Abstract
The South African government is evaluating the economy's performance over its first decade in power. This period can be characterised by a ‘double’ liberalisation: democratisation of the political process going hand in hand with liberalisation of the economy. This article provides a broad overview of the macroeconomic aspects of this liberalisation.
1 INTRODUCTION
Economic liberalisation might be expected to change the working of the macroeconomy – as the economy opens up, the foreign sector should begin to play a greater role in aggregate demand. Also, the lifting of constraints may disturb the established savings–investment process as savers and investors face new economic conditions. From this perspective, one of the questions that comes to mind when examining the macroeconomic policies, trends and events in the last ten years of the South African economy is whether or not there have been significant changes during the liberalisation period in demand‐side parameters, such as import coefficients and savings rates, along with jumps in flows, such as annual exports and investment. Following the methodologies presented in Berg & Taylor (Citation2001) and Davies & Rattsø (Citation2001), it is perhaps interesting to look at how output has responded to these shifts, using a simple three‐gap analysis and decomposition of aggregate demand ‘injections’ (investment, government spending, exports) versus ‘leakages’ (saving, taxes, imports). The key point is that in macroeconomic equilibrium, totals of injections must be equal to the total of leakages.
From the identification of contractionary and expansionary factors in effective demand, it is then possible to arrive at the economy's real financial balance. Changes in financial claims against the various institutions (the private, public and foreign sectors) require that some real economic variables have to adjust. On the other hand, for example, a contractionary stance of the rest of the world suggests that some other sector must increase liabilities or lower assets.
To examine what this means for the demand for labour, one can undertake a decomposition of employment growth. It is important here to bring the issue of productivity growth into the picture. In expanding sectors, productivity increases may not necessarily translate into reduced unemployment; in slow‐growing or shrinking sectors higher productivity may even result in a decline in employment. Finally, linking employment growth to productivity gains during the period of liberalisation may point to important shifts in earnings patterns.
This article provides an overview of macroeconomic performance and trends in South Africa since 1990. It employs various forms of decomposition analysis to describe important trends. After a brief background to the South African scene in the next section, we start our analysis with a three‐gap exposition of savings and investment in the country. This is followed by a decomposition of aggregate demand into leakages and injections, and a concluding section.
2 CONTEXT
It is useful to have a ‘story’ on which to peg any economic analysis. In the context of the overall review programme in which this article is being written, our stylised story is as follows. Like most economies in the new millennium, South Africa's main macroeconomic policy problem is how to manage globalisation. Most economies faced increasing globalisation pressures in the 1990s. Without taking a pro‐ or antiglobalisation stance, we would argue that globalisation confronts any national economic policy maker with the need to balance two important competing considerations. On the one hand, globalisation constrains national economic policies. Certain policies are inconsistent with the forces of globalisation and reduce the benefits that might be garnered from it. For example, a fixed exchange rate coupled with fiscal indiscipline can undermine export competitiveness, which is essential if the economy is to benefit from expanded global trade. Similarly, fiscal discipline is necessary to attract foreign investment, a key driver of globalisation benefits.
On the other hand, there are likely to be domestic social demands that need to be addressed. The logic of social policies often appears to run counter to the logic of globalisation, particularly in the short term and during any transitional process of opening up. Good policy is a matter of managing these two: how do policy makers ensure that legitimate domestic social concerns are not sacrificed in the pursuit of macroeconomic policies consistent with liberalisation and other forms of globalisation? Domestic social concerns are not only legitimate in their own right; failure to meet them may also threaten the success and sustainability of any benefits from globalisation.
History has perhaps sharpened this dilemma for South Africa, which faced – and still faces – the problem of managing ‘normal’ globalisation. However, the ending of apartheid compounded this problem in two ways. First, reintegration into the global economy, which started in 1990 after the African National Congress was unbanned, meant there was a double impetus for ‘globalisation’ – reintegration plus subsequent liberalisation (tariff reduction, financial liberalisation) that began in 1994. Secondly, the legacy of apartheid exacerbated both social expectations and the need to satisfy them.
From outside South Africa, it often appears that the policy debate has been between those who feel that immediate needs of social policy should take precedence over macroeconomic concerns, and those who give primacy to macroeconomic concerns. The argument for the latter – seen by many critics to be the stance of official economic policy in South Africa – is that growth requires macroeconomic stability, which will allow social concerns to be addressed more sustainably in the long term. The arguments for giving primacy to the social agenda are that ‘in the long term we are all dead’ – i.e. social needs carry a moral imperative to be addressed immediately – and that failure to address them will lead to social forces that will themselves threaten macroeconomic stability. As argued above, this will threaten the success of globalisation in itself. Bluntly speaking, failure of the public sector to ensure the provision of sufficient education and health care, in the name of fiscal discipline and an investor‐friendly environment, will result in a low‐productivity labour force that will be unable to reap the benefits created in certain sectors by globalisation.
It is always dangerous to present economic policies as sharply opposed alternatives. However, in public debates on policy, a clash is often perceived between macroeconomics and social policy. It is probably helpful to state explicitly our perspective on this. Policy interventions to promote social economic justice (if they work) contribute directly to improving social conditions. In this sense, their intended effects are transparent and apparently easy to understand. On the other hand, the contribution of macroeconomic policies to social conditions is indirect and thus less apparent. It is, therefore, sometimes difficult to understand why there might be a trade‐off between the two types of economic policy. It is also understandable why social activists oppose policy packages that reduce social interventions to maintain macroeconomic stability.
Macroeconomic instability comes about because aggregate demand exceeds aggregate supply by an excessive amount. Broadly speaking, excess aggregate demand means that the amount of goods and services that domestic residents want to use is greater than the amount available. Something has to give, either to increase the availability (e.g. bringing in more goods from outside) or to ration the shortage (e.g. through inflationary redistribution of income to reduce demand).
The exact way in which the economy attempts to adjust back to balance will depend on a host of factors, such as institutional norms in wage setting and various policy frameworks that are in place. Thus, for example, if the authorities maintain a fixed exchange rate, the instability is likely to manifest itself as a current account deficit and a debt crisis. With the price fixed, the excess demand is satisfied by drawing goods in from outside. This is only possible if debt is allowed to rise. If, however, the authorities allow the exchange rate to float, excess demand will manifest itself in a depreciating exchange rate. In effect, the depreciating currency raises the price of goods for domestic buyers and reduces the excess demand through a price mechanism.
Both these processes may be accompanied by inflation. A depreciating exchange rate can start off a cost‐push inflationary episode, which might be sustained depending on the relative strengths of various social forces trying to protect their real incomes. With a fixed exchange rate, that part of excess demand that cannot be satisfied by rising imports will cause domestic inflation. At a given exchange rate, domestic inflation will reduce competitiveness and place further pressure on the current account. This will cause either further depreciation or debt accumulation, depending on the exchange rate regime. There is thus a reinforcing feedback.
These responses impact negatively on social conditions. Inflation generally has a negative distributional impact. Diminishing export competitiveness leads at best to slower growth and at worst to job losses. The social response to these privations can set off various vicious circles.
It seems clear that, in principle, those concerned with social policy should also be concerned with macroeconomic stability. However, while it is both theoretically and empirically incontrovertible that macroeconomic instability is harmful for society, this does not mean that macroeconomic stability necessarily and automatically promotes good social conditions. It is necessary that the enabling environment it provides be used to the good. This may require social intervention on the part of the state. However, this does not imply that any social intervention will be beneficial. Many such interventions are irresponsible, either because their wider consequences are ignored, or because they are motivated by populist political interests rather than genuine concern for social welfare. Such policies do not provide sustainable solutions to the problems they purport to address. Insofar as they fail to consider their macroeconomic implications, they may contain within themselves the seeds of their own failure.
Any given change in a ‘standard’ measure of macro‐instability – the current account balance, budget deficits and inflation – can be explained by a number of different immediate ‘causes’. For example, an increased current account deficit must, by definition, mean that availability of foreign exchange has grown more slowly than its use. Questions arise, for example, whether the proximate cause was falling exports or rising imports. Perhaps both were rising, but exports not as fast as imports? Or perhaps the deterioration reflects changes in the invisible account? Similarly, the implications of a budget deficit and the appropriate policy responses are different if they came about because of rising current expenditure, falling revenue or rising capital expenditure. Distinguishing between these different immediate ‘causes’ gives some insight into underlying macroeconomic forces driving the macroeconomy.
We use a decomposition method that allows us to begin to distinguish between such immediate ‘causes’. The decomposition focuses on components of aggregate demand. In effect, it assumes that there is excess capacity in the economy, so that aggregate output is determined from the demand rather than the supply side. Although we have not tested this assumption, it is worth thinking about it. It assumes that any increase in demand can be met by increased use of existing capacity rather than by having to raise factor prices. This implies that, if we were able to control for capacity growth and other factors, aggregate output and the aggregate price level would not be correlated. One should therefore look at the relationship between these two variables. Our time‐frame (1993–2002) is inadequate for econometric analysis using annual data, but we can derive some impressionistic patterns from the data.
shows growth rates for nominal and real gross domestic product (GDP) and inflation rates, as measured by the GDP deflator. The rates may differ from the conventional wisdom in South African literature. They have been calculated as continuous growth rates [= ln(x1) – ln(x2)] and are thus smaller than the conventionally estimated year‐on‐year percentage changes. The advantage of the continuous growth rates is that the growth rate of a variable which is a product of components is equal to the sum of the growth rates of the components. In there is a noticeable change after 1993 – inflation rates are lower than previously while real growth is invariably positive. As we expect both these variables to be autocorrelated (one year's value is highly correlated with the previous year's), we also plot the growth rates ().
Growth rates for nominal and real GDP
Figure 2
Casual observation suggests that between 1993 and 1996 there was an inverse relationship between the two rates. Since 1996, however, there has hardly been any relationship – growth rates have varied between 0,8 and 4,2 per cent, but inflation has varied between 6,0 and 8,2 per cent. This possibly suggests that the inflation–growth nexus changed after 1996. Between 1997 and 2002, the coefficient of variation for inflation is 8,6 per cent while that for growth is 36,5 per cent. The correlation coefficient between the two is 0,99 for 1993–6 and 0,46 for 1997–2002. Prior to 1996, there seems to be a trade‐off between the two variables, in which higher growth is associated with lower inflation. After 1996 the relationship becomes less clear, with large variations in growth while inflation moved in a narrow band. Although this evidence is by no means definitive, it does not suggest that it is entirely inappropriate to proceed with the demand decomposition. It also suggests – very tentatively – that there might have been a change in the inflation–growth nexus after 1996.
A ten‐year review seems an appropriate time to examine how these (possibly) contradictory forces have been balanced in practice at the macroeconomic level. The decomposition methodology we employ allows us to begin to distinguish between sources of outcomes.
The focus of the study is from 1993 to the present. It is useful to divide the overall period into two: preliberalisation (1993–7) and the liberalised period (1998–2002). shows South Africa's trade ratio from 1981–2002. The trade ratio is measured as the sum of export and import values to GDP, in current prices, and is a commonly used crude measure of how open an economy is.
South Africa's trade ratio from 1981–2002
For the period under review one can see that the trade ratio started to grow in 1992, perhaps reflecting post‐apartheid reintegration. The slowdown in 1997–9 was probably related to the Asian crisis, but might also reflect the ending of the impetus provided by the abolition of apartheid. The acceleration after 1999 reflects both world recovery and domestic liberalisation policies starting to make an impact. The ratio's average annual rate of growth was 5,5 per cent between 1993 and 1996, 0,8 per cent between 1997 and 1999 and 9,8 per cent between 2000 and 2002.
3 A GAP ANALYSIS FOR South Africa
It is useful to begin our decomposition analysis with a ‘gap analysis’ of savings and investment, a non‐behavioural framework for showing interactions between private, public and foreign sectors. We start by setting out the framework and then look at the results it gives for South Africa. The gap framework is an accounting approach derived from the national accounts system. Our national income accounting identity gives:
(-1)
where Y is the GDP at market prices, M is imports of goods and services, C is final current expenditure (summed over households and the general government), I is gross capital formation (or domestic investment) and X is exports of goods and services.
This simply reminds us that, ex post, the total use of goods and services in any given period must be equal to their total availability. This identity is elementary to all economists. It is ex post because any increase in stocks (inventories) is investment and is thus used. We use the terms ‘availability’ and ‘use’ rather than ‘supply’ and ‘demand’ to emphasise the absence of any behavioural assumptions in the identity. It is sometimes suggested that the identity is not ‘true’ because of activities such as home and informal sector production. However, this is a question of measurement, not of the validity of the identity.
Rearranging and remembering that Y − C = S, i.e. domestic savings, gives:
(-2)
Domestic investment can only exceed domestic savings if imports exceed exports. As we are dealing with ex post magnitudes, this relationship derives from a consistent set of national accounts, not from any particular theory of economic behaviour. It does, however, carry a basic truth. A country can only use more goods and services than it produces (I > S or I − S > 0) if there is a net inflow of goods and services from outside (M > X or M − X > 0).
It is more usual to present the gaps as:
(-3)
In this form, use of domestic resources in excess of domestic availability shows up as a negative domestic gap (S − I < 0). Similarly, a negative foreign gap is the same as a current account deficit on the balance of payments (X − M < 0). The left‐hand side is called the ‘domestic gap’ while the right‐hand side is the ‘foreign gap’.
There are two final modifications we need to make before applying the framework. As national savings are normally measured as the difference between gross national disposable income (GNDI) and current (private and public) consumption, it is convenient to present identity (3) as:
(-4)
where NFA = net factor payments from abroad, and NTA = net transfers from abroad. This is simply a matter of being consistent in our measures of national income (Y) and savings (S) and the foreign sector (X and M):
| 1. | If Y = GDP, then S = gross domestic savings and X−M represents the balance of trade. | ||||
| 2. | If Y = GNP (= GDP+NFA), then S = gross national savings and we must add NFA to X−M. | ||||
| 3. | If Y = GNDI (= GNP+NTA), then S = gross savings, and we must add NTA to X−M+NFA. | ||||
The last of these is the normal usage of the term ‘savings’. It is also the most useful for us to employ in our later analysis.
This modification means that the right‐hand side of identity (equation 4) shows the full current account balance of the balance of payments. Strictly speaking, the right‐hand side of identity (equation 3) shows the balance of trade, not the current account balance.
Secondly, we can further decompose the domestic gap into a private and a public sector. In fact, provided our accounts are consistent, we can decompose it in numerous different ways. For example, it would be possible to break it down into provincial gaps if the data were available. In this case we first have to define the two sectors. In this article we equate the public sector with ‘general government’ and add public corporations in with the private sector. This is necessary because the data do not allow us to identify the income of public corporations separately.
We define government income (Yg) as government current income less interest payments on public debt, less government subsidies, less government current transfers to households, less government transfers to the rest of the world. This explicitly treats the government as a pure intermediary between benefactors and beneficiaries as far as transfers are concerned. We then define private income (Yp) as GNDI (at market prices) less government income (Yg).
Given the breakdown of current consumption expenditure between the private and public sectors as per the national account, we can then derive savings in the two sectors: Sg=Yg – Cg and Sp=Yp – Cp. This allows us to decompose the domestic gap into:
(-5)
Ideally we should present the data in constant prices. However, inconsistencies in the national accounts source data prevent this and so we rather present the data in current prices as percentages of GNDI. This amounts to assuming that the deflator is the same for all components of national income. Data sources are shown in .
Data sources
shows the three gaps for South Africa since 1990. To interpret the table, we begin with the foreign gap in column 10. This is the current account balance (CAB) on the balance of payments. Note that:
| 1. | The CAB is relatively small. The absolute value of the gap averaged 3,3 per cent of GNDI in the 1980s, 1,1 per cent between 1990 and 2002, and 0,8 per cent since 1997. This low ratio is what one would expect with a floating exchange rate regime – exchange rate adjustments prevent the CAB from departing too far from zero. It might be asked why it is not exactly zero. In part, this could reflect rigidities in the exchange rate adjustment process. It should be noted, however, that the year‐to‐year changes used in the decomposition reflect short‐term movements in the CAB, and thus reflect temporary disequilibrium in the market. | ||||
| 2. | The balance changes from being positive in the preliberalisation period to being negative in the liberalised period. The direction of this change is consistent with the expected impact of the financial liberalisation that took place simultaneously with the trade liberalisation. Financial liberalisation should lead to inflows of foreign capital. Thus the capital account of the balance of payments should be in surplus, which should be matched by a deficit on the current account. The fact that it is small shows that, whatever financial capital flows took place, the net flow of resources associated with this was small. This raises the complex issue of the ‘transfer problem’. A real capital inflow only takes place if there is an associated rise in imports. | ||||
The three gaps in South Africa (% of GNDI)
With the foreign gap being kept close to zero, the two domestic gaps are forced to be almost mirror images of each other. We see this in columns 3 and 6 of , where negative public sector balances match positive private sector balances. Three features merit comment:
| 1. | Private sector balances are positive throughout the period. In other words, private savings (by households and corporations) exceeded private investment. Because of the CAB constraint, this implies that the public balances were negative. | ||||
| 2. | The private gap declined as a percentage of GNDI throughout the period, while the relative public gap increased (i.e. became a smaller negative number). | ||||
| 3. | There is no discernable break in the trends. Rapid liberalisation in the second part of the period appears to have had little impact on the observed trends. | ||||
The second feature warrants looking at the components that lie behind the domestic gaps. As shows, both private and public investment remained relatively constant as shares of GNDI from 1993 onwards. The changing gaps were driven by the decline in the private savings rate and the rise in the public savings rate. This is vividly illustrated in .
Figure 4
It is important to emphasise that this analysis does not show causation, as there is no behaviour in the ‘model’. The correct interpretation at this stage is that, with the foreign gap being kept close to zero by exchange rate policy, any change in one of the domestic gaps has to be accommodated by an opposite change in the other. The framework is often used to show that a rise in the public deficit must cause a fall in the private surplus – a typical ‘crowding out’ scenario in which the private sector is ‘forced’ to finance a budget deficit. However, such an interpretation is only possible if one has a model – at least an implicit one – of how causality works in the economy. It could equally be argued that the high private surplus (that arises because of high real interest rates or simply lack of investment opportunities due to low domestic demand) ‘causes’ the public sector to appropriate resources for current and capital spending. What else can the private sector ‘saver’ do than buy government bonds if there are no other investment opportunities?
From it would seem that the attempt to reverse crowding out did not bring about the desired increase in private sector investment. Instead, private sector savings went down, i.e. every rand saved by the public sector is associated with a decline in private sector savings of more or less the same amount. A downward trend in private sector savings is often observed when financial or trade liberalisation takes place. Prior to liberalisation, limited spending outlets might lead to an artificial propping up of savings. In addition, opening up to foreign competition may squeeze profits and therefore savings by firms. Any changes in private savings, however, do not appear to have had any impact on private investment, suggesting that savings are not really a driver of investment in South Africa.
It is instructive to consider what the immediate effects might have been if the changes to the private gap required to match the concomitant change in the public and foreign gaps had come about with savings remaining constant and investment rising. A simple calculation shows that if the private gap over the period remained as it was but was brought about by the private savings rate remaining at its 1994 level (23,3 per cent) while investment rose accordingly, there would have been an additional R221 billion (in 1995 prices) of investment over the period 1994–2002. This is almost three times the level of investment in 1995, and would likely have had a significant effect on growth and employment.
The usefulness of this framework is that it provides a descriptive picture of where we need to start looking for explanations of the macroeconomic performance. In the following section we take the method of decomposition a little further.
4 STANCE AND MULTIPLIER ANALYSIS OF AGGREGATE DEMAND
The previous section suggested that a major macroeconomic trend in South Africa has been the reduction in the private surplus/public deficit. This appears to be driven by action on savings, in both the public and the private sectors, while investment in both sectors remained relatively constant. We now analyse this theme further.
To do so we use a technique developed by various authors in Taylor (Citation2001). From standard Keynesian analysis we know that any change in an exogenous demand component (an ‘injection’) will have a multiplier impact on aggregate demand. Its size depends on the size of the injection and the size of the multiplier. The latter depends inversely on the size of leakages relative to aggregate demand.
The technique we use allows us to decompose changes in the impact into those due to the multiplier and those due to the injection. In simple terms:
(-6)
where X is aggregate demand, k the multiplier and J exogenous final demand (the injection).
Comparing one period with the previous we can write
(-7)
Some simple manipulation allows us to write:
(-8)
or:
(-9)
Thus, the change in aggregate demand is decomposed into an ‘injection effect’ and a ‘multiplier effect’. The third term is an interaction term, which will typically be small.
shows the application of this to South Africa for the period under review. The first three columns show the levels of aggregate demand (GNDI plus imports), injections and the multiplier, respectively. Columns 4–6 show year‐on‐year changes in each of these, while columns 7–9 show the decomposition. As an example, consider 1996 where column 6 shows that aggregate demand rose by R33 billion. The decomposition shows that if injections had risen as they did (by R24 billion, column 4), but the multiplier had remained at its 1995 level (1,991, column 3), aggregate demand would have risen by R48 billion. Column 5 shows that the multiplier fell by −0,044 in 1996. If this had been the only change that occurred, aggregate demand would have fallen by R14 billion (column 8).
Overall decomposition of changes in aggregate demand (R billion, constant 1995 prices)
In most years, the two effects tend to offset each other, although in 1995 and 1997 they are reinforcing. Over the whole period, the injection effects dominate the multiplier effects. This is to be expected, as the nature of the multiplier limits the size of the changes that can occur. The value of the multiplier is constrained by the fact that the leakage parameters are virtually restricted to vary between 0,1 and about 0,4 or 0,5. Column 10 shows the ratio of the absolute size of the multiplier to the injection effects. There is no discernible trend, although in the last three years both effects seem to be larger than previously. One might want to see in this some early signs of a change in the relationship, but it is not statistically significant.
5 DECOMPOSING THE COMPONENT CHANGES
Following the gap analysis in the earlier section, the components of injections and leakages can be grouped according to the sector most ‘responsible’ for them – the private sector (investment and savings), the public sector (government expenditure and taxation) and the foreign sector (exports and imports). We can then further decompose equation 9 into changes emanating from each sector and those due to injection and leakage changes within each sector.
The rationale for doing this is that each set of actors is responsible for compensating or reinforcing changes. This is most clearly seen with regard to the public sector. The government may have the intention of cutting its expenditure in the context of fiscal reform, which will reduce aggregate demand. However, if at the same time it raises average tax rates due to better tax enforcement, the multiplier falls and the impact of the expenditure cut on aggregate demand is reduced. Similarly, increased private sector investment raises aggregate demand, whereas a rise in its propensity to save reduces aggregate demand. (Strictly, it reduces the multiplier that determines how exogenous changes impact on demand.) There are two alternative approaches to decomposing them and we briefly look at each.
5.1 Berg–Taylor decomposition
Berg & Taylor (Citation2001) suggest that one can think of a particular injection relative to its leakage parameter as reflecting the stance of the sector. Thus, the stance of the private sector is I/s, of the public sector it is G/t and of the foreign sector it is E/m, where I and E are defined as before, G is government current and capital expenditure, s is the private sector savings rate, t the tax rate and m the propensity to import. The stance shows the sector's ‘own’ contribution to aggregate demand – how much its injection translates into aggregate demand when passed through its ‘own’ multiplier. gives the relevant figures for South Africa. These figures have to be interpreted distinctly differently from the previous table.
Stances of the three sectors
The simplest way to think of the figures is that they show what demand would be if only that sector's injections and leakages operated. Thus, actual demand in 1997 was R709 billion. If government spending, exports and the tax and import parameters had all been zero so that only the private sector stance determined demand, it would have been R501 billion. Similarly, the public sector stance would have led to demand of R1,061 billion and the foreign stance to R669 billion.
Clearly, it does not make sense to think of the absolute values: the experiment of cutting out all sources of injection and all leakages except one is too difficult to contemplate. However, the trend relative to the average does provide some insight. The ‘average’ stance is simply actual demand. One can derive this by taking a weighted sum of the three stances, where the weights are the ratio of the sector leakage to total leakages. also presents the stances as ratios of this average, and they are plotted in .
Figure 5
The public sector stance is consistently above the average, showing that it was ‘pumping demand into the economy’. However, there has been a consistent decline in its relative size, showing that the amount by which it has boosted demand has fallen over the period. The average public stance for the preliberalisation period is 155,2 per cent. After liberalisation it falls to 121,0 per cent. For the private stance the opposite picture is true: it is consistently below the average, but rises over the period.
The foreign stance is stable and tracks the average. This is what one would expect with the exchange rate regime. The foreign stance is E/m. As m is defined as M/X, expressing the stance as a percentage of X reduces it to E/M. If M is constrained by exchange rate policy to be equal to E, the foreign stance is constrained to be equal to 100 per cent. As the CAB is not allowed to move much out of balance, imports and exports must match each other, so that increases in injections (raising the stance) must be matched by countervailing increases in leakages (reducing the stance).
For this article we have ‘exports’ reflect all current sources of foreign exchange (i.e. exports of goods and services plus factor payments from abroad plus transfers from abroad). Imports are similarly defined. This is employed to maintain consistency between national savings and the rest of the national accounts. If we measured only domestic savings (omitting foreign primary income flows and transfers), the foreign stance is raised to about 110 per cent of the average and the private stance is reduced accordingly. The trends are unaffected.
This preliminary analysis points to the changing roles of the private and public sectors as drivers of demand in the economy. The changes are relatively steady, rather than discontinuous around the break in trend due to our assumed change in liberalisation. An initial interpretation suggests that the change was driven by changing domestic policies rather than liberalisation. However, taking the Berg–Taylor decomposition further gives us more insight into the process.
While the stance of a sector gives us some idea of the impact of that sector, it ignores the relative size of the stance in the determination of overall demand. Thus, our figures above show that the private sector stance was well below the average, whereas the public sector stance was well above. However, which of these two really drove demand depends on their relative sizes. In other words, we could find the same results as above with a very small public sector and a very large private sector, or vice versa. Obviously the movement of actual demand would differ.
To take relative sizes into account we need to weight each stance. Some simple algebra shows that the relevant weight is the share of the sector's leakage in total leakages. shows the trends in this respect.
Figure 6
Some observations can be made:
| 1. | The relative weight of the public sector is low throughout the period. Thus, although the public sector's stance is expansionary, albeit declining, the weight of its contribution to demand reduces the impact. | ||||
| 2. | The falling relative weight of the private stance runs counter to the increasing trend in its size (seen in Figure 5). Thus, the increasing size suggests that the private sector has become more important as a driver of demand, while its falling weight suggests it has become less important. | ||||
| 3. | The weight of the foreign stance has increased noticeably. We saw that the stance was relatively neutral, as a result of exchange rate management. However, the trend in the weight suggests that its role as a driver of demand has risen. | ||||
| 4. | Although the trend in the weight of the public stance is not as pronounced, it does appear to be rising. Thus, while the public stance has fallen over the period, its weight has risen. The policy intention with regard to the public sector has been to reduce its role in the economy and to maintain fiscal discipline. The falling trend in the stance is consistent with this intention. However, the rising weight runs somewhat counter to it, suggesting that there has been a slight conflict between the two prongs of fiscal policy – the difference between expenditure and revenue and their size. It should be noted that this inconsistency does not necessarily imply inconsistencies in the actions of the government. Being the relative size of ‘its’ leakage, the weight of a sector's stance can change either because its leakage has been raised, or because the leakages of other sectors have fallen. It thus captures the net outcome of all leakage changes, rather than sector‐specific changes. It is perhaps appropriate to re‐emphasise here that the framework does not show causality. | ||||
What we have observed thus far is that the impact of different sectors on overall demand is mediated through ‘their’ injections and ‘their’ leakages. We can undertake one further decomposition that gives some insight into the relative importance of changes in these components.
5.2 Davies–Rattsø decomposition
Davies & Rattsø (Citation2001) follow standard decomposition lines by asking what the effect of the change in one component would have been if none of the other components had changed, instead of if none of the other components had existed, as in the Berg–Taylor decomposition described above. The algebra is derived by breaking equation 6 into its component parts. Thus:
(-10)
We want to examine the sources of change in X from one period to the next. The changes could come from changes in components of the multiplier or from changes in components of the injections. For illustrative purposes, we consider the private sector. If the only change had been a change in the private sector savings rate, the change in X would be all due to a multiplier effect. We could write:
(-11)
Similarly, if the only change had been a change in investment – a private sector injection effect – we could write:
(-12)
We can follow this decomposition through all six of the components of changes in final demand. Thus:
(--1)
provides the results for South Africa for the two periods, which are also illustrated in . The decomposed elements sum to overall growth in aggregate demand, as the decomposition is additive. If we begin with the overall effects, aggregate demand (GNDI) rose by 4,0 per cent per year in the first period and 3,5 per cent in the second (row 10). In the first period, the private sector was the main source of growth, accounting for 3,2 percentage points (row 11). This contribution dropped by almost half in the second period. In this period, the foreign sector was the dominant source, contributing 2,4 percentage points. This is what might be expected as the economy was opened up. In both periods, the net effect of the public sector was small (row 12).
Davies–Rattsø decomposition
Figure 7
Before looking at the decomposed effects, it is worth recalling what the gap analysis showed. There it was noted that the foreign gap was insignificant. How can this be reconciled with our finding now that it is significant? The gap analysis shows the movements in exports and imports. As has been stated repeatedly, a policy of ensuring that the CAB does not get significantly out of balance ensures that this foreign gap always remains around zero. The demand decomposition, however, separates the effects of these two components. On their own, each has a significant impact; taken together they offset each other.
Looking now at the decomposition into injection and leakage effects, we see that in the first period the injection effect of the foreign sector dominated the others. If only exports had risen in this period, with all leakages and all other injections remaining constant, then aggregate demand would have risen by 2,3 per cent per year on average (row 4). This effect is consistent with the expected impact of post‐apartheid reintegration. However, it was counteracted by the simultaneous rising propensity to import; the effect of this would have been to reduce aggregate demand by 1,7 per cent. Thus the impact of the foreign sector was somewhat muted.
By contrast, in this period the private sector contributed to aggregate demand both because investment was rising (contributing 1,4 per cent) and because the propensityto save was falling (contributing 1,7 per cent). Again, these findings are consistent with post‐apartheid expectations. The net effect was to make the private sector the largest source of growth in this period.
In the liberalised period (1998–2002) the pattern changed somewhat. Although the two effects counteracted each other in the foreign sector (as previously), the net effect was bigger. The expansion of exports was much faster than the rise in the propensity to import. In the private sector, the falling propensity to save remained dominant, while the injection from investment became smaller.
Most interesting is the role of the public sector. While the injections through government expenditure were positive in both periods, they declined in the latter period. Also, the leakages, through higher tax rates, have become more prominent during the latter period. As a result, total contribution has become negative in real terms.
These results seem to run counter to the standard ‘crowding‐out’ approach to public‐private sector interactions (which underlies the Washington Consensus). The policy thrust with regard to the public sector has been to reduce its size to make space for greater private sector action. However, suggests that the private sector has not expanded to fill the space vacated, as its total contribution to aggregate demand, although still positive, has declined.
The other leg of the liberalisation framework – opening up to the global economy – has been more successful in that, when comparing the two periods, the contribution made by the foreign sector has increased considerably. As mentioned above, this is the net effect of increased injections, offset by higher leakages through imports.
| a. | The falling savings rate raised its ‘own’ multiplier, so that stagnant investment contributed more. | ||||
| b. | However, falling savings relative to other leakages also meant it contributed less to demand. | ||||
| 1. | The public sector's role as a demand driver has been reduced over the period. | ||||
| a. | The rising ‘tax’ rate reduced its ‘own’ multiplier, but not enough to offset the effect of falling expenditure. | ||||
| b. | However, constant ‘taxes’ relative to other leakages meant that this sector contributed less to demand. | ||||
| 1. | The foreign sector's role has been increasing. | ||||
| a. | The rising import rate reduced its own multiplier, but this was offset by rising exports. | ||||
| b. | However, rising imports relative to other leakages increased the contribution of the sector to demand. | ||||
These conclusions suggest many questions regarding the changing macroeconomy of South Africa. We raise some of these questions in our final section.
6 POSSIBLE POLICY IMPLICATIONS
As the investigations undertaken in this article are based on a non‐behavioural methodology, they do not show cause and effect. One therefore needs to be careful when drawing policy‐relevant conclusions from them. At best, the analysis provides some guidance about the broad framework within which specific policies, based on more detailed research, must operate. More often, in fact, our policy implications suggest areas in which further knowledge should be sought before detailed policies can be designed. We have taken this approach in part because other studies in the ten‐year review provide more detailed investigations of specific policy areas. We also feel, however, that it is important for policy makers to consider the constraints imposed on specific policies by the requirements of macroeconomic consistency, and to consider the interrelatedness of policies with different targets – even policies designed to affect completely different sectors.
The main conclusion we have drawn from the demand analysis is that the past ten years have seen the public sector's role as a demand driver decline and that the ‘space’ thus created has been taken up by the foreign rather than the private sector. This raises some questions about the impact of policies followed over the past decade, as well as the direction policy should take in future. Our analysis allows only the broad objectives of such policy to be sketched, rather than the detailed instruments. We believe the evidence suggests that the consequences of reducing the role of the public sector have not been as positive as might have been hoped. However, the analysis does not provide evidence as to why this is the case. We have noted that the reduced public deficit was matched by a falling private savings rate rather than rising investment. As indicated earlier, this entailed a sizable ‘lost opportunity’ for investment, growth and employment creation. A central question for future policy has to be how to reverse this trend. Our article does not provide an answer, but might present some relevant considerations.
To understand what kinds of policies might be appropriate, it is useful to think about whether the observed trends were primarily the result of macroeconomic policies – which might be reversible in the short term, or of more deep‐seated structural features of the South African political economy – which might be somewhat more difficult to address.
We have speculated at various points in the article as to why the trends might be policy‐driven. Trade liberalisation and deregulation might cause savings rates to fall by removing constraints on spending, particularly for high‐income groups that might be expected to be important savers. At the same time, the policy of high interest rates targeted at restraining inflation may also have restricted investment. Even if investment is not particularly sensitive to interest rates, it is possible that the tight credit regime they imply also restricts investment.
If these were in fact the drivers behind the trends, policy makers should begin by reconsidering the high interest rate policy. Have the monetary authorities been overly cautious because of their concerns over inflation? If it is necessary to restrict demand, can the effects on investment be countered through the fiscal system? The message is relatively straightforward: if the stagnant investment relative to GDP is the result of policy, the policy should be reversed. The costs of the policy – in terms of foregone investment – should be compared with its benefits in terms of lower inflation.
However, even if investment could be raised this easily – if none of the other components of the three gaps was changed (by policy) – an adjustment process would be set off to rebalance the gaps. To control the process, policy should attempt to continue to reduce the public deficit (cutting expenditure or raising the effective tax rate) or to reverse the decline in the private savings rate. Alternatively, increased foreign investment could allow the foreign gap to move into deficit, accommodating the rising domestic investment. In our view, raising the savings rate would be the first prize. However, unlike investment, a policy‐induced decline in the savings rate is difficult to reverse. If, as we have speculated, it is due to liberalisation, it reflects a structural change adjusting to the new liberalised economy. In other words, the current configuration of the liberalised economy is consistent with a low savings rate. It would not seem sensible to try to reverse this by reversing the liberalisation. Rather, policies should be targeted at trying to raise the rate within the new economy. Further research into the nature of savings in South Africa needs to be undertaken. Policies addressing this issue might include encouraging the development of new instruments for mobilising savings or using the tax system to tackle declining corporate savings rates.
However, it is not at all clear that the causes of the trends are short term and reversible. The savings rate has been falling fairly consistently since 1985. As we have noted throughout the article, there does not appear to be a break in the trends as the second phase of liberalisation commenced. Casual evidence therefore suggests that the reasons for the trends are more structural. It is likely that both ‘uncertainty’ and ‘historical tensions’ between the new government and the old private sector play contributory roles. If this is the case, policies have to be much more broadly based. In essence, the first set of explanations says that the trends are due to management issues, while the second set is related to the whole social, political and economic environment. It would seem much more difficult to change the latter.
Detailed studies would have to be undertaken – or past studies reviewed – to come up with a clear list of what such structural inhibitions to increasing private sector investment and/or the private savings rate might be. It is likely that they all operate by raising uncertainty, which encourages both investors and savers to adopt a ‘wait and see’ policy. We do not know the effects that emigration has had on the savings rate, but given the economic status of the émigrés, it is likely to have reduced it. We would also want to explore whether ‘black empowerment investment’, which is necessary to reverse historical inheritances, has led to a phase in which investible funds are used primarily for transferring rather than creating wealth.
However, one of the immediate policy implications is that it may be necessary to reconsider the applicability of the ‘crowding‐out’ model to macroeconomic policy in South Africa. Two considerations are important. First, the standard basis for crowding out comes from a comparative static framework in which speed of adjustment is not considered. It is essentially assumed that as one sector contracts, the other simultaneously expands. If the response of the private sector is sluggish, the pace of contraction of the public sector has to be reconsidered. If it is too rapid, it may induce a deflationary contraction of the economy. Secondly, if there are structural obstacles that inhibit the expansion of the private sector into the space created by the reduced public sector, there may be a need for continued public sector intervention during a transitional period. Fiscal stability on its own is unlikely to stimulate investment.
This latter consideration suggests that policy makers should explore public–private partnerships to encourage private investment.
The foregoing discussion is focused on trying to expand the role of the private sector. Our analysis also showed that the foreign sector has become the dominant demand driver in the economy. This raises another set of policy considerations related to the increased exposure of the economy to global economic trends. The question is how policies should be designed to manage the risks this entails.
Liberalisation is intended to raise ‘microeconomic’ risk in the sense that firms are subject to greater competitive pressures. The risk we are concerned with is macroeconomic. Any decline in export demand caused by a downturn in the world economy will translate into a greater reduction in aggregate demand than it would have previously. In 1994, a 10 per cent decline in exports would, ceteris paribus, have caused a 3,8 per cent decline in aggregate demand; in 2002 it would have caused a decline of 5,1 per cent.
All our ‘analysis’ can suggest in this regard is that policy makers have to be aware of the changed circumstances and consider mechanisms by which any contractionary impulses imported from the global economy might be counteracted by domestic demand management. However, the area of country risk management is one that merits serious investigation.
References
- Berg, J, and Taylor, L, 2001. "External liberalization, economic performance, and social policy". 2001, CEPA Working Paper Series I, Working Paper No. 12, Centre for Economic Policy Analysis. New York: New School University. Available online at www.newschool.edu/cepa.
- Davies, R, and Rattsø, J, 2001. "Zimbabwe: economic adjustment, income distribution and trade liberalisation". In: External liberalization, economic performance, and social policy. 2001, In Taylor, L (Ed.), New York: Oxford University Press.
- "Quarterly Bulletin"South African Reserve Bank (SARB), undated, various years. Pretoria: SARB.
- Taylor, L, 2001. "External liberalization, economic performance, and social policy". 2001, New York: Oxford University Press.