Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Multi-Country Evidenceon the Effects of Macroeconomic, Financial and Trade Policieson Efficiency of Resource Utilization in the Developing Countries

Author(s):
Mathew Odedokun
Published Date:
July 1992
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

I. Introduction

This study examines the effects of selected macroeconomic policies on the economic efficiency in a sample of 81 developing countries by pooling cross-country data over various subperiods between 1971–90. (The complete list of countries and years covered is given in the Appendix on p. 42). To measure economic efficiency, the incremental output-capital ratio is used, while the macroeconomic variables being considered include export orientation, the size of the public sector, directed credit programs through development bank lendings, financial depth, the inflation rate, the real interest rate, and the real exchange rate distortion.

A viable way of fostering economic growth would be to promote economic efficiency as a centerpiece of macroeconomic policies. This requires identification of the direction of the effects each policy will have on efficiency. Another reason for identifying the effects of policies on the efficiency of resource utilization—and hence indirectly on economic growth—is that this approach is econometrically more feasible than to directly study their effects on economic growth. Real GDP growth measures the growth in overall economic activities, of the variables that interact with and against each other in the economy. A consequence of this interaction is an inability to precisely identify, or at times to even approximate, the effects of individual variables on growth. This is not the case with respect to the efficiency of resource utilization where a one-way causation effect is more prevalent, i.e., running from the variables to efficiency of utilization of the resources.

In light of the above, this study attempts to test for the effects of relevant macroeconomic variables and to identify the direction of their effects on the productivity of resources. It is hoped that the wide geographical and temporal coverage will enhance the reliability of the results.

The paper is organized in five sections and an Appendix. Section II provides a description of the approach adopted in this study; in Section III the empirical results are presented and evaluated; Section IV contains the summary and conclusions, while Section V briefly surveys existing studies in this particular area of the literature. The paper concludes with an Appendix on data sources and techniques of variable measurement.

II. The Approach of This Study

Economic growth is a product of both the quantum of resources and the efficiency with which the resources are utilized. In promoting growth, economic efficiency is not less consequential than the quantum of resources—in fact, it may have an even greater impact on growth. For instance, the data employed in the present study show that while the simple correlation coefficient between the annual growth of real GDP and the investment-GDP ratio (a measure of capital resources) over the generated 1,847 annual data points is only 0.167. On the other hand, the correlation coefficient between the growth of GDP and the incremental output-capital ratio (which measures the efficiency of capital resource utilization) is 0.868.

As pointed out above, panel data, i.e., the pooled cross-country and time series data, are employed to derive estimates for the sample. Two estimation methods are used: the fixed-effect, and the variance component techniques (see Maddala, 1977 for description). 2/ There are many reasons for pooling data, including the ability to report estimates under alternative specifications for each of the 81 sample countries, which may be too unwieldy to report for individual countries separately so that pooled data estimates would provide a concise summary. Also, within countries, temporal variations in some of the explanatory variables me be insufficient for the estimated effects to be discernable, despite their significant actual effects.

1. Index of economic efficiency

Efficiency in this study is measured conventionally as the incremental output-capital ratio (IOCR). The IOCR, which is sometimes referred to as the social rate of return on investment, is the orthodox index, at the level of applied research, of the efficiency of allocation of capital resources (see Agarwala (1983), Anderson (1987), Khatekhate (1988), Gelb (1989) and Gallagher (1991), where they have been similarly employed as performance variables in evaluating the effects of some policies). 3/ In fact, IOCR is often regarded in the literature on economic development as a measure of the efficiency of resources in general, instead of investment alone.

There is a mathematical relationship between the IOCR, the share of investment spending in the GDP (i.e., the investment ratio) and the real GDP growth, which is as stated below:

where K = capital stock, I = capital formation, and Y = real GDP. That is, IOCR is the economic growth divided by the investment ratio, or economic growth is the product of IOCR and the investment ratio. A policy implication of this is that the effects of policy and nonpolicy determinants of economic growth can be analyzed into those that operate via the IOCR or efficiency channel; those that operate through the investment ratio channel; and those that operate through both. An advantage of this approach is that the inherent spurious association between economic growth and many other variables in the economy (or the inherent causation that runs from economic growth to each of them) can be more easily avoided. That is, such an association need not be found with each of the IOCR and the I/Y when they are considered separately.

Thus, despite some inevitable conceptual limitations in the use of IOCR as the index of the efficiency of resource utilization, 4/ the approach adopted in this study is to examine the effects of selected macroeconomic policies on it. Using IOCR is an indirect way of inferring the effects of such policies on economic growth. The scope of the analysis is not extended to a detailed examination of the policy effects on the second component of economic growth—the I/Y; only where necessary will their effects on the I/Y and the dY/Y be examined.

2. Macroeconomic policy determinants of economic efficiency

The discussion now proceeds to those macroeconomic variables whose effects on efficiency are to be considered. Needless to say, the choice of variables is materially influenced by the type of data that are available for developing countries in general. While the existence of nonpolicy factors—economic and noneconomic—on the efficiency of resource utilization is recognized, 5/ our immediate concern is with the policy variables. Specifically, because of the general lack of data on nonpolicy variables, the discussion is confined to consideration of policy variables only.

The real interest rate (denoted as RINT and measured as nominal interest minus the inflation rate) is one of the policy variables being examined. As reviewed in Section V, some previous studies have included RINT as a determinant of the IOCR, e.g., Agarwala (1983) as a component of his chosen index of distortion; Gallagher (1991) as a component of his measure of the rent-seeking variable; Gelb (1989) and Khatekhate (1988) in conducting the Mann-Whiteney test. But while many authors have reported the positive effect of RINT on the IOCR, Khatekhate found an absence of a discernable relationship. Through the data employed in the present study, a reconciliation of such non-uniform findings will be undertaken.

Another policy variable to be considered is financial depth. Alternative measures of it are tried, as follows: the level of financial depth (denoted as FINDEPTH) computed as the ratio of the stock of liquid liabilities of the banking system to GDP; the growth rate of FINDEPTH; and the ratio of the change in (flow of) the real value of liquid liabilities to real GDP. 6/ While the real rate of interest is a “financial repressionist” measure, the financial depth variable is a “financial structuralist” index, using the terminology of Gupta (1987) (see Section V for a discussion of these approaches). Financial depth is expected to have a positive effect on the IOCR, based on the economic logic contained in the literature. Gelb (1989) included a measure of financial depth (computed as the ratio of the flow of the banking system’s liquid liabilities to gross domestic savings) and reported a positive effect on the IOCR.

A third measure relating to the domestic financial sector is the size of the directed credit in the economy. Theoretical discussions in the literature contain references to directed credit schemes as a way of hampering the functioning of the financial system and instances have been cited in the nontheoretical literature too (e.g., World Development Report, 1989) whereby directed credit policies are claimed to have distorted resource allocations. But, probably because of the difficulty in getting a reliable measure of the extent of directed credit, previous studies have not included it in a formal model. What is done in this study is to use the size of development banking in the economy (denoted as DEVBNK and computed as the stock of development bank lendings to the private sector in relation to the GDP) as an index of the extent of directed credit policy. According to the World Development Report (1989, p. 57), “development finance institutions have been perhaps the most common means of directing credit.” This paper alternatively employs the level form of DEVBNK and its growth rate and expects the effect on the IOCR to be negative.

Exchange rate policy or exchange rate distortion is another financial variable that is tested in this study. As contended by Agarwala (1983, pp. 18–19) concerning the studies on exchange rate policy, “unfortunately, the development role of exchange rate policy was not emphasized in development economics. Instead, the exchange rate was viewed mainly as an instrument for tackling balance of payments problems and sometimes (by and large with disastrous consequences) for controlling inflation.” He thus included a measure of exchange rate distortion as one of the seven components of his distortion index in explaining the crosscountry variation of IOCR. The exchange rate distortion was simply computed as the extent of deviation of real effective exchange rates from their 1973–74 base values, which has been faulted by some subsequent writers, e.g., Khatekhate (1988). Gallagher (1991) too included an index of exchange rate distortion (computed as the premium of the black market exchange rate versus the official rate weighted by total exports) as one of the four components of his index of rent-seeking employed in explaining cross-country variations of IOCR in Africa. While this measure of the exchange rate has much to commend it in principle, it is not adopted here mainly because of nonavailability or inaccessibility of data on black market exchange rate movements for the sample countries over the period of the study. Thus, instead, the measure of exchange rate distortion employed is a dummy variable (denoted as EXCHDIST), which takes a value of unity if an “inappropriate” exchange rate policy is adopted in a year and a zero value otherwise. The “inappropriate” policy is taken to be real exchange rate appreciation following a current account deficit in the preceding year or a real exchange rate depreciation accompanying a surplus in the previous year. (The current account position in the previous, instead of the contemporaneous, period is chosen because the measured present period current account situation could be the effect, as opposed to and/or in addition to being the cause, of the present period real exchange rate policy.) This measure of exchange rate distortion is expected to have a negative effect on the IOCR.

The final price variable considered is the rate of inflation (denoted as INF). Quite apart from the indirect effects that domestic price level movements may have as a component of the real interest rate and real exchange rate movements, they may also exert distortions via other channels. Thus, it is considered as a separate factor.

An export-orientation policy variable is included as a determinant of IOCR. It is alternatively measured as the ratio of exports to GDP (denoted as EXPT) and the growth rate of this ratio. The channels of the effects of exports on IOCR are not necessarily supposed to be the same as that of the real exchange rate distortion. The export variable is supposed to capture the effects of trade policies designed to promote exports and, in principle, is not supposed to be positively or negatively related to our measure of exchange rate distortions since the distortion can result from both “overvaluation” and “undervaluation” of the domestic currency. In line with the mainstream idea in the literature, the export variable is expected to be a nondistortionary variable and hence to exert a positive effect on the IOCR.

Finally, the effects of the size of the public sector (denoted as GOVT) on the IOCR are considered. In line with most studies, we measure this by the share of government spending in the GDP, and both this share and its growth rate are alternatively considered. Also, both the central government total expenditure and the general government consumption expenditure are alternatively employed for comparison in view of some inadequacies in each of them when considered in isolation. There is also the exigency of nonavailability of data on each variable for all of the sample countries. A priori, we are not in a position to categorically posit a negative or a positive effect that would be associated with the size of public sector on IOCR but if the balance of evidence from the existing empirical studies holds true it can be expected to have a negative effect on the IOCR.

3. Specification of the model

As pointed out towards the end of Section V, subsection 1, the existing theoretical studies provide little guidance about the mathematical or econometric relationship between economic efficiency and those factors being posited as its determinants. Neither is there an indication of how the explanatory variables are to be measured at the econometric level. This is in contrast to the theoretical literature on economic growth which contains mathematical relationships between growth and its determinants, which are amenable to econometric estimation. Thus, in a study like the present one, the functional form of the estimated equations has no such theoretical underpinnings but has to be based on economic and econometric “logic.” The same applies to the mode of measuring the regressors. Having said this, the form of our estimated equations is as presented below:

where IOCR = incremental output capital ratio; EXPT = export-GDP ratio; GOVT = government expenditure-GDP ratio; DEVBNK = end-of-period stock of the development banking system’s credit to the private sector as a ratio of GDP; FINDEPTH = financial depth or end-of-period stock of liquid liabilities of the banking system as a ratio of GDP; INF = inflation rate; RINT = real rate of interest; and EXCHDIST = real exchange rate distortion dummy variable. The notations having the lower-case letters (viz: expt; govt; devbnk; and findepth) are the growth rate equivalents of the corresponding notations (in ratio form) of those in the upper-case letters, e.g., expt is the growth rate of EXPT. (In the alternative equation estimates, both FINDEPTH and findepth are defined and measured as the ratio of the change in the real value of liquid liabilities to real GDP). The e and v are the error terms, while the a’s and b’s are the parameters to be estimated.

It can thus be seen that the difference between equations (2) and (3) is the substitution of the level form for the growth rate of those variables expressed as ratios of GDP. Growth rate equivalents have the attraction of exhibiting much less intercorrelation and are therefore not prone to multicollinearity problems when compared with the level form of the ratios—although, much of the discussions in the literature and at the policy level are often couched in terms of the level-form. 7/ In effect, the two equations are complementary.

In estimating equations (2) and (3) above, a few practical modifications are made. First, in appropriate cases, regional dummy variables are included (one each for sub-Saharan Africa, Asia, and Western Hemisphere, with a composite group of developing countries in North Africa, the Middle East and Europe serving as the reference or base region). Second, trend variable is included. Third, not all the seven regressors in each of the equations (2) and (3) feature simultaneously. For instance, estimates are made separately for countries where data on the interest rate or on the size of the development banking system are not available. Finally, we try to incorporate dynamic specification by including the lagged values of the regressors, with those lagged values whose coefficients are statistically significant being retained in the reported estimates. 8/

III. Empirical Results

Before reporting the estimates of equations (2) and (3) some “stylized facts” will be presented in sub section 1 below. Meanwhile, it should be pointed out that the annual data employed for the 81 sample countries over periods falling between 1961 and 1990 come from the IMF’s International Financial Statistics Yearbook (for the relevant years). The details about data sources and how the variables are measured as well as the list of countries included in the study are contained in the Appendix. All variables, other than the dummy and trend variables are pure fractions, instead of percentages.

1. Some “stylized facts”

In Table 1, a correlation matrix for the variables employed is presented. The table also contains the mean values of these variables which are presented in the last four rows for the entire period; pre-1974 (or pre-oil shock) era; pre-1981 and post-1980 periods in that order.

Table 1.Mean Values of and Correlation Matrix for the Variables
IOCRYINVDEPTexptGOVT1govtlGOVT2govt2DEVBNKdevbnkFINDEPTH1fIndepth1findepth2INFRINTEXCHDIST
0.8321.000
INV-0.0130.1781.000
EXPT-0.0020.0670.3021.000
expt0.0420.0610.0110.0801.000
GOVT1-0.135-0.1280.1280.276-0.0681.000
govt1-0.094-0.1300.0600.003-0.1450.1571.000
GOVT2-0.139-0.0600.2290.382-0.0390.7800.0351.000
govt2-0.065-0.0680.0840.016-0.0610.0670.3970.1331.000
DEVBNK-0.104-0.1370.1360.125-0.0570.1850.0190.143-0.0001.000
devbnk-0.059-0.0680.1640.029-0.1300.0380.2340.0440.1800.0461.000
FINDEPTH1-0.0850.0430.2930.410-0.0080.379-0.0050.547-0.0240.1630.0481.000
findepthl-0.121-0.1530.0200.003-0.0820.0930.3270.0470.2360.0570.3180.0801.000
findepth20.2000.2620.2030.176-0.0530.1610.1930.1960.1600.0630.2240.3980.6861.000
INF-0.156-0.167-0.116-0.1070.079-0.057-0.1720.051-0.088-0.051-0.220-0.074-0.3230.2921.000
RINT0.1210.1400.0730.044-0.0950.0940.2780.0890.1580.1000.1940.1270.3830.382-0.6421.000
EXCHDIST-0.052-0.0640.0120.004-0.1020.032-0.0130.040-0.0370.028-0.0530.000-0.013-0.0290.046-0.0381.000
Trend-0.205-0.1890.1240.114-0.0090.237-0.0520.298-0.0630.238-0.0930.272-0.041-0.0180.1870.0210.054
Dummies for:
Africa-0.050-0.108-0.055-0.071-0.0210.0660.011-0.0170.008-0.0950.032-0.247-0.019-0.126-0.077-0.157-0.017
Asia0.0640.1090.0410.0130.009-0.169-0.018-0.152-0.005-0.0400.0230.0000.0030.054-0.089-0.098-0.042
W. Hemisphere-0.051-0.082-0.160-0.158-0.004-0.236-0.010-0.2140.0140.001-0.098-0.230-0.010-0.1330.1980.0800.035
Others0.0530.1080.2270.1800.0210.3070.0140.386-0.0190.1310.0450.5020.0270.229-0.0670.0050.016
Mean Values
During:
1962-900.1670.1580.2340.3040.0140.1390.0070.2350.0170.0670.0370.3690.0200.0210.1570.0050.460
1962-730.2610.3370.2140.2450.0290.1220.0090.1820.0140.0510.0550.2920.0270.0240.0690.0000.397
1974-800.1610.0210.2520.3300.0060.1450.0140.2440.0220.0660.0480.3570.0150.0210.162-0.0510.513
1981-900.0900.0020.2300.3300.0000.1510.0040.2740.0140.0840.0250.4470.0280.0200.1930.0660.477
Notes: (1) At significance levels of 1 percent; 5 percent and 10 percent, a correlation coefficient le ntatistically different from zero if its absolute value is up to 0.07, 0.05 and 0.04 respectively.

The notation are defined as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT1 = general government consumption expenditure/GDP ratio; govt1 = growth of GOVT1; GOVT2 = central government total expenditure/GDP ratio; govt2 = growth of GOVT2; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVBNK; FINDEPTH1 = stock of banking system’s liquid liabilities/GDP ratio; findepthl = growth of FINDEPTH1; findepth2 = change in real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rate; RINT = real interest rate; EXCEDIST = exchange rate distortions; Ẏ = per capita real GDP growth; INV = investment/GDP ratio; and IOCR = incremental output-capital ratio.

Notes: (1) At significance levels of 1 percent; 5 percent and 10 percent, a correlation coefficient le ntatistically different from zero if its absolute value is up to 0.07, 0.05 and 0.04 respectively.

The notation are defined as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT1 = general government consumption expenditure/GDP ratio; govt1 = growth of GOVT1; GOVT2 = central government total expenditure/GDP ratio; govt2 = growth of GOVT2; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVBNK; FINDEPTH1 = stock of banking system’s liquid liabilities/GDP ratio; findepthl = growth of FINDEPTH1; findepth2 = change in real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rate; RINT = real interest rate; EXCEDIST = exchange rate distortions; Ẏ = per capita real GDP growth; INV = investment/GDP ratio; and IOCR = incremental output-capital ratio.

Attention should be drawn to the negative correlation of most of the regressors with the IOCR and with the per capita real GDP growth in Table 1. The ever declining values of these last two variables over the three subperiods also deserve attention—the per capita real GDP growth was practically nil during the 1981–90 decade.

2. Presentation of the equation estimates

The estimates of the equations (2) and (3) are as reported in Tables 2 and 4 below. Table 2 contains the variance component panel data estimates for all of the countries for the entire 1961–90 period. 9/ The variance component estimates for each of the 1962–73; 1974–80 and 1981–90 sub-periods are reported in Table 3—the fixed effect estimates are not qualitatively different from those of the variance component. Table 4, on the other hand, contains the fixed-effect estimates (which are also essentially the same as for the variance component) for the four regions viz: sub-Saharan Africa, Asia, Western. Hemisphere, and others (comprising North Africa, Middle East, and Europe). The main reason for presenting the estimates contained in Tables 3 and 4, in addition to those in Table 2, is to see whether the direction of the estimated effect of each regressor would be preserved by sub-dividing the entire data sets temporally and regionally.

Table 2.Estimates for All 81 Sample Countries 1961–90
EXFTexptexpt (t-1)GOVTgovtDEVBNXdevbnkdevbnk (t-1)FINDEPTHfindepthfindepth (t-1)INFINF (t-1)RINTEXCHOISTEXCHD1ST (t-1)Trend VariableRegional Dumry Variables for:R2 (& DW)
AfricaAsiaW. H M
ESTIMATES WHEN FINANCIAL DEPTH IS MEASURED AS GROWTH/LEVEL OF STOCK OF LIQUID LIABILITIES/GDP RATIO (N-1631; 79 Countries)
0.1120.119-0.221-0.5110.160-0.3970.199-0.031-0.018-0.006-0 0510.014-0.03613.4
(2 4)(2.6)(-3.5)(-7.2)(2.3)(-8.8)(4.1)(-2.0)(-1.2)(-7.3)(-1.8)(0.4)(-1.3)(1.48)
-0.655-0.135-0.2250.077-0.031-0.024-0.006-0.100-0.033-0.1008.6
0.058(-3.6)(-2,3)(5.6)(1.8)(-2.0)(-1.5)(-5.0)(-2.9)(-0.9)(-3.1)(1.53)
1.2
ESTIMATES WHEN FINANCIAL DEPTH IS MEASURED AS THE CHANGE IN REAL STOCK OF LIQUID LIABILITIES/REAL GDP SAVINGS RATIO
Estimates When Real Interest Rate and General Government Consumption Expenditure are Included as Regressors (N-1135; 71 Countries)
0.1430.242-0.5262.0524.343-0.4860.3020.175-0.0350.008-0.008-0.0090.0010.00814.7
(2.7)(4.5)(-6.8)(2.2)(5.0)(-3.8)(3.6)(1.4)(-2.1)(0.5)(-6.0)(-0.3)(0.02)(-0.2)(1.55)
0.1200.262-0.4562.8543.2130.468-0.0350.014-0.00g-0.0070.002-0.03013.2
(2.3)(4.9)(-6.0)(3.1)(4.0)(5.5)(-2.1)(0.8)(-7.1)(-0.2)(0.05)(-0.9)(1.54)
Estimates When Development Bank Lending Variables are Included a. Regressors (N-890: 48 Countries).
0.1490.066-0.202-0.1654.3623.692-0.1650.035-0.022-0.010-0.0110.009-0.0090.00517.1
(2.8)(1.2)(-4.3)(-3.7)(4.0)(3.3)(-3.7)(0.7)(-1.2)(-0.5)(-7.7)(0.2)(-0.2)(0.1)(1.39)
-0.124-0.4903.8663.216-0.1440.042-0.022-0.011-0.008-0.034-0.006-0.01113.1
(-2.0)(-3.1)(2.5)(2.8)(-3.3)(0.8)(-1.2)(-0.6)(-5.6)(-0.8)(-0.1)(-0.3)(1.46)
Estlmstes When Central Government Tc6.1 Expenditure 1. Included as Regressors (N-1426; 77 countries)
0 1450.176-0.1844.4021.995-0.1650.051-0.030-0.018-0.008-0.0020.029-0 00312.3
(3.1)(3.7)(-3.4)(5.6)(2.4)(-3.7)(1.0)(-1.9)(-1.2)(-7.2)(-0.1)(0.9)(-0.1)(1.42)
-0.084-0.2104.1412.377-0.1650.064-0.033-0.026-0.007-0.0290,005-0.03610.8
(-1.6)(-2.4)(5.2)(2.8)(-3.7)(1.8)(-2.1)(-1.7)(-6.0)(-0.9)(0.1)(-1.3)(1.47)
Notes: (1) t-valuirs are the figures in parentheses. At significance levels of 1 percent: 5 percent; and 10 percent a parameter estimate I. statistically different from sfro if its t-value in absolute term is up to 2.6; 2.0 and 1.6 respectively.

Variance component method is used in deriving the Panel Data Estimates.

In the /eat column, the figures presented are the R2 on top and DM statistic values b.neath in parentheses.

The dependent variables is the IOCR or incremental output-capital ratio while the notationn for regressors are as follows: EXPT - export/GDP ratio; expt - growth of EXPT; GOVT - government expenditure/GDP ratio; govt - growth of GOVT; DEVBNK - development bank lending/GDP ratio: devbnk - growth of DEVBNK: FINDEPTH - stock of banking system’s liabilities/GDP ratio: findepth - growth of FINDEPTH in the first four equation estimates; and findepth - change in real value of the banking system’s liquid liabilities/real GDP ratio in other equations; INF - inflation rate; RINT real interest rate; EXCHDIST - exchange rate distortion: and N - total number of observation.

Notes: (1) t-valuirs are the figures in parentheses. At significance levels of 1 percent: 5 percent; and 10 percent a parameter estimate I. statistically different from sfro if its t-value in absolute term is up to 2.6; 2.0 and 1.6 respectively.

Variance component method is used in deriving the Panel Data Estimates.

In the /eat column, the figures presented are the R2 on top and DM statistic values b.neath in parentheses.

The dependent variables is the IOCR or incremental output-capital ratio while the notationn for regressors are as follows: EXPT - export/GDP ratio; expt - growth of EXPT; GOVT - government expenditure/GDP ratio; govt - growth of GOVT; DEVBNK - development bank lending/GDP ratio: devbnk - growth of DEVBNK: FINDEPTH - stock of banking system’s liabilities/GDP ratio: findepth - growth of FINDEPTH in the first four equation estimates; and findepth - change in real value of the banking system’s liquid liabilities/real GDP ratio in other equations; INF - inflation rate; RINT real interest rate; EXCHDIST - exchange rate distortion: and N - total number of observation.

Table 3.Estimates For Different Sub-Periods
PERIODEXPTexptexpt (t-1)GOVTgovtDEVBNKdevbnkdevbnk (t-1)findepthfindepth (t-1)INFINF (t-1)RINTEXCHDISTEXCHDIST (t-1)Trend VariableR2DWNo. of Observations (& Countries)
Estimates fo Equations that Include General Government Consumption Expenditure and Reel Interest Rate
Pre-

1974,
0.113

(0 8)
0.160

(1.1)
-1.008

(-5.0)
6.107

(1.6)
6.394

(1.7)
-0.615

(-1 1)
0.548

(1.4)
0.775

(1.5)
-0.086

(-2 4)
0.012

(0.3)
-0,004

(-0,7)
21.11.40256

(39)
1974-

1980
0 208

(2 2)
0.154

(1 7)
-0.574

(-4 2)
0.330

(0.2)
6.438

(4.9)
0 026

(0.1)
0.207

(1.3)
0,508

(1.9)
-0.003

(-0 1)
0.042

(1.4)
-0.019

(-2.4)
18.41.37376

(63)
Post

1980.
0.038

(0 5)
0 298

(3.7)
-0.202

(-1.7)
1.390

(1.1)
0.176

(0.1)
-0.387

(-2 6)
0.214

(1.9)
0,154

(1.0)
-0.020

(-0 8)
-0.005

(-0.2)
0.016

(3.2)
20.01.52420

(68)
Estimates of Equations that Include the Sire of Development Banking
Pre-

1974.
0.196

(1 2)
0.074

(0.4)
-0.168

(-1.9)
-0.232

(-2.9)
3.816

(1.0)
1.941

(0.5)
-0.782

(-2 5)
0.038

(0.1)
-0.011

(-0.3)
-0.025

(-0.6)
-0,004

(-0.5)
14.21.09247

(37)
-0.190

(-1.1)
-0.551

(-1.2)
7.448

(1.9)
0.271

(0.1)
-0.440

(-1 4)
-0.134

(-0.3)
-0.003

(-0.1)
-0.026

(-0.6)
-0 005

(-0.7)
7.21.29247

(37)
1974-

1980.
0.412

(4.3)
0.232

(2.4)
-0.306

(-3.4)
-0.010

(-0.1)
3.989

(2.1)
7.898

(4.1)
-0.238

(-1 7)
0.195

(1.31
0.011

(0.37
0.021

(0.7)
-0.009

(-1.2)
20.81.24293

(45)
0.066

(0 6)
-0.663

(-1.5)
3.064

(1.5)
6.426

(3.3)
0.020

(0.1)
0.024

(0.2)
-0.001

(-0.02)
0.011

(0.3)
-0.011

(-1.4)
9.71.32293

(45)
Post-

1980.
-0.012

(-0 2)
0.017

(0.2)
-0.062

(-0.7)
-0.229

(-2.61
2.850

(1.8)
0.026

(0.02)
-0.137

(-2 8)
0.027

(0.5)
-0.028

(-1.0)
0.017

(0.6)
0.014

(2.5)
9.71.41350

(45)
-0.210

(-2 2)
-0.244

(-1.2)
4.298

(2.6)
0.408

(0.2)
0.419

(-3 2)
0.066

(1.2)
-0.026

(-0.9)
0.020

(0.7)
0.018

(3.3)
13.21.44350

(45)
Estimates of Equations the. Include Central Government Total Expenditure
Pre-

1974:
0.156

(1.41
0.247

(2.2)
-0.323

(-3.1)
3.429

(1.3)
-0.166

(-0.1)
-0.667

(3.8)
0.284

(1.2)
0.012

(0.4)
-0.027

(-0.9)
0.005

(0.9)
11.71.43426

(57)
-0.010

(-0.1)
-0.712

(-3.0)
3.519

(1.3)
0.680

(0.3)
-0.564

(3.3)
0.323

(1.5)
0.015

(0.5)
-0.037

(-1.2)
0.006

(1.1)
10.41.48426

(57)
1974-

1980:
0.246

(2.9)
0.105

(1.2)
-0.217

(-2.2)
3.267

(2.6)
4.634

(3.5)
-0.244

(-2.1)
0.124

(1.1)
-0.024

(-0.9)
0.044

(1.5)
-0.010

(-1.4)
11.81.29452

(73)
-0.063

(-0.7)
-0.217

(-1.2)
3.269

(2.6)
4.461

(3.4)
-0.117

(-1.0)
0.040

(0.4)
-0.025

(-0.9)
0.034

(1.2)
-0.008

(-1.2)
9.31.32452

(73)
Post-

1980:
0.061

(0.8)
0.187

(2.4)
-0.069

(-0.7)
4.584

(3.5)
-0.684

(-0.5)
-0.095

(-1.6)
0.029

(0.4)
-0.036

(-1.3)
-0.026

(-0.9)
-0.009

(1.7)
10.91.25544

(73)
-0.187

(-2.6)
-0.023

(-0.2)
5.345

(4.0)
0.822

(0.6)
-0.135

(-2.5)
0.084

(1.4)
-0.046

(-1.7)
-0.030

(1.1)
0.012

(2.3)
11.71.34544

(73)
Notes, (1) t-values are the figures in parentheses. At significance levels of 1 percent; 5 percent; and 10 percent, a parameter estimate is statistically different from zero of its t-value in absolute term in up to 2.6; 2.0 and 1.6 respectively.

In the last column, the figures presented are the total number of observations while the number of countries giving riae to the totals are the ones in the parentheses.

The dependent variable is the IOCR or incremental output-capital ratio while the notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVBNK; findepth = change in real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rata; RINT = real interest rate and EXCHDIST = exchange rate distortion.

Notes, (1) t-values are the figures in parentheses. At significance levels of 1 percent; 5 percent; and 10 percent, a parameter estimate is statistically different from zero of its t-value in absolute term in up to 2.6; 2.0 and 1.6 respectively.

In the last column, the figures presented are the total number of observations while the number of countries giving riae to the totals are the ones in the parentheses.

The dependent variable is the IOCR or incremental output-capital ratio while the notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVBNK; findepth = change in real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rata; RINT = real interest rate and EXCHDIST = exchange rate distortion.

Table 4.Estimates for Different Regional Groups of Countries
RegionEXPTexptexpt (t-1)GOVTgovtDEVBNKdevbnkdevbnk (t-1)findepthfindepth (t-1)INFINF (t-11RINIEXCHDISTEXCHDIST (t-1)Trend VariablesR2DWNo. of Observations (& Countries)
Estimates of Equation that Include General Government Expenditure and Real Interest Rate
Sub-Saharan Africa0.274

(3.0)
0.320

(3.4)
-0.238

(-1.8)
6.322

(2.2)
3.455

(1.2)
0.174

(0.2)
-0.042

(-0.2)
0.546

(0.7)
-0.069

(-2.0)
0.007

(0.2)
-0.007

(-1.4)
12.61.80372

(24)
Asia0.262

(2.7)
0.179

(1.9)
-0.384

(-3.1)
2 833

(1.8)
0.437

(0.3)
-1.360

(-3.5)
0.259

(1.3)
-0.595

(-1.7)
0.000

(0.0)
0.000

(0.0)
-0.004

(-2.4)
19.91.59233

(13)
Western Hemisphere0.009

(0.1)
0.166

(1.8)
-0.588

(-4.3)
5.418

(3.1)
4.937

(2.6)
-0.845

(-4.6)
0.586

(4.4)
-0.262

(-1.4)
-0.001

(-0.02)
0.003

(0.1)
-0.012

(-5.1)
29.81.23323

(18)
N. Africa, Middle East. and Europe0.010

(0.1)
0.201

(1.5)
-1.330

(-6.2)
1.015

(0.6)
5.362

(3.8)
-0.491

(-1.5)
0.319

(1.5)
0.786

(2.8)
-0.023

(-0.5)
0.075

(1.8)
-0.013

(-3.3)
30.91.47222

(17)
Estimates of Equations that Include the Size of Development Banking
Sub-Saharan Africa0.418

(2.7)
0.352

(2.4)
-0.293

(-2.3)
-0.319

(-2.7)
2.917

(0.7)
-0.290

(-0.1)
-0.984

(-2.8)
0.324

(0.9)
-0,037

(-0.7)
0.003

(0.1)
-0.007

(-1.6)
21.81.41176

(11)
0.826

(1.4)
0.229

(0.1)
2.874

(0.6)
-3.984

(-0.9)
-0.582

(-1.7)
0.473

(1.4)
-0.025

(-0.5)
-0.016

(-0.3)
-0.006

(-1.4)
8.81.54176

(11)
Asia-0.051

(-0.6)
0.033

(0.4)
-0.158

(-1.8)
-0.019

(-0.2)
4.662

(3.0)
0.837

(0.4)
-0.120

(-0.5)
-0.352

(-1.5)
0.008

(0.3)
0.021

(0.9)
-0.010

(-5.7)
23.21.66160

(7)
0.006

(0.1)
-1.170

(-3.3)
3.962

(2.5)
1.266

(0.7)
-0.171

(-0.8)
-0.341

(-1.5)
0.018

(0.8)
0.027

(1.2)
-0.007

(-3.5)
28.41.76160

(7)
Western Hemisphere0 016

(0.3)
-0.054

(-0.9)
-0.073

(-1.6)
0.126

(-2.7)
6.993

(4.7)
2.989

(1.9)
-0.120

(-3.3)
0.052

(1.2)
-0.011

(-0.5)
-0.044

(-2.2)
-0.011

(-8.8)
28.91.34398

(20)
0.199

(1.0)
-0.953

(-2.9)
6.537

(4.6)
2.237

(1.5)
-0.121

(-3.4)
0.046

(1.1)
-0.013

(-0.6)
-0.044

(-2.2)
-0.010

(-5.5)
28.91.40398

(20)
Africa. Mid-East, and Europe0.092

(0.7)
-0.039

(-0.3)
-1.270

(-7.0)
-0.084

(-0.6)
4.304

(1.7)
7.831

(3.4)
-0.771

(-2.4)
-0.099

(0.3)
-0.051

(-1.1)
0.038

(0.8)
-0.017

(-4.0)
39.01.65184

(11)
0.762

(2.3)
-0.430

(-0.8)
-0.538

(-0.2)
3.927

(1.5)
-0.410

(-1.1)
-0.813

(-2.4)
-0.065

(-1.3)
0.036

(0.7)
-0.012

(-2.4)
19.01.71184

(11)
Notes: (1) t-values are the figures in parentheses. At significance levels of 1 percent, 5 percent and 10 Percent, a parameter estimate is statistically different from zero of its t-value in absolute term as up to 2.6; 2.0 and 1.6 respectively.

In the last column, the figures presented are the total number of observations while the number of countries giving rise to the totals are the ones in the parentheses.

The dependent variables is the IOCR or incremental output-capital ratio while the notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVENK; findepth = change In real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rate; RINT = real interest rate and EXCEDIST = exchange rate distortion.

Notes: (1) t-values are the figures in parentheses. At significance levels of 1 percent, 5 percent and 10 Percent, a parameter estimate is statistically different from zero of its t-value in absolute term as up to 2.6; 2.0 and 1.6 respectively.

In the last column, the figures presented are the total number of observations while the number of countries giving rise to the totals are the ones in the parentheses.

The dependent variables is the IOCR or incremental output-capital ratio while the notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbnk = growth of DEVENK; findepth = change In real value of the banking system’s liquid liabilities/real GDP ratio; INF = inflation rate; RINT = real interest rate and EXCEDIST = exchange rate distortion.

In all cases, it is the t-values that are presented in the parentheses below the parameter estimates. The intercept terms are not reported for brevity.

3. Evaluation of the results

As can be seen from Tables 2 through 4, the DW statistics values generally do not fall short of 2, to the extent of indicating the existence of a positive linear serial correlation of the residuals. 10/ However, the explanatory power of the equations is generally low, as is seen from the R2 values. With these brief general comments, the direction of the effect of each explanatory variable will be examined in turn in the subsequent paragraphs. The variables are discussed chronologically as follows: exports, government expenditure, size of the development banking sector, financial depth, the inflation rate, the real interest rate and exchange rate variables.

In all cases, the coefficients of the growth of the export-GDP ratio (its contemporaneous and lagged values) are positive. These positive coefficients are also statistically significant in most cases. 11/ The same also applies (to a lesser extent though) to the export-GDP ratio expressed in the level-form. Thus, there is sufficient evidence to suggest that export expansion and, hence, export-oriented policy promotes efficiency of resource utilization. (Whether it increases the ivestment ratio and/or promotes economic growth will be examined briefly below).

Concerning the size of the public sector, the coefficients of the share of government expenditure in GDP (in both the level and growth-rate form) are negative and also statistically significant in all cases. This is so irrespective of whether it is the central government total expenditure-GDP ratio or the general government consumption expenditure-GDP ratio that is considered. In effect, the size of the public sector, as proxied by the share of government expenditure in the GDP, is found to have a negative effect on the efficiency of resource allocation. (Again, the direction of its effect on economic growth will be looked into briefly below).

The direction of the effects of the development banking credit-GDP ratio, in both the level and the growth-rate forms, is the same as that of the government expenditure-GDP ratio. In other words, directed credit policy (as measured by the magnitude of development bank lending) is found to hamper the efficiency of resource use. 12/

While the effects of the three macroeconomic policy variables examined so far are in line with expectations, or, at least, do not contradict them, the results are not so robust when alternative measures of financial depth are employed. Against the expected positive impact financial depth should have on economic efficiency, negative effects are detected when financial depth is measured by the ratio of the stock of the banking system’s liquid liabilities to GDP or the growth rate of this ratio. As is seen in Table 2, the coefficients of the contemporaneous values of the level and growth of the stock of liquid liabilities-GDP ratio are negative in those equations where they feature and are statistically significant in virtually every case. (Although the coefficients of the one-period lagged value of the growth-rate of the depth ratio are positive, their magnitudes and t-values fall short of the corresponding coefficients of their contemporaneous value. Thus, the overall effect of the growth of the depth ratio is still negative—lagged values beyond one-period have very insignificant coefficients and therefore are not retained in the reported estimates.) However, when financial depth is alternatively measured as the ratio of change in the real value of liquid liabilities, significant positive effects on efficiency are observed. This is seen in the Tables 2-4 estimates where the coefficients of the contemporaneous and one-period lagged values are generally significantly positive. 13/ Therefore, Tables 3 and 4 report only those estimates where this measure of financial depth is featured. Thus, we are led to infer from the results described in this paragraph that the common definition or measure of financial depth as the stock of liquid liabilities-GDP ratio is not appropriate or that financial intermediation may not promote economic growth through improvement of economic efficiency. However, the balance of the evidence supports the former inference, given the observed strong positive effects of the contemporaneous and lagged values of the alternative measure of financial depth.

The inflation rate is found to have an immediate negative impact on the incremental output-capital ratio. This immediate effect is, however, partially moderated or offset later, although the negative effect remains.

These inferences are based on the coefficients of the contemporaneous value of the inflation rate, which are significantly negative in all equations. On the other hand, the (absolutely smaller) coefficients of the one-period lagged value are generally positive and, in many cases, statistically significant. Thus, the inflation rate is found to hamper the efficiency of resource utilization through channels other than via the financial variables, such as the real interest rate, the real exchange rate or financial depth.

The real interest rate is found to have positive coefficients that are also statistically significant in most cases. 14/ In the few cases where negative coefficients are recorded, they are not statistically significant. Thus, subject to a minor qualification concerning the choice of data set in estimating those equations where real interest rate features as a regressor, there is reliable evidence showing that financial liberalization has a positive effect on the efficiency of resource use through an increase in real interest rate.

The real exchange rate policy distortion enters the equations with postulated negative coefficients in virtually all cases. This applies to both the contemporaneous and the one-period lagged values. These coefficients are also statistically significant in many cases. In those few equations where the coefficients are positive, they are not statistically significant. Thus, the evidence shows that a policy of not lowering the real value of domestic currency following a current account deficit or of not appreciating it following a surplus distorts the real value of the currency and hence adversely affects the efficient allocation of resources.

At this juncture, we move to the effects of regional dummies and trend variables. It can be seen from the estimates in Tables 2 and 4 that there is a general downward trend in the values of the incremental output-capital ratio. Table 3 estimates shed further light on this by showing that the downward trend prevailed only up to the end of the 1970s and that the 1980s are characterized by an upward trend. Also, as compared with the “conglomerate” region of North Africa, Middle East, and Europe, the incremental output-capital ratio is lower in the sub-saharan Africa, and Western Hemisphere but is essentially the same in Asia—the coefficients of the regional variables are negative (and sometimes statistically significant) for sub-Saharan Africa and Western Hemisphere but are not significant for Asia.

Now, the choice of data set to be employed in estimating the equations containing the real interest rate as a regressor, as referred to above, is discussed. It was found that if we include all observations for which figures on real interest rate are available, either an insignificant or a perverse effect of the real interest rate on efficiency of resource utilization is recorded. In fact, there is an implausible positive (instead of the expected negative) correlation of coefficients of approximately 1.000 between the real and the nominal interest rate and of 0.393 between the real interest rate and the inflation rate. It was also found that by ignoring the outliers (real interest rate corresponding to unusually high nominal interest rates in such countries as Brazil, Argentina, Bolivia, Chile and Israel during some periods), plausible correlation coefficients between the real interest rate and each of the nominal interest and inflation rates were now observed and that the coefficients of the real interest rate in the estimated equations assumed positive values. The more the outlier values were ignored, the more significantly positive the coefficients of the real interest rate in the equations became. 15/ A cut-off figure of below 100 percent of the annual nominal interest rate was eventually adopted in estimating the equations (and in deriving the correlation coefficients between the real interest rate and other variables as presented in Table 1). It is probably these outliers that account for the less than complete unanimity concerning the role of the real interest rate on resource allocation and economic growth in the empirical literature—e.g., as pointed out in Section V, while many authors reported a positive role, others like Khatekhate (1988) reported an absence of evidence in support of a positive role.

As mentioned earlier we shall make brief comments on the possible effects of the regressors employed in the study, on the capital formation ratio and on economic growth—having considered their effects on the incremental output-capital ratio. The equation estimates reported in Table 5 below form the basis for this review. The equations contain arguments that are similar to those presented in Tables 2-4. In addition, the real GDP growth equations contain labor force growth and an investment ratio that is in line with the usual neoclassical specification—the two are proxied by the population growth and a one-period lagged value of the investment ratio. The investment ratio equations also contain real GDP growth—proxied by its lagged value—as an argument to capture the accelerator effect. 16/

Table 5.Economic Growth and Inwtwent ftatio Equity on Estimates
POPINV

(t-1)
(t-1)
EXPTexptexpt

(t-1)
GOVTgovtDEVBNKdevbnkdevbnk

(t-1)
findepthfindepth

(t-1)
INFINF

(t-1)
RINTEXCHDISTEXCRDIST

(t-1)
TrendR2
0.169

(1.9)
0.020

(1.0)
0.014

(1.7)
0.021

(2.7)
-0.093

(-9.7)
1.090

(6.0)
0.640

(4.7)
-0.083

(-5.6)
0.026

(3.4)
-0.006

(-2.1)
-0.002

(-0.6)
-0.002

(-9.7)
16.2

(1.351
0.120

(1.3)
-0.001

(-0.1)
0.085

(2.5)
-0.309

(-6.9)
0.677

(6.4)
0.634

(4.6)
-0.034

(-4.5)
0.017

(2.2)
-0.005

(-1.9)
-0.002

(-0.7)
-0.001

(-6.3)
18.9

(1.37)
0.077

(0.8)
0.027

(1.0)
0.011

(1.2)
0.037

(3.9)
-0.104

(-7.7)
1.063

(6.0)
0.694

(4.6)
0.083

(5.2)
-0.004

(-1.3)
0.004

(1.5)
-0.002

(-7.4)
15.9

(1.63)
0.012

(0.1)
0.027

(0.8)
0.022

(2.3)
0.003

(0.3)
-0.087

(-5.2)
-0.033

(-3.9)
1.261

(5.9)
0.645

(3.6)
-0.022

(-2.6)
0.008

(0.8)
-0.003

(-1.0)
-0.001

(-0.2)
-0.002

(-8.7)
19.2

(1.50)
0.080

(0.6)
-0.025

(-0.8)
0.064

(2.6)
-0.158

(-3.2)
0.933

(4.4)
0.633

(2.9)
-0.020

(-2.3)
0.007

(0.7)
-0.003

(-0.9)
-0.0001

(-0.04)
-0.002

(-6.6)
14.8

(1.47)
INV0.060

(2.5)
0.017

(2.7)
0.007

(1.2)
0.379

(2.6)
0.515

(3.4)
-0.011

(-1.6)
0.013

(1.9)
-0.002

(-1.7)
13.9

(2.07)
INV0.069

(3.0)
0.156

(2.4)
0.432

(3.0)
0.514

(3.4)
-0.012

(-1.9)
0.013

(1.9)
-0.002

(-1.9)
13.8

(2.07)
INV0.047

(2.9)
0.476

(5.2)
0.342

(3.0)
-0.01$

(-2.5)
-0.002

(-2.1)
15.2

(2.05)
INV0.083

(4.3)
0.414

(3.4)
0.203

(1.9)
0.051

(4.2)
-0.003

(-2 6)
15.7

(2.01)
Notes, (1) t-values are the figures in parentheses. At a significance levels of 1 percent, 5 percent, and 10 percent, a paranoias estimate is statistically different from sere of it. t-value in absolute term is up to 2.6, 2.0 and 1.6 respectively.

In the last column, the figures presented are the R2 value. while the DW statistic values are the ones in the parentheses.

The dependent variable is the (total) real GDP growth (

) in the first 5 equation estimates and investment/GDP ratio (INV) in the last 4 while notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government consumption expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbok = growth of DENIM; findepth = change in real vave of the banking system’s liquid lisbilitisa/real GDP ratio; INF = inflation rate; RIOT real interest rate; EXCBDIST = exchange rate distortions; and POP = population or labor force growth.

The investment ratio equation was estimated through the Cochrane-Orcbut Iterative technique of correcting for autocorrelation.

Notes, (1) t-values are the figures in parentheses. At a significance levels of 1 percent, 5 percent, and 10 percent, a paranoias estimate is statistically different from sere of it. t-value in absolute term is up to 2.6, 2.0 and 1.6 respectively.

In the last column, the figures presented are the R2 value. while the DW statistic values are the ones in the parentheses.

The dependent variable is the (total) real GDP growth (

) in the first 5 equation estimates and investment/GDP ratio (INV) in the last 4 while notations for regressors are as follows: EXPT = export/GDP ratio; expt = growth of EXPT; GOVT = government consumption expenditure/GDP ratio; govt = growth of GOVT; DEVBNK = development bank lending/GDP ratio; devbok = growth of DENIM; findepth = change in real vave of the banking system’s liquid lisbilitisa/real GDP ratio; INF = inflation rate; RIOT real interest rate; EXCBDIST = exchange rate distortions; and POP = population or labor force growth.

The investment ratio equation was estimated through the Cochrane-Orcbut Iterative technique of correcting for autocorrelation.

As can be seen from estimates in Table 5, the directions of the effects of the various regressors on economic growth are the same as on the incremental output-capital ratio. Also, while the size of development banking is found to promote capital formation, its overall negative effect on economic growth suggests that its adverse effects on economic efficiency outweigh its positive impact on the quantity of capital formation. On the other hand, the effects of the inflation rate on economic growth through the efficiency and quantum of capital are found to reinforce one another. That is, they have negative effects on both incremental output-capital ratio as reported in Tables 2 to 4, and on the investment ratio as is seen in Table 5 (especially in the investment equation containing contemporaneous but no lagged value of the inflation rate). Similarly, the effects of the real interest rate on growth via the channels of the incremental output-capital and the investment ratios support each other—positive effects on both and, hence, on economic growth.

IV. Summary and Conclusions

Given the close relationship between the efficiency of resource utilization and economic growth and the relative econometric ease of identifying the effects of relevant factors on the former, this study has examined the effects of various macroeconomic policy variables on the economic efficiency. The study is based on the annual panel data for 81 developing countries over 1961 and 1990 period.

In particular, a case was made for indirect evaluation of the effects of policy variables on economic growth through a direct evaluation of their effects on the quantum of resources and efficiency with which the resources are used. The index of efficiency was defined as the incremental output-capital ratio and a list of macroeconomic policy variables and their effects on efficiency were examined. The highlights of the empirical results are as follows:

(a) The export-GDP ratio and, particularly, the growth of this ratio have positive effects on economic efficiency, suggesting that export-oriented policies promote efficiency.

(b) The government expenditure-GDP ratio and the growth of this ratio have negative effects on economic efficiency, suggesting that the size of the public sector is inversely related to efficiency.

(c) The size of development banking in relation to the GDP and its growth have negative effects on efficiency, suggesting that directed credit policies hinder economic efficiency.

(d) While a conventional measure of financial depth (computed as both the level and the growth of the ratio of the banking system’s stock of liquid liabilities to the GDP) is not found to have a positive association with economic efficiency. An alternative measure, computed as the ratio of the flow of the real value of liquid liabilities to real GDP, is found to have a very strong and consistent positive association with efficiency so that the balance of the evidence still points to an efficiency-promoting effect of financial deepening.

(e) The inflation rate has a negative effect on the economic efficiency, apart from the negative effect it may have through such financial variables as the real exchange rate and the real interest rate.

(f) The real interest rate improves economic efficiency.

(g) Economic efficiency is hampered by a policy of not depreciating the real value of domestic currency in response to a current account deficit or not appreciating it following a current account surplus.

(h) The directions of the effects of the above factors on efficiency are also the same as their directions on economic growth. Sometimes, particularly, in the case of development bank lending, the effects on growth via efficiency tend to be moderated or partly neutralized by opposing effects through the channel of the capital formation ratio. There are some cases, especially as it affects the real interest and inflation rates, where the effects on growth via the two channels reinforce each other.

As an underpinning of this study and a support to possible future studies, the next section provides a brief survey of the literature in the areas covered in this paper.

V. Brief Review of the Existing Studies

In the context of developing countries, the literature seems to have given prominence to the factors determining the quantum of resources, while relatively little attention has been given to their efficient utilization. For example, a sizable number of theoretical and empirical studies exist on the effects of policy and nonpolicy factors on capital accumulation or investment expenditure but there are relatively few similar studies on the causes of efficiency of capital utilization.

As the preceding study has focused on the effects that macroeconomic policies have on economic efficiency, the studies surveyed here concern distortions—particularly those induced by macroeconomic policies. The few studies reviewed here are grouped into theoretical and empirical studies, which are taken up in subsections 1 and 2 below. It should be noted, however, that this segmentation is far from being rigid or absolutely precise since some of the studies either belong to both categories or do not exactly fit into either of them.

1. Theoretical studies on policy-induced distortions and economic performance

Theoretical studies that address the effects of economic policies on resource allocation can be traced back to the early 1960s and even before that time. The genesis and a survey (up to the early 1970s) of such studies are contained in Magee (1973). What is particularly noteworthy is that such theoretical studies were then being conducted mainly within the framework of international trade theory so that they are microeconomic in orientation.

The tempo of such studies continued during the 1970s with the works of Bhagwati (1978), Balassa (1971), Krueger (1978), and so on. Also, during that decade, theoretical studies on the effects of policy distortions were no longer necessarily limited to international trade theory but were being extended or, rather, shifting to essentially macroeconomic spheres—especially the role of financial liberalization and the size of public sector. As these aspects have more relevance to the present study, the next few paragraphs will address them.

During the decade of 1960s, a lot of complementary propositions on the positive role of financial development on economic growth came into existence. These include the writings of Gurley and Shaw (1960), Goldsmith (1960, 1969), Patrick (1966) and Porter (1966). The central idea that pervades these studies is that financial development can bring about or is even a precondition for economic growth by increasing the investible surplus and by channeling resources to higher-yielding investments. This idea was later taken up and further developed in the celebrated writings of McKinnon (1973) and Shaw (1973), leading to the famous McKinnon-Shaw hypothesis that financial liberalization in the form of a high or rather a market-clearing real rate of interest is a prerequisite for economic growth. According to the classification adopted by Gupta (1987), while writers like Goldsmith (1966, 1969) belong to what he calls “financial structuralists” (by believing that financial intermediation affects economic growth directly quite apart from any effects on the real rate of interest), McKinnon and Shaw belong to the “financial repressionists” school (by emphasizing only price variables such as the real rate of interest and the real exchange rate as the channel through which financial liberalization promotes growth or by which financial repression retards growth). Before the McKinnon-Shaw hypothesis, it was widely believed that the returns on financial assets had negative effects on economic growth. It was McKinnon and Shaw who provided a theoretical framework that contradicted this view, e.g., by positing that financial assets are complements to instead of substitutes for physical assets. The initial McKinnon-Shaw framework has subsequently been refined by such writers as Galbis (1977), Fry (1982), Mathieson (1980), and Kapur (1976), but without changing the essence of the hypothesis. Whether they belong to the “structuralist” or to the “repressionist” schools, the general stance of these writers is that financial policies or reforms promote economic growth through the two main channels of increasing the amount of funds for investment, and by increasing the efficiency of resources by directing new and existing funds toward higher-yield investments.

There are some theoretical studies on the effects the size of the public sector has on efficiency and growth. One school of thought is of the view that (following the Keynesian belief about the positive effect expansion of government expenditure has on economic growth) an expansion in the size of the public sector promotes growth, while the opposing school posits that by hampering economic efficiency, an expansion in the size of public sector retards growth. These alternative propositions and the model framework on which they are based are presented and/or reviewed in Barro (1990), Landau (1986), Easterly and Wetzel (1989), among others.

Related to the theoretical studies on the effects of the size of the public sector are the studies on rent-seeking. According to Gallagher (1991, p. 55), the concept of rent-seeking “focuses on actions that seek to alter, enforce, maintain or circumvent government policies which entail transfers of income.” Such policies are identified to include government-sponsored monopoly and monopsony, credit and foreign exchange allocation schemes, tariffs, etc. The contention is that the resources being dissipated (in lobbying, cheating, bribing, evading, enforcing, etc.) as a result of such policies amount to wastes and deadweight economic loss. The theoretical frameworks of the effects of rent-seeking on efficiency and growth are discussed in Ekelund and Tollison (1981), Bhagwati (1982), Gallagher (1991), etc.

To conclude this discussion of the theoretical studies, it should be pointed out that virtually all the studies are based on an economic growth framework, particularly in the mathematical presentations of the models. Hardly any of the theoretical frameworks mathematically relates a measure of economic efficiency to those factors being postulated as efficiency determinants. As was pointed out in Section III, this has adverse implications when it comes to econometric testing of the effects of those factors on economic efficiency, as in the present study. Thus, we agree with Gallagher’s (1991, p. 31) submission that “the economic growth literature in some instances touches upon efficient use of capital, or resources in general, yet it does not explicitly incorporate allocative efficiency into estimated growth models.”

2. Empirical evidence on the effects of macroeconomic policies on efficiency

The initial microtheoretical exercises on the product and factor market distortions within the context of international trade theory have been subjected to some empirical tests, the pioneering ones being reviewed in Magee (1973) and some of the subsequent ones being those reported by Dougherty and Selowsky (1972), Floystad (1975) and de Melo (1977). As pointed out earlier, this microeconomic approach is of relatively little relevance to the present study and therefore need not be given further attention here.

Concerning the macro-oriented approach, empirical studies have reported on the role of trade policy or exports on economic growth and on the efficiency of resource utilization. Most of these studies have been reviewed by Moschos (1989), the World Bank (1987), and Jung and Marshall (1985), among others. In virtually all cases, a positive role played by export-oriented trade policy has been reported. However, there is a serious caveat concerning the methodology that is often adopted in conducting the evaluation. The initial approach of correlating export variables (e.g., real export growth or the share of exports in GDP) soon gives way to inclusion of real exports as an “input” in the usual neoclassical production to derive an estimable equation of the form:

where Y, I and X are the real values of GDP, gross capital formation and exports; L is the labor force; and a, b, c, and d are parameters and the dot (.) indicates the growth rate of the variable on which it appears.

By estimating equation (4), whether export-oriented policy has a positive effect on the resource allocation or not is to be inferred from the value of d. However, this approach does not tell whether the effect of export-oriented policy on economic growth is via an increase in the quantity of resources or through an increase in the efficiency of using the existing resources, or both. It was not until another framework was suggested by Feder (1982) that the channel of the effects of export-oriented policy on economic growth became clearer in the model to be estimated. The only postulated channels of the effects of exports on growth in Feder’s framework are through the differences between the productivities of resources employed in the export and non-export sectors and the external effects of export activities on the other sectors. Both channels are essentially variants of efficiency of resource use, without explicit reference to the channel of increase in the quantity of the resources. The estimated equation under this approach is as follows:

where e is the coefficient of the composite export variable (that is a product of the real export growth,

, and the share of export in the GDP, X/Y) and other notations are as defined earlier. The coefficient d now measures the external effects of the export sector on the remaining sectors, while the existence and extent of intersectoral difference in the efficiency of resources utilization are to be inferred from the coefficient e.

Despite the improvement that this formulation by Feder has on the earlier framework, it has a disadvantage of not being amenable to a situation where the effect of more than two sectoral policies (e.g., export and non-export sectoral policies) on efficiency are to be simultaneously evaluated. Also, it has some major defects in common with the earlier framework it attempts to build upon. One of the defects we consider sufficiently important to mention explicitly is the one that was empirically highlighted by Sheehey (1990). Sheehey started from the premise that the export variable that is often employed is an important component of the GDP in the national accounts and, as a result, there is an already built-in noncausal positive association between the measured export-oriented policy variable and real GDP growth. Further, the positive association that has often been detected between the measured export sectoral variable and economic growth is not unique anyway—the same association is likely to characterize economic growth and expansion in each of the other major components of the GDP. By employing cross-country data for 36 developing countries over the 1960–70 period, he was able to demonstrate that substitution of each of the other major components of GDP for the export variable in equations (4) and (5) above always leads to the same conclusion that shifting of resources to the production of each of them from the remaining sectors would enhance the overall efficiency in the economy! 17/ According to Sheehey, abstract, p. 111:

  • A strategy of export promotion has evolved into the new conventional wisdom. The alleged superiority of this strategy draws on an extensive list of empirical studies, an important strand of which consists of tests in cross-country format that use bi-variate correlations and/or production function-type regressions to demonstrate a strong positive relationship between exports and GDP growth. By showing that these same tests support the “promotion” of all major components of GDP, this note argues that these tests have no bearing at all on the export-promotion/import-substitution controversy.

Concerning the effects of the size of the public sector on the efficiency of resource utilization, a sizable number of studies have also been reported in the literature. Some of them and/or reviews of the notable ones can be found in Gemmel (1983), Kormendi and Meguire (1985), Grier and Tullock (1989), Landau (1986), Barro (1989), Easterly (1989), and Easterly and Wetzel (1989). The major problems with these kinds of studies are also generally the same as those concerned with evaluation of the effects of export-orientation, including the likelihood of an inbuilt noncausal association between the measured expansion in the size of the public sector and of economic growth. Also, the channels (whether via an increase in resources or through improved efficiency) of the effects of the size of the public sector cannot be inferred from the common empirical approaches—although the adoption of a Feder-type framework by authors like Ram (1987) to an evaluation of the size of the public (as opposed to the export) sector should go some way in rectifying this.

Related to the studies on the effects of the size of the public sector are some empirical studies that primarily address the impact of rent-seeking on resource allocation. 18/ However, such studies are too few. As has been rightly observed by Gallagher (1991, p. 52), “rent-seeking theories help explain why resources are not productively used, yet rent-seeking theorists have provided almost zero contribution to our understanding of growth. Finally, rent-seeking theories have been entirely too abstract.” One of the few empirical studies on the effects of rent-seeking on resource allocation can be found in Barro (1989). Also in a cross-sectional study of African countries, Gallagher (1991) computed a rent-seeking variable as a composite of international trade protection distortions, domestic agricultural price distortions, exchange rate disequilibrium, and below-zero real interest rate. He included it, together with the share of government spending in the GDP, as explanatory variables in the equation for efficiency of resource utilization (measured by the incremental output-capital ratio). Negative though weak effects of these variables on efficiency were detected.

In respect of the empirical studies on the effects of domestic macro-financial variables on resource allocation, such studies can be categorized into two main groups. They are those that evaluate the effects of the quantity variable (usually measured by the level of financial depth) and those that are concerned with the effects of the price variable (usually measured by the real interest rate). This classification corresponds to the two schools of thought identified by Gupta (1987) as “financial structuralists” and “financial repressionists”. The two groups are examined in turn in the next two paragraphs.

A recent review of most studies examining the relationship between financial depth and economic growth is contained in Balassa (1989). Some of these, like those reported by Fritz (1984) and Jung (1986), “merely” examined the direction of causation between economic growth and financial depth (or financial intermediation) while others like Jao (1978), Wai (1980), Gelb (1989) and Lanyi and Saracoglu (1983) adopted the approach of including financial depth as an explanatory variable in the economic growth equations. Yet, some others like McKinnon (1973) and IMF (1983) used a case-study approach in examining the relationship between financial depth and economic growth. In virtually all cases, a positive association between the measures of financial intermediation and economic growth are being detected. However, such approaches do not sufficiently equip us to jump to the conclusion that positive causation runs from financial intermediation to economic growth. Also, granted that positive causation runs from the financial variable to economic growth, we are not in a position to infer whether the channel is through an increase in the quantum of resources or through improvement of resource utilization efficiency or both. However, by employing the Feder-type framework to analyze the effects of the financial sector into channels relating to the financial-real sector productivity differential and the externality effects of the financial on the real sector (by replacing the export sectoral output in the Feder-type equation with the financial sectoral output) as in Odedokun (forthcoming), this problem may be catered for. But the problem of how to infer the direction of causation would still remain. Thus, the approach adopted by Gelb (1989) of direct evaluation of the effects of such financial variables on the efficiency of resource utilization would have much to commend it. Specifically, he included a measure of financial depth as an explanatory variable in an equation for the efficiency of resource utilization (measured by the incremental output-capital ratio) and detected a significant positive effect.

A sizable number of studies have also adopted the “financial repressionist” approach by examining the effects of the real interest rate on economic growth or by directly examining its effects on resource utilization. For instance, by classifying real interest rate regimes into positive, slightly negative and significantly negative, an IMF (1983) study reported a significant positive association between the real interest rate and economic growth. On the other hand, using a similar classification coupled with a non-parametric test methodology, Khatekhate (1988) was unable to detect any strong relationship between the real interest rate and growth. Earlier, Fry (1980) included the real interest rate in real GDP growth equations and a positive association was reported. Similar positive associations have been reported by Gelb (1989) and by Agarwala (1983) who included the real interest rate as one of the seven components of his computed index of distortion. However, for many reasons as rightly pointed out by Gelb (pp. 7–9), there is also the likelihood that it is the positive effect of economic growth on the real interest that is being identified in such studies, instead of the other way around. Also, granted that the causation is from real interest rate to growth, one is not in a position to infer whether the positive effect is brought about through increased investible resources or through improvement in the efficiency. The only studies that could overcome this problem would be those that include the real interest rate as a regressor in the equation for economic efficiency (usually, the incremental output-capital ratio) or the investment ratio, or both. As one of the variants of equation estimates reported, Fry (1979, 1981, 1984) and the Asian Development Bank (1988) included the real interest rate in the equation for incremental output-capital ratio (or its reciprocal) and found it had a positive effect on the economic efficiency (see Fry, 1988, p. 148 for a review of these studies). A similar equation estimate has been reported by Gelb (1989). However, all these estimates are based on a simple regression analysis that includes only the real interest rate as the regressor in the incremental output-capital ratio equation—which is equivalent to just a simple correlation analysis.

APPENDIX Data Sources and Techniques of Variable Measurement

All the data are from the IMF’s International Financial Statistics Yearbook, the relevant issues. The following explains how the variables are measured:

  • Incremental output-capital ratio: This is computed as change or first-difference of real GDP (line 99b.p of IFS) divided by the real value of gross capital formation, which is the nominal capital formation (lines 93e and 93i of IFS) deflated by the GDP deflator (line 99b.p of IFS).
  • Export-GDP ratio: This is nominal exports of goods and nonfactor services (line 90c of IFS) divided by nominal GDP (line 99b of IFS).
  • Government expenditure-GDP ratio: This is alternatively computed as the central government total nominal expenditure (line 82 of IFS or, in some cases, line 82z) divided by the nominal GDP and the general government (i.e., central, state and local governments) nominal consumption expenditure (line 91f of the IFS) divided by nominal GDP.
  • Development banking credit-GDP ratio: This is the ratio of end-of-year nominal value of the stock of claims of the development banking system on the private sector (line 42d of IFS) divided by nominal GDP.
  • Financial depth ratio: This is measured as the end-of-period nominal value of the wide money stock or M2 (line 351 of IFS), divided by nominal GDP. The second or alternative measure employed is also computed as the first difference of real value of M2 (i.e., M2 is first deflated by the GDP deflator before being first differenced) divided by real GDP. A more embracing measure of liquid liabilities of the whole banking sector (as opposed to that of the central bank and deposit money banks for which M2 stands) as provided in line 551 of IFS should have been employed, but few countries have data on this.
  • Real interest rate: It is the nominal interest rate minus the inflation rate. Where the data adequately exist, nominal deposit rate (line 601 of IFS) is selected. Otherwise, the nominal discount rate (line 60 of IFS) is chosen—with the proviso that its values are close to those of the deposit rate during the few periods when data exist for the deposit rate, otherwise the country is regarded as having no interest rate data.
  • Real exchange rate: This is the nominal domestic currency/SDR rate (line rb of IFS) divided by the domestic-foreign price ratio (derived by dividing the domestic GDP deflator by the industrial countries average GDP deflator as contained in the World Table section of IFS).

All the growth rates are computed as first-differences of the natural logarithm values of the corresponding level forms. Also, all variables are in pure fractions, not in percentages.

Countries and Years Covered by the Study

  • 1. Algeria, 1964–82
  • 2. Argentina, 1961–89
  • 3. Bahrain, 1975–88
  • 4. Barbados, 1972–86
  • 5. Benin, 1970–86
  • 6. Bhutan, 1980–88
  • 7. Bolivia, 1961–89
  • 8. Botswana, 1974–86
  • 9. Brazil, 1965–89
  • 10. Burkina Faso, 1972–85
  • 11. Burundi, 1971–90
  • 12. Cameroon, 1969–85
  • 13. Cape Verde, 1979–88
  • 14. Chile, 1961–88
  • 15. Colombia, 1967–90
  • 16. Costa Rica, 1961–90
  • 17. Cyprus, 1961–90
  • 18. Dominican Republic, 1963–90
  • 19. Ecuador, 1961–90
  • 20. Egypt, 1982–89
  • 21. El Salvador, 1961–90
  • 22. Fiji, 1966–88
  • 23. Ghana, 1961–88
  • 24. Guatemala, 1961–88
  • 25. Guyana, 1961–90
  • 26. Haiti, 1966–90
  • 27. Honduras, 1961–90
  • 28. Hungary, 1970–88
  • 29. India, 1961–89
  • 30. Indonesia, 1964–89
  • 31. Iran, 1964–88
  • 32. Israel, 1975–90
  • 33. Jamaica, 1961–88
  • 34. Jordan, 1961–90
  • 35. Kenya, 1967–89
  • 36. Korea, 1961–90
  • 37. Kuwait, 1970–88
  • 38. Liberia, 1964–86
  • 39. Libya, 1961–80
  • 40. Madagascar, 1961–85
  • 41. Malawi, 1961–90
  • 42. Malaysia, 1970–89
  • 43. Malta, 1961–88
  • 44. Mauritius, 1961–90
  • 45. Mexico, 1961–86
  • 46. Morocco, 1964–89
  • 47. Myanmar, 1967–88
  • 48. Nepal, 1976–89
  • 49. Nicaragua, 1961–87
  • 50. Niger, 1963–81
  • 51. Nigeria, 1961–89
  • 52. Oman, 1967–89
  • 53. Pakistan, 1961–90
  • 54. Panama, 1961–89
  • 55. Papua New Guinea, 1973–89
  • 56. Paraguay, 1961–89
  • 57. Peru, 1961–90
  • 58. Philippines, 1961–90
  • 59. Rwanda, 1968–90
  • 60. Saudi Arabia, 1968–89
  • 61. Senegal, 1967–86
  • 62. Sierra Leone, 1964–88
  • 63. Singapore, 1961–89
  • 64. South Africa, 1961–90
  • 65. Sri Lanka, 1961–90
  • 66. Swaziland, 1977–86
  • 67. Syria, 1963–88
  • 68. Tanzania, 1965–88
  • 69. Thailand, 1961–90
  • 70. Togo, 1970–87
  • 71. Trinidad & Tobago, 1966–69
  • 72. Tunisia, 1961–90
  • 73. Turkey, 1961–88
  • 74. United Arab Emirates, 1973–89
  • 75. Uruguay, 1961–89
  • 76. Venezuela, 1961–90
  • 77. Yemen Arab Republic, 1969–87
  • 78. Yugoslavia, 1968–88
  • 79. Zaire, 1961–89
  • 80. Zambia, 1961–87
  • 81. Zimbabwe, 1970–87
References

    AndersonD.Economic Growth and the Returns to InvestmentWorld Bank Discussion Paper No. 12 (World BankWashington, D.C.1987).

    Asian Development BankImproving Domestic Resource Mobilization Through Financial Development (ManilaAsian Development Bank1985).

    AgarwalaR.K.Price Distortions and Growth in Developing CountriesWorld Bank Staff Working Papers No. 575 (World BankWashington, D.C.1983).

    • Search Google Scholar
    • Export Citation

    BalassaB.Policy Reform in Developing Countries. (LondonOxford University Press1971).

    BalassaB.Financial Liberalization in Developing CountriesPPR Working Papers WPS 55World BankWashington, D.C.1989).

    BhagwatiJ.Anatomy and Consequences of Trade Control Regimes (National Bureau of Economic Research1978).

    BhagwatiJ.Directly Unproductive Profit-Seeking (DUP) ActivitiesJournal of Political Economy Vol. 90 No. 5 (1982) pp. 9881002.

    • Search Google Scholar
    • Export Citation

    BarroR.A Cross-Country Study of Growth, Saving and Government (National Bureau of EconomicResearch Working Paper No. 2855February1989).

    • Export Citation

    BarroR.Government Spending in a Simple Model of Economic GrowthJournal of Political Economy (October1990).

    de MeloJ.Distortions in the Factor Market: Some General Equilibrium EstimatesReview of Economics and Statistics (November1977) pp. 399405.

    • Search Google Scholar
    • Export Citation

    DoughertyC. and M.SelowskyMeasuring the Effects of Misallocation of LaborReview of Economics & StatisticsAugust1972 pp. 38690.

    • Search Google Scholar
    • Export Citation

    EasterlyW.Policy Distortions Size of Government and GrowthNovember 10, mimeo (World BankWashington, D.C.1989).

    EasterlyW. and D.WetzellPolicy Determinants of Growth: Survey of Theory and Evidence” (World Bank Working Paper Series No. 3431989).

    • Search Google Scholar
    • Export Citation

    EkelundR. and R.TollisonMercantilism as a Rent-Seeking Society: Economic Regulation in Historic Perspective (Texas A & M.1981).

    • Search Google Scholar
    • Export Citation

    FederG.On Exports and Economic GrowthJournal of Development Economics (>February/April1982).

    FloystadG.Distortions in the Factor Market: An Empirical InvestigationReview of Economics and Statistics Vol. 57 (May1975) pp. 20013.

    • Search Google Scholar
    • Export Citation

    FritzR.G.Time Series Evidence on the Causal Relationship between Financial Deepening and Economic DevelopmentJournal of Economic Development (July1984) pp. 91112.

    • Search Google Scholar
    • Export Citation

    FryM.J.The Cost of Financial Repression in TurkeySavings and Development Vol. 3 No. 2 (1979) pp. 12735.

    FryM.J.Saving, Investment, Growth and the Cost of Financial RepressionWorld Development Vol. 8 No. 4 (April1980) pp. 31727.

    • Search Google Scholar
    • Export Citation

    FryM.J.Interest Rates in Asia: An Examination of Interest Rate Policies in Burma India Indonesia Korea Malaysia Nepal Pakistan the Philippines Singapore Sri Lanka Taiwan and Thailand” IMF Asian Department (June1981).

    • Search Google Scholar
    • Export Citation

    FryM.J.Models of Financially Repressed Developing EconomiesWorld Development Vol. 10 No. 9 (1987) pp. 73150.

    FryM.J.Financial Saving, Financial Intermediation and Economic Growthin Domestic Resource Mobilization through Financial Development Vol. II (ManilaAsian Development Bank, Economics OfficeFebruary1984).

    • Search Google Scholar
    • Export Citation

    FryM.J.Money Interest and Banking in Economic Development (Baltimore and LondonJohn Hopkins University1988).

    GalbisV.Financial Intermediation and Economic Growth in Less Developed Countries: A Theoretical ApproachJournal of Development Studies Vol. 13 No. 2 (January1977).

    • Search Google Scholar
    • Export Citation

    GallagherM.Rent-Seeking and Economic Growth (Boulder, ColoradoWestview Press1991).

    GelbA.A Cross-Section Analysis of Financial Policies, Efficiency and GrowthPPR Working Paper WPS 202 (World BankWashington, D.C.1989).

    • Search Google Scholar
    • Export Citation

    GemmelN.International Comparisons of the Effects of NonmarketSector GrowthJournal of Comparative Economics No. 7 (1983) pp. 36881.

    • Search Google Scholar
    • Export Citation

    GoldsmithW.The Determinants of Financial Structure (ParisOrganization for Economic Cooperation and Development, OECD1966).

    GoldsmithW.Financial Structure and Development (Yale University Press1969).

    GrierK. and G.TullockAn Empirical Analysis of Cross-National Economic Growth, 1951–80Journal of Monetary Economics Vol. 24 (1989) pp. 25976.

    • Search Google Scholar
    • Export Citation

    GuptaK.L.Financial Development and Economic Growth in India and South KoreaJournal of Economic DevelopmentDecember1986 pp. 25976.

    • Search Google Scholar
    • Export Citation

    GuptaK.L.Aggregate Savings, Financial Intermediation, and Interest Rate”Review of Economics and Statistics (May1987) pp. 30311.

    • Search Google Scholar
    • Export Citation

    GurleyJ.G. and E.S.ShawMoney in a Theory of Finance (WashingtonBrookings Institution1960).

    International Monetary Fund IMFInterest Rate Policies in Developing Economies Occasional Paper No. 22 (Washington, D.C.1983).

    JaoY.C.Financial Deepening and Economic Growth: A Cross-Section AnalysisMalayan Economic Review Vol. 21 No. 1 (April1976) pp. 4758.

    • Search Google Scholar
    • Export Citation

    JungW.Financial Development and Economic Growth: International EvidenceEconomic Development and Cultural Change Vol. 34 (January1986) pp. 33346.

    • Search Google Scholar
    • Export Citation

    JungW. and P.MarshallExports, Growth and Causality in Developing CountriesJournal of Development Economics Vol. 18 (1985) pp. 112.

    • Search Google Scholar
    • Export Citation

    KhatekhateD.R.Assessing the Impact of Interest Rates in Less Developed CountriesWorld Development Vol. 16 No. 5 (1988) pp. 57788.

    • Search Google Scholar
    • Export Citation

    KapurB.K.Alternative Stabilization Policies for Less Developed EconomiesJournal of Political Economy Vol. 84August (1976) pp. 77795.

    • Search Google Scholar
    • Export Citation

    KormendiR. and P.McguireMacroeconomic Determinants of Growth: Cross-Country EvidenceJournal of Monetary Economics Vol. 16 (1985) pp. 14163.

    • Search Google Scholar
    • Export Citation

    KruegerA.O.Liberalization Attempts and Consequences (New YorkNational Bureau for Economic Research1978).

    LandauD. (1986) “Government and Economic Growth in Less Developed Countries: An Empirical Study for 1960–80Economic Development and Cultural Change Vol. 35 (October1988) pp. 3576.

    • Search Google Scholar
    • Export Citation

    LanyiA. and R.SaracogluInterest Rate Policies in Developing Economies” Occasional Paper No. 22 (Washington, D.C.IMF1983).

    • Search Google Scholar
    • Export Citation

    MaddalaG.Econometrics (McGraw-HillNew York1973).

    MageeS.Factor Market Distortions Production and Trade: A SurveyOxford Economic Papers (March1973).

    MathiesonD.J.Financial Reform and Stabilization Policy in a Developing EconomyJournal of Development Economics Vol. 7 No. 3 (September1980) pp. 35995.

    • Search Google Scholar
    • Export Citation

    McKinnonR.I.Money and Capital in Economic Development (WashingtonThe Brookings Institution1973).

    MoschosD.Export Expansion, Growth and the Level of Economic Development: An Empirical AnalysisJournal of Development Economics Vol. 30 (January1989) pp. 93102.

    • Search Google Scholar
    • Export Citation

    OdedokunM.O.Alternative Econometric Approaches of Analyzing the Role of Financial Sector in Economic Growth: Time-Series Evidence from 71 LDCsJournal of Development Economics (forthcoming).

    • Search Google Scholar
    • Export Citation

    PatrickH.T.Financial Development and Economic Growth in Underdeveloped CountriesEconomic Development and Cultural Change Vol. 14 (January1966).

    • Search Google Scholar
    • Export Citation

    PorterR.C.The Promotion of The Banking Habit and Economic DevelopmentJournal of Development Studies (July1966).

    RamR.Government Size and Economic Growth: A New Framework and Some Evidence from Cross-Section and Time Series DataAmerican Economic Review Vol. 76 (March1986) pp. 191208.

    • Search Google Scholar
    • Export Citation

    ShawE.Financial Deepening in Economic Development (Oxford University Press1973).

    SheeheyE.J.Exports and Growth: A Flawed FrameworkJournal of Development Studies (October1990) pp. 111116.

    WaiTun U.Economic Essays on Developing Countries (RockvilleMD. Sijthoff & Nordhoff1980).

    World BankWorld Development Report (World BankWashington, D.C.1987).

    World BankWorld Development Report (World BankWashington, D.C.1989).

1/

This paper was written while he was a visiting scholar at the Research Department of the IMF. The author wishes to thank Mr. Donald J. Mathieson for helpful comments and suggestions. The views expressed are those of the author and do not necessarily represent those of the Fund.

2/

Briefly, the fixed-effect model assumes that each country has its own intercept, which may or may not be correlated with other regressors. The variance component method, on the other hand, further assumes that these intercepts are not correlated with the other regressors.

3/

An alternative measure of the efficiency employed by Khatekhate (1988) is the “rental/wage ratio” computed as the real interest rate (as a proxy for the user cost of capital) divided by the real wage rate. However, this is not only an uncommon measure in the literature, but its computation requires data that are not available for most developing countries.

4/

Ideally, it is the actual output-capital ratio (as opposed to the incremental ratio) that is the appropriate index. However, data on capital stock for most developing countries do not exist.

5/

For instance, Gallagher (1991) considers the effects of composition of investment (by including investment in human capital); degree of cultural homogeneity, noneconomic shocks, and political plurality in his cross-country study of the determinants of IOCR in Africa—although the reported effects of these factors are generally hardly discernable.

6/

Other alternative measure of financial depth tried but not retained in deriving the reported estimates are as discussed later in the footnote on page 20

7/

By re-defining those variables expressed as ratios of GDP in per capita real terms, a typical result (estimated through fixed-effect panel method) obtained is as follows: IOCR=0.510expt+0.305govt+0.106devbnk+0.582findepth,(10.1)(4.2)(2.2)(6.1) where expt; govt; devbnk; and findepth are the growth in the real per capita values of exports; general government consumption expenditure; stock of development bank credits to the private sector; and stock of liquid liabilities of the banking system respectively. However, there is little or no economic rationale for measuring the regressors in this manner—inspite of the statistically very significant coefficients they have.

8/

We are not able to explore the effects of lagged values of EXPT, GOVT, DEVBANK and FINDEPTH (i.e., those variables are expressed as ratios of GDP) because of the high correlations they have with their contemporaneous values. For example, the simple correlation coefficients the one-period lagged values have with their respective contemporaneous values are 0.968; 0.982 and 0.984 for EXPT; GOVT (general government consumption expenditure-GDP ratio); DEVBNK and FINDEPTH. But the lagged values of the growth rates of these ratios are far from having such high correlations with their respective contemporaneous values and, hence, the effects of their lagged values are explored.

9/

Variance component method is used, instead of the fixed-effect technique, mainly because the latter does not enable us test for the effects of regional dummy variables.

10/

The DW values correspond to the OLS estimates as the software used does not provide DW values under the fixed-effect and variance component estimation methods.

11/

Also, the two-period lagged value of the growth of exports-GDP ratio has a significant positive effect. For example, the OLS regression on the current and past two values of the growth of exports-GDP ratio provides the following result:

IOCR=0.166 + (21.6)+ 0.053expt (1.7) + 0.088expt (t-1) (2.8) + 0.048expt (t-2) (1.5).

12/

We also explored alternative measures of the relative size of development banking. In all cases, either similar evidence or statistically insignificant effects—sometimes positive and sometimes negative—are detected. For instance, by measuring it as the ratio of first-difference of real value (nominal value deflated by the GDP deflator prior to taking the first-difference) of the stock of development banking credit to the private sector to real GDP, result is as follows:

IOCR = 4.378findepth (4.6) + 2.392findepth(t-1) (2.0) + 0.654devbnk (0.8) + 0.095debnk(t-1) (0.1)

Also, by measuring it as the ratio of the development banking system’s nominal flow of credit (i.e., the first difference of the nominal stock of credit) to the nominal GDP, the corresponding result is:

IOCR = 4.873findepth (4.6) + 2.674findepth(t-1) (2.4) - 2.203devbnk (-2.2) - 0.093devbnk(t-1) (-1.0)

where devbnk is the measure of development banking; findepth is financial depth measured as ratio of the first difference of real value of the stock of the banking system’s liquid liabilities to the real GDP; and t-values are in parentheses. Inclusion of other regressors does not change the results.

13/

It is pertinent to point out here the alternative measures of financial depth that were tried. When financial depth is alternatively defined as the flow of the nominal liquid liabilities (i.e., the first difference of the nominal stock values) in relation to the nominal GDP, statistical insignificance was observed.

Measuring it, as Gelb (1989) did, in terms of the ratio of the first difference of the real stock of the liquid liabilities (i.e., real value before first-difference) to the real value of domestic savings, does not yield statistical significance either.

Measuring financial depth from the perspective of the asset side of the banking sector balance sheet yields similar results (similar to when approached from the liability side) if the banking sector’s credit to the private sector is the asset employed. For instance, by defining FINDEPTH and findepth as the level and growth (respectively) of the stock of credit to private sector/GDP ratio, typical results are:

IOCR = -0.390FINDEPTH (-4.8)

IOCR = -0.121findepth (-2.9) + 0.021findepth(t-1) (0.5).

Similarly, if findepth is computed as the first-difference of real stock of credit to the private sector as a ratio of real GDP, a typical result is as follows:

IOCR = 0.572findepth (4.4) + 0.0244findepth(t-1) (1.9).

Inclusion of other regressors in the above (fixed-effect) equation estimates is found to make no appreciable difference.

14/

The statistically insignificant coefficients are due to the simultaneous inclusion of the real interest rate and inflation rates as regressors, in view of the high correlation between them. Statistically significant positive coefficients of the real interest rate are always obtained if the inflation rate is not included as a regressor in the equations.

15/

For example, by excluding observations that correspond to nominal interest rates above 300 percent the correlation coefficients between the real interest rate and each of the nominal interest rate and inflation rate become 0.540 and -0.171 while its correlation with the IOCR becomes positive (0.066) instead of the negative value of -0.064 when the outliers are included. By reducing the cut-off to observations corresponding to nominal interest rates below 200 percent, the correlation coefficients become 0.347; -0.387; and 0.081 respectively. These values change to 0.142; -0.642; and 0.121, respectively, if observations corresponding to a nominal interest rate of 100 percent and above are excluded, and so on.

16/

Export, exchange rate distortions and government expenditure variables are excluded from the investment ratio equations because there would be little or no economic reason for them. Inclusion of the financial depth variables in the investment equation is justified as being a measure of credit availability—the same justification for including the development banking variables. Real interest and/or inflation rates are conventional arguments in investment equations. Except for the financial depth and development banking variables, all the regressors in the economic growth equations are commonly employed.

17/

The major components of GDP are by end use in the form of investment, private consumption and government consumption as well as by industrial origin in the form of agriculture, manufacturing, construction, utilities and services.

18/

While studies evaluating the effects of credit rationing, foreign exchange rationing, the size of public sector, etc., on economic efficiency are also examining the effects of rent-seeking simultaneously, such studies are not being regarded here as belonging to those primarily concerned with the effects of rent-seeking. Evaluation of rent-seeking effects are only implicit and indirect in such studies.

Other Resources Citing This Publication