A. Introduction

1. Real estate prices in South Africa have risen strongly in recent years, growing by 18 percent per year since mid-1999. Empirical evidence for a group of industrial countries suggests that the probability of a housing price boom being followed by a collapse in housing prices is about 40 percent (IMF, 2003). Many of these house price corrections have been associated with a decline in economic activity or a period of financial instability. This raises a number of policy questions for South Africa. In particular, is the real estate market exhibiting signs of a speculative bubble? Would an abrupt property price correction pose a risk for the banking sector or lead to an economic slowdown? What can and should authorities do to minimize these risks? This paper addresses these questions by comparing the experience of industrial countries with asset price booms to developments in South Africa.

2. The paper is structured as follows: After presenting some stylized facts on property price developments in South Africa, the paper reviews briefly the link between asset price developments and economic activity. Then it looks at the ability of policymakers to identify asset price booms and to distinguish asset price boom periods that are in line with fundamental developments from periods that are commonly referred to as a “bubble”. In the next section, the paper investigates whether monetary policy should react to an increase in asset prices during an asset price boom period, and what practical issues policymakers face.

3. To preview the main findings, results suggest that there is no clear evidence that property is significantly overpriced in South Africa. Banks appear to be fairly resilient to potentially adverse developments in the real estate market. Developments of selected macroeconomic indicators in South Africa, however, appear to share some similarities with asset price boom periods in industrial countries that were followed by a period of weak economic growth or financial instability. As in industrial countries, a close monitoring of real estate developments is clearly warranted. Furthermore, it is suggested that additional analyses of ways to broaden and deepen the market for mortgages be made. Most notably, an increase in the variety of mortgage contracts, including the availability of longer-term fixed-rate contracts, could make the market more resilient to sudden shifts in interest rates. The analysis also highlights the desirability of broadening data availability.

B. South Africa: Some Stylized Facts

4. As is also common in some industrial countries, data on property price developments in South Africa are somewhat limited. Based on available data from Absa, the leading mortgage lender, the following stylized facts emerge (Absa 2004a-d) (Figure II.1):²

South Africa: Property Price Developments

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Absa, IMF.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

South Africa: Property Price Developments

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Absa, IMF.

• Download Figure
• Download figure as PowerPoint slide

South Africa: Property Price Developments

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Absa, IMF.

• Download Figure
• Download figure as PowerPoint slide

• Since 1980, house price developments appear to be in their second cycle. Following the house price boom in the early 1980s with its peak in 1984, house prices fell by over 40 percent in real terms within three years. During the following ten years, real house prices were more stable, though still on a gentle declining trend.

• Real house prices bottomed out in 1997. Over the last 5 years they have increased by some 11 percent per year in real terms. But, real house prices are still below their peak of 1984.

• As in industrial countries, house prices have fluctuated substantially over time. The average standard deviation of real annual house price changes is 9 percent, slightly higher than the average of almost 7 percent in industrial countries (IMF, 2004).

• On a regional level, house prices have exhibited some differences. Since 1998 nominal house prices more than doubled in Gauteng and almost doubled in KwaZulu-Natal; the lowest increase was recorded in Limpopo.

• Prices for existing houses tend to be lower than prices for new houses. This may be seen as an indication that existing houses are below their replacement value or that there is growing demand for new and, possibly, higher-quality housing.³

• The affordability of housing has declined. The ratio of house prices to gross household income increased to 6.1 in 2003, from 4.6 in 2000.

5. A number of economic factors have contributed to the strong rise in house prices. Monetary and credit policy was eased in 2003 and interest rates were lowered. Mortgage rates fell in line with official short-term interest rates and, as a result, the commercial banks’ variable mortgage rate fell to 11.5 percent in the first quarter of 2004 from over 23 percent at the end of 1998.⁴ Declining interest rates have provided a stimulus to the market. Demand has been reinforced by increased demand from an emerging middle class and foreign interest in coastal areas property. The magnitude of the overall effect of recent tax changes, such as reduction in the transfer duty for the third consecutive year in the 2004/05 budget, is more difficult to predict.⁵

6. Some concerns over price stability in the housing market and adverse economic consequences derive from three sources. First, as housing affordability has declined, some households may have become more vulnerable to economic shocks. If the value of mortgage debt and debt service relative to income increased, homeowners could be forced to sell (or driven in default), thereby increasing the supply of housing and driving prices down. Second, the possibility of rising interest rates as the economic recovery progresses and inflation pressures build, could affect household’s debt-service capacity and dampen housing demand. In particular, adjustable rate mortgages make households vulnerable to interest rate shocks. Third, the rise in property prices occurred against the background of some easing in monetary policy and strong domestic credit growth. During 2003, credit extended to the private sector increased by some 12 percent, if the investment category, which was distorted by regulatory and accounting changes, is excluded. Moreover, mortgage advances increased by 16.5 percent in 2003. As will be shown in this paper, in industrial countries, this constellation of macroeconomic factors has been associated with an increase in the probability of a house price correction.

C. Transmission Mechanisms

7. Asset price fluctuations may affect the real economy through a variety of channels. These include their effect on consumption via a wealth change, investment, and banks’ balance-sheets (Mishkin, 2001).

8. Changes in asset prices affect household consumption spending through a wealth effect, primarily in the form of stock market and real estate valuations. Recent studies for the United States find that a $1 decrease in stock market wealth leads to a reduction of spending of 4-7 cents; for other industrialized countries, the impact is somewhat less (e.g., IMF, 2002). The main effect is short-lived and dissipates over 1-3 years. Empirical analyses also suggest that the impact of a significant fall in real estate prices may be more important than an equivalent decline in stock prices (Case, Quigley, and Shiller, 2001), though this finding is not unchallenged (Ludwig and Slok, 2002). Available evidence for a group of emerging markets also suggests that there is a small, but statistically significant, relationship between stock market developments and private consumption. Over a three-year period, a 10 percent decline (increase) in stock prices is associated, on average, with a 0.2-0.4 percent decrease (increase) in private consumption (Funke, 2004). Preliminary estimates for South Africa also point to the existence of small stock market and real estate wealth effects. However, as in the case of industrial countries, the results are very sensitive to the choice of the observation period (see Box 1).

9. Large asset price swings also affect investment.⁶ With rising stock prices the market value of firms increases, relative to its replacement cost, and thus makes the financing of investment projects relatively cheaper. In contrast, in the case of a significant decline in stock prices, companies find it more difficult to raise equity and can only raise less money relative to the cost of equipment, thus investment may decline. Comparable mechanisms are in place in real estate markets. A higher price of housing, relative to construction costs stimulates housing construction because of higher profitability. The reverse is true for falling house prices. A sudden decline in property prices renders investment less attractive and reduces the profitability of the investment. As a result, investment may dry up and contribute to an economic slowdown.

10. A weakening of asset prices may have adverse effects on financial stability. Changes in asset prices may be transmitted to the real side of the economy through banks’ balance sheet effects. An increase in stock market or real estate wealth increases the available collateral and strengthens the financial position of borrowers. The higher level of available collateral tends to reduce the agency costs of the lender, diminishes borrowers’ incentives to engage in moral hazard, and leads to a reduction in the external finance premium (Wagner and Berger, 2002). Lenders’ willingness to supply credit increases and hence investment and consumer durable expenditure may increase. A decline in net worth, as a result for example of falling equity or property prices, may set into motion opposite effects.

D. Identification of Asset Price Boom Periods

11. An appropriate policy response to an emerging asset price boom depends on at least two factors. Policymakers must be able to identify asset price booms and, ideally, be able to distinguish between potentially harmful asset price booms (“high-cost” booms) and asset price booms that are likely not to result in an asset price bust and economic or financial instability (“low-cost” booms).⁷

12. There is no well-established definition of what an asset price boom is; and booms are sometimes associated with bubbles. From a theoretical perspective, a bubble refers to a situation where the price of an asset exceeds its fundamental value by a wide margin. However, making such assessment is fraught with complications. As a result, most empirical analyses follow a more pragmatic approach and focus on large and persistent increases in asset prices, using business cycle techniques (e.g. IMF, 2003) or large deviations from trend, to identify boom periods. In the latter case, trend measures have been derived on the basis of past growth rates or with the help of a Hodrick-Prescott filter (e.g., Bordo and Jeanne, 2002, and Detken and Smets, 2004)

13. The following empirical analysis focuses on the linkages between asset price booms and macroeconomic and financial developments in industrial countries in order to derive some lessons for South Africa. The analysis uses event-study methodology to compare the development of key macroeconomic indicators before, during, and after asset price boom periods. The analysis associates an asset price boom with a period where real asset prices are 10 percent above their annual trend value, based on a Hodrick-Prescott filter.⁸ For a group of 18 countries, we identify some 40 asset price boom periods during the 1970-2002 period.⁹ Asset price booms refer to aggregate asset prices, which are the weighted sum of property price and equity price developments. Using this methodology for South Africa suggests that the recent increase in property prices may indeed be classified as a property price boom, particularly since 2001. However, South Africa has not experienced a stock price boom. And, when equal weights are given to stock and house price developments, the situation could be characterized as an emerging asset price boom since 2003.¹⁰

E. Asset Price Booms, Inflation, Recessions, and Financial Instability

14. In general, monetary authorities may be concerned about asset price boom periods because of the macroeconomic repercussions. In particular, price booms may signal a rise in inflation or be followed by an asset price reversal that results in a slowdown in economic activity and, possibly, a period of financial instability.

15. Increases in asset prices could signal a rise in inflation or inflation expectations. If asset price developments are a leading indicator for inflation, a tightening of monetary policy may be warranted, to avoid an increase in inflation. In the context of a new Keynesian model with sticky prices and a financial accelerator, Bernanke and Gertler (1999, and 2001) show in a number of simulations that monetary authorities should indeed consider asset price changes if, and only if, they signal changes in inflation expectations. However, the empirical evidence on the relationship between asset price booms and inflation is mixed (see also Stock and Watson, 2003). Borio, English, and Filardo (2003) find that, on average, inflation does not increase significantly during, and directly after, the boom. Borio and Lowe (2002) suggest that causality may also run in the opposite direction. A disinflation period or a low inflation environment may be conducive to the development of an asset price boom. Most of the asset price booms in industrial countries during the late 1990s/early 2000s occurred during a period of disinflation or in a low-inflation environment. Developments in South Africa also appear to correspond to the situation where a boom period developed during a disinflation period. At the same time, there is some indication that inflationary pressures may be building.

16. To provide further evidence on the link between asset price booms and inflation, Figure II.2 compares inflation developments around the last year of the asset price boom for our sample of industrial countries. In each case, the last year of the asset price boom, t-5 depicts the situation five years before, and t+5, the situation five years afterwards. The data indicate that, on average, inflation has slightly increased during the asset price boom period. They do not, however, suggest that asset price booms are a clear leading indicator for future inflation, as inflation tended to fall after the boom period.

Inflation around Asset Price Booms

(In percent)

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Sources: SARB and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Inflation around Asset Price Booms

(In percent)

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Sources: SARB and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Inflation around Asset Price Booms

(In percent)

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Sources: SARB and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

17. Increases in asset prices are often followed by asset price corrections which, in turn, may lead to a slowdown in economic activity and may be accompanied by a period of financial instability.¹¹ From a policy perspective, it would be useful to identify regularities that help to distinguish between high-cost booms and low-cost booms. To this end, Figure II.3 compares the development of selected macroeconomic variables between boom periods that were followed by a recession and boom periods that were not followed by a recession.¹² The graphs suggests that asset price booms are more likely to be followed by recessions if the boom period in asset prices is:

• largely driven by a boom in residential property prices (peaks after stocks),

• characterized by an easing of monetary policy during the boom period and fairly strong credit growth,

• accompanied by some upward pressure on inflation, and

• followed by a sharp decline in credit and investment.

Asset Price Booms and Macroeconomic Indicators

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Own calculations. This figure characterizes the development of main macroeconomic indicators around asset price booms. The left column refers to booms that were followed by a recession whereas the right column depicts booms that were not followed by a recession. In each case, t is the last year of an asset price boom.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Asset Price Booms and Macroeconomic Indicators

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Own calculations. This figure characterizes the development of main macroeconomic indicators around asset price booms. The left column refers to booms that were followed by a recession whereas the right column depicts booms that were not followed by a recession. In each case, t is the last year of an asset price boom.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Asset Price Booms and Macroeconomic Indicators

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Own calculations. This figure characterizes the development of main macroeconomic indicators around asset price booms. The left column refers to booms that were followed by a recession whereas the right column depicts booms that were not followed by a recession. In each case, t is the last year of an asset price boom.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Asset Price Booms and Macroeconomic Indicators

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Own calculations. This figure characterizes the development of main macroeconomic indicators around asset price booms. The left column refers to booms that were followed by a recession whereas the right column depicts booms that were not followed by a recession. In each case, t is the last year of an asset price boom.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Asset Price Booms and Macroeconomic Indicators

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A002

Source: Own calculations. This figure characterizes the development of main macroeconomic indicators around asset price booms. The left column refers to booms that were followed by a recession whereas the right column depicts booms that were not followed by a recession. In each case, t is the last year of an asset price boom.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Developments in South Africa appear to share some of these characteristics, notably the loosening of monetary conditions and a pick up in inflation.

18. Asset price changes may also lead to financial imbalances and financial instability in the medium term.¹³ A number of banking crises in industrial countries were associated with a property price boom, including the crises in Japan (1992) and in the Nordic countries (e.g., Norway in 1987, Finland in 1991, and Sweden in 1991). At present, however, the banking sector in South Africa, which finances most real estate transactions, appears reasonably well protected against a possible drop in property prices:

• The mortgage debt of households has grown significantly less rapidly than the market value of housing; the ratio of household debt to market value of housing shrank to 42 percent in 2003 from 58 percent in 1999.

• Banks are well capitalized. The average risk-weighted capital-adequacy ratio stood at 12.2 percent at the end of December 2003.

• Net nonperforming loans (nonperforming loans less specific provisions) as a ratio of total loans was only 2.4 percent.

• Banking sector stocks have outperformed the total market index since 2003.

19. Still, the distress phase at the end of a boom period may be long-lasting, while economic activity is holding up well. Therefore, the link between asset prices and financial stability may be of a longer-term nature.

F. Practical Issues for Policymakers

20. If policymakers were able to identify high-cost asset price booms, they would be confronted with two choices. Policymakers can opt for a proactive monetary policy that aims at preventing the emergence of an “asset price bubble” or follow a “reactive policy” of aggressive easing once the asset price bubble has burst. A proactive policy rests on the ability of the authorities to influence asset prices in the desired manner.¹⁴

21. It remains open to debate as to the extent the monetary authorities can influence asset price developments with the necessary degree of precision. Given the complex nature of the transmission mechanism, the increase in interest rates necessary to halt an asset price boom is difficult to determine. A small increase in interest rates may have no significant impact on stock or property prices. It may even lead to a further boost in asset prices, as small increases may reassure consumers and investors that the central bank is “ahead of the curve”. Moreover, the existence of various asset classes complicates the appropriate timing of the move, in particular if asset price developments are not synchronized.¹⁵

22. In addition, (small) open emerging market economies may not face the same set of policy options as large industrial countries, in particular the United States. Although cross-border trading of real estate is more difficult than cross-border trading of securities, recent IMF research (2004) shows that property price developments are linked internationally through cross-border financial integration. Global interest rate developments and global economic activity appear to be major determinants of house prices in industrial countries. The average of the pairwise correlations of annual changes in real property prices between the United States and the other 17 industrial countries in our sample is 0.35. Between 1994 and 2002, the correlation between annual changes in real property prices in the United States and South Africa was 0.76.¹⁶ Cross-border linkages diminish the impact of domestic policy changes.

23. Experience from industrial countries suggests that there appears to be some reluctance from central bankers to admit that asset price developments play a significant role in their policymaking.¹⁷ From a central banker’s perspective, the risks involved are asymmetric. In the case of preemptive tightening, there is a risk that the central bank will be blamed for bursting sound economic developments (The Economist, 2002, September 5). On the other hand, monetary policy may well be applauded for easing aggressively during an asset price bust period. However, a policy that favors an aggressive lowering of interest rates when asset prices tumble and (explicitly) refrains from an increase when asset prices boom, may create moral hazard problems. As long as private investors anticipate that the central bank will aggressively lower interest rates in the case of a bursting asset price boom, the central bank would provide some sort of safety net to international investors. Investors have incentives to follow a riskier investment strategy (Trichet, 2003). This asymmetry has the characteristics of a put option and may, therefore, give additional stimulus to an asset price boom.

24. If the central bank aims at influencing asset price developments, an appropriate communication strategy may be critical. Unfortunately, economic theory offers few insights into the linkages between communication, public reaction, and economic outcomes. Anecdotal evidence suggests that policymakers run the risk of contributing to financial market volatility, if they respond inappropriately to asset price developments, as for example has often been the case where policymakers tried to support exchange rates at levels not consistent with fundamentals.

25. The communication strategy needs to take into account the likelihood that the link between asset price developments and financial stability is of a longer-term nature. A number of central banks in industrial countries regularly publish Financial Stability Reports. In March 2004 the South African Reserve Bank (SARB) published its first Financial Stability Review, which will help monitor developments in the financial system and help identify potential risks. Increasing sophistication of these reports will be a useful communication tool and may complement existing analyses for the conduct of monetary policy. As Mussa notes (2003, p. 50), it is important that any change in interest rates is undertaken and explained in the context of overall economic developments.

G. Policy Conclusions for South Africa

26. South Africa appears to share some of the characteristics (property price boom, easing of monetary policy, strong domestic credit growth) of asset price boom periods in industrial countries. These were often followed by a period of weak growth and, in some cases, banking sector fragility. At the same time, however, a closer analysis of available indicators suggests that aggregate asset price developments are, by and large, still in line with fundamentals.

27. The experience of industrial countries also suggests that a number of practical obstacles need to be overcome before a more proactive role of monetary policy is warranted. Because of the importance of global interest rates in explaining future movements in house prices, the role of monetary policy may be more limited in smaller economies. Open emerging market economies, therefore, may have to focus on the development of an appropriate regulatory environment and financial infrastructure to reduce the risk of future bubbles.

28. In South Africa, one risk relates to a sudden increase in interest rates, although at this stage the banking sector seems to be reasonably resilient to such a shock. The lag of long-term financing options makes borrowers more vulnerable to shifts in interest rates. To reduce this vulnerability, further analyses should be undertaken to investigate what changes may be needed to create an environment that would be conducive to a larger set of mortgage contracts. A larger variety of available mortgage contracts, including longer-term fixed-rate contracts, should allow for a more efficient allocation of interest rate risk.

29. As in many industrial countries, housing price data availability in South Africa is somewhat limited and relies on one private sector provider, Absa, the largest mortgage provider. A more systematic nationwide collection of data would provide more representative data. Moreover, if quality improvements are not adequately captured in housing market data, the measured increase in house prices may be overstated. At the same time, it would be important to collect comprehensive data on commercial property price developments. Commercial property prices also tend to follow boom bust cycles that are not necessarily synchronized with developments in residential property prices, but may lead to similar negative economic effects. Given the importance of the housing sector, improvements in the reliability of housing market data would be desirable.

Box II.1.

Is the Wealth Effect Important For Consumption in South Africa?

The analysis uses a simple, traditional consumption model of the form:

$\begin{array}{cc}\left(1\right)& C_t=\alpha +\beta Y_t\end{array}+\delta W_t+{\varepsilon }_t,$

where C is real private consumption expenditure, Y real GDP (as proxy for real disposable income), W is real wealth, and s is the error term. The standard interpretation is that the coefficient δ is the marginal propensity to consume out of wealth, i.e. the increase in consumer spending associated with an increase in wealth. Wealth is separated into stock market wealth and housing wealth, arguably the two most important wealth components. Stock market wealth is proxied by stock market capitalization, and housing wealth is proxied by real house prices, due to lack of availability of housing wealth data.¹⁸ The coefficients on the two wealth components may differ because of differences in ownership distribution, liquidity, and volatility of both wealth components.

The relationship is estimated using a vector-error correction model with 4 lags for the period 1985Q1 to 2004Q1. Tests for nonstationarity and cointegration were performed prior to the estimation. Results indicate that both types of wealth are statistically significant in the long run. The cointegration relationship gives the following results: the estimated coefficients on wealth represent elasticities, figures in parentheses are standard errors.

Table. Estonia: Macroeconomic Scenarios, 2000-2002

(In percent of GDP; unless otherwise indicated)

Table. Estonia: Macroeconomic Scenarios, 2000-2002

(In percent of GDP; unless otherwise indicated)

log(GDP)	log (stock market wealth)	log (house prices)
1.12	0.1090	0.0932
(0.1079)	(0.0132)	(0.0409)

View Table

Table. Estonia: Macroeconomic Scenarios, 2000-2002

(In percent of GDP; unless otherwise indicated)

log(GDP)	log (stock market wealth)	log (house prices)
1.12	0.1090	0.0932
(0.1079)	(0.0132)	(0.0409)

The estimated coefficients on housing wealth and on equity wealth are similar. Results suggest that a sustained 10 percent increase in real stock market wealth or real property prices leads to an increase of around 1 percent in real private consumption expenditure. In the vector error correction formulation, the adjustment coefficient is significantly estimated at -0.15, indicating a reversion of consumption towards its long-run relationship, at a speed of 15 percent every quarter.

To see whether the wealth effect has changed over time, the standard model was estimated over a second time period: 1994Q1-2004Q1. In this case, the coefficient on the house price was not significant. These findings mirror results for industrial countries that parameter estimates are unstable, and this may indeed suggest that the wealth effect on consumption in South Africa is fairly small.