Chapter

5. Projecting the Impact of Increased Aid on Economic Growth

Author(s):
Yongzheng Yang, Robert Powell, and Sanjeev Gupta
Published Date:
March 2006
Share
  • ShareShare
Show Summary Details

5.1. The Relationship between Aid and Growth

The debate about the effectiveness of aid in stimulating growth goes back many years, yet there remains considerable uncertainty about the aid-growth relationship. Some researchers suggest that there is either no effect or a negative one; others suggest a positive effect, but with diminishing returns. Still others argue that aid works to promote growth in some circumstances (when a country has good policies), but not in others.

Early skepticism about the role of aid in promoting economic growth focused on the potential disincentive effects of aid on investment and private sector development (Bauer, 1972). Some of the ensuing empirical research did in fact find little or no relation between aid and growth (Mosley, 1980; Singh, 1985). Similarly, Easterly (2001) questions the channels through which the “financing gap” model purportedly promotes growth, specifically whether aid flows lead to higher investment levels and hence to growth.13 There is evidence to indicate that, although aid can ease the liquidity and foreign exchange constraints to investment, it may actually worsen the incentive to invest, and hence be more likely to finance consumption.14

There remains great uncertainty about the relationship between aid and economic growth.

A related but distinct strand of the literature examines the impact of aid flows on the tradable goods sector. Rajan and Subramanian (2005a) find that aid flows do have adverse effects on growth, wages, and unemployment in labor-intensive and export sectors. However, their model does not allow for inferences regarding the overall growth rate.

The effectiveness of aid can be affected by the policy environment in the recipient country (Burnside and Dollar, 2000) and by the quality of governance (Radelet, 2004). Other studies have examined whether the impact of aid is conditional on other factors.15

5.1.1. The Impact of Different Types of Aid

Different types of aid are likely to have different relationships with growth, and not all aid is targeted at stimulating growth (Clemens, Radelet, and Bhavnani, 2004). A distinction can be drawn between three types of aid:

  • Aid provided in the context of disasters, emergencies, and humanitarian relief (for example, food aid), which may actually have a negative correlation with observed growth rates because it is often provided when a country is hit by a negative shock.
  • Aid that might affect growth but, if so, only indirectly and over a long time: Included in this group is aid to halt environmental degradation or to support democratic or judicial reform, and some aid to support health and education, which all may have an effect on labor productivity only over many years (Clemens, Radelet, and Bhavnani, 2004).
  • Aid that might reasonably be expected to affect growth within a fairly short time (for example, four years), including aid to build infrastructure such as roads, irrigation, ports, and electricity.

Aid inflows should be categorized according to whether 1) they can reasonably be expected to enhance growth in the short to medium term, 2) they are directed at longer-run growth, and 3) they support activities unlikely to be related to growth.

5.1.2. The Growth Effects of Scaling Up

To project the impact of scaling up on real growth rates, therefore, one might start with how the aid will be spent and the policy environment into which it will be disbursed. It is important to categorize the projected aid inflows and the associated higher budget expenditures into 1) those that can reasonably be expected to enhance growth in the short to medium term, 2) those directed at longer-run growth, and 3) those that support activities unlikely to be related to growth.

Cross-country econometric studies, which often use five- or ten-year averages of growth rates, may not provide sufficient guidance on the short-run impact of higher public spending on output in individual countries. Moreover, most empirical work on the impact of spending on growth provides estimates of the average impact of small changes in spending as a share of GDP, which may not directly translate into the impact of a substantial scaling up of aid, in which diminishing returns and other supply bottlenecks must figure more prominently.

5.2. Accounting for Diminishing Returns and Limits to Absorptive Capacity

The link between spending and growth reflects intermediate outputs that are subject to diminishing returns, supply constraints, or other bottlenecks.

Projecting the total response of output to scaled-up public expenditure requires accounting for diminishing returns to spending, the pace of convergence of output to its new steady state, and supply constraints and bottlenecks. An underlying notion is that the link between spending and growth reflects intermediate outputs (such as education or health capital or public infrastructure), the production of which can face constraints. Assessing these potential constraints in individual sectors is key to preparing a realistic assessment of the impact of scaling up.

Information on the extent of diminishing returns and the rate of convergence is subject to considerable uncertainty. Guidance is available on the initial response coefficient, but there are few estimated production functions for government spending or aggregate output in low-income countries that could inform the selection of the diminishing returns and convergence parameters.

5.2.1. Convergence Parameters

Existing estimates of convergence parameters are averages from cross-country studies and would have to be adjusted on a country-by-country basis to take account of local conditions and policy assumptions. Other things being equal, low-income countries have more scope to catch up with richer countries and should therefore be able to maintain a growth rate higher than the steady state for a long time. However, studies that control for other causes of growth (for example, strength of institutions) suggest that many low-income countries may be close to their steady state already. Therefore, convergence in response to a policy change may be more rapid, and the long-run impact of aid on growth may be correspondingly muted. Hence, measures that raise the steady-state level of income, such as strengthening governance and building up institutions, are important for ensuring that increased spending leads to sustained growth.

5.2.2. Diminishing Returns to Aid

The aid saturation point is the point at which the positive impact of aid falls to zero. This varies widely in different studies. The diminishing returns to aid are often captured in the empirical literature through the inclusion in an aid impact regression of a quadratic aid share variable along with the standard linear term. Clemens and Radelet (2003) summarize eight such studies relating the share of aid to per capita growth. All find a negative coefficient on the quadratic aid term, which, when combined with a positive coefficient on the level of aid, implies that the marginal return on aid is initially positive but then declines.16 The growth effects of most social sector expenditures, which have a long-term impact on growth, are more difficult to estimate. In addition, all existing saturation point estimates are derived from historical data, and saturation points in the future may be at significantly higher levels, particularly if aid absorption is accompanied by improvements in the policy environment and governance.

The aid saturation point—when the positive impact of aid falls to zero—is higher in countries with a good policy environment.

There is evidence that countries’ aid saturation points and absorption capacities are higher in good policy environments.17

5.2.3. Postconflict Situations

The sequencing and composition of aid require special attention in post-conflict countries. These countries face more severe constraints than the typical recipient of increased aid: basic institutions have to be rebuilt before attention can turn to achieving the MDGs. If rebuilding is successful, growth in these countries can rebound rapidly with restoration of law and order and the return of dislocated people. Aid is effective in these circumstances, but to be optimal, should change in composition over time (Chauvet and Collier, 2004). Furthermore, donors tend not to time aid properly—aid tapers off after three to five years, just when a country’s opportunity to build capacity and achieve sustained poverty reduction is best (Collier and Hoeffler, 2004). The amount of aid recommended for postconflict countries is quite similar to the level of a scaling-up program. This suggests that, after the first stage of rebuilding is complete, the guidelines in this paper are directly relevant for postconflict countries.

The basic challenges of a scaling up of aid are relevant to postconflict countries, after the initial rebuilding is complete.

5.3. Safeguarding Private Savings and Investment

Public investment can crowd in private investment in SSA countries. Crowding in likely reflects the complementarity of private investment with some components of public investment, especially infrastructure (Odedokun, 1997). Evidence on crowding in has been found across a variety of datasets and methodologies.18 Appendix Table A.2.6 summarizes some of the relevant studies.

Raising private sector savings and investment over the medium term will facilitate a country’s eventual graduation from relying on official sources of finance.

Strengthening the investment climate and the financial sector are important elements of any scaling-up scenario. In recent years, many low-income countries have prepared Diagnostic Trade Integration Studies (DTISs) as a means of integrating trade and investment issues into their development strategies.19 For example, the 2004 DTIS for Ethiopia recommended improvements in trade policies, the legal and regulatory environment, institutions, and trade-facilitation services, in order to encourage greater integration into the world economy and increased foreign direct investment.

More generally, measures should be included in every scaling-up scenario to strengthen financial sector institutions in order to raise private sector savings and investment over the medium term and to facilitate the country’s eventual graduation from relying on official sources of finance to relying on private sources. These measures should include instituting international standards for bank supervision, a competitive environment, and well-functioning financial markets, as well as taking steps to enhance the microfinance sector and to improve access to credit.

5.4. Raising Spending as a Share of GDP

Econometric evidence on the initial impact of public spending on growth is sensitive to the data and the methodology used. The link is clearest when public spending can be related to the stocks of the factors of production, such as physical capital, which are augmented by public investment. Many studies confirm a productive role for various types of infrastructure in low-income countries. Appendix Table A.2.7 summarizes studies that use either the production function or growth regressions and draw on data from Africa.20 Output responds positively to infrastructure, and there are also strong complementarities between different components of capital spending (Canning and Bennathan, 2000). The implication is that rates of return decline very quickly for an increase in any one component of capital. For low-income countries, electricity projects may yield the highest returns (Canning and Bennathan, 2000). There may also be large gains in growth from closing infrastructure gaps between average countries and regional leaders (Calderon and Serven, 2004).

Public spending has a greater impact on growth when it is directed toward the factors of production, such as physical capital, which are augmented by public investment.

Growth regressions find that the average impact on per capita growth of an increase of 1 percent of GDP in social sector or public investment spending is in the range of 0.5 to 1.0 percent over a five-year period. These studies are summarized in Appendix Tables A.2.7 and A.2.8.21

5.5. Confirming the Positive Impact of Aid

Case studies show that aid has a positive impact on growth, although it is difficult to precisely correlate the causes and effects given the variations among countries’ circumstances and their use of aid inflows. The multidonor Pro-Poor Growth study (Agence Française de Développement and others, 2005) finds that aid inflows fostered higher growth in the 1990s for the African case study countries. The underlying studies focus both on the impact of aid on growth and the ability of the poor to participate in growth. Aid impact on growth was strongest in Uganda, operating through reconstruction, improved economic management, social sector programs, and improvements in public administration. Aid also played an important role in relaxing constraints on growth due to debt burdens. Similarly, the Ghana case study finds that aid played the twin roles of supporting macroeconomic stabilization and boosting social sector programs that might otherwise have been cut due to scarce resources. The other case studies (Zambia, Burkina Faso, and Senegal) explore the impact of aid in less detail, but all note the importance of aid for financing health and education programs and thus supporting the human capital component of the growth process.

Aid can protect social sector programs and public investment in times of scarce resources or policy slippage.

The strength of public expenditure management (PEM) systems is also an important determinant of the growth impact of aid inflows. Studies of the fiscal impact of aid in Uganda, Zambia, and Malawi by the Overseas Development Institute (Fagernäs and Roberts, 2004) find the strongest discernible growth impact of aid in Uganda. For the other countries, the main benefit was the ability to protect some critical social sector and public investment programs at a time of overall fiscal stringency and policy slippage. One clear difference among the three countries was in the performance of the aid allocation mechanisms. Uganda comes much closer to a single integrated budget framework for all funding sources, allowing aid to be directed to its most productive use and accentuating the growth impact. In Malawi and Zambia, however, there persists a traditional assignment of aid inflows to specific development programs, even when these did not offer the highest growth impact.

Aid is most effective when it is delivered within a policy framework that supports efficient use of the additional resources.

Countries’ experiences with resource windfalls provide additional lessons for scaling-up countries, specifically by pointing to the need for effective expenditure management and for targeting of aid inflows. The analogy of aid inflows to natural resource inflows is imperfect, but both inflows generate similar policy challenges, including the risks of Dutch disease, misappropriation, and institutional deterioration resulting from diminished revenue mobilization incentives (Hausmann and Rigobon, 2003).22 Resource windfalls in developing countries can lead to a dissipation of resource wealth and even negative growth. This reinforces the importance of aid being delivered within a policy framework supportive of the efficient use of the additional resources.

13The “financing gap” or alternatively the “two-gap” model by Chenery and Strout (1966) identifies the gaps between savings and investment requirements and between foreign exchange earnings and import requirements as the major constraints to growth in developing countries.
14Empirical support for this idea can be found in Boone (1996).
15These include the effects of large export price shocks (Collier and Dehn, 2001), climatic shocks and trends in or the volatility of terms of trade (Guillaumont and Chauvet, 2001; Chauvet and Guillaumont, 2002), and the presence of totalitarian regimes (Islam, 2003).
16While considerable caution needs to be exercised in using these results because they are tentative, Hansen and Tarp (2001) show aid share saturation points between 14 and 27 percent of national income. Other studies find that saturation occurs at about 40 percent, well above the level of current aid inflows for most countries. For their short-impact aid component, Clemens, Radelet, and Bhavnani (2004) find the marginal impact of aid reaches zero at about 8–9 percent of GDP. But since short-impact aid is about half of total aid, the corresponding point for total aid occurs at about 16–18 percent of GDP, a range currently exceeded in only a small number of countries. Aid saturation points will be lower for postconflict countries, especially in the initial stages of reconstruction, but these aid flows will themselves help raise absorptive capacity over the medium term (as discussed in Section 5.2.3).
17Several of the studies surveyed in Clemens and Radelet (2003) allow the impact of aid on growth to depend on an indicator of the quality of the country’s institutions and policy stance. Collier and Dollar (2002) model a link between aid impact and the World Bank’s Country Policy and Institutional Assessment index, which ranges from 1 (lower quality) to 6 (higher quality). For a country with a score of 2, the saturation point is about 19 percent of GDP. With a score of 4.5, the saturation point occurs at 43 percent of GDP. This result is consistent with other evidence that the effectiveness of public expenditure depends on the quality of institutions (Baldacci and others, 2004). More recent evidence in Rajan and Subramanian (2005b) is more cautious about the existence of such a link.
18Greene and Villanueva (1991) note this effect for a small sample of developing countries in SSA, Asia, and Latin America, using a pooled approach. Hadjimichael and Ghura (1995) find it in a panel regression for a large SSA sample, while Ghura and Goodwin (2000), also using panel methods, find that the effect is present only for SSA, while crowding out was observed for Asia and Latin America. Belloc and Vertova (2004) investigate the effect at the country level and confirm crowding in for four of the five SSA countries they examine (the exception being Malawi).
19The Integrated Framework (IF) which sponsors the DTIS, was established in 1997 by six multilateral institutions (IMF, ITC, UNCTAD, UNDP, World Bank, and WTO) in order to facilitate the coordination of trade-related technical assistance to low-income countries.
20Of the individual developing country studies reported in Briceño-Garmendia, Estache, and Shafik (2004), all find that infrastructure has a positive effect on output, although none covers Africa specifically.
21The methodological debate associated with these studies has confirmed the need to control for reverse causality from output growth to public investment. Of particular relevance to SSA is the study by Gupta, Clements, Baldacci, and Mulas-Granado (2004) which covers 39 IMF program countries, of which 24 are in SSA. It finds a 0.7 percent effect on the growth rate of a 1 percent of GDP increase in capital outlays over a five-year period. In a larger sample of 120 countries, Baldacci and others (2004) simulate the increment to long-run growth from increases of 1 percent of GDP in education and health spending. For education, per capita GDP growth is, on average, 0.9 percent higher, and, for health outlays, it is 0.4 percent higher. These estimates are derived from the experiences of countries in which the elasticity of imports is less than one, and, hence, there is less than full leakage of higher aid flows to imports.
22The analogy of aid inflows and natural resource booms is imperfect because aid is normally provided by donors after agreement on appropriate policy conditions. Moreover, if it becomes clear that aid inflows are not being fully absorbed, they are likely to diminish.

    Other Resources Citing This Publication