FOREIGN CAPITAL INFLOW, DOMESTIC CREDIT AND PRIVATE INVESTMENT IN SWAZILAND

Adam B. Elhiraika *

Abstract: Historically, private investment in Swaziland relied heavily on foreign resources, which have been rapidly declining in the 1990s. This article examines the role of domestic credit and interest rate policy in promoting desired and sustainable investment rates in the country, in the context of the common monetary agreement. While foreign capital inflow has a significant effect in the long-run, private investment is found to be always responsive to changes in bank credit to the private sector. The paper argues for increased mobilisation of domestic resources and non-credit support to domestic enterprises as means for stimulating domestic investment and employment.

1. Introduction

The economy of Swaziland is characterized by a relatively well-diversified and developed financial system as well as a stable macroeconomic environment. Yet it has continued to depend heavily on foreign capital to finance investment needed to boost employment and economic growth. During the period 1976-1998, the average domestic investment rate was 27.9% of Gross Domestic Product (GDP), while the private investment rate was 17.5%. These rates have fluctuated widely over the years, possibly due to fluctuations in net foreign capital inflow, which represented 13.2% of GDP over the same period. Foreign capital flows are dominated by long-term direct investment.

The ponderous dependence on fluctuating foreign resources has raised concern regarding future employment creation and economic growth. The 1996 Annual Report of the Central Bank of Swaziland (CBS) noted that no substantial investment inflows have been seen since 1994, and since then the expansion of the economy relied on reinvestment undertaken by existing firms. The report remarked that such investment, in addition to being erratic, has focused on improving efficiency with little impact on employment. The reasons for falling capital inflows as seen by investors include infrastructure problems especially in the form of expensive and unreliable power supply, inadequate communication facilities, industrial unrest due mainly to pressures for political reform, and government bureaucracy which causes delays in the processing of trading licenses and work permits. The CBS report concluded that the removal of these impediments and availability of information needed by potential investors would create an environment that is conducive to foreign investment.

It has been argued that achieving a sustainable rate of growth in domestic investment through efficient mobilization and allocation of domestic resources is imperative for Swaziland to promote long-run economic growth and stability. `The emphasis of development policy should be on the supply side of the economy - through the mobilisation of capital and provision of human resources' (World Bank 1996, 17). The World Bank's (1996) study observed that encouraging domestic savings has not been a high priority given the large amount of savings that already exists in the country. Indeed the real deposit rate was negative over most of the period up to early 1990s. This is in sharp contrast with recent directions in finance and development literature, which suggests that the development of the domestic financial system can have a strong positive effect on the rate of economic growth (e.g., Ndikumana 2000; Clarke 1996; Montiel 1996). These effects come through one or more of such channels as increased investment efficiency as emphasised by, for example, Greenwood and Jovanovic (1990), increased savings volume, or efficient mobilisation of existing savings.

This paper examines, among other things, the role of the domestic financial system in Swaziland in supporting private investment in order to enhance economic growth and reduce the country's vulnerability to factors affecting foreign capital inflow. The analysis takes stock of the constraints, especially the limited room for independent monetary policy, due to Swaziland's membership in the Common Monetary Agreement (CMA). In addition to Swaziland, this agreement includes South Africa, Namibia and Lesotho. The currencies of Swaziland, Namibia and Lesotho are on a par with the South African Rand and there is free mobility of funds among these countries. Among a set of co-integrating variables the paper finds that the flow of domestic bank credit to the private sector has a positive and significant effect on private investment. After highlighting the potential and constraints of the domestic financial system, the paper emphasises the need for policies to enhance the availability of investment loans as well as non-credit support to domestic enterprises.

The following section looks at the economic performance, policy and investment environment in Swaziland, while section 3 reviews the literature. An empirical model of private investment for Swaziland is specified in section 4. Section 5 presents estimates of the model and analyses the findings. Finally, section 6 concludes with some policy recommendations.

2. Economic Performance, Policy and Investment Environment in Swaziland

The analysis in this paper attempts to answer two basic questions. First, what determines changes in private investment in Swaziland, and second, what role economic policy could play in achieving and sustaining a desired investment ratio. This section reviews the general economic and policy factors relating to private investment.

As the data in table 1 shows the economy of Swaziland has been relatively stable, though the data averages mask some huge year to year fluctuations. During the period 1976-1998, average inflation rate was 13%, the current account deficit was -13.2% of GDP, and government current budget deficit was -1.25% of GDP. The apparently high current account deficit was associated with high foreign capital inflows¹ largely in the form of foreign direct long-term capital investment. The country has always maintained a sizeable and generally increasing amount of foreign reserves, which averaged US$ 172 million per annum between 1976 and 1998.

The monetary data indicates a relatively low, but remarkably stable, ratio of broad money (M2) to GDP. For the whole period this ratio was 32%. However, this ratio is not a good indicator of financial deepening as it only reflects the ability of the financial system to provide a medium of exchange, and not necessarily credit services (Gregorio and Guidotti 1995, 438). A more important feature of the Swazi financial sector is the exceptionally high ratio of total bank deposits to M2. On average, this ratio was as high as 92.2% during 1976-98, indicating the accessibility of banking institutions. Furthermore, the share of interest-bearing deposits in total deposits amounted to 74% on average, rising from 71% in 1976-80 to 77% in 1991-95. But bank lending to the private sector relative to total bank deposit fluctuated widely between the lowest average of 63.4% in 1986-90 and 84% in 1991-95. The latter period witnessed gradual recovery in the economy after a recession that started in 1989 and bottomed out in 1992 (ECAS 1995, 2). Meanwhile, average bank reserves fluctuated between the maximum of 36.7% in 1986-90 to 21.2% in 1991-95 and to the minimum of 12.2% in 1996/98. The overall average reserve ratio of 30% for the whole period is quite high in view of the fact that the legal reserve ratio² varied between just 4.5% and 6% during the same period.

The 1991-95 period, which saw a high volume of bank lending also witnessed, increased real deposit and lending rates. The former rate rose from -8% in 1976-80 to the average of -3.9% in 1991-95, while the latter ranged between the minimum of -4.9% and the maximum of 2.5%, averaging 0.31% for the period 1976-95. In the period 1996/98, real interest rates were both positive and quite high. In fact the real lending rate has been positive since 1981, and has always exceeded the real deposit rate by a huge margin. The size of this margin is a pointer to a rather poor financial efficiency in the country. However, there is a casual indication in the 1990s that positive real interest rates, increased bank lending and real growth may move in the same direction.

For the entire period under review, Gross Domestic Investment (GDI) slowed down from a peak level in the 1970s to a bottom during 1986-90, but recovered well in 1991-98. Average GDI rate was 27.9% of GDP, private investment was 17.5% (see Figure 1). Gross Domestic Savings (GDS) financed about 16.4% of GDI, and the balance was met from foreign resources. This approximation of foreign resources is quite close to the actual net foreign capital inflow of 13.2% of GDP during the same period. The average real GDP growth rate of 5.84% was remarkably high compared to growth rates in developing countries during that period.

Available information suggests that the government of Swaziland has invested substantially in improving the country's infrastructure and human capital. Yet the trend of foreign capital flow into Swaziland is as unpredictable as Figure 2 demonstrates. The decline in private investment flows, especially since around mid 1980s, should be seen within the framework of a generally falling foreign capital flow to developing countries in general and Africa in particular. This decline followed the collapse of socialist economies and their integration into the world market system (see The African Development Bank Report 1997:91). Meanwhile, political developments in South Africa also appear to have a strong adverse impact on capital flows to Swaziland. During the time of apartheid and trade sanctions against South Africa, Swaziland was seen by many South African investors as a "safe haven". Since the political transformation in South Africa in early 1990s, this factor has disappeared and according to some observers there has been a partial reversal in the trend of capital flows between the two countries in recent years.

Table 1. Swaziland macroeconomic indicators, 1976-1998 (Base year=1990)

SOURCE: I.M.F. (1998 and 2000); World Bank World Tables (1990 and 1996); the Quarterly Bulletin of the Central Bank of Swaziland (various editions).

Notes: * Excluding gold reserves.

Table 2 presents some basic comparative macroeconomic indicators for Swaziland and some African countries for the 1990-1998 period. For Swaziland, the table shows an average real GDP growth rate of 3.14% per annum in 1990-98 and a per capita GNP of US$1390 in 1998. These figures together with other indicators such as GDI and GDS rates, the inflation rate and government deficit demonstrate that the Swazi economy performed well compared to African averages as well as other countries in the region such as Zimbabwe and Kenya. However, the difference between domestic investment and domestic savings clearly indicates the relatively heavy dependence of Swaziland on foreign resources.

Table 2. Some basic economic indicators in selected African countries, 1990-1998 (Base year=1990)

SOURCE: The African Development Bank 2000.

Note: * The Gross National Savings (GNS) rate in Swaziland for the same period was 23.41%.

The preceding discussion underscores the significance of reliance on and development of the domestic financial system in order to compensate for shortfalls in the availability of foreign capital and ensure a sustainable rate of domestic investment. Besides being relatively well diversified and developed, the financial system of Swaziland operates in a relatively deregulated and liberalized environment (World Bank 1996, ii). At present, this system encompasses four major commercial banks that are believed to be efficient but suffer from lending to a narrow corporate client base. In addition, there is a saving and development bank, a building society and an industrial development company that provide long-term finance. There are also contractual savings institutions, including pension and provident funds, and an insurance company. Finally, the financial system also includes savings and credit cooperatives that cater to the needs of small- and medium-scale borrowers in a manner that is believed to enhance the process of financial development and integration in the country (see Elhiraika 1999).

The financial system is managed by the CBS. There is, however, no independent monetary policy in Swaziland as a result of its membership in CMA. The agreement allows for unrestricted flow of funds between member countries, besides holding the Emalangeni on a par with the Rand, which represents 25% of the currency in circulation in Swaziland. Thus the CBS has little control over money supply, and no control at all over exchange rate policy. Efforts by the CBS have focused on maintaining the banks' surpluses in Swaziland, and not South Africa. Swazi banks often face excess liquidity because of perceived lack of investment opportunities. Until recently, commercial banks were required to hold 95% of their total liabilities in the form of domestic assets³. Meanwhile, interest rate policy has kept real prime lending and deposit rates always below those in South Africa, so as to attract more investors into Swaziland. The lending rate differential was notably high during the 1980s.

This policy is intended to subsidise investors perhaps at the expense of small savers. Yet it may also hurt domestic investment in terms of availability of credit. Businesses and large savers are able to redirect their surpluses to South African banks where real interest rates are higher. Moreover, this policy may have an adverse impact on the nature and efficiency of investment and job creation by encouraging more capital-intensive projects. There is clearly a trade-off between attracting investors and restraining banks from directing their funds to South African markets that offer higher interest rates. This raises a number of questions as to why banks suffer from excess liquidity, and what policy measures could be taken to help them invest their funds locally. However, to minimise the adverse effect of this policy Swaziland has significantly reduced the margin between its lending rates and South African rates in the early 1990s (Xuba 1994, 116).

To sum up, the economy of Swaziland has been relatively stable, highly open and well integrated with other economies in the region. For instance, the exports of Swaziland, mainly to South Africa, amounted to about 70% of GDP in the 1990s. Being highly open with no independent exchange rate and monetary policy the economy seems to be vulnerable to external shocks. Therefore, the promotion of the domestic financial system and increased domestic resource mobilisation is not only important for stimulating domestic investment, but also for promoting economic stability by reducing dependence on foreign resources.

3. Literature review and hypothesis

In the literature, private investment functions are generally modelled on the premise of the neo-classical models, the accelerator model, Tobin's Q-model or the expected profit model (see Serven and Solimano 1993). Ghura and Goodwin (2000) note that because of data limitations, especially for capital stock and appropriate measures of returns on investment, many empirical studies of investment behaviour in developing countries have used variants of the flexible accelerator model. The variables frequently included in private investment models can be grouped as financial and non-financial determinants. Financial variables include the rate of interest, the level of financial development and bank credit. The non-financial variables include real output, government investment, terms of trade, the real exchange rate, inflation, foreign debt, and foreign trade. The remainder of this section discusses the theoretical and empirical effects of these variables.

As explained by Ndikumana (2000, 383), the emphasis of recent investment literature on the role of finance for investment constitutes a major improvement over the traditional view that domestic investment is constrained primarily by domestic savings. The new view considers savings as a mechanism of supply of funds to the capital market that channels them into the investment process. Accordingly, the level of the development of the financial system as well as financial policy are hypothesised to play a more important role in influencing investment demand.

Several indicators of the level of financial development exist in the literature (see Levine and Zervos 1996). Again because of paucity of data in developing countries, many studies focus on two indicators: the ratio of broad money to GDP and bank credit to the private sector. However, as mentioned before, the broad money to GDP ratio indicates the ability of financial institutions to provide liquidity and not necessarily their role to allocate credit efficiently. A high M2-GDP ratio may be a symptom of financial underdevelopment while a low M2-GDP ratio may indicate a high degree of sophistication of financial markets, which allows individuals to economise on their money holdings (Bencivenga and Smith 1991). Moreover, the M2-GDP ratio is often correlated with bank credit, so that only one of them - normally the credit variable - is used in empirical models (see Gregorio and Guiddotti 1995, 438).

Changes in the volume of bank credit are expected to have a positive impact on private investment in developing countries. Many studies suggest that external finance is a more important source of enterprise finance (see Oshikoya 1994) and that because of relatively narrow financial markets credit policy affects private investment directly through the volume of credit available to firms. This implies that the availability of loans might be more important than the cost of borrowing

Since the early 1990s, many developing countries have embraced policies of financial reform and liberalization intended to promote capital markets and ensure positive real rates of interest. Though standard economic theory suggests a negative relationship between investment and the rate of interest, the effect of changes in the real rate of interest on private investment in developing countries may depend on the level of financial sophistication. In a financially repressed economy, rises in the rate of interest may stimulate an increased flow of deposits that enable increased investment to be undertaken (Fry 1998). But, a higher real interest rate may raise the cost of capital and discourage investment. Empirical evidence on the importance of the real interest rate variable is mixed (see Gibson and Tsakalotos 1994).

The accelerator model suggests a positive correlation between private investment and changes in real output because changes in real output signify changes in aggregate demand which investors attempt to satisfy (Ndikumana 2000). Increased output may also mean increased savings that can be used to finance domestic investment. Ample empirical evidence provides support to this hypothesis (e.g. Fielding 1997; Greene and Villanueva 1991). However, evidence also indicates that the income effect on investment is not always strong. For instance, Bleaney (1994) finds real output to have no significant effect on private investment in South Africa during the era of political unrest.

Theoretically the effect of government investment on private investment is ambiguous. Public investment may have a crowding out effect on private investment if the two compete for the same capital source. On the other hand, public expenditure on infrastructure, including power supply, roads and education, is likely to have a robust positive effect on private investment (see Hadjimichael and Ghura 1995). Inflation, as a variable that is closely related to government budget in developing countries, is regarded as an indication of the quality of government policy. High and unpredictable inflation rates are expected to adversely affect domestic investment by increasing the degree of uncertainty about the macroeconomic environment. Several studies confirm this effect empirically (e.g., Oshiokoya 1994; Greene and Villanueva 1991).

Although the relation between foreign capital inflow and domestic savings could be ambiguous, no studies find a negative relation between foreign capital flow and domestic investment. "This means that capital imports must finance some additional growth unless the productivity of capital falls drastically" (Thirlwall 1983, 298). An exception is the finding by Hadjimichael and Ghura (1995) that private investment is adversely influenced by foreign capital inflow. This may be explained on the basis of whether there is a net inflow or net outflow of capital. The latter as opposed to the former is likely to have a negative impact on private investment. The positive correlation between foreign capital inflow and domestic investment is particularly robust if capital inflows are dominated by private funds, which are often tied to specific projects.

Other open economy variables related to the foreign capital flow variable include foreign debt, terms of trade, the real exchange rate, and foreign trade. The literature postulates that increased foreign debt (relative to GDP or exports) would lower private investment demand because the resources used for servicing the debt crowed out public investment. To the extent that public and private investment are complimentary, increased foreign debt reduces private investment (Ghura and Goodwin 2000). More importantly, increased foreign debt is also an indication of debt overhang, which leads to expectations of increased future tax burden and reduces private investment demand. Similarly, deterioration in the terms of trade is envisaged to have a negative impact on private investment because it raises the cost of imports relative to income and the value of exports. Changes in the terms of trade may also worsen the current account position creating increased macroeconomic instability (see Hadjimichael and Ghura 1995).

The effect of real exchange rate changes on private investment is theoretically indeterminate. On the one hand, real exchange rate depreciation increases the cost of imported capital and reduces investment because most of the capital in developing countries is imported. On the other hand, real currency depreciation increases the profitability of the tradable goods sector and may stimulate investment in that sector.

The volume of international trade, as often measured by exports or imports, is an indicator of the degree of openness of the economy. It is hypothesised to have a significant positive impact on domestic investment (see Levine and Renelt 1992). Increased exports signify expansion of markets for domestic products and also provide the foreign exchange needed to import capital goods. However, the effect of imports on domestic investment is theoretically less obvious. It is expected to be negative if imports are dominated by demand for foreign consumer goods, but it would be positive if increased imports imply greater access to foreign capital goods.

Finally, human development variables as indicators of endogenous growth have usually been used in growth models, often yielding positive coefficients (Barro 1991). The few studies that have included human capital variables, such as school enrolment ratios, in investment equations also found positive and significant results (Mlambo and Nell 1995). Investment in human capital enhances technological progress and raises labour productivity. This is expected to stimulate private investment. In the next section I specify the private investment model focussing on the variables that are deemed to be relevant to the economy of Swaziland.

4. The Model

In addition to specification of the investment model, this section examines data sources and estimation procedure. As explained in the previous section, the flexible accelerator model has been the most widely used one (see Serven and Solimano 1993). The accelerator theory assumes a lagged adjustment of capital to its desired level (K^*_t) so that:

Where K_t = stock of private capital in the current period; K_t-1= private capital stock lagged by one period. The desired capital stock is determined by the long-run equilibrium condition that the value of the marginal product of capital has to be equated to the marginal cost of capital. "This condition is transformed such that K* depends on expected profits, which in turn depend on prices, output and other autonomous shocks" (Mlambo and Nell 1995). Thus empirical models of private investment have focused on the price and autonomous shock variables that affect expected profits and thereby the level of private investment. Due to data constraints, investment models for developing countries often ascribe greater importance to macroeconomic variables in explaining investment behaviour.

The specification of the private investment regression equation for Swaziland draws on the approaches adopted in previous studies (e.g.; Bleaney 1994; Shafik 1992; Ndikumana 2000) and the discussion in sections 2 and 3. The following equation gives the general form of the model:

PI_t = f(GDP, FCI, BC, EX, LR, INFL, GI, LR_D, PSER) (2)

Where PI = Private investment expenditure

GDP = Gross domestic product

GI = Government investment expenditure

FCI = Net foreign capital inflow

BC = The flow of bank credit to the private sector;

EX = Domestic currency value of real exports of goods and services

LR = real lending rate of interest

LR_D = real lending rate differential between between Swaziland (SD) and South Africa

(SA), defined as LR_SD-LR_SA

PSER = Primary school enrolment ratio (% of total)

INFL = The annual rate of inflation, defined as change in the consumer price index.

The measurement of variables conforms to standard national income definitions, but some variables require special remarks. Because of lack of data on net investment, private and public investments were measured as gross fixed capital formation in the two sectors, respectively. The total net flow of foreign resources (FCI) is the sum of all official and private flows to the country less repayment of past loans (amortisation) and profit repatriation. FCI is taken as the reverse of the balance of payment current account balance. I have used total bank credit to the private sector rather than long-term loans alone due to lack of a complete detailed series on bank credit. The inclusion of short-term lending is also justifiable since it may be used by businesses to finance working capital.

The effect of interest rates in Swaziland is often related to that of interest rates in South Africa since investors move freely between capital markets in the two countries. The real interest rate differential is meant to test the prediction that a positive differential would attract more investment into Swaziland. But there is no a priori reason for this to occur because as elaborated in section 2 the larger the interest differential the more funds are expected to move to South Africa. The real lending rate is measured⁴ as the difference between the nominal rate and the rate of inflation. The export variable rather than the import variable is included because it is expected to have a more direct effect on private investment. The primary school enrolment ratio is used in this study as an indicator of the level of human development in Swaziland, as dictated by data availability.

The foreign debt variable is not included because Swaziland has always maintained enough foreign reserves to service its debt and this variable is empirically highly correlated with the foreign capital flow variable. The real exchange rate and terms of trade variables are excluded because of the CMA agreement besides the lack of sufficient data on the latter variable.

While real output, foreign capital inflow, export and bank credit are expected to stimulate private investment, government investment, the real lending rate and the interest differential may take either sign. Inflation is hypothesised to have a negative effect on private investment.

Similar to Shafik (1992) and Fielding (1997), the process by which actual investment adjusts to desired investment is postulated to follow an error correction process. Error correction methodology is now widely applied in empirical research and proved to be useful for explaining a variety of long-run macroeconomic relationships. Engle and Granger (1987) show that although using stationary variables in regression equations is desirable, it may disguise low frequency information if some or all of the variables in the model are co-integrated. Most macroeconomic variables, such as income and investment, are normally co-integrated in the sense that they possess a long-run equilibrium relationship.

According to Granger (1986) any system of co-integrated variables can best be represented by an error correction mechanism in which the lagged residuals that are obtained from the underlying co-integrating relationship are added to the original vector of co-integrating stationary variables. The coefficient of the lagged residuals or the error correction mechanism (ECM) represents the process by which the dependent variable adjusts to its long-run equilibrium position. This procedure allows distinction between short-run and long-run relationships, and is therefore proved to be more relevant from policy perspectives.

The separation between short-run dynamics and long-run relationships is particularly important in the context of the present study because of the variety of macroeconomic changes that took place in Swaziland and other countries in the CMA region during the period under review. It is also useful in view of the importance ascribed by the paper to the role of economic policy in influencing future private investment in the country. In the context of investment literature, error correction models are also considered to be more appropriate than the traditional partial adjustment models for the following reasons (see Shafik 1992, 265). The ECM approach implies that firms incur no costs for changes that are planned, and provides a realistic representation of how rational, but fallible, agents make decisions.

Using the co-integration and error correction framework the following equation for private investment will be examined econometrically in the next section:

_ _t (3)

Where Z is a vector of explanatory variables, the term between brackets is the ECM component of the model, _ is the adjustment parameter (_ varies between zero and one), and _ is an error term.

5. Model estimation and results

5.1. Data Sources

Data used in the regressions consists of annual observations for the period 1971-1998. Most of the data is drawn from the International Financial Statistics (IFS) of the IMF, various years. These covered all balance of payments data, monetary data as well as real sector data such as GDP, gross capital formation, prices and nominal deposit and lending rates. After detecting some inconsistencies in the national income accounts as published in the IFS, a consistent set of such accounts, which embodies latest available revisions, was obtained from the Research Department of the CBS⁵. This set of both nominal and constant price data is also the only source of a complete series of gross private and public fixed capital formation by sector. Population, employment and human development data was taken from the World Bank's World Tables (1990 and 1996). All the data was updated for 1996-98 using information gathered from the IMF (2000) and the 1999 annual report of the Central Bank of Swaziland.

5.2. Stationarity and Co-integration Tests

Before estimating the model in equation 3, I have first tested for non-stationarity in all the variables using the Augmented Dickey-Fuller (ADF) test. This test helps us to ascertain the order of integration and the degree of differencing needed to make each time series stationary. A set of co-integrated variables was then selected using the Johansen (1988) multivariate technique. All real variables are expressed in constant 1990 price and natural logarithmic form, except interest rates.

The test statistics were applied to level as well as first differences, with the maximum lag of 2. The results in table 3 signify that most of the time series are non-stationary in logarithmic levels but have simple unit roots, i.e., I(1), so that only first differencing is required. These include private investment expenditure, gross domestic product, foreign capital inflow, the inflation rate, exports, public investment expenditure, commercial bank lending to the private sector, the primary school enrolment ratio, and the prime lending rates in Swaziland and South Africa. Accordingly, the difference between Swazi and South African lending rates is stationary without further differencing. This provides evidence of financial integration between the two countries in the sense that real interest rate movements in the two countries are closely related.

The Johansen procedure estimates a vector of autoregression (VAR) system and examines the possibility that there exist more than one co-integrating vectors in a multivariate model. There is ample evidence that the Johansen procedure is more robust than the alternative procedures including the two-step test of Engle and Granger (1987). However, as Carruth and Sanchez-Fung (2000) explain, even in the reduced form single equation model, the number of significant co-integrating vectors can be sensitive to the lag structure of the model. But there is often one vector that is consistent with the data.

Table 3: Unit Root Test (Estimation period 1971-1998)

Notes: * and ** denotes that the test is significant at the 55 and 1% levels, respectively. Lags varied between 1 and 2.

^(a) Difference between South African and Swaziland prime lending rates (LR_SA - LR).

^(b) Change in the logarithm of CPI. All the estimation uses PC-Give Professional version 9.2 (see Hendry and Doornik 1999).

The results from the Johansen co-integration test (both the eigenvalue and trace tests) are presented in table 4. All the variables included for the test have the same order of integration. Level stationary variables such as the interest rate differential between South Africa and Swaziland (LR_D) are excluded. The export (EX) variable was also excluded after the test was applied because it was found to be highly correlated with GDP. In addition, the primary school enrolment ratio, which is strongly trended, caused serious auto-correlation problems and was therefore dropped from the regression equation.

The maximum eigenvalue test has consistently indicated the existence of one significant co-integrating vector, but the trace test suggests the presence of more than one co-integrating vectors in equations 2 and 3 in the table. The two tests together signify that there is only significant co-integrating vector in all cases. For the sake of being brief, the alpha and beta coefficients of the Johansen test are not presented here. By normalising the co-integrating vectors on real private investment, I obtained the results shown in table 5. These results are discussed in the next sub-section.

Table 4. Multivariate co-integration tests

Notes: Max Eigen. = Maximum Eigenvalue.

* and ** denote significance at 5% and 1%, respectively. This implies rejection of the null hypothesis. Beta and alpha vectors are not shown for space purposes.

5.3 Discussion of Results

The results in table 5 suggest that the relationship between private investment and the explanatory variables is generally quite significant in terms of the t-values. The only notable exception is that the level of national income appears to have no strong relationship with private investment. This finding confirms earlier results by Bleaney (1994) who finds no strong relationship between income and private investment in South Africa. The weak relationship between private investment and aggregate income may be explained in terms of the motives behind foreign capital inflow to Swaziland, which financed a considerable part of private investment in the period under review. As mentioned previously, Swaziland was considered as a gateway for South African businesses during the time of international economic sanctions and trade embargo before the end of the apartheid era in 1990. For these businesses external markets were probably more important than domestic markets as represented by the output variable. The importance of foreign resources is confirmed by the positive and highly significant coefficient for the foreign capital flow variable. But the size of this coefficient is small implying that a 1-point increase in FCI raises private investment by 0.011 point.

Table 5. Normalised co-integrating vectors for investment

Notes: * Denotes significance at the 5% level. Figures in parentheses are the t-statistics.

^(a) The econometric package used does not give the standard errors associated with the constant term.

From policy perspective, the results suggest that the bank credit variable is the most potent factor stimulating private investment in Swaziland. The coefficient of this variable ranged between 0.7 and 1.5 and has always been statistically robust. It is therefore arguable that policy-makers in Swaziland can capitalize on the credit variable to influence private investment directly or indirectly. As Clarke (1996) and others argue the importance of the bank credit variable implies that there was some rationing in the credit market, and the relaxation of the credit constraint may boost private investment in the country.

Contrary to expectations by the financial repression hypothesis, the real lending rate has a negative and significant, but small, effect on private investment. This signifies the importance of the cost of capital for businesses. In view of the discussion in section 3 this finding is not surprising at all. The real lending, as opposed to the real deposit rate, has since 1981 been positive and generally increasing. This generated a high margin between lending and deposit rates. A high interest rate margin can boost bank profit at the expense of both savers and borrowers. The huge interest rate margin in Swaziland points out to the importance of policies to encourage competition among banks so that the margin is narrowed, and banking efficiency improved.

The government investment variable turned out with a significantly negative parameter, suggesting a crowding out effect on private investment. This result is theoretically plausible but rather surprising in the case of Swaziland, where public investment has been concentrated largely on infrastructure, education and other projects that are supposed to stimulate private investment. In general, crowding out occurs if the private and public sectors compete for the same investment opportunities and/or the same pool of funds. The literature also suggests that observed negative correlation between private and public investment might be an indication of counter-cyclical responses of government policy. Like, government investment, the inflation rate has a significant adverse effect on private investment. Accordingly, macroeconomic policy ought to give paramount attention to the control of inflation.

The econometric estimates of the dynamic error correction model are displayed in table 6. The residuals from level regressions (ECM_t-1) are included in a lag form in order to capture the process of adjustment. In general, the econometric results are significant and the diagnostic tests are good. The speed of adjustment implied by the ECM is between -0.4 and -0.47 per annum. The coefficients of the explanatory variables in the dynamic model are consistent with the long-run level regression in that, with the exception of the lending rate, all the independent variables retained their signs. Moreover, the statistical significance of the bank credit variable, government investment and inflation remained intact.

As shown in the table 6, changes in real output have a significant and large effect on short-term changes in private investment. This finding is theoretically consistent and in line with many empirical findings for developing countries by, for instance, Shafik (1992) for Egypt and Ghura and Goodwin (2000) for a group of developing countries. In terms of magnitude, GDP was followed by bank credit to the private sector, which turned out with a positive and highly significant coefficient. In the short-run, a 1% increase in bank credit to the private sector implies an increase in private investment by almost the same percentage.

The foreign capital flow and real lending rate variables seem to have no important effect on private investment in the short-run. Obviously, the foreign capital variable is not a policy variable. As the discussion in section 2 indicated, in addition to external factors, changes in foreign capital flows are probably related to structural and long-term developments rather than short-term fluctuations. This lends strong support to my argument regarding the role of domestic resource mobilisation in achieving sustainable economic development.

Increases in government investment are again found to have a strong, but small, adverse effect on private investment. Yet, it is still difficult to conclude that a crowding out effect does exist in view of the hypothesis of a counter-cyclical relationship between private and public investment. For example, the government may decide to invest in the transportation sector because of deficient private investment in that sector. The coefficient of the inflation variable confirms that inflation is detrimental to private investment and that the short-run effect of inflation is larger than the long-run effect. It may be the case that inflation news prompts investors to wait in the short-run. But, in the long-run, the effects of inflation are more predictable, and hence its effects are smaller especially if inflation rates are moderate.

Table 6. Dynamic equations for investment^(a) (difference regressions)

Notes: ^(a) The dependent variable is the difference of the logarithm of private investment. Figures in parentheses are the t-statistics. * and ** denotes significance at the 10% and 5% levels, respectively.

In summary, the regression results indicate that the variables used to explain private investment in Swaziland are generally significant either in the short-run or the long-run or both. In the long-run, foreign capital inflow and bank credit to the private sector have a strong stimulating influence on private investment, whereas government investment, the lending rate and the rate of inflation have strong but adverse effects. Aggregate income is the only variable that seems to have no significant effect in the long-run. The direction of the impact of the independent variables on private investment is always consistent. However, foreign capital inflow and the real lending rate showed no important impact in the short-run, while aggregate income was found to be the most powerful factor driving short-run changes in private investment, followed by the bank credit variable. The policy implications of the results are discussed in the next section.

6. Conclusion and policy recommendations

The objective of this paper was to investigate the role of the domestic financial system in stimulating domestic private investment in view of the declining and highly fluctuating foreign capital flows to Swaziland. The article also considered the role of economic policy relating to private investment, taking into consideration the constraints and opportunities arising from Swaziland's membership in the Common Monetary Agreement (CMA). Besides Swaziland the agreement includes South Africa, Namibia and Lesotho.

After analysing the macroeconomic environment and economic performance and policy, the paper specified and estimated an error correction model. The analysis demonstrates that the economy of Swaziland has been relatively stable, highly open and well integrated with other economies in the region. But, the data confirms that the country is heavily dependent on foreign resources compared to other countries in the region. This dependence together with the lack of independent exchange rate and monetary policy makes the economy vulnerable to external shocks. Consequently, this study argues that increased mobilisation of domestic resources is paramount for achieving a desired and sustainable rate of economic growth as well as for enhancing economic stability.

Econometric results suggest that bank credits to the private sector have a positive and significant impact on private investment in both the short and the long-run. Real GDP, as opposed, to foreign capital inflow, has no important effect in the long-run, but seems to have the greatest effect in the short-run. Government investment, inflation and the real lending were found to have significant adverse effects on private investment, but the lending rate is not important in the short-run. According to these results, policies relating to bank credit, GDP, government investment, the lending rate and inflation are expected to influence private investment decisions.

The large and significant bank credit's coefficient suggests the presence of a credit constraint, the relaxation of which may boost private investment. This finding lends support to a recent call by the Swaziland National Business Council⁶ (SNBC) that modalities should be found so that the excess funds resulting from high levels of liquidity by the country's commercial banks be mobilised in order to strengthen the growth of local enterprises.

The SNBC notes that private sector lending has remained the same in the last two years and that there was virtually no source of funds for emerging indigenous enterprises. Indeed although time and saving deposits account for about 70% of total bank deposits, there seems to be a high demand for loans in the country that is not satisfied by banks as indicated by the rapid expansion of savings and credit co-operative societies. Unfortunately, the SNBC provides no suggestions on how increased credit availability could be achieved. Due to the limited room for independent monetary policy in Swaziland, fiscal and other, policy measures may be used to promote the growth of both domestic savings and investment. These measures may include non-credit support to emerging as well as existing enterprises.

Assuming that balance of payments equilibrium has to prevail in the long-run and considering the implications of the CMA for capital mobility in the region, Swaziland may have to promote efficient and competitive investments through competitive, rather than subsidised, real interest rates. It was argued that the policy of lower deposit rates in Swaziland compared to South Africa might indeed depress domestic investment by lowering the availability of funds in the country and encouraging financial investment by Swazi banks and large individual savers in neighbouring South Africa.

This paper does not answer a number of important questions such as why banks in Swaziland consistently suffer from excess liquidity, and how to encourage increased lending to emerging indigenous entrepreneurs. These questions can be taken up in further research.

References

Barro, R. J. 1991. Economic growth in a cross-section of countries, Quarterly Journal of Economics 104, no. 2: 407-433.

Bencivenga, V. R., and B. Smith. 1991. Financial intermediation and endogenous growth. Review of Economic Studies 58, no. 2: 403-44.

Bleaney, M. F. 1994. Political uncertainty and private investment in South Africa. The South African Journal of Economics 62, no. 3: 188-197.

Carruth, A., and J. Sanchez-Fung. 2000. Money demand in the Dominican Republic. Applied Economics, no. 32:1439-1449.

Clarke, R. 1996. Equilibrium interest rates and financial liberalisation in developing countries, The Journal of Development Studies, no. 32: 391-413.

Economic Association of Swaziland (ECAS). 1995. A submission to the National Economic Review Commission: Executive summary. Economic Digest 6, no. 1: 1-13.

Elhiraika, A. B. 1999. The growth and potential of savings and credit co-operative societies in Swaziland. Development Policy Review 17, no. 4: 293-313.

Engle, R. F., and C. Granger. 1987. Cointegration and error correction: Representation, estimation and testing. Econometrica 55, no. 2: 251-276.

Fielding, D. 1997. Investment in Cameroon 1978-1988. Journal of African Economies 4, no. 1: 29-51.

Fry, M. 1988. Money, interest and banking in economic development. John Hopkins University Press.

Gibson, H., and E. Tsakalotos. 1994. The scope and limits of financial liberalisation in developing countries: A critical survey. Journal of Development Studies 30, no. 3: 578-628.

Granger, C. 1986. Developments in the study of cointegrated economic variables. Oxford Bulletin of Economics and Statistics 48, no. 3: 213-228.

Greenwood, J., and B. Jovanovic. 1990. Financial development, growth and the distribution of income, Journal of Political Economy 98, no. 5: 1076-1108.

Gregorio, J., and P. Guidotti. 1995. Financial development and economic growth. World Development 23, no. 3: 433-448.

Greene, J., and D. Villanueva. 1991. Private investment in developing countries. IMF Staff Papers, no. 38: 33-58.

Ghura, D., and B. Goodwin. 2000. Determinants of private investment: A cross-regional empirical investigation. Applied Economics no. 32: 1819-1829.

Hadjimichael, M., and D. Ghura. 1995. Public policies and private saving and investment in sub-Saharan Africa: An empirical investigation. IMF Working Paper 95/19.

Hendry, D. F. and J. Doornik. 1999. Empirical econometric modelling using Pc-GIVE. Timberlake Consultants, UK.

International Monetary Fund (IMF). 2000. Swaziland: Selected issues and statistical appendix. IMF Staff Report, no. 00/113, Washington, D. C., USA.

Johansen, S. 1988. Statistical analysis of cointegrating vectors. Journal of Economic Dynamic and Control 12, no. 6: 231-54.

Levine, R., and D. Renelt 1992. A sensitivity analysis of cross-country growth regressions. American Economic Review 82, no. 4: 942-963.

Levine, R. and S. Zervos. 1996. Stock market development and lLong-run growth. The World Bank Economic Review 10, no. 2: 323-339.

Mlambo, K., and K. Nell. 1995. Public policy and private investment in South Africa: An empirical investigation. A paper presented at the Conference on Transitional and Long-term Development Issues in South Africa, Johannesburg, December.

Ndikumana, L. 2000. Financial determinants of domestic investment in Sub-Saharan Africa: Evidence from panel data. World Development 28, no. 2: 381-400.

Oshikoya, T. W. 1994. Macroeconomic determinants of domestic private investment in Africa: An empirical analysis. Economic Development and Cultural Change, no. 42: 573-596.

Serven, L., and A. Solimano. 1993. Debt crisis, adjustment policies and capital formation in developing countries: Where do we stand? World Development 21, no. 1: 127-140.

Shafik, N. 1992. Modelling private investment in Egypt. Journal of Development Economics, no. 39:263-277.

World Bank. 1996. Swaziland financial sector study. Report no. 14985. SWA. The World Bank, Washington, D. C.

Xuba, P. M. 1994. Impact of foreign aid in Swaziland. Economic Digest 5, no. 4: 90-120.

Notes

* Department of Economics, CBE, United Arab Emirates University P.O. Box 17555, Al-Ain, United Arab Emirate. E-mail AELHIRAIKA@UAEU, AC.AE

1 Information gathered from CBS shows that during periods of low foreign capital inflow, private investment relied on substantial reinvestment of surplus generated by foreign direct capital investment in the country. Between 1981 and 1995, gross capital outflow was 5.4% of GDP, gross capital inflow was 17.7% of GDP, net long-term capital inflow was 14.67% of GDP, and total short-term and long-term reinvestment by foreign owned businesses was 19.3% of GDP.

2 Central Bank Quarterly Bulletin 1997,19. Mbabane, Swaziland.

3 According to the World Bank (1996), this measure is hardly effective because banks can continue to invest in South Africa and redirect their funds back to Swaziland at the last day of the months. Accordingly, the policy was abandoned in 1997.

4 Following Bleaney (1994, 196) and others real interest rate LR was alternatively calculated as LR=(1+i)/(1+INFL), where i is the nominal rate. But the two measures produced substantially similar results.

5 The CBS is the prime source of Swazi economic data published in the IFS.

6 The Swazi Observer, 9 September 1997, p.5