Domestic Debt in Low-Income Countries Andrea F. Presbitero∗

September 26, 2011 Very preliminary version - comments welcome

Abstract The potential consequences of the development of domestic debt markets in Low-Income Countries (LICs) are extremely relevant for policy-makers and international financial institutions, especially in light of a scaling-up of public investment in infrastructures. This paper introduces a new dataset on the stock of domestic debt in LICs over the period 1970-2010. With respect to the existing dataset, this one put together information from several different sources to expand the country and time coverage, devotes a careful attention to the problem of the zeros and addresses some inconsistencies between the existing datasets. The descriptive analysis of the evolution of domestic debt in LICs, especially over the last two decades, points out some interesting patterns. The reliance on internal financing has partially offset the reduction in external debt granted by bilateral and multilateral debt relief initiatives. Domestic debt increased at a lower and less volatile pace in countries with better policies and institutions. This pattern is mirrored by a greater capital accumulation, a faster financial development, and a stronger output growth. This descriptive evidence supports the hypothesis that the development of the domestic debt market can bring benefits only in presence of a stable macroeconomic environment, lack of political uncertainty and a developed financial system.

JEL Classification: E62, O11, O23 Key words: Domestic debt, Public debt, Low-Income Countries.

1

Introduction

Long term economic development in many Low-Income Countries (LICs), especially in Sub-Saharan African countries, has been impaired by the accumulation of large external public debts, as a result of several exogenous internal and external shocks, bad institutions, lack of fiscal discipline and bad policies. Starting from the mid-1990s, debt relief initiatives promoted by multilateral donors led to further debt reductions in addition to traditional bilateral debt relief (Arnone and Presbitero, 2010). As a result, in 2009, public and publicly guaranteed external debt in LICs was 25 percent of GDP, far lower than the values of 84 percent in 1994. During the same period, thanks to several factors, including favorable external conditions and strengthened macroeconomic policies and institutions, LICs as a whole experienced a decade of strong growth before the 2009 global financial crisis (Schindler et al., 2011; Pattillo, Poirson and Ricci, 2011). Even if there is not fully convincing evidence that debt relief boosted investment and growth in poor countries, debt reduction might have contributed easing institution-building and reducing debt ∗ Andrea F. Presbitero, Department of Economics and Social Sciences – Università Politecnica delle Marche (Italy), Money and Finance Research group (MoFiR) and Centre for Macroeconomic and Finance Research (CeMaFiR). E-mail: [email protected]; personal webpage: https://sites.google.com/site/presbitero/. I gratefully acknowledge financial support from the World Bank Knowledge for Change Program. I also thank Ali Abbas and Ugo Panizza for sharing their data and Aart Kraay for insightful suggestions.

1

overhang, at least in some countries (Presbitero, 2008, 2009; Pattillo, Poirson and Ricci, 2011). For these reasons, the increase of government domestic borrowing which occurred in several poor countries is often seen as a potential source of concern for debt sustainability and output growth (Arnone and Presbitero, 2010). External debt reduction has been partially offset by an increase in domestic debt following programs aimed at fostering the development of domestic markets for government securities (International Monetary Fund and The World Bank, 2001; UNCTAD, 2004; Borensztein, Levy Yeyati and Panizza, 2006). This trend, which became clear especially since the mid 1990s, has been already discussed by Panizza (2008) and Panizza (2010) with respect to a sample for developing countries. Notwithstanding the potential impact of domestic debt on output growth, with few exceptions (Abbas and Christensen, 2010; Presbitero, 2010), the empirical research on the growth effect of public debt in poor countries has focused on external public debt, mainly because of: 1) the relative small share of domestic debt in total public debt in developing countries, and 2) the lack of available and reliable comparable data on domestic debt for a sufficiently large sample of countries (Abbas and Christensen, 2010)1 . In fact, data availability on domestic debt in developing countries is particularly limited and for many countries the public debt data published in the official IMF and World Bank datasets “are plagued by missing observations, limiting their use for empirical research” (Jaimovich and Panizza, 2010). To overcome these limitations, this note keeps on the quest for better data (Panizza, Sturzenegger and Zettelmeyer, 2010) in domestic government debt, focusing exclusively on LICs, and 1) presents a new, comprehensive and up-to-date dataset on domestic debt (Section 2), 2) reviews some of the reasons behind the increase in internal financing in several poor countries and its potential effects (Section 3), and 3) discusses some interesting relations between domestic debt, external debt, and macroeconomic and institutional outcomes (Section 4). Future possible lines of research are introduced in the concluding section.

2 2.1

The dataset The existing datasets

This section reviews the existing sources gathering data on domestic debt stocks in LICs and introduces the new dataset. Few datasets collect detailed data on the structure of domestic and external public debt in emerging markets (Guscina and Jeanne, 2006), Middle East (Fouad et al., 2007), and Latin American countries (Cowan et al., 2006), while other recent databases gather information on domestic debt or total public debt in a broader set of industrialized and developing countries2 . Nonetheless, these sources differ in the definition of domestic debt, it being referred to central or general government3 , including just securitized debt or also advances from the Central Bank and commercial banks, and excluding or not State Owned Enterprises (SOEs) debt and contingent liabilities. One further issue is the definition of the zeros in the dataset, which could indicate either the absence of a domestic bond market, or a missing value. In the following paragraphs, the most important datasets will be presented, pointing out for each of them the country and time coverage, the domestic debt definition and the sources used, and their potential weaknesses. A summary of the most relevant information is presented in Table 1. 2.1.1

Christensen (2005) African Database

Christensen (2005) gathers data on domestic debt from IMF country reports, country desk databases and, when necessary, country desk economists and Central Banks, for a sample of 27 non-CFA SubSaharan African countries (of which 16 LICs) over the period 1980-2000. Domestic debt is defined as gross securitized domestic government debt, composed of treasury bills, development stocks and bonds. Hence, it excludes domestic debt arising from domestic arrears accumulation and direct advances from 1 Poor data availability, however, can be itself due to the limited interest of international financial institutions, at least until few years ago, on the issue of domestic debt in emerging and developing countries and to the lack of transparency of many governments in publishing time series data on domestic debt (Reinhart and Rogoff, 2009). 2 The best source on public debt structure in industrialized countries is the Task Force on Finance Statistics (TFFS). 3 The general government sector consists of all government units and all nonmarket nonprofit institutions that are controlled and mainly financed by government units, comprising the central, state, and local governments. The general government sector does not include public corporations or quasi-corporations.

2

the central bank and commercial banks4 . According to Christensen (2005, p. 520), “Angola, Botswana, the Democratic Republic of the Congo, Mozambique, and São Tomé and Príncipe did not have domestic government debt markets at the time of [data] collection” 5 . This dataset, even if rich of information – also on debt structure – and with few missing values, has a very limited country coverage and has not been updated after 2000. 2.1.2

Abbas and Christensen (2010) Domestic Debt Database

This dataset collects information on public domestic debt and total public debt on 143 LICs and Emerging Markets (EMs), over the period 1970-20076 . Data are extracted from the IFS monetary survey, usually published in the IMF country statistical appendices and staff reports. The definition for domestic debt used is commercial banks’ gross claims on the central government plus central bank liquidity paper. Specifically, the stock of public sector domestic debt is built aggregating the lines 22a, 42a, 20c and 40c of the IFS (Abbas, 2007). However, as the authors themselves state, “it is quite difficult to extract continuous and consistent DD series from these documents, as the coverage and timeliness of the appendices and staff reports varies significantly from program to non-program countries and from program to non-program “periods” for each individual country”. One weakness of this definition of domestic debt is that focusing exclusively on bank holdings of government debt could result in an underestimation of domestic debt, since the part held by retail investors and non-banking institutions is neglected. 2.1.3

IMF Historical Public Debt Database (HPDD) and the Public Sector Debt Statistics (PSDS)

Abbas et al. (2010) have recently built a comprehensive dataset at the IMF on total public debt, gathering information from several sources on the ratio of general government public debt over GDP for an unbalanced panel of 174 countries. The time coverage is extremely long, even if the country covered reduces to less than 100 going back to the 1970s and LICs data coverage starts generally in 1970. This dataset has the advantage of being built without recourse to extrapolations and interpolations, but the merging of different sources makes data not easily comparable between countries and over time (in this case, breaks - they are 81 - are highlighted in the dataset). One specific critical point is the heterogeneous definition of public debt across countries: the authors state to rely on general government debt, but this definition blends with central government going back in time and also with respect to low- and middleincome countries. In fact, data for a large set of (developing) countries since the 1970s are taken from the Abbas and Christensen (2010) dataset and from the WEO, both referring to central government debt. The main weakness of the HPDD is that domestic debt can be derived only as the difference between total public debt and external public debt, since the data refer exclusively to total public debt7 . 2.1.4

Panizza’s Public Debt Databases

Panizza has built an extensive dataset on the ratios of domestic and external public debt over GDP for 176 countries from 1970 to 20108 . This database is still ongoing and updates the previous datasets published by Panizza with several coauthors (Cowan et al., 2006; Jaimovich and Panizza, 2006; Panizza, 4 Other than information on debt stocks, for a sub-sample of countries the dataset includes also data on interest payments (available for 9 to 21 countries, depending on years) and on the nominal domestic interest rates (available for 14 to 22 countries, depending on years). Besides, there are information on bond holders and maturities for a cross-section referred to 2000. 5 Similarly, CFA countries generally did not have domestic debt markets, explaining their exclusion from the sample. 6 The dataset has been kindly provided by S.M. Ali Abbas. In their 2010 paper, Abbas and Christensen make explicit reference and use a previous smaller version of the dataset which collects information on public domestic debt and total public debt on 93 LICs and Emerging Markets (EMs), over the period 1975-2004. 7 Related to this project, the Public Sector Debt Statistics (PSDS) database, jointly developed by the World Bank and the IMF and launched in December 2010, brings together detailed quarterly public sector debt data of selected countries, initially mainly developing and emerging market economies. Country participation in the database is voluntary and, currently, 62 countries (among which 8 are LICs) have agreed to participate and 34 provided data. The database starts in 2000, but actual data are often present just for the most recent years, and it is updated quarterly. Gross Central Government debt is broken down by maturity, currency of denomination and residency. At the time of writing, Bangladesh, Kenya, Nepal and Uganda are the only LICs with available data on gross Central Government debt position by residency of the creditor in 2009 and/or 2010. 8 I gratefully thank Ugo Panizza for sharing his data on public debt.

3

2008; Jaimovich and Panizza, 2010). This version of the database has been built merging information from several different sources, starting with the IFS and WDI and then complementing them with data drawn from IMF Article IV documents, the GDF, Central Banks and Ministry of Finance websites, the Economist Intelligence Unit (EIU, especially for the 2010 update) and the Cowan et al. (2006) and the ECLAC/ILPES databases. With respect to the IMF HPDD (Abbas et al., 2010), the data compiled by Panizza have the great advantage of providing the breakdown of public debt into the external and domestic components. In particular, when not directly available, domestic debt is calculated as the difference between total and external debt (when total debt is lower than external debt, the latter is set equal to the former and domestic debt is set to zero). In general, the database provides useful indication about the lack of a domestic debt markets, since the zeros are actual zeros and should not be interpreted as missing values. However, a careful inspection of the actual data has shown that in several circumstances this is not the case. In order to have a broad coverage, this database shares with the HDPP the heterogeneous definition of domestic debt. The latter, in fact, is defined taking central government debt as a benchmark, no matter who the holders are, but using data on the general government and the non-financial public sector debt when the benchmark is not available9 . 2.1.5

Presbitero (2005) Domestic Debt Database

This dataset originally included 34 LICs and 38 MICs, for which there are reliable data on government domestic debt over the period 1979-2004. The dataset has been recently updated with information on LICs domestic debt up to 2010. The time coverage varies across countries and increases over time10 . The main sources of data are the Article IV Consultation and the IMF Staff Country Reports, together with Central Banks and Ministry of Finance publications and websites. IFS data are used only when they match-up (or show a very close similarity) with the existing data in the overlapping years, in order to extend the time dimension of the database. The definition of Central Government domestic debt is as in Christensen (2005) and comprises only securitized debt. This definition is available for 65 countries and includes mainly by Treasury Bills, Bonds, notes and government stocks, even if in some countries there are other special securities and consolidated debt. The main benefit of this aggregation is cross-country comparability, even if the presence of different levels of government could lead to the underestimation of actual domestic debt in fiscal decentralized countries11 . 2.1.6

Reinhart and Rogoff (2010) Debt Data

As part of an extensive project on sovereign and banking debt crisis, Reinhart and Rogoff (2010) have published the longest dataset for central government debt (as a share of GDP), even if for some countries there are information also (or exclusively) on general government debt. The dataset generally reports series on total public debt and on private plus public gross external debt, so that domestic public debt can not be calculated as a difference between the two series. According to the authors, the database, extensively described by Reinhart and Rogoff (2009), covers “annual data from as early as 1692 for the UK and 1719 for Sweden (depending on the country) to 2010 (updated with the World Bank’s July 2010 Quarterly External Debt Statistics and individual mostly official country sources, as detailed for public debt) for 70 countries” 12 . The database has been constructed merging data from a very broad 9 By contrast, the dataset published by Jaimovich and Panizza (2010, p. 2) on 89 countries over 1970-2005 “refer to gross central (as opposite as to general) government debt and therefore may not be fully comparable across countries with different levels of fiscal centralization. Second, unlike in Cowan et al. (2006), these data come from secondary (in most cases, official) sources and, hence, may reflect different definitions of gross debt and central government, or official misreporting – a drawback, it is worth noting, shared by IFS and WDI official statistics”. 10 Data have been put together in 2005 at PRMED at the World Bank as part of a research project coordinated by Jean-Francois Perrault and in which also Volodymyr Tulin was involved. For a more broader description of the dataset, see Arnone and Presbitero (2010). 11 It is not always possible to disentangle if the aggregate debt is made exclusively by securities. In some countries local governments and SOEs debt are highly relevant, as well as loans and other non-securitized instruments, making securitized central government debt the lower bound of the actual debt burden. For those reasons, a second dataset covers 72 countries and adopts a broader definition of domestic debt, following the approach used by the WEO. 12 See: http://terpconnect.umd.edu/ creinhar/Courses.html.

4

set of different sources, especially going back in time (see Reinhart and Rogoff, 2009, Appendix A.2, for country-specific detailed references), and has occasionally employed interpolations and other data transformations to build continuous series. Out of the 40 LICs listed in Table 1, the Reinhart and Rogoff (2010) data cover only six countries (Central African Republic, Ghana, Kenya, Myanmar, Zambia, and Zimbabwe), reporting data on total gross general/central government debt collected from several sources using different definitions of debt. Hence, given the limited coverage and, for the last decade, the adoption of the same source discussed so far, the Reinhart and Rogoff (2010) dataset is not included in Table 1. 2.1.7

The 2011 LICs domestic debt dataset

On the ground of the existing data on domestic debt, a final data set has been assembled, starting from the Panizza’s database, but including some significant changes. When possible, missing values has been filled with data collected from the IMF Article IV reports. For other countries, original missing values could be converted in actual zeros, building on the evidence provided by other sources or on the Deléchat et al. (2010) discussion of the development of domestic financial market. However, for several countries for which capital markets do not exist, such as for the Comoros and Eritrea (Deléchat et al., 2010, Table 2), Abbas and Christensen (2010), IMF country reports and other documents show the presence of domestic debt, explicitly reporting either domestic debt stocks or interest payments on domestic debt in the government financial operations Table. Hence, in building the final dataset, information drawn from IMF sources are taken as base reference. Finally, for several countries, data have been modified since the original ones significantly and strangely underestimate domestic debt with respect to other sources, often reporting domestic debt ratios equal to zero. In those cases, the data reported by Abbas and Christensen (2010) (whose definition, by contrast, should underestimate domestic debt with respect to the broader one adopted by Panizza), IMF staff reports and the updated Presbitero dataset are used to replace the Panizza’s data. In sum, the final dataset targets 44 Low-Income countries over the period 1970-2010 and includes information on the stock of government domestic debt, at best, for 41 countries (Somalia, North Korea and Afghanistan are the ones without data). All the other countries report domestic debt data for at least 10 years (Liberia), with 26 countries having 30 or more non-missing observations (Table 1). A visual representation of country coverage and of the increase of the average domestic debt over GDP ratio (GDP weighted) over time is summarized in Figure 113 .

3

Domestic Debt: Benefits and Risks

Although financial globalization involved mainly advanced and emerging economies, it also affected poor countries, even if to a minor extent (International Monetary Fund, 2007). Domestic public debt started increasing in LICs from the mid-1990s, in coincidence with an upsurge in financial liberalization: the Chinn-Ito Financial Openness index in LICs increased from -0.98 in the decade 1985-1994 to -0.58 in 1995-2004 (Chinn and Ito, 2010). The greater reliance on internal financing in emerging markets was a response to the financial crisis that hit many developing countries during the nineties. LICs, instead, were forced to tap the domestic securities market to finance structural current account and fiscal deficits, given the lack of access to external commercial lending, the insufficient foreign assistance and the impossibility to adopt expansionary monetary policy for countries under an IMF lending program (Arnone and Presbitero, 2010). In some countries, the recourse to internal borrowing was also motivated by the necessity to sterilize aid inflows, at the cost of rising interest rates (Hussain, Berg and Aiyar, 2009). Domestic debt accumulation has also been the result of explicit programs aimed at fostering the development of domestic markets for government securities (International Monetary Fund and The World Bank, 2001; UNCTAD, 2004; Borensztein, Levy Yeyati and Panizza, 2006). The rationale of these policies lies is the advantages that an internal credit markets could provide to the government and the economy. Deficit financing with domestic-currency denominated government securities is likely to reduce the vulnerability of a country to reversal in capital flows and limit the build-up of foreign-currency denominated debt, relieving the risks of exchange rate devaluations and monetary financing of budget deficits. A liquid and deep market with long-term instruments could mobilize domestic savings into the formal 13 The

dataset is available on the website: https://sites.google.com/site/presbitero/homepage/data.

5

Figure 1: Domestic Debt in Low-Income Countries: Evolution and Data Coverage 40

30 15

20

10 10

09 20

06 20

03 20

00 20

97 19

94 19

91 19

88 19

85 19

82 19

79 19

76 19

19

19

73

0

70

5

Number of countries with data

Domestic debt (% GDP)

20

Domestic Debt (GDP weighted) over GDP Double−exponential trend in the domestic debt ratio Countries with non−missing data point per year

Notes: Elaboration on the domestic debt dataset.

financial system, by providing domestic savers with an alternative to investing abroad. This could help reducing the cost of government financing and mitigate capital flight, stimulating capital accumulation. Finally, the shift from external to domestic borrowing could reduce the government’s external dependence, promoting political accountability and institutional reforms. The optimal debt structure should take into account the trade-off in terms of risks, costs and externalities between external and internal borrowing (Panizza, 2010; Abbas and Christensen, 2010). The domestic-bond maturity mismatch could substitute for the currency mismatch of foreign debt, given that many developing countries are unable to issue long-term government securities at a reasonable interest rate. The high costs of domestic debt, especially compared to concessional external financing and foreign aid, could crowd out public investment. In addition, in countries where a large share of domestic debt is non-indexed, the strong incentive to monetise government deficit may cause hyper-inflation episodes and trigger external debt crisis. Government borrowing could also affect the real sector through its effect on the financial sector. First, public sector borrowing could harm financial deepening, since developing country banks investing in government debt are more profitable but less efficient than the average (Hauner, 2006). Second, a growing domestic debt could crowd out lending to the private sector, especially to small and medium enterprises and rural borrowers, which have to rely on domestic savings and bank credit. This problem is intensified when banks and institutional investors are “forced” by the government to absorb “too much” domestic debt (Hanson, 2007). However, the International Monetary Fund (2005) found limited evidence that government recourse to domestic financing crowds out private sector borrowing in LICs. The balance between benefits and costs of domestic debt depends on the presence of prerequisites, such as a stable macroeconomic environment, an efficient money market, a broad investor participation and the presence of a sound legal, regulatory and supervisory framework (Guscina, 2008; Kose et al., 2009). However, macroeconomic and policy volatility are major issues in many LICs, making domestic debt a particular source of concern which should be addressed (Arnone and Presbitero, 2010).

6

4

Domestic Debt, External Debt and Macroeconomic Outcomes

4.1

The average trends

The establishment of the causal effects of domestic debt on the economy goes beyond the scope of this paper, whose aim is limited at presenting some stylized facts about domestic debt accumulation in LICs. On average, the stock of government debt fluctuated around 10 percent of GDP during the 1970s and the 1980s, then it started an upward trend in the last two decades, reaching 19.2% in 2010 (Figure 1). A first interesting issue regards the negative correlation between the evolution of domestic and external debt since the mid-1990s (Figure 2, panel a). Because of debt relief, external debt followed an opposite trend to domestic debt and shrunk from 80% of GDP in 1994 to almost 20% in 2009. As a result, the composition of total public debt was substantially altered and the share of domestic debt in total public debt increased from 11% in 1994 to 45.5% in 2009. The diagram reported in panel b does not corroborate the hypothesis of a reduction in bank credit due to government borrowing, but it might support the hypothesis the development of a domestic debt market could strengthen financial markets (Abbas and Christensen, 2010). In LICs, the sharp increase of domestic credit to the private sector, a measure of financial development, occurred over the last decade and followed the accumulation of domestic debt. Panels c, d and e of Figure 2 seem to suggest that the upward trend of the stock of domestic debt did not hinder domestic savings, capital accumulation and output growth, even if the lack of any counterfactual and of any control for covariates does not allow to infer any (lack of) causal relation. The diagram showed in panel c confirms that domestic debt can help mobilize national savings, even if the direction of causality is likely to work in both ways (Abbas and Christensen, 2010). Even if from panel d it is not possible to conclude that domestic debt nurtured capital accumulation, nonetheless the diagram suggests that debt service did not crowd out investment and that there has not been a debt overhang. On the contrary, domestic debt goes along with a boost in capital accumulation, indicating that internal financing could has been productively used to finance a much-needed scaling-up of investment in poor countries14 . Similarly, the build up of domestic debt does not seem to have hold back the growth acceleration of the last 15 years (panel e). Finally, the descriptive evidence presented in panel f suggest that the fear of inflationary episodes following domestic debt accumulation is exaggerated, since LICs as a group were able to stabilize inflation below 10 percent over the last decade, notwithstanding the increase in domestic debt. One possible interpretation of these patterns is consistent with the hypothesis that government domestic borrowing has been productively used to finance public investment and has fostered financial development, triggering private capital as well, as indicated by the increase in credit to the private sector. This mechanism might have contributed to the episodes of growth accelerations which characterized several poor countries over the last decade. As seen in the previous section, this story may be plausible. Nevertheless, it contrasts with the weak policy and institutional framework of many LICs, which should shift the balance in favor of the costs of domestic debt. Therefore, in the next section we are going to inspect whether the different institutional setting uncovers a more heterogeneous picture.

4.2

The role of the institutional setting

In order to assess whether the policy and institutional framework has an effect on domestic debt accumulation and on its potential impact on the economy, we classify countries in two groups according to their quality of policies and institutions, as measured by the Country Policy and Institutional Assessment (CPIA) index calculated by the World Bank15 . In particular, given that the index is available since 1987, each country is classified as having good (bad) policies and institutions whether the average 14 The

lack of data does not allow to validate the possibility of a crowding out of public investment. However, if public and private investment are complementary (Greene and Villanueva, 1991), considering total investment can provide a good approximation. 15 The CPIA score is a composite index measuring the extent to which country policies and institutions create a good environment for growth and poverty reduction. The CPIA indicators reflect the World Bank staff professional judgment, based on country knowledge, policy dialogue, and relevant publicly available indicators. To overcome criticisms related to lack of transparency and objectivity, in 2004 the CPIA process and methodology were extensively reviewed by an external independent panel and, starting from 2005, these data are fully disclosed and published in World Development Indicators (International Development Association, 2007).

7

Figure 2: Domestic Debt, External Debt, and Macroeconomic Outcomes 20

100

20

Private credit (% GDP)

09

06

20

03

20

00

20

97

20

94

19

91

19

88

19

85

19

Investment (% GDP)

09

06

20

20

03 20

00 20

97 19

94 19

91 19

88 19

85 19

20

10 10

Domestic Debt (GDP weighted) over GDP

Per capita real GDP growth (annual percentage change)

GDP Deflator (annual %)

09

06

20

20

03 20

00 20

97 19

94 19

91 19

88 19

85 19

82

79

19

19

76 19

73 19

70

0

Domestic Debt (GDP weighted) over GDP

(e) domestic debt and growth

Inflation rate

Domestic debt (% GDP)

30 15

19

09

06

40

5

20

20

03 20

00

97

20

94

19

91

19

88

19

19

19

19

85

−5

82

5

20

Per capita real GDP growth

10

79

82

(d) domestic debt and investment

0

19

79

Investment (GDP weighted) over GDP

15

76

19

Domestic Debt (GDP weighted) over GDP

Gross domestic savings over GDP

5

73

19

76

70 19

09

10

Domestic Debt (GDP weighted) over GDP

20

19

82

15

(c) domestic debt and savings

19

19

10

5

20

06 20

03 20

00

97

20

94

19

91

19

88

19

19

85 19

82 19

79 19

76

73

19

19

70

5

20

19

10

25 15

73

10

30

19

15

Domestic debt (% GDP)

15

20

Domestic savings (% GDP)

Domestic debt (% GDP)

(b) domestic debt and private credit

5

Domestic debt (% GDP)

79

Credit to the private sector over GDP

20

70

19

Domestic Debt (GDP weighted) over GDP

PPG External Debt (GDP weighted) over GDP

20

19

76

Domestic Debt (GDP weighted) over GDP

(a) domestic and external debt

19

19

70 19

09

06

5

20

03

20

00

20

97

20

94

19

91

19

88

19

85

19

82

19

79

19

76

19

73

19

19

70 19

20

10

20

5

10

73

40

30

19

10

40 15

19

60

Domestic debt (% GDP)

80 15

External debt (% GDP)

Domestic debt (% GDP)

50

(f) domestic debt and inflation

Notes: Elaboration on the domestic debt dataset. Data are all GDP weighted averages. To limit effects due to sample composition, only the observations with non missing values in all variables (domestic debt, external debt, investment, savings, GDP growth, inflation and private credit) are retained. The total number of observation is 944. With respect to inflation, 9 outliers are dropped.

score over the period 1987-2009 is above (below) 3.25, which is the sample median and corresponds to one of the institutional thresholds used in the World Bank-IMF debt sustainability framework (DSF)

8

to assess public debt sustainability (International Development Association and International Monetary Fund, 2010). A first finding which emerges from this sample split is the different trend of domestic debt accumulation in the two sample of countries (Figure 3). In the one with bad policies and a weak institutional framework, domestic debt increased substantially on the onset of debt relief initiatives, probably as a result of feeble alternative sources of financing, then the average stock of debt slightly declined and fluctuated around 15% of GDP. By contrast, at the end of the 1990s domestic debt was lower and more stable in countries with better institutions and policies, at least up to 2008-2009, when it increased considerably as a result of the global crisis. Figure 3: Domestic debt, by policy and institutional quality Bad policies and institutions

Good policies and institutions

Domestic debt (% GDP)

20

15

10

5

Domestic Debt (GDP weighted) over GDP

08 20

05 20

02 20

99 19

96 19

93 19

90 19

87 19

08 20

05 20

02 20

99 19

96 19

93 19

90 19

19

87

0

Quadratic fit

Notes: Elaboration on the domestic debt dataset. Data are all GDP weighted averages. To limit effects due to sample composition, only the observations with non missing values in all variables (domestic debt, external debt, investment, savings, GDP growth, inflation and private credit) are retained. The total number of observation is 944. Countries are classified with bad (good) institutions whether the overall CPIA index is lower than (equal to or greater than) 3.5.

These contrasting trends are reflected in the different paths of savings, capital accumulation, output growth and financial development, which are much more correlated with the dynamics of domestic debt in countries with a better institutional framework than in the ones with bad policies and institutions (Figure 4). From the 1990, the savings and investment rates are higher and less volatile in the sample of LICs with good policies and institutions (panels a and b). The strong increase in the private credit over GDP observed in the whole sample is fully attributable to countries with an average CPIA score equal to or above 3.25 (panel c): in those countries, the build-up of domestic debt did not crowd out credit to private sector, consistently with the International Monetary Fund (2005) evidence on a LICs group of 15 mature stabilizers. Hence, in the sample of good institutions countries real per capita GDP growth is stronger and less volatile (panel d ). From the mid-1990s, the upward trend in domestic debt goes hand in hand with fixed capital accumulation and a stronger output growth. By contrast, in countries with an average CPIA score below 3.25, domestic debt accumulation is reflected in a slower increase in the investment rate, in the lack of financial deepening and in a more volatile and lower growth rate.

9

Figure 4: Domestic debt and macroeconomic outcomes, by policy and institutional quality Bad policies and institutions

Good policies and institutions

Bad policies and institutions

Good policies and institutions

20

30

20

5

25

15 20

10

Investment (% GDP)

10

10

Domestic debt (% GDP)

15 15

Domestic savings (% GDP)

Domestic debt (% GDP)

20

15

5 5

Domestic Debt (GDP weighted) over GDP

Domestic Debt (GDP weighted) over GDP

Gross domestic savings over GDP

Investment (GDP weighted) over GDP

(a) domestic debt and savings Bad policies and institutions

20 11

20 08

20 05

20 02

19 99

19 96

19 93

19 90

20 11 19 87

20 08

20 05

20 02

19 99

19 96

19 93

19 90

19 87

20 11

20 08

20 05

20 02

19 99

19 96

19 93

19 90

20 11 19 87

20 08

20 05

20 02

19 99

19 96

19 93

19 90

19 87

10

(b) domestic debt and investment

Good policies and institutions

Bad policies and institutions

Good policies and institutions

60

6

20

20

10

20

5

3

15 0

GDP growth

30

Domestic debt (% GDP)

40

15

Private credit (% GDP)

Domestic debt (% GDP)

50

10 −3

5 −6

Domestic Debt (GDP weighted) over GDP

Credit to the private sector over GDP

Per capita real GDP growth (annual percentage change)

(c) domestic debt and private credit

08

11 20

05

Domestic Debt (GDP weighted) over GDP

20

20

99

02 20

19

93

90

96 19

19

19

1 19 1 87

20

05

08 20

20

02 20

99 19

96 19

90

93 19

19

87 19

11 20

05

08 20

20

02 20

96

99 19

19

90

93 19

19

87

19

11 20

08 20

02

05 20

20

96

99 19

19

90

93 19

19

19

87

10

(d) domestic debt and growth

Notes: Elaboration on the domestic debt dataset. Data are all GDP weighted averages. To limit effects due to sample composition, only the observations with non missing values in all variables (domestic debt, external debt, investment, savings, GDP growth, inflation and private credit) are retained. The total number of observation is 944. Countries are classified with bad (good) institutions whether the overall CPIA index is lower than (equal to or greater than) 3.5.

These patterns are consistent with the fact that both domestic and foreign investors require a stable and credible government and sound fiscal and monetary policies to hold domestic securities. Poor countries with a healthy institutional framework are better equipped to develop an efficient domestic market for government debt. This translates into a greater credit availability to the private sector, a larger volume of savings and capital accumulation, a higher investment efficiency and a stronger economic growth, consistently with the findings discussed by Abbas and Christensen (2010). The lack of some prerequisites in terms of monetary and fiscal policy and institutional stability, instead, are likely to offset the potential benefits of the development of domestic debt markets.

5

Conclusions

This paper introduces a new dataset on the stock of domestic debt in Low-Income Countries over the period 1970-2010. With respect to the existing dataset, this one put together information from several different sources to expand the country and time coverage, devotes a careful attention to the problem of the zeros and addresses some inconsistencies between the existing datasets. The descriptive analysis of the evolution of domestic debt in LICs, especially over the last two decades, points out some interesting patterns. The reliance on internal financing has partially offset the reduction 10

in external debt granted by bilateral and multilateral debt relief initiatives; thus, in 2010 the shares of domestic and external debt in low-income countries as a group were similar. Domestic debt increased at a lower and less volatile pace in countries with better policies and institutions. This pattern is mirrored by a greater capital accumulation, a faster financial development, and a stronger output growth. This descriptive evidence is consistent with the empirical findings provided by Abbas and Christensen (2010) for a sample of low-income countries and emerging markets. The positive correlation between domestic debt and investment and output growth, conditioned on sound policies and institutions, supports the hypothesis that the development of the domestic debt market can bring benefits only in presence of a stable macroeconomic environment, lack of political uncertainty and a developed financial system (Guscina, 2008). The potential consequences of the development of domestic debt markets in poor countries are extremely relevant for policy-makers and international financial institutions, especially in light of the muchneeded scaling-up of public investment in infrastructures (Buffie et al., 2011). Given the explicit descriptive nature of the evidence presented in this paper, a more careful econometric analysis of the potential benefits and risks of domestic debt, extended to the crisis years, could be particularly useful. Besides, this dataset could be used to investigate: 1) how domestic debt accumulation affects total public debt sustainability in LICs, and 2) which factors are more likely to be related to the development of domestic markets for government securities in poor countries.

References Abbas, Ali S. M. 2007. “Public Domestic Debt and Economic Growth in Lower Income Countries.” Department of Economics - University of Oxford. Abbas, Ali S. M. and Jakob E. Christensen. 2010. “The Role of Domestic Debt Markets in Economic Growth: An Empirical Investigation for Low-Income Countries and Emerging Markets.” IMF Staff Papers 57(1):209–255. Abbas, S. M. Ali, Nazim Belhocine, Asmaa A. ElGanainy and Mark A. Horton. 2010. A Historical Public Debt Database. IMF Working Papers 10/245 International Monetary Fund. Arnone, Marco and Andrea F. Presbitero. 2010. Debt Relief Initiatives - Policy Design and Outcomes. Global Finance Farnham: Ashgate. Borensztein, Eduardo, Eduardo Levy Yeyati and Ugo Panizza. 2006. Living with Debt - How to Limit The Risks of Sovereign Finance. Cambridge, MA: Harvard University Press. Buffie, Edward F., Andrew Berg, Catherine Pattillo, Rafael Portillo and Luis-Felipe Zanna. 2011. “Public Investment, Growth, and Debt Sustainability: Putting Together the Pieces.” International Monetary Fund. Chinn, Menzie D. and Hiro Ito. 2010. “A New Measure of Financial Openness.” Journal of Comparative Policy Analysis 10(3):309–322. Christensen, Jakob. 2005. “Special Data Section Domestic Debt Markets in Sub-Saharan Africa.” IMF Staff Papers 52(3):7. Cowan, Kevin, Eduardo Levy Yeyati, Ugo Panizza and Federico Sturzenegger. 2006. Sovereign Debt in the Americas: New Data and Stylized Facts. RES Working Papers 4480 Inter-American Development Bank, Research Department. Deléchat, Corinne, Gustavo Ramirez, Smita Wagh and John Wakeman-Linn. 2010. “How Global Financial Markets Affect Sub-Saharan Africa.” IMF Staff Papers 57(1):172–208. Fouad, Manal, Wojciech Maliszewski, Martin Hommes, Hanan Morsy, Martin Petri and Ludvig Soderling. 2007. Public Debt and Fiscal Vulnerability in the Middle East. IMF Working Papers 07/12 International Monetary Fund. Greene, Joshua and Delano Villanueva. 1991. “Private Investment in Developing Countries: An Empirical Analysis.” IMF Staff Papers 38(1):33–58. 11

Guscina, Anastasia. 2008. Impact of Macroeconomic, Political, and Institutional Factors on the Structure of Government Debt in Emerging Market Countries. IMF Working Papers 08/205 International Monetary Fund. Guscina, Anastasia and Olivier Jeanne. 2006. Government Debt in Emerging Market Countries: A New Data Set. IMF Working Papers 06/98 International Monetary Fund. Hanson, James A. 2007. The growth in government domestic debt : changing burdens and risks. Policy Research Working Paper Series 4348 The World Bank. Hauner, David. 2006. Fiscal Policy and Financial Development. IMF Working Papers 06/26 International Monetary Fund. Hussain, Mumtaz, Andrew Berg and Shekhar Aiyar. 2009. “The Macroeconomic Management of Increased Aid: Policy Lessons from Recent Experience.” Review of Development Economics 13(s1):491–509. International Development Association. 2007. Country Policy and Institutional Assessment - 2007 Assessment Questionnaire. Technical report The World Bank. International Development Association and International Monetary Fund. 2010. Staff Guidance Note on the Application of the Joint Bank-Fund Debt Sustainability Framework for Low-Income Countries. Technical report IMF and The World Bank. International Monetary Fund. 2005. Monetary and Fiscal Policy Design Issues in Low-Income Countries. Technical report International Monetary Fund Washington D.C.: . International Monetary Fund. 2007. Reaping the Benefits of Financial Globalization. Technical report International Monetary Fund Washington D.C.: . International Monetary Fund and The World Bank. 2001. Developing Government Bond Markets: A Handbook. Washington D.C.: The World Bank and International Monetary Fund. Jaimovich, Dany and Ugo Panizza. 2006. Public Debt around the World: A New Dataset of Central Government Debt. RES Working Papers 561 Inter-American Development Bank, Research Department. Jaimovich, Dany and Ugo Panizza. 2010. “Public debt around the world: a new data set of central government debt.” Applied Economics Letters 17(1):19–24. Kose, M. Ayhan, Eswar S. Prasad, Kenneth S. Rogoff and Shang-Jin Wei. 2009. “Financial Globalization: A Reappraisal.” IMF Staff Papers 56(1):8–62. Panizza, Ugo. 2008. Domestic And External Public Debt In Developing Countries. UNCTAD Discussion Papers 188 United Nations Conference on Trade and Development. Panizza, Ugo. 2010. Is Domestic Debt the Answer to Debt Crisis? In Overcoming Developing Country Debt Crisis, ed. Barry Herman, Jose Antonio Ocampo and Shari Spiegel. Number 188 Oxford: Oxford University Press. Panizza, Ugo, Federico Sturzenegger and Jeromin Zettelmeyer. 2010. International Government Debt. Business School Working Papers 2010-03 Universidad Torcuato Di Tella. Pattillo, Catherine A., Helene Poirson and Luca Antonio Ricci. 2011. “External Debt and Growth.” Review of Economics and Institutions 2(3). Presbitero, Andrea F. 2008. “The Debt-Growth Nexus in Poor Countries: A Reassessment.” Economics: The Open-Access, Open-Assessment E-Journal 2(30). Presbitero, Andrea F. 2009. “Debt-Relief Effectiveness and Institution-Building.” Development Policy Review 27(5):529–559. Presbitero, Andrea F. 2010. Total Public Debt and Growth in Developing Countries. MoFiR working paper 44. 12

Reinhart, Carmen M. and Kenneth S. Rogoff. 2009. This Time is Different - Eight Centuries of Financial Folly. Princeton, New Jersey: Princeton University Press. Reinhart, Carmen M. and Kenneth S. Rogoff. 2010. From Financial Crash to Debt Crisis. NBER Working Papers 15795 National Bureau of Economic Research, Inc. Schindler, Martin, Chris Papageorgiou, Hans Weisfeld, Catherine A. Pattillo, Nicola Spatafora and Andrew Berg. 2011. Global Shocks and their Impact on Low-Income Countries: Lessons from theGlobal Financial Crisis. IMF Working Papers 11/27 International Monetary Fund. UNCTAD. 2004. Debt Sustainability: Oasis or Mirage? New York and Geneva: United Nations.

13

14 1992-2010 1992-2010 1970-2007 1979-2007

1980-2000

1980-2000 1980-2000

36%

Coverage ratio

48%

1994-2004 1980-2003

1996-2002

1979-2004

1981-2004

1979-2004 1980-2004

1981-2004

1994-2001 1986-2004

1992-2004 1979-2004

1996-2004 1994-1997

1977-2004

50%

2000-2010

2000-2010

2000-2010

2000-2010 2000-2010 2000-2010

2000-2010 2000-2010 2004-2010 2004-2010 2000-2010

2005-2010

2008-2010 2000-2010

2000-2010 2000-2010 2000-2010

2004-2010 2004-2010

1990-2001

1981-2003 1984-2004 1982-2004

2000-2010

2007-2010

Different definitions, data from IMF country reports and CBs

1980-2003

1998-2004

CG securitized domestic debt

Presbitero - 2010 update

Datasets: debt definition and country-time coverage Presbitero - WB 2005

Notes: (1) The 2007 version of the dataset by Abbas has data on Mozambique over the period 1984-2004 and on Uganda over 1978-2004(2)

89%

1970-1980 & 1989-2007 1970-2007 1970-2007 1970-2005 1970-2007 1970-2007 1980-2007

1995-2007 1989-2007 2000-2007 1974-2007 1970-2007 1970-2007 1977-2003

1974-2007 1970-2007 1976-2007 1970-2007 1993-2007 1970-2007 1970-2007 1982-2007 1970-2007 1995-2007 1970-2006 1976-2007 1970-2006 1991-2005 1986-2007 1970-2007 1970-2007

Gross securitised claims on the CG (excluding the stock of Treasury securities held by the CB) plus all securities issued by the CB

1986-2006 1970-2007 1975-2007 1970-1986 & 1992-2007

1980-1999

1980-2000

1980-2000

1980-2000

1980-2000 1980-2000

1980-2000

1980-2000 1980-2000 1980-2000 1980-2000

1980-2000

1980-2000

Securitized domestic government debt

Abbas & Christensen (2010) (1)

Afghanistan Bangladesh Benin Burkina Faso Burundi Cambodia Central African Republic Chad Comoros Congo, Dem. Rep. Eritrea Ethiopia Gambia, The Ghana Guinea Guinea-Bissau Haiti Kenya Korea, Dem. Rep. Kyrgyz Republic Lao PDR Liberia Madagascar Malawi Mali Mauritania Mozambique Myanmar Nepal Niger Rwanda Senegal Sierra Leone Solomon Islands Somalia Tajikistan Tanzania Togo Uganda Uzbekistan Vietnam Yemen, Rep. Of Zambia Zimbabwe

Country

Christensen (2005)

Table 1: Domestic Debt Data Sets: Data Coverage in Low-Income Countries

91%

1998-2009 1970-2005 & 2008-2009 1975-2009 1970-1986 & 1992-2009 1998-2009 1992-2009 1992-2009 1970-2009 1976-2009

2000-2009 1970-1980 & 1989-2009 1970-2009 1970-2009 1970-2009 1970-2009 1970-2007 1980-2009

1995-2009 1989-2009 1978-83 & 2000-05 & 08-09 1974-2009 1970-2009 1970-2009

1974-2009 1970-2009 1976-2009 1970-2009 1996-2009 1970-2009 1970-2009 1982-2009 1970-2009 1995-2009 1970-2009 1976-2009 1970-2009 1990-2009 1986-2009 1970-2009 1970-1976 & 1979-2009

Gross general government debt (domestic + external)

IMF HPDD 2010

91%

1993-2010 1970-2005 & 2008-2010 1975-2008 1970-1986 & 1992-2010 1998-2009 1992-2010 1992-2010 1970-2010 1976-2005

1995-2008 1984-2008 1970-2005 & 2008 1970-2008 1970-2010 1970-2010 1970-2007 1984-2010 1980 & 1989-2009 1970-2008 1970-2009 1970-2008 1970-2010 1970-2007 1980-2008

1970-2008 1995-2008 1981-2010 1970-2009 1970-2010 1990-2008 1970-2008 1970-2009 1970-2010

1974-2010 1970-2008 1976-2009 1970-2008 1996-2009 1970-2008 1970-2008

CG domestic debt (any holder), but also GG and NFPS debt when the former is not available

Panizza (2011)