RDP 8606: The Nation's Wealth – Some New Calculations for Australia 3. New Calculations of Australia's Aggregate Personal wealth
June 1986
This section reports and interprets new calculations of Australia's aggregate personal wealth for the eighties. I make use of new statistical series recently published by government departments, stock exchange data, and other specially prepared series to generate personal wealth estimates which value all significant components of Australia's personal wealth at market prices or a close approximation. Like the H-B and Williams studies, I use a form of inventory approach, and in fact take the H-B procedures as a convenient point of departure.
(a) Valuation Procedures
The following components of personal wealth are separately identified and valued: residential land and dwellings; household durables; rural land, improvements, and inventories; other business fixed assets and inventories (including land); private holdings of government bonds, including local authority and semi-government issues; and Reserve Bank liabilities to the private sector. Domestic ownership of foreign assets are then added in, and the value of assets and inventories owned by the foreign sector is subtracted.
(i) The Housing stock
The value of residential housing is estimated by combining census estimates on dwelling numbers with a quarterly price index of dwellings compiled from sales data.[9] The census data are modified each quarter by estimates of completions and demolitions.[10] The price index reflects sales in Brisbane, Sydney, Melbourne and Adelaide. These cities contain about 75 per cent of the nation's dwelling stock. Table 2 reports housing stock valuations by Roger (1981) and Williams (1983) and the non-land dwelling stock valuations by H-B and the ABS [Walters and Dippelsman (1986)], to allow comparison with our new estimates (column 5), for selected years from 1973 to 1981.
Year End of June | (1) | (2) | (3) | (4) | (5) |
---|---|---|---|---|---|
Helliwell/Boxall | ABS | Roger | Williams | Piggott | |
1973 | 37,221 | 29,402 | 58,403 | 75,700 | – |
1977 | 78,996 | 62,893 | 117,978 | 149,500 | – |
1979 | 88,895 | 74,600 | 135,110 | 174,500 | – |
1981 | 118,277 | 99,948 | – | 250,600 | 297,456 |
1984 | 159,242 | 138,067 | – | – | 388,579 |
Sources: Col. 1 Reserve Bank series extending H-B estimates. |
The series in the first two columns are not comparable with columns 3–5 because they omit land value. Differences between the H-B and ABS estimates reflect different asset life distribution and depreciation assumptions. The series reported in columns 3–5, are derived from census data on quantities and Valuer General and other data on sale prices.
Roger (1981) does not provide details of his calculations, and it is therefore impossible to identify the source of the discrepancies between his estimates and the others. Williams (1983) and I both rely on price series derived by the Department of Housing and Construction (DHC), although the series we use differ in a number of respects. One possible source of the discrepancy between these two series is that earlier DHC price information used by Williams may have seriously underestimated the average value of Sydney dwellings, while this bias was eliminated in the more recent price series used here.[11] Since Sydney Prices receive the highest weighting of all the capital city series in deriving an overall index, the discrepancy between columns 4 and 5 could be partly accounted for by differences between the earlier and current DHC price series. In addition, Williams adjusts his DHC price series to take account of non-metropolitan housing, and separates houses and flats. He undertakes these calculations for 1976, at which point he estimates the average value of an Australian dwelling to be $32,100. If this figure is accepted, then a proportional adjustment using my data would reduce my June 1981 housing estimate from $297 billion to $274 billion. The limitations of these sales data for aggregate wealth estimation are discussed further below (p.18).
(ii) Household Durables
Two series on the value of household durables are maintained in Australia, one for the Australian Treasury, for use in the NIF-10 model, and another for the Reserve Bank. They do not differ significantly and I have used the RBA series in the study.[12]
The value of household durables is estimated using a perpetual inventory model, and they are therefore valued at replacement cost. This is defended as an approximation to market value, since the supply of most durables is elastic, at least in the long run. It is not a wholly satisfactory proxy, however: for example, exchange rate and tariff changes can alter the market value of durables such as motor vehicles, and non-reproducible assets, such as paintings and antiques, are not valued. A further point to bear in mind is that the sale prices of second hand goods are usually lower than replacement cost, because of the “lemons” argument [see Akerlof (1970)]. The probability that an item offered for sale second-hand is in some way defective (a lemon) is greater than the probability for the whole population of such items, and the price such an item commands reflects this probability difference. Essentially, the market fails because of asymmetries in information between buyer and seller. The seller knows the characteristics of the good he is offering; the buyer does not.
(iii) Rural Wealth
The capital value of farms, improvements, equipment and inventories has not been satisfactorily incorporated in either H-B or Williams. As pointed out earlier, Williams ignores all unincorporated wealth, including unincorporated rural land. H-B use a book value gross up procedure which extends their market valuation of corporate stock to the unincorporated sector, including Agriculture. This is unsatisfactory, since neither production techniques nor output price movements in the rural sector are well approximated by the corporate sector.[13]
To overcome these problems, I use a series on the market value of rural sector capital stock derived from an annual survey of the rural sector.[14] The survey explicitly values farms, including land, improvements, equipment and stock, and covers units which in total contribute around three quarters of gross farm product. Survey estimates are grossed up to national estimates using the gross farm product ratio.
(iv) Other Business Fixed Assets and Inventories
The valuation of corporate sector business fixed assets and inventories is undertaken using Sydney Stock Exchange (SSE) share market data.[15] These data comprise the market value of equity and the book value of net financial assets for companies covered by the SSE's STATEX service. The difference between these two totals gives a market value of business fixed assets and inventories. The STATEX service covers about 85 per cent of the gross profits of all companies listed on the SSE. Companies listed on the SSE account for about half all corporate activity.[16] Estimates of gross operating surplus for STATEX firms and for the whole economy are used to scale the STATEX figures to national corporate sector estimates. The reliability of this procedure depends on whether GOS/capital ratios are similar for the STATEX sample and for other companies. No direct market valuation is available for the wealth of non-farm unincorporated enterprises, including land. I proceed by assuming that, if the rural sector is excluded, corporate wealth is related to non-farm unincorporated wealth by the corresponding net capital stock ratio. Net capital stock estimates from the ABS[17] are used to gross up the value of incorporated business fixed assets and inventories to the total non-farm value.
This procedure ignores the value of the physical capital stock of financial enterprises, whether incorporated or not. Rough calculations suggest that this leads my aggregate estimates to understate the value of private wealth by 2–1/2 per cent to 3 per cent or $13 billion at June 1981. This is discussed further below.
(v) Government Bonds
It is conventional to include the market value of non-official holdings of government bonds as part of personal or private wealth. It is debatable whether government bonds do in fact represent net wealth, as Barro (1974) among others has highlighted. I have chosen to calculate and report the value of government debt, and to leave the question of interpretation to the user.
The face value of outstanding government bonds, issued by both the Commonwealth government and local and semi government authorities, is published regularly by the Reserve Bank, together with an analysis by type of holder which permits official and foreign holdings to be netted out. The Bank also maintains a series relating face to market value for Commonwealth bonds (excluding Australian Savings Bonds), and all Commonwealth issues can thus be valued at market prices.
Local and semi government authority issues are more difficult to value. These issues are not frequently traded; most purchasers select bonds of this type with the intention of holding them until maturity. As well, the maturity of these bonds is usually of greater duration than is the case with Commonwealth issues. These two considerations suggest that if the market value falls below face value for Commonwealth bonds, then the market face value ratio for local issues will be even lower, and that the difference between the market values of the two categories will be greater the higher is the interest rate. This is because the longer is the time to maturity of a bond, the greater the risk attached to it. In Table 3 I reproduce estimates of the market face value ratios of Commonwealth bonds, and “guesstimates” of the local and semi-government ratio, based on the above considerations, using the 10-year bond yield as the interest rate. Also reported are the total face values of non-official domestic holdings of each. The ratios are applied to face value totals to arrive at a market estimate of privately held government debt.[18]
Year End of June |
Commonwealth Bonds | Local and Semi-Govt Bonds | ||
---|---|---|---|---|
(1) Face value |
(2) Market-Face Value ratio |
(3) Face Value |
(4) Market-Face Value Ratio |
|
$m | $m | |||
1980 | 18,296 | 0.84 | 14,280 | 0.80 |
1982 | 20,309 | 0.78 | 18,866 | 0.70 |
1984 | 34,485 | 0.93 | 26,122 | 0.90 |
Sources: |
(vi) Reserve Bank Liabilities
The Reserve Bank's liabilities to the private sector include all currency notes and coin issued, statutory reserve deposits (SRDs) and other minor items. The Bank maintains a series for this wealth component, and I have used this without amendment.
(vii) Foreign sector adjustments
Two kinds of foreign sector adjustments must be made to complete a valuation of Australian owned personal wealth. Australian assets held overseas must be added in, and foreign claims on assets located in Australia must be subtracted.
Levels of Australian investment abroad are derived from ABS data. Some Australian holdings of foreign corporate equities are recorded at paid up rather than market value. The series should therefore be regarded as placing a lower bound on the value of Australian ownership of foreign assets.
The value of foreign claims on assets located in Australia is calculated from ABS data on foreign investment in Australia. Because corporate equities are recorded at paid up value in these series, a market valuation was established by deriving a market to paid up value in a sample of firms on the SSE. Table 4 reports the resulting series from 1980.[19] Unincorporated foreign holdings may not be valued at market prices, since there is no means of identifying movements in the prices of physical assets held by foreigners after purchase.
Year (at June 30) |
Australian Ownership of Foreiqn Assets $A million |
Level of Foreiqn Investment in Enterprises in
Australia $A million |
||||
---|---|---|---|---|---|---|
Corporate Equities | Otherb | Total | ||||
paid-up value | ratio of market to paid-up value | estimated market value | Gross Foreign Investment in Australia | |||
1980 | 3,438 | 5,811 | 3.7640 | 21,873 | 11,362 | 33,235 |
1981 | 4,124 | 7,526 | 4.1387 | 31,148 | 14,817 | 45,965 |
1982 | 5,249 | 8,679 | 2.7980 | 23,284 | 23,774 | 48,058 |
1983 c | 7,409 | 10,344 | 3.5217a | 36,428 | 35,940 | 72,368 |
1984 c | 9,362 | 11,015 | 3.8625a | 42,545 | 44,521 | 87,066 |
1985 c | 15,721 | 11,950 | 4.9907a | 59,638 | 66,227 | 125,865 |
a. estimated in movements equivalent to movements in share price
index. Sources: Australian ownership of foreign assets: Australian
Bureau of Statistics, “Foreign Investment, Australia”, Cat. No.
5306.0, various issues. |
(b) New Estimates of wealth
The new estimates of wealth, in aggregate and by component, are reported in Table 5. “True” estimates are reported for the end of June each year. For other quarters, linear interpolation was used for some components. Each component of Australia's personal wealth has been valued at market prices, together with claims on and liabilities to the public and foreign sectors. While much of the groundwork for these estimates had been laid in the H-B and Williams studies, the results reported here represent the first comprehensive market valuation of the Australian wealth stock.
Year | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) |
---|---|---|---|---|---|---|---|---|
Residential Land and Dwellings |
Household Durables |
Rural Wealth (a,c) |
Other Business Fixed Assets and Inventories (b,c) | Australian Ownership of Foreign Assets |
Non-official domestic holdings of Govt bonds (b) |
RBA Liabilities |
Total Personal Wealth |
|
80(1) | 241.4 | 30.2 | 53.3 | 99.2 | 3.7 | 28.1 | 6.8 | 462.7 |
(2) | 253.3 | 30.6 | 57.2 | 98.2 | 4.0 | 26.8 | 6.9 | 477.0 |
(3) | 275.8 | 31.4 | 59.3 | 97.4 | 4.1 | 28.3 | 7.0 | 503.3 |
(4) | 278.0 | 32.1 | 61.4 | 96.5 | 4.1 | 28.9 | 7.6 | 508.6 |
81(1) | 287.4 | 32.8 | 63.5 | 95.7 | 4.2 | 30.1 | 7.5 | 521.2 |
(2) | 297.5 | 33.6 | 65.6 | 94.8 | 4.2 | 29.2 | 7.6 | 532.5 |
(3) | 308.6 | 34.4 | 68.0 | 97.9 | 4.5 | 27.7 | 7.9 | 549.0 |
(4) | 319.8 | 35.9 | 70.5 | 101.0 | 4.8 | 30.3 | 8.4 | 570.7 |
82(1) | 324.6 | 36.8 | 72.8 | 104.1 | 5.1 | 29.6 | 8.2 | 581.2 |
(2) | 325.3 | 37.8 | 75.3 | 107.1 | 5.3 | 29.0 | 8.6 | 588.4 |
(3) | 329.2 | 38.7 | 78.2 | 103.1 | 5.8 | 33.2 | 8.5 | 596.7 |
(4) | 333.0 | 39.7 | 81.1 | 99.2 | 6.4 | 38.8 | 9.3 | 607.5 |
83(1) | 347.5 | 40.4 | 84.1 | 95.2 | 6.9 | 40.1 | 9.0 | 623.2 |
(2) | 355.7 | 41.6 | 87.0 | 91.2 | 7.4 | 40.3 | 9.1 | 632.3 |
(3) | 355.0 | 42.3 | 87.9 | 99.8 | 7.9 | 46.6 | 9.3 | 648.8 |
(4) | 363.5 | 43.1 | 88.8 | 108.4 | 8.4 | 54.5 | 10.3 | 677.0 |
84(1) | 374.9 | 43.9 | 89.7 | 117.0 | 8.9 | 55.3 | 9.8 | 699.5 |
(2) | 388.6 | 44.3 | 90.6 | 125.6 | 9.4 | 55.6 | 10.2 | 724.3 |
(3) | 400.7 | 45.0 | 93.6 | 123.2 | 11.0 | 59.2 | 10.6 | 743.3 |
(4) | 416.1 | 45.6 | 96.6 | 120.9 | 12.5 | 61.3 | 11.5 | 764.5 |
85(1) | 427.7 | 46.6 | 99.5 | 118.5 | 14.1 | 63.5 | 11.4 | 781.3 |
(2) | 439.9 | 48.0 | 102.5 | 116.2 | 15.7 | 59.9 | 11.7 | 793.9 |
Notes: |
The results suggest that earlier calculations of Australia's private wealth stock have substantially underestimated its value. At 30 June 1981, H-B (as revised) estimated Australia's wealth at $294.7 billion, while Williams (1983) reports a value of $360.5 billion. My calculations yield a value of $534.0 billion. Table 6 provides a comparison of all three series, both in aggregate and by selected components, which allows a partial reconciliation of these disparate findings. Discussion is confined to June 1981 estimates, since that is the most recent date for which all three series are available. Comparing the Williams results with the new estimates, we find that the estimate for residential land and housing is greater by $46.9 billion in the new calculations. Unincorporated wealth, not valued by Williams, accounts for $87.9 billion. These two items account for $134.8 billion of the $173.5 billion difference between the two estimates. This remaining discrepancy should be increased by $13 billion to $51 billion, to allow for the value of land and capital used by financial enterprises. This component is covered by the H-B and Williams estimates, but not by the new calculations.[20]
Year (at June 30) |
Piggott | Helliwell-Boxall | Williams | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Total | Housing | Wealth in Unincorporated Firms | Govt Debt* | Total | Housing | Govt Debt* | Total | Housing | |||
80(2) | 477.0 | 253.3 | 78.0 | 33.7 | 258.6 | 101.4 | 22.3 | 206.3 | 301.6 | ||
81(2) | 532.5 | 295.5 | 87.9 | 36.8 | 294.7 | 118.5 | 24.2 | 250.6 | 360.5 | ||
82(2) | 588.4 | 325.3 | 99.9 | 37.6 | 296.4 | 137.9 | 24.4 | – | – | ||
83(2) | 632.3 | 355.7 | 112.5 | 49.4 | 333.0 | 149.2 | 31.4 | – | |||
84(2) | 724.3 | 388.6 | 123.8 | 65.8 | 380.3 | 159.8 | 42.3 | – | – | ||
85(2) | 739.9 | 439.9 | 140.3 | 71.6 | 483.1 | 176.8 | 46.1 | – | – | ||
* Government bonds and Reserve Bank liabilities. Sources: Piggott: See text, and unpublished calculations. |
In aggregate wealth calculations, the 10 per cent discrepancy between the two results which remains unaccounted for would not be considered a major inconsistency, especially since the estimates were generated from very different approaches. The current results are, therefore, broadly consistent with the Williams estimates, once omissions by Williams and residential housing value revisions are taken into account.
In comparing the new estimates with the H-B series, the difference in the value of housing is a major component, accounting for $179.0 billion of the $239.3 billion difference. (Again, the remaining $60.3 billion discrepancy must be increased to $73 billion for comparitive purposes.) Rural land must account for part of this remaining difference, but no estimates of rural land alone are available for 1981. Earlier estimates suggest somewhat more than half rural wealth may be accounted for by the value of land. If we assumed a rural land value of, say, $35 billion (rural wealth was valued at $65.6 billion at June 1981), then there remains an unexplained discrepancy of about $38 billion. H-B neglected to include the market value of the non-official domestic holdings of local authority and semi-government bonds in their estimates, which at June 1981 were valued at $12.5 billion. Once the omissions of the H-B estimates have been valued, therefore, the discrepancy between the two series is largely explained for the date under consideration. The present approach has a good deal in common with H-B, so close agreement between the estimates after accounting for coverage differences might be expected. Nevertheless, it is reassuring that the differences in methodology adopted to achieve improved estimates do not dramatically affect the estimated value of those components valued at market prices by both series.
(c) A Critical Assessment
A number of weaknesses in the data and methods used for the new wealth estimates reported here can be readily identified, but three are of significant quantitative importance. They are the price index for residential housing, the valuation of non-farm unincorporated wealth by reference to the market value of corporate equities, and the omission of the wealth used by financial enterprises. They will be discussed in turn.
The use of the price series to value the aggregate stock of housing may be questioned on three grounds. First, the price indices refer to a sample of sales in four capital cities (Brisbane, Sydney, Melbourne and Adelaide), where site values might be thought to be higher than average. Second, a disproportionate number of new dwellings, relative to the total stock, will be sold in any one period, so the average age of dwellings sold will be less than the average for the stock, and the average floor size may be higher. A possible bias working in the opposite direction is that the average site value of newly completed homes is less than the average for the whole stock, since many new completions are located on the fringes of cities. A further point to bear in mind is that sales frequency may be greater when prices are high, so sales data could tend to bias upward the value of the housing stock.[21] While data limitations preclude any quantitative assessment of these possible biases, they should nevertheless be borne in mind in interpreting the results.
The second major weakness in the valuation procedure is the use of corporate equity market data to value non-farm unincorporated enterprises. The procedure used here is in the spirit of the H-B approach and although substantial improvements have been made, problems remain.[22] The most important of these is the possible impact of differential company and personal tax changes on the relative values of incorporated and unincorporated stock. A further issue is the extent to which expectations embodied in stock market changes can be extended to the unincorporated sector. Finally, the relative values of incorporated and unincorporated assets will be influenced in some degree by the greater liquidity of corporate equity. Once again, these issues cannot be quantified, but they are reservations which need to be recognised in assessing the reliability of the aggregate estimates.
The value of wealth used by financial enterprises cannot be valued using the methods employed here because of accounting conventions surrounding financial leasing and the inclusion of interest received in financial enterprise gross operating surplus. GOS estimates cannot be employed without double counting, since financial enterprises hold large claims on the trading sector. Net capital stock estimates for financial enterprises include the stock underlying financial leasing arrangements.
To gain some sense of the importance of this omission, I formed a rough estimate of the missing wealth based on the assumption that in 1966/67 financial leasing stock represented a negligible proportion of financial enterprise capital stock. I then used GOS financial enterprise data to scale the 1966/67 capital stock figures to give estimates for the 1980s. (The implicit assumption here is that the ratio of profits to value of physical capital stock has remained constant over this period.) I then estimated a market value of this stock by applying the wealth-capital stock estimate observed for corporate trading enterprises. This procedure generated an estimate of $13.4 billion for 1980/81, and of $21.1 billion for 1983/84 – between 2–1/2 per cent and 3 per cent of the estimated aggregate wealth.
Other problems with the calculations reported here are of less quantitative significance. They include certain omissions: the present value of public sector (pay-go) occupational pensions;[23] the wealth of private forestry, hunting and fishing, and incorporated Agriculture; the value of vacant land; and the value of non-reproducible durables such as antiques and paintings. Unvalued claims by the public on the private sector include public housing loans and public loans to the rural sector. Except for the first item, these are not thought likely to significantly affect the aggregate estimates.
Calculations of the kind considered in this paper are by their nature imprecise, and the results are not amenable to strict statistical tests to determine their reliability. It may nevertheless be useful to provide a subjective guide to the sensitivity of the aggregate estimates to errors in the valuation of components. Table 7 reports an exercise of this kind, again carried out for June 1981. It was assumed that the value of residential land and housing lay between the Williams and Piggott estimates, and that the calculation of financial enterprise physical asset wealth and unincorporated non-farm wealth could be in error by up to 50 per cent either way. Other errors were ignored. The results yield a range of $483 billion to $564 billion. The lower bound still exceeds the Williams estimate for 1981 by a third, and the corresponding H-B value by nearly two-thirds.
Estimated value | $533 billion | |
---|---|---|
Change for assumed maximum residential land and housing error |
−$46 billion – 0 | |
Financial enterprise adjustment | $7 billion – $20 billion | |
Non-farm unincorporated wealth adjustment | −$11 billion – $11 billion | |
Low | High | |
Aggregate wealth range | $483 billion – $564 billion |
Footnotes
The price index series, which dates from 1973, was prepared by the Department of Housing and Construction. The Department intends updating the series each quarter. [9]
The ABS publishes estimates of housing completions (ABS Cat. No. 8750.0). Demolition and other quantity estimates were supplied by the Indicative Planning Council. [10]
For example, in 1983, the DHC reported an average value of $66,000 for a Sydney dwelling, compared with $81,700 by the Real Estate Institute of Australia. The corresponding estimates for 1984 showed much closer agreement: $79,000 (DHC) and $82,200 (REIA). For the other states (Queensland and Victoria), there is close agreement between the two series for both years. This discussion is based on information provided by the ABS. [11]
For example, at the end of June 1981, the NIF-10 estimate was $33.7 billion, while the RBA recorded a value of $33.6 billion. [12]
In fact, in an earlier paper, Helliwell et al (1971) argued that the farm sector should be treated separately because “investment in primary production is determined by different factors than is other investment, and share market data are not of much use in establishing valuations” (p.6). [13]
This is conducted annually by the Bureau of Agricultural Economics. [14]
These data were provided by the STATEX service of the Sydney Stock Exchange. [15]
While the STATEX service covers the bulk of firms listed on the SSE, unlisted companies cannot be covered. These include a large number of overseas based companies operating in Australia but not listed on the exchange; exempt Australian companies; and a large number of smaller private companies. The STATEX service covers about 1,000 firms; the ABS survey, which the Bureau estimates covers 90 per cent of incorporated activity, includes around 5,000 firms. [16]
See Walters and Dippelsman (1986). [17]
It should be noted that the H-B series on government debt, as extended by the Reserve Bank, excludes all Local Government and Semi-Government bonds. [18]
This approach, and the series, were developed by Kim Hawtrey and Ian Russell of the Reserve Bank. [19]
See p.19 below for further discussion. [20]
I am grateful to Andrew Edquist for pointing this out to me. [21]
The substantial improvements include the separate valuation of rural wealth, the use of SSE data for valuing business wealth incorporating a far greater coverage than H-B found possible, and the use of GOS and net capital stock ratios for grossing up to national totals. H-B used book value ratios to move from their sample to national aggregates, and these have proved to be significantly affected by accounting conventions, such as asset revaluations, and by inflation. [22]
This is considered part of private wealth, since it represents a claim by the private on the public sector that is in the nature of a title. It is therefore distinct from government transfers such as the age pension. [23]