The macroeconomics of public sector deficits : the case of Ghana

Pollcy, Research, and External Affairs | WORKING PAPERS Macroeconomic Adjustment L and Growth Country Economics Department The World Bank May 1991 WPS 672 sa The Macroeconomics of Public Sector Deficits The Case of Ghana Roumeen Islam and Deborah L. Wetzel In developing countries, fiscal policy - in particular, the reduc- tion of public sector deficits - has been a key element of stabilization and adjustment programns. An empirical analysis of fiscal deficits in Ghana, where they have been a prominent feature, reveals their significant effects on both the real and financial sides of the economy. ThePolicy. Research, and Extenal Affair Cornplex distibutes PRE Working Paprs todisseaminate thefindings of wodk in progress and to encourage the exchange of ideas among Bank aaff and aU others intrested ir developmnent issues. These papers cany the names of the authors, reflect only their views, and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own. They should not be attributed to the World Bank, its Board of Directors, its management, or any of its member countries.  Policy, Research, and External Affalrs Mcocnombc Adjustment and Growth This paper - a product of the Macroeconomic Adjustment and Growth Division, Country Economics Department - is part of a PRE research project on "The Macroeconomics of the Public Sector Deficit" (RPO 675-31). Copies are availablefree from the World Bank, 1818 E Street NW, Washington, DC20433. Please contact Raquel Luz, room Ni 1-057, extension 34303 (170 pages, with figures and tables). Ghana's economic program after independence external lending was unavailable until 1984. emphasized public investment and spending as This policy led to high inflation, negative real the road to growth, a strategy that led to recur- interest rates, an overvalued currency, and the ring fiscal deficits and declining growth. By emergence of black markets. These forces 1983, per capita income was 10 percent lower further eroded the tax base and ultimately than in 1957. Since the 1984 Economic Recov- increased the deficit. ery Program, Ghana's fisco' deficits have declined and the public seLor has been rational- The authors also find that high levels of ized. Average growth rates have become inflation, combined with government restrictions positive. on private currency holdings, affect the demand for assets in Ghana, leading to a Laffer curve Islam and Wetzel provide two different effect in goveinment seigniorage: After a certain definitions of the fiscal deficit in Ghana. The point, an increase in the inflation rate actually first, more conventional approach aggregates the causes a reduction in seigniorage revenue. Yet components of the public sector, including the given the government's dependence on monetary central government, the social security and finance, reduced seigniorage meant more money national insurance trust, state-owned enterprises, creation and higher inflation. and the cocoa marketing board. However, because of the lack of data, this method of According to Islam and Wetzel: treating the deficit may understate its true value. *The fiscal deficit has had only little effect The second way looks at the total financing on private consumption; lagged consumption and flows to the public sector. Data on the central disposable income were more important. government debt are supplemented with data on the claims of the central bank and banking * Public sector investment in Ghana has system against state-owned enterprises and data mostly substituted for private investment. The on public external debt. current program of divestiture of state-owned enterprises shouli lead to an increase in private Islam and Wetzel examine the ways Ghana investment. chose to finance its deficits and how these affected the financial side of the economy. They * The fiscal deficit had a significant negative find that before implementation of the adjust- effect on the external side. The official real ment program of 1983, the government relied exchange rate tended to appreciate, the trade mainly on money creation for financing, though balance worsened, and the black market pre- this was more by default than by choice since mium rose. The PRE Working Paper Series disseminates the findings of work under way in the Bank's Policy, Research, and Extemal Affairs Complex. Anobjective of the series is to get these findings outquickly, even if presentations are less than fullypolished. The findings, interprctations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the PRE Dissemination Center  TABLE OF CONTENTS I. Introduction 1 II. Measuring the Deficit 8 The conventional approach to the public sector deficit 14 The consolidated public sectcr deficit 15 The central government 18 Social Security and the National Insurance Trust 25 The state-owned enterprises 27 The central bank 32 Defining the public sactor deficit based on its financing 34 Economic and policy determinants of public sector deficits 38 III. Fiscal Deficits and Financial Markets 51 Financing the deficit 53 Inflation, the demand for assets, and seignorage 63 The model 63 The demand for money and quasi-money 66 Simulation results 72 Conclusion 78 IV. Fiscal Deficits and Private Consumption and Investment 79 Fiscal deficits and private consumption 83 Fiscal deficits and private investment 88 V. Fiscal Deficits and External Sector 99 The model 114 The steady state 125 Empirical evidence 139 VI. Conclusions 150 References 165 Appendix 170 This is a revised version of a paper prepared for the World Bank Research Project 675-31 and presented at the World Bank conference on the "Macroeconomics of the Public Sector Deficit" held on July 11-13, 1990. We are grateful for the comments and sugrestions of William Easterly, Klaus Schmidt-Hebbel, Vikram Nehru ar.d the conference participants.  I. INTRODUCO>QO A pervas *e macroeconomic phenomenon in both industrial and developing countries has been the growth and persistence of fiscal deficits. The issues surrounding fiscal deficits are certainly not new, but the economic developments of the past decade have led to renewed interest in fiscal themes. In the developed countries, the growth of the U.S. federal deficit provided the impetus for a reassessment of the effect of fiscal deficits on economic activities. In the developing countries, fiscal policy, particularly the reduction of fiscal deficits, has been one of the cornerstones of short-term stabilization and medium-term adjustment programs. The macroeconomic theory concerning fiscal deficits has undergone a considerable transformation since Keynes emphasized fiscal policy in his General Theory. Rational expectations and the proposition of Ricardian equivalence put the effectiveness of traditional demand ranagement policies into question. Much of the discussion over fiscal policy in the past fifteen years has addressed the question of whether fiscal deficits and the way in which they are financed will have an effect on economic activity. While the theoretical debate continues, in practice fiscal deficits continue to be an important issue. In developing countries in particular, economic programmes in recent years have emphasized demand reduction via reductions in fiscal deficits. The underlying assumption is that fiscal  2 deficits will have an effect on demand. The idea that fiscal deficits are something that can be measured and managed is implicit in that assumption, yet as experience in both developed and developing countries shows, deficits may not be so easy to measure nor to control. This study presents an empirical investigation of fiscal deficits in Ghana and assesses their effect on the financial and real side of the economy. The case of Ghana is interesting because of the major role that fiscal deficits have played in her economic history. At the moment of independence Ghana was perceived to be at the forefront of the African countries. Her population was well educated and economic infrastructure was strong. Under Nkrumah, the first government led a statist economy that emphasized public investment and public spending as the road to economic growth. As time went on, governments became more and more profligate and deficits increased. Over the same time period, economic conditions degenerated. In 1983, the nadir was reached -- per capita income was over ten percent lower than its 1957 level.1 In 1984, the government initiated the Economic Recovery Programme (ERP) in an attempt to rescue Ghana's economy. An important component of this programme has been the reduction of fiscal deficits and the "rationalization" of the public I For an interesting review of Ghana's economy since independence see Rimmer (1989).  3 sector.2 The ERP is still in place and fiscal deficits have been reduced relative to their levels in the 1970s. Average annual growth of real GDP has averaged 6 percent in the post- 1983 period as compared to -3.4 percent in the period from 1978 to 1983. Given the multitude of factors at work in the Ghanaian economy, it is difficult to attribute the decline and renewal of the economy to any one factor. We will instead try to evaluate the impact that fiscal deficits may have had on certain macroeconomic variables which may in turn have affected Ghana's economic performance. The study is set out in three sections. In the first section, we consider various measures of the fiscal deficit in Ghana. An ideal measure would include all components of the public sector including central, state and local government, decentralized agencies, state-owned enterprises and even the central bank. Although the data for all these levels of the public sector are not available, we construct a measure of the consolidated fiscal deficit from the available information. We then construct an alternative measure of the fiscal deficit b>ssed on the flow of funds to the public sector. In the second part, we focus on fiscal deficits and financial markets. We first consider how the way Ghanaian deficits have been financed may influence certain 2 Other components of the programme are devaluation of the exchange rate, liberalization of the trade regime and , more recently, financial reform.  4 macroeconomic variables, particularly inflation. We then assess how inflation affects demand for money and quasi-money. A simple model i6 constructed that allows us to assess the relationsh:? between fiscal deficits, inflation, and seignorage. The third part of the study addresses the effect of fiscal deficits on private consumption ( NB: private investment is discussed in the companion paper by Ms. Islam). The relevance of the variouRa theories concerning the effect of fiscal deficits on private sector demand is considered by estimating a consumption function. The relationship between fisca1 deficits and the external sector (the real exchange rate and the trade balance) is also considered in the companion paper. The empirical nature of the study calls for a caveat. Generally, the data for developing countries is not always as complete and as consistent as developed country data. Data difficulties and limitations are particularly acute in African countries and Ghana is no exception. z. general, the most reliable data sources available were used, but there is no guarantee that the measurement of any given variable was consistent over time. Given the severe lags in accounting and general instability in the Ghanaian economy during the period that we cover (1970-1988), it is quite likely that there is a fair amount of "noise" in the data used, which may affect our regression results. In addition, because of data limitations,  5 we tend to use fairly simple equations for estimLtion. While thase equations may not represent the frontier of econometric modelling, they do let us glean some information about the basic relationships betweent fiscal deficits and other macroeconomic variables. The principal conclusions of our anal;sis are as follows: 1.) Because of the lack of data, the conventional approach to treating the deficit may present a misleading picture in Ghana. Determining the deficit using financing flows to the public s3ctor provides an alternative measure of the claims that the public sector makes on resources that differs considerably from the conventional measure. Using either measure, Ghana has had significant public sector deficits over the past two decades. 2.) Lack of access to external lending and weak domestic financial markets imply that the bulk of the public sector deficit was financed by money creation. AcceQs to external lending allowed the government to substitute foreign financing for money creation after 1984, but money creation did not fall as much as one would expect. In the mid-late seventies, the correlation between money creation and inflation was high, suggesting that fiscal deficits contributed to the high levels of inflation at the time. This relationship is less clear cut in the eighties, but this is not surprising given some  6 substantial supply shocks during the period, 3.) Inflation has a significant, negative Influence on the demand for money and the demand for deposits. The domestic interest rate was not found to have any significant effect on either the demand for money or the demand for deposits. One possible explanaticn of these results is the highly regulated nature of Ghana's financial markets until recently. 4.) A simple model set out in section II indicates that high rates of inflation provided the Ghanaian government with seignorage revenue only 1 to 2 percent of GDP greater than it would have received with an inflation rate of 20 percent. Since 1984, the level of seignorage has not differed greatly from the level that would have existed at a 20 percent rate of inflation. The model also suygests that had the government not appropriated private currency in the currency conversion of 1979, seignorage revenue would have been higher throughout the period. 5.) Lagged consumption and disposable income were found to be the principal determinants of private consumption. The results of this estimation indicate that neither pure Ricardian equivalence nor the pure Keynesian theory 'nold. The significance of disposable income indicates that liquidity constraints affect consumption decisions. Numerous fiscal  7 variables were tested as set out in the research proposal, but none were found to be significant. 6.) Private investment was found to be negatively affected by public investment, thus indicating that some crowding-out did occur in Ghana. Private investment is also positively influenced by a relaxation of credit constraints and by corporate tax revenues. (This suggests t.-.at corporate tax revenues might be highly correlated with corporate profits.) The real interest rate was not found to have a significant influence on private investment. 7.) Public expenditure has had a significant impact on the official real exchange rate, the trade balance, and on the black market premium. Increased public expenditure was found to have a negative effect on the trade balance. It was also found to have a negative effect on the official exchange rate (higher public expenditures tended to appreciate the real exchange rate). The empirical results also show that a rising public sector deficit along with stringent restrictions on foreign exchange transactions lead to a very high black market premium.  8 XI. MEASURING THE DEFICIT Implicit in the notion of targeting the fiscal surplus or deficit3 is the idea that one is able to obtain a reasonable measure of the deficit. This is not necessarily as straightforward as it might seem. Keynesian theory initially vtewed the deficit as exogenous. It soon became clear that the deficit was actually endogenous, given talat tax revenres and government expenditures are partly determined by the level of economic activity. Various r7easures of the deficit have heen developed to adjust for cyclical movements in the economy and to incorporate the effects of certain macroeconomic variables (particularly inflation). As ir normally the case, the most appropriate measure of the deficit depends upon the purpose of the analysis. Tanxi and Blejer (1984) discuss what they call the "conventional deficit" with reference to the definition set out in the International Monetary Fund's "Draft Manual on Government Finance Statistics". This measure arranges the payments and receipts of the government sector accounts as follows: Fiscal Deficit - (Revenue + Grant,' - (Expenditure on Goods 3 For the rest of the discussion we will refer only to deficits as they are generally the more common phenomenon.  9 and Services + Transfer Payments + Net Lending), or alternatively, Fiscal Deficit = Borrowing + Net Decrease in Cash Uloldings - Amortization This approach does have its shortcomings. First, the definition emphasizes cash flow concepts rather than accrual concepts of accounting. At times, the cash flow concept may not fully reflect underlying trends. For instance, if a government purchases goods and services and delays payment (builds up arrears), the cash concept may not reveal in the current year that the level of spending has changed. Tanzi and Blejer point out that while capturing the monetary impact of the budget, the cash concept may not capture the income- reating (i.e., the Keynesian) impact. Thay note that in the heyday of Keynesian economics (the mid-sixties), the accrual concept was generally preferred to the cash concept. The classification of grants as a revenue source rather than as a financing item is also a practice that is questionable. Grants are not usually permanent sources of income and therefore may fluctuate. (In Ghana they have fluctuated dramatically over the years.) Unless grants are a guaranteed source of revenue, it might be advisable to  10 classify them as a financing item. Similar questions may be raised as to whether net lending should be included as part of the public sector's deficit. If the net lending is extended to the private sector then it does not necessarily reflect public sector usa of resources, but rather, it reflects the public sector's role as a financial intermediary. If the net lending is extended to the public sector (say from the central government to a state-owned enterprise) then it does imply a public sector claim on resources, in which case net lending probably should be included on the expenditure side. In practice, in many developing countries at least, lending between the various levels of the public sector is often not repaid and thus effectively becomes a net transfer. Tanzi and Blejer note a final issue concerning this measure of the deficit. When inflation is significant it may be difficult to distinguish in an economic sense between amortization payments and interest payments. As all interest payments are considered as an expenditure item and no allowance is made for the repayment element implicitly included in the interest payment, the size of the deficit may be overstated. The World Bank's "World Development Report 1988" discusses different measures of the deficit.4 It first cites the Rublic sector borrowinga reuirement (PSBR) as a useful 4 See World Bank (1988), p.56.  11 indicator of the public sector's net use of financial resources. The PSBR represents the total excess of expenditure over revenue for all government entities, all of which must be financed by new borrowing net of repayment of previous debt. This measure is also referred to as the "consolidated public sector deficit". In the calculation of the PSBR expenditure includes wages of public employees, spending on goods and fixed capital formation, interest on debt, transfers and subsidies. Expenditure does not inclade amortization payments on government debt or accumulation of financial assets (net lending). Revenue includes taxes, user charges, interest on public assets, transfers, operating surpluses of public coijanies, and sales of public assets. Revenue does not include the drawdown of cash reserves. A measure that is often used is the "orimary deficit". This subtracts all interest payments from the PSBR in order to obtain a measure of the current policy stance. The argument is that the interest payments currently being made reflect past policy decisions rather than present policy. In order to evaluate current policy, these payments should not be included in the deficit measure. Another concept of the deficit excludes only the inflationary component of interest payments. Finally, the Report discusses the "structural deficit". This measure presents the deficit adjusted for up- and down- turns in the business and/or commodity cycle and for factors  12 that might cause temporary deviations from the trend level of expenditure and tax revenue. Such temporary deviations might be caused by any temporary expenditure or tax policy such as a tax amnesty or a decision to withhold government sector wages. While in theory the structural deficit is clear cut in practice it is often difficult to calculate, particularly in developing countries. Recent research5 has emphasized the importance of obtaining a complete picture of the public sector claim on resources by including all levels of the public sector. Thus the consolidated public sector deficit should include data not only on the central government accounts, but on regional and local accounts as well as on decentralized public agencies and state-owned enterprises. The measures of the deficit discussed to this point have largely been pragmatic ones in that they provide measures that are for the most part practicable given the average level of data availability. Both Boskin (1982) and Buiter (1983) note that these measures, and the way they are calculated, may be quite far from the analytical concepts that are used in the theoretical debate over the effects of the fiscal deficit.6 5 See Easterly (1989a, 1989b), Marshall and Schmidt-Hebbel (1989) and World Bank (1988). 6 Boskin (1982) also makes the important point that, given the difficulties inherent in obtaining a measure of the deficit, econometric analyses of the impact of the deficit may be based on analytically inappropriate concepts or on substantial measurement error and that the issues at hand are not being analyzed or tested in an appropriate manner.  13 Buiter (1983) goes into some detail in presenting a set of stylized accounts that corresponds to an estimate of the comprehensive net wealth or permanent income accounts for the public sector. From these accounts he derives the government budget constraint. He essentially extends the permanent income hypothesis to the public sector.7 Buiter argues that in a first best world not only private agents, but governments and international organizations as well, would decide on spending, saving, lending, production and portfolio allocation constrained only by comprehensive wealth or permanent income. Buiter goes on to argue that both the conventional analysis of the public sector balance sheet8, and the comprehensive wealth accounts that he outlines, should be incorporated into an analysis if the public sector. The conventional accounts provide a guide to the binding constraints on public sector behavior in any given period. The comprehensive accounts provide an indication of the real net worth of the public sector over time and hence optimal policy when the only constraint is permanent income. We have not attempted to construct these comprehensive wealth accounts for Ghana. 7 The permanent income hypothesis set out by Friedman in 1957 argues that current consumption decisions are determined by expected lifetime income (as opposed to only current income). The theory in its simplest form is represented by the optimization of intertemporal consumption subject to lifetime wealth. While the theory ordinarily discusses private consumption, Buiter extends it to public sector decisions as well. 8 These typically contain only marketable financial assets and liabilities.  14 We first consider in detail the data that are available for Ghana and construct a measure of the consolidated public sector deficit from the income and expenditure accounts of the government. The definition of the deficit used follows that of Tanzi and Blejer (1984). We then consider the effects that economic and policy variables have had on the conventional deficit based on the framework set out in Marshall and Schmidt-Iiebbel (1989). Because the conventional measure of the deficit underestimates the true public sector deficit, we then construct a measure of the public sector deficit from the financing side using information that is available on the total stock of domestic and external public sector debt which provides an alternative assessment of public sector claims on resources. 1.) The conventional aDRroach-to the public sector deficit Discussions of the fiscal deficit in the literature on Ghana generally focus on the accounts of the central government with reference made to the importance of other parts of the public sector (e.g. state-owned enterprises).9 Reports of the international organizations include more detailed information on components of the public sector such as the Social Security and National Insurance Trust and the Cocoa Marketing Board, but no attempt at consolidation is 9 See Green (1987), Huq (1989), Killick (1978), and Rimmer (forthcoming).  15 made. There are a number of reasons why a consolidation of the public sector has not been undertaken. One reason is that data on the state-owned enterprises, which constitute a sizable portion of the industrial sector, are not available until 1984, and even then they are not complete. Another difficulty is that some of the accounts of public sector institutions are kept on a fiscal year basis while others are kept on a calendar year basis. a) The consolidated public sector deficit With these difficulties kept in mind one can at least gain an idea of the order of magnitude of the fiscal deficit by taking into consideration the data that are available. Figure 2.1 shows the public sector revenue, expenditure, and deficit (as a percent of GDP) that consolidates all the available information including data on the central government accounts10, the Social Security and National Insurance Trust accounts, and data on the net profits or losses of the Cocoa Marketing Board (through 1988) and the Ghana Industrial Holding Corp (through 1986)11. Very little data is available 10 Note that the data presented on the central government exclude capital expenditure financed through external project aid and the corresponding grants and loans. To the extent that these loans a.-e disbursed, the figures presented may underestimate capital expenditure and thus may give a figure for total public sector expenditure that is underestimated. The exclusion of this information becomes problematic in the years after 1984, when project aid into Ghana increased dramatically. 11 The consolidation added up the central government accounts and the social security accounts. The net operating surpluses (deficits) of the Cocoa Marketing Board and the Ghana Industrial Holding Corp. were added (subtracted) to (from) the revenue side.  FIGURE 2.1 PUBLIC SECTOR: REV., EXP. & DEF. 1969/70- 1988 28- 26 24 0~2 18 0 106/017/517/018 16 14 0~~ 12 z~ - hi~ 997 94/517/018 C)~ ~ ~~a RV X E  17 on local government revenues and expenditure in Ghana. In general, the government has traditionally been very centralized and the central government has provided over half of the revenue of the local governments in the form of budgetary transfers. The absence of data on local government finance therefore does not pose major problems for our measure of the fiscal deficit. In contrast, the absence of data on the state-owned enterprises other than the Ghana Industrial Holding Corp. does imply that we are missing an important part of the claim on public sector resources. The effect of this absence will be discussed below. Considering Figure 2.1 which is based on a consolidation of the above-mentioned accounts, we see that for the majority of the past two decades public sector expenditure has surpassed revenue and that public sector deficits have been significant. Consolidated gublic sector expenditure was approximately 19 percent of GDP in the 1969/70 fiscal year and peaked at 26 percent of GDP in 1975/76. It dropped sharply after 1975/76, reaching a minimum of roughly 8 percent of GDP in 1983. Since 1983 consolidated public sector expenditure has risen and has reached a plateau at about 14 percent of GDP. While an attempt was made to take intergovernmental transfers into account, the breakdown of information did not allow for the complete removal of all double counting. Prior to 1984, data on a fiscal year basis and that on a calendar year basis were consolidated by classifying calendar data under the fiscal period ending with the given year. For example, 1975 yearly data are classified with 1974/75 fiscal year data. It was not possible to reclassify fiscal year data into calendar year data.  18 Consolidated gublic sector revenue peaked in 1970/71 as a result of a sharp increase in revenue from import and export taxes on cocoa. Consolidated public sector revenue declined from this point onward reaching a trough in '.80/81 of about 5 percent of GDP. After 1983, revenue collection improved sharply rising to 18 percent of GDP in 1987 and then dropping slightly to 17 percent of GDP in 1988. Based on this data, the public sector deficit (expenditure minus revenue) moved from a surplus of a little bit more than one percent of GDP in 1970/71 to a deficit of 13 percent of GDP in 1975/76. The deficit dropped to about 4 percent of GDP in 1979/80 but then inrreased sharply in 1980/81 due to a sharp drop in public sector revenue. After 1982, the deficit declined and moved into a surplus of approximately 2 percent of GDP and has remained at about that level. Due to the lack of data on the state-owned enterprises and due to the fact that all capital expenditure financed by tied external loans or grants is excluded, we are fairly safe in assuming that the deficit that results from the consolidation of these accounts is not giving a complete picture of the resources that are being claimed by the public sector. Before going on to alternative approaches, let us consider briefly the components of the consolidated deficit that we have just described. i) The central government Given the absence of detailed accounts on the state-owned  19 enterprises,the pattern shown by the consolidated public sector is dominated by the revenue and expenditure of the central government. Central government expenditure as a share of GDP is only slightly less than that for consolidated public sector expenditure as seen in Figure 2.2. This is explained by the fact that data on the Cocoa Marketing Board and on the Ghana Industrial Holding Corp. are in the form of net profits or losses. They were thus included in the consolidation on the revenue side. The difference between consolidated public sector expenditure and _entral government expenditure reflects the expenditure of the Social Security and National Insurance Trust. A breakdown of central government expenditure over the period from 1969/70 to 1988 is presented in Table 2.1. The most striking point brought out by the expenditure breakdown is the dominance of central government expenditure on consumption over expenditure on investment. Throughout the period central government consumption remained steady wnile investment by the central government varied considerably in line with general economic conditions. Neither interest payments nor current transfers take a large part of central government expenditure, although interest rates have increased in the 1980s. Net lending is low throughout the period. As seen in Figure 2.2, central government revenue as a share of GDP has varied considerably over the past two decades. It peaked in 1970/71 at almost 20 percent of GDP,  FIGURE 2.2 CENTRAL GOV., REV., EXP. & DEF. 1969/70 - 1988 28- 26 24- 22- 20- 16 0.- 0 o 1~6 o 1~4 Z 1~2 0.~ -2 1969/70 1974/75 1979/80 1984 0 CGREV + CGEXP Q CGDEF  OWBU