IV Empirical Determinants of Growth, Savings, and Investment

Martin Mühleisen, Dhaneshwar Ghura, Roger Nord, Michael Hadjimichael, and E. Ucer
Published Date:
June 1995
  • ShareShare
Show Summary Details

This section reports the results of an empirical investigation of the factors that influenced growth, savings, and investment in sub-Saharan Africa during 1986–92. A multiple regression framework is used to separate out the effects of macroeconomic stability on growth, savings, and investment while controlling for the effects of other variables. The empirical methodology used is outlined in the Appendix. Actual data covering 39 countries during 1986–92 were used.52 The time span covered by this study is somewhat short, and, thus, the empirical findings should be viewed as preliminary and interpreted with caution. However, these results tend to indicate that the behavior of growth, private investment, and domestic savings in sub-Saharan Africa is consistent with certain theoretical regularities. The findings also confirm the results of several previous empirical studies of growth, savings, and investment. In particular, these aggregates reacted favorably to a stable macroeconomic environment during 1986–92. The evidence suggests that macroeconomic stability is conducive to higher rates of savings and investment and to faster rates of economic growth. Nevertheless, the direction of causality between macroeconomic policy variables and foreign aid on the one hand, and growth, savings, and investment on the other hand, are not clear-cut. A full investigation of these causal relationships is beyond the scope of this study. However, the broad correlations presented in this section are indicative of the important links between macroeconomic stability and growth, savings, and investment.

Determinants of Growth

Over the years, two opposing schools of thought have emerged on the causes of the poor growth performance in sub-Saharan Africa. Some observers have blamed external factors beyond the control of policymakers (such as a weak demand for the primary commodities exported by sub-Saharan African countries and unfavorable weather). Others have suggested that policy failures (reflected in high and variable inflation rates, high budget deficits, overvalued real exchange rates, and administered interest rates and agricultural producer prices) were at the heart of the region’s poor growth performance.53

In a recent paper, Fischer (1991) has suggested that macroeconomic policies can influence economic growth through their effect on the volume of investment or on the efficiency of investment. Using pooled time-series and cross-section data for a diverse group of developing countries, Fischer provides empirical support for both transmission channels. Various other studies, using data for industrial and developing countries, have provided support for the positive effects of macroeconomic stability on economic growth.54 Cross-section empirical studies of growth performance that have included a dummy variable for African countries have invariably found a significant negative sign on its estimated coefficient, indicating that over periods as long as two to three decades African economies have grown on average much slower than the rest of the world (e.g., Barro (1991) and Fischer (1991)). A recent paper by Easterly and Levine (1993) found that the dummy variable for African countries remained significant even after controlling for the effects of economic policies, human capital, factor endowments, and initial conditions.55 One possible explanation for the weak performance of Africa in comparison with other developing countries is the relatively slower contribution to growth from capital accumulation and the modest improvement in total factor productivity (Table 16).

Table 16.Contribution to Trend Output Growth, 1971–911(Annual percent changes)
All Developing

AfricaAsiaLatin America
Trend output growth5.
Capital contribution2.
Labor contribution1.
Total factor productivity1.
Source: IMF, World Economic Outlook, May 1993.

In this subsection, the channels through which macroeconomic policies affect growth in sub-Saharan Africa are investigated. The explanatory variables included are the ratio of private investment to GDP (PIY), the ratio of government investment to GDP (GIY), population growth (PG), secondary school enrollment ratio in 1970 (SEC70), the ratio of the budget deficit to GDP (DEFY), inflation (INF) and the square of INF (INFSQ), the terms of trade (TTG), a dummy for countries of the CFA zone (CFADUM), a dummy for the sustained adjusters (SUSDUM), a dummy for countries with protracted macroeconomic imbalances (IMBDUM), a dummy for droughts (DRY), and an index of political rights and civil liberties (FREE). (See Table 17 for the definitions of the variables.) Table 18 reports the regression results; regressions (l)-(4) use pooled time-series and cross-section data, and regression (5) uses pooled cross-section and period average data.56 The rest of this subsection discusses the results.

Table 17.Sub-Saharan Africa: Definitions of the Variables Used in the Regressions1
YGGrowth in real GDP.
YGPCGrowth in per capita real GDP.
PIYPrivate investment as a ratio to GDP.
GIYGovernment investment as a ratio to GDP.
NSYNational savings as a ratio to GDP, where national savings are defined as investment plus the current account balance.
DSYDomestic savings as a ratio to GDP, where domestic savings are defined as national savings minus grants.
RYPC$Per capita real GDP, expressed in U.S. dollars.
PGPopulation growth.
SEC70Secondary school enrollment ratio in 1970. Data for the 1970 enrollment ratios are missing for Namibia, Sao Tome and Principe, and Seychelles. Also, since data for school enrollment are not available for 1970 for Malawi, the 1976 figure is used instead; for Cape Verde, the 1975 figure is used. (Source: World Bank, World Debt Tables.)
DEFYOverall budget deficit (including grants) as a ratio to GDP.
INFAnnual percentage change in the consumer price index (annual rate of inflation).
INFSDStandard deviation of INF. Note that this variable is time-invariant.
RERGPercentage change in the real effective exchange rate.
RERGSDStandard deviation of RERG. Note that this variable is time-invariant.
BMYBroad money as a ratio to GDP.
TTGPercentage change in the terms of trade.
DETXExternal debt as a ratio to total exports.
ODAYOfficial development assistance as a ratio to GDP.
DRYA dummy variable as a proxy for the extent of inadequate rainfall. It takes a value of 1 if the percentage change in the per capita food production is less than zero, and it assumes a value of zero otherwise. The index of per capita food production was obtained for 1986–91 from the World Bank, World Debt Tables.
DEPENDDependency ratio, defined as the ratio to total population of those below 15 and over 64. The proxy used to measure DEPEND is the ratio to total population of the difference between total population and the labor force (LABOR). Data for LABOR were unavailable for Sao Tome and Principe and for Seychelles. The labor force data were obtained from the World Bank, World Debt Tables.
FREEIndex of political freedom and civil liberties for 1986–91, obtained from McColm and others (1991), which uses the following methodology for calculating this index: countries are ranked on a seven-point scale for levels of political rights and civil liberties, with ratings summarized as free, partly free, and not free. For the purpose of the current study these three categories have been as-signed the (arbitrary) numeric codes 2, 1, and 0, respectively. Data for this variable are missing for 1992. Political rights are defined as the rights to participate meaningfully in the political process, such as the right of all adults to vote and compete for public office, and for elected representatives to have a decisive vote on public policies. Civil liberties are defined as rights to free expression, to organize or demonstrate, and to a degree of autonomy such as is provided by freedom of religion, education, travel, and other personal rights. For a more complete description of the procedures used in the construction of the index, refer to the above publication and also to Gastil (1987). A similar political index has been used in empirical work (e.g., Kormendi and Meguire (1985)).
SUSDUMDummy variable for countries judged to be sustained adjusters during 1986–93: Benin, Burundi, the Gambia, Ghana, Kenya, Lesotho, Malawi, Mali, Mozambique, Niger, Senegal, Tanzania, Togo, and Uganda.
IMBDUMDummy variable for countries with protracted imbalances during 1986–93: Burkina Faso, Cameroon, Cape Verde, Central African Republic, Chad, Comoros, Congo, Côte d’lvoire, Equatorial Guinea, Ethiopia, Gabon, Guinea, Guinea-Bissau, Madagascar, Nigeria, Rwanda, Sao Tome and Principe, Sierra Leone, and Zambia.
CFADUMDummy variable for CFA franc countries: Benin, Burkina Faso, Cameroon, Central African Republic, Chad, Côte d’lvoire, Comoros, Congo, Equatorial Guinea, Gabon, Mali, Niger, Senegal, and Togo.
Table 18.Estimates of the Growth Equation1

(Dependent variable: growth rate of per capita real GDP) (Estimation method: generalized least squares)2

Regression Number
Explanatory Variable(1)3(2)3(3)3(4)3(5)4

The results indicate that the macroeconomic environment matters for growth. Thus, countries that had lower rates of inflation and budget deficit ratios, and faster convergence of the actual real effective exchange rates to their respective equilibrium levels, experienced faster rates of growth than the others. The negative effect of inflation lends support to the predictions of cash-in-advance models, thus confirming the results of many previous studies.57 The percentage change in the real effective exchange rate can be regarded as a proxy for the effects of movements of that rate toward its equilibrium rate58 and out ward orientation.59 Also, the budget deficit ratio, the rate of inflation, and the percentage change in the real effective exchange rate all have significant effects on growth, even when they are all included in the same regression, implying that these policy variables are measuring different empirical effects. It is quite possible that, in addition to capturing the effects of the size and the level of government inefficiency, the budget deficit ratio is also capturing the crowding-out effect associated with large budget deficits.

Another important result concerns the effects of the rates of private and government investment on economic growth. The results of regression (1) indicate that both the private and government investment ratios are significantly correlated with growth. The significance of government investment is robust, regardless of specification, confirming a similar result found by Easterly and Rebelo (1993). Also, the coefficient of the ratio of government investment to GDP is positive and significant when policy variables are included in the same regressions, implying that this variable has effects on growth other than through the efficiency channel. Thus, the ratio of government investment has a volume effect on growth; that is, higher rates of government investment lead to higher rates of growth. However, the magnitude and significance level of the estimated coefficient for private investment fall when policy variables are included together in the same regression. This result needs to be interpreted with caution; it does not imply that private investment is not an important variable in determining growth. Rather, once account is taken of the effects of the macroeconomic policy environment, private investment does not have an independent influence on growth. Thus, it appears that policy variables had a greater effect on the efficiency of private investment than on the volume of private investment during 1986–92.60,61

There are also indications that the marginal productivity of government investment is significantly higher than that of private investment; the coefficient of government investment in the regressions is five to six times higher than that of private investment. However, this result may be due to the fact that measures of private investment include investment by public enterprises.62 Contrary to the results of this study, Khan and Reinhart (1990) and Khan and Kumar (1993) found private investment to be more efficient than public investment for a diverse group of developing countries during 1970–79 and 1970–90, respectively. In these two studies, public enterprise investment was excluded from the private investment series. However, they did not control for the effects of macroeconomic stability on growth. In addition, using data from a diverse group of countries, Easterly and Rebelo (1993) found that the ratio of public investment (excluding investment by public enterprises) to GDP had a positive, strong, and robust effect on growth. Their estimated coefficient is, however, much larger than the ones found in this paper by a factor of about three, implying that public investment in sub-Saharan Africa has been about three times less efficient than for the world in general.

The external environment also appears to have influenced growth in sub-Saharan Africa during 1986–92. The estimated effect of changes in the terms of trade, which is felt with a one-year lag, is positive and significant at the 5 percent level.6364 This finding lends support to the notion that terms of trade losses have been, in part, responsible for the poor growth performance of sub-Saharan African countries during 1986–92.65 However, the adverse effects of losses in the terms of trade on per capita growth seem to have been offset by depreciations in the real effective exchange rate.

Structural and institutional reforms also appear to affect growth performance, after controlling for the effects of macroeconomic stability, the rate of investment, and the level of human capital. The dummy variable representing the sustained adjusters is statistically significant (albeit at the 10 percent level), indicating that the countries in this subgroup had a significantly higher average growth rate than other African countries (about 0.7 percentage point a year). This finding supports the view that broadly based structural reforms alleviate the impediments to private sector development and stimulate economic growth. In contrast, countries with protracted macroeconomic imbalances had significantly lower growth rates. Also, the dummy variable for the CFA franc countries is consistently negative and statistically significant. For this group of countries, the average per capita growth rate was about 2 percentage points a year lower than the average for sub-Saharan Africa during 1986–92, although they had significantly lower levels of inflation than other countries in the region. As a group, the CFA franc countries had average government investment rates, secondary school enrollment ratios, inflation rates, and degree of political freedom during 1986–92 that were significantly lower than the average for sub-Saharan Africa. Also, the real effective exchange rate of this group of countries was more appreciated than for the average of the region. Since the regressions control for all of these factors, the dummy for the CFA franc countries could be picking up the effects of the private sector’s low confidence in government policies, and/or the impact of structural rigidities.

The measure of human capital is positively and significantly correlated with growth.66,67 As expected, population growth lowers per capita growth, notwithstanding the positive contribution to growth exerted by the associated faster growth of the labor force. One way to attenuate this effect in the period ahead would be to raise investment in human capital, as implied by the significant negative correlation between the level of human capital development and population growth.68 Regression (3) expands the number of explanatory variables to include two noneconomic variables: FREE69 and DRY.70 The estimation results suggest that progress toward political freedom enhances economic growth. This variable could be capturing the effects of the existence of market-friendly legal and institutional frameworks that facilitate growth-enhancing activities.71 The school enrollment ratio loses its statistical significance in this regression, although its effect remains positive; this could be because it is highly (positively) correlated with the measure of political freedom.72 Adverse weather was also important in explaining the region’s growth performance. Inadequate rainfall lowered by 1.1 percentage points a year the average growth rate in sub-Saharan Africa during the period under review.73

The growth equation was also estimated on the basis of period average data for 1986–89 and 1990–92 for all the country-specific variables, with a view to eliminating the year-to-year variability in the data and capturing the underlying trends. Regression (4), which gives the estimation results, confirms the robustness of the results of the regressions based on pooled time-series and cross-section data. In addition, the absolute value of the coefficients of most explanatory variables and their statistical significance are higher, while the coefficient of determination is substantially higher.74 In fact, the coefficient of private investment becomes significant in this regression, which includes the variable indicating the degree of political freedom.75 However, the coefficients on the dummies SUSDUM and IMBDUM are not significant in this regression.

Table 19 reports the estimated beta coefficients of the explanatory variables and the contribution of the latter to per capita growth, based on regression (4) of Table 18. The beta coefficients are unit-free and measure the relative impact of explanatory variables on per capita growth. Of the three macroeconomic policy variables—the inflation rate, the deficit ratio, and changes in the real effective exchange rate—the third had the largest impact on growth. The combined relative effect of the policy variables was by far the largest of all explanatory variables. It is noteworthy, in particular, that the beta coefficient of changes in the real effective exchange rate was larger in absolute terms than the beta coefficient of changes in the terms of trade.

Table 19.Beta Coefficients and Contribution of Explanatory Variables to Per Capita GDP Growth
Explanatory Variable1Estimated Coefficient2Estimated Beta Coefficient3Estimated Mean of the VariableContribution to Per Capita Growth4

Since the various explanatory variables behaved rather differently during 1986–92, it is also interesting to evaluate the contribution of each of these variables to per capita growth at the sample mean. The results of this exercise are shown in the last column of Table 19. It is clear that rapid population growth and adverse rainfall were responsible to a large extent for the region’s poor per capita growth performance. On the policy front, the real effective exchange rate contributed positively to per capita growth. However, this positive contribution was overshadowed by the adverse effects of the budget deficit ratio and inflation, which together contributed to a substantial lowering of per capita growth. The losses in the terms of trade also contributed to the poor growth performance in sub-Saharan Africa, but this adverse effect was more than offset by the depreciation of the real effective exchange rate. The terms of trade effect was also exceeded markedly by the combined negative contribution of the budget deficit ratio and inflation. Thus, after population growth and unfavorable weather, inappropriate macroeconomic policies were the second most important contributing factor to the poor per capita growth performance in sub-Saharan African countries during 1986–93. It should also be noted that increases in the private investment ratio and political liberalization had strong positive effects on growth.

The results presented in this subsection reinforce the importance of growth-oriented adjustment policies designed to help countries attain and maintain stable macroeconomic conditions (including the maintenance of an appropriate exchange rate) and the acceleration of structural reforms, so as to stimulate the expansion of government and private investment. While lowering the budget deficit would help to enhance economic growth, doing so by cutting government investment could actually hurt the growth process. The empirical results also suggest that, other things being equal, an increase in government investment, financed by higher tax receipts without an increase in the budget deficit ratio, would promote growth. Many sub-Saharan African countries are characterized by narrow tax bases, weak tax administration, and a proliferation of tax exemptions, and thus have significant potential for raising tax receipts by broadening the tax base, improving tax administration, and rationalizing the tax system. Such reforms would increase government revenue and investment without necessarily raising tax rates, which would tend to undermine private investment. Increasing government expenditure on education and health would help raise human capital and contribute to growth, both directly and indirectly, by slowing population growth over the long term. In addition, a stable macroeconomic environment, by raising the efficiency of private investment, can speed up the process of achieving sustainable growth.

Determinants of Savings and Private Investment

Section II provided an extensive discussion of the behavior of aggregate and sectoral savings and investment balances in sub-Saharan Africa and in certain subgroups during 1986–93. It noted that, notwithstanding the generally poor performance observed in the aggregate trends of savings and investment during this period, these aggregates exhibited a considerable degree of diversity in the countries considered. An understanding of the sources of these differences would help to identify ways of enhancing the performance of savings and investment, and thus economic growth in the region. In this context, a principal objective of this subsection is to investigate empirically the role of macroeconomic policies in explaining the diversity in experiences regarding the behavior of savings and private investment. The recent surge in interest on the policy determinants of economic growth in developing countries has also increased interest in analyzing the effects of the same factors on aggregate investment.76 The main message that emerges from the existing empirical literature is that, broadly speaking, the same macroeconomic policy indicators that are correlated with growth over periods of as long as two to three decades are also correlated with the rate of investment.77

A number of macroeconomic variables have been included in savings and investment studies to account for the effects of monetary, fiscal, and exchange rate policies. Table 20 lists some of the variables that have been used in recent empirical studies and summarizes their expected effects on private investment. The current study includes the following variables in the empirical investment equation: the rate of inflation (INF); the standard deviation of inflation (INFSD); the overall budget deficit (including grants) as a ratio to GDP (DEFY); government investment as a ratio to GDP (GIY); the standard deviation of the percentage changes in the real effective exchange rate (RERGSD); the stock of foreign public debt as a ratio to exports (DETX); the square of DETX (DETXSQ); and broad money as a ratio to GDP (BMY). Also, the following variables are included in the empirical savings equation: INF, INFSD, DEFY, GIY, DETX, DETXSQ, BMY, the percentage change in the terms of trade (TTG), and the dependency ratio (DEPEND). The regression results are reported in Table 21 and are discussed in the rest of this subsection.

Table 20.Empirical Determinants of Private Domestic Investment for Developing Countries
Suggested Variable in the RegressionMechanism/Rationale/EffectSource1
Monetary policyReal interest rate
  • • Under restrictive credit policy, higher interest rates would imply higher user cost of capital due to an increase in the real cost of bank credit or an increase in the opportunity cost of retained earnings.

Greene and Villanueva (1991)
Volume of direct credit to the private sector
  • • Under financial repression, interest rate may be a poor proxy for the direction of monetary policy as well as the user cost of capital. Direct credit may be more appropriate.

Bier (1992)
Fiscal policyBudget deficit
  • • Crowding out of private sector investment through higher interest rates or lower credit availability.

Fischer (1991 and 1993);
Government spending on infrastructure investment
  • The mix of tax increases and spending reductions may adversely affect private sector; also through the complementary nature of public and private investment

Blejer and Khan (1984); Sakr (1993)
Exchange rate policyReal depreciation/appreciation
  • • A high dependence on imported capital and intermediate goods, along with a low share of traded goods in total investment, would result in a contraction of investment in the face of a real depreciation while a real appreciation would create an unsustainable investment boom.

  • • A depreciation would cause an increase in real debt burden and a decline in the firms’ (in the home goods sector) net worth.

  • • Higher variability would imply higher uncertainty.

Serven and Solimano (1992)
  • • Creates rent-seeking behavior and induces households and firms to divert resources from productive activities to other activities that allow them to reduce the burden of the inflation tax.

  • • Higher viability would imply higher uncertainty.

Fischer (1991 and 1993)
GrowthGrowth rate of GDP
  • • Proxy for aggregate demand developments.

Serven and Solimano (1993)
Foreign capital inflowsLevel of foreign direct investment
  • • Eases domestic financing constraint Causes crowding-in effects by creating linkages and externalities.

Sakr (1993)
External debtDebt service-GDP ratio
  • • Debt-overhang effect: investors’ perceptions of future returns will be affected as concerns arise about the country’s ability to meet future contractual debt servicing without increasing taxation.

Greene and Villanueva (1991)
Table 21.Estimates of the Investment and Savings Equations1(Estimation method: generalized least squares)2
Dependent Variable=Private Investment Rate (PIY)Dependent Variable=Domestic Savings Rate (DSY)
Explanatory Variable(1)(2)(3)(4)(5)(6)(7)(8)(9)
Lag(DETXSQ)5.9 × E-7*5.2 × E-76.0 × E-7**5.1 × E-7*7.3 × E-77.0 × E-7**6.2 × E-7***7.1 × E-7**4.0 × E-7

The results suggest that macroeconomic stability plays an important role in stimulating savings and private investment. Thus, the rates of savings and private investment are enhanced in an environment where the rate of inflation and budget deficit ratio are low.78,79 In addition, macroeconomic uncertainty, as measured by the standard deviation of inflation or the standard deviation of changes in the real effective exchange rate, has adverse effects on savings and investment.80,81 This result is an indication that uncertainties about the returns or the direction of policies have deleterious effects on savings and investment. Thus, progress toward macroeconomic stability plays a crucial role in affecting domestic savings and private investment for sub-Saharan African countries. This confirms similar results for total investment by Fischer (1991) and Kormendi and Meguire (1985), and for private investment by Greene and Villanueva (1991) and Ozler and Rodrik (1992). It should be noted that once account is taken of the effects of macroeconomic policies, output growth has no independent effect on the rate of private investment, confirming a similar result by Ozler and Rodrik (1992).

Another interesting result relates to the positive and significant effect on savings and investment of the ratio of broad money to GDP, confirming the potential for payoffs relating to the ongoing financial deepening in African economies. The real interest rate was used as an additional variable to capture the effects of financial liberalization and deepening. This variable, however, was highly correlated with the rate of inflation (with a correlation coefficient of 0.96), indicating that during 1986–92 nominal rates in sub-Saharan African countries adjusted rather slowly to changes in economic fundamentals and that, on average, changes in inflation were dominating the movements in real interest rates.82

The estimation results support the complementary effects of public investment on private investment, confirming similar results by Blejer and Khan (1984), Greene and Villanueva (1991), and Tun Wai and Wong (1982) for a diverse group of developing countries, and Sakr (1993) for Pakistan.83 Thus, government investment in sub-Saharan African countries, by providing positive externalities, fosters both private capital accumulation and economic growth. In addition, both the budget deficit ratio and the public investment rate are significant when included in the same regression for private investment. One implication is that, although lowering the deficit would be beneficial to private investment, achieving this objective by cutting government investment expenditure would be counterproductive, as evidenced in particular by the experience of the CFA franc countries (see Section II). Another implication of the joint significance of the budget deficit and public investment ratios is that the former is most likely capturing a crowding out effect.

As regards the effects of external debt, the results are consistent with the debt-overhang hypothesis. Increases in the ratio of external public debt to exports dampen both private investment and domestic savings, confirming similar results for private investment by Borensztein (1990a and 1990b), Greene and Villanueva (1991), and Oshikoya (1994). Furthermore, this variable has a nonlinear effect on private investment and domestic savings; the results indicate that even low ratios of external debt have counterproductive effects on domestic savings and private investment.84

Countries that implemented structural reforms also made progress in raising savings and private investment. The sustained adjusters as a group had savings rates significantly higher than the average. In contrast, the countries with protracted macroeconomic imbalances as a group had average rates of savings significantly below the average for the region. The CFA franc countries had significantly lower rates of private investment and domestic savings than the average. As account has already been taken of differences in the policy stance, the dummy variables are most likely capturing the effects of the differentiated progress registered by these country groups in alleviating structural and institutional impediments to private sector development.

Changes in the terms of trade did not have significant effects on savings during 1986–92.85 Fry (1985 and 1986) found the terms of trade to have significant positive effects on national savings rates in Asia. Also, Fry (1989) found similar results for a group of highly indebted developing countries. However, his results on the effects of the terms of trade and the results of this study are not directly comparable. His objective was to capture the income effects of changes in the terms of trade, which he measured as the rate of change of the terms of trade divided by the lagged ratio of imports of goods and services to GNP measured at current prices. The objective of the current study is to measure the pure effects of changes in the relative international prices of the exports and imports of the economies concerned. The income effects of the changes in the terms of trade for the current group of countries would be expected to be captured in part by the growth rate (YG) and per capita real GDP (RYPC$).

The domestic savings rate is positively correlated with the per capita level and growth of real GDP. Broadly speaking, the same macroeconomic policy variables that affect the growth rate of real GDP also affect domestic savings, suggesting that these policies have both direct and indirect effects on the savings behavior. Also, the estimation results indicate that domestic savings rates are higher in countries at higher levels of development (measured by their real GDP per capita, expressed in U.S. dollars), confirming similar results by Fry (1978 and 1980).86 The ability of an economy to mobilize domestic savings to finance investment activities depends in part on its level of development; the lower the level of a country’s development, the fewer domestic resources it has to finance private investment activities.

The regression results of this study indicate that the dependency ratio has a significant negative effect on the rate of savings. A number of other studies have also investigated the effects of demographic factors on the rate of savings.87 The evidence to date is not definitive and remains somewhat controversial. The results appear to be highly dependent on the sample and time period considered, the specification used, and the variables included or excluded.

Overall, the empirical findings of this subsection indicate that maintenance of a stable macroeconomic environment is critical to efforts toward mobilizing domestic savings and encouraging private investment, and thus toward laying the foundations for sustained economic growth.

Effects of Foreign Aid on Savings, Investment, and Growth

Foreign aid to sub-Saharan Africa has been substantial during the past decade (Table 22). Foreign assistance flows, based on data compiled by the Organization for Economic Cooperation and Development (OECD), have increased sharply in recent years, reaching 11.5 percent of GDP in 1992 (14 percent, excluding Nigeria) from about 6 percent of GDP in the mid-1980s (7 percent, excluding Nigeria).88 The results of the previous subsections indicate that, besides a stable macroeconomic environment, government investment (including investment in human capital) can play an important catalytic role in increasing domestic savings, private investment, and economic growth. The narrow domestic revenue base in many sub-Saharan African countries has meant that government investment remains largely foreign-financed, and foreign assistance is likely to continue to play an important role in the economies of the region.

Table 22.Sub-Saharan Africa: Official Development Assistance(In percent of GDP)
Sub-Saharan Africa
Excluding South Africa6.
Excluding South Africa and Zaïre5.
CFA franc countries7.
Non-CFA franc countries, excluding Zaïre5.
Positive per capita growth countries5.
Negative per capita growth countries, excluding Zaïre7.
Sustained adjusters10.913.916.018.019.318.419.616.6
Low macro imbalances countries4.
Protracted imbalances countries, excluding Zaïre3.
Source: OECD, Geographical Distribution of Financial Flows to Developing Countries, 1987/92.

This subsection examines the contribution of foreign aid to savings, investment, and growth in sub-Saharan Africa.89Killick (1991) notes that during 1980–88, sub-Saharan Africa received on average $22 a person a year in foreign assistance, compared with $5 a person for other developing countries; during the same period, foreign aid to sub-Saharan Africa amounted to 33.5 percent of gross investment, compared with 3.3 percent for other developing countries. Previous studies have found that, in comparison with other regions, the effectiveness of foreign aid has been low in sub-Saharan Africa. The studies by Cassen (1986), Gupta and Islam (1983), and Mosley (1987) have found the relationship between aid and economic performance (including growth) in sub-Saharan Africa to be much weaker than in other developing regions.

The economic impact of foreign assistance and its effectiveness in stimulating investment and economic growth have been a recurring subject of debate. Bauer (1981) argues passionately against foreign aid and attributes much of the failure of development programs to the flood of foreign assistance. Most other studies are less definitive, however. Early investigations, using data mainly from the 1960s, have tended to show a negative relationship between foreign aid and savings. Using data for 44 countries, and a time period spanning 1953–66, Weisskopf’s (1972) econometric results support the hypothesis that the impact of foreign capital inflow on ex ante domestic savings is significantly negative. Ex post, however, his results hold only for those cases where savings, rather than the balance of payments, are the binding constraint on investment. Gupta and Islam (1983) use a simultaneous-equation model to account for the fact that, because of the interdependence between savings and growth, foreign capital affects savings and growth directly as well as indirectly. Their overall conclusion is that the impact is indeterminate, with the results showing a weak positive impact of foreign aid on growth, while domestic savings appeared to be somewhat negatively affected. In addition, Fry (1978 and 1980) and Giovannini (1985) found the effect of foreign assistance on domestic savings to be negative and statistically significant. Schmidt-Hebbel, Webb, and Corsetti (1992) found a statistically significant negative effect of foreign savings on household savings in the context of developing economies.

A comprehensive review of aid effectiveness by Cassen (1986) concludes that the statistical relationship between foreign aid and economic growth is weak but attributes the result largely to shortcomings in the methodology applied in most research. Criticizing the cross-section studies that show little link between foreign aid inflows and growth, Cassen argues that on an individual country basis, particularly in South Asia, the relationship is much more significant. Mosley (1987) also concedes that there is no statistically significant correlation between foreign aid on the one hand and economic growth and savings on the other. Citing what he calls the “micro-macro paradox,” he contrasts this with the well-documented success of many development projects in the field. He concludes that the fungibility of foreign aid has contributed to the shifting of some domestic expenditures to less productive purposes and calls for channeling aid to countries that have a proven track record of using it effectively and for increased policy conditionality in the use of aid funds. Nord and others (1993) focus on a subset of more successful adjustment programs in sub-Saharan Africa and conclude that large increases in foreign aid did not appear to have led to a decline in domestic savings rates. The World Bank (1994), which also adopted an approach of dividing sub-Saharan Africa into groups characterized by their adjustment performance during 1987–91, found that although part of foreign savings financed consumption, domestic savings rates in the better-performing countries improved during the period.

In the face of the recorded large increases of foreign aid to sub-Saharan Africa, two principal questions arise. First, did the increased foreign assistance stimulate or crowd out domestic savings? Second, did investment increase proportionately, and did foreign aid affect private investment differently from public investment? This leads to an additional question: Did foreign aid stimulate economic growth?

The regression estimates, based on pooled time-series and cross-section data for 39 countries in sub-Saharan Africa during 1986–92 and the same country groupings as before, yield some interesting results. A simple ordinary least-squares estimation is shown in Table 23.90 For sub-Saharan Africa as a whole, the impact of foreign assistance on domestic savings is negative. However, the negative impact is concentrated in those countries with protracted imbalances and negative per capita growth, whereas in the group of sustained adjusters, foreign aid appears to have stimulated domestic savings. Government investment is, not surprisingly, strongly related to foreign aid, which is explained by the high proportion of government capital expenditure financed by aid. The results for private investment are mixed, possibly because of data weakness. They do show, however, that the impact of aid is likely to be positive for the group of sustained adjusters. In countries with negative per capita growth performance over the period, in contrast, foreign assistance appears to have discouraged private investment.

Table 23.Partial Effects of ODA on Savings, Investment, and Growth1(Estimation method: ordinary least squares)
Estimated Coefficients for
Dependent VariableExplanatory VariableSub-Saharan Africa [N=234]2Countries with low macro imbalances [N=36]2Sustained adjusters [N=84]2Countries with protracted macro imbalances [N=114]2Countries with positive per capita GDP growth [N=132]2Countries with negative per capita GDP growth [N=102]2
Private investment/GDPLag(ODAY)-0.0500.7840.530*-0.1250.239-0.581***
Government investment/Lag(ODAY)0.290***0.7850.345***0.295***-0.0200.296***
Domestic savings/GDPLag(ODAY)-0.296***5.163***0.620***-0.593***-0.314***-0.846***
Per capita GDP growthLag(ODAY)0.0420.8060.0580.149***-0.0750.122*

Using a multivariate framework similar to that applied in the previous subsections yields very similar results (Tables 24 and 25). Domestic savings and private investment are both adversely affected by foreign assistance flows. Part of the negative impact of foreign aid on private investment could be due to the effect of the debt overhang on private investment behavior, since, in sharp contrast with the results presented in the previous subsection, the external debt ratio is no longer significant in this framework.91 The effects of the other variables on private investment remain robust. In particular, the budget deficit ratio continues to have a significant negative effect, which indicates that foreign aid itself is not crowding out investment. Also, in contrast with the findings of the previous subsection, the dummy variables for the group of sustained adjusters and the countries with protracted imbalances are significant and have the expected signs.

Table 24.Estimates of the Effects of ODA on Investment and Savings1(Estimation method: generalized least squares)2
Dependent Variable
Private investment rateDomestic savings rate
Explanatory Variable(1)(2)(3)(4)
DETX32.1 × E-58.7 × E-50.00040.002
DETXSQ31.2 × E-71.6 × E-77.5 × E-81.8 × E-7
Table 25.Estimates of the Effects of ODA on Growth1(Dependent variable: growth rate of per capita real GDP)(Estimation method: generalized least squares)2
Results of Estimation with
Explanatory VariableTime-series



period average


The simple correlation between foreign aid and growth in sub-Saharan Africa is zero for this group of countries, confirming previous findings at high levels of aggregation. Also, at the level of the country subgroups, results are mixed, and the only statistically significant results suggest that for the poor performers—that is, the countries with protracted imbalances and/or negative per capita growth over the period—foreign aid enhanced economic growth (Table 23).92 The nonlinearity of the relationship between foreign aid and growth is shown in Table 25.93 In a multivariate framework, which introduces several additional possible influences on growth, the results indicate that foreign aid stimulates growth initially; beyond a certain threshold, however, the impact on growth appears to be negative.94 This would mean that too much foreign assistance can hurt an economy, possibly because of the limited capacity of many sub-Saharan African countries to absorb foreign resources. Most of the foreign resources must pass through the public sector in one way or another, and it is likely that arising distortions are magnified beyond a certain level of inflows.95 This threshold should not be seen, how-ever, as an optimal level of foreign aid. Even at lower levels of aid, concerns about the effectiveness of its use remain pertinent, particularly in countries where a large proportion of foreign assistance is channeled through the public sector. If, as suggested before, the impact of investment on economic growth lies mainly in the efficiency of its use, rather than in its volume, then this is likely to hold for foreign aid as well.

Foreign assistance has been extended to sub-Saharan Africa largely in response to the low levels of per capita incomes and the deep-rooted constraints to economic development faced by these countries, rather than linked to actual economic performance. Bilateral foreign aid, in particular, is based mainly on humanitarian considerations. Cash nonproject assistance (balance of payments support or adjustment lending), although rising in recent years, has remained modest in relation to total assistance, partly because performance-linked assistance primarily takes the form of debt relief or debt rescheduling and is not included in the data on ODA.

The experience of the past decade suggests that sound economic policies have had a profound impact on the effectiveness of foreign aid in promoting economic development. The narrow domestic resource base of many sub-Saharan African countries, and their urgent need for investment in basic infrastructure and social services, means that foreign assistance will continue to play an important role in the region in the near future. In the face of limited donor resources, the need for such assistance is increasing rapidly, as more developing countries in Africa and elsewhere (including the former Soviet Union) are embarking on intensified reform efforts. In such an environment, foreign assistance will tend to be linked, at least implicitly, to economic performance. This underscores the need for sub-Saharan African countries to pursue stable macroeconomic policies and implement broadly based structural adjustment programs so as to maximize the beneficial effects of foreign assistance and attract adequate aid inflows. Such policies would also strengthen domestic savings over time and help these countries reduce their reliance on foreign aid.

The challenge for sub-Saharan Africa is to make judicious use of the foreign assistance that is available, concentrating on providing public goods, such as infrastructure, health, and education services, while at the same time creating an environment that encourages private sector initiative. Over time, investment will need to depend more on private sector resources, both domestic and foreign.

South Africa and Zaïre were excluded from the analysis.

A recent World Bank study on Africa (World Bank (1994)) has noted that the generally weak economic performance of the countries in the region can be attributed more to inappropriate domestic policies than to external shocks.

See, for example, the papers by Kormendi and Meguire (1985) and Grier and Tullock (1989). The paper by Levine and Renelt (1992) lists other relevant studies; the same paper, using cross-section data for a large group of industrial and developing countries and the Learner extreme bounds analysis, has shown that the empirical links between macroeconomic variables and economic growth are quite fragile. It remains to be established, however, if the same conclusion would hold if the sensitivity analyses were done instead with pooled time-series and cross-section data for groups of countries in the different regions of the world and for different time periods.

There is only limited recent empirical evidence for the effects of macroeconomic policies on growth in sub-Saharan Africa. The papers by Easterly and Levine (1993), Ghura (1992), and Ghura and Grennes (1993) are exceptions. The paper by Wheeler (1984) made an early attempt to investigate the relative importance of policies and external forces in explaining growth in the region. Recent empirical studies on the determinants of growth in sub-Saharan Africa have investigated the role of exports (Fosu (1990)), capital instability (Fosu (1991)), political instability (Fosu (1992)), and export instability (Gyimah-Brempong (1991)).

The subperiods over which the averages of the series are taken are 1986–89 and 1990–92.

De Gregorio (1993), Fischer (1991), Ghura (1992), Grier and Tullock (1989), and Kormendi and Meguire (1985) found significant negative effects of inflation on growth. Barro (1991) and Fischer (1991), for a group of developing countries, and Easterly and Levine (1993), for sub-Saharan Africa, found significant negative effects of the budget deficit ratio on growth.

There is only a limited amount of empirical evidence on the effects of devaluations on output and employment. The results of the existing studies are inconclusive. Some studies report expansionary effects of devaluations on output. These include the studies by Gylfason and Schmid (1983), Conolly (1983), Nugent and Glezakos (1982), and Nunnenkamp and Schweickert (1990). Other studies, however, have found that devaluations have contractionary effects. These include the studies by Branson (1986) for Kenya; Sheehey (1986) and Solimano (1986) for Latin America; and Agenor (1991) and Khan and Knight (1982) for a diverse group of developing economies. The study by Edwards (1986) found that devaluations have a neutral effect on output.

A significant positive effect of outward orientation on growth has been reported by Agarwala (1983); Alam (1991); Cottani, Cavallo, and Khan (1990); Dollar (1992); Edwards (1988 and 1992); Feder (1983); Ghura (1992); Ghura and Grennes (1993); Knight, Loayza, and Villanueva (1993); Ram (1987); and Roubini and Sala-i-Martin (1991). Also, a survey by Edwards (1993) lists empirical studies that found outward-oriented trade strategies to be beneficial to economic growth. However, he points to several weaknesses of the existing studies, such as misspecified equations, simultaneity bias, questionable proxies, and problems of causality.

As noted by Kormendi and Meguire (1985), if a policy variable works mainly through the efficiency channel, the inclusion of the private investment ratio in the growth equation would raise the significance of the coefficient of the policy variable, but would not substantially change the value of its coefficient. However, if a policy variable works mainly through its effect on the volume of investment, the inclusion of the private investment ratio would lower the significance and magnitude of the coefficient of the policy variable. Using the total investment ratio, De Gregorio (1993) finds that the effect of inflation on growth worked through its effect on the efficiency of investment; De Gregorio and Guidotti (1992) find that financial intermediation affects growth through its effect on the efficiency of investment; and Fischer (1991) and Kormendi and Meguire (1985) find support for the effects of policy variables through both the efficiency and volume channels.

It would be interesting to investigate whether the same result would hold with data covering a longer period of time. The paper by Ghura (1992), using data for 33 countries in sub-Saharan Africa during 1970–87, shows that the effect on per capita growth of the total investment ratio is positive and significant when the latter is included in a regression together with macroeconomic policy variables. This result would seem to indicate that over the long run both the volume and efficiency of total investment matter for growth in the region.

Private investment is taken as a residual between total investment and government investment. Since the data for investment by public enterprises are not available, it is not possible to separate out this type of investment from the private investment series.

Contemporaneous changes in the terms of trade did not have a significant effect on growth.

The effects of the terms of trade on growth have been investigated by other empirical studies. Ghura and Grennes (1993) found that growth in sub-Saharan Africa was affected more by domestic policies than by the terms of trade during 1972–87. Fry (1986), in the context of Asian countries during 1961–83, and De Gregorio (1992), in the context of Latin America during 1950–85, found no support for the direct effect of the terms of trade on growth. Also, in the context of Latin American countries, Edwards (1983) found statistically significant relationships between the terms of trade and growth for only two of the six countries considered.

Changes in the terms of trade could also affect economic growth through their impact on the real effective exchange rate. Edwards (1989) and Khan and Ostry (1991) discuss the effects of the terms of trade on the real exchange rate.

Following the empirical literature on growth, human capital is measured by an initial level of education, in this case the level of secondary school enrollment ratio in 1970. It is expected that the effects of human capital on growth, especially that of secondary school enrollment, occur with a lag; thus the choice of the 1970 school enrollment ratio for the time period investigated is reasonable.

The positive effects of human capital development have been found by many other researchers (e.g., Barro (1989) and Mankiw, Romer, and Weil (1992)). In the context of sub-Saharan Africa, Ghura (1992) reports a positive and significant effect on growth of life expectancy at birth.

This correlation coefficient is minus 0.28 and highly significant for this group of countries.

Gastil (1987) and McColm and others (1991) have constructed indices of political rights and civil liberties for many countries, including all the countries included in this study. The index developed by McColm and others (1991) is used in the empirical analysis of this study to investigate the effect of political liberalization on growth. Table 17 describes the methodology used by McColm and others (1991) to construct the political reform index.

See Table 17 for a description of how DRY is measured. Owing to data limitations for the variables DRY and FREE, this regression covers 1986–91.

Botswana and Mauritius are two of the outliers as regards the degree of political freedom; they both had a significantly higher degree of political freedom than the rest of the countries during 1986–91. When these two countries are removed from the sample, the results do not change.

The secondary school enrollment ratio is still significant at the 5 percent level for the one-tail test.

Fry (1986), using the rate of change in agricultural output adjusted for the relative size of agriculture as a measure of weather, found that it had a significant positive effect on growth in the Asian context during 1961–83.

This is due to the removal of the year-to-year variability from the data. This smoothing of the series (especially for growth) maps the original data onto a tighter range, thus allowing a better fit.

This result is not robust because, as shown below, the inclusion of foreign assistance in this regression weakens the significance of private investment.

A recent book by Serven and Solimano (1994) provides a comprehensive survey of the current research on the determinants of private investment. The paper by Oshikoya (1994) investigates the macroeconomic determinants of private investment in eight African economies.

Note that the negative effect of inflation on domestic savings is significant only when the dummy variable for sustained adjusters or for countries with protracted imbalances is included.

The studies by Gupta (1987), Lahiri (1989), and Schmidt-Hebbel, Webb, and Corsetti (1992) investigated the effects of inflation on savings in developing countries. The results are mixed and generally statistically insignificant.

The percentage change in the real effective exchange rate did not have a significant effect on the rate of private investment for this sample of countries during 1986–92. Solimano (1989), using time-series data for Chile, found empirical support for a negative effect of the real exchange rate on private investment in the short run. However, his findings suggest that in the medium term, private investment is stimulated by a real depreciation.

It must be noted that when RERGSD is included in the regression for private investment, the effect of the budget deficit ratio remains negative but loses its significance.

The rigidity of interest rates in the context of sub-Saharan Africa has been due to a number of factors, in addition to government controls. These factors include the oligopolistic nature of the domestic banking system; inadequate banking supervision; and thin domestic money, credit, and capital markets.

Blejer and Khan (1984) and Sakr (1993) have distinguished between government investment on infrastructure and other government investment; they have found that the former has a larger positive impact on private investment, while the latter can actually be counterproductive to private investment. Similarly, investigating the responsiveness of farm output to government investment in infrastructure, Antle (1983) and Binswanger and others (1987) provide empirical evidence that investment in rural infrastructure (such as roads and irrigation networks) has a significant positive impact on farm output in developing countries.

Fry (1989) found that a low ratio of foreign debt to GDP stimulated the rates of national savings and total investment for a group of 28 highly indebted developing countries during 1967–85. The paper by Aghevli and others (1990), using data for 86 developing countries during 1982–88, found that countries without debt-servicing problems had significantly higher rates of national savings.

In a pooled analysis of eight African economies, Oshikoya (1994) found the effect of the changes in the terms of trade on private investment to be insignificant.

Greene and Villanueva (1991) included the level of real per capita GDP in their estimated private investment equation, with a view to capturing the effects of changes in savings, and found its effect to be statistically insignificant for 1975–87.

See the papers by Aghevli and others (1990), Collins (1989), Fry (1986), Lahiri (1989), Leff (1969), Mason (1988), Ram (1982), and Rossi (1989). See also the survey article by Hammer (1986).

As elsewhere in this paper, the data should be regarded with some caution. Foreign assistance flows used in this subsection are based on reporting by donor countries to the OECD. They are likely to be biased upward, to the extent that some donors may exaggerate the value of their aid in kind, and they include technical assistance, which for some countries can account for as much as one-third of foreign aid. On the other hand, the data do not include debt rescheduling, which in many countries provided substantial balance of payments relief during the period under review.

The issue of the sustainability of foreign assistance itself, particularly for countries that rely heavily on foreign aid, is not addressed.

The high correlation between foreign aid and domestic economic variables indicates potential endogeneity problems; this study attempts to address these problems by lagging foreign aid by one period—see Appendix I for details.

The foreign assistance and external debt variables are highly correlated, with a correlation coefficient of 0.64.

This result should be interpreted with caution, because growth rates were by definition very low in these countries during the period.

Using a generalized least-squares estimation method, and growth of real per capita GDP as the dependent variable, both foreign aid and the square of foreign aid are introduced among other explanatory variables. Both are significant, but the square of foreign aid carries a negative sign.

The first estimated regression in Table 25 yields a threshold of foreign aid equivalent to about 25 percent of GDP.

It would be interesting to examine whether the threshold beyond which foreign aid harms economic growth is influenced by the composition of the aid inflows (e.g., food aid and project aid against balance of payments support), but this question is beyond the scope of this study.

    Other Resources Citing This Publication