DRAFT NOT FOR CITATION IMPACT OF THE HIPC DEBT INITIATIVE ON MACRO ECONOMIC VARIABLES: THE CASE OF UGANDA Michael Atingi-Ego Bank of Uganda Ego@bou.or.ug ABSTRACT Following its consistency in pursuing sound macro-economic policies and commitment to structural reforms, Uganda become the first country to qualify for debt relief under the HIPC Initiative in 1998. As a result, the government deficit as a percentage of GDP increased from 6.7 to 12.8 in 1997/98 and 2001/2 respectively as the government expenditure on poverty reducing sectors more than doubled. As a result of the additional liquidity injection created from this process, the conduct of monetary policy became complicated with real interest rate rising and real exchange rate appreciating. From our empirical analysis, we see that the increase of government expenditure and foreign inflows by one percent saw the real exchange rate appreciate by 0.24 and 0.2 percent respectively. On the other hand, a 1% appreciation of real exchange rate has led to deterioration of noncoffee export performance by 4 percent. As developing countries therefore look forward to accessing Millennium Development Goals resources, the other side of this assistance should carefully be looked into as it could have several implications for the long-run sustainability of the private and external sectors of their economies. 1 DRAFT NOT FOR CITATION IMPACT OF THE HIPC DEBT INITIATIVE ON MACRO ECONOMIC VARIABLES: THE CASE OF UGANDA. 1.0 Introduction In 1996, the World Bank and IMF conceived a multilateral mechanism for extending debt relief to the Heavily Indebted Poor Countries (HIPC). The primary objective of this initiative is to reduce the poor countries debt burdens to sustainable levels in order to enable them have more resources to implement polices required to overcome constraints to sustainable economic growth. A key condition for obtaining the HIPC debt relief is that a country has to sustain strong macroeconomic performance and must have put together a comprehensive framework for structural and social reforms. In particular a country qualifying for debt relief must have comprehensive policies towards poverty alleviation through social sector reforms for which the saving made available from debt relief will be used. Therefore Uganda’s consistency in pursuing sound macroeconomic policies and its commitment to structural reforms enabled it to become the first country to qualify for debt relief under the HIPC Initiative in 1998. Prior to the HIPC debt initiative, Uganda implemented several strategies to reduce its debt burden. These strategies have included among others, borrowing on concessional terms, rescheduling with the Paris Club creditors during which a total of US$ 184 million was rescheduled. Between 1981 and 1987, the Uganda government negotiated debt rescheduling with the Paris Club creditors during which a total of US$ 184 million was rescheduled. Uganda also gained access to the Enhanced Toronto terms in January 1989, under which about 30% of the eligible maturities were written off, and the maturity of the remaining debt was reduced to 14 years with an 8 year grace period. A further write off of 50% of the outstanding maturities was obtained under the June 1992 Paris Club agreement. In July 1993, using the debt reduction facility of the World Bank, Uganda also bought back US$ 153million of commercial debt at substantial discount. Uganda has also benefited from other write-off and cancellations granted by the Paris Club countries, which led to bulk of its debt to the OECD being written-off. In 1995, Uganda also received additional debt relief from the bilateral donors, via the Multilateral Debt Fund (MDF) to help service multilateral debt obligations. In spite of the above efforts, the external debt indicators as indicated in appendix table 2 continued to deteriorate as the country accumulated more debt to finance development programmes in the 1990s. By the end of 1995/96, Uganda remained heavily indebted with the external debt stock amounting to US$ 3.5 billion while the stock arrears of neared US$ 250 million. Multilateral debt as a percentage of debt stock was high averaging 77% in 1997/98. The ratio of debt to export of goods also remained high although it declined from 1526.6% in 1992/93 to 535.5% in 1996/97. The Multilateral Development Fund and HIPC were set up to substantially address the issue of multilateral debt in order to give substantial relief to countries that were pursuing prudent macroeconomic policies and to further sustain growth in these economies. Although it is still in its early stages, the HIPC debt initiative has helped bridge the resource gap in Uganda and has thus helped increase government expenditure on poverty reducing programmes. HIPC has therefore released resources that would have otherwise been used for debt service for investment in poverty reducing social sector programmes. To date, these resources, which are channeled through the Poverty Action Fund (PAF), 2 DRAFT NOT FOR CITATION are an integral part of the Medium Term Expenditure Framework (MTEF), which comprises a rolling three year plan for the Government of Uganda. In light of the above, this paper attempts to qualitatively and quantitatively assess the extent to which the HIPC debt initiative through increased budgetary provisions has impacted on macro economic variables in Uganda. The paper is structured as follows: Section 2 discusses HIPC Initiative in Uganda, use of freed resources and mechanism used to disburse these resources. Section 3 then discusses the impact of increased budgetary provisions as a result of HIPC on macro economic variables. Section 4 gives empirical results of the impact of increased government expenditure on some macroeconomic variables such as the exchange rate and interest rates and their impact on the private sector activities. 2.0 The HIPC Debt Relief Initiative As earlier mentioned, in 1996 the World Bank and the IMF conceived the HIPC initiative as a multi-lateral mechanism for extending debt relief to the heavily indebted poor countries that were pursuing prudent macro economic polices. The key guiding principles of the initiative included additionality of debt relief, maintenance of the financial integrity of the multilateral financial institutions and burden sharing on a fair and equitable basis. Uganda’s consistency in pursuing sound polices and its commitment to structural reforms enabled her to become the first country to qualify for debt relief under the HIPC initiative in April 1998. Relief was based on the Net Present Value (NPV) of debt to exports with the sustainable level of debt assessed at 202% of exports. To reach this level, the amount of relief required was US$ 347 million in NPV terms, which is equivalent to total saving of US$ 650 million, spread over a thirty year period. In the first year, 1998/99, Uganda received US$ 45 million in debt relief and US$ 40 million was expected to be received annually for three successive years. The first HIPC was expected to provide an “exit strategy” from the debt rescheduling process or, leave beneficiary countries with sustainable debt levels. Post-HIPC, a country would be expected to service its debt and still be able to direct sufficient resources to poverty reduction. However, because of some unforeseeable factors such as the El Nino weather phenomenon, changes in global interest rates and a decline in international commodity prices, it was difficult for Uganda to have post-HIPC sustainable debt levels. Other heavily indebted poor countries also faced varying calamities, which prevented them from achieving debt reductions to sustainable debt levels. The HIPC Debt relief Initiative was then reformed (hence the Enhanced HIPC) to enable it benefit more countries and deliver deeper debt relief more quickly. The key elements of the Enhanced HIPC initiative are: • • • • Lowering the debt sustainability thresholds to a Net Present Value (NPV) of Debt at no more that 150% of exports of goods and services. This is to provide greater safety cushion and increased prospects for permanent exit from unsustainable debt. Faster debt relief starting from decision points Floating completion points; and A strong link between debt relief and poverty reduction 3 DRAFT NOT FOR CITATION In order to meet the last element of the Enhanced HIPC package, countries are expected to prepare national poverty reduction strategies. Uganda did accomplish this, and the international financial institutions and the donor community, accepted her strategy as a viable and comprehensive one. Uganda then became the first country to benefit from the enhanced HIPC initiative in April 2000. Annual debt relief is estimated at US$ 55 million in addition to the US$ 40 million from the original initiative. Table 1 summarises the Debt relief under the first HIPC and the enhanced HIPC initiative. Table 1 A. HIPC Initiative Multilateral Bilateral B. Enhanced HIPC Initiative Multilateral Bilateral Total Debt relief Debt Relief Under the HIPC Initiative Nominal Debt Service Relief Million US$ 650 1,300 HIPC Debt Relief, NPV Terms (Million of US$) 347 274 73 656 1,950 546 110 1,003 Completion Point April 1998 May 2000 Use of Debt Relief Resources from both the HIPC and the Enhanced HIPC are channeled to poverty reduction programmes through the Poverty Action Fund (PAF). At the advent of HIPC debt initiative, the Uganda Government established a Poverty Action Fund (PAF) through which savings made from debt relief would be channeled to finance poverty reduction programmes in accordance with the Poverty Eradication Action Plan (PEAP) and Poverty Reduction Strategy Paper (PRSP) priorities. Although the original intention was to create transparent mechanisms for ensuring that the resources saved from the HIPC initiative are channeled fully to poverty eradication programmes, the PAF has attracted additional donor support for poverty reducing programmes over and above the regular programmes of some donors. HIPC debt relief funds one third of the total resources in PAF. The PAF resources are spent on PEAP/PRSP priority programmes such as primary education, primary health care, agriculture extension, rural road programmes, rural water supply and sanitation services and enhanced accountability. The share of the PAF in total spending rose rapidly from 17% in 1997/98 to 31% in 2000/2001. Government has pledged to continue increasing this share over time. All PAF expenditures are an integral part if the governments budget, as envisaged by the PEAP/PRSP, which provides a framework for sector wide plans and investment programmes, which in turn are translated into actual budgets as contained in the Medium Term Expenditure Framework (MTEF). The contribution of savings from HIPC debt relief to the PEAP/PRSP priority spending on the provision of social services and support to activities aimed at enhancing the ability of the poor to increase their incomes, channeled through PAF are shown in Table 2 4 DRAFT NOT FOR CITATION Table 2: PEAP Expenditures (Billions of Ushs.) Total Expenditure (Excluding arrears) Of which: PAF HIPC % Contribution of PAF to Total expenditure % Contribution of HIPC to total expenditure % Contribution of HIPC to PAF resources Measures to increase incomes Rural Roads Implementation of the Land Act Agriculture Extension Micro Finance/ Restocking Programmes Measures to Improve Quality of Life Primary Health Care Water and Sanitation Primary Education Other District Grants Monitoring and Accountability Exchange rate UShs per US$ 1997/98 141.38 1998/99 231.84 1999/00 302.69 2000/01 439.8 2001/02 624.61 97.84 44.64 42.2 193.98 65.55 64.1 298.8 159.16 67.9 400.4 127.3 64.1 19.3 21.7 36.2 20.4 45.6 33.8 53.3 31.8 8.96 8.42 20.39 20.18 0.54 0.47 0.21 0.49 31.78 24.58 2.7 4.49 7.29 35.79 28.67 3.0 4.12 11.64 65.63 42.7 7.93 15.6 12.54 128.25 4.16 3.86 120.23 203.28 20.24 12.34 169.83 0.87 3.7 1150 7.68 1362 251.06 21.42 17.57 211.57 0.5 2.0 10.57 1512 349.96 57.17 35.33 254.66 2.8 24.71 17.67 1762 459.46 104.43 53.13 293.7 8.2 53.28 26.01 1773 Source: Ministry of Finance, Planning and Economic Development On average HIPC relief constituted 25% of total expenditure over the period 1998/99 to 2001/02 and 42% of PAF resources over the same period as indicated in Table 2. It is therefore clearly evident that the savings from the HIPC debt relief have significantly supplemented budgetary resources and have enabled the Ugandan Government to increase expenditure in poverty reducing sectors and this has contributed significantly to the poverty reduction efforts. Over time the PAF expenditures have been rising and estimates for 2001/02 are put at UShs.609 billion (approximately US$ 330 million). 3.0 Increased Government Expenditure and Impact on Macro Economic Variables The increase in government expenditure and the associated fiscal deficit has become a contentious issue in Uganda. The increased expenditure as a result of the HIPC initiative has been associated to an increasing fiscal deficit. In a paper “Is Budget Deficit Reduction Necessary”1 six measures of the deficit were highlighted including: • Overall deficit (including grants), defined as total revenue plus grants minus government expenditure which rose rapidly from 1.4% of GDP in 1997/98 to 5.7% of GDP in 2001/02 • Primary deficit similar to overall deficit (including grants) but excludes domestic and external interest payments from the computation of government expenditure which rose from 0.26% in 1998/99 of GDP to 4.6% in 2001/2002 1 Martin Brownbridge (2000) 5 DRAFT NOT FOR CITATION • • • • Overall deficit (excluding grants) is defined as domestic revenue minus expenditures and is a key deficit measure for Uganda rose from 6-7% of GDP but began to rise rapidly in 1998/99 and stood at 13% of GDP in 2001/02 Current deficit is current revenue minus current expenditure this deficit increased to 1.7% of GDP in 2001/2 which indicates that the domestic revenues could not even cover governments current expenditures. GOU Budget Deficit excludes the externally financed development expenditures and is defined as domestic revenue minus the GOU budget. This was less around 1 % of GDP between 1994/95 and 1997/98, but then increased to 7.3% of GDP in 2001/2 Domestic deficit is defined as the GOU deficit excluding the interest payments on external debt and government imports financed directly through the BOU but including arrears and PN payments. This deficit was less than 1% of GDP between 1993/94 and 1998/99. but then began to rise rapidly to 6.2% of GDP in 2001/2. An analysis of all six measures revealed that the fiscal deficit in Uganda has enlarged and the increase is largely attributable to the increased availability of donor aid, mainly in form of budget support, including debt relief, which has been used to expand the GOU budget. 4.0 Implications of Increased Government Spending on Macro economic variables The increase in Government spending since HIPC became operational in the late 1990’s and the associated rise in the fiscal deficit, certainly generated some important benefits in that debt relief had enabled GOU to increase expenditure in poverty reducing sectors and this has contributed significantly to the poverty reduction efforts. However it also created macro economic problems. This was because higher Government spending, even if funded by donor aid, creates liquidity in the domestic economy. This liquidity must be mopped up – the technical term is sterilized – by the Central Bank if it is not to lead to large inflationary money supply increases. As shown in Appendix table 1, the fiscal deficit almost doubled between 1997/98 and 2001/02. The widening of the fiscal deficit led to tenfold increase in the liquidity created by Government’s fiscal operations – that is Government expenditures in the domestic economy minus domestic revenues. Liquidity created by programmed fiscal operations increased from Ushs.68 billion in 1998/9 to Ushs.664 billion in 2001/2. This liquidity had to be mopped up by BOU if it was not to cause inflation using a combination of foreign exchange sales to the Inter-bank Foreign Exchange Market (IFEM) and net issued of treasury bills. Net foreign exchange sales also increased very rapidly over the last five years rising from US$ 5 million in 1998/99 to the programmed level of almost $254 million in 2001/2. However, actual foreign exchange sales were much lower at US$198 million, partly because of concerns by the BOU over the level of foreign reserves in the face of sharply lower than programmed disbursements of budget support, and because of concerns for the stability of the exchange rate and potential exchange rate appreciation. With foreign exchange sales being lower than programmed, TB sales had to increase to control the rise in base money. At the margin almost all additional TB issues have to be held by the commercial banks because non-bank demand for TB’s is very limited. The net issuance of Treasury Bills climbed very quickly from the late 1990’s from Ushs.37 billion in 1997/98 and by 2001/2, the net issuance of Treasury Bills increased to a massive Shs.268 billion leading to volatile treasury bill interest rates. 6 DRAFT NOT FOR CITATION Sales of foreign exchange and treasury bills in order to mop up excess liquidity caused by increased government expenditure have implications on macroeconomic variables and these are as follows; 1. Increased foreign exchange sales have inevitably appreciated the real exchange rate. This in turn has damaged the competitiveness of private sector exporters, which is crucial for boosting economic growth. As shown in appendix Table 1, the RER appreciated as sales of foreign exchange in the market increased. In addition, examining the trends in the price indices for the major components of GDP shows that prices for non-traded goods in Uganda have grown much faster than prices for traded goods as depicted by appendix Figures 1 (a) and (b). This implies that price incentives within the domestic economy have shifted away from traded goods production and towards non-traded goods production in the last few years. The increased fiscal deficits since 1997/98 took place at the same time that Uganda ‘s external terms of trade fell sharply, by 35% between 1997/98 and 2001/2, largely because of a fall in world market coffee prices. Compared to 1997/98 prices for exports and imports, the terms of trade shock cost Uganda around 6% of GDP in 2001/2, which is of roughly the same magnitude as the increase in government spending funded from increased donor aid. Therefore in terms of aggregate spending in the economy, the external terms of trade shock acted to offset the impact of the fiscal expansion, although the terms of trade shock and the fiscal expansion did not have symmetrical effects at the sectoral level. The aid funded fiscal expansion led to an increased trade deficit from 10.1% of GDP in 1997/98 to 15.6% of GDP in 2001/2. Because of the terms of trade loss however, the trade deficit measured in constant prices actually declined marginally between 1997/98 and 2001/2. We cannot ignore the fact that a shift in relative prices from tradable to non-tradable will undermine the GOU objective of creating a dynamic export led economy. Private sector led-export promotion is central to the Medium Term Competitiveness Strategy (MTCS) and this objective should not be compromised by an excessive fiscal deficit. Sterilization of the fiscal deficit through the sales of foreign exchange is not a sustainable policy since it would mean that the whole economy, and not just the Government budget, would inevitably become over-dependent on a continuous large inflow of donor funds intermediated through the Central Bank. This will affect the structure of both production and demand in the economy. Yet from experience, actual disbursements of donor aid usually fall short of what has been programmed – the reality over the last few years is that disbursements in each year have averaged less than 60% of what had been programmed at the start of that year. The reliance on donor funds to finance activities in Medium Term Budget Framework will be setting the economy up for a serious crisis if, for any reason, these donor funds did not materialize in full. Moreover, this crisis would not be confined to the Government budget, but it would spill over into the foreign exchange markets and into the wider economy. 2. The huge increase in Treasury Bill sales has also squeezed the funds available in the banking system for lending to the private sector – hence the fiscal deficit is directly crowding out private sector borrowing. Commercial Bank holdings of TB increased from 23% to 32% of their asset portfolio between June 2000 and June 2002. This was a period in which Private Sector Credit 7 DRAFT NOT FOR CITATION (PSC) slowed down over the period 1999/00 and 2001/2 combined while bank loans to the private sector fell from 30% to 23% of the banks asset portfolios. They may have been other reasons independent of the widening fiscal deficit, for the stagnation of PSC over the last two years such as weak demand for credit from credit worthy borrowers, but it is difficult to escape the conclusion that even if demand for private sector credit had been strong enough to support the programmed PSC growth, the banks would not have had the resources to supply this credit because of their increased holding of TB’s and therefore growth in PSC would have still been chocked off. Given the difficulty in selling the programmed volume of forex to the IFEM, the only way in which room could have been created for faster growth in PSC would be through a tighter fiscal stance, which would have allowed net TB issues to be lower, thereby filling up resources in the banks asset portfolios for more lending to the private sector. In the next section, empirical results showing how increased Government expenditure has affected export performance, the price of non-tradables and private sector credit is reviewed. 5.0 Increased Government Expenditure, Real Exchange Rate Appreciation and Export Performance. 5.1 Empirical Results The assessment of the impact of increased Government expenditure on export performance and the price of non tradables versus price of tradable goods, was done in a study by Atingi Ego and Sebudde(2000)2. An approach based on Edwards (1989) and El Badawi (1994) model of the real and equilibrium exchange rate was used to assess whether developments in the Real Effective Exchange Rate (REER) were in line with economic fundamentals and hence consistent with external viability of the economy. This required computing an exchange rate that evolves with the permanent component of fundamentals in the economy, which results in the simultaneous attainment of both internal and external equilibrium. REER was defined as the nominal exchange rate adjusted for the price differential between Uganda and her major trading partners. The fundamental variables that determine Uganda’s real exchange rate movements were defined as terms of trade, the degree of openness, government expenditure and sustainability of capital flows. Estimation A three-phase approach towards estimating the equilibrium exchange rate (ERER) was adopted. (i) estimation of the REER by cointegration and the Error Correction Method (ECM) approach, (ii) filtering transitory factors from the fundamentals and (iii) estimating the ERER using the permanent components of the fundamentals. The two filter methodologies employed are Hodrick Prescott, (HP 1984) approach and the Moving Average methodology used by El Badawi, 1994. The deviation is what constitutes the magnitude of misalignment in the exchange rate. Evidence provided by the time series of both the real exchange rate and the nominal effective exchange rate for the period 1970-1999, depicts the shilling generally depreciating except in the 1987/88 financial year and the 3 year period between 1992 and 1994. The real exchange rate (REER) has closely followed a similar trend. This evidence could, however, be misleading since a 2 Uganda’s Equilibrium Real Exchange Rate And Its Implications for Non-Traditional Export Performance 8 DRAFT NOT FOR CITATION real appreciation following a successful liberalization could either arise from changes in the economic fundamentals or from other non fundamental factors. However changes arising out of the latter lead to misalignment of the RER from its equilibrium. The real effective exchange rate has also been increasing showing the prices of non tradable has been increasing. The time series properties of the variables were investigated and order of integration determined using the Dickey Fuller(DF) and the Augmented Dickey Fuller (ADF) unit root test. And these are summarized in table 2 below. Table 2: Variable Time Series Properties of the Variables Sample Period: 1970-1999 Variable name ADF (X) ADF Order of X (∆X) Integration Real Effective Exchange LREER -2.6951 -4.1549 I (1) Rate Openness LOPEN -1.7911 -3.5776 I (1) Real Government LGE -2.8732 -3.5776 I (1) Expenditure Terms of trade LTT -1.6172 -4.1248 I (1) Foreign Exchange Inflows LFLOW -1.4011 -4.0543 I (1) Bank Credit Growth DLCREDIT -2.1581 -3.7777 I (1) Critical value at 5% level of significance -3.5796 -3.58671 Where: LREER, LOPEN, LGE, LTT, LFLOW and LCREDIT are respectively the logarithmic notions of REER, OPEN, GE, TT, FLOW and CREDIT and the letter L denotes the logarithmic notion of the variable as defined earlier Having established the order of stationarity, the I(1) variables were entered into the co integrating vector to establish the existence of estimating long-term relationship using the Johansen (1988) procedure. Using a VAR of 2 a long run relationship for REER was established as: LREER= 0.503*LOPEN – 0.238*LGE -0.201*LFLOW-0.83*LTT …………… (1) The results show that a 1% increase in openness will depreciate the REER by 0.503% while a 1% increase in foreign exchange inflows, terms of trade or government expenditure will exert appreciative pressure on the REER amounting to 0.201%, 0.83% and 0.238% respectively. (See appendix table 2.3 and 2.4 for details of other results) Going by appendix figure 2 we see that the level of misalignment in the REER reduced significantly as the Ugandan government undertook several trade reforms in the 1990’s. However, the appreciative pressures observed in the REER in the period 1977 to 1981, 1984 to 1988 and 1991 to 1996, could, to a large extent, be attributed to the significant improvements in the terms of trade arising from the improved unit export prices over this period. For the latter part of the 1990s the increased government expenditure and foreign exchange flows appear to have exerted appreciative pressures on the exchange rate. To the extent that increases in government expenditure exert appreciative pressures on the REER, it would appear that the increased government expenditure during the, 1990’s impacted appreciative pressures on the REER. 9 DRAFT NOT FOR CITATION The appreciation pressures on the REER emanating from increased government expenditure could be on account of the fact that the bulk of this expenditure is spent on non-tradables such education, health, water and sanitation, rural roads and including construction and the supply of some of these are fixed in the short to medium term. Consequently, their prices have arisen leading to appreciation pressures on the REER as shown in appendix figures 1 (a) and (b). On the other hand, the donor financing of this expenditures through increased flows has appreciated the exchange rate via increased foreign sales. As earlier mentioned, this results from the monetary authorities actions to sterilize the shilling liquidity created by increased expenditures through increased sales of foreign exchange. These results confirm that an increase in government expenditure as a result of HIPC relief funded by the donor community led to an increase in foreign flows have indeed exerted appreciative pressure on REER thereby indicating an increase in price of non-tradable goods relative to price of tradable goods. The levels of misalignments in the exchange rate3 shown in appendix figure 2 and the performance impact on non-traditional exports are summarized in appendix tables 2.8 (A-H). We perform a Granger causation test on these two variables and find that there is a unidirectional causation from misalignment in the exchange rate to non-traditional exports. Table 3: Cointegration for the non-traditional exports (NTE) and the level of the ERER (Moving Average) – VAR=4. A. Johansen Maximum Likelihood Test for cointegartion based on Maximal Eigenvalue Null Alternative Statistic r=0 r=1 18.5809 r<= 1 r=2 95% Critical Value 14.9000 1.8704 90% Critical Value 12.9120 8.1760 6.5030 B. Test for cointegration based on Trace of the Stochastic Matrix Null Alternative r = 0 r>= 1 Statistic 95% Critical Value 20.4513 17.9530 90% Critical Value 15.6630 r<= 1 r = 2 1.8704 8.1760 6.5030 C. Normalized Co-integrated Vectors (β-vector) and Adjustment Matrices (αvectors) (i) LNTE LREER -1.00000 0.53904 4.1114 -0.08797 Further, given that the Ugandan economy has experienced huge foreign exchange inflows, which our analysis has treated as a fundamental factor driving the equilibrium exchange rate, one would wonder whether such inflows have not had a negative impact on the export sector performance. To ascertain this, and whether the level of the equilibrium rate itself has any impact on the performance of non-traditional exports, an analysis of the performance of the export sector in relation to the movement in the equilibrium exchange rate itself has been done through cointegration and error correction. 3 measured as the deviation of the REER from the ERER 10 DRAFT NOT FOR CITATION As depicted by table 3, the single co-integrating relationship found confirms the fear that appreciation of the ERER negatively affects the export sector performance. The long run relation would thus be given as: LNTE = ……………………….(2) 4.1114*LREER This suggests that a 1% deprecation in the LREER increases non-traditional exports by 4.1%. From the results in equation (1) above, which depicts capital flows (FLOWS) to be negatively related to the REER, we can argue further that the increased foreign exchange flows which have appreciative pressures on REER, have not augured well with the non-traditional export sector. However, our short run analysis presented in Table 3, suggests that the changes in the REER has no significant policy implications on changes in non-traditional exports. It is current changes in non-traditional exports depend largely on their past values and the impact of the long run presented by the ECM term. Table 12: Ordinary Least Squares Estimation with Changes in ERER(HP method) Dependent variable is DNTE 16 observations used for estimation from 1984 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT .40844 .15498 2.6354[.023] DNTE(-1) -.24606 .17685 -1.3913[.192] DNTE(-2) -.51526 .17935 -2.8729[.015] DNTE(-3) -.34472 .20343 -1.6946[.118] ECM(-4) -.40252 .18100 -2.2239[.048] R-Squared .46019 F-statistic F( 4, 11) 2.3444[.119] R-Bar-Squared .26389 S.E. of Regression .41301 Residual Sum of Squares 1.8764 Mean of Dependent Variable .16364 S.D. of Dependent Variable .48139 Maximum of Log-likelihood -5.5571 DW-statistic 2.4252 CONCLUSION This paper empirically suggests that the HIPC Debt initiative has had an impact on macro economic variables in Uganda. Increased government expenditure financed by increased donor flows has appreciated the REER rate by magnitudes of 0.24 and 0.20 percent respectively for every one percent increase in them. On the other hand, for every one percent appreciation on the real exchange rate has seen the performance of non-coffee exports deteriorate by 4 percent. There are some lessons for poor countries to learn as they look forward to accessing the millennium development goals (MDG’s) resources. Increased government expenditure on account of MDG resources are likely to create and complicate macro-economic management of the resultant higher interest rates and appreciating exchange rate, thereby crowing out the private sector which could compromise the governments policy of the private sector export led growth in reducing poverty. Developing countries hoping to access the MDG resources should therefore do so carefully by accessing only those quantities that will not fully crowd out their private sectors and compromise country efforts to generate sustainable external sectors while ensuring increased absorption for the resources. The latter will lead to lowering the cost of doing business while maintaining the export competitiveness of their countries. 11 DRAFT NOT FOR CITATION References Adlung and Carzaniga (2000), Trade in Health Services under GATS- Past and Future Atingi Ego (2002), The Role of the HIPC Debt Relief Initiative in bridging the Resource Gap for the Financing of Poverty Reduction Programmes and the Role of SMES in PovertY Reduction: The Case of Uganda Atingi Ego and Sebudde (2000), Uganda’s Equilibrium Real Exchange Rate and ITS Implications for Non-Traditional Export Performance Bank of Uganda (2002), Private Capital Flows Survey 2001 Bank of Uganda (2002, 2001), Annual Supervision Report Bank of Uganda (2002, 2001,2000), Annual Report Edward S (1989). Real Exchange Rates, Devaluation and Adjustment: Exchange Rate Policy in Developing Countries. Cambridge, MA: MIT El Badawi I(1994) Estimating long run Equilibrium Exchange Rates published in Williamson J. (1994) Ch. 5 Estimating Equilibrium Exchange Rates Muwanga Zake and Ndhaye (2001), The HIPC Debt Relief Initiative Uganda’s Experience Nanyonjo J (2001). The HIPC Debt Initiative: Uganda’s Social Sector Reforms and Outcomes UNU/WIDER Discussion Paper no. 2001/138 Republic of Uganda (2000). Poverty Eradication Action Plan, PEAP. Republic of Uganda(1996/97-2002/2003). Background to the Budget Republic of Uganda(2002). Poverty Reduction Strategy Paper (PRSP) Progress Report Republic of Uganda(2002). Statistical Abstract 12 DRAFT NOT FOR CITATION Appendix 1. TABLE 1: Government Deficit, Liquidity Creation and Sterilisation, Real Exchange Rate and Terms of Trade Fiscal Deficit (excluding Grants) (as % of GDP) Liquidity Created by Fiscal Operations (Billions of Shillings) Foreign Exchange Sales by BOU US$ Millions Net Treasury Bill Issues (Billions of Shillings) Stock of TB’s (Billions of Shillings) Growth in Private Sector Credit (%) Base Money-start of the year (Billions of Shillings) Interest Cost of Domestic (Billions of Shillings) Terms of Trade* (% change) Real Exchange Rate (% change) 1997/98 6.7 1998/99 7.8 1999/00 11.2 2000/01 11.5 63 58 212 418 664 5 25 117 174 198 37 36 133 168 268 112 148 281 449 717 24.8 332 33.6 373 6.2 433 9.4 442 4.2 550 29.1 21.4 29.6 57 91.5 98.2 -2.6 140.4 21.0 115.2 17.4 129.4 -7.8 114.5 -0.6 100 -22.7 121.3 5.9 74.3 -25.7 121.0 -0.3 61 -17.9 *- Terms of trade define as the relative prices of non-trade versus traded goods. 13 2001/02 12.8 DRAFT NOT FOR CITATION Table 2: Composition of External Debt and Debt Indicators Total External Debt Stock (millions of US$) Multilateral Debt (millions of US$) Bilateral Paris Club (millions of US$) Bilateral Non-Paris Club (millions of US$) Commercial Banks (millions of US$) Commercial Non Banks (millions of US$) Other Loans (millions of US$) Debt Service (millions of US$)** Accumulation of arrears out of current maturities (millions of US$) Exports of Goods (millions of US$) Exports of Goods and Non-factor Services (millions of US$) Debt Indicators (Percentage) Multilateral debt as a % of total debt stock Bilateral Paris Club as a % of total debt stock Bilateral Non-Paris Club as a % of total debt stock Commercial Banks as a % of total debt stock Commercial Non Banks as a % of total debt stock Other Loans as a % of total debt stock Total External debt as a % of export of goods & services Total External debt as a % of export of goods Total external debt as a % of GDP Total debt service as a % of export of goods and services Total debt service as a % of GDP 1992/93 2637.2 1993/94 2999.3 1994/95 3386.9 1995/96 3515.8 1996/97 3660.2 1997/98 3664.4 1998/99 3499.6 1815.9 281.8 415.6 17.1 62.0 44.8 169.7 46.2 2156.1 332.0 398.4 1.3 38.2 73.4 167.8 17.1 2487.9 380.2 407.8 7.7 27.2 76.2 128.1 19.9 2655.2 350.6 404.1 3.0 26.2 75.9 142.2 35.3 2727.6 346.0 417.3 1.0 26.2 74.7 155.9 20.8 2826.9 324.5 461.8 2.5 26.7 22.0 166.5 27.4 2782.6 288.3 361.7 5.4 35.3 26.3 152.9 52.8 172.8 253.5 264.7 200.3 593.0 664.8 588.0 723.3 683.5 837.5 458.4 633.9 549.2 734.4 68.9 10.7 15.8 0.6 2.4 1.7 1040.3 1526.6 87.4 71.2 5.2 71.9 11.1 13.3 0.0 1.3 2.4 1497.4 1133.2 80.5 42.8 3.6 73.5 11.2 12.0 0.2 0.8 2.2 509.5 571.2 64.2 26.4 2.7 75.5 10.0 11.5 0.1 0.7 2.2 486.1 597.9 64.0 23.4 2.6 75.9 9.6 11.6 0.0 0.7 2.1 437.0 535.5 64.3 19.8 2.6 77.1 8.86 12.6 0.07 0.73 0.6 578.1 792.1 58.8 27.4 2.7 79.5 8.2 10.3 0.2 1.0 0.8 476.0 636.6 60.4 24.2 2.0 *Projection Source: Bank of Uganda and Ministry of Finance Planning and Economic Development 14 199 DRAFT NOT FOR CITATION 15 DRAFT NOT FOR CITATION Figure.1(a) 180.0 170.0 160.0 150.0 140.0 130.0 120.0 public construction private construction 110.0 100.0 90.0 1997/98 1998/99 1999/2000 2000/01 2001/02 2002/03 Figure 1(b) 140.0 130.0 120.0 Non traded/traded goods (prod) non traded/traded goods (exp) 110.0 100.0 90.0 1997/98 1998/99 1999/2000 16 2000/01 2001/02 2002/03 DRAFT NOT FOR CITATION Appendix Table 2.2: Uganda’s Nominal and Real Effective Exchange Rates Year NEER 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 REER 0.16 0.16 0.16 0.19 0.22 0.26 0.28 0.50 0.73 1.26 2.30 4.18 5.12 6.13 7.62 Year 134.17 177.03 179.74 161.60 144.13 120.82 97.39 125.34 94.60 94.20 125.95 193.79 187.41 179.97 143.18 NEER 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 14.45 31.44 100.00 234.57 471.69 923.39 1525.73 2111.29 2031.16 1836.05 1838.84 1855.93 1940.08 2417.42 REER 114.96 91.01 100.00 83.97 103.73 183.96 252.33 258.66 290.78 276.70 257.30 255.67 266.84 337.93 AppendixTable 2.3: Johansen and Juselius Cointegration Procedure Results VAR = 2; Sample Period: 1970-1999 Additional I(0) variables in VAR: DDLCREDIT Null Alternative -T.ln(1-µi) 5%c.v 10% c.v Σ -T.ln(1-µI) 5%c.v. 10% c.v r=0 r≤1 r≤2 r≤3 r≤4 r =1 r =2 r =3 r =4 r =5 30.41 22.83 14.72 7.27 1.67 33.46 27.07 20.67 14.07 3.76 30.90 24.73 18.59 12.07 2.69 76.90 46.50 23.66 8.94 1.67 68.52 47.21 29.68 15.41 3.76 64.83 43.94 26.79 13.33 2.69 Appendix Table 2.4: Estimated Normalized Co-integrated Vectors (β-vectors) and Adjustment Matrices (α-vectors) in the Johansen Estimation Lreer (i) (ii) -1.000 0.030 -1.000 0.536 lopen -27.164 0.057 0.503 0.166 lge -1.707 0.00345 -0.238 0.192 lflow ltt 24.797 -0.917 -0.048 0.056 -0.201 -0.831 0.181 -0.042 The long-run relationship for REER is consequently derived from vector (ii) and is expressed as: lreer = 0.503*lopen - 0.238*lge - 0 .201*lflow – 0.831* ltt…..…(14) 17 DRAFT NOT FOR CITATION Appendix Table 2.5: Over-parameterized Model by Ordinary Least Squares Estimation Ordinary Least Squares Estimation Dependent variable is DREER 28 observations used for estimation from 1972 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT ( C ) 1.1159 .28512 3.9136[.001] DREER(-1) -.32737 .20670 -1.5838[.133] DOPEN .69081 .30527 2.2629[.038] DOPEN(-1) .39689 .35056 1.1322[.274] DGE -.13547 .18216 -.74369[.468] DGE(-1) -.022452 .14249 -.15756[.877] DTT -.068505 .13207 -.51869[.611] DTT(-1) -.21167 .13140 -1.6109[.127] DFLOW -.16009 .25799 -.62053[.544] DFLOW(-1) -.016842 .25252 -.066698[.948] DDLCREDIT .12331 .093630 1.3170[.206] ERROR(-2) -.44979 .11462 -3.9243[.001] R-Squared .76953 F-statistic F(11, 16) 4.8566[.002] R-Bar-Squared .61108 S.E. of Regression .16034 Residual Sum of Squares .41133 Mean of Dependent Variable .028457 S.D. of Dependent Variable .25710 Maximum of Log-likelihood 19.3576 DW-statistic 1.5246 Durbin's h-statistic *NONE* * Test Statistics * A:Serial Correlation * B:Functional Form * C:Normality * D:Heteroscedasticity Diagnostic Tests * LM Version *CHI-SQ( 1)= 3.2207[.073] *CHI-SQ( 1)= 5.4258[.020] *CHI-SQ( 2)= 2.7136[.257] *CHI-SQ( 1)= 1.0650[.302] * F Version *F( 1, 15)= 1.9496[.183]* *F( 1, 15)= 3.6053[.077]* * Not applicable * *F( 1, 26)= 1.0281[.320]* Appendix Table 2.6: Variable Deletion Test (OLS case) Dependent variable is DREER List of the variables deleted from the regression: DTT DGE DGE(-1) DFLOW 28 observations used for estimation from 1972 to 1999 Regressor Coefficient Standard Error INPT ( C ) 1.0969 0.25855 DREER(-1) -0.36152 0.18157 DOPEN 0.46081 0.11820 DOPEN(-1) 0.37497 0.15672 DTT(-1) -0.18870 0.10590 DDLCREDIT 0.15544 0.078986 ERROR(-2) -0.43861 0.10379 Joint test of zero restrictions on the coefficient of deleted variables: Lagrange Multiplier Statistic CHI-SQ( 5)= 3.9695[.554] Likelihood Ratio Statistic CHI-SQ( 5)= 4.2807[.510] F Statistic F( 5, 16)= .52860[.751] 18 * DFLOW(-1) T-Ratio[Prob] 4.2425[.000] -1.9911[.060] 3.8987[.001] 2.3926[.026] -1.7818[.089] 1.9679[.062] -4.2262[.000] DRAFT NOT FOR CITATION Figure 2: Uganda's Equilibrium Exchange Rate(ERER) and Real Exchange Rate(RER) 1975-1999 6 5.5 5 4.5 4 Year reer erer(MA) 19 erer(HP) 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 1979 1978 1977 1976 3.5 1975 Exchange Rate (in logs) 6.5 DRAFT NOT FOR CITATION Appendix Table 2.7: Year Non-Traditional Exports and Exchange Rate Misalignments Total Exports (mil US$) Coffee Exports NonDegree of Misalignment(%) Traditional Moving Hedrick(mil US $) Exports average Prescott Filter (mil US $) 1980 319.4 312.2 2.5 57.3 45.4 1981 229.3 224.2 2.5 15.5 22.2 1982 347.3 341.7 1.4 11.2 8.9 1983 367.7 340.2 14.2 12.6 1.5 1984 407.3 375.6 14.8 38.5 26.9 1985 347.8 325.8 6.62 78.8 47.9 1986 394.9 382.2 4.5 90.4 72.2 1987 333.6 321.0 6.6 73.0 61.2 1988 266.3 259.2 1.016 79.9 67.7 1989 277.7 262.8 7.1 48.6 32.9 1990 177.8 140.4 25.23 19.4 -5.1 1991 173.2 117.6 32.5 8.7 -26.7 1992 151.2 95.4 35.6 3.7 -36.9 1993 200.0 114.5 61.7 11.7 -29.7 1994 462.9 356.9 85.2 4.7 -24.9 1995 560.3 419.0 115.2 -11.0 -17.5 1996 639.3 396.1 211.1 -14.6 -12.2 1997 575.6 310.4 209.6 -5.4 4.2 1998 510.2 295.0 161.9 -14.5 -0.2 1999 536.9 285.5 194.7 -18.7 -1.9 Source: Bank of Uganda Annual Reports and computations by authors Note: Degree of misalignment ((+) is over-valuation and (-) is under-valuation. 20 DRAFT NOT FOR CITATION Table 2.8(A – H): Results of the Co-integration and , error correction analysis for Nontraditional Exports and Exchange Rate Misalignments 2.8A. Time Series Properties of Over-valuation and Non-traditional exports Variable Variable name ADF(x) ADF(∆x) Over-valuation by MA Over-valuation by HP Non-Traditional Exports/Total Exports ratio Critical Value at 5% OVER(ma) OVER(hp) NTE -1.3396 -2.4049 -3.6386 -3.1046 2.9970 2.9970 Order of cointegration I(1) I(1) I(1) 2.8B. Johansen and Joselius Cointegration with restricted intercepts and no trends in the VAR Cointegration LR Test Based on Maximal Eigenvalue of the Stochastic Matrix 17 observations from 1983 to 1999. Order of VAR = 3 List of variables included in the cointegrating vector: List of variables included in the cointegrating vector: LNT LOVER(ma) Intercept LNT LOVER(hp) Intercept List of eigenvalues: 0.56132 0.32141 0.0000 List of eigenvalues : 0.60391 0.34111 0.0000 Null Alternative Statistic R=0 R<= 1 r=1 r=2 95% C.V. 90% C.V. Null 14.0078 15.8700 13.8100 R=0 6.5917 9.1600 7.5300 Alternative Statistic R<= 1 r=1 r=2 95% C.V. 90% C.V. 15.7438 15.8700 13.8100 7.0923 9.1600 7.5300 2.8C. Johansen and Joselius Cointegration with restricted intercepts and no trends in the VAR Cointegration LR Test Based on Trace of the Stochastic Matrix 17 observations from 1983 to 1999. Order of VAR = 3. 17 observations from 1983 to 1999. Order of VAR =3. List of variables included in the cointegrating vector: List of variables included in the cointegrating vector: LNT LOVER(hp) Intercept LNT LOVER(ma) Intercept Null Alternative Statistic 95% C.V 90% C.V Null Alternative Statistic 95% C.V 90% C.V R=0 r>= 1 22.8361 20.1800 17.8800 R=0 R<= 1 r=2 7.0923 9.1600 R<= 1 2.8D. 7.5300 r>= 1 20.5994 r=2 6.5917 20.1800 17.8800 9.1600 Estimated Cointegrated Vectors in Johansen Estimation (Normalized in Brackets) Cointegration with restricted intercepts and no trends in the VAR Vector β α LNTE -1.000000 β α -1.0000 0.102 LOVER(ma) -5.6711 LOVER(hp) -7.2596 - Intercept -1.1938 Intercept -1.6526 - 21 7.5300 DRAFT NOT FOR CITATION 2.8E. Ordinary Least Squares Estimation of Over-parameterized Model(HP) Dependent variable is DNT 17 observations used for estimation from 1983 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT .52487 .23213 2.2612[.045] DNT(-1) -.46660 .23773 -1.9627[.075] DNT(-2) -.55844 .32363 -1.7256[.112] DHP(-1) -4.8434 2.5958 -1.8658[.089] DHP(-2) 2.3976 1.7727 1.3525[.203] ECM2(-3) .10208 .23152 .44091[.668] R-Squared .46272 R-Bar-Squared 0.21851 S.E. of Regression .85372 F-stat. F( 5, 11) 1.8947[.175] Mean of Dependent Variable .26467 S.D. of Dependent Variable .96572 Residual Sum of Squares 8.0172 Equation Log-likelihood -17.7332 Akaike Info. Criterion -23.7332 Schwarz Bayesian Criterion -26.2328 DW-statistic 1.6885 Diagnostic Tests * Test Statistics * LM Version * * A:Serial Correlation*CHSQ( 1)= .36405[.546]*F( 1, 10)= .21883[.650] * B:Functional Form *CHSQ( 1)= 3.3530[.067]*F( 1, 10)= 2.4570[.148] * C:Normality *CHSQ( 2)= 1.7471[.417]* Not applicable * D:Heteroscedasticity*CHSQ( 1)= .23669[.627]*F( 1, 15)= .21180[.652] F Version 2.8F. Variable Deletion Test (OLS case) and Ordinary Least Squares Estimation - HP Dependent variable is DNT List of the variables deleted from the regression: DHP(-2) ECM2(-3) 17 observations used for estimation from 1983 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT .43709 .22058 1.9815[.069] DNT(-1) -.47696 .23388 -2.0393[.062] DNT(-2) -.57230 .26610 -2.1507[.051] DHP(-1) -3.9094 1.6270 -2.4028[.032] Joint test of zero restrictions on the coefficients of deleted variables: Lagrange Multiplier Statistic CHSQ( 2)= 2.4908[.288] Likelihood Ratio Statistic CHSQ( 2)= 2.6933[.260] F Statistic F( 2, 11)= .94418[.418] R-Squared .45323 R-Bar-Squared .27097 S.E. of Regression .82457 F-stat. F( 4, 12) 2.4868[.099] Mean of Dependent Variable .26467 S.D. of Dependent Variable .96572 Residual Sum of Squares 8.1589 Equation Log-likelihood -17.8821 Akaike Info. Criterion -22.8821 Schwarz Bayesian Criterion -24.9651 DW-statistic 1.7094 22 DRAFT NOT FOR CITATION 2.8G. Ordinary Least Squares Estimation with Over-valuation under MA method Dependent variable is DNT 17 observations used for estimation from 1983 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT 0.35079 .24706 1.4198[.181] DNT(-1) -0.37425 .24447 -1.5309[.152] DNT(-2) -0.42294 .26471 -1.5977[.136] DMA(-1) -3.7375 2.0915 -1.7870[.099] DMA(-2) -0.38659 1.7986 -.21494[.833] Diagnostic Tests * Test Statistics * LM Version * A:Serial Correlation*CHSQ( 1)= 2.7485[.097] * B:Functional Form *CHSQ( 1)= .85294[.356] * C:Normality *CHSQ( 2)= .12499[.939] * D:Heteroscedasticity*CHSQ( 1)= .10728[.743] * F Version *F( 1, 11)= 2.1214[.173] *F( 1, 11)= .58106[.462] * Not applicable *F( 1, 15)= .095261[.762] 2.8H. Variable Deletion Test (OLS case) and Ordinary Least Squares Estimation - MA Dependent variable is DNT List of the variables deleted from the regression: DMA(-2) 17 observations used for estimation from 1983 to 1999 Regressor Coefficient Standard Error T-Ratio[Prob] INPT .36394 .23043 1.5794[.138] DNT(-1) -.36946 .23435 -1.5765[.139] DNT(-2) -.42165 .25475 -1.6551[.122] DMA(-1) -3.8415 1.9587 -1.9612[.072] Joint test of zero restrictions on the coefficients of deleted variables: Lagrange Multiplier Statistic CHSQ( 1)= .065199[.798] Likelihood Ratio Statistic CHSQ( 1)= .065324[.798] F Statistic F( 1, 12)= .046200[.833] ----------------------------------------------------------------------------------------------------------------------------------------------R-Squared .29848 R-Bar-Squared .13660 S.E. of Regression .89735 F-stat. F( 3, 13) 1.8438[.189] Mean of Dependent Variable .26467 S.D. of Dependent Variable .96572 Residual Sum of Squares 10.4680 Equation Log-likelihood -20.0004 Akaike Info. Criterion -24.0004 Schwarz Bayesian Criterion -25.6668 DW-statistic 1.4041 For all diagonistics, the following signs stand for : A: Lagrange multiplier test of residual serial correlation B: Ramsey's RESET test using the square of the fitted values C: Based on a test of skewness and kurtosis of residuals D: Based on the regression of squared residuals on squared fitted values 23