation induced by sanctions can suddenly spark a significant increase in de facto debt 

costs, leading to higher risk of debt defaults and debt-finance costs for all business (Gaillard 

2011; Ahlquist 2006). Corneli and Tarentino (2016) argue, for instance, that high external debt 

significantly increases the likelihood of a liquidity shock, which is notable because even if it did 

not have this effect directly, the imposition of sanctions may easily cause it to be so. 

Additionally, the term-structure of debt is also considered to be an important indicator of 

financial risk. For instance a number of works have demonstrated that when debt is largely 

financed via short-term loan terms, sudden shocks to financial capital can exacerbate crises 

because businesses are then unable to “rollover” their debts and thus adapt by curtailing their 

financial activities. This leads to increased risk of crises (Ekici and Nemlioğlu 2017; Benmelech 

and Dvir 2013), in addition to significantly higher lending costs as the ratio of short term debt 

increases (Wang, Chiu, and King 2017). For governments, this means that default risk is higher 

and in the event of economic difficulties, raising necessary funds becomes especially difficult. 

Consequently, high levels of short-term debt would make it more difficult to survive the onset 

of economic sanctions.  Combined, these debt conditions regarding overall external debt and 

the its related term structure portends significant costs for the economy. It thus leads to the 

final hypotheses of the paper: 

H4: Human Rights Sanctions will lead to lower levels of repression as the levels of 

external debt increases 

 

135 

 

H5: Human Rights Sanctions lead to lower levels of repression as ratio of short-term to 

total debt increases. 

I have now stated the relevant financial vulnerability conditions which indicate the degree to 

which sanctioned governments perceive they can insulate their support base from economic 

costs. While the literature argues that leaders can try to use other policies like tax incentives or 

subsidies to offset the economic harm of sanctions, to the extent that each of these financial 

vulnerability conditions is binding for an economy it implies a very severe constraint on the 

leaders’ ability to successfully offset this increased vulnerability. Further, as each of these 

conditions are having an influence (rather than considering these in pure isolation), it further 

complicates the ability of the government to be able to deploy sufficient resources to offset the 

combination of these costs. This would especially be the case during a recession, when 

government resources dwindle via automatic stabilizers.  

An important issue is that the concept of financial vulnerability which I have outlined above is 

such a multifaceted issue that it is better represented by an index which accounts for all of 

these vulnerability dimensions simultaneously. Examining each issue in isolation could easily 

lead to obscure inferences because other aspects of the international economy may 

counterbalance it. This is even implied by the theory itself, since I posit that foreign exchange 

reserves represents financial insurance which enables governments to insulate their supporters 

from the financial costs of sanctions. Thus, high levels of liabilities may increase vulnerability, 

but ignoring the levels of reserves may not give an accurate picture. Because of this fact, there 

is tremendous advantage to combining these measures into a unified index describing the 

aggregate level of financial vulnerability for a country at a specific point in time. Thus, one main 

 

136 

 

contribution of the paper is to create an Index describing financial vulnerability, the FVI Index, 

and use it to examine in more straightforward fashion the impact of financial vulnerability on a 

countries susceptibility to economic coercion. Therefore, I create an overall financial 

vulnerability index (FVI), an examine the broadest overarching hypothesis of the paper: 

H6: Human Rights Sanctions lead to lower levels of repression as the overall level of 

financial vulnerability increases. 

Research Design and Data 

I utilize the Threat and Imposition of Economic Sanctions (TIES) database version 4. Similar to 

Peterson (2014) I transform this data, which is based on sanctions episodes as the unit of 

analysis, into country-year format. To examine the impact of sanctions, I mainly utilize 

sanctions which were imposed principally due to concerns regarding human rights abuses and 

are imposed by western democracies or the EU. For data on government repression, I utilize 

the Varieties of Democracy version 6.2 measure “Physical Violence Index”, which is created by 

Bayesian factor analysis45. I use this measure, rather than more traditional measures of human 

rights abuses such as PTS or CIRI, mainly because it is arguably a more robust and precisely 

estimated measure which offers greater variability and has a longer time span, permitting the 

use of all years of the TIES data conditional on the covariates. It combines measurement on the 

use of political killings by government agents in addition to government sanctioned torture. In 

the analysis, I use as the dependent variable the future 5 year average of Physical Violence, in 

order to show persistence in the changes of political repression. As a robustness measure, I also 

utilize Christopher Fariss’ measure of human rights practices, which accounts for changes in the  

                                                           

45 This measure is described with more detail on page 74 of codebook 7. 

 

137 

 

reporting of human rights conditions over time (Fariss 2018b). Because I present both measures 

in the main output tables, it should be noted that these measures predict opposite effects: one 

is a measure of repression (VDEM Violence) whereas the Fariss measure is an estimate of 

respect for human rights.  

For the financial vulnerability conditions, the first measure of vulnerability, the level of external 

liabilities to GDP ratio, comes from the external wealth of nations46 database (Lane and Milesi-

Ferretti 2006). The rest of the financial vulnerability measures – foreign reserves/Debt, External 

Debt/GNI, Short-term Debt/Total Debt, and Current Account Deficits– are acquired from the 

World Bank’s WDI database. The measures of total liabilities, foreign reserves and external debt 

are logged because the distributions of these variables are skewed, whereas the other 

measures are fairly symmetric. 

To create the FVI Index (Financial Vulnerability), I normalize each of the financial vulnerability  

measures and then add them together with equal weight47, multiplied by 100. This creates an 

index which theoretically ranges from 0 to 100, although in practice the maximum is 

approximately a value of 71. 

As controls, I use economic growth from the World Bank, in addition to the Geddes regime 

typology measure from the GWF database. From the Varieties of Democracy database I use the 

electoral democracy index as a democracy indicator, and from VDEM I also use a measure of 

International War based on COW criteria. A dummy variable for civil wars and separately oil 

                                                           

46 https://www.imf.org/en/Publications/WP/Issues/2016/12/31/The-External-Wealth-of-Nations-Mark-II-
Revised-and-Extended-Estimates-of-Foreign-Assets-and-18942 
47 I invert the foreign reserves variable, so that the direction of the relationship between reserves and 
vulnerability is the same as the other variables in the Index. 

 

138 

 

wealth comes from Bodea, Elbadawi, and Houle (2017). To capture additional incentives for 

repression I also utilize the Banks’ measure of “violent dissent”, acquired from Clay and 

DiGiuseppe (2017) replication data.  

Because the literature on economic sanctions has been concerned with the potential for 

unmodeled omitted-variables bias as a confounder when modeling repression policies, and 

because the expected impact of human rights sanctions partly implies a within country effect, I 

mainly utilize Bayesian Fixed Effects models to examine these relationships. I do, however, also 

estimate several random coefficient, random intercept models and show that the relationships 

hold with either specification. All models used a burn-in period of 40,000 simulations and 

100,000 simulations for inference, and use standard diffuse and uninformative priors on all 

parameters. Additionally, while I do not present these in the main text, I performed standard 

visual and statistical tests of MCMC convergence to assess the validity of producing summary 

estimates from the MCMC chains. For all models reported in this paper, the MCMC diagnostic 

criteria support the inference that the chains converged48. Because I posit that financial 

vulnerability conditions the effect of sanctions on human rights practices, each of the models 

takes the form of an interaction effect between human rights sanctions and the financial 

vulnerability condition. I present output tables of these effects, but the main interpretation will 

rely on marginal effects plots, all of which report 95% Bayesian credible intervals. The 

maximum time sample ranges from 1961 to 2011, though some of the external wealth of 

nations measures only begin in 1970. All of the models of the paper exclude Western Europe 

                                                           

48 This is more precisely stated by noting that I failed to find evidence of non-convergence, since convergence 
of MCMC chains can never be proven. 

 

139 

 

and North America from the sample, since these countries are the originators of human rights 

sanctions and their inclusion in the sample only biases the estimated effects. 

The concern over endogeneity is an issue which has arisen in the literature on economic 

sanctions, as is demonstrated in Peksen (2017), though it does not have a large impact on the 

study in his case. This is because a concern over endogeneity in the case of how sanctions 

influence human rights practices in this case should lead to conservative estimates, meaning 

that it shouldn’t be expected to bias the estimates in favor of my theory. For instance, 

Nooruddin (2002) argues that selection bias is a reason why some have had difficulty finding a 

positive effect of economic sanctions. In essence, the argument is that countries who are willing 

to be coerced do not allow themselves to reach the sanctions stage, and so this tool of foreign 

policy may be effective even while most researchers have not found a positive effect. Thus, 

actually finding a positive influence of sanctions is against expectations, according to this 

concern.  

Results 

First, I present the evidence regarding all of the hypotheses regarding the FVI subcomponents 

which correspond to hypotheses 1:5. Table 3.1 reports the output of Bayesian linear fixed 

effects models on the future 5 year average of the Physical Violence Index (repression). These 

estimates report the summaries of the posterior density estimates of the parameters, with the 

mean and standard deviations. I also report the probability that a coefficient estimate is 

greater/less than 0, depending on the direction of the estimate. In this context, the probability 

is simply the number of simulated values which lie to the left or right of zero, which 

approximates inference from a frequentist perspective. Thus, these display the estimated 

 

140 

 

coefficients of the estimated interaction with each FVI condition and human rights sanctions. So 

far the results are strongly supportive of theoretical expectations. Table 3.1 reports that all but 

one measure (short term debt) exhibit significant relationships and lie in the expected 

direction. Foreign Reserves and CA Deficits both had fully 100% of posterior simulated draws on 

the correct side of 0, while Liabilities and External Debt had 97% of simulated draws on the 

correct side. Short term debt displays a mean estimate of essentially zero and only 54% of 

draws were on the theoretically correct side. This is not too surprising in the sense that the 

short term nature of debt may not matter considering other important factors. For instance, if 

overall debt is low, then arguably the importance of the short-term debt ratio is dubious. While 

I do not report the estimate in the table, I further explore the role of short-term debt by then 

estimating a triple interaction model of [HR Sanctions * Short-term Debt * External Debt]. This 

would be a sensible estimate because the term-structure of debt should clearly rise in 

importance with the level of overall debt. This triple interaction seems to reveal what we would 

expect to find based on theoretical expectations. The coefficient is negative and 97% of 

simulated posterior draws are on the correct side. 

Thus, we see that the imposition of sanctions leads to reduced repression over 5 years when 

Liabilities, External Debt, and CA Deficits are high, and these estimates clearly show that the 

preponderance of these estimates are below zero. Conversely, we also see that sanctions 

imposition when the level of foreign reserves is high increases the level of repression. This 

makes sense based on theoretical expectations, because these resources allow the government 

to try and stabilize the economy and lend to major financial institutions in an effort to weather 

the sanctions episode. In effect, this allows them to insulate their supporters from the negative  

 

141 

 

 

142 

MeanSDPrMeanSDPrMeanSDPr(Intercept)4.4320.548100%(Intercept)5.1580.646100%(Intercept)4.460.807100%(lag) Physical Violence0.5880.014100%(lag) Physical Violence0.5940.016100%(lag) Physical Violence0.6030.017100%HR Sanctions-0.0220.01198%HR Sanctions-0.0090.0267%HR Sanctions0.0740.05392%CA Deficits0094%Short Term (ST) Debt-0.0010100%(log) Ext Debt-0.0070.00398%Sanctions * CA Deficits-0.0040.001100%Sanctions * ST Debt00.00154%Sanctions * Debt-0.0270.01597%Democracy-0.2570.019100%Democracy-0.2120.022100%Democracy-0.1960.022100%International War0.0230.008100%International War0.0320.009100%International War0.020.0198%Civil War0.0050.00679%Civil War0.0010.00758%Civil War00.00752%(log) GDP pc0.0010.00456%(log) GDP pc-0.0020.00567%(log) GDP pc-0.0020.00564%Growth Rate0054%Growth Rate0093%Growth Rate0085%Time-0.0020100%Time-0.0020100%Time-0.0020100%N = 4,662N = 3,094N = 2,606# of Countries = 134# of Countries = 93# of Countries = 92MeanSDPrMeanSDPr(Intercept)5.9950.813100%(Intercept)5.4870.745100%(lag) Physical Violence0.5790.016100%(lag) Physical Violence0.5890.018100%HR Sanctions0.1380.06898%HR Sanctions-0.1320.043100%(log) Liabilities/GDP0.0060.00490%(log) Reserves/Debt00.00353%Sanctions * Liabilities-0.0340.01897%Sanctions * Reserves0.0370.013100%Democracy-0.2280.022100%Democracy-0.2060.024100%International War0.0310.01100%International War0.0210.01197%Civil War-0.0020.00759%Civil War0.0010.00758%(log) GDP pc-0.0020.00563%(log) GDP pc0.0010.00660%Growth Rate0097%Growth Rate0092%Time-0.0030100%Time-0.0030100%N = 3,937N = 2,323# of Countries = 132# of Countries = 84Table 3.1: Effect of Sanctions on Government Repression (FVI Subcomponents)CA DeficitsShort Term (ST) DebtExternal Debt/GDPReserves/DebtLiabilities/GDPDependent variable is the future 5 year average of the (VDEM) Physical Violence Index. Models estimated are Bayesian linear models with country fixed effects 

 

 

 

 

 

143 

MeanSDMeanSDMeanSDMeanSD(Intercept)6.040.475(Intercept)13.3371.497(Intercept)-22.7874.863(Intercept)2.3780.049(lag) Physical Violence0.6980.013(lag) Physical Violence0.4450.024(lag) Human Rights0.630.019(lag) Physical Violence0.3120.004HR Sanctions0.2920.066HR Sanctions0.4770.119HR Sanctions-0.9610.393HR Sanctions0.4080.012FVI0.0000.000FVI0.0010.001FVI-0.0120.002FVI0.0000.000Sanctions * FVI-0.0090.002Sanctions * FVI-0.0130.003Sanctions * FVI0.0290.01Sanctions * FVI-0.0120.000Time-0.0030.000(log) Population0.140.039(log) Population-0.2970.128Democracy-0.3640.003Democracy-0.2490.031Democracy0.2480.08(log) Population-0.0980.003Personalist Regime0.0160.012Personalist Regime0.0990.04Personalist Regime0.0710.001Party Regime-0.0280.011Party Regime0.0410.037Party Regime0.0160.001(log) GDP pc0.030.008(log) GDP pc-0.0170.028(log) GDP pc-0.0130.001Growth Rate0.0000.000Growth Rate0.0020.001Growth Rate-0.0010.000International War0.0150.013International War-0.0080.043International War0.0290.001Civil War0.0110.008Civil War0.0060.028Civil War0.0320.001Violent Dissent-0.0010.002Violent Dissent-0.0230.006Violent Dissent0.0000.000Oil Wealth-0.0360.009Oil Wealth0.0730.03Oil Wealth-0.0350.001Time-0.0080.001Time0.0140.003N = 2,515N = 1,611N = 1,611N = 1,611# of Countries = 101# of Countries = 80# of Countries = 80# of Countries = 80Table 3.2: FVI Index and the Impact of Government Repression(A) FVI Basic(B) FVI Full Model(C) FVI - Fariss Measure(D) HLM Model 

effect of sanctions, which would indicate their freedom to increase repression in order to crack 

down on dissident formation. Additionally, exploration of the range of the interactions reveals 

that that the effects have meaningful effects at various ranges of data. 

Next, Table 3.2 presents the results from several different specifications. To conserve space in 

the table, I omitted the column for the probability of a positive/negative effect of the 

coefficient and simply bolded all measures estimated to have at least a 90% probability of being 

positive or negative. I note, however, that the choice of 90% probability does not influence any 

of the interaction effects, all of which had 100% of simulated posterior draws below zero. In an 

effort to maximize the sample size as well as show that the results are not driven by overfitting 

bias, I estimated a basic fixed effects model wherein the only included measures were the 

interaction of sanctions and FVI, the lagged level of repression, and a time trend. Model 3.2 (B) 

shows these same results but expanded to include a broader range of control variables, and 

Model 3.2 (C) shows the estimated effects of the future 5 year average of the Fariss Human 

Rights measure. Across these models we see evidence consistent with theoretical expectations, 

that sanctions lead to reduced levels of repression for countries which are financially 

vulnerable. We should note that in model 3.2 (C), using the Fariss measure, this estimated 

interaction effect is positive, as expected. Because the Fariss measure predicts respect for 

human rights, it shows the inverse relationship as that indicated by the VDEM index of Physical 

Violence. Thus the opposite sign is anticipated. Model 3.2 (D) shows the results of the Bayesian 

random coefficient, random intercept model, for which I allowed the estimated impact of FVI to 

vary across countries. Here we see estimates consistent with the other estimations.  

 

144 

 

Finally, inferences are best made by examination of the marginal effects plots. Figure 3.1 shows 

the marginal effects produced by simulating the range of the interaction from the hierarchical 

model in Table 3.2 (D), although marginal effects plots from all models are highly similar. Thus, 

these are the full models but estimating for both VDEM Repression and then in the second plot 

for Respect for Human Rights practices. We see that sanctions have much different impacts 

depending on the level of financial vulnerability. When FVI is low and countries are financially 

secure, then sanctions increase the future level of repression significantly, in keeping with the 

literatures previous findings. However, as FVI increases approximately past its mean value, the 

future level of repression declines significantly. Conversely in the Fariss plot (B), we don’t find 

strong evidence that sanctions reduce future levels of human rights practices when FVI is low, 

 

145 

 

but we see that as FVI increases sanctions significantly improve the level of respect for human 

rights. 

Consequently, we can see that the sub-measures of vulnerability seem to have a strong 

mediating influence on the impact of human rights sanctions on government repression, 

although in some cases it depends on context. We recall that short-term debt is only significant 

in a triple interaction with the level of debt, and the marginal effects plot for liabilities does not 

appear strongly supportive although it appears very close to conventional levels of significance 

(and, were we to reduce the credible intervals to 90% widths, it would be so). However, 

considering the multidimensional nature of overall risk the best analysis resides from the FVI 

index itself. In considering the FVI index we find the strongest evidence and is robust to 

multiple model specifications, and appears to have a strong impact across a range real world 

values in the data.  

Conclusion 

The literature on the use of economic sanctions has always largely held that these sanctions 

negatively influence human rights abuses, in addition to a range of undesirable features such as 

eroding democracy and political rights, media freedoms, and potentially harmful social costs. 

Recently, it has been shown that human rights sanctions can have positive effects on human 

rights practices but for external actors. This paper thus makes an important contribution to the 

literature by demonstrating how sanctions can meaningfully lead to positive human rights 

improvements even for sanctioned countries. By examining specific cases of vulnerability and 

then developing an index of this vulnerability, we can more clearly discern how leaders fear 

they cannot sustain the costs of sanctions when their economic is financially vulnerable. This 

 

146 

 

also leads us to an improved understanding of the specific tools which may be more useful in 

deploying sanctions in ways that lead to the positive outcomes sanctions are designed to 

impose. 

 

 

 

 

 

 

147 

 

 

BIBLIOGRAPHY 

148 

 

BIBLIOGRAPHY 

 

Agnello, Luca, and Ricardo M. Sousa. 2012. “How Do Banking Crises Impact on Income Inequality?” 

Applied Economics Letters 19 (15): 1425–29. https://doi.org/10.1080/13504851.2011.631885. 

Ahamed, Liaquat. 2009. Lords of Finance: The Bankers Who Broke The World. Penguin Random House 

Publishing.  

Ahlquist, John S. 2006. “Economic Policy, Institutions, and Capital Flows: Portfolio and Direct Investment 

Flows in Developing Countries.” International Studies Quarterly 50 (3): 681–704. 

Alfaro, Laura, Anusha Chari, and Fabio Kanczuk. 2017. “The Real Effects of Capital Controls: Firm-Level 
Evidence from a Policy Experiment.” Journal of International Economics 108 (September): 191–
210. https://doi.org/10.1016/j.jinteco.2017.06.004. 

Alfaro, Laura, Sebnem Kalemli-Ozcan, and Vadym Volosovych. 2007. “Capital Flows in a Globalized 

World: The Role of Policies and Institutions.” NBER, May, 19–72. 

Bailey, Michael A., Anton Strezhnev, and Erik Voeten. 2017. “Estimating Dynamic State Preferences from 

United Nations Voting Data.” Journal of Conflict Resolution 61 (2): 430–56. 
https://doi.org/10.1177/0022002715595700. 

Benmelech, Efraim, and Eyal Dvir. 2013. “Does Short-Term Debt Increase Vulnerability to Crisis? 

Evidence from the East Asian Financial Crisis.” Journal of International Economics 89 (2): 485–94. 
https://doi.org/10.1016/j.jinteco.2011.12.004. 

Blomberg, S. Brock, Gregory D Hess, and Athanasios Orphanides. 2004. “The Macroeconomic 

Consequences of Terrorism.” Journal of Monetary Economics 51 (5): 1007–32. 
https://doi.org/10.1016/j.jmoneco.2004.04.001. 

Bodea, Cristina, Ibrahim Elbadawi, and Christian Houle. 2017. “Do Civil Wars, Coups and Riots Have the 

Same Structural Determinants?” International Interactions 43 (3): 537–61. 
https://doi.org/10.1080/03050629.2016.1188093. 

Carneiro, Cristiane Lucena, and Laerte Apolinário Jr. 2016. “Targeted Versus Conventional Economic 

Sanctions: What Is at Stake for Human Rights?” International Interactions 42 (4): 565–89. 
https://doi.org/10.1080/03050629.2015.1036989. 

Catão, Luis A. V., and Gian Maria Milesi-Ferretti. 2014. “External Liabilities and Crises.” Journal of 

International Economics 94 (1): 18–32. https://doi.org/10.1016/j.jinteco.2014.05.003. 

Clay, K. Chad, and Matthew R. Digiuseppe. 2017. “The Physical Consequences of Fiscal Flexibility: 

Sovereign Credit and Physical Integrity Rights.” British Journal of Political Science 47 (4): 783–
807. https://doi.org/10.1017/S0007123415000502. 

 

149 

 

Corneli, Flavia, and Emanuele Tarantino. 2016. “Sovereign Debt and Reserves with Liquidity and 

Productivity Crises.” Journal of International Money and Finance 65 (July): 166–94. 
https://doi.org/10.1016/j.jimonfin.2016.02.016. 

Davis, J. Scott, Adrienne Mack, Wesley Phoa, and Anne Vandenabeele. 2016. “Credit Booms, Banking 
Crises, and the Current Account.” Journal of International Money and Finance 60 (February): 
360–77. https://doi.org/10.1016/j.jimonfin.2015.09.008. 

DiGiuseppe, Matthew. 2015. “The Fiscal Autonomy of Deciders: Creditworthiness and Conflict 
Initiation.” Foreign Policy Analysis 11 (3): 317–38. https://doi.org/10.1111/fpa.12044. 

DiGiuseppe, Matthew, and Patrick E. Shea. 2016. “Borrowed Time: Sovereign Finance, Regime Type, and 

Leader Survival.” Economics & Politics 28 (3): 342–67. https://doi.org/10.1111/ecpo.12081. 

Dominguez, Kathryn M. E., Yuko Hashimoto, and Takatoshi Ito. 2012. “International Reserves and the 

Global Financial Crisis.” Journal of International Economics 88 (2): 388–406. 
https://doi.org/10.1016/j.jinteco.2012.03.003. 

Dreger, Christian, Konstantin A. Kholodilin, Dirk Ulbricht, and Jarko Fidrmuc. 2016. “Between the 

Hammer and the Anvil: The Impact of Economic Sanctions and Oil Prices on Russia’s Ruble.” 
Journal of Comparative Economics 44 (2): 295–308. https://doi.org/10.1016/j.jce.2015.12.010. 

Drury, and Li. 2006. “U.S. Economic Sanction Threats Against China: Failing to Leverage Better Human 

Rights.” 2006. https://onlinelibrary-wiley-com.proxy1.cl.msu.edu/doi/abs/10.1111/j.1743-
8594.2006.00033.x. 

Du, Brian, and Darius Palia. 2018. “Short-Term Debt and Bank Risk.” Journal of Financial and 
Quantitative Analysis 53 (2): 815–35. https://doi.org/10.1017/S0022109017001132. 

Early. 2009. “Sleeping With Your Friends’ Enemies: An Explanation of Sanctions‐Busting Trade.” 

International Studies Quarterly 53 (1): 49–71. https://doi.org/10.1111/j.1468-
2478.2008.01523.x. 

———. 2011. “Unmasking the Black Knights: Sanctions Busters and Their Effects on the Success of 

Economic Sanctions1.” Foreign Policy Analysis 7 (4): 381–402. https://doi.org/10.1111/j.1743-
8594.2011.00143.x. 

Early, and Spice. 2015. “Economic Sanctions, International Institutions, and Sanctions Busters: When 
Does Institutionalized Cooperation Help Sanctioning Efforts?” Foreign Policy Analysis 11 (3): 
339–60. https://doi.org/10.1111/fpa.12038. 

Ekici, Oya, and Karun Nemlioğlu. 2017. “Emerging Economies’ Short-Term Private External Debt as 

Evidence of Economic Crisis.” Journal of Policy Modeling 39 (2): 232–46. 
https://doi.org/10.1016/j.jpolmod.2017.01.002. 

Elliott, Kimberly Ann. 1998. “The Sanctions Glass: Half Full or Completely Empty?” International Security 

23 (1): 50–65. https://doi.org/10.2307/2539262. 

 

150 

 

Escriba-Folch, and Joseph Wright. 2010. “Dealing with Tyranny: International Sanctions and the Survival 

of Authoritarian Rulers.” International Studies Quarterly 54 (2). https://onlinelibrary-wiley-
com.proxy1.cl.msu.edu/doi/abs/10.1111/j.1468-2478.2010.00590.x. 

Fariss, Christopher J. 2018a. “The Changing Standard of Accountability and the Positive Relationship 

between Human Rights Treaty Ratification and Compliance.” British Journal of Political Science 
48 (1): 239–71. https://doi.org/10.1017/S000712341500054X. 

———. 2018b. “The Changing Standard of Accountability and the Positive Relationship between Human 

Rights Treaty Ratification and Compliance.” British Journal of Political Science 48 (1): 239–71. 
https://doi.org/10.1017/S000712341500054X. 

Gaillard, Norbert. 2011. A Century of Sovereign Ratings. 2012 edition. London: Springer. 

Grauvogel, Julia, Amanda A. Licht, and Christian von Soest. 2017. “Sanctions and Signals: How 

International Sanction Threats Trigger Domestic Protest in Targeted Regimes.” International 
Studies Quarterly 61 (1): 86–97. https://doi.org/10.1093/isq/sqw044. 

Hatipoglu, Emre, and Dursun Peksen. 2016. “Economic Sanctions and Banking Crises in Target 

Economies.” Defence and Peace Economics 29 (2): 171–89. 
https://doi.org/10.1080/10242694.2016.1245811. 

Houle, Christian. 2009. “Inequality and Democracy.” World Politics 61 (4): 589–622. 

Jeanne, Olivier. 2007. “International Reserves in Emerging Market Countries: Too Much of a Good 

Thing?” Brookings Papers on Economic Activity 2007 (1): 1–79. 
https://doi.org/10.1353/eca.2007.0013. 

Jung, Kuk Mo, and Ju Hyun Pyun. 2016. “International Reserves for Emerging Economies: A Liquidity 

Approach.” Journal of International Money and Finance 68 (November): 230–57. 
https://doi.org/10.1016/j.jimonfin.2016.06.020. 

Kabadayi, Burhan. 2013. “The Effects of Sovereign Ratings Changes on Turkey’s Stock Market.” Journal 

of Applied Finance and Banking 3 (6): 87–96. 

Kaempfer, Lowenberg, and Mertens. 2004. “International Economic Sanctions Against a Dictator.” 

Economics & Politics 16 (1): 29–51. https://doi.org/10.1111/j.1468-0343.2004.00130.x. 

Lane, Philip, and Gian Milesi-Ferretti. 2006. “The External Wealth of Nations Mark II : Revised and 

Extended Estimates of Foreign Assets and Liabilities, 1970–2004.” IMF. 2006. 
https://www.imf.org/en/Publications/WP/Issues/2016/12/31/The-External-Wealth-of-Nations-
Mark-II-Revised-and-Extended-Estimates-of-Foreign-Assets-and-18942. 

Li, and Drury. 2004. “Threatening Sanctions When Engagement Would Be More Effective: Attaining 

Better Human Rights in China.” International Studies Perspectives 5 (4): 378–94. 
https://doi.org/10.1111/j.1528-3577.2004.00185.x. 

 

151 

 

Marinov. 2005. “Do Economic Sanctions Destabilize Country Leaders?” American Journal of Political 

Science 49 (3). https://onlinelibrary-wiley-com.proxy1.cl.msu.edu/doi/full/10.1111/j.1540-
5907.2005.00142.x. 

McLean, Elena V., and Taehee Whang. 2010. “Friends or Foes? Major Trading Partners and the Success 

of Economic Sanctions.” International Studies Quarterly 54 (2): 427–47. 
https://doi.org/10.1111/j.1468-2478.2010.00594.x. 

Mortimer, Joanne Stafford. 1967. “VIII. Commercial Interests and German Diplomacy in the Agadir 

Crisis.” The Historical Journal 10 (3): 440–56. https://doi.org/10.1017/S0018246X00027692. 

Neuenkirch, Matthias, and Florian Neumeier. 2015. “The Impact of UN and US Economic Sanctions on 

GDP Growth.” European Journal of Political Economy 40 (December): 110–25. 
https://doi.org/10.1016/j.ejpoleco.2015.09.001. 

Nooruddin, Irfan. 2002. “Modeling Selection Bias in Studies of Sanctions Efficacy.” International 

Interactions 28 (1): 59–75. https://doi.org/10.1080/03050620210394. 

Oatley, Thomas. 2010. “Political Institutions and Foreign Debt in the Developing World.” International 

Studies Quarterly 54 (1): 175–95. https://doi.org/10.1111/j.1468-2478.2009.00582.x. 

Obstfeld, Maurice, Jay C. Shambaugh, and Alan M. Taylor. 2010. “Financial Stability, the Trilemma, and 

International Reserves.” American Economic Journal: Macroeconomics 2 (2): 57–94. 
https://doi.org/10.1257/mac.2.2.57. 

Peksen, Dursun. 2009. “Better or Worse? The Effect of Economic Sanctions on Human Rights.” Journal of 

Peace Research 46 (1): 59–77. 

———. 2010. “Coercive Diplomacy and Press Freedom: An Empirical Assessment of the Impact of 

Economic Sanctions on Media Openness.” International Political Science Review 31 (4): 449–69. 
https://doi.org/10.1177/0192512110372610. 

———. 2017. “Autocracies and Economic Sanctions: The Divergent Impact of Authoritarian Regime Type 

on Sanctions Success.” Defence and Peace Economics 0 (0): 1–16. 
https://doi.org/10.1080/10242694.2017.1368258. 

Peksen, Dursun, and A. Cooper Drury. 2009. “Economic Sanctions and Political Repression: Assessing the 

Impact of Coercive Diplomacy on Political Freedoms.” Human Rights Review 10 (3): 393–411. 
https://doi.org/10.1007/s12142-009-0126-2. 

———. 2010. “Coercive or Corrosive: The Negative Impact of Economic Sanctions on Democracy.” 
International Interactions 36 (3): 240–64. https://doi.org/10.1080/03050629.2010.502436. 

Peksen, Dursun, and Byunghwan Son. 2015. “Economic Coercion and Currency Crises in Target 

Countries.” Journal of Peace Research 52 (4): 448–62. 
https://doi.org/10.1177/0022343314563636. 

 

152 

 

Peterson, Timothy M. 2014. “Taking the Cue: The Response to US Human Rights Sanctions against Third 

Parties.” Conflict Management and Peace Science 31 (2): 145–67. 
https://doi.org/10.1177/0738894213503432. 

Reuter, Ora John, and Jennifer Gandhi. 2011. “Economic Performance and Elite Defection from 

Hegemonic Parties.” British Journal of Political Science 41 (01): 83–110. 

Rodrik, Dani. 2006. “The Social Cost of Foreign Exchange Reserves.” International Economic Journal 20 

(3): 253–66. https://doi.org/10.1080/10168730600879331. 

Schultz, Kenneth A., and Barry R. Weingast. 2003. “The Democratic Advantage: Institutional Foundations 

of Financial Power in International Competition.” International Organization 57 (1): 3–42. 
https://doi.org/10.1017/S0020818303571065. 

Shea, Patrick E., and Paul Poast. 2017. “War and Default.” Journal of Conflict Resolution, June, 

0022002717707239. https://doi.org/10.1177/0022002717707239. 

Svolik, Milan. 2008. “Authoritarian Reversals and Democratic Consolidation.” American Political Science 

Review 102 (2): 153–68. https://doi.org/10.1017/S0003055408080143. 

Svolik, Milan W. 2012. The Politics of Authoritarian Rule. Cambridge University Press. 

Tang, Min, Narisong Huhe, and Qiang Zhou. 2017. “Contingent Democratization: When Do Economic 

Crises Matter?” British Journal of Political Science 47 (1): 71–90. 
https://doi.org/10.1017/S0007123415000095. 

Teixeira, João C. A., Francisco J. F. Silva, Manuel B. S. Ferreira, and José A. C. Vieira. 2018. “Sovereign 

Credit Rating Determinants under Financial Crises.” Global Finance Journal 36 (May): 1–13. 
https://doi.org/10.1016/j.gfj.2018.01.003. 

Wang, Chih-Wei, Wan-Chien Chiu, and Tao-Hsien Dolly King. 2017. “Debt Maturity and the Cost of Bank 

Loans.” Journal of Banking & Finance, November. 
https://doi.org/10.1016/j.jbankfin.2017.10.008. 

Wood. 2008. “‘A Hand upon the Throat of the Nation’: Economic Sanctions and State Repression, 1976–

2001.” International Studies Quarterly 52 (3): 489–513. https://doi.org/10.1111/j.1468-
2478.2008.00512.x. 

 

 

 

 

 

 

153