ECONOMIC GROWTH AND THE STRUCTURE OF TAXES IN
SOUTH AFRICA: 1960 - 2002
STEVEN F. KOCH , NIEK J. SCHOEMAN AND JURIE J. VAN TONDER
Abstract
One tenet of taxation is its distorting effect on economic behaviour. Despite the economic inefficiencies
resulting from taxation, it is widely believed that taxes impact minimally on the economy's growth rate.
Evidence in developing countries generally supports this view. In this paper, we present evidence that
tax distortions in South Africa may be much more severe. Using tax and economic data from 1960 to
2002 and a two-stage modelling technique to control for unobservable business cycle variables, we
examine the relationship between total taxation, the mix of taxation and economic growth. We find that
decreased tax burdens are strongly associated with increased economic growth potential; in addition,
contrary to most theoretical research, decreased indirect taxation relative to direct taxation is strongly
correlated with increased economic growth potential. J.E.L. Classification: H21, 047 Keywords: tax
burden, tax mix, translog, Data Envelopment Analysis
1. INTRODUCTION
ELEMENTARY ECONOMIC ANALYSIS of taxation primarily focuses on the tax burden. Under most
situations, the primary burden of a tax is a decrease in economic activity, referred to as deadweight loss.
A simple extension of the standard model, allowing for taxes on all goods and activities, implies a
reduction in economic activity in every market in the economy, and, therefore, taxes would be expected
to negatively affect economic growth in an economy. However, that simple analysis ignores a number of
other issues. Most importantly, if the government uses the collected tax revenues to fund investment in
social goods, especially goods resulting in external benefits (infrastructure, education and public health,
for example), the economic growth rate could be positively influenced by taxation. Furthermore, if
money is transferred from people with low marginal utilities of income (rich) to people with high
marginal utilities of income (poor), while revenues are additionally used to fund public investment, then
the economy can gain from this 'double-dividend' of taxation. Therefore, measuring the
The authors would like to thank two anonymous referees for their thorough and helpful comments.
Their comments have greatly improved this paper. The authors would also like to thank Renee van
Eyden, Moses Sichei, Marc Ground, and Cristina Moolman for their help with the empirical modelling,
and Zurica Cluasen-Robinson for her help uncovering relevant tax policy materials. In addition, Neil
Cohen, Jan van Heerden, James Blignaut, and participants at the 2003 Biennial Conference of the
Economic Society of South Africa provided useful insights, which helped improve this paper. No
attempt is made to implicate anyone, other than the authors, for any errors that may still remain in the
paper.
Corresponding author; Associate Professor of Economics, Department of Economics, University of
Pretoria, Pretoria 0002, Republic of South Africa, steve.koch@up.ac.za
Professor of Economics, Department of Economics, University of Pretoria, Pretoria 0002, Republic
of South Africa, niek.schoeman@up.ac.za
Senior Lecturer in Economics, retired, Department of Economics, University of Pretoria, Pretoria
0002, Republic of South Africa.
openUP
economic effect of taxation ought to be a simple exercise in determining whether or not the
benefits of government expenditures exceed the costs of taxation.
Unfortunately, it is not easy to directly determine the net benefits of taxation and,
therefore, to determine whether or not taxes are benevolent. For example, if taxation pushes
economic activity underground, we would find that it lowers economic growth when, in fact,
it is merely shifting economic activity from the measured economy to the unmeasured
economy, i.e., we are overstating the negative effects of taxation on economic growth. On the
other hand, if many markets are characterized by low elasticities, so that economic activity is
not significantly affected, but the government purchases imported goods with the tax
revenues, then GDP and economic growth will be lower due to the dual leakages, taxation
and importation. Therefore, a complete analysis of the economic effects of the fiscus on an
economy would examine the impacts of expenditures and taxes. The results reported in this
paper represent the first stage of research into these issues conducted on the South African
economy. Rather than comparing the benefits of government expenditure with the costs of
taxation, we examine the relationship between taxation and economic growth.
Any analysis of the relationship between taxation and economic activity must account for
other variables. In particular, economic growth in an economy could be high due to
international factors; in the case of South Africa, inflation in the developed world in the late
1970s and early 1980s led to increased demand for gold, which, in turn, led to increased
economic growth. Unfortunately, it is impossible to include data on all of the things that
might have contributed to or detracted from economic growth in South Africa during the
time period of our analysis. For that reason, in this paper, we report the results from a two-
stage model that controls for non-included variables in a different way. The two-stage
procedure is similar in concept to instrumental variables estimation, although the application
is different, due to parametric considerations, from two stage least squares.
We proceed by investigating South African tax policy, the expected effect of those tax
policies and relevant research in section 2. In section 3, we discuss the analytical framework
used to measure the impact of taxes on economic growth. We provide a preliminary analysis
of the data, including the control for unobservable economic variables, in section 4. We
present and discuss the results of the analysis in section 5, and provide concluding remarks in
section 6.
2. BACKGROUND
(a) Economic Growth and Taxes: A. Review of the Literature
Theoretically and empirically, it has been shown that taxes affect the allocation of resources,
and, often, distort the underlying behaviour of economic agents. Regarding economic growth,
taxes influence the labour-leisure trade-off and investment decisions. Due to the importance
of labour and capital in determining economic growth, it is surprising that the relationship
between economic growth and taxes has not featured more prominently.
In what is now referred to as Harberger's superneutrality conjecture, Harberger (1964: 62-
63) stated, ".. .this boils down to the question of how significantly the rate of growth could be
influenced by plausible changes in the mix of direct and indirect taxation. I think that the
answer is not very much." He further showed that changes in the mix of direct and indirect
taxes do not alter labour supply or investment rates
openUP
sufficiently and influence rates of economic growth only negligbly. Harberger's
superneutrality conjecture has been supported theoretically as well as empirically. Hall
(1968) develops a savings-consumption model finding that tax changes will only lead to
transitory changes in growth. Hall, who applies a neoclassical growth model, relies on
exogenous technical change and population growth which, by assumption, are not likely to
be significantly influenced by the economy's tax structure.
1
Models of endogenous growth, however, may provide a theoretical link between economic
growth and tax policy.
2
For example, Lucas (1990), Jones and Manuelli (1990), King and
Robelo (1990) and, more recently, Yamarik (2001) all argue that economic growth is
retarded by taxation. Analyses by Pecorino (1993, 1994) suggest economic growth could be
increased significantly, from 1.53 per cent per annum to 2.56 per cent per annum in the US,
if the tax mix were shifted away from income taxes towards non-distorting consumption
taxes. However, Stokey and Robelo (1995) and Mendoza, Milesi-Ferrett, and Asea (1997),
who also make use of endogenous growth models in their analyses, find negligible negative
effects of taxation on economic growth in developed economies. As an example of the small
effects, Engen and Skinner (1996) report that a 5 per cent reduction in average tax rates in
the US would likely add 0.25 per cent per year to economic growth.
3
An important feature of
these models is the assumption that taxes are returned to consumers, efficiently — an
assumption that may not be realistic in developing countries.
Although a consensus appears to be emerging regarding the impact of tax policy on
economic growth in the developed world, primarily the US, less analysis has taken place in
the developing world. However, the research that has been undertaken suggests that the
impact of taxation in developing economies is larger than it is in developed economies.
Marsden (1990) groups 20 developing countries into high tax and low tax regimes and finds
that the low tax group averaged 7.3 per cent growth, but the high tax group only 1.1 per cent.
Wang and Yip (1992) show that the structure of taxation is more important than the level of
taxation in explaining economic growth in Taiwan from 1954 to 1986. Their empirical
estimates show significant and negative impacts of specific taxes on economic growth, but
the effect of total taxation is not significant. Kim (1998) compares economic growth and
taxation in the US with economic growth and taxation in Korea. According to his analysis, 35
per cent of the difference between US and Korean economic growth can be explained by
differences in the tax structure between the two countries. Kerr and MacDonald (1999) find
mixed evidence that the ratio of indirect taxes to direct taxes negatively and significantly
affects economic growth in seven Asian economies. On the other hand, Tanzi and Shome
(1992) uncover no obvious uniformities between the tax policies of eight Asian economies
concluding, "...tax structure may become largely irrelevant when macroeconomic
1
More specifically, if exogenous technical progress is the main determinant of economic
growth, tax policy can only affect long-run income and not long-run economic growth, primarily
because capital accumulation only covers depreciation and population growth in these models.
2
Endogenous growth models allow for continued capital accumulation, beyond depreciation
and population growth in steady-state, e.g., capital accumulation depends upon the net return to
investment, which is affected by tax policy.
3
Although the yearly effect is small, accumulation of that quarter percent over 36 years from
1960 implies an overall increase of 7.5 per cent of GDP, approximately $500 billion in 1996
(calculation reported in Engen and Skinner, 1996).
openUP
problems become predominant, and the distortions created by the tax system become of a second order
magnitude..."
Due to the structural problems in African economies and the results of Tanzi and Shome's analysis,
we might expect the effects of taxation to be minimal in Africa. Although Africa has received less
attention than Asia, Skinner (1987) estimates the effect of taxation in Sub-Saharan Africa over the
period 1965 to 1982. He finds that personal and corporate income taxes have a significant and negative
effect on economic growth; trade taxes also reduce economic growth, indirectly, in the region, while
sales and excise taxes have no significant effect on economic growth. Based upon Skinner's analysis,
we tentatively conclude that the tax structure may not be largely irrelevant on the continent, as was
implied by Tanzi and Shome.
For a number of reasons, the impact of taxation in the developed world is likely to be different from
the impact in the developing world, especially in Africa, and, therefore, taxation in Africa, which has
received little attention, merits further study. Importantly, developing countries do not have the
infrastructure to adequately police tax compliance; thus, shifts in tax policies in developing countries,
especially increases in income taxes, are likely to push economic activity underground.
4
Furthermore,
governments in developing countries may not return taxes back to the public in an efficient manner
(e.g., by not adequately investing in public goods),
5
or governments might be corrupt or otherwise not
trustworthy (e.g., by squandering resources on lavish residences, by changing tax policies in an ad hoc
manner, or taking control of economic resources).
6
Finally, government agents have the incentive to
increase the tax base of taxed activities. In the case of developing countries, which often rely on
corporate taxes imposed on large (often state-owned) companies, the tax structure provides incentives to
increase the profits of these companies, often to the detriment of competition, which could have
significant economic growth effects.
7
(b) South African Tax Policy: 1960-2002
Tax policy in South Africa has been scrutinized and revised continuously since 1960.
Given the political and structural changes that have occurred since the complete
democratisation of the country in 1994, fiscal policy will be dissected into two time
periods.
(i) Fiscal Policy Preceding 1994 Preceding 1994, South Africa operated under apartheid, and the
government continually increased expenditures on defence, in an effort to stabilize the country
through covert military operations. Government expenditure
4
This inframarginal behaviour is not limited to developing economies, Small and Cartaga (n.d.)
estimate that a one unit increase in the tax burden (from 29-30 per cent) in New Zealand leads
to a 0.1 per cent increase in the size of the informal economy.
5
Skinner (1987) finds that a 5 per cent increase in public investment, financed through taxation,
reduces growth by a approximately 0.6 per cent in Sub-Saharan Africa between 1974 and 1982.
6
McMillan and Masters (2000) develop a model of economic growth and government policy
based on time-consistent behaviour showing that government commitment devices improve the
incentives for economic agents to invest, beyond subsistence levels, and, thus, are likely to
improve economic growth.
7
Gordon and Wilson (1999) develop a theoretical model showing that incentives within
government are likely to be just as important in explaining government behaviour as are
incentives in explaining other economic agent behaviour.
openUP
increased from 20 to 27.5 per cent of GDP, during the period 1970 to 1994, while total
taxation increased from 17 to 22 per cent of GDP.
8
Between 1970 and 1994, budget deficits
averaged 3.6 per cent of GDP, with a maximum of 7.3 per cent in 1993. Fortunately,
minimum reserve requirements provided a captive market for borrowing, which alleviated
the need to collect additional revenue. Nonetheless, due to a substantial increase in revenue
collection, the debt/GDP ratio gradually declined to 29.6 per cent in 1982, from 43.6 per cent
in 1970, after which it increased again to 43.5 per cent in 1994.
9
Government appointed a
few commissions to investigate the fiscal situation and provide future policy guidance, with
the Franszen Commission (1970) pointing to the potential negative impact of increased tax
burdens.
10
As far as indirect taxes are concerned, government initially relied on sales duties with
high rates and a very narrow base. By 1978, the sales duties had been replaced by a 4 per
cent general sales tax (GST). The GST broadened the indirect tax base and allowed for the
reduction in marginal personal income tax rates.
Another commission, the Margo Commission, undertook a similar review of fiscal policy
in 1987. The Commission operated during a period of rampant inflation and foreign
disinvestments, a result of international pressure against apartheid. The Commission
recommended a number of changes, although they were similar in nature to the proposals of
the Franszen Commission.
11
Two of the most important changes regarding taxes in South
Africa resulted from this report: the GST was dropped in favour of an invoice-based value
added tax (VAT), and a reduction in company tax rates to 35 per cent (although the reduction
might have been mitigated by the simultaneous introduction of the secondary tax on
companies (STC), which was an incentive for companies to retain earnings for investment
purposes).
(it) Fiscal Policy from 1994 Following the transition to complete suffrage in 1994, fiscal
policy was redefined and, in order to assist the government with tax policy, a new
commission, the Katz Commission, was appointed. Various tax law changes and
amendments followed, all aimed at broadening the tax base, reducing government borrowing
pressures, closing loopholes, and improving the neutrality of the tax base, all aimed at
improving economic performance. The government adopted the residence principle of
taxation from the beginning of 2001 and, shortly before, introduced foreign dividend
taxation, successfully extending the tax base to include foreign source income. The intention
was, in addition to expanding the base, to improve fairness and establish a global presence
for South African corporations.
The government also pursued various supply-side policies, allowing for accelerated
depreciation allowances and tax holidays as part of its macroeconomic strategy for
8
All percentages are extracted from the South African Reserve Bank ( www.reservebank.co.za).
9
The impact of absorbing the former Homelands debt between 1990 and 1995 raised the ratio
from 36.9 to 49 per cent.
10
In particular, the Franzen Commission recommended: a reduction in the progressiveness of
the direct tax structure, abolishment of the source basis of taxation, a shift toward indirect
taxation, a broadening of the income tax base through redefinition; all, but the last proposal,
were adopted
11
The Margo Commission recommended: fringe benefit taxation, lower and fewer personal
income tax rates with fewer deductions, equal treatment of men and women (especially,
marriage neutrality), no capital gains taxes, a modification of GST, a capital transfer tax, and
retention of the source basis of taxation.
openUP
