AN ANALYTICAL FRAMEWORK FOR DEBT

SUSTAINABILITY AND DEVELOPMENT (FIRST DRAFT)

Paper commissioned for the UNCTAD project ‘Capacity Building for Debt Sustainability in Developing Countries’. Valpy FitzGerald University of Oxford July 2005

2

1. Introduction 1.1 The Motivation for this Paper This paper has been commissioned to provide the analytical framework for the UNCTAD project ‘Capacity Building for Debt Sustainability in Developing Countries’. UNCTAD is engaged in building the capacity of policy makers in developing countries to analyse the external debt process in the overall context of long-term development, linking the management of debt to macroeconomic and structural policies.1 This project thus adopts a critical approach in that it does not take the narrow view of debt sustainability as being solely reached by reducing an excessive current level of debt. Rather it implies that debt sustainability is an integral part of a successful development strategy, closely linked to export growth. There are at least three good reasons for developing country governments to borrow abroad: (i) the economic return on public investment in developing countries is superior to the cost of borrowed capital, and growth can be accelerated by prudent use of debt without excessively reducing current consumption levels; (ii) domestic firms cannot easily borrow abroad (particularly small and medium enterprises) and terms are better for sovereign borrowers, so it is efficient to use debt for on-lending to productive sectors, particularly exports; and (iii) the externalities from public investment in infrastructure, health, education etc are large and positive, but cannot be captured by direct foreign investment – particularly benefits accruing to poor people. Foreign private investors also gain benefits from investing in developing country sovereign debt, as the rates of return are higher than those obtainable on OECD government bonds, while the inherent risk arising from default probabilities can be mitigated by appropriate diversification of portfolios. However, this option is available only to ‘emerging markets’ – that is middle-income countries and a few large low-income countries. As a result of failures in international capital markets (arising from contract enforcement, information asymmetry and economic externalities) most low-income countries do not have access to external private capital except for foreign direct investment in natural resource sectors. In consequence bilateral aid donors and multilateral development banks act as financial intermediaries to provide loans on suitable terms based on their own ability to raise funds in global capital markets. Debt has to be repaid in foreign exchange, hence trade plays a critical role. The relationship between external borrowing and trade (whether export promotion or import substitution) is the key to a successful external debt strategy, as external indebtedness cannot be sustainable in the long run, if the development strategy does not lead to an increase in foreign exchange earnings above the level of domestic resource requirements, in order to release resources to repay the debt. The point of departure of a sustainable debt strategy is, therefore, a clear vision by the government of the country's development trajectory. None the less, debt burdens have risen to unsustainable levels in many developing countries, leading to protracted

1 See Tran-Nguyen (2004).

3

renegotiations and calls for debt cancellation. There is therefore an urgent need for a fresh look at the issue of debt sustainability linked to development strategies. 1.2 The Current Empirical Context for External Debt Analysis It is clearly essential to take into account the specific situation of different types of debtors (middle-income countries, low-income countries, HIPCs), in part because the types of lenders are diverse (with distinct objectives and leverages), but also because their economic structures (e.g. export sectors) and exogenous shocks are not the same. In this sub-section we take a brief look at the aggregate data organised by regional groups, income levels, and debt difficulties. This disguises many individual country problems of course, but does give a good idea of the overall issues. As Table 1.1 indicates, the debt burden in relation to exports is three times higher in low-income countries than in middle-income countries. However middle-income countries owe three quarters of all developing country debt. Thus the ‘debt problem’ is a middle-income country issue if considered from the point of view of the world economy, but a low-income country issue from that of economic development.

Table 1.1 External Debt and Exports by Income Level, 2003

Exports

(US$ bn) External Debt

(US$ bn) Debt/Exports

(%)

Low income countries 176 523 297% Middle income countries 1813 1815 100% Total developing countries 1999 2339 117% Source: World Bank World Development Report, 2005

The total value of external debt and debt service varies widely by region and by debtor status, at Table 1.2 indicates. By 2003 the net debt flow is quite low compared to outstanding debt in all three regions identified here2, but only in Developing Asia are reserves sufficiently large (particularly since the mid-1990s financial crises) to cover all debt liabilities. In Latin America (and by extension in ‘UDC’ countries with recent debt difficulties) reserves do not even cover debt service, leading to serious liquidity difficulties. In Africa (and by extension the HIPC group) reserves do cover debt service but this is not very useful because their debt is not traded. The very low level of effective debt forgiveness actually implemented to date is also evident from Table 1.2. Finally, it is worth noting that if the overseas assets of the private sector were recorded and entered here, the net asset position of developing countries would be positive – and in this sense there is no ‘developing country debt issue’ as such, but rather a serious public debt problem.

2 The CIS and Eastern Europe were in fact the main net borrowers in 2003.

4

Table 1.2 External debt of developing countries by region and debtor status, 2003 (US$bn) Total of

which Developing Asia

Latin America & Caribbean

Africa UDC HIPC

External debt 2724.3 695.7 759.0 278.0 804.2 145.8 Official Reserves 1412.6 670.1 196.2 90.9 168.7 19.7 Debt service 437.8 105.5 174.3 25.2 112.3 3.6 Net external borrowing 91.5 18.7 0.6 3.8 3.8 2.0 Exceptional financing 32.4 6.2 14.4 6.7 13.0 5.1 Source: IMF World Economic Outlook (WEO) May 2005. Notes: ‘UDC’ are Unsustainable Debt Countries (author’s definition) with arrears and/or rescheduling during 1997-2001; HIPC are ‘highly indebted poor countries’ under consideration by the WB and IMF for debt cancellation; ‘debt service’ is actual payments of interest on total debt plus amortisation payments on long-term debt, incorporating exceptional financing; ‘exceptional financing’ is arrears on debt service, rescheduling of debt service and debt forgiveness (this latter is only 11.0 US$bn).

The external debt structure varies in two dimensions – maturity and creditor. As Table 1.3 shows, most debt is ‘long term’ (that is, maturity of one year or more) and has an average maturity of the order of ten years.3 Africa and the HIPC countries rely most on official creditors while Asia and Latin America have a predominance of private lenders. Within this latter category, bonds predominate over bank credit – although the difference is not great in practice from the point of view of the borrower.

Table 1.3 Structure of External Debt, by maturity and creditor, 2003 Total of

which Developing Asia

Latin America & Caribbean

Africa UDC HIPC

Short term 377.9 106.3 89.7 19.4 34.8 3.3 Long Term 2344.9 589.4 669.3 258.6 769.4 142.5 Total debt 2724.3 695.7 759.0 278.0 804.2 145.8 Official creditors 1021.9 292.6 204.5 213.1 491.3 132.0 Private creditors: bank credit

722.1

161.9

185.8

42.5

183.2

10.8

bonds 960.0 241.3 368.7 22.4 129.7 3.0 Source: IMF WEO May 2005

In relation to exports, it is clear that the major ‘debt overhang’ difficulties are encountered in Latin America and Africa, and where most of the UDCs and HIPCs are to be found. In relation to debt service, however, it is Latin America that has the major problem (and by extension the UDCs) because Africa (and thus the HIPCs) has softer aid-related debt terms – and indeed does not fully service its debt. The differences in both implicit interest rates paid and amortisation rates reflect the differences between middle-income countries accessing private lending and low income countries relying on official lenders on the one hand, and the higher default risk in Latin America compared to Asia on the other. 3 See Table 6.1 in Section 6.

5

Table 1.4 Indicators of debt in relation to trade, 2003 Total of

which: Developing Asia

Latin America & Caribbean

Africa UDC HIPC

(% of exports) External Debt 111.3 73.2 199.3 144.6 208.9 321.1 Interest Payments 4.3 2.6 10.8 4.1 6.8 3.6 Amortisation 13.6 8.5 34.9 9 22.3 6.6 (% of debt) Interest 3.9 3.6 5.4 2.8 3.3 1.1 Amortisation 12.2 11.6 17.5 6.2 10.7 2.1 Source: IMF WEO May 2005

1.3 The structure of this paper This paper focuses on external debt. It does not examine domestic debt, even though with currency convertability, public debt sold locally and denominated in domestic currency is in effect a contingent claim on foreign exchange reserves – albeit at an undefined exchange rate. The growth of domestic debt in developing countries in recent years, due both to restrictions on foreign borrowing and the premature liberalisation of domestic financial systems under donor pressure, is a source of considerable concern - as it has led to both excessively high local interest rates and the diversion of domestic savings away from productive investment.4 Nor does this paper address debt crises as such, nor the subsequent renegotiations and restructurings. These are the topic of part II of this UNCTAD study. None the less, many of analytical results for prudent debt management discussed below are relevant to debt recovery because this cannot be sensibly planned except in the context of a sustainable debt trajectory. The structure of this paper is as follows. Section 2 provides the appropriate national accounting framework for debt analysis, and then sets out the traditional ‘gap models’ (savings, trade and fiscal) along with a sketch of the modern critique of this approach. Section 3 then outlines the modern intertemporal approach to debt analysis and derives the optimal debt level in relation to output and exports for an open developing economy, followed by an explanation of the two ‘golden rules’ for external debt management. The macroeconomic consequences of external debt are addressed in Section 4, starting with the key impact on real exchange rates and followed by that on fiscal balances and income distribution. Section 5 opens by explaining how credit rationing in global capital markets means that debt levels are not determined by borrowers, and goes on to analyse the impact of interest rate and trade shocks under these circumstances. Finally, Section 6 derives the policy conclusions of this paper for both domestic governments and the international community.

4 See IMF (2003).

6

2. Debt and the ‘finance gap’ model 2.1 Accounting for debt Debt accounting is quite complex because debt flows – inflows of fresh debt capital and outflows of debt service – enter into the process of savings and investment, the balance of payments (both current and capital account) and the fiscal framework. In the savings and investment balance, net debt flows act as ‘external saving’. In the current account debt interest payments are an outflow of factor income. In the capital account the net debt flows create changes in the net external asset position. In the fiscal accounts, gross debt inflows are an external resource, while amortisation and interest payments are major expenditure items. Further, net debt flows in a particular year, in combination with debt inherited from previous, determine debt for next year – thus introducing a dynamic element. These are all accounting identities and tell us nothing about the behaviour of the various components: in other words they are not a ‘model’. However, these identities do clarify the complex relationship between debt and the domestic economy, and also underline the fact that the components must be reconciled - in other words, ‘add up’. We use the following nomenclature: Y aggregate output (i.e. GDP) C aggregate consumption X exports of goods and services

M imports of goods and services S domestic saving I investment (gross fixed capital formation) i interest rate on external debt δ amortisation rate on external debt D external debt level G government expenditure T government revenue R official foreign exchange reserves

We start with the aggregate demand-supply balance

XICMY ++≡+ [2.1] which, when the savings-investment identity is inserted, yields the ‘accumulation balance’

MXSI

CYS

−≡−−≡

[2.2]

The balance of payments on current account includes not only exports and imports of goods and services, but also factor income5: to simplify exposition we only include

5 Including not only income from capital but also from labour – that is, worker remittances.

7

here the interest payments on (public) external debt at this stage. Note that is these are included in [2.2], then the definition of savings (S) is national savings and that of output (Y) is GNP. The current account (CAB) and the capital account (KAB) are opposite and equal so as to ensure the ‘balance of payments’ actually does balance:

iDMXCAB −−≡ [2.3] and the capital account is

RDKAB ∆−∆≡ [2.4] so that

0≡+ KABCAB [2.5] Note that private capital flows (and the corresponding factor payments) can be inserted into this accounting framework very simply. Foreign private assets (A) include portfolio holdings overseas (sometimes misleadingly called ‘capital flight’) and FDI abroad by domestic companies, an increasingly important phenomenon6 (UNCTAD, 2005). Foreign private liabilities (L) include both foreign borrowing by domestic firms and inward FDI. The full capital account can thus be written7

}{ ARLDKAB ∆+∆−∆+∆≡ [2.4a] However, in the rest of this paper we shall assume that the government is the only external debtor – not least because the ‘external debt’ statistics reflect public sector and publicly guaranteed external debt. Then we have the fiscal balance, which in this framework we assume to be closed only by foreign borrowing – thus excluding monetary issue (‘seignorage’) and domestic borrowing from consideration: iDGDT +≡∆+ [2.6] Last, but far from least, we have the law of motion for the external debt itself (D) in terms of its previous period value (D-1), and new borrowing (∆D) and the depreciation rate (δ)

11

11

−−

−−

+−≡∆∆+−≡DDDD

DDDD

δδ

[2.7]

6 See UNCTAD (2005). 7 The current account identity would also have to be rewritten of course to include the corresponding private factor income inflows and outflows.

8

2.2 ‘Financing gap’ models of debt and growth Financing gap models provided the basic analytical framework for both lenders and borrowers working on the support of development strategies from the 1960s through the 1980s.8 In these models9, the objective of the planner is to maximise the rate of output growth (y) subject to the constraints imposed by domestic savings (i.e. the capacity to invest), the external sector (i.e. the capacity to import) or fiscal balances (i.e. the capacity to spend).

The savings constraint exists because available funds are determined by the propensity to save (s) and the inflow of external finance (F), which in turn determines the maximum level of investment (I) that can be undertaken and thus the rate of growth (y). The external constraint exists because the level of imports (M) cannot exceed the foreign exchange available from exports (X) and capital inflows (F). Exports are assumed fixed in the short term, due to capacity constraints and/or limited external markets. The availability of imports determines the maximum level of output (Y) for a given import propensity (m). The fiscal constraint exists because growth depends on public investment (either because it is bulk of investment in poor countries, or because it is essential in order to promote private investment in middle income ones) which is a given share (π) of total investment; but public investment and thus growth is constrained by budgetary balance (Z) requirements.

Note that in consequence external finance (F) thus acts as a source of ‘external saving’, as a form of ‘import support’ and as a source of fiscal revenue. Through these channels F affects both the level of investment and the rate of growth of GDP. This model still informs most of the empirical policy debate about aid, debt and foreign investment. The planners problem is thus to maximize y where

ttt

tttt

t2t1t

ttt

t1-tt

tt

IYtgZ

DiDDFIm + Ym = M

YgY s= S

I + K = K

Kk = Y

πδ

τ

+−=+−−=

−+

−−

)(

)(

)(

11

[2.8]

subject to the three constraints

8 Avramovic and others (1964) is a good survey of the traditional methodology for analysing the relationship between debt and growth. 9 There is a large literature on these models, which originates with the Harrod-Domar growth model (the savings constraint) and continues with Chenery & Strout (1955) who set out the external constraint model, which was then extended for the fiscal constraint. Good formal expositions of all three ‘gap models’ are given by Bacha (1990) and Taylor (1994).

9

tt

ttt

ttt

FZF + X M

F + S I

≤≤≤

[2.9]

The outcome depends on which of the three constraints actually binds at any one point in time, which is an empirical issue. The savings-constrained maximum growth rate ( *

sy ) is straightforward to derive:

⎥⎦⎤

⎢⎣⎡ +−+=

Y

Fgskys )(* τ [2.10]

The main concern of aid-related policy modelling in most LDCs is the externally-constrained maximum rate of growth ( *

ey ), which we find by substituting to yield

⎥⎦⎤

⎢⎣⎡ −+

1* m

Y

FX

m

k = y

2e [2.11]

Finally, the fiscally constrained rate of growth ( *

fy ) is given by

⎥⎦⎤

⎢⎣⎡ −+ )(* gY

Fk = y f τπ

[2.12]

All three growth rates are of course increasing in net debt inflows (i.e. 0/ >∂∂ Fy ), but with different derivatives. Which binds depends on the parameter values. There is some reason to believe that in the poorest economies the savings constraint binds, and then external, and finally fiscal as economic development advances. The effect of net debt flows is progressively greater in each of these three stages because generally

F

y

F

y

F

y

m

sef

∂∂>

∂∂>

∂∂

<<***

2 1π [2.13]

2.3 The Limitations of ‘Financing Gap’ models The ‘financing gap’ model continues to form the basis for the trade, aid and growth linkages in medium-term macroeconomic programming model used by the World Bank: the Revised Macroeconomic Standard Model (RMSM).10 It also informs the

10 See the Addison (1989) and http://www.worldbank.org/data/rmsm/index.htm for an updated version of this paper plus other RMSM documents.

10

short term monetary programming framework used by the IMF, which focuses on fiscal side.11 These two models still form the essential analytical underpinning for the mission reports of the two Bretton Woods institutions on stabilisation and adjustment programmes.12 The UNDP makes estimates of the external financing requirements of poor countries on a similar basis when preparing for meetings of donor consortia. However, the last decade has seen growing criticism of the limitations of these models, which no longer correspond either to modern macroeconomic theory or to macroeconomic policy practice in open economies. Indeed from a neo-classical viewpoint this tradition is regarded as invalidating the proposals from the Bank and the Fund on additional lending and debt forgiveness.13 However, their persistence is doubtless due in large part to their analytical simplicity and the fact that the parameters can be estimated easily and quickly from available macroeconomic data in developing countries.14 Without going so far as to reject the financing gap models, it is possible to identify four areas of weakness which need to be remedied in order to produce a sounder conceptual framework and analytical model for quantifying debt sustainability. These are:

◦ First, the coefficients in the behavioural equations (particularly the constraints) are assumed to be stable and exogenous, rather than endogenously determined. In the case of savings, empirical evidence and Keynesian theory suggest that domestic saving (and thus consumption) in fact adjusts to the level of fixed investment and foreign inflows of capital.15 Again, the fiscal balance can always be adjusted by varying government expenditure.

◦ Second, in the external balance of trade, exports are assumed to be given and

imports to depend only on the level of economic activity. This ignores the effect of the real exchange rate on both import and export volumes, and thus the possibility of adjusting to foreign exchange shortages without having to reduce growth.16 It also underplays the role of exchange rates in determining the fiscal balance.

◦ Third, financing gap models assume that extra external finance always

contributes to growth, by simply and directly adding to savings funds, import capacity or fiscal resources and allowing investment – and thus growth – to rise. However, it is well established that external resources often in practice

11 See IMF (1987), which in turn derives from Polack (1957). See also Baquir and others (2003) for the growth linkages in IMF models. 12 See Aghenor and Montiel (2003) for a recent survey, and Khan and others (1990) for a formal statement of the relationship between the two models. 13 See Easterly (1999). 14 Particularly since Bank and Fund missions have only a few days in which to prepare macroeconomic forecasts of the countries upon which they are reporting, and thus have not the time (or the data) to construct, calibrate and test econometrically a full macroeconomic model of the kind used by OECD treasuries. 15 See FitzGerald (2003). 16 See Dornbusch & Helmers (1988).

11

lead to increased consumption.17 By extension, the investment undertaken may not lead to increased exports and thus debt repayment capacity.

◦ Fourth, and most seriously from an analytical viewpoint, the financing gap

model does not allow for intertemporal optimisation by economic agents: that is, the fact that households, firms and governments take investment, saving and borrowing decisions looking forward over many years. This is the basis of modern macroeconomics in general and for small open economies in particular; and allows resource allocation behaviour to be endogenised.18

3. Sustainable debt levels

3.1 The optimal debt level and export capacity The modern approach to debt sustainability starts from the same foundation as the modern macroeconomic theory of open economies, where apparent balance of payments disequilibria in the short run can be seen as part of an intertemporal equilibrium based upon expectations by economic actors about the future. The small open economy is composed of overlapping generations of households optimising consumption and saving over time, while firms make investment decisions based on profit maximisation.19 Current account surpluses (or deficits) generate net asset (or liability) positions with the rest of the world, which in turn affect the future behaviour of firms and households. If there is free access to international financial markets at a given interest rate (i) and no issues such as debt default, then the country obeys the Fisherian maxim and separates the decision to invest from the decision to consume.20 Focusing here on the decision to invest, firms choose their investment strategy so as to maximise the wealth of their shareholders when measured at world interest rates. The intertemporal equilibrium strategy21 amounts to selecting an investment rate (k*) that is a solution to

17 At least since Griffin (1970). 18 See Obstfeld & Rogoff (1995) and Sen (1994). 19 This is now a standard formulation: see for instance Obstfeld & Rogoff (1995). 20 The savings rate depends on the social rate discount factor and the intertemporal elasticity of substitution of consumption on the one hand, and the (world) interest rate on the other. See Sen (1994). 21 See Cohen (1994) for the derivation.

12

tt

ttt

tt

ttk

QIk

KIK

KQQ

dtJQit

/

)(

))(exp(max0

≡−=

=

−−∫∞

δ&

[3.1]

where Q is the level of net output, J the cost of installing new capital, K the capital stock and δ the rate of depreciation. In order to find a tractable solution to this general problem, we have to specify the relevant functional forms. We start off by defining national income (W) as output minus debt interest costs where debt also plays a role in capital formation such that

0)(''

0)('

<>−=

DW

DW

iDYW

[3.2]

in order that a maximum should exist.22 This formulation is also convenient because the two constraints reflect declining absorption capacity and debt overhang effects respectively. Under these conditions, the optimal debt level will be defined by the condition for maximising W with respect to D:

0)( =

∂∂+−

∂∂=

∂∂

D

iDi

D

Y

D

W [3.3]

In other words, debt should be contracted up to the point where the marginal addition to output equals the marginal addition to interest costs. Obviously, the higher the interest rate, the lower the resulting optimal debt level, while the larger the positive impact of that debt on output, the higher the optimal debt level. To find the optimal debt level, we start with a standard23 endogenous-growth production function of the form

aKY = [3.4] Leaving aside the last term in [3.3] and thus assuming that the interest rate is unaffected by the debt level we have the maximisation condition as

a

i

D

K

iD

K

K

Yi

D

Y

D

W

=∂∂

=−∂∂

∂∂=−

∂∂=

∂∂

0 [3.5]

22 Otherwise the optimal debt level would be infinite, of course. 23 See Rebello (1991) for the basis of the ‘AK’ model used here, and Aghion & Howitt (1999) for a comprehensive survey of modern endogenous growth theory.

13

The key issue is thus the effect of debt on investment. We shall examine the particular case where the domestically funded capital (K1) is already installed, there is no previous debt, and the authorities contemplate moving in one period to the optimal debt level, by providing extra capital stock (K2) funded by external debt 21 KKK += [3.6] External debt (D) is then contracted. A fixed proportion (λ) of this is used to fund the installation of new productive capital (directly as in rural infrastructure, or indirectly as loans to exporters): the rest for other activities such as social investments (health etc), coverage of current deficits or non-economic infrastructure. The cost of this productive public investment (J) is a quadratic function of the investment rate: 24

⎟⎟⎠

⎞⎜⎜⎝

⎛+=

=

1

22 2

1K

KKJ

JD

φλ

[3.7]

So we can now specify the objective function [3.1] as

⎟⎟⎠

⎞⎜⎜⎝

⎛+−+=

1

2221 2

1)(K

KK

iKKaW

φλ

[3.8]

and differentiating with respect to K2 yields the optimal solution25 in terms of the ratio (γ) between debt-funded capital and ‘domestically funded’ capital:

γλφ

φλ

=⎟⎠⎞

⎜⎝⎛ −=

=⎟⎟⎠

⎞⎜⎜⎝

⎛+−=

∂∂

11ˆ

01

1

2

1

2

2

i

a

K

K

K

Kia

K

W

[3.9]

Note that the optimal capital structure coefficient (γ) can be negative – which would imply accumulating foreign assets instead of borrowing abroad.26 We find the optimal debt-to-ouput ratio (θ), by substituting [3.9] into [3.7] and [3.4]:

( )

)1(

)1(ˆ

ˆ

)1()ˆ(ˆ

1ˆ

1ˆ

ˆ

121

11

22

γλγφγθ

γ

γφλγφ

λ

++==

+=+=

+=⎟⎟⎠

⎞⎜⎜⎝

⎛+=

aY

D

aKKKaY

KK

KKD

[2.10]

24 Heijdra and van der Ploeg (2002), p. 40; and also Cohen (1994), p. 490. 25 This can be seen as being set by the ‘social planner’ or by firms maximising profits. 26 As natural resource exporters do.

14

Clearly θ is increasing in γ, and thus the optimal debt-output ratio will be lowered by an increase in the interest rate (i) as we would expect, but will be raised by an increase in the proportion of debt funds allocated to productive investment (λ) or in the overall productivity of capital (a). This result can be is generalised to a steady-state growth situation because this means that Y/K is constant (and thus both components of capital grow at the output growth rate), and thus D/Y must be constant. If the optimal debt level (θ) is higher, then debt can be safely raised and economic growth is higher. Overall capital productivity requires some further comment in the context of this study. We assume a simplified form of the externally constrained economy discussed in the previous section such that:

KX

XM

mYM

αβ=≤=

[3.11]

where exports are a function of the proportion (β) of the capital stock located in the export sector with known productivity (α) in a similar ‘AK’ growth model to that used above. Substituting {3.4] and [3.9] into [3.11]

γλαβφ

αβ

=⎟⎠⎞

⎜⎝⎛ −=

=

11ˆ

1

2

imK

K

ma

[3.9a]

In other words, the optimal debt level rises with the proportion β of debt-funded capital stock allocated to the export sector. However, it falls with an increase in interest rates or the import coefficient. To put this another way: long-run debt solvency – and thus the avoidance of debt crises arising from trade or capital market shocks – requires the allocation of a higher proportion of the funds raised not only to productive investment but to investment in the export sector. Finally, we can also define the optimal debt service ratio (ω) from this result, where

X

Diˆ

ˆ)( δω += [3.12]

by substituting [3.10] and [3.11] into [3.12] to yield:

αβ

θδδω mi

X

Y

Y

Di )(

ˆ

ˆ

ˆ

ˆ)( +=+= [3.13]

Note that the optimal debt service ratio (ω) will decrease with a higher interest rate (i) because as we have seen this has a powerful negative influence on the optimal debt-output ratio (θ).

15

3.2 The ‘golden rules’ for debt sustainability The ‘law of motion’ for external debt from the previous section (equation [2.7]) can be expressed in terms of the primary27 current account balance (P) on the assumption that this debt is the only form of external finance:

11)1( −− ++= ttt PDiD [3.14]

which by repeated substitution yields

∑−

= +=

+

1

0 )1()1(

t

tt

tt

t

i

P

i

D [3.15]

and when n goes to infinity, the present value of debt (i.e. the left hand side of the equation) goes to zero and we retrieve the intertemporal balance of payments constraint

∑∞

= +=

00 )1(t

tt

i

PD [3.16]

In other words, all debt must eventually be paid back. However, in practice, developing country financial authorities and debt managers have to work on a tighter time scale and without the luxury of searching for optimal solutions. The ceiling on ‘prudent’ debt is conventionally expressed as a share of output28 or as a ratio of debt service to exports, the former reflecting longer-term ‘solvency’ considerations and the latter shorter-term ‘liquidity’ ones. Once the prudential ceiling (d) on the debt output ratio has been reached, debt management strategy is logically not to exceed it. Thus the ‘golden rule’ is that

dY

D

Y

D

t

t

t

t =≤−

−

1

1 [3.17]

For a given rate of output growth (y) and expressing the primary deficit as a ratio (p) of output we have

pdy

id

Y

P

Yy

Di

Yy

PDi

Y

Dd

t

t

t

t

t

tt

t

t

−++≥

−++=

+−+=≥

−

−

−

−

1

1

)1(

)1(

)1(

)1(

1

1

1

1

[3.18]

so that the ‘golden rule’ for the debt-ouput ratio is

27 That is, excluding interest payments. 28 The most celebrated example being of course the Maastricht Treaty limiting public sector debt of EMU members to 60 percent of national income.

16

dyiy

idp )(1

1

1 −≈⎟⎟⎠

⎞⎜⎜⎝

⎛ −++≥ [3.19]

In other words, a primary deficit (p<0) can only be safely incurred if the growth rate is higher than the interest rate (y>i). If we express the rule in terms of the current account balance proper, as a proportion (c) of output then the golden rule becomes

ydc

idpc

−≥−=

[3.20]

In other words, that the maximum current account deficit as a proportion of GDP is the rate of growth multiplied by the prudent debt-GDP ratio. We can now turn to the second ‘golden rule’ related to the ratio of debt service to exports. The derivation is very similar to that for the first rule, but expressed in terms of the second ceiling (σ): Once the prudential ceiling (d) on the debt output ratio has been reached, debt management strategy is logically not to exceed it. Thus the ‘golden rule’ is that

σδδ =+≤+

−

−

1

1)()(

t

t

t

t

X

Di

X

Di [3.21]

For a given rate of export growth (x) and expressing the primary deficit as a ratio (p') of exports we have

pix

i

X

Pi

X

Di

x

i

Xx

PDii

X

Di

t

t

t

t

t

tt

t

t

′+−++≥

+−+++=

+−++=+≥

−

−

−

−

)(1

1

)()(

1

1

)1(

})1){(()(

1

1

1

1

δσσ

δδ

δδσ

[3.22]

so that the ‘golden rule’ for the debt service ratio is

σδδ

σ+−≈⎟

⎠⎞

⎜⎝⎛ −

++

+≥′

i

xi

x

i

ip 1

1

1 [3.23]

In other words, and more generally that a primary deficit (p'<0) can only be safely incurred if the export growth rate is higher than the interest rate (x>i). If we express the second rule in terms of the current account balance proper, as a proportion (c') of exports

17

σ

δ

σ

+−≥′

−′=′

i

xc

ipc [3.24]

3.3 Convergence and expectations Debt authorities attempt to adopt at least a medium-term view, and when their debt levels are above the prudent limits, then a ‘convergence’ strategy must be adopted in order to reach these limits within a reasonable number of years. Suppose that we wish to reach to the prudential limit (d) of the debt-GDP level over a number of years from the present level (d') by reducing the debt by a proportion u each year over n years, then

n

d

ddu

/1

⎟⎠⎞

⎜⎝⎛

′−′

= [3.26]

and the first golden rule is re-expressed as:

dyuc

dyuip

)(~)(~

−≥−+≥

[3.27]

Whether new debt is sustainable depends, therefore, on expectations about the future growth rates of output and exports on the one hand, and on future interest rates and terms of trade on the other. As we have seen, the debt level will affect growth itself in an optimal solution, so also the use to which the debt is to be put, and thus future productivity, are also crucial assumptions. Debtors and creditors presumably agree on these forecasts before signing a debt contract. For instance, if we expect the rate of change of the terms of trade (h) to have a projected value in the future, then this is distinguished from export volume growth ( x ) and [3.24] is rewritten as

σδ+

+−≥′i

xhc [3.24a]

For a debt contract to be agreed upon by debtor and creditor, both must agree on the projected parameter values, or if they disagree, at least the overall outcome must be anticipated as profitable to both sides. But events do not always turn out as well as expected.29 Three key examples are the debt crises of the early 1980s and the middle 1990s, and the present plight of the HIPCs. In the 1970s debtors and creditors thought that in view of the shortage of raw materials, terms of trade would improve while interest rates would remain low. However, in the 1980s a radical shift in US monetary policy led to rising interest rates 29 It is notorious that the international financial institutions always provide high and positive growth rates for exports and output from debt countries, despite experience to the contrary.

18

and falling terms of trade: debt levels that had been previously regarded by both borrowers and lenders to be sustainable, turned out to be unsustainable. In the mid-1990s, the growth prospects of many ‘emerging markets’ were again felt to be very good, interest rates were again low, but much of the investment was in non-traded sectors, which with overvalued exchange rates led to export growth falling below interest rates and generated a liquidity crisis. Last but not least, many very poor countries were lent large sums by aid donors and international financial institutions in order to support economic development, which the donors felt at the time was sustainable (if not, they should not have lent). When exports did not grow as expected, these poor countries were forced into effective default and ‘HIPC’ status. Expectations on both sides are thus crucial to the lending/borrowing decision – there can be no ‘over-borrowing’ without ‘over-lending’. 4. External Debt and Macroeconomic Management

4.1 Macroeconomic consequences of external debt In the previous section, we have explored the effect of external debt on economic growth and derived the prudential rules required to ensure that the debt is sustainable in the long and medium while achieving the maximum increase in incomes. The focus was on the investment process and the dynamics of debt over time. In this section we are concerned with the shorter-run macroeconomic consequences of debt, where the effects on relative prices (particularly the real exchange rate) and demand are the main concern.30 Debt flows and stocks have strong macroeconomic effects, where generally debt capital inflows expand the economy and outflows of interest and amortisation contract it. The net effect depends, therefore, on the point in the debt cycle. The simplest way to see this is to look at a model of the small open developing economy, where the balance of payments is made up of exports (X), imports (M) and capital flows (F) as well as debt service. Exports depend on world demand (Q) and the real exchange rate (e); imports on domestic output (Y) and the real exchange rate. External debt then affects this structure through four distinct channels:

i) the effect of debt service (iD) in reducing resources available for imports and thus curtailing output (Y);

30 We are concerned here with external debt and developing countries of course: for an excellent study of public debt management in developed countries, where the issues are distinct, see Missale (1999).

19

ii) the effect of the capital inflow from increased debt (∆D) in increasing available resources and thus allowing Y to expand;

iii) the effect of the proportion of the debt stock (ε) which is allocated to the export sector;

iv) the effect of these two flows for the real exchange rate (e) and thus secondary consequences for both exports and imports.

We have, then, the following setup

DF

iDMFX

emYmM

DxexQxX

∆=+=+

+=++=

21

321 ε

[4.1]

The way in which this model ‘closes’ depends upon the policy regime followed by the authorities.31 Here we examine the alternatives of variable or fixed real exchange rates. A floating real exchange rate is usually the result of a floating nominal exchange rate being used to achieve balance of payments equilibrium; although it can be the result of combining a nominal anchor with variable domestic inflation. Either way, the real exchange rate adjusts to clear the external account, consistent with a stable output level (Y = Y*) – indeed this is the policy objective. In this first case, the real exchange rate (e) is thus found from [4.1] by appropriate substitution. Differentiating with respect to the debt level (D) then shows the effect of an increased debt level.

01

*

22

3

22

301

<−

−+−=∂∂

−−−∆−+=

mx

ix

D

e

mx

DxQxDiDYme

ε

ε

[4.2]

The derivative is unambiguously negative32 in this case: in other words, the macroeconomic effect of increased debt to cause the real exchange rate to appreciate. The effect on exports, and thus debt service ratio, then depends on the net offset between the negative influence of the revalued exchange rate (the so-called ‘Dutch Disease’ effect) and the positive effect of the debt resource allocation to the export sector (εD).

31 Technically we have a model with four equations, five endogenous variables (X, M, Y, F, e) and one instrumental variable (D), so there is one degree of freedom for a policy rule, which provides what is in effect the fifth equation. 32 Remember that m2 < 0.

20

2222

3

2

11

mx

i

mx

x

D

ex

D

X

−−−⎟⎟

⎠

⎞⎜⎜⎝

⎛−

−=

∂∂+=

∂∂

ε

ε [4.3]

This underlines the importance, already stressed above, of allocating a sufficiently high proportion (ε) of debt funding to the export sector – or at least to traded output such as competitive import substitution33 – so that the revaluation effect of debt inflows is at least counterbalanced. Otherwise, the debt service burden will be increased both by higher debt interest charges and reduced exports because. Assuming that the economy is at its prudential debt service limit (σ), the net effect of increased debt on the debt service ratio is:

( )

⎭⎬⎫

⎩⎨⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛−

−−−−−=

∂∂−=

∂∂

22

3

22

2

111

/

mx

x

mx

ii

X

D

X

X

iD

X

i

D

XiD

εσσ [4.4]

From this the proportion (ε) of debt to be allocated to exports in order to keep the debt service ratio from increasing is clearly:

322

22

)(

)1()(ˆ

xmx

imxi

−−−+−≥

σσε [4.5]

We now turn to the second case of the fixed real exchange rate (e = e*), either as an explicit policy strategy to maintain export competitiveness or as the result of effective inflation targeting combined with a nominal anchor. In this case, output (Y) becomes an endogenous variable and adjusts so as to keep demand at a level consistent with balance of payments equilibrium in [4.1].

0/)1(

/})(*{

13

1223

>+−=∂∂

−−++∆=

mxiD

Y

miDmxeDxDY

ε

ε [4.6]

The result is clear: the effect of debt increases on output is unambiguously positive. Indeed, this corresponds to the ‘external gap’ model of Section 2 above. The contrary is also true of course, a reduction in the debt level (or net flow) leads to a fall in output. This is the way in which capital market shocks are transferred to the domestic economy: the multiplier effect of capital flows on output (and employment) is very strong because the reciprocal of the import coefficient (m1) in [4.6] has a value between 0.2 and 0.3 for most developing countries. Note that a greater higher export orientation of investment (ε) will have the effect of increasing the multiplier too, due to the greater trade-dependence of the economy.

33 Which in this model would be equivalent to reducing m1.

21

The relevant prudential debt ratio in this case is the debt-output ratio (d). The net effect of a debt increase will then depend on the offset between increased debt and increased output, with a higher allocation (ε) to exports - and thus output through the imports that these exports fund – reducing the resulting increase in the debt-output ratio.

( )

⎭⎬⎫

⎩⎨⎧ +−−=

∂∂−=

∂∂

1

3

2

11

1

1/

m

xid

Y

D

Y

Y

D

YD

YD

ε [4.7]

The condition for stabilising (or even reducing) the debt-output ratio is thus again that a sufficiently large proportion of debt funds be allocated to export sector investment:

3

1 )1(ˆ

dx

idme

−−≥ [4.8]

In sum: on the one hand, debt flows have a strong macroeconomic impact, although the exact form depends on the policy regime; and on the other, a sufficient allocation of debt funds to the export sector is essential if prudent debt-output and service-export ratios are to be maintained. 4.2 Fiscal Consequences of External Debt So far we have treated the macroeconomy as a single entity: we now need to separate out the fiscal sector, because as the contractor of debt, the effects upon public sector accounts are key to macroeconomic equilibrium. Recalling the fiscal balance in our national accounting framework (Section 2, equation 2.6), and writing it in terms of the gross debt flow (F΄) rather than net of amortisation (F), we have:

FTDiG ′+≤++ )( δ [4.9] Clearly an increased gross debt flow (F΄) in itself allows a fiscal expansion (i.e. G to rise); but of course accumulated debt itself generates large budgetary items which in some cases become the largest single item of expenditure34 and crowding out other expenditure categories. As the debt is denominated in foreign currency and the rest of the budget items in domestic currency, we include the real exchange rate. Netting out the debt amortisation flows, and assuming

34 For all developing countries in 2003, the average ratio of external debt service to GDP was 6% (see Table 1.4 in Section 1). The average ratio of tax revenue to GDP (t) was 15 % and of public health expenditure to GDP was 3% in that year (WDI, 2005).

22

◦ a strict budgetary rule that only allows a maximum domestic currency fiscal deficit (q) to be financed from seignorage and/or domestic borrowing

◦ the prudent debt-output ratio (d) is maintained ◦ tax revenue is a given share (τ) of national income

we can write [4.9] as YdeYqieDG ∆++≤+ )(τ [4.10] Diving through by Y and rearranging, we have the share (g) of government expenditure in national income in terms of the familiar debt parameters (d, i) and the rate of growth (y) of output:

diyeqg )()( −++≤ τ [4.11]

Absent a serious tax reform (∆τ) or a more relaxed monetary stance (∆q) then the government expenditure share (g) in national income becomes highly dependent on the debt parameters on the one hand, and the growth rate (y) and the real exchange rate (e) – both of which are themselves strongly affected by the debt strategy – on the other. In particular, an increase in e consequent upon devaluation when the rate of output growth is low (i.e. y < i) – a common occurrence during debt crises - will have a strongly negative fiscal impact on expenditure. This subordination of government expenditure to debt management needs has at least three very serious consequences:

a) it is very difficult to give true priority to increasing social provision in general (and poverty reduction in particular) by expanding real health and education expenditure faster than population growth;35

b) it is not possible to engage in an active counter-cyclical fiscal policy in order to reduce the impact of exogenous shocks on investment and growth by expanding infrastructure expenditure to maintain capacity utilisation;

c) and long-term planning of public expenditure is rendered meaningless, reducing the efficiency of public services, undermining infrastructure provision and wasting scarce administrative skills.

4.3 Debt and Income Distribution The primary macroeconomic effect of debt on relative prices is through the real exchange rate as discussed above, and thus on the return to factors (land, labour and capital) in the traded and non-traded sectors. When debt flows are positive, the exchange rate appreciates, and prices move in favour of non-traded sectors – indeed the ‘Dutch Disease’ effect tends to stimulate speculative investment and employment

35 Despite claims that the PRSPs give priority to the poor, at best they only do this within an expenditure target conditioned by macroeconomic strategy.

23

in sectors such as real estate while depressing traded sectors, particularly exports. This is why we have stressed the importance of using debt funds directly to stimulate investment in the export sector. The impact of GDP expansion on employment as a whole is clear enough, but the effect of changes of the real exchange rate on real wages is rather more complex. Consider an economy with two goods with their corresponding prices, foreign (pf) and domestic (pd), and a nominal exchange rate (E)

d

f

p

Epe = [4.12]

The real wage (ω) is defined in terms of the nominal wage (w) and the consumer price level (pc)

cp

w=ω [4.13]

where consumer prices depend upon the proportion (µ) of the domestic good in the consumption basket fdc Eppp )1( µµ −+= [4.14]

The price of the non-traded domestic good (pd) is a markup on unit wage costs, so that

jwpd = [4.15]

We can now combine these, substituting (4.12) and (4.15) into (4.14) and then plugging the result into (4.13), so as to derive the link between the real exchange rate (e) and the real wage rate (ω)

1}])1({[ −−+= ej µµω [4.16] In other words, real currency depreciation (that is, e rising) must also mean a real wage reduction, and vice versa. The fact that overvaluation from capital inflows is politically popular, and undervaluation (especially large real devaluations associated with stabilisation policies after financial crises) is politically unpopular is merely a reflection of this distributional logic. This result can then be linked back to debt via Equation 4.2 above, from which it is clear that an increased debt flow will cause the real exchange rate to depreciate (and thus e to fall) and in consequence real wages will rise.

01

)1(

)1(

22

32

2

>−

−+−=

∂∂−−=

∂∂

mx

ixD

e

Dερπω

ρπωω

[4.17]

24

More generally, it provides an explanation for why it is not always possible to set the real exchange rate at the level required to clear the current account, and thus the ‘external constraint’ discussed in Section 2 is a real constraint, despite the neoclassical critique.36 5. Debt Vulnerability and External Shocks 5.1 Determinants of debt flows So far we have been working on the assumption that developing countries can choose the level (D) of debt that they contract, at a given interest rate (i). This is the conventional assumption in economic analysis, and indeed in policy debates when reference is made to ‘over-borrowing’. In fact neither of these is true: lenders determine the volume of debt lending and the interest rate is not given. In effect international debt flows are ‘credit rationed’. ‘Official’ lending (that is by bilateral donors or multilateral organisations) is always determined by the lender on her own criteria, although these should in principle support sustainable development and thus coincide with those of the borrower.37 However, it is not up to the borrower to decide on the debt level. The overall volume of official lending is determined by the institutional strategy of the lender, who must maintain loan volumes (lending out repayments) and indeed increase them in order to expand. Within this total, regional and country allocations will depend upon both the lenders’ technical appreciation of development prospects38 and thus the sustainability of debt on the one hand, and the geopolitical pressures of donor governments on the other. Given a ceiling of official lending from donors in any one period, developing country governments tend to contract debt up to this limit. It is in this sense that ‘credit rationing’ exists. It is extremely rare for developing countries – particularly small low-income countries without access to private capital markets – to turn down official lending proposals. The interest rate and maturity of official debt is also set by the lender, usually on a subsidised basis: this may be similar to long-term market rates in

36 McCombie & Thirlwall (1994) demonstrate other ways in which the balance of payments constraint does bind in practice. 37 Although there is a regrettable tendency for lenders not only to think that they ‘know best’ and have a longer-term view, but also that they can better represent the interests of the poor, compared to a local democratic government. 38 Hopefully along the lines sketched in the two previous sections...

25

OECD countries but does not include a risk premium;39 or can be at lower than market rates for poor countries. Eligibility is decided by the donor. As far as international (private) debt markets for bonds issues by and bank loans to developing country governments (‘sovereigns’) credit rationing also obtains. This is the result of a market equilibrium where creditors attempt to maximize their expected returns on loans subject to the risk of non-payment. Uncertainty in the loan market creates adverse selection, as the two sides have different perceptions of risk and lenders cannot distinguish between borrowers as to their ability to repay in the future. In addition, OECD investors appear to less tolerant of risk in foreign markets than they are of risk on their own markets, leading to an inefficient allocation of their portfolios known as ‘home bias’ which further limits the access of developing countries to international capital markets.40 Consider an initially upward-sloping supply schedule of bank loans or bond purchases in the space of the interest rate plus spread and the volume of lending shown in Figure 5.1 below.41 The competitive international banking market is made up of many non-collusive bank lenders and borrowers. Banks are price-takers in deposit markets but set lending rates (i.e. spreads) to maximise expected returns. Higher lending rates have an adverse selection effects on borrowers and thus increase default risk along with higher levels of indebtedness – both due to interest rate burden on reserves and because the incentive to default rises with debt and interest rate. So the debt supply schedule will be backward-sloping beyond a certain point. Banks could in principle charge different rates for different risk classes, but have imperfect information on fundamentals (e.g. default risk) as well as fear of covariant risk between borrowers (contagion). The return spread (r) is the interest income on bonds (or bank loans) in excess of the riskless over the riskless rate.42 The demand schedule (Dd) for debt is the backward-sloping supply curve discussed in the previous paragraph. Competitive lenders maximise their debt holdings at the point (Dd , r*): at this ‘price’ (i.e. return spread) the potential supply of debt assets from developing countries (Ds) is in excess of demand (Dd) – in other words, the willingness to borrow exceeds the willingness to lend even though the market is in equilibrium.

*)(*)( rDrDD sd <= [5.1]

39 The World Bank, for instance can borrow (and thus lend) at rates similar to the US government on New York markets although not because the Bank’s lending is efficient but rather because its bond issues carry a US Treasury guarantee. 40 See FitzGerald & Babilis (2005). 41 See Folkerts-Landau (1985). Of course the aggregate will be made up of all developing countries, some of which do not need credit at all (e.g. Brunei), some of whom have permanent access (e.g. Taiwan) and others who can never access private funds at all (e.g. Tanzania), but most are sometimes ‘in’ and sometimes ‘out’ depending on overall market sentiment, but are always willing to borrow more at the going rate. 42 Typically the US government bond rate.

26

Figure 5.1 Credit Rationing in Global Debt Markets

In effect, the market interest rate in foreign currency (if ) to emerging market borrowers will contain three elements: the riskless world rate (iw) and a risk premium (ρ). The risk premium is the product of the perceived 43 probability of default (π) and investors’ own degree of risk aversion (A).44 Perceived default risk depends upon precisely the indicators we have discussed in the previous section such as the debt-GDP ratio (d) and the debt service ratio (σ).45 Thus we have:

),( σπππρ

ρ

d

A

ii wf

==

+=

[5.2]

Clearly any increase in the two debt ratios will increase default risk (which is why the loan supply curve is backward-sloping) and thus not only raise interest costs but also reduce loan availability. Note that the risk premium (ρ) depends on forecasts of debt default probability, and thus on expectations of export and output growth (as we have seen in Section 4) on the one hand, and on the risk tolerance of investors on the other. If the capital account is ‘open’ in the sense that capital can flows freely into and out of the economy, then the domestic interest rate (id) in denominated in domestic currency adjusts through arbitrage to an equilibrium with the ‘dollar’ rate (if) plus the expected rate of change of the exchange rate ( EE /& ) expected by international investors and thus affected by market sentiment (e.g. contagion) as well as host country fundamentals: EEAdiEEii wfd /).,(/ && ++=+= σπ [5.3]

43 Perceived by lenders of course, where ratings agencies can play an important role. 44 Thus the risk premium is only equal to the underlying default risk if the financial market is strictly risk-neutral and there is perfect information; so that yield spreads should not generally be interpreted as measures of ‘country risk’ – see Cunningham and others (2001). 45 As well as cruder liquidity measures such as the ‘quick ratio’ mentioned in Section 1.

27

This means that external debt and global capital market conditions will have a strong influence on domestic interest rates, and thus fiscal costs and private investment levels. Domestic policy space is sharply reduced in consequence – particularly monetary policy but also the fiscal stance. 5.2 Global capital market shocks The most obvious capital market shock is when developed country interest rates shift, affecting interest rates paid by developing country sovereign borrowers (if) in two ways as Equation 5.3 indicates: first by simply raising (lowering) the riskfree rate (iw) and secondly by raising (lowering) the default risk (ρ) because the debservice ratio (σ) has increased (decreased). However, the form of capital market shock that is much more important in practice is when the demand for emerging market debt shifts due changes within developed country markets – such as changing regulations, fluctuations in risk aversion or contagion from other debtors. These lead to ‘horizontal’ shifts in the asset demand curve in Figure 5.1. Indeed, the greater part of aggregate shifts in the demand emerging market assets can be attributed to events in the home capital markets: changes in risk tolerance and investor confidence as well as shifts in interest rates and wealth levels on the one hand, and trading behaviour in the form of herding and momentum trading on the other. The macroeconomic and distributional consequences for emerging markets are disproportionately large: they are in effect a major externality.46 This results from the asymmetry in international capital markets: while these flows are relatively small in relation to home economies, they are very large in relation to host markets. Capital flows have strong macroeconomic effects in developing economies that are much larger than their share of GDP or market capitalisation might suggest because of the sensitivity of the macroeconomy to the balance of payments, as we have seen in the previous section. The effect of these shocks is exacerbated by hysteresis47 – due to the irreversibility of investment and wage-price stickiness, a downswing does not lead the economy back to where it was before the upswing. Fluctuations in the real exchange rate associated with short-term capital flows also lead firms to misallocate investment between the traded and non-traded sectors, with negative consequences for growth.48 However, perhaps the most serious negative effect of debt on growth is felt directly through the balance of payments, but rather through the effect on investor uncertainty

46 See FitzGerald (2001). Interestingly, this is the position taken by the IMF in the World Economic Outlook for 1998 (‘Financial Crises: Characteristics and Indicators of Vulnerability’), although by 2005 the World Economic Outlook has become much more sanguine, attributing most of emerging market volatility to domestic fundamentals. 47 A model of this process is set out in Chapter 6 of FitzGerald (2003). On the macroeconomic theory of hysterisis and path-dependency see Heijdra & van der Ploeg (2002), Chapter 2.2 and Appendix A.6.4. 48 See FitzGerald & Perosino (1999).

28

about future macroeconomic circumstances49 and policy regimes50 when the debt level is above the prudential limit – commonly known as ‘debt overhang’. Even if the government cannot credibly pre-commit to repay debt, it may be able to pre-commit by investing itself in growth before borrowing, so that both domestic investors and foreign lenders are more optimistic about growth prospects. Inherited debt will reduce optimal investment due to the ‘tax’ on returns arising from service obligations on old debt. This reduces the future ability to repay and increases the benefits of default; both of which will reduce the willingness of lenders to extend further loans. By extension, reducing the debt burden will increase investment and future ability to repay: thus creating a situation where it is in the interest of lenders to unilaterally forgive a proportion of the debt in order to secure full service on the rest.51 5.3 Global Trade Shocks This is not the place to discuss the origins of global trade shocks, but these can clearly take various forms independently of a country’s trade strategy. They include:

◦ sudden movements in export prices, particularly for primary commodities, due to demand shifts in developed countries or supply changes by other producers;

◦ unexpected shifts in import prices, particularly those for essential commodities such as oil;

◦ gain or loss of access for exporters to particular developed country markets, due to changes in trade barriers or domestic (e.g. health) regulations.

These shocks obviously have an immediate effect on the debtservice-export ratio by changing the denominator: a sudden fall in primary commodity prices for instance will raise this ratio even though debt service itself has not changed, and can render a previously sustainable debt unsustainable. A second-order effects depend upon what policy response the authorities take. Following the framework in Section 4 above, they have four options in face of an export price drop:

◦ incur more debt in order to sustain import levels and maintain the level of economic activity; this in turn appreciates the real exchange rate (or at least prevent if from depreciating) so exports do not rise as producer prices remain depressed; and thus the debt service ratio will rise due to both lower exports and higher debt;

◦ maintain the debt level and allow the currency to depreciate in order to restore current account equilibrium and stimulate exports; in this case exports do not fall and the debt service ratio is maintained; but the fiscal balance is worsened, requiring social expenditure cuts; and the income distribution worsens with declining real wages and inflation;

49 See FitzGerald, Jansen & Vos (1994). 50 See Rodrick (1991). 51 See Sachs (1989).

29

◦ maintain the debt level and stabilise the real exchange rate, cutting the level of economic activity in order to depress imports, prevent inflation and balance the current account; exports decline (the producer price has fallen) and the debt service ratio rises, as will the debt-income ratio; also employment falls along with output

◦ any one of the above combined with a reallocation of debt funds towards exports with good markets, so as to maintain export growth and thus reduce the debtservice ratio.

Which policy option is adopted thus determines the impact of a trade shock on debt sustainability. The domestic policy choice between exchange rate shifts and demand management depends on local economic structures and political economy circumstances, as well as the pressure from international institutions. The third ‘golden rule’ in this context is well known: “treat negative shocks as permanent and positive shocks as temporary”. It is clearly better to reduce debt in response to improved trade conditions rather than increase it when they deteriorate. None the less, developing country governments frequently do the exact opposite: increasing debt during downswings and not reducing it again in upswings. Above all, the tendency to apply public debt to non-traded sectors (encouraged by the international institutions) reduces the ability to cope with trade shocks. 6. Conclusions: policy principles for debt management in development strategies 6.1 Debt policy parameters in practice The two main indicators we have used for prudent debt management are shown in Table 5.1. It is clear that those economies with unsustainable debt in practice (that is, those which were in arrears and/or undertook rescheduling during the 1997-2001 period) still have very high debt-GDP ratios approximating the conventional upper bound of 60 percent, although the ratio for the HIPCs is even higher. This ceiling derived from experience of when countries get into major debt difficulties is in fact considerably lower than that set by the World Bank52 in the context of the HIPC initiative.

52 See World Development Finance 2004: ‘moderately indebted’ countries are those with a ratio of the present value of contracted debt payments (PV) to GNP of over 132 percent and of PV to exports of goods and services (XGS) of over 48 percent; while ‘highly indebted’ countries have PV/GNP of over 220 percent and PV/XGS of over 80 percent. No explanation is given for how these exact figures are derived. These ratios are also difficult to compare with the IMF data used in this paper because the ratio of PV to nominal debt depends on the terms of the debt itself. However, and equivalent debt/gdp ratio for ‘highly indebted countries’ would appear to be about 150 percent, and 50 percent for the service/export ratio.

30

Those in the former category with private creditors are mainly in Latin America: their debt service ratios are higher and the average maturity shorter, as well as having higher interest rates. Indeed the Latin American UDCs are mainly suffering a liquidity problem, reflected in the fact that the ratio of reserves to short-term obligations (or ‘quick ratio’ as it is known by debt traders) is less than unity, making them susceptible to speculative attack. The African HIPCs in contrast, appear to be insolvent rather than illiquid: the inability to repay principle giving very long implicit maturities (i.e. years required to pay off debt at present rates). In marked contrast, Asia appears to be both solvent and liquid.

Table 5.1 Indicators of debt vulnerability, 2003 Total of

which: Developing Asia

Latin America & Caribbean

Africa UDC HIPC

External Debt (% of GDP)

38.1 25.4 43.9 49.9 63.2 86.9

Debt Service (% of exports)

17.9 11.1 45.8 13.1 29.2 10.2

Interest Payments (% of debt)

3.9 3.6 5.4 2.8 3.3 1.1

Implicit maturity (years) 8.2 8.6 5.7 16.1 9.4 48.7 Liquidity Ratio 1.73 3.16 0.74 2.04 1.14 2.86 Source: own calculations based on IMF WEO May 2005. Notes: ‘UDC’ are Unsustainable Debt Countries’ with arrears and/or rescheduling during 1997-2001; HIPC are ‘highly indebted poor countries’ under consideration by the WB and IMF for debt cancellation; ‘interest rate’ is average of interest payments divided by debt outstanding: long-term average interest rates in advanced economies averaged 5 percent in this period; ‘implicit maturity’ is outstanding debt divided by amortisation payments; ‘liquidity ratio’, is the ratio of reserves to payments during the year (short term debt principle plus service of long-term debt).

None the less, as Table 6.2 shows, sustainability in all areas and for all debtor classes has clearly improved over the past decade. This appears to be due to export growth and import contention (allowing current account balances to move into surplus) rather than significant reduction on debt level as such. Indeed all regions appear to be running current account deficits that are less than what the prudent rules would indicate. Although this restraint is doubtless due to credit rationing on the part of creditors and excessive stabilisation efforts on the part of debtors, it does imply that there is room to initiate a new debt cycle. This should clearly be accompanied by prudent macroeconomic policy of the type outlined in the previous section.

31

Table 6.2 Changes in debt sustainability 1996-2003 Total of

which: Developing Asia

Latin America & Caribbean

Africa UDC HIPC

Debt Levels Debt/GDP 1996 37.8 31.2 35.0 69.0 51.0 126.9 2003 38.1 25.4 43.9 49.9 63.2 86.9 GDP growth 1996-2005

5.1 6.6 2.6 3.9 3.5 4.8

CAB/GDP 1996 - -1.9 -2.2 -1.1 - - 2003 - 3.1 0.3 -0.1 - - ‘prudent value’ (d) -1.9 -1.9 -1.0 -2.3 -2.0 -5.1 Debt Service Service/exports 1996 21.5 13.9 46.7 20.3 29.2 22.6 2003 17.9 11.1 45.8 13.1 29.2 10.2 Export growth 1996-2005

10.8 12.0 7.0 8.1 8.1 7.1

CAB/exports 1996 - -7.3 -14.7 -3.6 - - 2003 - 8.9 1.4 -0.29 - - ‘prudent value’ (σ) -2.1 -1.5 -3.2 -1.4 -2.4 -1.24 ‘Golden rule’ export growth (x) 10.8 12.0 7.0 8.1 8.1 7.1 interest rate (i) 4.8 4.1 6.6 4.0 4.0 1.5 sustainability (x>i) > >> ≈ > > >> Source: Authors calculations from IMF WEO May 2005. Note: for definitions, see Section 3 above. The debt/gdp and service/export levels used in the calculation of prudential CAB and ‘golden rule’ are the simple average of 1996 and 2003.

6.2 Policy implications for developing countries Debt levels must clearly be kept within prudent limits, but governments should also make credible commitments to keep within these constraints, employing appropriate legislation if necessary. This is essential to reduce uncertainty for domestic firms, who are the main vehicles of investment on which growth depends, as a debt overhang and the prospect of deflationary stabilisation policies and debt restructuring (or even moratoria) imply future losses of sales, profits and asset values. Debt should be contracted on the longest terms possible. The cost of servicing should be kept at a minimum, but vulnerability to future capital market or world trade shocks

32

should also be a key consideration. Thus higher interest rates may be a reasonable price to pay for loans of longer maturity if vulnerability can be reduced thereby.53 The use of funds generated by external debt should be geared to ensuring repayment capacity. As we have seen, this means that a substantial proportion of these funds should be allocated to the support of export growth. This does not imply that the government should be directly engaged in export production but rather that these funds should be used to support appropriate infrastructure provision, supply of long-term credit to exporters and training for the workforce. The support of export growth also involves maintaining a competitive real exchange rate, which has implications both for nominal exchange rate management on the one hand, and for wage bargaining policy on the other. The political economy constraints on excessive reduction of real wages can be countered by appropriate commitments to output and thus employment growth. A low real interest rate and an expansionary credit policy are needed to support the required investment rate. This in turn means that the domestic financial system should to some extent be shielded from international capital markets. The recent popularity of inflation targeting as the central stabilisation policy in emerging market economies does not help reduce debt vulnerability because it has the effect of increasing the cyclical effect of capital flows. The opening to capital flows and a floating exchange rate arising from financial liberalisation has meant the use of single monetary policy instrument (the interest rate, combined with rigid fiscal and reserve ‘rules’) in emerging market economies to achieve price stability. This precludes not only countercyclical monetary or fiscal policy, but also the use of the exchange rate to maintain export competitiveness - the key to prudent debt management. There is thus a strong argument for emerging market authorities to adopt a counter-cyclical monetary stance in response to capital flows. This would involve real exchange rate targeting, bank credit regulation and an active fiscal stance and can be shown be more effective in supporting growth and investment.54 For such a policy to be successful in middle-income countries with substantial short-term private capital flows there is thus a strong practical case for intervention in order to reduce the volatility of capital flows.55 These controls are now usually based on price measures, particularly taxes, while quantitative instruments have become less common. Open-market operations have also proved quite successful in this regard, and can be complemented by the active use of reserve requirements and public sector deposits. Domestic regulatory systems for banks and securities markets (including corporate borrowing abroad) are also important supportive instruments. 53 Missale (1999) demonstrates how this principle has been applied in OECD countries. 54 See FitzGerald (2005b). 55 See FitzGerald (2005a).

33

6.3 Policy implications for the international community Although this study is not directly concerned with the international financial architecture, there are a number of policy areas that can only be addressed by the international community. Substantial debt reduction has not yet been forthcoming, even for HIPC countries, due to difficulties in budgetary assignments for the corresponding asset write-downs. This is an internal accounting matter for OECD countries and requires urgent solution. Further debt restructuring can reduce the liquidity problem of debt service pressure on the current account, but it does not reduce the investment disincentives from debt overhang and may even make them worse by increasing uncertainty.56 Given that export growth is the key to prudent debt management, access to OECD markets for developing country exporters is clearly crucial to their ability to contract debt prudently and accelerate economic growth and poverty reduction. The same is true of measures to reduce speculative fluctuations in primary commodity prices. Ideally, these would be combined with linkage of debt repayments to export levels – at least in the case of those to official creditors.57 Insofar as a large element of capital shocks to developing countries originate within OECD financial markets, it is important to recognize these externalities. A preliminary step in reducing the impact of these exogenous shocks would be for the IMF to provide temporary finance on a larger scale, more quickly and with less conditionality in order to ensure smooth debt management. In the longer run, it is essential to deepen the market for developing country debt within OECD countries by: lengthening the tenor, ensuring liquidity and encouraging pension and insurance funds holdings.58

56 This effect far outweighs any potential moral hazard implicit in ‘front loading’ debt forgiveness. 57 In practice markets are very unlikely to accept sovereign bonds with yields linked to commodity exports, certain primary commodities can be used as collateral. 58 This of course would also be in the interest of developed country portfolio investors by reducing the riskiness of high-yielding sovereign debt.

34

References Aghion, P. and P. Howitt (1998) Endogenous Growth Theory Cambridge, MA: MIT Press Addison, D. (1989) “The World Bank Revised Minimum Standard Model: concepts and issues”, Working Paper Series 231, Washington DC, World Bank [an updated (1999) version of this paper plus other RMSM documents can be found on: http://www.worldbank.org/data/rmsm/index.htm] Agenor, P-R and P. J. Montiel (1999) Development Macroeconomics, 2nd edn Princeton, NJ: Princeton University Press Avramovic, D. and others (1964) Economic Growth and External Debt Baltimore MD: Johns Hopkins UP Bacha, E.L. (1990) ‘A Three-Gap Model of Foreign Transfers and the GDP Growth Rate in Developing Countries’ Journal of Development Economics 32 Baqir, R., R. Ramcharan and R. Sahay (2003) ‘IMF Program Design and Growth: What is the Link?’ Research Workshop on Macroeconomic Challenges in Low Income Countries Washington DC: International Monetary Fund Buiter, W.H. (1990) Principles of Budgetary and Financial Policy New York: Harvester Wheatsheaf Chenery, H.B. and A.M. Strout (1955) ‘Foreign assistance and economic development’ American Economic Review 65. Cohen, D. (1994), "Growth and External Debt", in Van der Ploeg, F. (ed.), Handbook of International Macroeconomics, Oxford: Blackwell Cunningham A, Dixon L, Hayes S. 2001. Analysing Yield Spreads on Emerging Market Sovereign Bonds. Bank of England Financial Stability Review. 175-186. Dornbusch, R. and F. Helmers (1988) The Open Economy: tools for policymakers in developing countries, New York, OUP. Easterly, W. (1999) ‘The ghost of the financing gap: the growth model used in the international financial institutions’ Journal of Development Economics 60: 423-438 FitzGerald, V., K. Jansen and R. Vos, (1994) ‘External constraints on private investment decisions in developing countries", in J.W. Gunning, H. Kox, W. Tims and Y. de Wit (eds), Trade, Aid and Development Basingstoke: Macmillan FitzGerald, V. (2001) ‘Short-term capital flows, the real economy and income distribution in developing countries’ pp. 29-51 in S. Griffith-Jones, M. Montes and A. Nasution (eds) Short-term Capital Flows and Economic Crises Oxford: Oxford University Press for UNU/WIDER

35

FitzGerald, V. (2003a) Global Markets and the Developing Economy Basingstoke: Palgrave FitzGerald, V. (2003b) ‘The instability of international demand for emerging market assets’ pp. 191-213 in R. Ffrench-Davis and S. Griffith-Jones (eds) From Capital Surges to Drought: Seeking Stability for Developing Economies Basingstoke: Palgrave for UNU/WIDER FitzGerald, V. (2005a) ‘Policy issues in market based and non-market based measures to control the volatility of portfolio investment’ pp. 241-274 in C.J. Green, C. Kirkpatrick and V Murinde (eds) Finance and Development: Surveys of theory, Evidence and Policy Aldershot: Edward Elgar FitzGerald, V. (2005b) ‘Monetary models and inflation targeting in emerging market economies’ in Arestis, P., J. McCombie and M. Baddeley (2005), The New Monetary Policy, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. FitzGerald, V. (forthcoming) ‘International Risk Tolerance, Capital Market Failure and Capital Flows to Emerging Markets’ in A. Shorrocks (ed.) Thinking Ahead: the Future of Development Economics Basingstoke: Palgrave for UNU/WIDER FitzGerald, V and S. Babilis (2005) “Risk appetite, home bias and the unstable demand for emerging market assets” International Review of Applied Economics volume 19.4 FitzGerald, V. and G. Perosino (1999) ‘Trade liberalisation, employment and wages: a critical approach’ in G. Barba-Navarretti, R. Faini and G. Zanalada (eds) Labour Markets, Poverty and Development Oxford: Clarendon Griffin, K.B. (1970) ‘Foreign Capital, Domestic Savings, and Economic Development’, Oxford Bulletin of Economics and Statistics 32 (2) Heijdra, B.J. and F. van der Ploeg (2002) The Foundations of Modern Macroeconomics Oxford: Oxford University Press IMF (1987) ‘Theoretical aspects of the design of Fund-supported adjustment programs’, IMF Occasional Paper 55, Washington DC, International Monetary Fund. IMF (2003) ‘Public debt in emerging markets – is it too high?’ World Economic Outlook 2003, Chapter III Khan, M. S., P. J. Montiel and N. U. Haque (1990) ‘Adjustment with growth: relating the analytical approaches of the World Bank and the IMF’, Journal of Development Economics, 32: 155-179. McCombie, J. and A.P. Thirlwall, (1994) Economic Growth and the Balance of Payments Constraint Basingstoke: Macmillan Missale, A. (1999) Public Debt Management Oxford: Oxford University Press

36

Obstfeld, M. and K. Rogoff, (1995) ‘The intertemporal approach to the current account’ in G.M. Grossman and K. Rogoff (eds) Handbook of International Economics vol.III, Oxford: Blackwell Polak, J.J. (1957) ‘Monetary analysis of income formation and payments problems’ IMF Staff Papers 6(1): 1-50 Rebello, S. (1991) ‘Long run analysis and long run growth’ Journal of Political Eocnomy 94: 500-21 Rodrik, D. (1991) ‘Policy uncertainty and private investment in developing countries’ Journal of Development Economics, 36, pp. 229-242 Romer, P.M. (1986) ‘Increasing returns and long run growth’ Journal of Political Economy 94: 1002-1037 Sachs, J. D. (1989), Developing Country Debt and the World Economy, Chicago IL: University of Chicago Press for NBER Sen, P. (1994) ‘Savings, investment, and the current account", in F. van der Ploeg (ed), The Handbook of International Macroeconomics, Oxford: Blackwell Taylor, L. (1994) ‘Gap models’, Journal of Development Economics vol 45 Taylor, L. ed (1993) The Rocky Road to Reform: Adjustment, Income Distribution, and Growth in the Developing World Boston MA: MIT Press for UNU/WIDER Tran-Nguyen, A.N. (2004) ‘Debt Servicing-Capacity of Developing Countries: Conceptual Approaches and Measurement Problems’ Geneva UNCTAD, processed UNCTAD (2005) World Investment Report 2005 Geneva: UNCTAD World Bank (2004) Global Development Finance 2004 Washington DC: World Bank