Estimates of Household Sector Wealth for South Africa, 1970–2003¶

Literature › Works

Janine Aron; John Muellbauer (2006). Estimates of Household Sector Wealth for South Africa, 1970–2003. Review of Income and Wealth, Series 52, Number 2. Blackwell Publishing (International Association for Research in Income and Wealth).


Type	journal article
Scope	Economics and economic history of apartheid
Method	descriptive quantitative
Period studied	1970–2003
Regions	South Africa
Themes	macro fiscal finance debt, inequality poverty unemployment, methods data sources, sanctions divestment, transition post apartheid
Citation key	`aron2006`

Argument¶

The paper builds the first coherent quarterly market-value balance sheet for the South African household sector, 1970–2003. It does so by cumulating Reserve Bank flow-of-funds data from assumed 1969 benchmarks, converting book to market values with asset price indices, and testing sensitivity to alternative benchmarks. Relating asset-to-income ratios to asset prices, interest rates, tax and regulation shows a rising weight of pension wealth, a falling weight of directly held securities, a housing boom and bust after the early-1980s gold boom, and rising household debt alongside falling liquid assets from the early 1980s to late 1990s. The authors argue that such wealth estimates are needed for consumption and debt models, and that wealth fluctuations track gold prices, the 1985 debt crisis, sanctions and financial liberalisation.

Evidence and method¶

South African Reserve Bank flow-of-funds (National Financial Account), Capital Market Statistics, Quarterly Bulletin, unpublished Reserve Bank household debt and constant-price housing stock data, ABSA house price index, JSE share and bond indices, 1995 and 2000 Statistics South Africa income and expenditure surveys, and US and UK benchmarks (Federal Reserve, ONS/CSO).

Place in the debates¶

Measuring household wealth and saving behaviour in emerging markets (asset and credit channels of monetary transmission) (see Data reliability and econometric method). Argues South Africa's financial depth makes wealth effects relevant despite deep inequality, so market-value household balance sheets are needed; the authors take the Reserve Bank to be the right body to publish them.
Effect of sanctions and the 1985 debt crisis on the domestic economy (see Sanctions, disinvestment and the anti-apartheid movement). Treats sanctions and the debt crisis as constraints on capital and trade that weakened growth and real house prices and raised the relative price of durables, but does not test this causally.

Findings¶

No official market-value household balance sheets exist for South Africa; the authors build the first coherent set. (p. 1)

We construct the ﬁrst coherent set of the main elements of household-sector balance sheet estimates at market value for South Africa.
Spending is highly concentrated at the top of the distribution, so the affluent dominate aggregate behaviour. (p. 2)

the top quintile in 2000 and 1995 (in parenthesis) were 64 percent (63 percent), the top two quintiles 82 percent (82 percent), and the top three quintiles 92 percent (93 percent)
Earlier consumption estimates give significant long-run wealth propensities to spend. (p. 5)

We found statistically highly signiﬁcant long-run propensities to spend of around 0.12 out of liquid assets minus debt, around 0.06 out of illiquid ﬁnancial assets and around 0.06 for housing wealth.
There are no household asset surveys, so wealth must be built from flow-of-funds data and assumed benchmarks. (p. 7)

There are no household surveys of asset ownership; and data on the estates of the deceased have not been used to obtain estimates of personal wealth.
Wealth-to-income fluctuations are tied to gold prices and political and economic crises, including the 1985 debt crisis and sanctions. (p. 16)

When economic and political difﬁculties increased in the 1980s, and the debt crisis of 1985 and international trade and ﬁnancial sanctions severely constrained access to capital and trade, growth weakened and real house prices began a long-term decline.
Housing assets boomed and then fell in the early 1980s after the gold price spike. (p. 18)

Housing assets experienced a boom and then bust in the early 1980s, after gold prices rose temporarily from around $300 to over $800 dollars per ounce.
Liquid assets earned negative real after-tax returns for two decades while marginal tax rates rose. (p. 18)

the real after-tax return on liquid assets has been negative from the early 1970s to the early 1990s—apart from a brief spell in 1984–85
Financial liberalisation from the early 1980s lifted the debt-to-income ratio even as real interest rates were high. (p. 17)

The removal of quantitative controls over credit in the early 1980s, associated with a move to controlling credit expansion via higher interest rates, induces a positive correlation
Trade sanctions between 1985 and 1990 raised the relative price of durables while the real durables stock fell relative to income. (p. 19)

Trade sanctions between 1985 and 1990, help to account for the rise in the relative price of durables
The equity share of pension assets roughly doubled, and pension wealth tracks equity returns, tax incentives and deregulation. (p. 19)

It also shows the rising proportion of pension assets invested in equities—from 20 percent in the early 1970s to over 50 percent by the 1990s.
State pension rights were extended to all races from 1989 reforms. (p. 13)

In a series of reforms since 1989, state pension rights, providing a minimum of income support for those over retirement age, have been extended from the white population only, to all races.

Evidence ledger¶

Sources used¶

Source	Kind	Period	How used
South African Reserve Bank, National Financial Account (flow of funds), 1970–91 annual, quarterly from 1992	official statistical series	1970–2003	Household-sector financial flows are cumulated from benchmark stocks (equations 1 and 2) into asset stock estimates.
South African Reserve Bank, Capital Market Statistics 1948–92 (and Quarterly Bulletin)	official statistical series	1948–2003	Supplies stock data at book value (and some market value) for pension funds, long-term insurers, unit trusts, participation bonds and local authority/public enterprise securities.
South African Reserve Bank, Quarterly Bulletin: household debt, personal disposable income, bank deposits by depositor type	official statistical series	1970–2003 (debt published from 1991Q3; deposits by depositor from 1991Q3)	Provides household debt, income denominators and the 1991Q3 benchmark for personal broad money holdings.
Unpublished SARB data: pre-1991 household debt, mortgage debt, consumer credit, constant-price housing stock	official statistical series	1960–2003	Used to construct household debt ratios and gross housing assets (constant-price stock converted using house prices).
Public Finance Statistics (ownership of end-1969 government stocks)	official statistical series	end-1969	Gives the 1969 benchmark for household holdings of short-term and long-term government stock.
Treasury estimate of household holdings of long-term government bonds	official statistical series	recent (unspecified)	Used as a second benchmark, requiring a modified revaluation adjustment for bond holdings.
JSE government bond price indices (from 1980), SARB bond price index (from 1999), JSE all share index (spliced to FTSE South Africa all share index in 2002)	stock market financial data	1980–2003	Price indices used to revalue book-value bond and equity stocks to market value; pre-1980 bond indices derived from price-yield relationships.
ABSA Bank Ltd house price index and annual ABSA survey of land-to-building value ratio	other	1966–2003	Medium-house price index converts constant-price housing stock to current values; land share calibrated to the ABSA survey.
SARB annual durables purchases data (four categories) and price indices/deflators	official statistical series	1946–2003	Perpetual inventory method with assumed service lives yields the market value of consumer durables stocks.
World Bank database on financial development and structure (Beck et al., 1999), 2003 data	international trade or finance data	2003	Table 1 compares South African financial development indicators with industrial and emerging economies.
Statistics South Africa, Income and Expenditure Surveys of October 1995 and October 2000	household or labour survey microdata	1995, 2000	Shows concentration of household spending in the top quintiles, justifying a focus on affluent households' asset access.
U.S. Federal Reserve and U.K. National Accounts (Blue Book) balance sheets, and U.K. CSO (1978) household balance sheets	secondary literature compilation	1969–2002	Supplies the 80% and 85% household shares of broad money and the 1969 equity ownership ratios used to calibrate South African benchmarks.
Mouton Report (1992) and Katz Commission Third Interim Report (1996)	commission or official report	1992, 1996	Cited for underfunding of official pension funds before the 1990s and for the taxation of pension fund income after 1996.
Estate-based wealth studies: McGrath (1982) for Natal 1972; Van Heerden (1996) for Transvaal 1985	secondary literature compilation	1972, 1985	Mentioned as the only estate-data studies of wealth; found to give little information on asset composition so not used.
Aron and Muellbauer (2000a, 2000b; 2004) econometric models and appendices	secondary literature compilation	1970–2003	Provides the econometric evidence on wealth effects in consumption and debt equations and the technical appendices on benchmarks and price indices.
Prinsloo (2000, 2002) and Van der Walt and Prinsloo (1993, 1995) SARB papers	commission or official report	1993–2002	Used for household debt data and components, durables service lives, and after-tax returns on liquid assets.

New data¶

Quarterly market-value household sector balance sheet estimates for South Africa, 1970–2003: liquid assets, debt, pension wealth, directly held securities, housing and consumer durables, built by accumulating flow-of-funds data from assumed 1969 benchmarks and converting book to market values with asset price indices.

Inference¶

Quantification: Statistical analysis.

Time-series construction of stocks via stock-flow identities and revaluation with price indices from assumed benchmarks, with sensitivity analysis on benchmarks. Trends are interpreted through graphical comparison with relative prices, returns, taxes and regulation, plus simple correlations and reference to the authors' earlier econometric work.

Key claims¶

Claim	Basis	Strength
Market value estimates of household wealth can be built back to 1970 despite no official balance sheets, and these are the first coherent set for South Africa. (p. 285)	SARB flow of funds, Capital Market Statistics, Quarterly Bulletin and unpublished SARB series, cumulated from assumed 1969 benchmarks.	Partial data
Pension wealth rose strongly relative to income while directly held securities trended downward. (p. 285)	Capital Market Statistics pension stocks revalued with price indices; equity benchmark assumed (15-fold/25-fold), shown in Table 2.	Partial data
The debt-to-income ratio rose and the liquid-asset ratio fell from the early 1980s to the late 1990s, linked to financial liberalization. (p. 301)	SARB household debt and deposit series (unpublished pre-1991), with correlations with real interest rates and inflation.	Partial data
Housing wealth relative to income fell after 1983 mainly due to the decline in real house prices, tied to gold price booms and busts and interest rates. (p. 302)	SARB constant-price housing stock and ABSA house price index; Figure 3; gold price movements cited.	Illustrative or anecdotal
The rise in pension assets to income is explained by total returns on equities and bonds, regulatory relaxation of asset ratios, and tax advantages. (p. 303)	Log total return indices and decade correlations (>0.87), Figures 4–5; regulatory and tax changes described.	Partial data
Household spending is concentrated in affluent households, so aggregate monetary transmission is dominated by those with access to assets and credit. (p. 286)	Statistics South Africa Income and Expenditure Surveys 1995 and 2000 expenditure quintile shares.	Systematic data
South Africa's financial development is comparable to industrial economies and above emerging markets. (p. 286)	World Bank financial structure database for 2003, SARB mortgage data (Table 1).	Systematic data

The author's caveats¶

The authors state that no official balance sheets or household asset surveys exist and benchmarks are assumed (deposit share 80/85%, equity ratio 15:1 or 25:1, with large differences in net wealth). Book versus market valuation is mixed in insurer data in 1985–91. Foreign assets, unincorporated business assets and unlisted equity are excluded. Pension rights at defined benefit are omitted.

Gaps in the evidence¶

The 1969 benchmarks, especially for directly held equities and household broad money, are borrowed from U.S./U.K. experience rather than South African data, so levels (not just trends) are uncertain; share register surveys or estate data would test them. Household asset surveys, tax records, or flow-of-funds revisions would anchor levels but largely did not exist then. Excluded foreign, unincorporated business and farm assets, plus racial distribution of wealth, are not covered, which matters for an apartheid-era economy and for the interpretation of trends.

Replicability: Data described but not public.