Analytical Overview and Literature Survey

Assessing the macroeconomic impact of scaling up foreign aid is important, since it may have significant effects on competitiveness, economic growth, and poverty reduction. The spending and absorption of aid-financed expenditures in the short to medium term could cause a significant appreciation of the real exchange rate, thereby discouraging the expansion of exports and hurting long-term growth.³ This is often called the “Dutch disease” effect of aid, because the decline of the tradable goods sector relative to the nontradable sector could impair productivity growth, which is particularly strong in the export sector owing to the high rate of learning by doing in this sector. This has been borne out by empirical studies, including for African manufacturing firms. According to Rajan and Subramanian (2005), the potential adverse effects of aid on competitiveness and exports explain the weak link between aid inflows and growth in developing countries.

Because aid largely accrues to governments, the macroeconomic consequences are determined by what governments do with the aid (Adam, 2005). If it is entirely saved, or if the ultimate recipient of the aid, whether in the public or the private sector, spends the entire increase on imports, the real exchange rate will be unaffected, at least initially. On the other hand, it is far likelier that aid inflows will boost demand for both imports and domestically produced (nontradable) goods. If the public sector has a greater propensity to consume domestically produced goods, the demand for nontradable goods could be stronger if aid is used to finance increased public expenditure rather than direct transfers to households or tax cuts. In either case, the mechanism is the same so that differences in outcomes are a matter of degree (Adam, 2005).⁴

As noted by Heller (2005), fiscal policy can become even more complicated in a high-aid environment. If government scales up basic services—an action associated with hiring workers, delivering services to the public, and maintaining new infrastructure—it faces the challenge of what to do if and when donors do not sustain the aid. If governments find it difficult to reduce expenditures from aid-financed levels, the pressure for greater domestic financing of the deficit may increase significantly. Buffie and others (2006) argue that if the public suspects that donor financing of scaled-up and partially irreversible public spending commitments may be temporary, an aid surge may be highly inflationary in the short run. Monetary policy alone cannot handle this problem, because with thin and undeveloped bond markets, sterilization causes interest payments and future seigniorage requirements to balloon, and the resulting increase in expected future inflation creates inflation problems in the near term. Barring interventions that directly address concerns about aid volatility (Bulíř and Hamann, 2005) or the reversibility of fiscal commitments (for example, through temporary wage contracts), some degree of near-term fiscal restraint is a necessary component of a successful strategy. According to Buffie and others (2006), the right policy package combines a critical minimum degree of fiscal restraint with either reverse sterilization (buying back some internal debt) or a reserve buffer stock scheme (allocating some of the aid inflow to reserve accumulation). It is also important to project an exit strategy—that is, the macroeconomic path the country envisages following after scaled-up aid flows fall to more normal levels or, perhaps, if aid flows have been front-loaded, to lower-than-normal levels. Given the vast socioeconomic needs and infrastructure gaps in Mozambique, as in much of sub-Saharan Africa (Ndulu, 2007), the exit strategy is best accomplished through higher domestic revenue mobilization in order to maintain long-term fiscal sustainability, particularly when countries are below their potential for raising tax revenues.⁵

The short-run impact on demand of spending aid could be mitigated—or even reversed—depending on the supply side response over the medium to long term. When learning by doing also has externalities in the non-tradable sector, the long-term adverse impact of spending aid will be limited, even if the real exchange rate appreciates in the short term (Torvik, 2001). Investments in physical and human capital, both in the government and in the private sector, can increase productivity not only in the tradable sector but also in the nontradable sector, potentially offsetting the initial loss of competitiveness. If productivity gains from aid-financed public investment can be secured so that the export sector’s share of total output expands in the medium to long term, the issue is simply an intertemporal one, at least in the aggregate; future growth in the export sector will compensate for the temporary growth-retarding effects of a short-run real exchange rate appreciation (Adam, 2005). Containing unproductive spending and targeting the poor, to efficiently absorb larger amounts of foreign aid and improve public service delivery, are critical, and will require a well-functioning public financial management (PFM) system, including at the subnational level.⁶

Real exchange rate misalignment should be avoided because it may be more costly than shifts in the equilibrium real exchange rate. The interaction of the demand- and supply-side effects of aid means that the real exchange rate may “overshoot” its long-run value and may, in fact, move in the opposite direction, so that a short-run appreciation is followed by medium-term depreciation. Temporary movements of this kind can be very costly because firms face high adjustment costs and are unable to access credit from underdeveloped financial markets to finance the short-run losses caused by unfavorable temporary real exchange rate movements, even if these movements are anticipated. In such circumstances, firms may run down their capital, lay off skilled workers, or, in a worst-case scenario, close down completely, even though the long-run prospects for the tradable sector may be strongly favorable. Hence, short-run movements in real exchange rates may have permanent effects on the structure of production and growth. Bleaney and Greenaway (2001) estimate a negative effect of lagged exchange rate misalignment on growth covering 14 sub-Saharan African countries over 1980–95, while overvaluation might hurt investment even though it lowers the price of imported capital goods. If firms falsely believe temporary real exchange rate movements to be permanent, they incur costs as they first move into (what they think is) the booming sector and then out again when the temporary effects pass. It is important to distinguish here between the volatility of aid flows and the volatility of the real exchange rate itself. It is the latter that matters for intersectoral resource allocation decisions. Whether the former mitigates or exacerbates the latter depends on whether aid is pro-or countercyclical (Adam, 2005).

The dynamic macroeconomic effects of large aid flows in sub-Saharan Africa remain largely an empirical issue, but the literature has yielded somewhat inconclusive results, warranting further analysis. Econometric estimates of the impact of aid on the real exchange rate often show what Bulíř and Lane (2002) have referred to as “traces” of aid-induced real exchange rate appreciation, while evidence of aid-financed contraction of the export sector is equally mixed (Arellano and others, 2005; and Yano and Nugent, 1999). The small or insignificant Dutch disease effects may be explained by the tendency for the real exchange rate to depreciate as a result of policy reforms (for example, trade liberalization) associated with aid conditionality that are difficult to capture but that bias the effects of aid and fiscal, monetary, and exchange rate policy responses to aid downward.⁷ Aid inflows can also play an important role in financing imported inputs, which may affect short-run supply responses as otherwise idle capacity is brought into use or leads to stronger productivity growth in the nontradable sector as noted above, potentially triggering a depreciation in the real exchange rate (Adam, 2005). As shown in Prati and Tressel (2006), sterilization may also be an effective tool for depreciating the real exchange rate in the short run, although attempts to target the real exchange rate below its equilibrium level are likely to be associated with higher inflation or real interest rates (Calvo, Réinhart, and Végh, 1995). An excessive reliance on open market operations is also likely to crowd out private sector credit (Christensen, 2004), thereby reducing private investment (Servén, 2002) and, ultimately, undermining economic growth in low-income countries (Abbas and Christensen, 2007).

Following the literature, the next section attempts to estimate the fiscal and macroeconomic effects of aid in Mozambique over the past decade or so by using quarterly data in the post-stabilization phase (1996–2006), in an attempt to overcome some of the weaknesses of previous analyses of foreign aid in low-income countries, such as small sample biases and structural breaks that make estimates sensitive to different test specifications. The insights in the section “Aid and Macroeconomic Management in Mozambique” will help guide a discussion in the following section of the potential macroeconomic implications and policy trade-offs of a further scaling up of aid to Mozambique.

Aid and Macroeconomic Management in Mozambique

Identifying the fiscal effects of aid is a prerequisite to understanding the macroeconomic effectiveness of aid (McGillivray and Morrissey, 2001). However, there has been limited analysis of the fiscal effects of aid in Mozambique.⁸

There is a growing literature on how aid affects the fiscal behavior of governments (reviewed in McGillivray and Morrissey, 2001). The most common approach is through fiscal response models (FRMs). These studies tend to find that aid ultimately leads to increased spending, and total spending often increases by more than the value of aid (McGillivray and Morrissey, 2001). There is evidence that aid has had a beneficial impact on investment and recurrent spending in sub-Saharan African countries (Commission for Africa, 2005). IMF (2005) suggests that Mozambique spent most of the aid it received. The degree of spending was calculated by the widening in the government fiscal deficit, net of aid, that accompanies an increase in aid. IMF (2005) shows that public expenditures actually increased, on average, more than the increment in net aid inflows, leading to a substantial widening of the fiscal deficit net of aid during the aid surge of 2000–02. The aid surge was roughly equally distributed between current and capital expenditures. However, this study focuses on short periods when aid surged and takes a relatively simplistic approach compared with the fiscal response literature of aid. It may therefore have missed more complicated dynamic effects (Mavrotas, 2002).

The fiscal effects of aid in Mozambique are assessed by estimating an FRM within a vector autoregression (VAR) modeling framework as in Osei, Morrissey, and Lloyd (2005). The variables of the fiscal response VAR model are ordered as follows: foreign aid (AID), government expenditure (Expenditures), tax revenue (Tax), and the change in net credit to the government from the banking system (NCG). Note that since nontax components of revenue and nonbank borrowing are omitted, we are not estimating an identity. We also estimate a model in which foreign aid is divided into grants and loans as well as project aid and budget support. Total government expenditure is disaggregated into capital and current components. We use quarterly data over the period 1996Q1–2006Q3, with all the variables measured in constant 2004 prices expressed in millions of new meticais (MT). Data on domestic fiscal variables and external financing are from the IMF.

The dynamic effects of aid shocks can be evaluated using impulse response functions. An analysis of the time-series properties of the variables reveals that the variables are integrated of order one or I(1), except for AID and NCG, which are stationary I(0). However, cointegration tests did not identify a significant cointegrating vector between the variables. Therefore, the series are first differenced for estimation purposes to avoid the spurious and inconsistent regression problem (Hendry, 1995).⁹ Results of Granger causality tests are somewhat mixed but clearly point to aid Granger-causing fiscal variables but not vice versa, suggesting that aid disbursements have not been influenced by the budget balance over the period. The results also suggest that shocks to foreign aid are exogenous to the system, rather than determined by it, and offer statistical support for the legitimacy of the impulse response analysis of aid shocks in Mozambique below. The structural shocks are recovered from the VAR residuals using the Cholesky decomposition of the variance-covariance matrix.¹⁰

Plots of the impulse response functions for a one standard deviation shock in aid are shown in Figures 6.1 and 6.2. A one standard deviation shock corresponds to about MT 1,200 million (about 1 percent of GDP in 2004). Figure 6.1 shows that the impulse response to an aid shock of one standard deviation gives a full pass-through (change in spending t periods after the shock over an initial percent change in aid) within four quarters. In fact, as observed in IMF (2005), public expenditures actually have a tendency to increase more than the original shock to aid in Mozambique. However, in contrast with the findings in IMF (2005), the aid shock seems to result in more capital expenditures, with only about one-third of aid financing current spending within a year (Figure 6.2).¹¹ Not only does AID result in an increase in capital spending but it also seems to lead to an increase in domestic taxation. Moreover, these results are largely unaltered, whether they relate to loans or grants.¹² The picture is somewhat less clear regarding the impact of AID on NCG, possibly because of the smoothing of expenditures or issuance of domestic debt.

View Full Size

Fiscal Effects of Foreign Aid

(Total Expenditures)

Source: Author’s calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Fiscal Effects of Foreign Aid

(Current and Capital Expenditures)

Source: Author’s calculations.

• Download Figure
• Download figure as PowerPoint slide

The spending of foreign aid could undermine competitiveness, although such effects can be mitigated, depending on the supply-side response over the long term. Given that aid inflows have been spent by the government in Mozambique, the next issue is whether these expenditures have induced a significant appreciation of the real exchange rate and thus discouraged the expansion of exports, thereby hurting long-term growth. The results presented in Chapter 10 show that government spending (both current and total expenditures) has tended to lead to the appreciation of the fundamental equilibrium real exchange rate, but some of those effects are mitigated by trade liberalization and negative terms of trade movements. ¹³ In addition, even though the share of Mozambique’s total exports in world trade has increased, the strong growth of megaprojects accounts for most of this increase.¹⁴ The share of non-megaproject exports in world trade and the ratio of non-megaproject exports to GDP have remained roughly constant. Concerns therefore remain regarding the impact of scaled-up aid on Mozambique’s competitiveness. These results are somewhat different from those presented in Benito-Spinetto and Moll (2005), who simulate a simple 1-2-3 model in the spirit of Devarajan and others (1994) and indicate that Dutch disease appears not to be an important factor in Mozambique. They argue that the supply-side effects of aid in the form of investment in health care and infrastructure could help mitigate a real exchange rate appreciation. The lack of visible evidence that supply-side effects are strong enough to dampen appreciation pressures on the real exchange rate stemming from the demand-side impact of aid spending may, however, be explained by well-known lagged (and possibly nonlinear) effects, as well as by the rudimentary state of Mozambique’s PFM systems, which, at least until recently, were weaker than those in most other sub-Saharan African countries.¹⁵

Even the short-run response to aid is dependent on a number of factors, including the degree of aid absorption.¹⁶ As discussed in Chapter 5, the impulse responses of the estimated dynamic stochastic general equilibrium (DSGE) model for Mozambique show that the degree of absorption of aid-financed spending can have very different effects on the short-term dynamics of the exchange rate, output, and inflation. To see how those effects may have played out in Mozambique, we estimate an identified six-variable VAR from 1996Q1 to 2006Q3.¹⁷ The ordering of the variables embodies two key identifying assumptions—that is, that aid shocks contemporaneously affect all variables in the system and that economic variables do not respond contemporaneously to policy variables, except for the exchange rate. As an asset price, the exchange rate is expected to respond faster to policy shocks than real economy variables. The following ordering is chosen: foreign aid, real GDP¹⁸ (or real non-megaproject exports), inflation, reserves, domestic debt, and the nominal exchange rate. The structural shocks are recovered from the VAR residuals using the Cholesky decomposition of the variance-covariance matrix. While the ordering is certainly debatable, and a non-recursive structural VAR could have been used for identification, the estimation provides a generalized view of the impact of aid shocks in Mozambique. In addition, it is fairly well accepted that many macroeconomic variables do not respond instantaneously to policy shocks, particularly when we consider quarterly time intervals (see Christiano, Eichenbaum, and Evans (2005) for the general approach).

Aid-financed expenditures have been mostly absorbed in Mozambique, with only hints of Dutch disease affecting the export sector (Figures 6.3 and 6.4). The impulse responses suggest that aid (most of which has been spent) has, at least initially, resulted in a nominal exchange rate appreciation and, possibly, lower output. Inflation declines initially but picks up gradually. A similar picture emerges when one considers real non-megaproject exports instead of our proxy for real GDP, suggesting only modest contractions of the export sector in response to aid shocks. Interesting, most of the foreign exchange associated with aid inflows appears to have been sold by the central bank within five quarters of the initial aid shock, facilitating a textbook spend-and-absorb response to aid. The results do indicate, however, that the central bank may have some-what smoothed sales of foreign exchange in the very short term through temporary net domestic debt issuance (that is, sterilization), possibly to avoid excessive exchange rate volatility in a thin market. Temporarily mopping up excess liquidity through sterilization but ultimately selling the foreign exchange may have helped keep a rise in prices at bay and minimized the likelihood of crowding out private sector credit at the same time. This contrasts with the response to aid shocks of most other non-CFA countries in sub-Saharan Africa, which seem to have relied more heavily on sterilization (IMF, 2005), contributing to unfavorable debt dynamics and, probably, greater crowding out of the private sector. Overall, the macroeconomic impact of aid shocks seems to go roughly in the same direction as the model simulations for Mozambique in Chapter 5, in the case where aid has been absorbed through sales of foreign exchange by the central bank.

View Full Size

Mozambique: Macroeconomic Effects of Foreign Aid

(Total Exports)

Source: Author’s calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mozambique: Macroeconomic Effects of Foreign Aid

(Non-Megaproject Exports)

Source: Author’s calculations.

• Download Figure
• Download figure as PowerPoint slide

Illustrative Scaling-up Scenarios

The purpose of illustrative scaling-up scenarios for Mozambique is to identify the potential macroeconomic implications and key measures and policies that would help the country absorb a higher level of aid and use it efficiently. A full costing of the MDGs has not been undertaken for Mozambique, but the country appears to be well placed to achieve the income poverty MDG (halving the poverty rate by 2015) with present resources and policies.¹⁹ Therefore, we consider the macroeconomic consequences of an arbitrary but moderate and sustained increase of external finance for illustrative purposes.²⁰ This contrasts with scaling-up scenarios presented for a few other countries (for example, Mattina, 2006) based on costing the MDGs and targeting a specific growth rate to meet the income poverty MDG. The focus of this chapter is to discuss the macroeconomic implications and policy trade-offs involved in spending and absorbing a modest scaling up of foreign aid, which currently amounts to about 15 percent of Mozambique’s GDP.

Macroeconomic Implications and Policy Trade-offs

The impact on medium- to long-term growth of a further scaling up of aid is difficult to gauge, but research points to areas for further analysis and prioritization of reform strategies. A number of research papers that have undertaken growth accounting exercises for Mozambique (for example, Chapter 3 of this book; Manoel and others, 2005; and World Bank, 2005) suggest that physical capital accumulation and total factor productivity have played a fundamental role in explaining the post-conflict growth acceleration supported by the first generation of structural reforms.²¹ They also confirm the possibility of achieving about 7 percent annual growth in the medium term and 5–6 percent in the long term, which would be sufficient to halve the poverty rate by 2015, assuming strong productivity growth continues, given relatively low productivity in agriculture and man-ufacturing (Chapter 2 of this book; and Eifert, Gelb, and Ramachandran, 2005). On the specific issue of the impact of scaling up, Chapter 3 highlights the upside potential from more public investment (depending on the ratio of current expenditures to capital spending), and crowding-in private investment and spillovers from infrastructure and human capital accumulation. While the formal modeling of absorptive capacity is in its infancy (for a discussion, see Bourguignon and Sundberg, 2006), a preliminary estimation in Chapter 3 also shows the possible negative consequences of a further scaling up in Mozambique operating through private investment and the growth of the quality of human capital. The former is intended to reflect diminishing returns while the latter reflects labor market effects, including the challenge for Mozambique of maintaining educational quality in the face of rapid aid-financed expansion of the school network (see Arndt, Jones, and Tarp, 2007). Recent evidence suggests that primary education completion rates and the quality of education are relatively low, probably because of high pupil-to-teacher ratios and the growing number of untrained teachers, although innovative steps have been taken recently to address the shortage of trained teachers.²²

The literature and evidence presented in the section “Aid and Macroeconomic Management in Mozambique” above call for paying particular attention to potential Dutch disease effects related to aid-financed scaling up of expenditures and thus ways to mitigate such effects. While it is fairly well accepted that to spend and absorb is likely to be the best response to an aid shock in normal circumstances (see Buffie and others, 2006; IMF 2005; and Chapter 5 of this book), such a response could still entail a significant loss of competitiveness and thus dampen exports and real GDP growth. Therefore, we analyze this issue further by simulating the estimated DSGE model for Mozambique in Chapter 5 in response to a persistent aid shock (autocorrelation coefficient of 0.7) that raises aid by 2 percent of steady-state GDP.²³ For illustrative purposes, we consider three scenarios (Figure 6.5). In the first scenario, we assume that there are no productivity spillovers from public investment. The second scenario assumes a modest level of spillovers (the baseline in Chapter 5). In the third scenario, we consider a high level of productivity spillovers from public investment. All scenarios assume the full spending and absorption of foreign aid, the usual assumption in scaling-up scenarios. Note that the function form, as in Prati and Tressel (2006), assumes that productivity is an increasing function of the size of public investment expenditure, so the higher levels of productivity could come from greater spillovers (for example, related to infrastructure investments) or a decrease in the ratio of current expenditures to capital spending.²⁴ The unbroken line describes a scenario where there are no productivity spillovers; the thick broken line assumes a modest level of spillovers; and the light broken line depicts a higher level of productivity spillovers from public investment in response to aid shocks (Figure 6.5).

View Full Size

Illustrative Scaling-up Scenarios

Source: Author’s calculations.

• Download Figure
• Download figure as PowerPoint slide

Greater productivity-enhancing public investment expenditure could ameliorate or even reverse the Dutch disease effects of additional government spending in Mozambique. The impulse responses in Figure 6.5 on a quarterly time interval clearly show the role productive government investment expenditures can play in mitigating the Dutch disease effects of aid-financed government spending by dampening appreciation pressures on the real exchange rate and thus stimulating stronger export performance. Moreover, as expected, consumption and output are unambiguously higher with greater productivity spillovers, while the pickup in inflation following an initial decline (as in the section “Aid and Macroeconomic Management in Mozambique”) is milder. At this point, it is important to note that analysis of this kind comes with a number of caveats. For example, we assume for simplicity and tractability that productive government spending affects the transitional dynamics of the economy but not the steady state. Although this makes sense when considering fairly persistent but stationary aid shocks as observed in the past in Mozambique, one could also consider the impact of a permanent increase in aid on steady-state variables, an area for future research.²⁵ Finally, it is important to note that none of the outcomes of productivity spillovers from public investment are assured; the composition of government expenditure and efficiency of public service delivery will be critical in determining outcomes. However, the discussion emphasizes the importance of prioritizing productive government investments and accelerating the implementation of second-generation reforms in PARPA II to increase returns to investment:

• Infrastructure and human capital accumulation. Since 2000, about 65 percent of scaled-up spending has been allocated to the priority sectors (for example, education, health care, and infrastructure) identified in Mozambique’s first Poverty Reduction Strategy Paper—Plano de Acgão para a Redugão da Pobreza Absoluta I (PARPA I)—for 2000–05. The number of children in primary school has doubled; infant and maternal mortality has been reduced; and antiretrovi-ral (ARV) treatment has begun to be provided to Mozambicans infected with HIV. These achievements have been financed, in part, by resources made available by the Heavily Indebted Poor Countries (HIPC) Initiative. While substantial progress has been made, large human capital and infrastructure gaps remain. Comparisons of Mozambique with the fast-growing Asian economies indicate the importance of expanding secondary education and addressing acute infrastructure gaps, particularly in communication and transportation networks (Table 6.1).

• Improving agricultural productivity lies at the heart of growth prospects and poverty reduction (Chapter 2). Much of the increase in crop income has been achieved through area expansion and greater use of available labor, both family and hired. As a result, basic food crops widely grown by smallholder farmers, predominantly for subsistence, have exhibited relatively stagnant yields (output per hectare). Improving the security of land tenure and increasing the use of new technologies associated with cash crops mostly grown under contract by agroindustrial firms could help boost productivity. Smallholders could also increase land productivity and crop income by diversifying into profitable cash crops, many of which are tied to contract farming schemes. The use of productivity-enhancing inputs (particularly fertilizers, seeds, and irrigation) and development of rural credit and input markets could facilitate diversification to such crops. Further development of the agriculture sector would require the development of good roads, storage facilities, and extension services. Mozambique also has the potential to further cultivate high-value vegetables and flowers for export to Europe by improving quality and the marketing infrastructure.²⁶

• Strengthening the business environment is key to building a manufacturing base and generating employment. Sustained growth accelerations tend to be associated with manufacturing exports, but Mozambique’s manufacturing base remains underdeveloped because of a lack of competitiveness. Low factory-floor productivity levels are explained by high indirect costs and output losses, including those stemming from burdensome regulations that squeeze firms’ value added and reduce total factor productivity (TFP), and the use of obsolete and ill-maintained equipment and absence of modern management techniques (see Chapter 9, Figure 9.9). The wide gap in business competitiveness between Mozambique and the fast-growing Asian economies and competitors in sub-Saharan Africa, as measured by the World Bank’s Doing Business indicators, suggests that the authorities’ reform strategy must center on lowering the costs of doing business by (1) streamlining burdensome regulatory practices; (2) reducing the costs of hiring and firing workers; (3) removing infrastructure bottlenecks; (4) further reducing the cost of, and expanding access to, financial services; and (5) reducing expropriation risks.

• Strengthening public financial management systems and undertaking wider public sector reforms will be vital to ensure that resources efficiently reach the most economically and socially productive priority sectors, including at the subnational level, where some of the scaled-up resources will ultimately be spent (Box 6.1). Enhancing administrative capacity is also essential to the effective absorption of scaled-up donor assistance. To improve the link between policy objectives, appropriations, and performance indicators, the government is embarking on a transition to program-based budgeting. The government could proceed on a pilot basis with a few line ministries in the 2008 budget. In support of this initiative, the medium-term fiscal framework (CFMP) will also require strengthening through a better costing of policies and more comprehensive sector strategies. In addition, the significant role played by state-owned or public-participating enterprises²⁷ in the provision of public services, and their impact on macroeconomic developments—including domestic and external borrowing levels, and fiscal risks—argue for including them in the budgetary accounts or, at a minimum, enhancing the monitoring of their activities. Finally, the international community has a role to play by including donor-financed projects in the single treasury account and e-SISTAFE (see Box 6.1 for a description of e-SISTAFE) to improve timely fiscal reporting on all development expenditures and thus gauge their impact better.

Table 6.1.

Increasing Returns to Public Investment: Lessons From Asia

Source: Johnson, Ostry, and Subramanian (2007). Note: SADC = Southern African Development Community; SSA = Sub-Saharan Africa; ASEAN-4 = Indonesia, Malaysia, the Philippines, and Thailand.

¹Gross enrollment ratio.

²Data are for the most recent period available.

³Data refer to year closest to growth takeoff.

Table 6.1.

Increasing Returns to Public Investment: Lessons From Asia

			Infrastructure
	Human capital		Telephone mainlines (Per 1,000 people)	Total road surface area (In kilometers)
	Primary education1 (In percent)	Secondary education1 (In percent)
	Sub-Saharan Africa2
Mozambique	86.0	8.7	4.0	38.0
SADC average	102.0	42.5	66.0	227.5
SSA average	91.4	31.8	35.6	166.4
	Selected Asian countries that have sustained growth acceleration3
China	112.6	45.9	2.0	123.0
ASEAN-4 average	83.4	22.6	6.9	184.5
Indonesia	80.0	16.1	1.5	151.6
Malaysia	88.7	34.2	13.8	261.2
Thailand	81.4	17.4	5.5	140.6

Source: Johnson, Ostry, and Subramanian (2007). Note: SADC = Southern African Development Community; SSA = Sub-Saharan Africa; ASEAN-4 = Indonesia, Malaysia, the Philippines, and Thailand.

¹Gross enrollment ratio.

²Data are for the most recent period available.

³Data refer to year closest to growth takeoff.

View Table

Table 6.1.

Increasing Returns to Public Investment: Lessons From Asia

			Infrastructure
	Human capital		Telephone mainlines (Per 1,000 people)	Total road surface area (In kilometers)
	Primary education1 (In percent)	Secondary education1 (In percent)
	Sub-Saharan Africa2
Mozambique	86.0	8.7	4.0	38.0
SADC average	102.0	42.5	66.0	227.5
SSA average	91.4	31.8	35.6	166.4
	Selected Asian countries that have sustained growth acceleration3
China	112.6	45.9	2.0	123.0
ASEAN-4 average	83.4	22.6	6.9	184.5
Indonesia	80.0	16.1	1.5	151.6
Malaysia	88.7	34.2	13.8	261.2
Thailand	81.4	17.4	5.5	140.6

Source: Johnson, Ostry, and Subramanian (2007). Note: SADC = Southern African Development Community; SSA = Sub-Saharan Africa; ASEAN-4 = Indonesia, Malaysia, the Philippines, and Thailand.

¹Gross enrollment ratio.

²Data are for the most recent period available.

³Data refer to year closest to growth takeoff.

An Exit Strategy and Fiscal Sustainability

Domestic revenue needs to gradually increase to enable Mozambique to end its dependence on donors in an orderly manner. Donors could decrease foreign aid from present levels after the 2015 target date for achieving the MDGs. As a result, the strategy for exiting from aid dependence should aim to raise domestic revenue during the next eight years so that Mozambique will be able finance recurrent spending, at a minimum, from its own resources, without recourse to unsustainable domestic borrowing (or a disruptive expenditure contraction) to offset declining external assistance.

To evaluate whether the authorities’ long-term revenue target as described in World Bank and IMF (2007) is sufficient to maintain a scaling up of aid-financed expenditures in a sustainable manner, we assume a fiscal framework with additional expenditures of US$200 million from 2008 to 2015 compared with the baseline in IMF (2007).²⁸ Given the absence of a costing of the policies and sectoral interventions needed to achieve the MDGs, the increased expenditures are assumed to be divided between current and capital spending in a 1:2 ratio based on the historical distribution of aid shocks estimated above.²⁹ This is a simplifying assumption, and the ratio would no doubt be different for different programs, but it fulfills the requirement of planning for the recurrent expenditures and hiring of additional personnel in priority sectors targeted by the MDGs.

The authorities’ ambitious revenue targets are adequate to gradually reduce Mozambique’s dependence on donors without accumulating unsustainable debt in the event of a moderate scaling up of expenditures. The projected revenue ratio of about 19Vi percent of GDP in 2015 (increasing from 14 percent of GDP in 2006) would cover both a scaled-up level of recurrent spending and a portion of the country’s public investment pro-gram.³⁰ In addition, even if one assumed that all of the scaled-up spending was financed through concessional external borrowing, the external debt-to-export ratio would still remain below the indicative performance-based debt-burden thresholds (the ratio of the net present value of debt to exports is 150 percent for Mozambique), which take into account the empirical finding that the debt levels that a low-income country can sustain increase with the quality of its policies and institutions. This highlights the fiscal space available to scale up concessional external borrowing focused on achieving the MDGs, partly as a result of aid delivered under the Multilateral Debt Relief Initiative (MDRI). Mozambique’s prudent external debt-management strategy of avoiding recourse to nonconcessional external borrowing has also helped and is assumed to continue. In addition, the low level of domestic public debt in Mozambique (about 8V2 percent of GDP), which is due partly to the willingness of the central bank to mop up excess liquidity through foreign exchange sales and thus absorb foreign aid, has also contributed to a low level of debt distress. It should be noted, however, that exports, revenues, and real GDP growth will play an important role in defining the scope for expanded real government spending in the context of fiscal sustainability and an acceptable level of aid dependency.³¹ Given this, and to safeguard the exit strategy, spending plans would need to be carefully reassessed regularly in light of the growth and Dutch disease effects of the initial expansion in spending as well as realized revenue outcomes. As discussed in World Bank and IMF (2007), any nonconcessional financing of large infrastructure projects would need to be considered case by case, based on the projects’ economic returns, impact on debt sustainability, and potential effects on the financing decisions of donors and concessional lenders.

Box 6.1.

Public Financial Management (PFM) Reforms in Mozambique

The public expenditure and financial accountability (PEFA) assessment of 2006 shows that PFM systems in Mozambique have undergone major improvements in recent years. The government has been implementing a nationwide PFM reform called SISTAFE, initially with substantial IMF technical assistance and more recently with financing from a multidonor common fund. Mozambique’s medium-term fiscal framework (CFMP) is also closely aligned with the priorities of Mozambique’s Poverty Reduction Strategy for 2006–09 (PARPA II). It was approved by the Council of Ministers for the first time in 2006 and subsequently in 2007, making it a credible tool to guide the preparation of annual budgets. A government financial management information system (e-SISTAFE) was rolled out to all ministries at the central and provincial levels in 2007, facilitating better monitoring of expenditures. It is also possible to identify programs through the budget formulation module of e-SISTAFE, and to track priority expenditures defined in PARPA II on a real-time basis.

According to the authorities’ medium-term PFM action plan and budget (APB), e-SISTAFE will be progressively rolled out to districts and municipalities (initially to 37 districts in 2008), as well as to state organs at the central level in 2008. An integrated payroll database compatible with e-SISTAFE was developed following the completion of the civil services census in April 2007. The validation of the database by the Administrative Tribunal is now expected for 2008. However, the salary payments via e-SISTAFE will start in 2008 while the validation process continues to eliminate the fiduciary risk identified by the PEFA assessment.¹ A limited number of separate foreign currency accounts are now operational within the single treasury account (CUT), facilitating the inclusion of donor-financed projects on CUT. As part of Mozambique’s decentralization efforts, a National Decentralization Strategy, including a review of intergovernmental fiscal relations, will be prepared by end-2008. A new wage policy to sharpen incentives and increase accountability is also being designed with the help of the World Bank. Implementation of the new procurement system up to the district level will also be continued. Finally, the capacity of internal and external audit bodies is being increased, with specific milestones set for 2007–08.

View Full Size

Overview of 2004 PEFA Scores and Identified Potential Scores for 2006

Source: PEFA (2006).

• Download Figure
• Download figure as PowerPoint slide

Projected SISTAFE Outputs, 2006–09

1. Financial and budget execution in operation in all ministries at the central, provincial, and district levels, and in the municipalities and public enterprises.

2. The module of payment of salaries and pensions implemented.

3. The budget-formulation module implemented (Phase II).

4. The asset-management module and procurement interface implemented.

5. The ATM revenue-management module implemented.

6. Program budgets under preparation by ministries.

7. The debt-management module implemented.

8. The internal audit-module implemented.

9. The Data Processing Center operating as an effective and sustainable unit.

10. Project Implementation Unit for SISTAFE operating as an effective and sustainable unit.

Source: UTRAFE, Action Plan and Budget Report, 2006–09. (UTRAFE, the Technical Unity of Financial Administration of the State, was created to manage the reform of SISTAFE.) ¹ Payments of salaries and pensions continue to be executed centrally by the accounting department of the Ministry of Finance.

Mozambique would need to support this ambitious revenue increase by continuing revenue administration reforms and strengthening tax policies (particularly the fiscal regime for exploitation of mineral and petroleum resources). Varsano and others (2007) estimate the tax gap for Mozambique. Actual tax collection is considered to be a function of taxing capability—that is, the maximum collection possible given the country’s economic, social, institutional, demographic, and other characteristics—and of the effort exerted to mobilize funds for public use (determined by the tax laws and how strictly they are enforced). The study concludes that Mozambique’s tax effort (the ratio of actual collection to taxing capability) is about 50 percent, one of the lowest in sub-Saharan Africa, suggesting that there is significant scope for raising revenues, up to a level of nearly 22 percent of GDP. Schenone (2004) also estimates the difference between potential collection and actual collection, which is lower because of weaknesses in the tax laws (exemptions and nonassessments) and non-compliance (evasion). The study finds that noncompliance accounts for a much larger portion of the gap than the tax laws. The implication is that although there is room for increasing revenues through tax policy measures, the prospects are much greater for measures that increase the efficiency of the tax administration.³² There is a possibility that Mozambique could achieve medium-term revenue targets by widening its tax base with-out resorting to tax rate increases. The next phase of revenue administration reforms (2007–10), which will be supported by a multidonor common fund, will focus on implementing the strategic plan of the central revenue authority (ATM) and will be important in this regard. The ATM will be established in three stages: a transition period to end-2007, a gradual integration of the tax and customs agencies to take place during 2008, and further strengthening and consolidation in 2009 and 2010.

Mozambique’s vast natural resources and the present boom in world commodity prices are providing the country with a tremendous opportunity to attract greater foreign direct investment (FDI) in the mining and petroleum sectors. The goal should be to maximize the fiscal returns and economic linkages of FDI while minimizing environmental damage and social dislocation. Now that investor confidence in Mozambique is stronger, the authorities recognize that the fiscal contribution of new projects in these sectors could be substantially increased by reducing the generous tax exemptions put in place to attract private investment during the post-civil war period.³³ As discussed in Chapter 8, Mozambique is making important strides in this regard by putting in place a new fiscal regime for the mining sector and building the capacity of personnel to negotiate effectively with multinationals through the use of financial models and model contracts.

A significant amount of FDI has also flowed to the country’s petroleum sector.³⁴ Following agreements negotiated in 2000, the Pande-Temane gas field installations, processing facilities, and gas export pipeline to South Africa were commissioned in 2004, representing some US$1 billion in FDI. The Pande-Temane project (including the pipeline) is currently in a period of cost recovery and high debt service, so government revenues have been modest, comprising mainly petroleum production tax and a small amount of corporate income tax, but these are set to increase significantly as production ramps up and joint venture investors recover their initial development costs.³⁵ Further exploration is continuing and sufficient additional reserves have been identified that the joint venture partners³⁶ are considering a possible 50 percent expansion of the volume of annual gas sales to about 180 million gigajoules. Mozambique has also secured more than US$300 million in petroleum exploration commitments from the recent Rovuma Basin round, which may add US$200 million in planned drilling and evaluation activities to contracts awarded earlier. Mozambique has the potential to become a significant regional gas producer. Companies are exploring both for oil and for significant gas deposits that might supply liquefied natural gas (LNG) export facilities.

Even though fiscal incentives were granted for petroleum exploration and the investors benefit from stability assurances, the current combination of economic terms for exploration- and production-concession contracts (EPCCs), if properly administered, should yield returns that are competitive (for both the government and the companies) by international standards. In addition, the authorities are reviewing the fiscal regime for this sector and intend to put in place a comprehensive fiscal regime that would be embodied in the general tax law to avoid case-by-case negotiation of petroleum tax and production-sharing terms.

The mining and petroleum projects in the pipeline, including the multibillion dollar Moatize coal mine project now under development, could yield revenues in the range of 2–4 percent of GDP after 2010. Further oil and gas discoveries would result in significant revenue over the long term.

Policymakers need to take the budgetary importance of foreign aid and its volatility into account when designing the appropriate macroeconomic policy response to a scaling up of aid if it is much larger than the scenarios considered here. Foreign aid (both grants and loans) accounts for about 50 percent of total expenditures and about 120 percent of revenues in Mozambique. An increasing share of this net aid flow is also in the form of direct budget support, as opposed to project financing (see Chapter 7), and often explicitly or implicitly conditional on the expansion of public expenditure programs that tend to exhibit relatively high inertia, such as the expansion of health and education services. In the absence of institutional mechanisms to lock in high aid levels or guarantee rapid fiscal adjustment in the event of a decline in aid, it may be impossible to rule out the possibility that a further aid surge could be fiscally destabilizing. In this situation, a further large scaling up of foreign aid—even one that ultimately proves highly persistent—brings with it the expectation of an increase in future domestic financing requirements and seigniorage (Chapter 4). While a moderate scaling up of aid may not be a cause for concern in a country that has adequate international reserves and low public debt, which could help cushion a reversal of aid, a larger aid surge could sharply reduce money demand in the short run and lead to both high inflation and private capital outflows (Buffie and others, 2006). To avoid such a scenario, the authorities may consider smoothing the expenditure pattern and saving part of the aid surge in international reserves to be spent and absorbed at a later date.

Conclusions

Mozambique has fully spent and mostly absorbed scaled-up foreign aid (ranging between 10 percent and 20 percent of GDP) over the past decade or so. The additional expenditures have allowed Mozambique to scale up basic services, including doubling the number of children in primary school, reducing infant and maternal mortality, and beginning to provide ARV treatment for HIV infections, while sustaining economic growth of 8 percent a year on average and reducing the poverty headcount index from 69 percent in 1997 to 54 percent in 2003. The pursuit of prudent macroeconomic policies and first wave of structural reforms has helped Mozambique keep up this growth momentum by maintaining macroeco-nomic stability and a sustainable fiscal and external position. Looking forward, illustrative scaling-up scenarios highlight the need to carefully manage a further scaling up of foreign aid.

A modest scaling up of foreign aid inflows (about 2 percent of 2008 GDP) could continue to be spent on programs that are designed to help Mozambique achieve the MDGs and absorbed. A scaling up of spending, whether financed by donors or not, will put pressure on the real exchange rate to appreciate and thus has the potential to hurt the export sector and long-run growth. Therefore, there is a need to elicit a supply response in the most productive priority sectors (for example, agriculture, education, health care, and infrastructure as defined in PARPA II) and mitigate potential Dutch disease effects by strengthening PFM systems while embarking on a public sector reform program to improve efficiency and public service delivery. Enhancing productivity growth requires not only greater productive public investment but also complementary second-generation reforms aimed at strengthening the business environment and buttressing agricultural production. It could, however, be the case that policymakers view the level of macroeconomic variables (for example, inflation and the real exchange rate) as well as macroeconomic volatility, even in a spend-and-absorb aid scenario, to be too high from the point of view of consumer welfare and long-term growth, depending on developments in the real economy. This could be particularly relevant if the scaling up is significantly larger or less persistent than the aid shock considered here or if microeconomic capacity constraints become binding in some sectors. In such a case, the authorities may consider smoothing expenditures and saving part of the aid in international reserves to be spent and absorbed at a later date.

Debt relief and prudent macroeconomic management have provided the fiscal space for Mozambique to sustain scaled-up spending, albeit with a continued need to consolidate long-term fiscal sustainability. The HIPC Initiative and MDRI have reduced Mozambique’s debt and allowed the country to maintain a relatively high level of expenditures (about 25–30 percent of GDP), which are being financed through concessional external borrowing and foreign grants. The authorities’ strategy of absorbing foreign aid through sales of foreign exchange and avoiding recourse to nonconcessional external borrowing has also helped create an environment conducive to scaled-up expenditures. However, given the low tax to-GDP ratio and the need to guard against aid volatility and gradually reduce dependence on donors, an annual average revenue increase of 0.5 percent of GDP should continue to be targeted through the CFMP by widening the tax base and improving revenue administration. Beyond the PARPA II period, strong growth and increased fiscal revenues from mega-projects will help Mozambique maintain long-term fiscal sustainability and possibly allow it to start tapping international capital markets. This approach will provide an exit strategy from aid dependence in the long run and ensure that Mozambique can finance at least recurrent spending from its own resources.

The lessons from Mozambique’s experience of managing foreign aid inflows include the following:

• A full spending of scaled-up foreign aid, if carefully managed, could help a country make vast strides in human development and poverty reduction in a short time without resulting in significant macroeco-nomic absorption problems and microeconomic capacity constraints, although the latter may be reflected with a lag and call for a coordinated approach to capacity building, particularly training of frontline workers (for example, teachers, nurses, and agriculture extension workers).

• Prudent macroeconomic policies and well-sequenced structural reforms are key to maintaining macroeconomic stability and sustain-ing rapid broad-based growth.

• The willingness of a central bank to sell foreign exchange associated with aid inflows to mop up excess liquidity, and thus mostly absorb foreign aid, could avoid building up unsustainable domestic debt and crowding out the private sector without a significant loss of competitiveness.

• A prudent external borrowing strategy and encouragement of FDI, particularly in the natural resource and infrastructure sectors, can help consolidate long-term fiscal sustainability and gradually reduce dependence on donors.

• Sustained large foreign aid inflows (concessional external borrowing and grants) in the range of 10–20 percent of GDP need not result in a weaker revenue effort and fatigue with reforms, including more difficult second-generation institutional reforms such as public financial management, if the country takes a longer-term perspective and the international community continues to provide appropriate technical assistance.