This paper's basic concern is with aid effectiveness and with the absorption of aid into the economies of recipient countries. It asks how the literature on the fiscal impact of aid adds to an understanding of these issues, and what light fiscal impact studies might shed on the nature and mechanics of the aid-growth relationship. It surveys aspects of the literature on aid and growth, budgetary choice, fungibility and fiscal impact and the empirical methodologies used in the last aspect. In doing so it serves as a theoretical and methodological introduction to fiscal impact case studies of three African countries - Malawi, Uganda and Zambia - which are presented in separate ESAU working papers, but whose main findings and conclusions are summarised here. The empirical conclusions of these studies are based on time series data from the 1970s to the early 2000s.
All three case-study countries have experienced economic troubles, and have pursued similar programmes of economic reform in the later 1980s and the 1990s. They have had defective, imperfectly consolidated, budgetary processes, and have been subject to fiscal indiscipline. They have all received large aid inflows relative to GDP, including substantial balance-of-payments and budget support, particularly in the early 1990s. Only Uganda, the most successful reformer, has successfully emerged from its difficulties into sustained growth and poverty reduction.
The weight of opinion among contributors to the aid-growth literature suggests that aid is effective in promoting growth, and that this is probably achieved, at least in part, through investment. However, there is ambiguity about the mechanics involved and about the flows of funds induced by aid which support them. Most aid is official, and is provided to governments for use in the financing of projects and programmes which ought to feature in their budgets - though in practice some are omitted. Budgets are therefore the first, and main, link in the chain of causation. Empirical evidence about the impact of public expenditure on growth is inconsistent, and the results are poorly explained. There is, moreover, no consensus in the literature about the micro-foundations of the budget process.
Much of the fiscal impact literature assumes that the objective of budget planners is to minimise a quadratic loss utility function. Previous authors have derived systems of structural and reduced-form equations from the first order conditions. They have estimated the reduced-form equations using instrumented estimates of budget planners' exogenous targets for the main fiscal aggregates.
This paper suggests that a more realistic schematic representation of the process is a Stone-Geary model in which budget planners are required, as a minimum, to finance certain basic mandatory or entitlement expenditures and, having done so, to allocate the additional resources available so as to equate the marginal social benefit of each outlay, thus maximising social welfare. This proposition is demonstrable in countries where budgets are planned in two parts, with dichotomies of funding and purpose, as has been the case in the case-study countries. The Stone-Geary welfare function is Cobb-Douglas with constant returns, so that discretionary expenditure shares are equal to marginal social benefits.
The country case studies estimate the structure and magnitude of the fiscal impact of aid using vector autoregression (VAR) analysis which is suitable for the analysis of time series of interrelated endogenous variables with lagged effects. It avoids prior assumptions about the underlying structural model and the use of estimated target variables.
A practical problem encountered in the econometric analysis is that, in all three countries, official development assistance (ODA) disbursed by donors has consistently exceeded by a wide margin the (grant and net loan) financing of budgets. The explanations for this include the use of ODA for debt and debt-service reduction, and the failure of donors and recipient countries to ensure that disbursements of project, programme, emergency and technical assistance paid directly to suppliers by donors, are properly recorded in budgets. A further complication is that all three countries have made use of non-ODA loans (both official and commercial) to finance public expenditures, and these are confused with ODA in the budget data. The econometric analyses of the three countries, therefore, quantify separately the impact on fiscal aggregates of both the external finance reported in national budget sources and of ODA as recorded by donors and reported by OECD.
The omission from budget receipts of significant proportions of official aid disbursements implies that recipient countries' public expenditure accounts also omit certain classes of expenditure. This should be a matter of concern to the donors of budget support, alongside other 'fiduciary' concerns.
The case studies demonstrate that the prime effect of external financing/aid in all three countries has been to increase the size of their development budgets - in some specifications by more than the magnitude of the aid injection. The impact on recurrent budget outlays has been either positive, but lower than that on the development budget (Uganda, Zambia), or negative (Malawi). Domestic revenues have risen with aid in Malawi and Uganda (in Malawi only in the long run),and have fallen in Zambia. However, the causality here is probably indirect and is potentially spurious. The effect of aid on domestic borrowing has been case-specific: in Malawi, the effect has been to reduce borrowing, in Uganda to leave it unchanged and in Zambia to increase it.
This paper points to case-study evidence that the practice of fiscal dichotomy inherited from the 1960s and 1970s - with recurrent budgets being financed from domestic revenues and development budgets from aid and other external sources - has persisted in some degree in the three countries, in spite of the institutional amalgamation of their ministries of finance and development planning. Only in Uganda has the practice been eroded by recent budget reforms. This lends some weight to a broad Stone-Geary interpretation of their budget processes, with mandatory and entitlement (payroll, pensions etc.) expenditures which are hard to compress financed in the main from within the (relatively stable) recurrent budget, and discretionary new projects and programmes from the (more volatile) development budget. As donors have attached greater importance to the operation and maintenance of government services, and have increased the share of (volatile) fungible programme support in their aid disbursements, so recipient countries have increased the share of non-investment expenditures in their development budgets (Malawi and Uganda, but not Zambia).
The persistence of dual budgeting, and the rigidity in budget management that it implies, has implications for the donors of general budget support. This support is predicated on an assumption of budgetary fungibility, permitting the efficient deployment of all resources for the production of public services. This is rendered invalid by compartmentalisation.
The interpretation placed on the excess of incremental development expenditure over aid injections is more speculative. It may arise from the over-optimistic planning of development budget expenditures relative to actual receipts of external financing, or simply from the 'gearing' effect of aid on public expenditure whereby domestically financed 'counterpart' funds have to be mobilised to complement and accompany external financing. A third possibility, as in Malawi, is that the development budget substitutes for the recurrent budget in financing certain classes of operating expenditure.
The empirical results yield plausible indications in two of the three countries about the effect of aid on macroeconomic stability. In Zambia, they suggest that the government failed to use aid to stabilise the economy because it was followed by a weakening in domestic revenue mobilisation and an increase in domestic borrowing. In Malawi, on the other hand, the results suggest the opposite, namely lower borrowing and higher revenues. Nevertheless, Malawi has remained prone to macroeconomic instability. In Uganda, the estimated effect on macroeconomic stability has been insignificant, but is interpreted as supportive of the country's success in stabilisation. The econometric analysis suggests that the impact on domestic revenue has been positive, whereas on domestic borrowing it has been insignificant in the long run. Caution is required in using these results, however, because budgets in all three countries are imperfectly consolidated, omitting significant fiscal and quasi-fiscal transactions, and because fiscal and monetary data on governments' domestic borrowing have been unreconciled.
The VAR/VEC-based econometric analysis highlights the presence and apparent importance of direct and indirect lagged effects in public finance - extending over periods of several years. These are country-specific, and not readily perceived by budget planners, but the evidence underlines the possible longer-term implications of current budget choices. In particular, the effect of aid in raising development expenditure by more than the size of the injection is found to have been persistent.
The country studies, combined with earlier research results, show that fiscal effects research casts only limited direct light on the mechanics of the aid-growth relationship. They indicate variety in the pattern of effects, suggesting that different growth outcomes can be associated with very similar patterns of fiscal impact and vice versa. The impact of aid must therefore ex ante be considered indeterminate. This paper does, however, point out the probably significant contrast between (i) the relatively stable flow of external assistance to Uganda and the more volatile flows to Malawi and Zambia, (ii) the (recently) disciplined, purposeful and effective planning of public expenditures in Uganda, contrasting with the persistent fiscal indiscipline and haphazard expenditure planning in the other two countries, and (iii) Uganda's success in using aid to rehabilitate economic infrastructure and institutions. The paper concludes, however, that these differences alone do not account fully for Uganda's superior growth and poverty-reducing growth record, which is due also to the more attractive environment for enterprise that has been created. In other words, the effectiveness of aid has been dependent on the policy environment in these three recipient countries.
Could the similarity of estimated impacts of aid on development expenditure arise from the statistical methodology used rather than from the underlying institutional and behavioural similarities of the budget processes in the three countries? The authors deny this, noting that the same statistical methodology applied recently to Ghanaian data identified response patterns in Ghana that were somewhat different from those found in Malawi, Uganda and Zambia.
The survey section of this paper draws attention to the differences in fiscal response patterns between different countries found in the literature. The conclusions of this paper and the accompanying country studies lend weight to other evidence of country specificity, though with the rider that inherited institutional similarities may pre-dispose to similar fiscal impact responses.