Aid, fiscal policy and macroeconomy of Uganda: a cointegrated vector autoregressive (CVAR) approach

Bwire, Thomas (2012) Aid, fiscal policy and macroeconomy of Uganda: a cointegrated vector autoregressive (CVAR) approach. PhD thesis, University of Nottingham.

PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (2MB) | Preview


While confronting the question of aid effectiveness, an important issue (but often ignored) in the context of a developing country like Uganda is which GDP measure would be most reliable as this is crucial for measuring the macroeconomic impact of aid. The most commonly used GDP measure in the aid-growth literature is typically from World Development Indicators (WDI) or Penn World Tables (PWT) (being considered the most reliable or the easiest to obtain). However, disparities in GDP from alternative sources are common and in practice one has different estimates of the level, change and growth of GDP for the same country over the same period. This is of a particular concern especially in developing countries (without exception) where the informal and subsistence sectors are a large share of the economy (Jerven, 2010) and where not all transactions in the formal sector are recorded (MacGaffey, 1991), and the quality of data is still very poor and measurement perceptions of macroeconomic aggregates are varied and weak (Mukherjee, White and Wuyts, 1998). Because the source chosen for GDP may affect inferences on growth and economic performance for African countries, the thesis entry point was an analysis of alternative sources of GDP, and aimed to construct a consistent GDP series for Uganda. The extent of discrepancy in GDP estimates was investigated, and the year on year percentage GDP growth rates, including percentage and average growth rate discrepancies were derived, with a particular focus on sub-periods when GDP from alternative sources diverge most.

Although UBOS and WDI real UGX GDP year on year growth rate estimates had a 3.6 percentage point average absolute discrepancy per year, they are consistent, similar and cointegrated. In fact, over 1970-76 and 2000-08 the two series are very close, and they are quite close for 1978-83 and 1993-99. Therefore, either series can be considered to represent trends in the size of the macroeconomy. However, the UBOS real series is smoother and produces a more stable measure of GDP than does the WDI series and it is the underlying source from which macroeconomic data is sought by the international agencies, including WDI. Given this, the less volatile UBOS real series (real UGX GDP/U) was preferred especially as there was less need to incorporate dummies in the rest of the thesis. Fiscal data and private consumption (our preferred measure of growth) in the thesis were derived from this same source. Two dynamics relationships, i.e. one between foreign aid and domestic fiscal variables, and the other between foreign aid, domestic fiscal variables, exports and private consumption in Uganda are assessed using annual data over the period 1972 to 2008. ACVAR model is employed and executed using CATS in RATS, version 2.1and E-views 7.2. Features of the data over 1972-79, a period characterized by political and economic instability in Uganda and the effect of policy shift due to structural adjustment programme and the Museveni regime in Uganda are reflected in the analysis.

Considering first the core fiscal variables, we find that aid and fiscal variables form a long-run stationary relation and the role of structural changes remain unclear as the policy shift dummy seems unimportant for the long-run fiscal relation. A test of structural links between aid and fiscal variables reveals that aid is a significant element of long-run fiscal equilibrium, and the hypothesis of aid exogeneity is not statistically supported. In the long-run, aid is associated with increased tax effort, reduced domestic borrowing and increased public spending, although aid additionality/illusion hypothesis remains inconclusive given the nature of the DAC measure of aid used here. A decomposition of the common trends shows that shocks to tax revenue are the pulling forces, while empirical shocks to domestic borrowing, government spending and aid are the pushing forces of the fiscal system. In terms of policy, it is crucial for the donors to increase the reliability and predictability of aid in order for Uganda to improve fiscal planning and reduce the need to resort to costly domestic borrowing. In addition, one way to make inference on the relationship between aid and spending more clear is for donors to coordinate aid delivery systems and also make aid more transparent.

Finally, we extended the fiscal analysis and also considered how aid, mediated by the fiscal variables, and exports impact on the growth of the private sector- a relationship a kin to the growth response to aid in Uganda. Results show that aid and the Ugandan macrovariables are significantly cointegrated, and a battery of sensitivity and robust checks demonstrate that the cointegration rank is 2. These are formally identified as representing respectively the statistical analogue of the budgetary equilibrium among the core fiscal variables and the link between aid, fiscal variables, exports and growth in private consumption. Using this rank condition, the hypotheses of long-run exclusion of aid and aid exogeneity are optimally tested within a system of equations, but these are not statistically supported. With particular reference to the growth relation, we find broad support that aid has had, in the long-run, a positive impact on the private sector, albeit indirectly through public spending, and deficit financing is associated with ‘crowd in’ effect linked to public investment spending. However, the belief that ‘earmarking’ aid to investment spending contributes to achieving target growth rates may be exaggerated. It is the productivity, not the level of investment that matter. On the contrary, aid may have an important role in supporting consumption spending, and this happens to be more beneficial to growth in Uganda than may be commonly acknowledged. The role of structural changes remains unclear as the policy shift dummy seems unimportant for the long-run fiscal and growth relations, but may matter for the short-run.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Morrissey, W.O.
Lloyd, T.A.
Keywords: Economic assistance, fiscal policy, economic conditions, Uganda
Subjects: H Social sciences > HC Economic history and conditions
Faculties/Schools: UK Campuses > Faculty of Social Sciences, Law and Education > School of Economics
Item ID: 12918
Depositing User: EP, Services
Date Deposited: 03 Apr 2013 15:26
Last Modified: 15 Dec 2017 06:02

Actions (Archive Staff Only)

Edit View Edit View