WorldDevelopment, Vol. 20, No. 10, pp. 1481-1488, 1992. 030.5-750X/92 $5.00 + 0.00 Printed in Great Britain. 0 1992 Pergamon Press Ltd

Impact of Foreign Aid on the Fiscal Behavior of

LDC Governments

HAIDER AL1 KHAN

and

EIICHI HOSHINO* University of Denver

Summary. -The purpose of this paper is to examine the Impact of foreign aid on the behavior of the recipient govcrnmcnts. Specifically. we look at a sample of less-developed countries in South and Southeast Asia and study their response to foreign aid received. Their rcsponsc is measured in terms of their public investment and consumption as well as taxation. Our paper attempts to make two new contributions to the literature. First. we try to estimate the impact of foreign aid on a sample of five South and Southeast Asian countries. Secondly. we improve on the earlier method of a combination of single-equation and two-step least squares by estimating the whole system with a nonlinear three-step least squares method. The results confirm the hypothesis that aid does affect consumption, investment and taxation of these govcrnmcnts, but they also show that the relationship is a complex one. In particular, grants and loans have different effects on invcstmcnt and taxation

1. INTRODUCTION

The purpose of this paper is to examine the impact of foreign aid on the behavior of the recipient governments. Specifically, we look at a sample of LDC’s in South and Southeast Asia and study their response to foreign aid received. Their response is measured in terms of their public investment and consumption as well as taxation. Both official and unofficial grants and loans from bilateral and multilateral sources are included in aid. It is postulated that the govern- ments’ expenditures and revenue efforts are affected by foreign aid through a reallocation within the categories of public expenditures and revenue raising.

Various previous researchers have found evi- dence for considerable dissipations of aid in the forms of government consumption (Please, 1967; Papanek, 1973: Weisskopf, 1972a, 1972b; and Heller, 1974). Many researchers concentrate at the macro level on the relationship among foreign aid, domestic savings and economic growth (see, for example, Griffin and Enos, 1970). Recent work by Mosley, Hudson and Horrel (1987) showed that while the conclusions regarding such putative relationships may be

valid, there is nevertheless an interesting ques- tion regarding the effect of foreign assistance on intermediate policy variables such as public investment, government consumption and taxa- tion. We pursue their lead and try to trace such relations in a simultaneous equation model applied to a sample of five South and Southeast Asian less-developed countries (LDCs). In doing this we follow an emerging trend in applying rigorous econometric modeling techniques to explore foreign aid issues in this region.’

The model follows Heller (1974, 1975) in that the behavioral equations are the results of max- imizing behavior by the policy maker. Our paper attempts to make two new contributions to the literature. First, we try to estimate the impact of foreign aid on a sample of five South and Southeast Asian countries. This has not been done within the context of an econometric model before. Second, we improve on Heller’s earlier method of a combination of single-equation and two-step least squares (2SLS) by estimating the whole system by a nonlinear three step least squares (3SLS) method.’ The results confirm the

*Final revision accepted: October 28, 1991.

1481

1482 WORLD DEVELOPMENT

hypothesis that aid does affect consumption, investment and taxation of these governments, but they also show that the relationship is more complex. In particular, grants and loans have different effects on investment and taxation.

In the next section we describe the model and derive the system of equations. The data, estima- tion procedure and results are described in sections 3 and 4 respectively. Summary and conclusions follow.

2. THE MODEL

We make the following assumptions regarding the actions of the public policy makers in our sample countries. They maximize their utility subject to: (a) alternative uses of public resources such as

provisions for economic growth, spending for social and economic services, and for the maintenance of political institutions.

(b) the distribution of total GDP between the private and the public sector.

(c) alternative modes of domestic financing, such as taxation and borrowing.

(d) alternative types of external assistance. such as grants and loans.

Let us postulate the following general utility function for the policy makers at any time period t:

U = f (Ig, 7-3 G,.. G,, B, A,, AZ)

where lg = Public investment expenditure T = Total tax revenue G,. = “Civil” consumption

G, = Socioeconomic consumption B = Borrowing from domestic sources

Al = Total foreign grants to the public sector from all sources

AI = Total foreign loans to the public sector from all sources

Each variable is in real terms and refers to time period t unless otherwise stated.

The three expenditure categories, as in the budgets of Heller’s Anglophone countries, re- flects a functional distinction in the budget of many Asian countries. IX is the public sector’s contribution to the planned growth rate. It is measured by the gross capital formation in the public sector plus net loans to the others sectors.

G, by definition includes all current expendi- tures which are not intended for capital forma- tion in the conventional sense, but rather are spent toward socioeconomic ends. This will include total government consumption (defined below). Therefore G, does include expenditure

on health and education, i.e. some investments on human capital.

G,. is the expenditure of the state to sustain itself. It includes administrative expenses, debt servicing, police security, defense forces and some transfer payments.

On the revenue side tax policy is assumed to be a manipulable instrument. Hence the endoge- nous status of the tax variable. Borrowing is assumed to be a restriction (to be explained later) yielding disutility to the policy makers.

External finance through A, and A2 are viewed as exogenous to the model. These are controlled by aid giving agencies which are motivated by both political and economic factors.

We use the following form of the utility function:

u = a(, + a, (1, - I,*) -5 ([,, - I,&,*)? - 2

a3 (T - T*) --?- (T - T*)' + a5 (G, - 1

G,.*) - __ (G, - G,.*)’ + u7 (G,, - G,*) - ?

-““(G, - G,*)’ - aq (B - B*) -? ? 7

(B - B*)2 a, 2 0 (1) The above form, though lacking in inter-

dependency between variables, is heuristically accurate and simple. The variables with an asterisk represent the target level for each vari- able. It must be acknowledged, however, that work with nonquadratic functional forms is still in a very early stage. hence other functional forms (with various asymmetries) should also be considered in the future.’

The relationships yielding the target variables are as follows:

IX+ = alI Y,_, + aI2 Ip (2)

T+ = aI3 Y,_, + aI1 M,_, (3)

G,.* = a15 G,.,-r (4)

G,* = aI6 E + aI7 Y,_, + aIx (Y, - Y,_,) (5)

where M = Value of real imports E = Primary school enrolments (in

units of 10,000 students)

L = Real private investment expendi- ture

l,q* can be expected to be influenced by the

desired rate of growth of GDP/capita. Within a Harrod-Domar framework Y,_r can be used as an instrument because IX* will be positively related to Y,_, , given a target growth rate. The relation between I, and I,, may be complemen- tary or substitutional. G,,, as we previously mentioned, includes investment in human capi- tal. E is used as a proxy variable to measure the level of this expenditure.

IMPACT OF FOREIGN AID 1483

variables subject to constraints (7) and (8) using Lagragian multipliers yields:

6U

61, a1 - Q2 ug - I,*) + h2 = 0 (9)

6lJ - = a5 - a6 (G,. - Cc*) + h, = 0 (10)

bG,

G,. is assumed to be noneconomic in nature. It is likely to be the outcome of a complex interaction between the various domestic (e.g., political parties, private interest groups, bureau- crats, military, etc.) and perhaps international (e.g., foreign advisors, aid donors, etc.) political actors. Modeling these processes would be a heroic task. Rather than doing this, here the simplifying assumption that G,* increases at a constant proportion on a fiscal year basis is made. Empirically, this seems to be reasonable as there is little variation around the trend rates of growth in G,. for all the countries in the sample.

6U ~ zz

6G,, a7 - ox (G,, - G,*) + )iz = 0(11)

bU ~ = - a - a4 (T - T*) - h,

6T (1 -’ PI) - h2P, = 0

blJ

(12)

- = - al, - a,,, (B - B*) - h, = 0 (13) 6B

The planned level of taxes are determined by previous year’s income (which is an underesti- mate given positive rate of growth) and the level of previous period’s import. It is assumed that ex ante borrowing target is zero.”

We have introduced the institutional con- straints which force the policy makers to operate only in the feasible region of their decision mapping. The simplest assumption will be:

T + B + A, + A7 = In + G, + G,

This will be unrealistic because of empirical evidence suggesting that borrowing for recurrent budget is not practiced and the level of loanable funds to the public sector within the economy is a small, in fact negligible.

6U ~ = I, - B - (1 - P,)T - (1 - P2)A,

hh, - (1 - P3)Az = 0 (14)

6U - =

% G,, - G,. - P,T - PzA, - p-Y42 = 0 (1%

Solving (9)-(15) we get the following set of structural equations:

Bl) +

PI +

An alternative set of constraints will be:

Ifi = B + (1 - P,)T + (A, + A*) G, + G, = P,T oap,a1 (6)

(16)

where 1 - P, = the proportion of the recurrent budget saved for investment purposes.

Weisskopf (1972a, 1972b) contends that there is significant substitutability between consump- tion and investment elements of foreign aid. The following constraint set (our final set) incorpor- ates this suggestion:

I32 Plah a4 T=----- + -(Cc* - G,) +-

Pi P3 1%

aI0 T* + ~ {[I - PI1 [Ix - (1 - PI) A, - (1 - Pi) ,421) + &i (17)

Zb’ = B + (1 - P,)T + (1 - P2)A, + (1 - P3)

A2 (7)

G, + G,. = P,T + P2 A, + P3 A2 (8)

a1 - a9 a2 aI0 Is = + Ig* +

a2 + aI0 a2 + aI0 a2 + a10

((1 -PI) T+ (1 -P~)A, + (1 -A

Maximizing (1) with respect to the policy A,) + ~4 (18)

1484 WORLD DEVELOPMENT

(3, = (ai - a; + a<, (I - P,))

(33 = IQ + UIII (1 - P,)‘}

The above system of simultaneous equations (16)-(18) are estimated by three stage least squares method.’ As is well known unless a simultaneous equations system is exactly identi- fied (that is, unless there is one to one corres- pondence between the parameters of the reduced form system expressing each endogenous vari- able in terms of varius exogenous variables only, and the structural equation parameters) methods other than single equations ordinary least squares (OLS) (or indirect least squares) must be used. The present system is overidentified. Therefore, for some structural parameters one can not expect to derive a unique value from the reduced form equation parameter estimates. The 2SLS procedure solves this problem by introducing additional instrumental exogenous variables so that exact identification is once again possible. Since this, however, is done equation by equa- tion the estimated system as a whole may or may not exhibit minimum variance for the sampling distribution of parameters. In order to ensure this an estimate of the variance-covariance mat- rix is taken from the second stage. Next, the system as a whole is estimated by following the procedure of Generalized Least Squares (GLS). As a result the variance after the third stage is lower than or equal to that after the second stage. Furthermore, in our estimation the GLS correc- tion procedure for heteroskedasticity and auto- correlation are also followed. In the case of heteroskedasticity the differences in variance arising from country size differences are cor- rected. In the second case serial correlation in the time-series data is corrected for first-order auto- correlation. We now turn to a discussion of the data set and the results.

3. THE DATA

We use a sample of five South and Southeast Asian countries: India, Pakistan, Bangladesh, Sri Lanka and Malaysia. Since the time-series data for each country are not sufficient and the cross- section is small,” we combine the two and use a

regression model for pooled time-series, cross- section data.

The data sources are primarily the budgetary statistics of the country governments, supple- mented by UN Yearbooks, IMF statistics, and OECD statistics.

All aid data have been converted to real terms by dividing current values by deflators from IMF (195576). In each case the initial year price base is treated as 100. All figures have also been converted to dollars by using the IMF published exchange rate for the appropriate years. Loan data include both concessional and nonconccssio- nal transfers and except for official aid grants and loans also include those from private sectors.

4. THE EMPIRICAL RESULTS

The estimated equations described the govern- ments’ consumption, investment and taxation behavior in the presence of foreign aid. Tables I- 3 describe these estimated equations. Each equa- tion is estimated four times, allowing for distinc- tions between official and total grants and loans, and for bilateral vs. multilateral aid.

It should be mentioned first that the hypothesis regarding the flexible allocation of aid between government consumption and investment is veri- fied. Absolute values of all the P,‘s (i = 1, 2, 3) arc between 0 and 1 (see Table I) and signifi- cantly different from one at 0.05 level. Thus aid money is distributed between various corz.slrr>tp- ICVZ and imwmmf expenditures. Further, aid may divert other revenues from investment to consumption. The estimate of PI equal 0.X8 implying that only 12% of the taxes flow to the investment in the presence of aid. There is a difference between loans and grants. however, as can be seen from the negative sign of the loan coefficient, P1. Loans do encourage investment by actually pulling away from the current budget.

The results seem to hold generally regardless of whether aid is official or both official and private. and whether aid is bilateral or multilate- ral (see Table 2). In order to see more directly

Aid component P, 1’7 P;

Pooled total (A, + A:) 0.8X 0.3x -0.21

Pooled official (A, + A,) 0.01 0.77 0.04

Bilateral (A ,) 0.86 0.14 0.12

Multilateral (A?) 0.X6 0.35 0.16

IMPACT OF FOREIGN AID 1485

WORLD DEVELOPMENT

Table 3. Revenue equakm parameters

ah allI Aid components Constant Y,-, __ -

131 Pi

Pooled total -3.2t 0.31* 0.8* 1.6k (A, + A,) (-3.32)t (12.34) (3.8) (2.3)

Pooled official -2.1* 0.25* 0.75* 1.66* (A, + A,) (-2.3) (10.06) (3.Y) (3.8)

Bilateral - 1.9* 0.29* 0.76* 1.71* (A,) (-3.2) (11.02) (5.06) (4.6) Multilateral -2.2* ( 0.26* 0.7x* 1.75* (AZ) (-4.0) (13.04) (6.08) (5.22)

Revenue -3.2 + 0.31Y,. , + 0.8(0.704G,,,_, - 0.88G,) equation + 0.192 (lx - O.l2A, + 1.21 A,) + 0.21 M,_, (Pooled total)

M, I

0.21* (2.8)

0.12* (2.9)

0.15* (3.52) 0.1x*

(3.06)

“Significant at 0.05 level ?I values in parentheses.

how investment is affected we substituted for I,* in equation (18) and arrived at equation (19) for I, which is as follows:

l,q = -0.8 + 0.074T + 0.32A, + 0.8SA2 + O.O9Y,-, + 0.181,

Turning to this investment equation we see that both grants and loans affect public invest- ment positively. Loans have, however, a more positive effect than grants. Our previous observa- tions regarding the erosion of investment effect of tax in the presence of foreign aid is confirmed. Indeed in the presence of private investment and a tiarrod-Domar type effect (given by the coefficient of y,_t) the contribution of tax re- venues to public investment is even weaker (7.4%). In case of investment also the stratifica- tion of the aid sample according to official vs. total or bilateral sources did not lead to markedly different results.

Finally, Table 3 shows the effect of aid on the taxation of these governments. Interestingly, grants reduce taxation effort, while loans in- crease it. The explanation for this is quite simple. Rational policy makers use nonrepayable grant money to reduce the tax burden, but the burden of loan payments induce them to increase taxes.’

5. SUMMARY AND CONCLUSIONS

A number of interesting conclusions with important policy implications can be reached from this study. Our results confirm the hypoth- esis that foreign aid affects both the expenditure

and the revenue side of the recipient govern- ments’ budgets. On the consumption side, gener- ally foreign aid is treated as an increase in income, and given positive income elasticity in the government sector there is an increase in consumption. The marginal propensity to con- sume out of foreign aid, however, turns out to be less than one. Therefore, some public investment is also financed out of aid money.

Though we could not find any differences between the aid given by multilateral and bilat- eral sources, there are important differences between the other two components of aid, namely loans and grants with regards to how they are used. Both loans and grants are used for consumption as well as investment purposes. But ceteris parihus, 85 cents out of a dollar of loan received goes to investment as opposed to 32 cents of grants. Thus, if the purpose of aid is to generate investment, it is served much better by making loans to the LDC governments than by giving grants.

Finally, aid also affects the tax efforts of these LDC governments. Interestingly, again grants reduce the tax burden while loans increase it. At the same time, the comparison of welfare effects of grants vs. loans may not he straightforward. Since grants do not have to be repaid immedi- ately the reduction of taxes may seem to be more welfare enhancing than the effects of loans. Since loans. however, lead to relatively more invest- ment and hence more income in the future the net discounted benefits may very well be higher in some cases. This is an interesting area of future research.

IMPACT OF FOREIGN AID 1487

NOTES

1. See respectively, Gang and Khan (1991), Aslam equal the values of target variables. Gang and Khan (1987), Rana and Dowling (1988), Pack and Pack (1989) try to move in this direction without complete (1990). success.

2. Gang and Khan (1991) also use a nonlinear 3SLS 4. Of course, B* > 0 is compatible with the model. procedure. B* = 0 simplifies later analysis.

3. For readers who are more familiar with the usual quadratic loss function from the stabilization literature a few words of explanation about the insertion of the linear forms may be useful. It has been long known that the symmetry in the quadratic loss function implies behavioral restrictions in that exceeding or falling short of the target is treated equivalently. Insertion of the linear terms introduces a slight asymmetry. This by no means implies, however, that the unconstrained objec- tive function is maximized (or, minimized if we take the loss minimazation approach) when actual values equal the targets. A completely satisfactory solution requires asymmetric loss functions which also give an uncon- strained minimum when values of decision variables

5. Heller (1974) used the 2SLS procedure. Executing the third stage increases the efficiency of parameter estimates. Corrections for heteroskedasticity. and auto- correlations are also made.

6. The cross-section has been kept small deliberately to make it as homogeneous as possible. The time-series for each country is as follows; India, 1955-76; Malaysia, 196876; Pakistan, 1955-70; Sri Lanka, 1955-76; and Bangladesh, 1972-76.

7. In an intertemporal model the tax burden could be imposed on future generations rather than the present one.

REFERENCES

Aslam, N., “The impact of foreign capital inflow on savings and investment: The case of Pakistan,” Pakistan Development Review, Vol. 26 (1987).

Central Statistical Office, Government of Pakistan, 25 Years of Pakisran in Sratistics, 1947-72 (Karachi: Government of Pakistan, 1972).

Central Statistical Organization, Department of Statis- tics, Government of India, S/atis/ical Abstract of [he Indian Union (IHO-72) (New Delhi: Government of India, 1972).

Gang, I. N. and H. A. Khan, “Foreign aid, taxes and public investment,” Journal of Developmenl Econo- mics, Vol. 34 (1991).

Gang, I. N. and H. A. Khan, “Modeling foreign aid and Development Expenditures,” Paper presented at American Economic Association Meetings (1989).

Government of Bangladesh, Annual Budgetary Sratis- tics (Dhaka: Government of Bangladesh. 1972-76).

Government of India, Annual Budgetary Slatisucs, (New Delhi: Government of India. 1955-76).

Government of Malaysia, Annual Budgetary Statistics, (Kuala Lumpur: Government of Malaysia, 1968-76).

Government of Pakistan, Annual Budgetary Sratistics, (Rawalpindi and Islamabad: Government of Pakis- tan, 1955-70).

Government of Sri Lanka, Annual Budgetary Sratistics, (Colombo: Government of Sri Lanka 1955-76).

Griffin, K. B. and J. L. Enos, “Foreign assistance: Objectives and consequences,” Economic Develop- ment and Cuhural Change, Vol. 18 (1970). pp. 313- 337.

Heller, P. S., “A model of public fiscal behavior in Developing Countries: Aid, investment and taxa- tion.” American Economic Review, Vol. 65 (1975), pp. 429-445.

Heller, P. S., “Public investment in LDC’s with

recurrent cost constraint: the Kenyan case,” Quar- terly Journal of Economics, Vol. 88 (1974). pp. 251- 277.

International Monetary Fund, International Financial Sfatistics (Washington, DC: IMF, various years).

Mosley, P., J. Hudson. and S. Horrel, “Aid, the public sector and the market in less developed countries,” Economic Journal. Vol. 97 (1987). pp. 616-641.

OECD, Geographical Distribution of Financial Flows to Developing Counrries (Paris: OECD, various years).

OECD, National Accounts of Less Developed Coun- tries, IWO-1966 (Paris: OECD, 1968).

OECD, OECD-ICVA Directory: Development Aid of Non-governmental Non-profit Organizalions (Paris: OECD. 1967).

OECD, Statistics of Balance of Payments, 1950-1961 (Paris: OECD, 1964).

Pack, H. and J. R. Pack, “Is foreign aid fungiblc? The case of Indonesia,” Tlze Economic Journal. Vol. I00 (1990).

Papanek. G. F., “Aid, foreign private investment, saving and growth in less developed countries,” Journal of Political Economy. Vol. 81 (1973) pp. 12t&l30.

Please, S., “Saving through taxation - Mirage or reality,” Finance Developmenf, Vol. 4 (1967). pp. l- 10.

Rana, B. R. and J. M. Dowling, “The impact of foreign capital on growth: Evidence from Asian Developing Countries.” The Developing Economies, Vol. XXVI-I ( 198X).

United Nations, Directory of International Statistics (New York: United Nations. 1975).

United Nations. Guide 10 Basic S/atisfics in Countries of the ECAFE Region (Bangkok: ECAFE, 1969).

1488 WORLD DEVELOPMENT

United Nations, Compendium of Social Stutistics 1967- I977 (Recut+1 de Sfutisiiques Sociu1e.s) (New York: United Nations. 1977).

United Nations, Methods Used in Compiling the United Nutions Price Indexer for Basic Commodities in international Trude (New York: United Nations. 1958).

United Nations, Demographic Yeurbook (New York: United Nations, various years).

United Nations, Statistical Yearbook (Annuuire Stutistt-

clue) (New York: United Nations. various years). Weisskopf, T., “The impact of foreign capital llow on

domestic savings in Underdcvelopcd Countries,” Journal of Internutional Economics, Vol. 2 (lY72a), pp. 25-38.

Wcisskopf, T., “An econometric test of alternative constraints on the growth of underdeveloped coun- tries,” Review of Economics and Statistics, Vol. 54 (1972h) pp. 67-78.