This Selected Issues paper on Trinidad and Tobago highlights that real GDP growth accelerated slightly from 2.4 percent in 1995 to 3.2 percent in 1996. In both years, growth was mainly driven by a good performance of the non-oil sector, which expanded by 3 percent and 3.6 percent, respectively. Construction, distribution, and tourism grew at an especially rapid pace. Manufacturing showed an uneven performance, growing in 1995, but stagnating in 1996, which was owing to the differential effects of trade liberalization on its various subsectors.


This Selected Issues paper on Trinidad and Tobago highlights that real GDP growth accelerated slightly from 2.4 percent in 1995 to 3.2 percent in 1996. In both years, growth was mainly driven by a good performance of the non-oil sector, which expanded by 3 percent and 3.6 percent, respectively. Construction, distribution, and tourism grew at an especially rapid pace. Manufacturing showed an uneven performance, growing in 1995, but stagnating in 1996, which was owing to the differential effects of trade liberalization on its various subsectors.

III. Trinidad and Tobago: Savings and Investment Functions

A. Introduction

70. One of the major challenges facing Trinidad and Tobago is to reduce the high level of unemployment as indicated in Section II of this report. An increase in the rate of economic growth especially in the labor intensive sectors of the economy is crucial in order to realize this objective. Increased savings and investment are essential components to increasing the rate of economic growth. This section describes developments in savings and investment in Trinidad and Tobago, evaluates their determinants over the period 1973-95, and focuses on the policy variables that influence them.

71. The period 1973-95 is of particular interest because it straddles extremes in the Trinidad and Tobago economy’s fortunes. The initial period (1973-82) was characterized by the oil boom in 1973-82, which laid the seeds of some of the problems experienced in subsequent periods; a second period was associated with a sharp economic decline from 1983 to the early 1990s; and a third period was characterized by an economic recovery supported by an intensification of stabilization and structural adjustment measures which were initiated in the mid-1980s. The oil windfall led to high levels of savings and investment, and the government assumed increasing importance in investment activities. Public investment relative to GDP rose to unprecedented levels which became unsustainable once the terms of trade swung against the country’s main export.

72. This section is organized as follows, sub-section B describes the observed trends in savings and investment in Trinidad and Tobago, sub-section C analyses the factors determining the behavior of savings and investment in Trinidad and Tobago, and sub-section D draws some conclusions concerning policies to increase savings and investment.

B. Trends in Savings and Investment

73. The sharp rise in oil prices in 1973 raised domestic income and led to an increase in national savings. The bulk of the increase in national savings can be ascribed to the public sector which was the major domestic beneficiary of the oil windfall through taxes paid by the oil companies. National savings reached an average of about 27 percent of GDP in 1973-83 and declined to an average of 15 percent of GDP during 1984-90, but recovered to about 17 percent of GDP in 1990-95. Private sector savings fluctuated within a fairly narrow range (12-16 percent of GDP). Government savings ranged from more than 20 percent of GDP in some years in the 1970s to negative levels by several percentage points of GDP from the mid-1980s to the early 1990s (Figure 12a).

Figure 12.
Figure 12.


(In percent of GDP)

Citation: IMF Staff Country Reports 1997, 041; 10.5089/9781451837544.002.A003

Sources: Trinidad and Tobago Authorities; and Fund staff estimates.

74. Gross capital formation in Trinidad and Tobago was characterized by a period of rapid growth during the 1970s and by a decline to the 1973 level in the 1980s as per capita income fell. Total investment averaged 24 percent of GDP annually in 1973-83 but declined to an annual average of 17 percent of GDP in 1984-90 before declining further to an average of 14 percent of GDP in 1990-95 (Figure 12b). Much of this decline was due to a drop in government investment, as falling oil prices reduced the government’s ability to save and required the government to exercise expenditure restraint.

75. During 1973-83, government investment increased substantially to an average of about 13 percent of GDP compared to less than 5 percent of GDP in 1973. These investments took the form of participation in and full ownership of enterprises, especially energy-related. This level of investment was unsustainable after the oil revenue declined sharply and other revenues contracted in line with the general slow down in economic activity. Capital expenditure by the government bore the brunt of fiscal adjustment when the terms of trade shifted against the country, and it declined to an average of less than 5 percent of GDP during 1984-90, before declining even further to an annual average of less than 1½ percent in 1990-95. Such a low level of capital expenditure does not cover depreciation of the capital stock and raises serious concerns about the effects of decaying infrastructure on economic growth.

76. Private investment on the other hand, was much less volatile, with its average level roughly unchanged at about 12 percent of GDP in 1973-95. Before 1990, private sector investment was concentrated in the energy sector, protected import-substituting industries such as food processing, assembly type industries and textiles, and services.

77. Foreign direct investment, which was about 5 percent of GDP at the beginning of the oil boom in 1973, peaked at more than 8 percent of GDP in 1975, reflecting investment activities in the energy sector. In the following eight years, foreign direct investment slackened to about 4 percent of GDP before dropping sharply to an annual average of less than 1 percent of GDP during the period 1983-88. The level of foreign direct investment in the oil sector picked up substantially in 1989-95, as petroleum taxes were lowered. Foreign direct investment peaked at 11 percent of GDP in 1994 before declining to about 6 percent of GDP in 1995 (Figure 12c). The recent shifts largely reflected the price developments in the world petroleum market. Additionally, the government’s divestment program and liberalization measures promoted private investment in tradable goods sectors and export-oriented activities.

C. Factors Explaining Savings and Investment in Trinidad and Tobago


78. There are various approaches to explaining saving with the main differences among them being determined by the time frame over which the analysis of consumption behavior is based. Dissatisfaction with current income as a predictor of current consumption led to the shift in focus toward the permanent income hypotheses (PIH) and the life-cycle income hypotheses (LCH) both of which are premised on the notion of intertemporal utility maximizing behavior on the part of households. The PIH explains fluctuations in the savings ratio by the slow adjustment of consumption to variations in income while the LCH explains it in terms of a desire to smooth consumption over a lifetime. These intertemporal approaches to saving now dominate the literature on savings.16 The analysis of savings behavior in developing countries has to contend with problems associated with the fact that these economies are undergoing rapid structural changes, that in low income countries most households are primarily production units and not just consumption units, and that conditions of perfect competition may not hold.

79. Although the various theories suggest different formulations of savings functions, there are some explanatory variables which are common to all.17 Most empirical work has tended to adopt an eclectic approach. The range of variables used includes income growth, per capita income, government saving, interest rates on deposits, exports, financial development, inflation, foreign savings, external debt, population growth, the ratio of the non-working to the working population, changes in the terms of trade, and public investment.

80. Among the issues that researchers have sought to settle are the extent to which consumption is influenced by current income or whether it fully reflects intertemporal considerations; the relevance of the Ricardian equivalence with respect to private sector saving and government dissaving; the impact of income growth, demographic factors, government tax and expenditure policies, availability of foreign exchange, terms of trade, and the rate of inflation on the savings rate, and whether saving is interest elastic.

81. This analysis of the determinants of savings in Trinidad and Tobago focuses on fiscal variables (the overall budget deficit (GBUD), current balance (GSY), and investment by the government (IG)), demographics (DEPEND AND POPGR), real GDP per capita in U.S. dollars (YPCUS) and GDP growth (RYG), interest rates (RDR), a measure of financial deepening (BMY), inflation (INFL), changes in the real effective exchange rate (CREER), foreign savings (ECA), changes in the terms of trade (CTOT), and the debt to GDP ratio (DBTY) and debt service (DBTS). The general functional form for the domestic savings rate investigated here is as shown in equation (1).18


82. The expected signs are indicated below the respective variables and the format of the regression output is shown in equation (2). Domestic saving is expected to be positively affected by income growth, fiscal surpluses, well developed financial intermediation, per capita income, and real deposit rates, and negatively affected by the level of foreign savings, the rate of inflation, the dependency ratio and population growth, debt ratios, and appreciation of the real effective exchange rate. The impact of changes in terms of trade can be positive or negative as discussed below.

83. The estimated equations for both domestic savings and private savings explain most of the variation of the two dependent variables and have good fits as reflected by their adjusted coefficients of determination which fall within the range 0.78-0.90 and satisfy tests for stability. Detailed diagnostic tests are shown in Attachment I.19 As shown in equation (2), economic growth, per capita income, the level of monetization of the economy, its external current account on the balance of payments, the level of external debt and the constant term were statistically significant and with the expected signs.


84. While statistically insignificant, the overall budget balance had the expected sign. Inflation, the dependency ratio, changes in the terms of trade, real deposit rates, external debt service ratio, population growth, and changes in the real effective exchange variables included in the general specification referred to above were statistically insignificant and were excluded from the selected equation. Their signs tended to bounce around in the various specifications tried.

85. The Trinidad and Tobago data suggest that the domestic savings rate is a positive function of growth in real income, per capita income in U.S. dollars, and a negative function of the debt ratio and of foreign savings as measured by the external current account.20 The negative impact of foreign savings on domestic savings can be explained by the tendency of abundant foreign savings to lessen the need for increasing domestic savings, especially public savings.21 The significant and negative coefficient of the debt ratio is consistent with the debtoverhang hypothesis, suggesting that high debt ratios can be a drag on both savings and investment rates.22 If economic agents read the existence of high debt ratios as a sign of higher taxes in the fixture, this might motivate them to increase their savings. However, the results in this case seem to suggest another interpretation in terms of perceptions about the prospects of the economy. The debt ratios in Trinidad and Tobago rose from an average of 15 percent of GDP in 1973-83 to 44 percent of GDP in 1984-95 and that may well have been interpreted as a harbinger of worse economic conditions. In this situation, expectations about macroeconomic balances, especially prices and the exchange rate, might have motivated people to acquire real assets by drawing down their savings or reducing their savings rates.

86. The initial formulation of independent variables to explain private saving is shown in equation (3). Private saving is expected to be positively influenced by the growth of income, foreign saving, financial intermediation, per capita income, real deposit rates, and government saving; and negatively influenced by budget deficits, government investment, appreciation of the exchange rate, and the demographic variables. As discussed below, the impact of inflation and changes in terms of trade on private savings are indeterminate.


87. The regression equation (4) for the determination of the private savings rate shows that the government’s overall balance (GBUD), the external current account balance (ECA), per capita income (YPCUS), and government investment (IG) were statistically significant and with the expected signs, except for the coefficient of the current account variable which turned out to be positive and different from its behavior with respect to domestic savings. This suggests different motivations for the saving behavior of the government and private economic agents. However, the positive coefficient is consistent with the notion that rising export earnings can raise the propensity to save and is consistent with the findings of other studies.23 More detailed diagnostic statistics are presented in Attachment II.


88. As noted in Goldsbrough et al (1996), empirical studies generally refute the idea that changes in public saving are fully offset by changes in private saving (Ricardian equivalence). The overall budget position variable in the case of private savings has a negative sign and thus suggests some degree of Ricardian equivalence in the Trinidad and Tobago data. The impact of inflation on the private saving rate is ambiguous. This is because nominal interest rates are usually increased as a monetary policy instrument to contain inflation and these high interest rates may in turn lead to higher savings. On the other hand, lower savings may be a reaction to the uncertainty caused by episodes of high inflation. In any case, the inflation variable, though positive in the case of domestic savings and negative in the case of private savings was statistically insignificant in the Trinidad and Tobago data.

89. While positive as expected, the coefficient of the rate of economic growth variable was statistically insignificant as a determinant of the private saving rate. Other variables included in the general specification were also statistically insignificant and not included in the selected equation.


90. Theoretical explanations of investment range from the simple accelerator model which characterizes investment as a function of the difference between the initial actual level of capital stock and the desired level of capital, to approaches such Tobin’s Q model, the internal liquidity model, and the expected profits model. Tobin’s Q model relates investment demand to financial variables through the q ratio. This ratio represents the relation of the rate of return on investment to the cost of capital In contrast, the internal liquidity model relates the desired stock of capital to the availability of internal funds as follows:


where α is the desired ratio of capital to internal funds available for investment and L is a measure of liquidity. The expected profits model relates the desired capital stock to the market value of the firm as follows:


where ά is the desired ratio of capital to the firms’ market value and V is the market value of the firm.

91. These models of investment demand assume developed financial markets and readily available data on capital stocks and cost of capital. A number of studies have pointed out the shortcomings of some of these models, especially in a developing country context.24 In particular, financial markets in these countries are underdeveloped, institutional problems often arise in foreign exchange and labor markets, the share of public investment is often quite large relative to the private sector investment, and data on the cost of capital and the stock of capital are sparse.

92. Private investment is influenced by economic conditions, while public investment is a policy variable that the authorities can adjust. Research by Khan and Reinhart (1990) lends support to the idea that private investment has direct linkages to economic growth. Studies on the relationship between public and private investment have confirmed the notion of complementarity between the two. Blejer and Khan (1984) found that the level of private sector investment was a positive function of the trend level of government investment and established the long-run complementarity but short-run substitutability between the two variables. Aschauer (1989) concluded that higher public sector investment raises the national investment rate above the level chosen by rational agents and induces an ex ante crowding out of private investment, but it also raises the return to private capital which in turn crowds in private investment.

93. Most of the empirical work done on investment seems to use some variant of the accelerator model and takes as a point of departure the flexible accelerator model of Chenery and Koyck. One of the strengths of the accelerator model derives from support of its basic tenets by substantial empirical work and its malleability. The range of variables commonly tested in investment functions include economic growth, per capita income, the level of prices, interest rates, fiscal policy variables, foreign savings, external debt, exports, exchange rate movements, and foreign direct investment. The determinants of private investment (Ip) examined in this study include economic growth (RYG), real per capita income level in U.S. dollars (YPCUS), inflation (INFL), real lending rates (RLND), debt burden as measured by the debt service ratio (DBTS) and the proportion of external debt to GDP (DBTY), the rate of public sector investment (IG), overall budget deficit (GBUD), government saving (GSY), reserves in months of imports (RSVS), the share of oil exports (OLEX), changes in the real effective exchange rate (CREER), real credit to private sector (CRPVT), foreign direct investment (FDI), and the lagged dependent variable (Ip_1).

94. RYG and YPCUS are measures of income and its rate of change and are used as proxies of aggregate demand, INFL captures the diversion of resources from productive activities to rent seeking behavior and/or various forms of hedges against inflation, IG, GBUD, GSY capture the substitution or complimentary effects of fiscal policy to private investment; RLND and CRPVT capture the impact of monetary policy and the financial system on private investment. Higher real interest rates imply a rise in the cost of capital as well as an increase in the opportunity cost of retained earnings. In situations of financial repression, the volume of credit may be a better indicator of monetary policy than the cost of credit. CREER captures the impact of a change in the real exchange rate on investment behavior. In cases of high dependence on imported capital goods, a depreciation would result in a decrease in investment as well as an increase in the debt burden, while an appreciation would have the opposite effect.25

95. The measure for available foreign reserves (RSVS) captures the impact of foreign exchange availability on investment. In some developing countries, investment is often constrained by the lack of foreign exchange and this is especially true in situations where exchange controls and other restrictions are applied. OLEX reflects the extent to which investment might be positively influenced by the existence of abundant natural resources. DBTS and DBTY capture the negative effects of a rising debt burden on investors’ confidence and therefore willingness to invest in a given economy.

96. A general form of an investment equation was tested including all variables and the expected signs of the coefficients are shown in equation (10). The regression format is presented in equation (11) and more detailed diagnostic tests are presented in Attachment III. In this formulation, private investment is postulated to be positively influenced by economic growth, per capita income, government budget surpluses, availability of credit, foreign reserves, exports, and foreign investment; and negatively influenced by inflation, external debt variables, and the cost of credit. A priori, the effects of government investment and exchange rate movements are indeterminate.


97. As can be observed from the regression equation (11) below, the overall fit of the equation was good and the coefficient of determination (R2) adjusted for degrees of freedom was 0.88.


98. Regression results show that among the tested explanatory variables for private investment, growth in real GDP, investment by the government (IG), real lending rate (RLND), inflation (INFL), credit to the private sector (CRPVT), external debt (DBTY), and lagged private investment (Ip_l) are all statistically significant and had the expected signs. The real growth in GDP has a positive influence on the behavior of private investment, while lagged private investment has a negative effect, which is consistent with the theoretical expectation of the accelerator principle that investment is positively related to the level of current income and negatively to the existing capital stock. While correctly signed, the income per capita variable was not statistically significant in the selected equation.

99. As indicated above, the relationship between private and public investment can be one of substitution or complementarity. The results of the regression suggest that the relationship was one of substitution during the period covered by the Trinidad and Tobago data. This is plausible, especially after 1973 when there was a large increase in public investment activities. There is a negative relationship between the real lending rate and investment behavior, suggesting that higher real lending interest rates tend to discourage investment activities. The negative coefficient for inflation suggests a negative impact of inflation on the rate of private investment.

100. The fact that both the real lending interest rates and inflation were found to be statistically significant determinants of private investment in Trinidad and Tobago is consistent with the existence of financial repression in the early part of the period covered by these estimates. The significance of credit to the private sector as a determinant of private investment is expected, given the existence of financial repression in Trinidad and Tobago during the period in question. With the financial liberalization that has occurred in recent years, the expectation would be for the interest rate variable to take on much more significance than the availability of credit in determining private investment.

101. The significance of the external debt variable is consistent with the notion that resources required to service debt potentially crowd out investment, and that a large ratio—such as the 56 percent of GDP level that Trinidad and Tobago’s debt reached in the late 1980s—may mean a debt overhang which could make economic agents anticipate large future tax liabilities, induce capital flight, and thus raise the implicit cost of domestic capital and negatively impact private investment. Other variables in the initial specification proved to be insignificant and their signs bounced around. An alternate equation with lagged per capita income in U.S. dollar terms, government investment, and debt service as explanatory variables, and including a lagged adjustment process, exhibited a poor fit (an R2 of 0.36). In this equation, only the estimated coefficient of debt service was statistically significant and the income variable was negative.

D. Conclusions

102. The regression results suggest that the main explanatory factors for the behavior of the domestic savings rate in Trinidad and Tobago during the period reviewed were economic growth, per capita income, public sector savings and foreign savings, and the degree of monetization of the economy as measured by the inverse of the demand for money function. The private saving rate was shown to be influenced by fiscal policy variables (GBUD, GSY, and IG), per capita income, and foreign savings. Private investment was shown to be mainly influenced by economic growth, public investment, real lending rates, inflation, the availability of credit, external debt, and its own past levels.

103. These findings are generally in line with theoretical expectations, consistent with a number of other studies, and confirm the importance of the macroeconomic environment in determining the outcomes in these variables. However, some caution is in order in the interpretation of these results. Although the selected equations pass a battery of tests, a limitation has to do with the lack of variables to specifically account for the structural changes that the economy went through during the period covered. Subject to the data and methodological limitations of the study, the results point to policy measures that authorities may deploy to address deficiencies in the savings-investment process. For example, targeting a strong fiscal position, a low level of inflation, a stable exchange rate, and efficient public investment in infrastructure would be conducive to higher savings rates and increased investment. However, such public investment should not crowd out private investment, hence the need to target higher levels of public savings.

104. The principle findings of this investigation support the conventional wisdom that the establishment of a stable macroeconomic environment through judicious fiscal, monetary, and external policies provides a desirable environment for the increase in savings and investment, which in turn provide the basis for economic growth. In the case of Trinidad and Tobago, consideration has to be given to the type of investment and its sectoral composition as well. This is because, in addition to the benefits of growth itself, much depends on the employment generation that follows from that growth.

ATTACHMENT I Diagnostic Tests for Domestic Savings, the Sample is: 1973 to 1995

Equation 2

Modelling Domestic Savings by OLS; The sample is: 1973 to 1995


RSS = 41.68796214 for 7 variables and 23 observations

Variance instability test: 0.211; Joint instability test: 1.221

Information Criteria: SC = 1.549; HQ = 1.2903; FPE = 3.3985

R2 adjusted = 0.779

AR1-2F(2,14) = 2.7147 [0.1009]

ARCH 1 F(1,14) = 0.031932 [0.8607]

Normality Chi2(2) = 3.0717 [0.2153]

Xi2 F(12, 3) = 0.48105 [0.8434]

RESET F(1, 15)= 0.56199 [0.4651]

Normality test for Residual

Sample size 23

Mean 0.0000

Standard Deviation 1.346299

Skewness -0.525224

Excess Kurtosis - 0.777568

Minimum -3.036890

Maximum 1.653973

Normality ChP2(2) = 3.0717[0.2153]

ATTACHMENT II Diagnostic Tests for Private Savings, the Sample is: 1973 to 1995

Equation 4

Modelling Private savings (PSY) by OLS; The sample is: 1973 to 1995


RSS = 32.556 for 6 variables and 23 observations

Variance instability test: 0.113; Joint instability test: 1.437

Information Criteria: SC = 1.165; HQ = 0.944; FPE = 2.415

R2 adjusted = 0.897

AR 1- 2F(2, 15) = 1.3583 [0.2870]

ARCH 1 F(1, 15) = 0.012805 [0.9114]

Normality Chi2(2)= 1.4776 [0.4777]

Xi2 F(10, 6)= 0.71083 [0.6984]

RESET F(1, 16)= 0.056496 [0.8151]

Normality test for Residual

Sample size 23

Mean 0.0000

Std. Devn. 1.220166

Skewness -0.338122

Excess Kurtosis -0.858192

Minimum -2.536034

Maximum 1.744895

Normality Chi2(2) = 1.3994[0.4967]

ATTACHMENT III Diagnostic Tests for Private Investment, the Sample is: 1973 to 1995

Equation 9

Modelling Private Investment (Ip) by OLS; The sample is: 1974 to 1995


R2 = 0.786 F(8, 13) = 5.9637 [0.0024] σ = 1.02677 DW = 2.32

RSS = 13.70537538 for 9 variables and 22 observations

Variance instability test: 0.137283; Joint instability test: 1.12792

Information Criteria: SC = 0.791263; HQ = 0.450071; FPE= 1.48555

R2 adjusted = 0.882

AR 1-1 F(1, 12) = 0.73194 [0.4090]

ARCH 1 F(1,11) = 0.14829 [0.7075]

Normality CM2(2)= 5.3158 [0.0701]

RESET F(1, 12) = 0.055445 [0.8178]

Normality test for Residual

Sample size 22

Mean 0.0000

Std. Devn. 0.789286

Skewness -0.684669

Excess Kurtosis 1.233374

Minimum -2.222291

Maximum 1.595252

Normality Chi2(2) = 5.3158 [0.00701]

ATTACHMENT IV Data Sources and List of Variables

The main sources of the data are IMF, International Financial Statistics and Recent Economic Development Reports, and the Central Statistical Office publication, ‘The National Income of Trinidad and Tobago.”

article image


  • Aghevli, Bijan B., James Boughton et al, “The Role of National Saving in the World Economy: Recent Trends and Prospects” Occasional Paper 67. (Washington: International Monetary Fund, March 1990).

    • Search Google Scholar
    • Export Citation
  • Aschauer, David A.,Does Public Capital Crowd Out Private Capital?”, Journal of Monetary Economics 24 (1989) pp. 171188.

  • Doornik Jurgen and David Hendry, PcGive 8.0: An Interactive Econometric Modelling System, Institute of Economics and Statistics, University of Oxford, London, International Thomson Publishing.

    • Search Google Scholar
    • Export Citation
  • Giovannini, Alberto,The Interest Elasticity of Savings in Developing Countries: The Existing Evidence,World Development, 11(7), July 1983, pp. 601607

    • Search Google Scholar
    • Export Citation
  • Goldsbrough David, et al, “Reinvigorating Growth in Developing Countries: Lessons from Adjustment Policies in Eight Economies” Occasional Paper 139, IMF, Washington DC, July, 1996.

    • Search Google Scholar
    • Export Citation
  • Greene, Joshua and Delano Villanueva,Private Investment in Developing Countries: Empirical Analysis,IMF Staff Papers, Vol. 38, No. 1, March 1991.

    • Search Google Scholar
    • Export Citation
  • Gupta, Kanhaya, L.Foreign Capital Inflows, Dependency Burden, and Saving Rates in Developing Countries: A Simultaneous Equation ModelKyklos, Vol. 28-Facs.2, pp. 358374, 1975.

    • Search Google Scholar
    • Export Citation
  • Hadjimichael, Michael, Dhaneshwar, Ghura et al, “Sub-Saharan Africa, Growth, Savings, and Investment, 1986-93” Occasional Paper 118 (Washington: International Monetary Fund, January 1995).

    • Search Google Scholar
    • Export Citation
  • Jansen Karel, Finance, Growth, and Stability: Financing Economic Development in Thailand, 1960-86, Gower Publishing Company Limited, England, 1990.

    • Search Google Scholar
    • Export Citation
  • Jorgenson, Dale, W.,Economic Studies of Investment Behavior,The Journal of Economic Literature, Volume IX, December 1971, No. 4 pp. 11111148.

    • Search Google Scholar
    • Export Citation
  • Khan, Mohsin S., Carmen M. Reinhart,Private Investment and Economic Growth in Developing CountriesIMF Working Paper WP/89/60, (Washington: International Monetary Fund, July 26, 1989).

    • Search Google Scholar
    • Export Citation
  • Kennedy, P., A Guide to Econometrics, The MIT Press, 1985.

  • Laumas, P.Exports and Propensity to Save,Economic Development and Cultural Change, July 1982.

  • Levacic, R. and Rebman A., Macroeconomics: An Introduction to Keynesian-Neoclassical Controversies, 1982 The Macmillan Press Ltd., London.

    • Search Google Scholar
    • Export Citation
  • Lee, J.K.,Exports and the Propensity to Save in LDCs,The Economic Journal, June 1971, pp. 341351.

  • Maizels, A., Exports and Economic Growth in Developing Countries, Cambridge: Harvard University Press, 1968.

  • Masson, Paul, R., Tamin Bayoumi, and Hossein Samiei,International Evidence on the Determinants of Private Saving,IMF Working Paper WP/95/51, (Washington: International Monetary Fund, May 1995).

    • Search Google Scholar
    • Export Citation
  • Mikesell R.F. and Zinser J. E.,The Nature of the Savings Function in Developing Countries: A Survey of the Theoretical and Empirical Literature,The Journal of Economic Literature, Vol. XI, March 1973, No. 1.

    • Search Google Scholar
    • Export Citation
  • Ogaki, Masao, Jonathan D. Ostiy, and Carmen M. Reinhart,Saving Behavior in Low and Middle-Income Developing Countries: A Comparison,IMF Working Paper WP/95/3, (Washington: International Monetary Fund, January 1995).

    • Search Google Scholar
    • Export Citation
  • Yusuf, S. and Peters R. K.,Saving Behavior and its Implications for Domestic Resource Mobilization; The Case of the Republic of Korea,World Bank Working Papers, Number 628, 1984.

    • Search Google Scholar
    • Export Citation
  • Yotopoulos, P. and Nugent, J., Economics of Development: Empirical Investigations, Harper and Row Publishers, New York, 1976.


See IMF staff studies for the World Economic Outlook, September 1995 for a detailed discussion of these two approaches.


See Jansen (1990) for a detailed discussion on the limitations of the various theories and the range of explanatory variables identified.


A list and descriptions of the variables used in the regressions are presented in Attachment IV. The period covered is 1973-95 and the method of estimation used for the multiple regressions in this exercise is Ordinary Least Squares (OLS), t statistics are shown in parenthesis below the estimated regression coefficients and two stars and one star indicate significance at the 1 and 5 percent levels respectively. This also applies to the other regression equations in this section. All variables were expressed as rates of change or in scaled form relative to GDP so as to remove spurious correlation. The data series used are annual.


The equations are selected on the basis of the model selection criteria specified in PC-Give which includes Schwarz (SC), Hannan-Quinn (HQ) and the final prediction error (FPE). These criteria are similar to the use of R2, adjusted R2 and the variance of the disturbance term (σ̂2) but tend to penalize for additional parameters more than the degrees of freedom adjustment to the standard error of the equation. Equations with the least SC, HQ, and FPE are selected as the “best.”


In a world of perfectly elastic supply, foreign savings are determined by the difference between national saving and investment. However, in the real world the direction of causalility may be reversed and foreign savings could become a potential determinant of national saving. See IMF (1995) for an elaboration of this issue. Therein also are cited examples of empirical evidence in support of the negative coefficient (e.g., Fry (1978) and Giovannini (1985)).


This contrasts with the result in the equation for private savings.


See Hadjimichael et al and references confirming similar results.


The works of Laumas (1982), Lee (1971), Maizels (1968) and Papanek (1973) support this view which is also consistent with the Laursen-Metzler-Harberger effect that negative terms of trade shocks lower saving via adverse effects on wealth and income.


For example see Greene and Villanueva (1991), and Blejer and Khan (1984).


On the other hand, a real exchange rate depreciation improves the profitability of the tradable goods sector, which in turn is expected to attract increased investment in that sector. Thus, the impact of exchange rate movements on private investment is ambiguous.