Corruption and Political Turnover in a Sovereign Default Model

Transcription

1 Corruption and Political Turnover in a Sovereign Default Model Adams Adama School of Social Sciences, The University of Manchester, Department of Economics, 3rd Floor, Arthur Lewis Building, Oxford Road, Manchester, M13 9PL, United Kingdom. adams.adama@manchester.ac.uk Abstract This paper looks at the interactions between government spending, government borrowing, political corruption and political turnover. Incorporating these factors in a sovereign default model, we show that sovereign borrowing and default decisions, together with business cycle fluctuations, are affected by the level of corruption. We find that, following a negative technology shock, less corruption leads to a greater fall in household consumption, whilst more corruption leads to a greater fall in politicians consumption. Counter intuitively, we establish that a change in power from a more corrupt to a less corrupt government is more likely to cause default than the reverse. This is due to the fact that a more corrupt government wants to borrow more than a less corrupt government because of its greater inclination to steal public funds. Given that defaulting restricts future access to borrowing, the incentive of a more corrupt government to default is tempered relative to that of a less corrupt government. In addition, a change in power from a more corrupt to a less corrupt government means that there is relatively more debt to pay off. Another result is that, a more corrupt policy maker has to borrow more to achieve the same level of welfare for the economy as a less corrupt policy maker. Finally, we find that households suffer welfare loss as a result of corruption and over borrowing. To support the findings from our theoretical analysis, we estimate an empirical model using data on sovereign default, corruption, political stability and other macroeconomic variables. Our empirical results produce evidence of a positive effect of corruption on sovereign default. Keywords: government, corruption, weak institutions, sovereign default JEL Classification Numbers: D72,D73,H6,B22 1

2 Contents 1 Introduction 5 2 Data Methodology and Estimation Strategy Sovereign Default Corruption Government Stability Macroeconomic factors Empirical results Robustness checks Robustness Using Classical Methods Robustness Using Bayesian Method The Model Government Households Firms The Government s Decision Problem Solution Recursive Equilibrium Results Calibration Corruption, Political Risk, and Default Risk Corruption and Business Cycles Conclusion 22 References 23 7 Appendix Definition of Terms Used in Regression Computational Method List of Figures List of Tables Baseline Results

3 7.4.2 Test of Robustness Bayesian Estimates Using Student-t Priors Bayesian Estimates Using Gamma Priors List of Figures 1 Default set against bond issuance for more and less corrupt governments Default set against output for more and less corrupt governments Bond issuance and price Economy s Welfare under more and less corrupt types Value function for corrupt and non-corrupt governments Welfare loss from corruption Impulse Responses to a Negative Technology Shock Value function for patient and impatient types under same level of corruption Objective function for more and less corrupt governments Objective function for corrupt and non-corrupt governments Welfare and Default Debt Levels List of Tables 1 CALIBRATION Summary statistics Cross-correlations Results for Determinants of Sovereign Default Continuation of Results for Determinants of Sovereign Default Test of Robustness: Results for Determinants of Sovereign Default Test of Robustness: Results for Determinants of Sovereign Default Test of Robustness: Results for Determinants of Sovereign Default Test of Robustness: Marginal Effects Test of Robustness: Marginal Effects Test of Robustness- INTERACTIONS :Results for Determinants of Sovereign Default Test of Robustness- INTERACTIONS :Results for Determinants of Sovereign Default Test of Robustness- INTERACTIONS :Results for Determinants of Sovereign Default Test of Robustness Using Bayesian Linear Regression (Student-t Prior) Test of Robustness Using Bayesian Linear Regression (Student-t Prior) Test of Robustness Using Bayesian Linear Regression (Student-t Prior)

4 17 Test of Robustness Using Bayesian Linear Regression Using Student-t Prior Test of Robustness Using Bayesian Linear Regression( Student-t Prior) Test of Robustness Using Bayesian Linear Regression (Gamma Prior) Test of Robustness Using Bayesian Linear Regression (Gamma Prior) Test of Robustness Using Bayesian Linear Regression (Gamma Prior) Test of Robustness Using Bayesian Linear Regression(Gamma Prior) Test of Robustness Using Bayesian Linear Regression(Gamma Prior) Test of Robustness:Excluding Interest Rate Spreads Test of Robustness:Excluding Interest Rate Spreads Model Mean Default Values Model moments

5 1 Introduction Business cycles in emerging economies vary from their developed counterparts. Higher interest rates and interest rate spreads, greater output volatility, higher corruption, lower GDP Per Capita, greater political instability, weaker institutions and higher frequencies of sovereign default are the common characteristics of emerging economies. Sovereign default has attracted the attention of academics and policy makers for a long time, and attention has been reignited by recent events in the world. This paper seeks to contribute to the literature on sovereign default by studying the role of corruption in influencing default decisions. It does so within the context of a dynamic stochastic general equilibrium (DSGE) model which allows for the random political turnover of alternative types of government that have different propensities to be corrupt. Corruption manifests as the embezzlement of public funds that are otherwise used to finance productive public expenditures. The results of the analysis highlight the important role that corruption can play in determining the risk of default, a role that is evidenced by the data. Sovereign default has been attributed to various reasons, both economic and political. Hatchondo et al. (2009) introduced political factors into the modeling of sovereign default which they claimed to have accounted for defaults in Brazil in 2002 and Argentina in The authors present a DSGE model with political turnover, where different government types are distinguished in terms of their degree of impatience which influences their willingness to pay off debt. In turn, this affects the equilibrium bond spread through the direct impact on each type s borrowing opportunities. It is shown that sovereign default may be triggered when the government changes from a more patient to a less patient type, provided that there is a enough political stability. The model also illustrates a non-monotonic relationship between political stability and default probability which is in line with the political turnover models of Amador (2003)and Cuadra and Sapriza (2008), where political stability affects sovereign bond spreads through future utility flows. The foregoing analyses yield important insights, but one issue that is not addresses, but which merits attention, is the role of corruption in influencing the risk of sovereign default. This is particularly pertinent for emerging economies, and there are good reasons for thinking why corruption may be important for default decisions. For instance, a corrupt policy maker may be willing to borrow substantial funds 5

6 even with high interest rate spreads in order to create room for stealing. As a result, corruption may lead to a higher incidence of default if loan repayments become excessive. Ciocchini et al. (2003) studied the effect of corruption on sovereign bond spreads in emerging economies and observed that high levels of corruption are associated with high spreads. Importantly, the authors argue that corrupt governments need liquidity which results in high borrowing even with very high spreads. Furthermore, if corruption leads to a drain on loaned funds and other sources of government income, then it may limit the government s ability to meet its debt obligations. For instance, in Russia more than US$4 billion in IMF loans apparently disappeared shortly before Russia s default in 1998 (Ciocchini et al. (2003) ). Similarly, Haque and Sahay (1996), Tanzi and Davoodi (1997)and Johnson et al. (1998) observe that higher levels of corruption are associated with lower tax revenue, which would in turn lower the government s ability to repay its debt. As argued in Hatchondo et al. (2009), political turnover may have significant implications for sovereign default decisions. One observation is that most of the countries that default seem to be the most corrupt. In addition, the data suggests that there is a strong correlation between sovereign default, corruption and political turnover. In spite of the above, little or no attempt has been made to explore the interlinkages between corruption and the risk of default. This paper seeks to do so. The analysis is based on a DSGE model in which different government types alternate in power. In contrast to most other models, we endogenise output, rather than treat it as an endowment. In particular, we suppose that output depends on productive public expenditures, the provision of which requires public funding in one form or another. Corruption takes the form of a government s illegal consumption of public funds, which impacts on both public goods provision and the decision to default. The government borrows in order to smooth household consumption and may be either of two types - a more corrupt or a less corrupt type - which alternate in power with a certain probability. The government in office makes a decision about whether or not to default by trading off the benefits (avoidance of debt repayments) and the costs (restricted access to future borrowing) associated with this. We show how borrowing and default decisions, together with business cycle fluctuations, are affected by the level of corruption. We establish that higher levels of corruption are associated with higher levels of borrowing and higher sovereign default risk. Counter-intuitively, we find that this risk is higher when there is political turnover from a more corrupt to a less corrupt government, rather than 6

7 vice-versa. We also compute the welfare loss to households as a result of corruption and show that this loss is increasing in the amount of government borrowing. To set the scene for our theoretical analysis, we first conduct an empirical investigation into the determinants of sovereign default. Using data on a broad sample of emerging economies, we find that, amongst other factors, both corruption and political stability have a significant positive effect on sovereign default. The rest of the paper is organized as follows. In section 2 we present the empirical analysis of the determinants of sovereign default. In section 3 we set out the theoretical model that we use to study the relationship between sovereign default and corruption. In section 4 we solve the model. In section 5.1 we calibrate the model and present our main results in section5.2. In section 6 we make some concluding remarks. 2 Data This section looks empirically at the determinants of sovereign default.our main objective is to evaluate the extent to which corruption has direct or indirect effects on the risk of default. The literature on the empirical determinants of credit spreads in emerging markets is broad. Ciocchini et al. (2003) studied the effect of corruption on sovereign bond spreads in emerging economies and observed that high levels of corruption are associated with high spreads. Edwards (1984), Min (1998); Rowland and Torres (2004) find that macroeconomic factors have significant impact on credit spreads. However, political factors and corruption have not received much attention in the empirical literature.baldacci et al. (2011) find in their studies that lower political risk is associated with tighter spreads. Verma (2002) finds that more democratic countries tend to default more, whilst Bordo and Oosterlinck (2005) find that half of the default incidents that occurred during 1880 to 1913 were around political turnovers. We use data from World Bank Development Indicators(WDI), Standard and Poors, Moodys, Transparency International and International Country Risk Guide(ICRG) from We consider the 7

8 following variables as potential determinants of sovereign default: corruption, government stability and political rights, together with various controls such as external debt, annual GDP growth and GDP Per Capita. We focus on emerging economies, where the selection of countries is based on the availability of data for the period of study( ), and also controlling for outliers. This gives us 31 countries for the period 1984 to 2008 which include: Ukraine, Uruguay, Moldova, Peru, Paraguay, Pakistan, Dominican Republic, Cote Divore, Venezuela, Ecuador, Indonesia, Russia, Cameroon, Argentina, Algeria, Bangladesh, Botswana, China, Egypt Arab Republic, Ethiopia, Gabon, Gambia, India, Kazakhstan, Kenya, Lebanon, Malaysia, Senegal, Sri Lanka, Tanzania and Zambia. 2.1 Methodology and Estimation Strategy We specify the econometric model as: de f i = α + δc i + γy i + βx i + ε i where de f i = the number of defaults, C i = vector of measures of corruption, X i = vector of controls, Y i includes government stability index(gsi) and political rights(pr) and ε i = error term. 2.2 Sovereign Default We use data from Moody s and Standard and Poor s as a measure of default. We look at sovereign default episodes for 31 emerging economies from 1984 to Using this data, we establish the following measures of sovereign default: (A) default if a country defaults year-after-year; (B)default if a country defaults after 3 years; (C) default if a country ever defaults during the sample period. Based on measure A we we come out with 775 observations. Measure B leads to 279 observations. And measure C gives 31 observations. We are likely to face endogeneity problems when we use A and B. This endogeneity problems may arise as a result of the fact that an announcement of default by a country could lead to a high corruption rating. To correct for this we can use measure C, where we regress current default on immediate past explanatory variables. We do this by choosing from the data each observation that precedes a year of 8

9 default for all the countries that default. For the countries that never default we take an average for the period Corruption We use three main measures of corruption. The first is the Transparency International Corruption Perception Index(TPI) which measures corruption on a scale of 1-10, 10 being the least corrupt. The second is International Country Risk Guide Index(ICRG) on a scale of 0-6 which we convert to 0-10, with 10 being least corruption. The third is the World Bank Index (cppi) which measures corruption on a scale of -2.5 to 2.5 with 2.5 indicating most transparency. However, the World Bank measure of corruption and the Transparency International measure of corruption are strongly correlated and we therefore concentrate on one of them. Since the World Bank measure expands more into the past and we prefer that to the Transparency International measure. 2.4 Government Stability In our attempt to examine the effect of political turnover on sovereign default risk, we use government stability index(gsi) data from the International Country Risk Guide(ICRG) and Political Rights (PR) Index from Freedom House, for the period 1984 to The GSI data is measured from 0-12, where 12 indicates the most stable government whilst the Political Rights is measured from 1-10, where 10 is the worst; which we convert to 0-10, 10 being the best. 2.5 Macroeconomic factors Several macroeconomic variables which affect sovereign default are used as controls in our estimation. Included amongst there are the logarithm of external debt, logarithm of GDP Per Capita, annual GDP growth, interest rate spread, real interest rate all from WDI. To capture the effect of domestic borrowing on sovereign default we use the real interest rate, and interest rate spreads( i.e. the difference between domestic lending rate and deposit rate in the economy). Real interest rate has been used in the literature 9

10 as explanatory variable for determinats of sovereign default(see Verma (2002)). Since Real interest rate and interest rate spreads have high positive correlation, we do not include both of them in the estimation at a time. 2.6 Empirical results The baseline results of our model are reported in tables (4) and (5). The central argument of this paper is that corruption and political turnover affect sovereign default decision. In each of the two tables, we present the results based on OLS. Column 11 of table (4) shows our OLS results when we exclude all macroeconomic variables. The results in column 11 show that the coefficient on World Bank measure of corruption carries the expected negative sign and it is significant at 10%. This implies that higher corruption increases sovereign default. The coefficients on government stability and Political Rights are both significant at 5%. It is generally thought that government instability is conducive to sovereign default but our results show it is otherwise which is consistent with Hatchondo et al. (2009) findings. In our setting the level of borrowing could even be higher due to political stability coupled with corruption for the reason given above. Similarly, one would expect the sign on political right to be negative as good political rights of the people will mean less corruption and less default. However we argue that the positive sign is right for similar reasons given for government stability. The results in the rest of the columns in tables (4) and (5) are where we include one or more macroeconomic determinants in our regression.the coefficients on corruption, government stability and political rights have the right signs and are significant in the rest of the results where we include one or more macroeconomic variables. The coefficient on external debt has the expected (positive) sign and is significant in all columns where it is included in the regression except in column 2 in table (4) and column 10 in table (5). This is inline with the view that excessive borrowing leads to default. Another macroeconomic determinant of sovereign default is the interest rate spread. As argued in Ciocchini et al. (2003), a high bond spread leads to a high risk of default. The sign on the interest rate spread is positive and significant in all cases which supports this view. The real domestic interest rate coefficient has the expected(positive) sign in all 10

11 cases where is it included in the regression. However it is only significant in column 3 and 2 in tables (4) and (5) respectively. The positive sign implies that high interest rates lead to high default risk. The coefficient on GDP Per Capita is positive and significant through out tables (4) and (5). The possible explanation could be that higher GDP Per Capita does not necessarily mean the people are earning. There could be a few rich people or foreigners who actually own all the resources. This implies that higher GDP per Capita could still lead to higher default risk. GDP per capita has also been argued to work as a proxy for a country s level of political stability(see Kalliomäki (2012)). Given this argument, we expect it to have a positive effect on default. Similarly, when we include both GDP growth and GDP Per Capita in the regression, the sign is always positive and sometime significant.this could be the effect from GDP Per Capita. However, when we exclude GDP Per Capita alone the signs of GDP growth in columns 5, 6 and 10 of table (5) are negative and significant at 10%. The negative sign is in line with findings in the literature(see Tomz and Wright (2007) ), that low GDP growth leads to default. 2.7 Robustness checks In this part of the paper we check for robustness using (i) a probit model, (ii) different measures of corruption, (iii) interactions effect and (iv) Bayesian analysis Robustness Using Classical Methods To check for robustness we first run a probit regression. Tables (7) to (10) are the results from the probit regression and their respective marginal effects. The results show that the coefficients on the ICRG measure of corruption are negative and significant in all columns where it is included in the regression in tables (7) and (8). The World Bank measure of corruption is negative and only significant in column 1 of table (7). This implies that the more transparent a country is, the less the risk of sovereign default. The coefficients on government stability and political rights are both significant at 5% or 10%. It will generally be thought that government instability will affect sovereign default decision positively, it is however the otherwise. As we argued above high stability of government(see Hatchondo et al. (2009)) 11

12 is good for sovereign default. We argue also that the positive sign on political rights is right for similar reasons given for government stability. The coefficient on external debt has the expected (positive) sign and is significant in all columns in table (8) except in column 8 but not significant in table (7). The sign on the interest rate spread is positive and significant in table 8 where we use the ICRG measure of corruption, but insignificant in table (7) where we use the World Bank measure of corruption. Similarly, real domestic interest rate coefficient has the expected sign and is significant in all columns in table (8) where it is included in the regression. The coefficient on GDP Per Capita is significant in column 6 of table (6). Similarly, the coefficient of GDP growth is negative and significant in columns 3 and 4 of table (7), and only column 3 of table (8). It is interesting to point out that the all coefficients in our probit model are larger than those from our baseline regression results. Tables (9) and (10) are the marginal effects of the results in tables (7) and (8) respectively. This is particularly important in our analysis of sovereign default risk. Holding all other things constant, a change in corruption from 0 to 1 is likely to increase default risk by 76%, 26% and 32% in columns 1, 2, and 7 respectively in table (9). A similar interpretation applies to rest of the columns in tables (9) and (10). The results in tables (11) to (13) relate to the interactions effect. When we interact the ICRG measure of corruption with GDP growth the interaction term is negative and significant at 5% in column 7 of table (13). This implies that where GDP growth is low coupled with high corruption, sovereign default is likely to be high. However, the World Bank measure of corruption is not significant when interacted with GDP growth. Our interaction of political rights and government stability is positive and significant. This is to imply that when government stability is high and at the same time we have better rights of the people, the risk of sovereign default is still likely to be high. The positive and significant coefficient when we interact government stability or political rights with GDP Per Capita implies that even when GDP Per Capita is high, there is still the risk of sovereign default when political rights is better or when government is highly stable. As we argued earlier, politically stable government is conducive to sovereign default. We could also argue that once we interact political stability with another proxy for political stability we expect to have a positive sign. 12

13 Despite our attempt to correct for endogeneity by using measure C, interest rate spreads could still be argued to be endogenous. It is important we point out again that the interest rate spreads we use here is the difference between lending and deposit rates in the domestic economy and not the spreads in the international bond market. We therefore believe that interest rate spreads here is not endogenous. However, to clear any doubt regarding the endogeneity of the interest rate spreads, we drop interest rate spreads from our regression in tables (24) and (25). All explanatory variables except external debt have the expected signs and are significant in all columns in tables (24) and (25). External debt has the right sign in all columns and significant in columns 5 and 6 in table (24) and in column 1 in table (25). Generally, our results do not change much with the exclusion of interest rate spreads. More importantly, excluding interest rate spreads does not change effect that our main 1 explanatory variables have on sovereign default Robustness Using Bayesian Method A limitation of our empirical analysis is so far the small nature of the sample size. To further check for robustness of the estimates from our classical methods we use a Bayesian method which is argued to provide a better solution to the problem of small sample. Tables (14) to (18) and (19) to (23) are the results from the Bayesian Linear Regression using student-t priors and gamma priors respectively. The coefficient on corruption is negative in all cases which shows that less transparency(more corruption) results in higher sovereign default. Except in table (15), the coefficient on government stability is positive in all cases with the student-t priors. This supports our baseline results. Similarly the sign on political rights consistently remains positive through out. The signs of the coefficients on the interest rate spread is positive whilst the signs on GDP Per Capita are positive in all cases except in table (18) where we regress the dependent variable on corruption, political rights and GDP Per Capita. Tables (19) to (23) are the results for using gamma priors. In all cases we provide the 2.5% and 97.5% percentiles of the posterior means. The 2.5% and the 97.5% percentiles approximate the 5% highest posterior density intervals (HPDI) or 5% credible sets which imply that our coefficients are significant at 1 corruption and political stability 13

14 5% level. 3 The Model We consider an economy with infinitely-lived households, perfectly competitive firms, and alternating types of government. The basic structure of our model follows that of Eaton and Gersovitz (1981) as used recently by Arellano (2008), Yue (2006), Aguiar and Gopinath (2006) andhatchondo et al. (2009). Our model differs from those frameworks by incorporating government expenditures and government corruption into the analysis of sovereign default. The basic assumption is that the economy borrows to smooth consumption through the central government s trading of risk free bonds. There is a single stream of a stochastic and tradable good which is produced by firms and consumed by households and the government. There is a shock to productivity which accounts for business cycle fluctuations. The government has an opportunity to default on its debt, but doing so is costly because it leads to a loss in output and future lower bond prices. Corruption takes the form of a government s pilfering of public funds which reduces the provision of productive public goods. There exist two types of political rulers which alternate in power and which differ according to their propensity to be corrupt. Both types are partially benevolent in the sense that they care about not just the utility of households but also their own utility from corrupt consumption. At the beginning of period t, a corrupt ruler j in power observes the current state of the economy and borrows to smooth consumption. During period t, ruler j observes the shocks and the debt level, and decides to default or not. If a default is chosen, the economy subsequently suffers an output loss and lower bond prices in the future. At the end of the period t, there is a probability that the ruler will be replaced by the alternative government 3.1 Government Without corruption, the total amount of funds available for government spending comprises taxes on firms profits, plus government borrowing, less debt repayments. The budget constraint of the government 14

15 is therefore g t = τπ t + q jd (γ j,t,y t,b t+1 )b t+1 (1 d)b t (1) where g t denotes public funds, τ is the tax rate, Π t denotes profits, b t denotes debt, b t+1 denotes borrowing, q j is the bond price and d is a default decision parameter which takes the value 1 when there is default and 0 when there is no default. Equation[1] gives the total amount of public funds that are potentially available for productive public expenditures. But not all of these funds are used for these expenditures: some of the funds are looted to finance the consumption of corrupt political leaders. Denoting this consumption by c g j,t the actual amount of government expenditure is given by ĝ t, where c g j,t + ĝ t = g t (2) In what follows, we assume that c g j,t and ĝ t are proportionally related to g t such that: ĝ t = θ j g t (3) c g j,t = (1 θ j )g t (4) The term θ j measures the government s propensity to be corrupt which is specified in the calibrated version of the model as exp( γ j ). The larger is γ j, the lower is θ j and the more corrupt is a government. 3.2 Households We assume a continuum of infinitely-lived households who derive utility from consumption, c t and who supply one unit of labour to firms in return for a wage, w t. The government borrows on behalf of households. The representative household s expected utility is given as E 0 [ t=0 β t u(c t )] (5) where u(c t ) = log(c t ) and its budget constraint is given by 2 c t = w t + (1 τ)π t (6) 2 Π t + w t = y t w t + w t = y t 15

16 3.3 Firms Firms produce output, y t, using labour, l t, and the services provided by productive public goods, ĝ t. Following Barro (1990), Mauro (1996) and, Farida and Ahmadi-Esfahani (2007) the production function is given by y t = A t ĝ 1 α t l α t (7) where A t is a technology shock which evolves according to log(a t ) = ρlog(a t 1 ) + ε At,t 4 The Government s Decision Problem Following others, we make the standard assumption that defaulting is costly not only because it imposes a direct output loss on the economy and lower bond prices in the future which are standard assumptions in the sovereign default literature. We denote by λ t the output loss due to default, and by H the history of default, which takes the value 1 when default has occurred and 0 when default has not occurred. Combining the household budget constraint[6] with the government s budget constraint[1] yields the resource constraint on the economy as c t + g t q jd (y,b t+1 )b t+1 (1 Hλ)y t (1 d)b t (8) Using equations[3], this may be re-written as c t + 1 (1 θ) c g j,t q jd (y,b t+1 )b t+1 = (1 Hλ)y t (1 d)b t (9) Thus household consumption is made up of net output, plus government borrowing, less debt repayments, less consumption by corrupt politicians. Evidently, the higher is corruption( i.e. the lower is θ j ), the lower is the level of households consumption. The government is partially benevolent in the sense that it cares about not just the utility of households, but also its own utility from its corrupt consumption of public funds. Its objective function is given by E 0 β t U(c t,c g j,t ) (10) t=0 where U = (1 γ j )u(c t ) + γ j u(c g, j,t ) = (1 γ j )logc t + γ j c g, j,t. The government s decision problem is to choose c t, c g j,t,b t+1 and d to maximize [10] subject to [9], where y t is given in [7] and ĝ t = ( θ j 1 θ j )c g j,t from [3] and [4]. 16

17 4.1 Solution We use the value function iteration method to solve the dynamic problem. Let V j (A,b,H) denote the value function of a government of type j at the beginning of period t when it is in power. Similarly, let ˆV j (A,b,H) denote the value function of a government of type j at the beginning of period t when it is in opposition. Finally, let P denote the probability of political turnover. Since the two government types may differ, the value functions V j and ˆV j do not need to coincide. When the government in power finds it optimal to default, its value function is given by V j1 (A,H) = max b,c g, j,t {U[(y(1 λh) + q j1 (b,y)b 1 (1 θ) c g j,t ),c g j,t ] +(1 P)βEV j (A,b,1) + PβE ˆV j (A,b,1)} On the other hand, when the government in power does not find it optimal to default, its value function is given by V j0 (A,b,H) = max b,c g j,t {U[(y(1 λh) b + q j0 (b,y)b 1 (1 θ) c g j,t ),....c g j,t ] + (1 P)βEV j (A,b,H) + PβE ˆV j (A,b,H)} We compute V j (A,b,H) as V j (A,b,H) = max { V j1 (A,H),V j0 (A,b,H) } The value function of a government j when it is not in power depends on the optimal behaviour of the government in power, denoted by i. If type i finds it optimal to default, the value function of type j is given by ˆV j (A,H) = U[(y(1 λh) + q i1 (b i1 j (H,y))b i1 (H,y) 1 (1 θ) c g i,t )] +PβEV j (A,b i1 (H,y),1/y) + (1 P)βE ˆV j (A,b i1 (y,h),1/y) If type i does not find it optimal to default, the value function for type j takes the form ˆV j (A,b,H) = U[(y(1 λh) b + q i0 (b i0 (b,y,h),y)b i0 (b,y,h) 1 (1 θ) c g i,t )] +PβEV j (A,b i0 (b,y,h),0/y) + (1 P)βE ˆV j (A,b i0 (b,y,h),0/y) The government decides to default when the value of default is greater than the value of no default, otherwise there is no default, in which case the government repays its debt. Let d j (b,y(ĝ),h) denote the optimal default decision of a government of type j. Then d j (b,y(ĝ),h) = { 1 if V j1(a,h) > V j0 (A,b,H) 0 if V j1 (A,H) V j0 (A,b,H) Bond prices satisfy the lenders zero-profit condition which, adapting from Hatchondo et al. (2009), 17

18 is given by q jd (b,y(ĝ),γ j ) = r [1 P(γ j) d i j (y,b,h)f(dy /y) (1 P(γ j )) d j (y,b,h)f(dy /y)] Basically, a country s bond price depends negatively on the level of corruption in the country,γ j, and the credit history, H, that the future government inherits which depends on the default decision, d j in the current period. It will be recalled that default in the current period decreases future output and affects participation in the financial market which could therefore affect future default decisions. Obviously, bond prices also depend on the type of government in power(i.e. either more corrupt or less corrupt). Information about the current government type in power not only gives information about the probability distribution of future government types but also affects the default probability distribution for the subsequent period. The probability of a government being replaced is defined such that E c Y aio = Γ i P i where Γ i, E c,y aio, P i are the level of corruption, election cycle, average number of years in office and the probability of being replaced respectively. 4.2 Recursive Equilibrium Recall that V (A,b,H) : R R is the life time value function that starts with current output y, asset position b and credit record H, where R = B {0} AUB {1} A. The asset price is such that 1 Q = {q/q(b,y,γ j ) : B Y Γ [0, 1+r ]} where q Q. Let j = L,M denote the index of the two types of government(less corrupt and more corrupt). A stationary recursive equilibrium is defined as : 1) the set of value functions V j (A,b,H),V j1 (A,H),V j0 (A,b,H)and, ˆV j (A,b,H) for all j = L,M ; 2) the policy functions c t, ĝ t,b j0 (b,y,h),b j1 (y,h), c g, jtand the default decision rule d j (b,y(g),h) for all j = 18

19 L,M; 3) and the bond price q jd (b,y,γ j ) for all j = L,M, such that : (a)v j (A,b,H),V j1 (A,H),V j0 (A,b,H) and, ˆV j (A,b,H) satisfy the functional equations (b) c t, ĝ t,b j0 (b,y,h),b j1 (y,h) c g j,t and d j (b,y,h) solve the dynamic programming problem; (c) the bond price q jd (b,y,γ j ) satisfies the lenders zero profit condition. 5 Results 5.1 Calibration We calibrate the model to resemble an emerging economy that is known to display corruption and sovereign default. To this end, we base our calibration on Hatchondo et al. (2009) for the case of Argentina. The output loss parameter is taken to be The risk-free interest rate is taken to be 1%. The parameter values that govern the stochastic process for the technology shock are chosen so as to depict the behaviour of GDP. Table 1: CALIBRATION Description Parameter Value Corruption Index γ j [0 1] Interest rate r t 0.01 Autocorrelation coefficient ρ 0.95 Standard dev of innovation σ Output loss λ Discount factor β 0.95 Stable government P s 1.5/100 Unstable government P u 2.5/100 The degree of political stability is taken as a free parameter, and two scenarios are considered. The first is a politically stable economy where the ruler is expected to rule for an average of 16 years; the second is a politically unstable economy where the ruler is expected to rule for an average of 10 years. The corruption parameter ranges from 0-1 with higher values indicating a more corrupt government. 19

20 5.2 Corruption, Political Risk, and Default Risk. Here we present an analysis of whether changes in political situation associated with changes in the level of corruption can trigger sovereign default. We simulate the model for 1,000,000 periods. Counter intuitively, we show that when there is political stability but a possibility of political turn over, a change in regime from a more corrupt government to a less corrupt government is more likely to trigger default than a change in regime from a less corrupt government to a more corrupt government. This is because a more corrupt government tends to borrow more in order to fill its own pockets.the possibility of remaining in power tempers the government s incentive to default as the government will face punishment in the form of lower future bond prices and loss of output. In addition, a more corrupt government appropriates more public funds from other sources of revenue. The upshot is that there is more debt for any new government. Figure(1) shows that the probability of default increases with γ j ( the corruption index). Figure 1 plots the default set against debt whilst figure(2) plots the default set against output for more corrupt and less corrupt governments. Together, they show that default is higher with more debt and lower output respectively.they also show that default is higher with higher corruption; at the same level of debt, default risk is higher with more corruption. The intuition is that low corruption reduces the share of government consumption and this in turn increases private consumption. The relatively high weight on private consumption reduces the risk of default as the policy maker can more easily repay the debt. On the other hand, when corruption is high, there is less weight on private consumption and the policy maker has greater incentive to default and keep the money for himself. For any given low level of output, the risk of default is higher for a more corrupt government. Since corruption affects output negatively in our model, higher level of corruption means the graph shifts inside and up, implying that risk of default increases. Figure(3) is a plot of the bond price for stable and unstable governments. The bond price is affected by the decision to default, the level of corruption and the political turnover of the government. High political stability and low corruption are associated with a high bond price. That is, the more stable and less corrupt is the political environment, the higher the price at which a policy maker issues debt. The 20

21 more corrupt and the most unstable government type compensates the lender for default contingency by offering a low bond price. In figure 3, at any given level of debt the most corrupt and unstable government faces the least bond price followed by the moderately corrupt and stable government. Also consistent is that high bond price is associated with high debt repayment practice and hence low default risk. Figure(4) plots the welfare function of the economy. A more corrupt policy maker finds it optimal to choose bond levels that are far above the less corrupt type s bond level. In essence the more corrupt policy maker has to borrow more to achieve the same level of welfare for the economy as the less corrupt policy maker can achieve. This is a further reason why corruption can lead to default. To achieve the same level of welfare, the more corrupt government that decides to default will do that with more assets as compared to the government not defaulting. It is also observed from figure(5) that comparing the case without corruption at all and a case where there is very negligible corruption, the value function is only higher for the corrupt when the government borrows more. For low levels of debt the economy without corruption seems to do better. Figure(11) shows that the less corrupt default at low debt level than the more corrupt. Further analysis of the implications of corruption on sovereign default risk is carried out using figure(6). Comparing the case of our model where there is an infinitesimal level of corruption to a model where corruption is not modeled, the welfare loss is high where more borrowing is done. In fact the welfare loss is negative where less amount of bonds are issued. Finally, the results in table (26) show that not only do changes in power from more corrupt to less corrupt government more likely to cause default, but changes from unstable to stable government is also more likely to cause default than otherwise. 5.3 Corruption and Business Cycles We now look at the implications of corruption for business cycles in the economy. Essentially, the effect of corruption is to amplify the effect of a negative shock that hits the economy. Figure( 7) shows the impulse responses from a negative technology shock. A negative technology shock reduces output on impact but the fall in output is higher with more corruption. High corruption means that politicians consumption increases with more than in the case of low corruption. Conversely, households consumption 21

22 increases more with intermediate and low levels of corruption than with high level of corruption after the shock. Productive government spending increases more with high corruption. Debt increases on the impact of a negative shock and the increase is greater when corruption is higher. This is as a result of the effect of the negative shock on output coupled with high corruption which together lead to more borrowing. Table (27) shows the model moments. Output and debt are more volatile when corruption is highest. 6 Conclusion This paper seeks to provide an explanation for the possible interaction between sovereign default and corruption. The key factors that determine the likelihood of default is corruption and government spending. We find that, once government spending and corruption are taken into account, the risk of sovereign default increases through the negative effect of corruption on the bond price and government spending and borrowing. We also find that when a negative shock hits the economy, less corruption leads to a greater increase in households consumption whilst more corruption leads to a more increase in government consumption. Interestingly, we establish that changes in power from a more corrupt to a less corrupt government is more likely to cause default than the reverse. Finally, we find that a more corrupt policy maker has to borrow more to achieve the same level of welfare for the economy as a less corrupt policy maker achieves. Our empirical results support the positive effect of corruption on sovereign default decision. The policy implication of the paper is that corruption and government spending have important roles in explaining sovereign default, and that efforts by international organizations to prevent sovereign debt crises should not only involve assistance in the design of better fiscal policies, but also tackle the root of the problem which is corruption. 22

23 References Acemoglu, D. and Johnson, S. (2004). "Institutions as Fundamental Cause of Long-Run Growth. NBER Working Papers Series Acemoglu, D. and Verdier, T. (1998). Property Rights, Corruption and Allocation of Talent: A General Equilibrium Approach. The Economic Journal 450, Acemoglu, D. and Verdier, T. (2000). The Choice Between Market Failures and Corruption. The American Economic Review 1, Aguiar, M. and Gopinath, G. (2006). Defaultable Debt, Interest Rates and the Current Account.. Journal of International Economics 69 (June): Alfaro, L. and Kanczuk, F. (2005). Sovereign Debt as a Contingent Claim: A Quantitative Approach.. Journal of International Economics 65 (March): Amador, M. (2003.). A Political Economy Model of Sovereign Debt Repayment.. Stanford University mimeo. Arellano, C. (2008). Default Risk and Income Fluctuations in Emerging Economies.. American Economic Review 98 (June): Baldacci, E., Gupta, S., and Mati, A. (2011). Political and fiscal risk determinants of sovereign spreads in emerging markets. Review of Development Economics, 15(2): Barro, R. (1990). "Government Spending in a Simple Model of Endogenous Growth.". Journal of Political Economy 98, s103-s117. Blackburn and Forgues-Puccio (2010). Financial Liberalisation, Bureaucratic Corruption and Economic Development. Journal of International Money and Finance. Blackburn, K. and Powell, J. (2011). Corruption, Inflation and Growth. Centre for Growth and Business Cycles Research, Department of Economics, University of Manchester, UK. Bordo, M. and Oosterlinck, K. (2005). Do political changes trigger debt default? and do defaults lead to political changes? In Conference on the Political Economy of International Finance PEIF. Ciocchini, F., Durbin, E., and Ng, D. (2003). Spreads?. Does Corruption Increase Emerging Market Bond Cole, H. L., Dow, J., and English, W. B. (1995). "Default, Settlement, and Signalling: Lending Resumption in a Reputational Model of Sovereign Debt.. International Economic Review 36 (May):

24 Cuadra, G. and Sapriza, H. (2008). Sovereign Default, Interest Rates and Political Uncertainty in Emerging Markets.. Journal of International Economics 76 (September): Cule, M. and Fulton, M. (2005). "Some Implications of the Unofficial Economy-Bureaucratic Corruption Relationship in Transition Countries.". Economics Letters 89, Dixit, A. and Londregan, J. (2000). Political Power and the Credibility of Government Debt.. Journal of Economic Theory 94 (September): Djankov, S., Glaeser, E., La Porta, R., Lopez-de Silanes, F., and Shleifer, A. (2003). The New Comparative Economics.. NBER Working Papers Series Eaton, J. and Gersovitz, M. (1981). Debt with Potential Repudiation: Theoretical and Empirical Analysis.. Review of Economic Studies 48 (April): Economist., T. (1999). Argentina s Next Steps.. The Economist. Economist, T. (2005). Argentina s Debt Restructuring. Tough Deal.. The Economist. Edwards, S. (1984). Ldc s foreign borrowing and default risk: an empirical investigation. Esfahani, H. (2000). "Institutions and Government Controls.. Journal of Development Economics 63, Farida, M. and Ahmadi-Esfahani, F. (2007). Modelling Corruption in a Cobb-Douglas Production Function Framework.. Australian Agricultural and Resource Economics society, 51 st Annual Conference. February 13, Gelos, R. G., Sahay, R., and Sandleris, G. (2004). Sovereign Borrowing by Developing Countries: What Determines Market Access?. IMFWorking Paper 04/221. Hall, R. E. and Jones, C. I. (1999). Why Do Some Countries Produce so much more Output Per Worker than Others?. The Quarterly Journal of Economics. Haque, N. U. and Sahay, R. (1996). Do Government Wages Cuts Close Budget Deficits? The Cost of Corruption.. IMF Staff Papers, Vol. 43, No 4. December, pp Hatchondo, J. C. and Martinez, L. (2009.). "Long-Duration Bonds and Sovereign Defaults.. Journal of International Economics 79 (September): Hatchondo, J. C. and Martinez, L. (2010.). The Politics of Sovereign Defaults.. Economic Quarterly Volume 96, Number 3 Third Quarter 2010 Pages