The Volatility of Capital Flows in South Africa: Some Empirical Observations

The Volatility of Capital Flows in South Africa: Some Empirical Observations By Michael Nowak African Department International Monetary Fund Prepared for BER Conference, Johannesburg, June 1, 2001

The Volatility of Capital Flows in South Africa:

Some Empirical Observations

By

Michael Nowak African Department International Monetary Fund

Prepared for BER Conference, Johannesburg, June 1, 2001

The views expressed in this paper are those of the author and do not necessarily represent those of the International Monetary Fund Research assistance was provided by Ms Pamela Mjandana and Mr Nehrunaman Pillay

The Volatility of Capital Flows in South Africa: Some Empirical Observations

I SUMMARY AND CONCLUSIONS

In the past several years, South Africa has taken a number of steps, such as budget-deficit reduction and adoption of a more flexible exchange rate regime, that have helped reduce its vulnerability to adverse external shocks In addition, the SARB has made considerable strides

in reducing the net open forward position (NOFP), which has represented a major source of external vulnerability, has been lowered significantly The NOFP, however, remains

relatively large As such, it continues to represent a source of concern to investors that is reflected in South Africa’s sovereign risk spreads The SARB has, therefore, expressed its commitment to achieving further reductions

This note examines alternative options available to the SARB for reducing the NOFP On the basis of some preliminary statistical findings, it argues that:

?? Regular pre-announced foreign exchange purchases may help bring about up-front

cuts in borrowing spreads, but could place undue pressure on the rand, the external current account, and the real economy

?? Care should be taken in targeting capital inflows, such as FDI, that may be perceived

as stable; while such flows may indeed be less volatile than other capital flows, they still tend to show little persistence over time However, in the case of certain large one-off inflows of FDI, such as privatization proceeds, there may be a strong basis to suppose that the inflows will not be reversed It makes sense, therefore, to purchase some or most of these inflows

?? The evidence tentatively suggests that incoming FDI may be partially offset by

long-term capital outflows, possibly reflecting cover operations by investors However, when the picture is extended to include all other capital flows, there is no significant offset

?? Additional external borrowing may be appropriate in modest amounts

II E XTERNAL VULNERABILITY AND BORROWING SPREADS

Over the course of the past several years, macroeconomic policies in South Africa have helped reduce the vulnerability of the economy to adverse external shocks and contagion from other emerging markets The public finances have been brought firmly under control,

an inflation-targeting regime is now in place, and the exchange rate has been allowed to float with a significant gain in external competitiveness The banking system is strong and healthy

In addition, South Africa’s medium- and long-term external debt remains low in comparison with other emerging market economies (Table 1) And the SARB’s net open forward position

(NOFP), which is a measure of its short-term foreign currency exposure, has been reduced by nearly two-thirds since the currency crisis of 1998

Nevertheless, the NOFP and other indicators of international reserve adequacy remain a source of concern for investors that is reflected in South Africa’s sovereign risk spreads (Figure 1) The relationship between the NOFP and borrowing spreads has been quantified

by Jonsson (2001),1 who found that a reduction of US$1 billion in the NOFP was typically associated with a decline in spreads of nearly 15 basis points.2 This means that the cut in the NOFP from its recent peak of US$23 billion (18 percent of GDP) in September 1998 to US$9 billion (7 percent of GDP) at end-April 2001 would have helped lower spreads by over

200 basis points Moreover, if the NOFP were to be reduced to zero, spreads could fall a further 135 basis points Narrower spreads should translate into lower long-term interest rates

III REDUCING THE NOFP: THE OPTIONS

The benefits of lowering the SARB’s short-term foreign currency exposure are not seriously contested What is perhaps more debatable is the best means by which the SARB should acquire the foreign exchange needed to reduce the NOFP At least three options are

available:

?? To make regular purchases of dollars in the foreign exchange market according to a

predetermined schedule

?? To acquire dollars in the foreign exchange market from capital inflows that are

considered relatively stable and persistent (so-called “cool” money)

?? To use the proceeds from additional medium-or long-term external borrowing

Option 1: Regular purchases

The SARB could buy dollars in the spot market on a regular periodic basis in predetermined amounts until the NOFP had been reduced to an acceptable level This would have the

benefit of establishing a clear timetable for NOFP cuts and help allay concerns that the

process of NOFP reduction might be open-ended Moreover, if the profile of purchases were

1

Jonsson (2001), “The Risk Premium in South African Long-Term Interest Rates,” IMF, mimeo.”

2

This study identified other important determinants of South Africa’s risk spreads These included external borrowing (spreads are raised), reductions in the budget deficit (spreads are lowered), contagion from other emerging markets, as captured by the Emerging Bond Market Index (spreads are raised, but impact is small)

to be announced ahead of time, and if the exercise carried with it sufficient credibility, the favorable impact on spreads and interest rates would likely take place up-front

Despite its appeal, there are downside risks with such an approach By stepping into the market to purchase foreign exchange, irrespective of prevailing market conditions, the SARB would be adding to pressure on the rand at times of weak investor sentiment On such

occasions, the SARB may need to attract capital into the country by offering improved rates

of return This could be done either by forcing a step exchange rate depreciation (and thereby generating the expectation of a strengthening of the rand) or by raising interest rates Under either of these options, it would probably be both difficult and undesirable for the SARB to maintain the higher rates of return for any sustained period of time To do so could introduce excessive exchange rate and/or interest rate volatility and, in the process, undermine the SARB’s inflation-targeting strategy or hurt growth performance Eventually, the return on capital would need to decline, and the capital inflows would be reversed The external current account would, therefore, need to provide the necessary foreign exchange flows through a large real currency depreciation, and this could place considerable strain on the real

economy

In short, to expect the current account to improve by the equivalent of up to 7 percent of GDP, even over a few years, is neither realistic nor advisable Moreover, macroeconomic policy would be “held hostage” to the objective of reducing the NOFP Not surprisingly, the SARB has chosen not to pursue this option

Option 2: Identifying “cool” capital inflows

Rather than stepping into the market on a regular basis, one option might be for the SARB to wait until South Africa enjoyed capital inflows and to make an assessment as to whether these inflows were likely to stay This approach would enable the NOFP to be lowered without subjecting the rand to undue downward pressure or having to raise interest rates keeping macroeconomic policy predictable

The question that arises is what information would be useful to the SARB to help it make a decision about when, how much, and what type of capital inflow to buy Specifically, how can the central bank make use of prior knowledge relating to the volatility and persistence of capital flows? And how would the interaction between different types of capital flows affect the decision to buy foreign exchange?

In principle, foreign direct investment (FDI) is commonly perceived as less volatile and less likely to be reversed than so-called “hot money,” such as short-term capital It has been argued that, by their nature, FDI flows are driven by long-term sentiment and are more likely

to be “stable” than short-term flows Moreover, to the extent that FDI entails physical

investment in plant and equipment, it is “illiquid” and, therefore, relatively difficult to

reverse

However, the international evidence on whether FDI flows tend to be “cool” is mixed On the one hand, in a study of industrial and developing economies (which did not include South

Africa), Claessens, et al (1995) concluded, that for most of the countries reviewed, FDI was

as volatile as other types of flows and that it was generally neither persistent nor predictable.3

On the other, an IMF study argued that, during the recent Asia and Mexico currency crises, FDI was relatively more stable, and less likely to be reversed, than other types of flows.4 Let

us focus on these points

Volatility

The table below presents standard deviations (as a measure of volatility) for different

components of South Africa’s capital account for the period since 1994 when capital controls

on nonresidents were liberalized.5 The standard deviations are lowest for FDI and long-term capital (which consists primarily of bank loans), suggesting that these flows are indeed less volatile than other capital transactions Surprisingly, perhaps, equity flows are markedly

more volatile than FDI even though the distinction between the two is somewhat arbitrary. 6

South Africa: Means and Standard Deviations for Capital Flows, 1994 Q1–2000 Q3

FDI

Equity

Debt

Long-term

Short-term

-70.49 467.51 443.83 -111.40 107.27

499.23 789.42 1,037.78 356.06 897.90

Persistence

When considering buying inflows of capital, the central bank may wish to know the

likelihood of that type of inflow being repeated over a given period of time To test for

persistence, we examined the autocorrelation properties of the major components of South Africa’s capital account (Figure 3) The tests indicate that none of the components, including

3

Claessens, Dooley, and Warner (1995), “Portfolio Capital Flows: Hot or Cold?” World Bank Economic Review, Vol 9 (January)

4

IMF, 1998, “International Capital Markets: Developments, Prospects, and Key Policy

Issues,” Washington, D.C

5

The data are quarterly and available from the IMF’s Balance of Payments Yearbook

6

FDI covers an investment that involves an “effective voice in management,” usually

considered to be 10 percent or more of voting power in local enterprises

FDI, exhibit any persistence Each type is essentially “white noise.” In other words, capital

that comes into South Africa in one quarter is just as likely to be followed in the next quarter

by an outflow as it is by another inflow This is not a reason to avoid targeting capital inflows (evidence of negative autocorrelation, however, probably would be) Rather, it suggests that,

as a general rule, the SARB would need to exercise considerable caution when buying

inflows to reduce the NOFP

Nonetheless, in certain cases involving large one-off inflows of FDI, the SARB may have

sufficient prior information to make an assessment as to whether the inflow is unlikely to be

reversed This would probably be the case, for example, with privatization proceeds or

inflows associated with the restructuring of De Beers’ shareholdings

Interaction with other flows

Even if the SARB knew that a particular inflow was not going to be reversed, it would be

important to know whether or not this inflow was likely to give rise to a corresponding

outflow elsewhere in the capital account, or even the current account In the case of FDI, this could happen if foreign investors wished to avoid taking an open position in South Africa;

they might decide to take cover by borrowing locally and buying foreign assets or repaying

foreign loans.7

This is a testable proposition that appears to be partially correct The table below reports the correlation coefficients between the various components of South Africa’s capital account It suggests that long-term capital, which consists primarily of bank loans, is the only form of

capital that is negatively correlated to FDI in any significant degree (the partial correlation

coefficient is about 40 percent) By itself, this should provide the SARB with some pause for thought before buying FDI inflows

South Africa: Correlation Matrix for Capital Flows, 1994 Q1–2000 Q3

FDI Equity Debt Long-Term Short-Term All excluding FDI FDI

Equity

Debt

Long-term

Short-term

All excl FDI

1.000

-0.147

0.347

-0.398**

-0.054

0.061

1.000 -0.037 0.159 -0.280 0.503**

1.000 -0.356 -0.527*

0.465**

1.000 -0.103 -0.262

1.000 0.106 1.000

* Significant at 5 percent level

** Significant at 10 percent level

7

Alternatively or the recipient of the inflow may decide to reinvest the funds overseas,

exchange control regulations permitting

However, the findings also indicate that all other components of the capital account, when taken together, do not offset FDI This point should not be overstated since there is no obvious explanation as to why FDI and debt security (i.e., fixed income assets) flows, but not other types of capital, are positively correlated Further research using more sophisticated statistical techniques could shed further light on these relationships

It is also possible that FDI inflows are offset by purchases of imports This may be the case, for example, when FDI takes the form of physical investment The data do indeed indicate that FDI and the current account are negatively correlated; the correlation coefficient is, however, relatively small (just over 10 percent).8

Option 3: External borrowing

Empirical work conducted in the IMF indicates that medium-and long-term external

borrowing by the public sector tends to raise spreads, but that the impact diminishes

significantly as the maturity of the debt increases.9 The benefit of external borrowing to reduce the NOFP stems primarily from an improvement in the maturity structure of South Africa’s total foreign currency exposure It does not lead to any overall reduction in exposure and should, therefore, be undertaken in modest amounts Borrowing in the longer maturity ranges would have the most beneficial net impact on spreads

8

The partial correlation coefficient is essentially the same for both quarterly data (1994: Q1–2001: Q3) and for annual data (1985–2000)

9

Jonsson (2001), op cit