Determinants of Sovereign Defaults - 3

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Determinants of Sovereign Defaults: What Does the Empirical Evidence Tell Us?

Author(s): Sujata Verma


Source: Economic and Political Weekly, Vol. 37, No. 48 (Nov. 30 - Dec. 6, 2002), pp. 4817-
4822
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4412903
Accessed: 18-09-2016 00:52 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend access to
Economic and Political Weekly

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
Determinants of Sovereign Defaults
What Does the Empirical Evidence Tell Us?
Developing-country defaults on their external debt repayments are a function of both
their 'willingness to pay' as well as their 'ability to pay'. While the ability to repay can be
captured by economic factors like the grth rate of the economy and external shocks,
willingness to repay is more difficult to measure. In this paper, it is emphasised that
political factors are important determinants of the willingness to repay and should be treated
as an explanatory variable explicitly. A mnultivariate probit model is estimated that tests the
hypothesis that the probability of default is inversely related to the level of democracy in the
debtor country. It is shown that the evidence doesn't support the hypothesis.

SUJATA VERMA

may not be willing to do so. Eaton Among


and the important political variables
Introduction are how democratic the political regime is
Gersovitz (1981) were the first to emphasise
this distinction been 'ability to pay' and
andthe extent of political stability in the
overeign debt crises have become 'willingness to pay'. If a country incurs a
debtor country.
commonplace in the past two de negative output shock, and is unable to It was recognised in the literature on
cades. Many Latin American coun- make repayment, it can be captured 'sovereign
by risk' that the political circum-
tries defaulted on their debt in the 1980s, economic indicators such as slowdown in stances in the debtor nation must enter the
calculation of risk in making loans to a
followed by the east Asian crisis in the late the rate of growth of output and low rates
1990s and more recently in Russia. Inter- of investment. Financial ratios like a low foreign nation. It is difficult to quantify
national credit markets differ from the such risk systematically as there are both
reserve/exports ratio or a high debt servic-
domestic credit markets in a fundamental ing/exports ratio indicate that the countryconceptual as well as measurement prob-
way. Out of respect for sovereign immu-is facing difficulties in making repayments.lems. Political systems in different coun-
nity, the creditors cannot punish the sov- Political factors such as type of politicaltries are very diverse in nature. Democracy
ereign defaulter. Then why do debtorregime influence the willingness to repayis a multi-dimensional concept and its
nations repay their foreign debt? What isloans. For instance, Hungary is an example definition is controversial in the existing
the incentive mechanism at work here? of a country that never defaulted in spite literature. It has been defined in terms of
The objective of this paper is to explore of facing severe economic problems. For political processes, for example, the pro-
empirically the determinants of sovereign four consecutive years, 1990-1994, it hadcess of selection of the executive. It has
a negative rate of growth of per capitaalso been associated with the multiplicity
defaults. In particular, the focus is on how
a sovereign's decision to default or to GDP, and was on the verge of a default, of political parties, compared with a dic-
repay its external debts is influenced by yet it didn't default. One of the plausibletatorship. Alternatively, democracy has
its political system. reasons is that Hungary was in the processbeen defined in terms of extent of political
There exist a number of studies, which of transformation from a communist state rights and civil freedoms of the nations.
have focused on the influence of political to a multi-party democracy, and needed the There does not exist any single measure
factors on the economic growth of coun- support of the western world. This is an of democracy that is satisfactory. For
example of how political considerations instance, for many monarchies, a non-
tries. Robert Barro in his 1994 paper
explored the question of whether democ-affected the decision not to default. elective head can guarantee civil freedoms
racies promote internal contract enforce-We can distinguish between two kinds to the people. Another example is of
In determinants of sovereign defaults: Hungary, where democratisation of an
ment, and concluded that they did not. of
Probability of default = f(economic fac-
this paper the following question is ana- erstwhile communist state was accompa-
lysed 'Do democracies promote external tors, political factors). nied by the mushrooming of political
contract enforcement?' The promise toThere exist two distinct sets of variables parties. Given these difficulties in measur-
repay its foreign debt is taken as an ex- - economic and political - that are used ing democracy, we propose the use of four
to explain sovereign defaults. Financial
ample of the external contract. It is shown alternative measures of democracy (de-
that the empirical evidence does not sup- ratios like the degree of indebtedness and fined in Section II) to overcome the mea-
port the hypothesis that democracies are interest-reserve ratio are good indicators surement problems and to obtain meaning-
of the debt servicing capacity of debtor
less likely to default on their foreign debt ful results.
repayments. nations. Structural variables like the per This paper is organised as follows:
A debtor nation could default either capita GDP and fiscal deficit describe Section II discusses the theory and metho-
because it is unable to repay the loancountry
or characteristics. All these are im- dology followed. It also contains a discus-
a country may be able to repay its debt portant
but economic explanatory variables. sion of the structural and political factors

Economic and Political Weekly November 30, 2002 4817

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
affecting the probability of default. Sec- We do not have the observations on the equated with political stability, but this
tion III outlines the statistical contours.
left-hand-side of equation (1), since pro- may not be true, for instance, in Singapore
Section IV summarises the results. First bability that default will take place is not where autocracy has proved to be a stable
known. A random variable, YI, is defined political system. Two variables measuring
the results of the basic regression are laid
out, followed by sensitivity analysis in such that Yi =1 if default occurs and zero political instability included in the study
Section V. Section VI concludes. The data otherwise. are Changes in Effective Executive and
sources are listed in the appendix. Yi = OIX = Xiifa + PiX, + e, < 0 ...(3) Coups. For a detailed discussion of the
definition and data sources of all the
II
i = 1|X = X iifa + PiXi + Ei > 0 ...(4) political variables, refer to the appendix.
The value of Z in equation (2) is a linear
Theory and Methodology
function of the exogenous variables and Ill
The objective of this empirical study is thus the value changes with Xi according Statistical Contours
to determine the factors that influence the to the sign and magnitude of i.
sovereign's decision to repay or default on The rate of change is not constant. A In this section, the summary statistics
its sovereign debt. This study is based on transformation of co-efficients is used highlighting the differences in structural
data for 30 developing countries over a that yields and political characteristics among coun-
20-year period from 1975 to 1995. This tries are given in Table 1. Variables con-
dP (YI = 1)
interval is chosen so as to cover the period = o(z i) .(5) sidered include total external debt/gross
of significant defaults by many Latin dXi national product, number of reschedulings
American countries in the early 1980s. of debt in the sample period, security price
The dependent variable is taken to be the In the probit model the term Pi appears of the secondary debt, political freedom
probability of default. Historically, there as a multiplicative factor and determines index, democracy index, pluralism, party
have been few cases of outright default.1 the sign of the effect. The transformed Pi fractionalism index and numberofcoups in
Instead of a sudden snapping of borrower- can be used to find the effect of a unit the sample period. The average values of
lender ties, what we see is debt renego- change in the exogenous variable on the these variables are calculated over the sample
tiations and agreements on rescheduling probability of default. We made use period of followed by the standard errors.
of debt between them. In most empirical STATA statistical package to obtain theTable 1 contains data on the mean average
studies,2 actual rescheduling has been used transformed equation. over the 20-year period (or the longest
as a proxy for measuring the probability sample available in this period) and the
of default. Probit analysis has been used Factors Affecting Probability mean standard deviation for each of the
as the statistical tool since the dependent of Default 30 countries in the sample. From the tale
variable, namely, whether the country we observe that 17 of the 30 nations
rescheduled or not, can take only limited Structural Variables: The structural ex-rescheduled their debt during the sample
values. Recognising the shortcomings of planatory variables included in Xi includeperiod. The second column shows the
this approach, we make use of the second- the external debt - GNP ratio, growth rate extent of indebtedness of various nations;
ary market price of sovereign debt as an of GDP, interest payments-export ratio,the higher the total external debt/gross
alternative dependent variable while check- debt service-export ratio, budget surplus,national product ratio, higher the level of
ing for robustness of the results. interest rates (Libor), and the average indebtedness. China and Singapore are
The probit model3 assumes that the growth rate of the industrialised nations.among the least indebted nations whereas
probability of default, Pd' is related to a The rationale and the data sources are
Panama, Egypt and Morocco are among
vector of economic and political variables given in the Appendix. the most indebted. Columns 3 and 4 show

Xi, I= 1, 2, ...n, by the following functional Political Variables: Given the problems
the alternative dependent variables used.
form: in measuring democracy, we define demo- It is clearly seen that the countries with a
cracy in terms of four alternative measures.
high number of reschedulings also tended
1 +co
Political Rights Index and Democracy to be the ones whose debt sold at a large
P = F(Z) = _ fe(-U2/2)du=h(z) ...(1)
\(27r) oo Index reflect the definition of democracy
discount in the secondary markets (Ecua-
in terms of political and civil freedoms dor, Peru, Morocco, Argentina, Brazil).
where u is a random variable normally guaranteed by a political regime to its Columns 5 and 6 show the level of
distributed (with mean zero and unit vari- people. Two other measures capturingdemocracy
the according to political and civil
ance) and Z is a linear combination of the process of political participation include
rights. A lower value signifies high level
Xi, which is obtained from the inverse of Political Pluralism and Party Fraction- of democracy. Costa Rica, Israel, Portugal,
the cumulative normal function: alisation Index. Trinidad and Tobago, and Venezuela are
It is important to make a distinctionamong the most democratic nations and
Z = F-l(PI) = ct + Pi Xi ...(2) between the political regime variables and
Hungary, China, Indonesia, Kenya, Tuni-
Pd represents the probability of loanpolitical instability variables. Whereas the sia and Yugoslavia are among the least
default. Default will be more likely to former refers to whether a country isdemocratic according to this criterion.
occur the larger the value of the index Z. democratic or not, the latter variables are When level of democracy is measured by
Thus Pd can be interpreted as an estimateindicated by the number of coups, assas-the openness of the political process, in
of the conditional probability that a coun-sinations or changes in the heads of state, columns 7and 8, highervalues are indica-
try will default on its loans, given the and can occur irrespective of the kind of tive of higher level of democracy. In terms
information contained in Xi. political regime. Democracy is sometimesof this criterion, Costa Rica, India, Israel,

4818 Economic and Political Weekly November 30, 2002

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
Portugal, Trinidad and Tobago and Ven- Among the non-political variables, bud-the Libor is contemporaneous, its lower
ezuela are among the most democratic value may reflect the incentive for the
get surplus and growth rate in industrialised
nations, and communist countries, Kenya, countries are not significant explanatorydebtor to renegotiate loans at a lower cost
Tunisia and Uruguay are among the least variables. In terms of the debt burden ratios,than the cost of previously borrowed funds.
democratic. Five of the countries in our the probability of default is directly related
sample showed evidence of successful coupsto the level of indebtedness and interest/ V
and are taken as evidence of unstable poli-export earnings ratio. The probability of
Sensitivity Analysis and
tical systems; these countries are Thailand, default is negatively related to the growth Extensions
Peru, Turkey, Ecuador and Argentina. Arate of the debtor country, as expected. It
high value of political factionalism indexis puzzling that Libor has a negative Sensitivity of results to alternative speci-
for these countries is not indicative of
coefficient, which implies that probability fications of explanatory variables: Since
democracy, but of an unstable politicalof default is higher for lower value of interest-exports ratio and total external debt
system. Next, we turn to the results ob-Libor. One explanation could be that since ratio were highly significant, these were
tained from running the probit regressions.
Table 1: Summary Statistics

IV Country TED/ Number Security Political Demo- Plyralism Party Coups


GNP of Re- Price Freedom cracy Fraction
Probit Estimation Results schedulings Index Index Index
Argentina 56.9 0.24 40.76 2.8 6.2 6.6 40.8 0.04
This section presents a multivariate probit 21.4* 0.43* 17.6* 1.8* 3.3* 3.8* 29.7* 0.21*
analysis to estimate how various macro-Hungary 62.5 0 100 4.47 8.52 4.28 20.22 0
12.22* 0* 0* 1.91* 3.45* 3.7* 32.8* 0*
economic and political variables changeBrazil 37.02 0.19 44.37 3 6.2 7.8 65.25 0
the probability of occurrence of default, 8.6* 0.4* 20.8* 1* 1.64* 0.72* 15.9* 0*

holding constant the other explanatorySri Lanka 64.2 0 100 2.95 6.9 8.4 37.8 0
9.3* 0* 0* 0.86* 1.75* 1.32* 20.4* 0*
factors. Chile 80.02 0.09 79.7 4.9 9 3.38 26.5 0
Table 2 reports the probit equation 34.3* 0.3* 14.6* 1.9* 3.39* 4.42* 38.4* 0*
China 10.3 0 100 6.4 12.9 2 0 0
estimation results for four alternative model
6.2* 0* 0* 0.59* 1.09* 0* 0* 0*
specifications based on different defini-Colombia 37.65 0 76.9 2.2 5.5 9.7 56.2 0
9.39* 0* 11.7* 0.53* 0.98* 0.84* 6.12* 0*
tions of democracy. Since probit coeffi-Costa Rica 98.8 0.23 47.6 1 2.14 11 58.27 0
cients are not easily interpretable, the effects 37.2* 0.43* 19.6* 0* 0.35* 0* 5.5* 0*
Ecuador 93.9 0.28 35.9 3 5.8 7.2 67.7 0.04
of changes in the regressors on the pro- 21.8* 0.46* 16.2* 1.8' 2.5* 3.6* 34* 0.21*
bability of default, evaluated at the meanIndia 93.98 0 100 2.4 5.71 9.57 54.99 0
21.8* 0* 0* 0.7* 1.4* 0.5* 17.7* 0*
values, are reported in percentage terms.
Indonesia 53.45 0 100 5.42 10.7 6 64.77 0
The estimated coefficients for each model 16.8* 0* 0* 0.74* 1 0* 6.49* 0*
are reported along with the number ofIsrael 89.1 0 100 1.85 4.19 9.6 75.2 0
7.8* 0* 0* 0.3* 0.4* 0.5* 3.7* 0*
observations and the pseudo R-square. Kenya 79 0.04 100 5.47 10.6 4.47 12.33 0
Two groups of variables are considered: 22.9* 0.21' 0* 0.67* 1.27' 1.88' 26* 0*
Korea 33.2 0 100 3.9 7.95 7.3 59.6 0
structural and political. The number of 17.8 * 0 * 1.6* 2.81* 1.8* 9.9* 0*
observations ranges from 437 to 457Malaysia 46.3 0 100 3.6 7.85 7.3 30.9 0
14* 0* 0* 0.8' 1.06* 0.5* 9.8* 0*
depending on the availability of data of the
Mexico 51.8 0.14 56.9 3.33 7.47 9.3 47.09 0
regressors. Pseudo R-square ranges from 17.9* 0.35* 12.9* 0.7* 0.6* 0.9* 10* 0*
0.25 to 0.29, and suggests a moderateMorocco 99.3 0.28 56.4 4.3 9 5.6 80.6 0
25.2* 0.45* 14.35* 0.7* 1.1* 1.8' 4.2* 0*
degree of explanatory power of the model. Panama 105.1 0.09 40.8 4.9 9.09 4.19 45.4 0
The results indicate that all the political 26.3* 0.3* 21.3* 1.54* 2.6* 2* 45.3* 0
Peru 63.9 0.23 26.3 3.7 7.6 6.6 49.4 0.09
variables are statistically significant at 95 18.65* 0.4* 25.2* 1.7* 2.4* 3.4* 25.09* 0.3*
per cent level of significance. This impliesPhilippines 71.05 0.23 59.5 3.8- 7.7 6 25.1 0
13' 0.43* 12.4' 1.2* 2.1* 2.6* 18.8' 0*
that they are important explanatory vari-
Portugal 54.5 0 100 1.6 3.5 10.2 66.1 0
ables. The Political Rights Index has a 15.1* 0* 0* 1.2* 1.6* 0.9* 8.7* 0*
negative coefficient, implying that democ- Singapore 18.5 0 100 4.4 9.7 6.1 3.7 0
2* 0* 0* 0.51' 2.2* 1.3* 3.6* 0*
racies have a higher probability of default. Thailand 37.3 0 100 3.56 7.4 6.6 68.8 0.14
This result also holds true for the broader 5.7* 0* 0* 1.35* 1.9* 2.1 34.2* 0.35*
Trinidad and 35.9 0.09 100 1.3 3 10.3 38.2 0
measure in terms of Democracy Index. Tobago 16.2* 0.3* 0* 0.49* 0.86* 1.11* 15.8* 0*
Pluralism has a positive coefficient, whichTunisia 60.8 0 100 5.6 10.5 3.4 2 0
10.7' 0* 0* 0.5* 0.8* 1.2* 6.4* 0*
implies that more the political pluralism Turkey 43 0.14 100 2.9 7.1 7.4 53.9 0.047
(more democracy), higher the probability 8.9* 0.35* 0* 1.2* 1.9* 3.2* 23.9' 0.21*
of default. Party Fractionalism Index has Egypt 104.8 0.09 100 5 9.8 6.4 32.1 0
18.8* 0.3* 0* 0.89* 1.2* 1.1* 10.4' 0*
a positive coefficient, more fragmentedUruguay 48.68 0.9 74.7 3.4 6.9 4.3 35,8 0
the parties (more democratic regime), 18.3* 0.3* 19.1* 1.9' 3.68* 4.29* 35.03* 0*
Venezuela 66.1 0.04 57.7 1.47 3.8 9.4 62.5 0
higher is the probability of default. In terms 12.2' 0.21' 12.3* 0.8* 1.16* 0.75* 4.03* 0*
of pseudo R2, the definition of democracyYugoslavia 36.3 0.19 58.12 5.8 11 4.1 11.7 0
7.4* 0.4* 30.4* 0.4* 0.8* 0.3* 28.6* 0*
in terms of Party Fractionalism Index has
the best goodness of fit. Note: Mean Average (years: 1975-1995) and Mean St

Economic and Political Weekly November 30, 2002 4819

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
used in all regressions. The following the government guarantees the former but market at huge discounts. The size of the
alternative specifications were used: not the latter. Lee uses commercial banks'
discount reflects the lack of confidence
(1) Rate and level of growth in the debtor rescheduling cases [see also Berg and Sachsthat these debts will ever be fully repaid
country was measured by three alternative (1988)] as the dependent variable. We[Ozler and Huizinga 1991]. Thus the lower
specifications: growth rate of per-capita have taken official and commercial debt the price at which debt is sold relative to
GDP, level of per capita GNP, or the In- rescheduling separately and also totalits face value, the lower its creditworthi-
vestment-GDP ratio. rescheduling as alternative ways of mea- ness, and higher the probability of default.
(2) Cost of funds was measured usingsuring the probability of default. The data on the secondary market prices
four alternative specifications: London For total rescheduling as well as for is published weekly by Solomon Brothers
Inter Bank Rate on US dollar deposits official rescheduling, all the four measures and is available from 1986 onwards. The
yearly and six-monthly, Libor lagged oneof democracy are found significant and results of regressions of various structural
year and the cost of borrowed funds cal-show that more democratic counties have variables and the four measures of democ-
culated by deflating the Liborby the export a higher probability of default. For com- racy on the secondary market price are
price index following World Bank metho-mercial rescheduling, political freedom summarised in Table 3. The dependent
dology. index and pluralism are not significant variable is the security price of the second-
(3) Inclusion of the total debt service- variables. Thus the results are not altered ary market country debt. Two groups of
exports ratio. in any fundamental way by making a variables are considered: structural vari-
We found that the results on the political distinction between rescheduling on offi- ables and political variables. The OLS
variables remained unaltered. Democracy cial and commercial debt. coefficients along with their standard errors
Index showed a negative coefficient, while Use of secondary market price of debt are reported.
Pluralism, and Party Fractionalism Index as an alternative dependent variable: The number of observations ranges
showed a positive coefficient. So the Annual data on rescheduling of debt mayfrom 95 to 103 depending on the avail-
empirical results are shown to be robust not accurately measure the probability ofability of data of the variables. R-square
to alternative definitions of the structural default since there is a time'lag betweenranges from 0.37 to 0.50, and suggests a
variables. default and a formal rescheduling. Thismoderate degree of explanatory power of
Distinguishing between official and problem can be overcome by using thethe model.
commzercial debt rescheduling: Lee (1991) secondary market price of third world The sign on the Political Freedom Index
points out that it is important to distinguish country debt as the dependent variable.and Democracy Index is positive, which
between official and commercial debt, as The third world country debt is sold on theshows that democracies have a lower

WOMEN IN UNORGANISED SECTOR


PROBLEMS & PROSPECTS

* Jawahar Lal Singh * Ravi Kumar Pandey * Arun Kumar Singh


Women labour generally work in the unorganised sector. This sector is *2002
characterised by the
various factors viz., long hours of work, wage discrimination, lack of job security,
? 331 pages no limit of
minimum wages, lack of minimum facilities at work place, ill treatment, heavy
* Rs. 700/- physical work,
sexual exploitation, etc. In this book, an attempt has been made to examine the socio-economic
* Hard Bound
status of women workers engaged in unorganised sector and to study the socio-economic
* ISBN 81-87365-03-X
transformation of women workers, of their standard of living, children, health, education, life
expectancy and status in the family, etc. Measures and yardsticks for further strengthening
E] employment opportunities of women workers engaged in unorganised sector have also been
suggested.

APPROACHES TO HIGHER EDUCATION


*J. L. Singh *R. K. Pandey *A. K. Singh
Higher education promotes social and economic development by building human and technical
capabilities of a society. Higher education has an important role to play in facilitating
* 2002 technical
and institutional changes by helping to integrate advanced training and research with shifting
* 269 pages
economic realities. In this book, it has been sought to examine main issues and aspects
* Rs. 650/-
related to higher education which include policy, network and financing in the global perspective
* Hard Bound
in general and Indian context in particular. Further, the dimensional perspective related to
* ISBN
higher education and its policy have been examined since planning era and the 81-87365-02-1
need for new
policy for higher education at different levels has also.been suggested.

B-7. Saraswati Compl~er, Suhbhash Chowk,


Laxmni Nagar New Delhi- 11 (092. Phone: 2025199
E-mail: sunrisepublications(ahotmaail. coin

4820 Economic and Political Weekly November 30, 2002

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
creditworthiness and hence a higher political instability in these countries. The destabilising. A coup in the debtor country
probability of default. The negative sign inclusion of political instability does not is necessarily destabilising, but has no
on Party Fractionalisation Index and alter the results obtained earlier for the definite impact on the probability of
Pluralism further reinforces this result. An- political regime variables. default.
other interesting result is the positive sign Coups were not a statistically significant Sensitivity of probability of default to
of the coefficient on growth rate of variable. Changes in executive head has level of income in debtor country: Next,
industrialising countries, implying that a a negative coefficient. This implies thatwe classify the debtor countries into
lending boom affects the security prices those countries in the sample which hadthree sub-groups according to their level
favourably. a higher number of changes in the heads of per capita GNP, namely, low-income
Inclusion of political instability vari- of government in the year tended not tocountries, middle-income countries and
ables: Number of changes in the executive default. However, changes in executivehigh-income countries. The purpose of
head in a year and number of coups - these head also includes change in elected this sub-division was to find out if coun-
two variables are taken as measures of the government and so may not always be tries similar in terms of average income
per person showed similarity in the re-
Table 2: Determinants of Probability of Default - Probit Estimation Results
sponse of the government towards repay-
(1) (2) (3) (4) ment of debt. Among the structural vari-
Structural Factors
ables considered were investment/GDP,
Real GDP growth -0.0039* -0.0039* -0.00285* -0.0040* interest-exports ratio, total external debt-
(.0018) (0.0018) (0.0015) (0.0018) GNP ratio and budget surplus. The politi-
Interest/Exports 0.0034* 0.0033' 0.0030* 0.0036*
(.0009) (0.009) (0.0008) (0.00094) cal variables considered were Democracy
External Debt/GNP 0.0015* 0.0015* 0.0011 * 0.0015* Index, Party Fractionalism Index, Change
(0.0003) (0.00033) (0.00031) (0.00033) in Executive Index and Coups.
Budget Surplus 0.00257 0.00220 0.00072 0.0031*
(0.0024) (0.0024) (0.0019) (0.0024) It is interesting to note that none of the
LIBOR -0.0136' -0.013' -0.010' -0.0129* political variables were significant for the
(0.0035) (0.0035) (0.0032) (0.00359) high-income debtor countries. For the low
Industrialising country rog 0.0079 0.0075 0.0052 0.0083
(0.0077) (0.0076) (0.0064) (0.0077) and the middle-income debtor countries,
Political Factors it was found that the probability of default
Political Freedom Index -0.010*
was positively related to democracy, de-
(0.0052)
Democracy Index -0.0063* fined in terms of 'party fragmentation'.
(0.0030)
Party Fractionalism Index 0.00094*
VI
(0.00032)
Pluralism 0.00533*
Concluding Remarks
(0.0031)
Number of observations 457 457 437 457
Pseudo R2 0.257 0.260 0.2964 0.254 In this paper, it has been demonstrated
that political factors play a significant role
Note: * indicates significance at 5 per cent level. All variables
in a sovereign's decision to default. Given
in parentheses.
four alternative definitions of democracy,
Table 3: Determinants of the Probability of
countries in the sample with a more Default-
demo- O

(1) (2) (3) (4) cratic regime tended to default more. This
result is robust to different definitions of
Structural Factors:
Growth rate of GDP 2.969 2.88 3.024 3.127 democracy, for various specifications of
(0.613) (0.593) (0.657) (0.696) the regression equation and for different
Interest/Exports -0.2730* -0.0411 * -0.308* -0.165*
definitions and measures of the dependent
(0.571) (0.559) (0.614) (0.645)
Real interest rate -0.0016 -0.00177 -0.0020 -0.00175 variable.
(0.00071) (0.00068) (0.00076) (0.00079) Thus, similar to Barro's contention that
Total external debt/GNP -0.1868 -0.219 -0.11597 -0.1223*
democracies do not promote internal con-
(0.0868) (0.0847) (0.091) (0.094)
Libor -0.4176* -0.9529* -2.2* -2.466* tract enforcement, the empirical evidence
(1.489) (1.442) (1.56) (1.704) points to the failure of democracies to
Budget surplus/GDP -1.479' -1.5222 -1.952 -1.390*
(0.755) (0.727) (0.8011) (0.866)
enforce external contracts. A plausible
Growth rate of 0.022* 0.0879* 0.9957* 1.782* explanation is that democracies do not
Industrial countries (3.606) (3.481) (3.85) (4.171) efficiently utilise the borrowed resources
Political factors:
Democracy index 4.53 to be able to repay back. Running a cor-
(0.846) relation between Democracy Index and
Political freedom index 8.61
investment-GDP ratio gives a coefficient
(1.40)
Pluralism -4.35 of +0.20, implying that democracies in-
(1.1989)deed have a lower investment rate.
Party fractionalism index -0.3839 The measurement of political regime is
(0.113)
Number of observations 103 103 103 95 subject to many conceptual problems and
R2 0.46 0.50 0.38 0.37also suffers from inadequate data. Also,
Note: * indicates
the
significance
use of probit
at
analysis
5
forper
panel data c
in parentheses. set has many shortcomings that need to be

Economic and Political Weekly November 30, 2002 4821

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms
addressed. Considerable work also nieeds pacity. These measures are obtained from The second variable is the Party Fractiona-
to be done in terms of refinement of lism Index, defined as,
International Financial Statistics, published
n
measures of political variables. Despite
by the IMF.
these constraints on the empirical model- (4) The budgetary surplus of the central
F = 1- (t)2
i=1
ling of sovereign defaults, important government
con- is an indicator of the domestic
where ti is the proportion of members
clusions can be drawn from the empiricaleconomic situation. It is calculated as a
associated with the ith party in the lower
evidence. While the 'ability' of a sover-
ratio of the budgetary surplus and gross house of legislature and n represents the
eign to repay debts is measurable bydomestic
its product, both in nominal terms. total number of parties. The more frag-
It is obtained from the IFS.
structural variables, a measure of its 'will-
mented the countries political parties, the
ingness to pay' must include the will (5) of It is surprising that not many earlierhigher the index.
studies looked at the cost of borrowed
the political regime to enforce external (8) Two measures of political instabil-
contracts. funds or lending rates as an explanatory
ity are: change in effective executive (It
The less developed countries form a
variable. We took Libor (one-year London
is the numberof times in a yearthateffective
heterogeneous group in terms of country
inter bank offer rate on US dollar deposits)
control of the executive power changes
characteristics. The preceding empirical
as representing the lending rate. We also
hands. Such a change requires that the
analysis doesn't take into account considered
the six-monthly Libor and also new executive be independent of the pre-
lagged it one year. The cost of borrowed
differences in the less developed countries decessor) and Coups (The number of extra-
funds can also be calculated by deflating
in geographical region, country size, popu- constitutional or forced changes in the top
Libor by the export price ratio of each
lation, extent of poverty and social ineq- government elite and/or its effective con-
uity, historical background and other country
other for all the years under consider-
trol of the nation's power structure in a
factor. These factors may influenceation,the following the methodology usedgiven by year. Unsuccessful coups are not
patterns of borrowing and lending in thethe
World Bank. The data source for these
included.) The data is taken from the
variables was the IFS.
international credit markets. For instance, Bank's Data Set. l1D
it has been observed that the US has aided (6) Historically, international lending
Latin American countries facing a crisishas been following a cyclical movement, Address for correspondence:
and Germany has bailed out some of theso it is important to measure whether the vermasujata@ hotmail.com
east European nations. Japan has provided developed world is in recession or boom.
loans to Asian LDC's. This may be reflec- This can be captured by the growth rate Notes
tive of investor preference in lending to of the industrialised nations. These were
more familiar countries. Sometimes there obtained from the IFS. One of the most 1 Cuba, North Korea.
have been instances of lenders favouring commonly used measures of democracy 2isSee Verma (2000) for a complete review of the
their former colonies, for example, British in terms of political rights [Barro 1994]. existing literature.
3 See Balkan (1995).
lending to India. Thus historical ties matter. We have used political rights index as one
So far, the attention of economists has measure of democracy, but have also
References
been on whether default history matters. included a broader measure including both
But perhaps looking at a borrower's po- political and civil rights to form a democ-
Balkan, E (1995): International Bank Lending
litical history may shed some light on its racy index. The advantage of the second and Country Risk.
default behaviour. measure is that if for instance in a mon-
Barro, Robert J (1994): 'Democracy and Growth',
archy, political rights are low yet there is NBER Working Paper No 4909.
Appendix high civil freedom, and this will be re-Berg, A and Jeffrey Sachs (1988): 'The Debt
flected in this measure. The data are Crisis: Structural Explanations of the Country
Sources of Data
Performance', Journal of Development
obtained from the annual surveys by Economics 29, pp 271-306.
Raymond
(1) The level of total external debt as Gastil, etc. All the countries
Cline, W (1984): 'International Debt: Systematic
a proportion of GNP is a measure ofhavethe been rated on a scale from 1 to Risk
7, and Policy Response', Institute of
International Economics, Washington DC.
where 1 is the highest level of political
magnitude of indebtedness and is clearly
Eaton, Jonathan and Mark Gersovitz (198 1): 'Debt
rights/civil liberties. Thus the democracy
an important determinant of the probability
with Potential Repudiation: Theoretical and
of default. Total external debt/grossindex
na- takes values from 2 to 14, where the
Empirical Analysuis', Review of EcOnomic
level of democracy decreases with Studies
tional product is taken from the World an (48), pp 289-309.
Debt Tables for various years. increase in the index. Verma, Sujata (2000): 'The Architeture of the
(7) A definition of democracy based on International Capital Markets', PhD
(2) Other important debt-burden ratios
Dissertation: University of California, Santa
political and civil rights does not capture Cruz.
considered are interest payment/exports
and total debt service/exports. Both the
mea-process of political participation. Hence,
Kharas, H (1984): 'The Long Run Creditworthiness
two
sure the ability of the debtor country tomeasures of democracy are obtained of Developing Countries: Theory and Practice',
repay debt out of the export. The fromdatathe cross-sectional time series data Quarterly Journal of Economics, No 3,
source is the World Debt Tables.
pp 415-439.
archives: Bank's Data Set. Political Plu-
Ozler, S and H Huizinga (1991): 'Bank
ralism is a composite index indicating Exposure, Capital and Secondary Market
(3) The growth rate of gross domestic
polyarchy and is made up of effectiveness Discounts on Developing Country Debt',
product is a good measure of the repaying
of legislature, competitiveness of nomi-Working paper.
capacity of the debtor country. Per capita
GNP [used by Cline (1984)] and invest- Lee, Suk Hun (1991): 'Using Terms of
nating process, party coalitions and party
Rescheduling as Proxy for Partial Reneging
legitimacy. It is on a scale from 0- 10
ment- GDP ratio [used by Kharas (1984)] on LDCs Debt in a Test of Willingness to Pay
are alternative measures of repaying showing
ca- increasing pluralism from 0 to 10. Model'.

4822 Economic and Political Weekly November 30, 2002.

This content downloaded from 128.210.126.199 on Sun, 18 Sep 2016 00:52:58 UTC
All use subject to http://about.jstor.org/terms

You might also like