Review Journal

The Determinants of the composition of public debt in developing and

emerging market countries
By Kristine Forslund, Lycia Lima, Ugo Panizza
Review of Development Finance I (2011)207-222

Reviewed by Iman Sufrian

NPM 1206313974
A. BACKGROUND/INTRODUCTORY
The use of domestic public debts have had a long history of being a financing
source in large sample of countries back in 19 th century (Reinhart and Rogoff; 2008).
This particular field of study topic is important because Reinhart and Rogoff (2008)
also stated that the accumulation of a large domestic debt was often contributed to
external debt crisis and large inflationary periods that hindered the achievement of
welfare society in affected countries.
There have been numbers of research that attempt to study the determinants
of debt composition in developing and emerging countries. The authors of this paper
mentioned four studies related to this paper. Those four papers are Burger and
Warnock (2006), Claessens et. Al. (2007), Eichengreen and Luengnaruemitchai
(2004) and Borensztein et. al (2008). The findings of the previous four studies are as
follows:
a. Using a cross-sectional data of 49 countries which consists of 27 emerging
countries and 22 advanced countries, Burger and Warnock (2006) find that
policies and institutions play an important role in the development of the local
government bond market. In particular, they find that low inflation, rule of law, and
country size are positively correlated with the development of the domestic
government bond market and that the fiscal balance GDP growth are negatively
correlated with the size of the government bond market;
b. Claessens et al. (2007) use panel data of 36 countries (of which 12 are emerging
countries and 24 advanced countries) for the 1993-2000 period. They also find
that country size, size of the banking system (as measured by total deposit over
GDP), good institutions, low inflation, flexible exchange rates and fiscal burden
are positively correlated with the size of the domestic bond market. Moreover,
1

they find that countries with flexible exchange rates tend to have large domestic
bond markets.
c. Eichengreen and Luengnaruemtchai (2004) also use panel data techniques of 41
countries over the period 1990-2000. They find that country size and institutional
quality are positively associated with the development of the domestic bond
market. However, contrary to Claessens et al. (2007), they find that lower
exchange rate volatility is positively correlated with the size of domestic bond
market and argue that this might be due to the fact that a fixed exchange rate
lowers currency risk and may encourage foreign participation. They also find that
countries without capital controls ten to have larger bond market.
d. Borensztein et.al (2008) expand theirs sample and distinguish between the
determinants of the development of markets in government, corporate and
financial sector bonds rather that considering the bond market as a single
aggregate. They find that country size is significantly correlated with the size of
bond market but that the relationship is non-linear. They also find that bond
market development is positively correlated with trade openness, total public
debt, institutional quality, lack of capital control, and the privatization of pension
system. Moreover, they also find that level of domestic interest rate is negatively
correlated with market capitalization even though there is no significant
correlation between banking spread and the size of government bond market.
However, when focus on emerging economies, they find that country size become
very much less correlated with the size of public debt.
This paper has similar objectives with the above mentioned papers. The
objectives of this paper are as follow:
a. To document recent trends in the composition of public debt across a large
sample of developing and emerging market countries;
b. To test whether there are empirical regularities that explain the choice between
domestic and external public debt
However, there are difference between this paper and the four papers discussed
above. Those differences are:
a. The definition of public debt. The previous papers focused on bond market
development while this paper focuses on total public debt.

b. This paper covers up to 95 developing countries, of which 33 are low-income

countries. While the four previous papers examine a relatively small group of
emerging market countries. In addition, the analysis of the previous papers jointly
included developing and industrial countries (except the study conducted by
Borenztein at.al, 2008). The inclusion of industrial countries in dataset analyzed is
suspected as the cause of variance in bond market development between these
two groups of countries. Therefore this paper excludes the industrial countries
and focusing on developing countries.
c. The difference of methodology used in this paper compared to the previous four
papers. While previous studies either focused on cress-country differences
(Burger and Warnock, 2006) or jointly looked at cross-country and with countries
differences, this paper also used fixed effect estimation which allows for the
isolation of the determinants of changes within countries.

B. METHODOLOGY
Initially, the authors of this paper conducted a wide literature review other
researches about Public Debt.
Based on literature review, the author of this paper then constructed a model
that determined public debt composition. They also constructed controls variables
which classified into five different categories.
Subsequently, the model is empirically tested using the panel of developing
countries for the 1990-2007 period to capture both the domestic and external
component of public debt.
The authors of this paper used both random effect and fixed effect models.
They also tested for difference across groups of countries by splitting the sample
between low income developing countries and middle-income developing countries.
Based on the estimation of the model the authors then analyse the result and
draw conclusion about the determinants of domestic public debt in developing
countries.

C. JOURNAL CONTENT
Determinant of debt composition
The authors of this paper conducted an estimation of public debt compositions
by regressing domestic public (measured as a share of total public debt) over a set
of country characteristics. Control variables were classified into five different
categories:
a. Macroeconomic imbalances. The proxies of this variable were inflation, current
account balance, government balance, total public debt and its square and
exchange rate misalignment;
b. Country size and level of development. The proxies of this variable were GDP,
GDP per capita, M2 over GDP and corruption;
c. Crises and external shocks. The proxies of this variables were a banking crisis
dummy, a sovereign default dummy, a dummy that captures sudden debt
explosion, a dummy that captures sudden debt reductions, the growth rate of the
real exchange and the term of trade index;
d. Openness. The proxies of this variables were a measure of de facto trade
openness and a measure of de jure capital account openness;
e. Exchange rate regime. However, it was not clear about the indicator used to
proxy this variable.
In all regressions the authors controlled for global factors by including year fixed
effect. In the random effects regressions the authors also included a set of regional
dummies, where East Asia and Pacific was the excluded dummy.
The authors of this paper acknowledged two caveats in their analysis. Those
two caveats were:
a. Causality concerns. The authors realized that they could not claim causal
relationship from explanatory variables to debt composition.
b. The fact that supply and demand effects often could go in opposite directions.
Therefore, it was difficult to have a clear prediction on the relationship between
explanatory and dependent variables.

Hypothesis
Based on the model above, which describe the determinants of domestic debt
share, the author of this paper then constructed hypothesis about relationship of
each indicators with domestic debt shares. The expected relationships are described
as follow:
a. The authors expected to have negative relationship between Log of inflation and
domestic debt share;
b. Current account balance was expected to have positive correlation with domestic
debt share;
c. However, the relationship between domestic debt share and the government
balance was expected to be uncertain;
d. The relationship between the level of debt and domestic debt share was positive
and nonlinear relationship;
e. Exchange rate misalignment was expected to have positive relationship with the
domestic debt share;
f. Exchange rate was expected to have positive relationship with the domestic
debts share;
g. The expected signs for the second set of explanatory variables and the domestic
debts share were positive;
h. The effect of banking crises on domestic debt share was positive;
i. The default dummy was expected to have positive relationship with domestic
debt share;
j. The author expected to have a positive correlation between debt contraction
dummy and domestic debt share.
k. Sudden debt explosion was expected to have a positive sign in the model;
l. The term of trade was expected to have a positive correlation with domestic debt
share;
m. However, the author of this paper did not have clear expectation for the
relationship between domestic debt share;

Results
The authors of this paper presented the empirical result for both models:
random effect model and fixed effect model. Based on both empirical result model,
the authors of the paper then summarized the result as follow:
a. Control variables played a limited role in explaining cross-country differences in
the composition of public debt
b. Inflationary history had no statistically significant effect on the composition of
public debt. There was no evidence that countries with a history of high inflation
have lower shares of domestically issued debt
c. The composition of public debt was driven by the presence of capital control.
Countries with high level of capital controls there was no statistically significant
correlation between domestic debt share and inflationary history. In countries with
low and intermediate level of capital controls, inflationary history had a negative
and statistically significant impact on domestic debt share. This suggests that
capital control have a negative effect on the development of domestic debit
market in countries characterized by high policy credibility but might help
developing the bond market in countries characterized by low policy credibility.

D. SUMMARY
Based on above parts, it can be seen that the wide range of literatures studies
mentioned in this paper has provided some insights about the relationship between
each control variable and domestic debt share. However, since this study is lacked of
micro foundation, it is difficult to explain about how exactly each control variable and
explanatory variable interacts. The behaviour of the economic agents (such as
government and investors) that affect the variables in the model is not systematically
analysed. Moreover, some inconsistent result between this paper and the previous
papers about the relationship of certain control variables and the explanatory
variable will likely make the analysis of the result become even more difficult.
Further study that includes micro foundation of the relationship between
control variable and explanatory variable will improve the understanding of how
those variables interact.
6

Policy Implication
Nevertheless, this study may have some important policy implication, in
particular with regard to the finding that inflationary history has no statistically
significant effect on the composition of public debt. This paper explores the
determinants of this finding and shows that the results are driven by the presence of
capital controls.
This paper shows that in countries with high level of capital controls there is
no statistically significant correlation between domestic debt share and inflationary
history. However, in countries with low and intermediate level of capital controls, the
authors found that inflationary history has a negative effect on the development of
the domestic debt market in countries characterized by high policy credibility but may
help developing the domestic bond market in countries characterized by low policy
credibility.
Although, the authors recognize that the positive effect is likely to vanish if
countries that impose capital control continue to adopt irresponsible policies.
However, in countries that are seriously to improve policy credibility, capital control
can be considered as an additional policy instrument in building a local debt market.

REFERENCES
Kristine Forslund, Lycia Lima, Ugo Panizza ,The Determinants of the composition of
public debt in developing and emerging market countries, Review of
Development Finance I (2011)207-222