Public spending and economic growth: evidence from Ghana (1970–2004)

Abstract

Governments undertake expenditures to pursue a variety of objectives, one of which is economic growth. This paper examines aggregated and disaggregated expenditure on economic growth in Ghana over the period 1970–2004. Expenditure on education and health represents human capital development, while expenditure on roads and waterways captures infrastructure development. The study reveals that the aggregated government expenditure retarded economic growth. The study's findings show that expenditures on health and infrastructure promote economic growth, while those on education had no significant impact in the short run. In addition, the political economy variables−namely the nature of governance (democracy) and political instability (years of changes in government and military dictatorship)−proved significant in explaining Ghana's economic growth over the study period.

Keywords:

• Economic growth
• public spending
• human capital development
• political instability

1. INTRODUCTION

Economists have been particularly interested in the factors that cause countries to grow at different rates and achieve different levels of wealth. One of the possible explanations for the differences in wealth is how much governments spend and what they spend it on. Diamond (1989), basing his argument on growth, which is the major goal of any government, says policy-makers need to know the relative contribution of various components of expenditure to their country's economic growth and performance.

In pursuing growth objectives, governments in developing countries spend an average of 26 per cent of the Gross Domestic Product (GDP) on goods and services. In 1991, the World Bank estimated that this figure had risen by eight percentage points over a 15-year period (World Bank, 1992; see also Devarajan et al., 1996). The magnitude and growth of this figure has prompted the need to critically examine the relationship between the size of government expenditure and economic growth. In Ghana, the government spent an average of approximately 22 per cent over the period 1970–2004 (see Table 1). Although cuts in government expenditure may sometimes be necessary or desirable, it is the social sectors such as education, health and housing that suffer from the cut in most developing countries. Thus the study of public expenditure is crucial for identifying which components to cut as fiscal restraint and expenditure switching measures.

Table 1: Some measurements of government size and expenditure elasticity of GDP

Year	Nominal government expenditure (current prices)	Growth rate of nominal government expenditure	Government expenditure (% of GDP)	Real government expenditure	Real government expenditure index (1990 = 100)	Growth of real government expenditure	Recurrent expenditure (% of total)	Development expenditure (% of total)	Expenditure elasticity of GDP
1970	0.4679	18.4	21.0	467.900	183.9	18.4	65.3	34.7	0.9
1971	0.5238	12.0	21.0	523.800	181.6	11.9	64.0	36.0	3.4
1972	0.5431	3.7	19.3	543.100	141.4	3.7	56.2	53.8	24.5
1973	0.5485	1.0	15.7	548.500	117.0	1.0	71.2	28.8	0.9
1974	0.7541	37.5	16.2	377.050	70.0	–31.0	73.5	24.5	0.3
1975	1.1462	52.0	21.7	573.100	109.3	52.0	69.3	30.7	0.8
1976	1.484	29.5	22.7	742.000	143.7	29.5	67.2	32.8	1.6
1977	2.1366	44.0	19.1	712.200	148.3	–4.0	77.0	23.0	1.8
1978	3.1648	48.1	15.1	263.733	40.6	–63.0	81.4	18.6	1.0
1979	4.2957	35.7	15.2	226.090	88.8	–14.3	87.3	12.7	6.0
1980	4.668	8.7	10.9	166.714	57.8	–26.3	82.0	18.0	1.1
1981	7.7193	65.4	10.6	126.546	33.0	–24.1	89.5	9.5	0.8
1982	9.530	23.5	11.0	128.784	27.5	1.8	90.4	10.6	2.1
1983	14.7553	54.8	7.9	89.426	15.9	–30.6	90.8	10.2	0.6
1984	26.694	80.9	9.8	115.558	22.0	29.2	85.0	15.0	0.4
1985	45.763	71.4	13.3	180.169	34.9	55.9	84.0	16.0	1.0
1986	70.6607	54.4	13.8	222.904	46.4	23.7	86.1	13.9	1.0
1987	102.135	44.5	13.7	230.553	35.5	3.4	78.9	21.1	1.0
1988	143.897	40.9	13.7	247.247	97.1	7.2	77.1	22.9	1.5
1989	196.466	36.5	13.9	269.501	105.9	9.0	75.7	24.3	0.6
1990	254.473	29.5	12.5	254.473	100	–5.6	77.9	22.1	0.3
1991	340.262	33.7	14.1	288.358	113.3	13.3	77.5	22.5	0.6
1992	498.813	46.6	17.8	383.997	133	33.2	74.8	25.2	0.8
1993	760.911	52.5	21.0	468.830	122.1	22.1	78.4	21.6	1.0
1994	1141.312	50	22.1	563.055	120.1	20.1	73.5	26.5	1.0
1995	1698.700	48.8	22.0	525.588	93.3	–6.7	71.8	28.2	1.6
1996	2446.279	44.0	22.2	516.420	98.2	–1.7	75.0	25.0	0.4
1997	2908.965	18.9	20.6	480.186	93.0	–7.0	78.1	21.9	1.8
1998	4513.200	55.1	26.1	649.942	135.4	35.4	76.3	23.7	2.3
1999	4964.300	10.0	24.1	635.959	249.9	–2.12	71.5	29.5	0.5
2000	7675.400	54.6	28.3	785.448	308.7	23.51	66.9	33.1	1.7
2001	9053.700	18.0	23.8	696.582	273.7	–11.33	70.4	29.6	0.4
2002	12194.900	34.7	25.0	809.432	318.1	16.20	76.6	23.4	1.2
2003	18946.200	55.4	28.6	1017.431	299.8	25.7	72	28	0.6
2004	23952.700	26.4	30.5	1149.497	351.7	13.0	68	32	0.7
Source: Computations by the author using data from the Ghana Statistical Service, African Development Indicators (2005 and 2006) and various budget statements (Government of Ghana).

In developing countries, fiscal deficits have been blamed for many of the problems that beset them in the 1980s – indebtedness, debt crises, high inflation, poor investment performance and low growth. The major source of fiscal deficit in many countries has been, undeniably, that government expenditures did not match revenue receipt. Writing in the 1890s, Adolph Wagner formulated the law of ‘increasing extension of state activity’, referred to as Wagner's Law (Wagner, 1892). He observed that in the early stages of development, governments tend to increase public expenditure because of the mounting political pressure for social progress. This is one of the theories that emphasise economic growth as the fundamental determinant of public-sector growth, and it has since been the focus of many empirical studies (Lin, 1994).

Interest in the relationship between the size of government expenditure and economic growth is gaining momentum in public finance literature. It has been examined by Robinson (1977), Landau (1983), Ram (1986), Barro (1991) and Devarajan et al. (1996), among others; and more recently by Ghali (1999), Al-Yousif (2000), Dalamagas (2000), Folster and Henrekson (2001) and Yasin (2003), among others, using diverse methodologies. However, this relationship arouses a great deal of controversy in the growth literature. It was in the aim of elucidating some of the controversial issues that the author examined the relationship between public spending and economic growth in Ghana over the period 1970–2004.

2. STATEMENT OF THE PROBLEM

Privatisation has recently become an important policy option for improving the allocation of resources, mitigating budget deficits and encouraging the development of the private sector in most developing countries. In contemporary Ghana and other developing countries, the private sector is regarded as the ‘engine of economic growth’. The importance of the private sector has been re-echoed in all the current Ghanaian policy documents, such as the VISION 2020 (1995–2020), the Coordinated Programme for Economic and Social Development of Ghana (2003–2012) and now the Growth and Poverty Reduction Strategy (2006–2009). The question that begs for an answer is: Do we need increased government activity in contemporary market economies such as ours? Policy-makers are divided as to whether government expansion promotes or impedes economic growth.

Proponents of bigger governments assert that government programmes provide valuable public goods such as education and infrastructure. Grossman (1988) underscores some of the main ways governments facilitate growth: they provide a legal and social framework, defence, police services and judiciary; enforce property rights; correct the inadequacies of an unrestrained marketplace; develop economic infrastructure; regulate externalities; and transfer payments for maintaining social harmony and improving the productivity of the labour force (see also Dalamagas, 2000). In addition, when public and private capital formation are truly complementary, government projects may stimulate entrepreneurs and enhance private investment and hence promote economic growth (Taylor, 1988; Lindauer & Valenchik, 1992).

The proponents of smaller governments, on the other hand, argue that a larger government will impede economic growth because many government operations are inefficient and not in line with the true public interests. For instance, the bureaucracy of decision-making in the public sector, the political practice of promoting the interests of cohesive minorities at the expense of society and the behaviour of interest groups who lobby for resources to be diverted into rent-seeking activities, so as to redistribute income to their advantage, all have the potential to cause inefficiencies in the provision of government output. In addition, many of the government's fiscal and monetary policies tend to distort economic incentives and lower the productivity of the system (Ram, 1986). In this context, taxes and transfers are viewed as distorting market prices and thus reducing incentives for employment and investment.

Which side of the argument is right? This paper seeks to contribute to the debate by ascertaining the relationship between government expenditure and economic growth, both theoretically and empirically, in Ghana over the period 1970–2004.

The author is a proponent of increased state activity and holds the view that increasing public spending in pro-poor interventions such as education, health and infrastructure promotes economic growth in the long run. Empirical findings from this paper attest that health and infrastructural expenditure enhanced Ghana's economic growth over the period 1970–2004.

3. OVERVIEW OF GOVERNMENT EXPENDITURE IN GHANA OVER THE PERIOD 1970–2004

There has been a consistent increase in nominal government expenditure throughout the 35-year period (see Table 1). From a level of ¢0.4679 billion in 1970, nominal government expenditure increased to an unprecedented level of ¢4,513.2 billion by 1998 and reached a peak of ¢23,952.7 billion in 2004.1 The increase in nominal government expenditure year by year was further buttressed by positive growth rates throughout the period. The highest growth rate in nominal government expenditure during the period under review – 80.9 per cent – occurred in 1984, probably due to high productivity and the increase in government consumption precipitated by the 1983 drought and famine. The lowest growth rate of nominal government expenditure – approximately 1 per cent – was recorded in 1973, probably due to the relatively low inflation of 17.3 per cent.

The sustained increase in the nominal government expenditures and their respective growth rates give credence to the incrementalist approach to budgeting. Under this approach, the current year's budget is a function of the previous year's value: a certain percentage or adjustment is made to the previous values to obtain the current budgetary allocations, and as a result current nominal government expenditure is expected to increase.

However, the analysis of both the government expenditure pattern and growth changes when they are converted to real values. The real government expenditure values were obtained by deflating the various nominal expenditures by the Consumer Price Index for the various years (1990 Consumer Price Index = 100). The real expenditures, unlike the nominal ones, necessitate effective comparison. The highest real government expenditure value of ¢1,149.497 billion was recorded in 2004, and the lowest value of ¢89.4261 billion in 1983.

It is interesting to note that real government expenditure for the period 1970–1977 was on average higher than the average values for the period 1979–1991 but lower than the 1992–2004 average (see Table 1). The real government expenditure value of ¢1,149.497 billion in 2004 is about 13 times higher than the real value for 1983 (¢89.4261 billion). Again worth noting is that the years 1977, 1981 and 1983 recorded triple-digit inflation of 116.5 per cent, 116.5 per cent and 122.8 per cent, respectively. As expected, the rate of government expenditure over the period fluctuated. This is in sharp contrast to the robust growth rates of nominal government expenditure over the same period. The highest nominal growth rate of approximately 81 per cent in 1984 was translated into a real growth of only 29 per cent. In 1981, the nominal growth rate of government expenditure was 65.4 per cent while the real growth rate was −24.1 per cent. The period 1991–2004, on the other hand, on average witnessed a robust increase in real expenditure.

In the sectoral classification of government expenditure into recurrent and development expenditures, the recurrent component as a percentage of the total took the lion's share relative to development expenditure. Over the period, recurrent expenditure averaged about 80 per cent of total government expenditure.

Government expenditure ratios (expressed as a percentage of GDP) increased over the post-Economic Recovery Programme period, recording a significant jump in 1993 (a year after Ghana returned to constitutional rule) when the expenditure/GDP ratio increased by over 4 percentage points vis-à-vis the pre-reform average of 16.4 per cent. It has since been increasing steadily, averaging 24.8 per cent between 1994 and 2004.

In terms of expenditure elasticity of GDP, the period average of 1.8 indicates that GDP was expenditure elastic.

Examination of the components of public expenditure in Table 2 reveals that the public wage bill accounts for the lion's share, averaging about 27 per cent of the total in the pre-Economic Recovery Programme period and reaching about 34.9 per cent in 1992. Between 1994 and 1999, the wage bill as a percentage of total government expenditure averaged approximately 24 per cent, exceeding the combined average expenditure on education and health as a percentage of total expenditure over the same period (23 per cent). Interestingly, the period 2000–2004 saw a further rise in the wage bill, averaging 25 per cent, thus surpassing the 1994–1999 average. Among the various reasons for the increased wage bill are the general increase in real wage and the introduction of the ‘Extra Duty Allowance’ for medical and paramedical workers, which has since been suspended because of its budgetary implications.

Table 2: Trends and structure of public expenditure in Ghana, 1970–2004: some major components

Year	Roads and waterways	Social security/social welfare	General administration	Interest payment	Wage bill^a	Defence	Health	Education	Transport and communication	Mining, manufacture and construction	Interest payment^a	Health^a	Education^a	Wage bill^a
1970−1983	7.0	−	18.5	12.4	27.8	7.7	8.6	19.1	3.5	−	2.0	−	2.9	−
1984	7.1	4.2	17.5	13.0	19.8	6.0	8.5	20.2	1.9	2.5	1.3	0.8	2.0	2.0
1985	5.6	4.9	18.5	11.1	32.7	6.5	8.7	22.3	4.0	1.3	1.5	1.2	3.0	4.4
1986	3.8	5.3	13.1	16.0	39.5	6.5	8.3	24.0	3.8	1.5	2.2	1.1	3.3	5.5
1987	8.4	6.4	9.7	10.4	37.5	4.2	9.3	26.5	2.9	1.8	1.4	1.3	3.6	5.1
1988	8.6	6.9	13.5	8.3	34.2	3.2	9.0	25.7	1.8	1.2	1.1	1.2	3.5	4.7
1989	7.6	7.3	13.0	9.5	34.6	3.1	10.1	24.3	1.2	1.1	0.8	1.4	3.4	4.8
1990	6.3	7.2	13.1	10.7	34.6	3.5	10.1	25.5	1.3	1.0	1.3	1.3	3.2	4.3
1991	8.3	7.0	12.3	12.6	31.5	4.5	8.4	23.2	1.3	0.9	1.8	1.2	3.2	4.4
1992	10.3	7.0	12.2	12.3	34.9	3.6	7.8	24.0	0.8	1.0	2.2	1.4	4.3	6.3
1993	9.1	9.0	10.3	17.9	33.0	3.5	7.8	20.8	0.8	1.1	3.5	1.5	4.0	6.5
1994	7.1	7.2	14.7	20.2	28.7	3.2	4.9	18.7	0.6	2.6	4.4	1.1	4.1	6.3
1995	11.2	8.3	12.6	19.4	27.9	3.5	6.6	19.4	0.3	0.8	4.2	1.4	4.3	6.1
1996	11.2	7.1	13.9	23.7	26.9	3.0	5.0	18.4	0.3	0.9	7.5	1.5	4.0	8.5
1997	8.9	7.2	12.9	29.0	27.8	3.5	4.5	19.0	0.2	0.6	6.0	1.4	3.9	5.7
1998	7.4	5.1	10.6	23.8	22.6	3.6	4.2	15.7	0.2	0.5	6.2	1.1	4.1	5.9
1999	8.5	−	−	21.3	21.5	3.0	5.3	18.3	−	−	5.6	1.3	4.4	5.2
2000	10.2	−	−	27.0	18.9	4.2	5.6	20.5	−	−	7.5	1.6	5.8	5.3
2001	11.2	−	−	24.0	18.6	4.0	5.8	16.5	−	−	7.8	1.4	4.1	4.4
2002	11.1	−	−	23.5	32.5	3.7	9.2	14	−	−	6.1	2.3	3.9	8.1
2003	12.6	−	−	22	29.1	3.9	4.9	16.2	−	−	6.5	1.4	4.0	8.3
2004	13.0	−	6.7	21	−	3.8	8.0	18.2	−	−	6.4	2.4	5.5	8.7
Note: % of total expenditure and/or as % of GDP.
^aExpressed as a percentage of GDP.
Source: Computations by the author using data from the Ghana Statistical Service, African Development Indicators (2005 and 2006) and various budget statements.

In 2001 the Minister of Finance highlighted the increase in the share of the wage bill in total expenditure with its concomitant ghost names:

The government losses an estimated ¢300 billion every month through the insertion of ‘ghost names’ onto workers’ payrolls. The amount represents 10 per cent of the about ¢3 trillion that the government expends on civil/public servants every month. (Daily Graphic, 14 December 2001)

The statement clearly demonstrates how the issue of ghost names has exacerbated the wage bill. It is noteworthy that ghost names are negligible or non-existent in the private sector due to employers’ high degree of vigilance.

Available evidence shows that for several years between 1980 and 2002 the minimum wage was less than US$1. Certainly, both inflation and exchange rates show that workers’ minimum wages were eroded in real terms. Thus, for decades, Ghanaian workers’ pay, in both the public and the private formal sectors, was persistently low due to macroeconomic instability (Government of Ghana, 2005). For instance, by the second quarter of 2000, when the exchange rate depreciated to about ¢5500 to US$1, the minimum wage was still pegged at its 1999 level of ¢2900 or US$0.5. In 2001, marginal gains were made in real incomes following the revision of the minimum wage from ¢2900 to ¢5500 (about US$0.75).

Since 2003 the minimum wage has exceeded US$1 both in nominal and real terms. In fact, over the period 1980–2002 the real minimum wage declined in 13 of the 23 years. The most important reason for the increased minimum wage in real terms is the current macroeconomic stability.

Interest payments became an important component of government expenditure especially during the 1990s. This is a significant portion of government expenditure and thus needs to be given urgent attention if appreciable progress is to be made in reducing poverty and accelerating economic growth. It is no wonder that the country opted for the Highly Indebted Poor Countries initiative in March 2001, reaching a decision point in February 2002.

Addison and Osei (2001) outline two reasons for the increase in Ghana's interest payments. Firstly, high domestic interest rates, which had averaged over 30 per cent in the 1990s, significantly more than the early 1980s average of about 13 per cent, induced the supply of private funds. Secondly, Ghana's external debt ratio of about 87 per cent of GDP in 1997 was almost three times the average for the pre-reform period. This aggravated the debt-servicing ratio, which increased from an average of about 11 per cent of GDP during 1970–1983 to about 31 per cent of GDP in 1998. Having reached the decision point of the Enhanced Highly Indebted Poor Countries initiative and some subsequent debt relief, it was expected that debt servicing (interest payment) would fall over time in order to help accelerate domestic development.

However, over the period 2002–2004 the interest payment amounted to 6.5 per cent of the GDP. This unexpected trend is attributable to huge domestic borrowings that occurred during this period via the government's Treasury Bill instrument. Prudent macroeconomic policy from 2001 reduced the Treasury Bill rate from its highest value of 42.77 per cent per annum to 14.89 by 2005 (International Monetary Fund, 2006). The high Treasury Bill rate in the 1990s crippled entrepreneurial initiative because of its risk-free nature. Wealthy businessmen and entrepreneurs preferred investing their money in Treasury Bills since they were assured of over 40 per cent returns, which entrepreneurial profits could not equal. The banking institutions were no match for the government in terms of soliciting for funds from the public via savings. The 9 per cent deposit rate was no match for the over 40 per cent returns from government Treasury Bills, and this crowded out the private sector in the process. The sharp fall in the Treasury Bill rate over the period 2000–2004 rendered it less attractive and thus lessened subsequent domestic debt servicing.

Health and education expenditures (as a percentage of GDP) recorded slight increases, although the importance of these relative to other expenditure items has declined. For example, the share of public health spending in total expenditure decreased from an average of about 9 per cent during 1984–1991 to about 5 per cent in 1997, while the average for the 1998–2004 period was 6.1 per cent.

The share of public spending on education in total government expenditure decreased from about 19.1 per cent in the pre-Economic Recovery Programme era to about 15.7 per cent in 1998. Aside from the decline in health and education expenditures, intra-expenditure allocations had been biased towards urban areas, university education and tertiary healthcare. In 1993 a World Bank report on Ghana indicated that only 25 per cent of the total health budget was earmarked for primary health and preventive care (World Bank, 1995). However, the situation was obviously better in education – the proportion spent on basic education was about 62 per cent in 1989 compared with about 44 per cent in 1984 (Addison & Osei, 2001). Granted that the share of spending on primary education and healthcare as a proportion of their respective sector totals has increased over the years, it still does not merit a reduction in the share of education and health expenditures in the total budget.

A World Bank poverty assessment report on Ghana in 1995 noted that social spending was not well targeted to the poor. Public expenditure on health was too low and the existing spending was urban biased (World Bank, 1995). Public health expenditure as a percentage of the GDP still remains comparatively low, averaging 2.3 per cent during 1995–2002. With the increasing menace of malaria, and recently HIV/AIDS, one may expect public spending on health to increase appreciably as many a country strives to achieve the Millennium Development Goals for health.

In the area of education, the period 1998–2000 saw public expenditure as a percentage of GDP averaging 4.1. This is markedly better than the pre-reform average of 2.9 per cent but not different from the levels attained between 1993 and 1996. Unlike the situation that prevailed in the 1980s and 1990s, when intra-educational expenditures favoured tertiary education, the period 2002–2004 saw pre-primary and primary education receiving 39.2 per cent, secondary education 37.4 per cent and tertiary education 18.0 per cent of the total expenditure on education (World Bank, 2005).

One of the key explanations for the decline in public funding of tertiary institutions is the introduction of the Academic Facility User Fees in 1999 and the proliferation of private universities within the past decade. This has lessened the burden on government, which hitherto was solely responsible for financing tertiary education.

The shift in emphasis from tertiary to pre-primary and primary education has improved Ghana's chances of meeting Millennium Development Goal 3, which is to ‘achieve universal basic education’. With 8 years more to go, Ghana's gross primary school enrolment hit 87.5 per cent in 2004 (Government of Ghana, 2006). The introduction of incentives such as the School Feeding Programme, in partnership with New Partnership for Africa's Development (NEPAD) on a pilot basis in selected primary schools has further increased primary enrolment between 2005 and 2007. What is of concern is the fairly significant spread between the gross and net enrolment rates. The World Bank (2006) estimated that Ghana's net enrolment rate was a modest 58 per cent in 2004.

Expenditure on defence steadily decreased from the pre-reform high level of 7.7 per cent of total expenditure to an average of 3.5 per cent between 1990 and 1998. Between 1999 and 2004, defence expenditure averaged 3.8 per cent. Fortunately, Ghana does not have a pressing security concern, given that its immediate neighbours (Côte d'Ivoire, Togo and Burkina Faso) are relatively stable and peaceful. Consequently, military spending has been low and can be expected to stay that way, thus sparing Ghana the burdens the more insecure sub-regions of sub-Saharan Africa have to carry.

Expenditure on general administration or general services was also steadily reduced from the pre-reform high level of 18.5 per cent of total expenditure to 10.6 per cent in 1998. It is envisaged that the share of expenditure on general administration in total expenditure will decline further as the efficiency of administering public services increases. Transport and communication recorded the least share in total expenditure, recording an average of 3.5 per cent in the pre-reform period and a high of 4.0 per cent in 1985, with the lowest share being 0.2 per cent in 1997 and 1998.

In summary, it could be said that some progress has been made in reducing the size of the public-sector wage bill but pragmatic efforts should be made to channel increased resources to priority sectors such as health, education and infrastructure.

4. THEORETICAL FRAMEWORK

The concepts of economic growth and development have often been used to describe a country's economic performance. While the two concepts are closely related, there is a fundamental distinction. Todaro and Smith (2003:793) describe economic growth as ‘the steady process by which the productive capacity of the economy is increased over time to bring about rising levels of national output and income’ and economic development as ‘the process of improving the quality of all human life’.

Thus economic growth is primarily a quantitative measure (the rate of change of real GDP), while economic development is a combination of quantitative and qualitative measures. Economic development incorporates economic growth in addition to fundamental changes in the structure of the economy, such as a more egalitarian income distribution, a decrease in unemployment and a reduction in poverty. This paper, however, focuses on economic growth (mainly because of the ease of measurement) and how public spending could influence the growth process in Ghana.

There are no generally accepted economic theories capable of explaining, with any degree of success, the process of economic growth. As in research on other aspects of growth, empirical research on the relationship between government spending and economic growth is hampered by the lack of good economic theory. As Carr (1989) notes, theory is unable to settle the debate about the precise role the government sector plays in the economic growth process. Hence, the issues involved have been increasingly viewed as empirical, with inconclusive outcomes.

In essence, economic theory postulates a rationale for government provision of goods and services based on the failure of markets to provide the desired level of public goods and services, to internalise externalities and to cover cost when there are significant economies of scale (Stiglitz, 1988). In the traditional Keynesian macroeconomic model, growth theory maintains that public spending, particularly of a recurrent nature, contributes positively to economic growth. For instance, a high level of government consumption is likely to increase employment, profitability and private investment through the multiplier effect on aggregate output. Government spending raises aggregate demand, leading to an increase in output, depending on the size and effectiveness of the expenditure multiplier (Branson, 1989).

In a simple open macro-economy, the Keynesian aggregate output accounting framework is represented as follows:

where Y is the aggregate output, G is the autonomous government expenditure, X is the exports, M is the imports, (X−M) is the net exports, and C is the consumption, which consists of an autonomous and induced part, that is:

where a is the autonomous consumption and Y ^d is the disposable income.

Totally differentiating Equation (1):

Assuming da, dI and d(X−M) are constant:

where m is the basic expenditure multiplier. This tells how much aggregate output changes as a result of a change in government expenditure. Thus, according to the Keynesian framework, the impact of changes in government expenditure on aggregate output can be summarised in three propositions.

• An increase in government expenditure will raise aggregate output. However, the extent of the increase will be determined by the quantum of the expenditure multiplier.

• An increase in tax will decrease aggregate output by the size of the expenditure multiplier.

• Either an increase in government expenditure or a reduction in taxes will, ceteris paribus, increase aggregate output by the size of the expenditure multiplier.

The argument in favour of increasing expenditure as compared with tax cuts is that expenditure injections such as government consumption or investment expenditures will give impetus to other economic activities such as employment creation. The basic rule for the growth-promoting public sector is that its activities should complement rather than compete with those of the private sector (avoiding the crowding-out effect).

Thus an important role for the government is to provide certain investments in human capital such as primary education, public and primary healthcare, and infrastructure. In addition, certain expenditures on the legal system, public order and civil administration are necessary to ensure a stable environment in which investment with higher payoffs can be encouraged.

5. MODEL

In studying the relationship between government spending and economic growth in Ghana, this paper uses a framework analogous to Ram's (1986). His model is built on a two-sector production function, originally developed by Gershon Feder (1983) to explore the relationship between exports and economic growth. This model assumes that the economy consists of a government sector (G) and a non-government sector (C). Output in the government sector is produced with a factor combination of labour (L) and capital (K), while output in the non-government sector is produced with labour (L) and capital (K), with an ‘externality’ effect on output stemming from the government sector. The production functions for the government and non-government sectors can be written as follows:

where subscripts denote sectoral inputs. If the total inputs in the two sectors are given as:

then total output (Y) is given by the summation of outputs in the two sectors, and thus:

The model also assumes that the relative factor productivity in the two sectors is different, such that:

where uppercase subscripts denote partial derivatives of the functions with respect to subscripted inputs. For instance, G _L denotes ∂G/∂L (the marginal productivity of labour in the government sector) or its discrete analogue ΔG/ΔL. From Equation (8), it is obvious that the sign ∂ indicates which sector has higher marginal factor productivity. By inference, a positive ∂ means that the factor productivity in the government sector is higher than that in the non-government sector.

Manipulating the production functions and combining Equations (5)–(8), we can derive the following approximation for an aggregate growth equation:

or writing δ′ for δ/1+δ, Equation (9a) can be written as:

where the dot over the variable indicates its rate of growth; for example, (ġ) denotes ∂g/g or its discrete equivalent Δg/g. In fact, ġ denotes the rate of change of real GDP. The coefficients α, β and ϑ are of the kind usually found in simple growth models. In this case, α is the marginal product of capital (K) in the C sector, β is the elasticity of non-government (private) output (C) with respect to labour (L), and ϑ equals C(G/C) and is the elasticity of non-government output with respect to G.

If δ′=ϑ, we obtain a special case where Equation (9b) reduces to:

In Equations (9b) and (10), ϑ gives only the externality effect of government size and not total effect. If it is assumed that C _G, rather than ϑ, is the constant parameter, Equation (9b) can be written as:

Obviously, the coefficient of G (G/Y) in Equation (11) is different from the coefficient of that variable in Equation (9b). Typically, the coefficient in Equation (9b) is likely to be much smaller than the coefficient in Equation (11). In addition, the advantage of Equation (11) is that, unlike Equations (9) and (10), one can obtain the overall impact or effect of government size directly from the coefficient of G (G/Y). However, the disadvantage is that one cannot obtain separate estimates of the externality effect and the factor productivity differential. This weakness will have no impact on this paper, since the focus is to obtain the overall effect of the government expenditure variable on economic growth. It is also notable that while collinearity between G and G (G/Y) may lower precision in the estimation of Equation (9b), neither Equation (10) nor Equation (11) has that weakness.

Since the main focus of this study is to obtain the direction and magnitude of the overall effect of government size on economic growth, Equation (11) was adopted and the estimated coefficient of G (G/Y) – that is, (δ′+C_G)–was treated as a composite coefficient, say β. For empirical comparisons, Ram also estimates an equation that has been widely used in the literature (for example, in Landau, 1983). This can be written as:

Equations (10) and (11), or what might be called the Ram model, measure the size of government by the percentage change in government expenditure, while the standard Equation (12), otherwise called the Landau model, measures it by the ratio of government expenditure to GDP.

An export orientation variable, political instability and the nature of political governance dummies, which are instrumental for determining growth in sub-Saharan Africa, were added to the Ram (1986) model to reflect the realities of the politico-economic environment in Ghana during the sample period. The inclusion of the export variable and political instability were necessitated by the existing literature and hence justified in this study.

Functionally, the model is specified as follows:

Equation (13) includes specified government expenditure variables instead of the aggregate government expenditure (GGY), to avoid multicollinearity.

The final regression equation and expected analytical results are as follows:

where g is the rate of change of real GDP (economic growth), α is the constant term, L is the Labour Force (proxied with the rate of growth of population), XR is the rate of growth of real exports and IY is the ratio of investment to GDP (both private and public). GGY is the rate of change of real government expenditure weighted by the share of government expenditure in GDP. RG, EG and HG are the shares of total expenditures on infrastructure, education and health, respectively.

The political instability dummy (POLINS) is introduced by assigning 1 to years in which the military intervened and/or change of government occurred, and 0 to years without political change. The nature of political governance is represented by a dummy (DEMOCRA) with any year of multi-party (parliamentary rule) democracy assigned a value of 1, and 0 otherwise. It is expected that POLINS will inhibit while DEMOCRA promotes growth.

Finally, E _t is the stochastic term with standard properties. The variables were chosen with recourse to both growth theory and the general empirical growth literature. The ordinary least-squares method of estimation is used to estimate the regression coefficients.

6. DATA

The data are annual and range from 1970 to 2004 (a 35-year period). The sample was chosen mainly on the basis of data availability. Unless otherwise specified, all the data are drawn from Ghana Statistical Service and the International Monetary Fund's International Financial Statistics, African Development Indicators and various Government of Ghana Budget Statements (Government of Ghana, 1985–2005). In the empirical analysis, we test whether the specified components of government expenditure, education, infrastructure and health are associated with higher economic growth. Hence our key explanatory variables are the impact of the aggregate expenditure as well as the specified components on the economic growth process in Ghana.

The rate of change of real GDP was taken as a proxy for economic growth, which is the dependent variable. The aggregate government expenditure variable is measured as the rate of change of real total government expenditure weighted by total expenditure to GDP ratio (Ram, 1986). Fixed capital formation as a ratio of GDP and population growth rate are used as proxies for investment and labour force, respectively. The use of the latter is dictated by the lack of data on the labour force in Ghana. The external orientation of the economy is captured by the rate of change of real exports. The Consumer Price Index (1990 = 100) is used as a deflator to construct real magnitudes.

Since the variables are in growth rates or weighted by GDP, they are no longer time series and thus the conventional time series properties are not applicable.

7. DISCUSSION OF RESULTS

Tables 3 and 4 provide the results for the estimated growth model with the aggregate government activity and disaggregated components, respectively. A diagnostic test summary for both equations indicates the absence of autocorrelation and autoregressive-conditional heteroscedasticity and the presence of homoscedasticity. The equations also passed the normality chi-squared test. This implies that the residuals in the regressions are white noise.

Table 3: Estimation of the growth model with aggregate expenditure

Independent variable	Coefficient estimate	t statistic
Constant	−2.22	−0.89
L	0.70	1.13
XR	0.04	1.62
IY	0.30∗∗	2.36
GGY	−0.50∗∗	−2.38
R ^2 = 0.24	DW = 1.72
Note: DW, Durbin–Watson statistic.
Source: Author's own computation.

Table 4: Estimation of growth model with disaggregated expenditures plus political economy variables

Independent variable	Coefficient estimate	t statistic
Constant	−3.17	−0.54
L	0.38	0.64
XR	0.05∗	1.97
IY	0.01	0.03
RG	0.18∗∗	2.73
EG	−0.30	−0.92
HG	1.15∗∗	2.03
POLINS	−3.62∗∗	−2.16
DEMOCRA	4.11∗	1.88
R ^2 = 0.43	DW = 1.810
Note: DW, Durbin–Watson statistic.
Source: Author's own computation.

In addition, the models passed the RESET test at the conventional 5 per cent level of significance. This proves that the models are correctly specified. Therefore the null hypothesis of model mis-specification is rejected, and thus the linear model specified is correct and the use of ordinary lest squares is justified.

In Table 3, the model explains only 24 per cent of the variations in real output in Ghana over the period 1970 –2004. It also suggests that about 76 per cent of the variation in real output over the period is explained outside the model. Although the explanatory power of the model is low, it is not strange because growth models using variables in rates or ratios have lower explanatory power than natural logarithm and other models.

In addition, our concern is to test the statistical significance of the aggregate government activity variable. In line with the neoclassical growth theory, the labour force proxy has the correct sign but is insignificant. Although this study is centred on the relationship between public expenditure and economic growth, the inclusion of the external orientation variable is justified. The rationale for this is the premise that the government can create the enabling environment through its expenditure outlays to promote exports to help generate sufficient foreign exchange for the country's growth process. Although this result was not anticipated, it was not surprising. In related studies, Tsikata (1996) and Baah-Nuakoh (2000), using similar models, found the external orientation variable insignificant in Ghana if the growth of government is introduced into the model.

Empirically, the ratio of investment to GDP proved to be significant in the simplified growth model. The robustness of the investment variable might be attributable to the complementary relationship between government expenditure and public investments. In the growth literature, the coefficient of this variable is described as the marginal productivity of capital. Thus the marginal productivity of capital is found to be significantly high over the period.

The relationship between the aggregate government activity variable and economic growth has been controversial and inconclusive irrespective of the size of government chosen. Empirically, this variable was negatively and statistically significant at the 5 per cent level. Although this result was not expected, it is not unexplainable. It implies that government's activity in terms of aggregate rate of expenditure stifled Ghana's economic growth. The rationale is that an increase in government expenditure will increase the amount of distortionary taxation to finance such expenditure, and hence will reduce growth. The argument is based on the assumption that a higher government spending acts as a proxy for a higher rate of taxation, which may throttle economic growth.

While aggregate government expenditure provides utility to households, government consumption expenditure reduces economic growth because the higher taxes needed to finance the consumption expenditure decrease returns on investment (Devarajan et al., 1996). In addition, economic growth is retarded when governments grow too large relative to technical efficiency (Barro, 1990). Thus, increases in government expenditure without a corresponding increase in managerial efficiency have been the bane of Ghana's economic growth. Given that about 10 per cent of the total wage bill is paid to ‘ghost’ workers, and the fact that expenditure disproportionately favours wage bills, interest payment on general debts and recurrent expenditure in general, the aggregate government expenditure may not have a positive impact on economic growth.

Table 4 provides the regression results for the disaggregated government expenditure variables. The model explains 43 per cent of the variations in real output in Ghana over the period 1970–2004. The results reveal that the three most significant variables explaining Ghana's economic growth are spending on health, infrastructure and political instability. Expenditure on public health positively and significantly influenced Ghana's economic growth at least over the range of this data-set.

Public spending on health is an important ingredient of human capital development. The state of health of the population is important in determining overall productivity, whether public or private. In addition, government spending on infrastructure also influenced growth positively and significantly. Infrastructural development, particularly roads, reduces transaction costs and deepens trade. If rural dwellers are the beneficiaries of infrastructural enhancement, this has a far-reaching impact on poverty reduction. The education variable was not significant. This might be because there is a lag in the response of education to economic growth. It takes some years for a child or adult in the process of acquiring skills today to influence productivity. Obviously, the mechanism is not instantaneous.

As expected, political instability retarded Ghana's economic growth over the study period. This is in sharp contrast to the finding by Tsikata (1996) that political instability and/or democracy has no effect on growth. The possible explanation for this result is that the data were extended to include more information on democracy and political instability.

Sowah (1994) corroborates the inverse relationship between growth and political instability to the effect that all the years of negative growth rates but one were under military dictatorships (see Table 5 for details). This might explain why the investment variable is insignificant. Normally investments, both domestic and foreign, diminish under military and unconstitutional governments. This is because under such regimes investors are not sure of the marginal productivity of their investments and thus may not invest, and in extreme cases may disinvest.

Table 5: Political changes and economic performance in Ghana (1965–2004)

Year	Inflation	Real GDP growth (%)	Exchange rate (US$)	Government expenditure (% GDP)	Political change
1965	22.7	1.37	1.02	−
1966	14.8	0.09	1.02	−	MT
1967	−9.7	−3.00∗	1.02	21.3	A
1968	10.7	6.42	1.02	23.5	A
1969	6.5	5.85	1.02	19.8	E
1970	3.0	6.76	1.02	21.0
1971	8.8	5.56	1.82	21.0
1972	10.7	−2.79∗	1.28	19.3	MA
1973	17.0	15.21	1.15	15.7	A
1974	18.7	3.39	1.15	16.2
1975	24.5	−12.86∗	1.15	21.7	P
1976	62.1	−3.52∗	1.15	22.7	A3
1977	116.5	2.26	1.15	19.1
1978	73.1	8.48	2.75	15.1	P
1979	54.4	−3.17∗	2.75	15.2	ME
1980	50.1	0.00	2.75	10.9	MT
1981	116.5	−1.79∗	2.75	10.6	A4
1982	22.3	−7.20∗	2.75	11.0	A3
1983	122.8	−4.4∗	30	7.9
1984	39.6	8.7	50	9.8
1985	10.4	8.6	59.99	13.3
1986	24.6	5.20	90.01	13.8
1987	39.8	4.79	176.06	13.7
1988	31.4	6.23	229.89	13.7
1989	25.2	5.1	303.03	13.9
1990	37.2	3.3	344.83	12.5
1991	18.0	5.3	390.63	14.1
1992	10.1	3.9	520.83	17.8	E
1993	25.0	5.0	819.67	21.0
1994	24.9	3.8	1052.63	22.1
1995	27.9	4.5	1446	22.0
1996	46.2	5.2	1740	22.2	E
1997	27.9	4.2	2250	20.6
1998	17.7	4.6	2346	26.1
1999	12.4	4.4	3500.69	24.1
2000	25.2	3.7	5346.27	28.3	E
2001	35	4.2	7103.06	23.8
2002	32.9	4.5	7987.29	25.0
2003	23.6	5.2	8524.86	28.6
2004	11.9	5.8	8768.72	30.5	E
Notes: ∗Slump in output. MT, military takeover; Ai, attempted coup (number in the year); P, palace coup; E, general elections.
Source: Sowah (1994), updated and modified by author from 1995 to 2004.

On the other hand, exports and democracy weakly influenced growth positively (at the 10 per cent level of significance). This provides a useful insight into the growth problem in developing countries such as Ghana, to the effect that democracy (parliamentary rule) and/or political stability is a prerequisite for economic growth.

To resolve the issue of causality, a pairwise Granger-Causality test was conducted (see Table 6). The finding gives support to the Wagnerian thesis that economic growth Granger-Causes growth in aggregate government expenditure. In addition, there is evidence that economic growth Granger-Causes expenditure on infrastructure. This study did not find any evidence of causality between the other variables of interest.

Table 6: Results for Granger-Causality test

Null hypothesis	F statistics	Probability
GGY does not Granger-Cause g	2.65	0.10
g does not Granger-Cause GGY	3.78	0.02∗∗
g does not Granger-Cause RG	3.59	0.03∗∗
RG does not Granger-Cause g	1.29	0.30
g does not Granger-Cause EG	0.54	0.66
EG does not Granger-Cause g	2.15	0.12
g does not Granger-Cause HG	0.34	0.80
HG does not Granger-Cause g	2.12	0.12
∗∗Significant at 5% level.
Source: Author's own computation.

8. SUMMARY AND CONCLUSIONS

The purpose of this paper was to shed light on the relationship between government expenditure and economic growth in Ghana. It focused on growth because it is important for a country to know the contribution of the aggregate and different components of government expenditure to the growth process.

The study therefore empirically investigated the effects of three key components of government expenditure (education, infrastructure and health), alongside the aggregate government expenditure on Ghana's economic growth process. The variables that significantly influenced Ghana's economic growth from the simplified model were investment and aggregate government spending. The aggregate expenditure variable had a significant negative association with economic growth.

In the leading model (i.e. the model with all the variables under consideration), public spending on health and infrastructure proved significant in explaining economic growth over the sample period. In addition, the two political economy variables, political instability and the nature of governance (democracy or autocracy), had a significant association with Ghana's economic growth.

The study provided some insights into the growth problem in Ghana. Accordingly, the following recommendations are provided on the basis of the research outcomes.

The government of Ghana should practise a mixed policy of expenditure cutbacks and switching. While aggregate government expenditures inhibit economic growth, the expenditure allocations to health and infrastructure promote growth. It is therefore imperative to increase expenditure allocations to pro-poor interventions in health, infrastructure and education, while maintaining defence expenditure or reducing it slightly.

All of the various benefits from heavily indebted poor countries and debt relief should be channelled to these pro-growth sectors. In addition, the 2.5 per cent of the value-added tax that is the cornerstone of the Ghana Education Trust Fund could be increased modestly to actualise the government's objective of making primary education truly free and compulsory while at the same time achieving the Millennium Development Goal of Universal Basic Education.

To accelerate infrastructural development, we should first and foremost strive to achieve robust economic growth. This is because the causality runs from economic growth to infrastructure.

To actualise our potential growth, it is pertinent to accelerate the domestic savings rate to realise the desired investment. The current savings rate of about 10 per cent of GDP is inadequate for a robust economic growth. The government should therefore sustain the current macroeconomic stability so as to maintain the prevailing interest on the Treasury Bill instrument rate to create leverage with the financial institutions, since the public and private sectors solicit for funds from the same market. The hitherto wide gap between the Treasury bill and the savings rates was a disincentive to savings, and possibly crowded out private investment.

Finally, sustaining the young democracy is a necessary precondition for economic growth and development, because poor economic performance was associated with military regimes. For Ghana to continue to compete successfully with its neighbours and other countries in the sub-region in attracting foreign direct investment, stable democracy is non-negotiable.

Notes

¹See Table 5 for the ¢ (Ghanaian cedi)/US$ exchange rate dynamics.