RDP 9109: Estimates of Private Sector Wealth 2. Estimating Private Sector Wealth

Tim Callen

October 1991

Five separate components of total private sector wealth are identified and estimated. These are the stock of dwellings, the stock of consumer durables, business assets (which are adjusted for overseas holdings of domestic assets and Australian holdings of foreign assets), non-official holdings of government bonds and non-official holdings of currency.

(a) The Dwelling Stock

Data for the outstanding number of dwellings are available for June 1981 and June 1986 from Census results. These figures are combined with the quarterly ABS series on the number of new dwellings completed and with estimates of demolitions and sales by housing authorities to the private sector to obtain a quarterly series on the number of dwellings.[1]

The market value of the stock of dwellings is then obtained by multiplying the number of dwellings by a suitable price series. The series used here is a weighted average of established dwelling prices in capital cities and other areas. This approach differs from most previous studies, where a series relating to the price of established dwellings in capital cities has been used. The latter is likely to overestimate the values: for instance, according to the Commonwealth Bank/Housing Industry Association (CBA/HIA) figures, the median price of an established dwelling in capital cities in June 1990 was $133,400 while outside capital cities it was $92,900.

The price series used here has been spliced together from series calculated by the CBA/HIA and the Real Estate Institute of Australia (REIA). The CBA/HIA publish a series on the median price of established dwellings in capital cities from 1984.[2] They have also recently started to publish information on the median price of established dwellings in other areas. This, however, is only available from December 1989. To calculate the series back to 1984, it is assumed that the growth rate of prices in other areas is the same as that for capital cities. For the period 1980 to 1984 the growth rates in the REIA series for capital city prices are used to backdate the CBA/HIA series.

To determine the appropriate weightings for capital cities and other areas in the overall dwelling price series, information was obtained on the number of outstanding dwellings in each capital city from the 1986 census. The weights calculated were 62 per cent for capital city prices and 38 per cent for other area prices. This gave a value for the dwelling stock of $741 billion in the June quarter 1990.[3]

Foreign ownership of the dwelling stock should be deducted from the total to obtain domestic ownership. However, because no data are available on foreign ownership of dwellings, no adjustment is made.

(b) Consumer Durables

The stock of consumer durables is defined as the sum of the stock of household durables and the stock of motor vehicles. Piggott took these data from the Reserve Bank RBII model database. This database is available only up until the June quarter of 1988. From then, the figures are updated as follows. The constant price stock is calculated each period by adding the constant price estimates of expenditure in these categories (taken from the National Accounts) to the previous period's depreciated stock. The previous period's stock is depreciated according to the depreciation rates in the Treasury NIF10 model.[4] These are 25 per cent per annum and 28.6 per cent per annum for consumer durables and motor vehicles respectively. The current period's stock is then converted to current prices by use of the relevant price deflator.

(c) Business Assets

The estimates for business assets derived here differ from those made by Piggott in several ways. Firstly, Piggott excluded the wealth of financial enterprises. Second, the companies in his sample held a disproportionately low amount of commercial property. Given that the market value of commercial property rose particularly quickly in the late 1980s, estimates of the market value of business assets based on Piggott's sample would be increasingly understated over those years. Third, although he subtracted the foreign ownership of domestic business assets, he did not include the domestic ownership of foreign business assets in his estimates of business wealth (although such an adjustment was made to total wealth).

The estimates of domestically-owned assets of incorporated and unincorporated business, whether domiciled in Australia or abroad, presented here consist of five components. These are: non-rural equipment and inventories; non-rural non-dwelling construction (NDC), including the land on which the structures are built; rural business assets; domestic ownership of business assets domiciled abroad; and (minus) foreign ownership of business assets domiciled in Australia.

(i) Non-rural equipment and stocks

The first step was to calculate the ratio of the market value to historical value of the assets of a sample of large, non-financial, non-mining companies.[5]

The second step was to multiply this by ABS estimates of the historical value of all equipment (whether held by incorporated or unincorporated enterprises but excluding that held by the rural sector) plus the historical value of all non-rural inventories.[6] There are two implicit assumptions in this procedure. The first is that the ratio of market to historical values for unincorporated enterprises is the same as that for corporations. The second is that the sample companies' holdings of equipment are representative of the total business sector's holdings.

These assumptions seem reasonably plausible. However, the sample companies' holdings of NDC are likely to under-represent commercial property. For this reason, the aggregate market value of NDC is calculated separately as follows.

(ii) Non-rural NDC

There are major differences across industry categories in both the type of NDC held and the value of the associated land. For this reason, the aggregate market value of NDC was derived as the sum of that for three categories: banking, finance and property; retail and wholesale; and all other non-rural.

The holdings of NDC of the sample used above was assumed to be representative of the holdings of the second and third categories. Hence, the sample ratio of market to historical values referred to above was used to calculate the total market value of NDC for each of these categories. For banking, finance and property, however, a disproportionately high amount of the holdings of NDC is in the form of office buildings. For this reason, the sample ratio used above is inappropriate and another ratio, based on a sample of companies which hold office buildings almost exclusively was used.[7] This then gives three component estimates of the market value of NDC but with each excluding the value of the land on which the structure is built.

This will obviously lead to an underestimate of the value of non-dwelling construction and will underestimate the share of business wealth in the total estimates. Hence each of the components needs to be grossed up by an estimate for land.

Obtaining a value for land proved difficult and any attempt at incorporating it is somewhat arbitrary. Data were obtained from the NSW Valuer-General's (VG's) Department on site values and rental values for representative office, industrial and retail buildings in Sydney, Newcastle and Wollongong. These data were used in conjunction with data on yields on office, industrial and retail buildings in Sydney obtained from Baillieu, Knight, Frank.[8] Using the data on yields and rent, a price for each structure was calculated and the site value as a proportion of this price was obtained. A simple average for all the areas listed in the VG's Department report was then calculated.[9]

In 1989, land represented 26 per cent, 33 per cent and 31 per cent of the value of the property for retail, office and industrial structures respectively. These are proportions of the market value and hence should be added to the market value of non-residential construction. It was assumed that office buildings were representative of the banking, finance and property sector, retail of the retail and wholesale sector and industrial for the remainder in grossing up the value of non-dwelling construction to incorporate land. These proportions were assumed to hold throughout the sample period.

The total market value of non-rural equipment, inventories, NDC and associated land can thus be written as:

(Total equipment and inventories at historical cost)t * sample 1 ratiot
+ (Retail and wholesale NDC at historic cost)t * sample 1 ratiot * (1/(1–0.26))
+ (“Other” NDC at historic cost)t * sample 1 ratiot * (1/1–0.31))
+ (Banking, finance and property NDC at historic cost)t * sample 2 ratiot * (1/1–0.33))

(iii) Rural wealth

In the rural sector, nearly all enterprises are unincorporated and of a completely different nature to those covered by the samples. In this case, Piggott's approach of estimating rural wealth separately was followed. Piggott's estimates of rural wealth are used and are updated from information from the “Farm Surveys Report” carried out annually by ABARE.[10] Rural wealth is calculated as the product of average farm capital and the number of farms in each sector.[11] The series that can be constructed from recent surveys is not directly comparable to that calculated by Piggott, so growth rates are used to update the Piggott series.[12]

(iv) Domestic ownership of foreign business assets

This series should be added to total business wealth as the profits from these assets are recorded in the national accounts as domestic income. The series used is calculated from the ABS publication Foreign Investment, Australia.

(v) Foreign ownership of domestic business assets

Finally, foreign ownership of domestic business assets needs to be deducted from the total value. This series is obtained from the same ABS release referred to above.

Summing the first four categories and subtracting the fifth gives an estimate of the total market value of business assets. The figure of $525 billion in June 1990 compares with the Treasury's estimate of $229 billion.[13] There are four reasons why the estimate here is higher. Firstly, it includes a value for land in non-residential construction. Secondly, property held by the banking, finance and property sector is valued from a sample of companies which hold a high proportion of office buildings rather than from the non-mining, non-finance sample. Thirdly, rural wealth is estimated separately rather than from the non-mining, non-finance sample. Fourthly, Australian holdings of overseas assets are included in business wealth.

While the estimate presented here is substantially higher, it should be noted that it is more consistent with national accounts estimates of gross domestic product. The ratio of business wealth to dwelling wealth is 0.71, compared to 0.28 in the Treasury's estimates. If one thinks of GDP as the return on wealth (including human wealth), business wealth should be about 2.5 times as large as dwelling wealth. This follows from the observation that the gross operating surplus of corporate and unincorporated business is about 2.5 times as large as imputed and actual rent, which can be thought of as the return on the dwelling stock. It may be that the preferential taxation of owner-occupied housing causes its required pre-tax rate of return – as reflected by national accounts rent data – to be much less than the required pre-tax rate of return on business assets. But it is difficult to believe that this fully reconciles the observed multiple with the implicit (national accounts) multiple of 2.5. Hence, it is likely that, despite being large relative to previous estimates, the estimate of business wealth presented here still understates its true share of total wealth.

(d) Non-Official Holdings of Government Bonds

As noted by Piggott, it is debatable whether the private sector considers holdings of government bonds as net wealth (see Barro (1974)). However, to be consistent with previous authors an estimate is provided.

The private sector has two main claims on the public sector; their holdings of Commonwealth Government securities (CGS) and their holdings of Local and Semi-Government securities. Data on CGS at face value classified by holder are available in the Reserve Bank Bulletin. Private holdings at face value are calculated as total holdings less holdings by the RBA, major commonwealth trust funds and other public authorities. To obtain an estimate of the market value of this stock a market/face value ratio needs to be applied. It is assumed that this ratio is a constant of 0.9 for ease of calculation.[14]

Data on Local and Semi-Government securities at face value are available on an annual basis from the Reserve Bank Bulletin.[15] The holdings of securities (as opposed to borrowings) are again taken to a market value by assuming a market/face value ratio of 0.9.

Finally, foreign ownership of government bonds are deducted. Foreign ownership is taken from the ABS publication Foreign Investment, Australia.

(e) Notes and Coin in Circulation

This represents private sector holdings of currency. It is calculated as the sum of notes and coin in circulation (notes and coin on issue less holdings by the RBA).

Footnotes

The stock of outstanding private dwellings in any period is equal to the previous stock plus the number of new dwellings completed plus sales to the private sector by housing authorities less demolitions. Data on sales by housing authorities and on demolitions are not readily available. Instead, they are derived as a residual to keep the census data consistent with the completions data, i.e. the residual is calculated to keep the June 1981 census estimate of the number of existing dwellings plus the number of completions over the period consistent with the stock of dwellings estimated in the June 1986 census. The residual is assumed to be constant each quarter and to apply outside the June 1981 to June 1986 period. [1]

The use of a median rather than a mean dwelling price will bias downwards the estimate of the value of the stock of dwellings. However, a time series on mean dwelling prices is not available as far as I know. Piggott's estimates used a mean price but this covered prices only in Sydney, Melbourne, Adelaide and Brisbane. Conversations with the sources he quoted suggest that such a series could no longer be constructed. The extent of the bias depends on the skewness in the distribution of dwelling prices. The more skewed to the left the distribution is, the greater the mean relative to the median. Yates (1991), using information from the 1988/89 Household Expenditure Survey (HES), calculated the mean value of owner-occupied dwellings at $136,367, while the median was $100,000. [2]

Yates (1991) has made use of information contained in the 1988/89 Household Expenditure Survey to make an estimate of housing wealth. She estimated owner-occupied housing wealth to be $537 billion in 1988/89. To obtain a figure for total housing wealth this needs to be grossed up by the proportion of non-owner occupiers. The HES estimates the owner-occupation rate to be 72.7 per cent. Using this figure gives an estimate of $739 billion for the value of the total dwelling stock. As she points out, a number of assumptions need to be made to obtain this figure. The most important is that the quality of the owner-occupied stock is the same as that of the rental stock. It is more likely that the owner-occupied stock is of a better quality and better located than the rental stock, implying this estimate is overstated. It is difficult to make a direct comparison of the figures presented here and the Yates' estimate because the HES is compiled over a whole year, not at any particular point in time. Assuming the HES applies to June 1989 then the estimate here is lower by some $40 billion. If an average of the June 1988 and June 1989 estimates are used, the estimate here is $125 billion lower than Yates' estimate. An alternative way to gross up the estimate of owner-occupied housing wealth is to use the national income estimates of imputed and monetised rentals. This gives an estimate for the value of the total dwelling stock of $707 billion. This is only $10 billion higher than the estimate made here for June 1989. This method should overcome the problem of assuming that the owner-occupied and rental stocks are of the same quality. (I would like to thank John Piggott for suggesting this method to me.) [3]

Note that durables are depreciated whereas the housing stock is not. This is because the housing price reflects the price of the existing, depreciated stock whereas the durables price is for new durables only. [4]

The market value is calculated as the market value of equity plus the book value of debt. The historical value is calculated for each company in the sample by subtracting the asset revaluation reserve from the book value of the capital stock.
The market value of this sample forms the numerator of Tobin's q as calculated by Dews (1986). Financial and mining companies are excluded from the sample because their asset portfolios are significantly different. The former hold a disproportionately high amount of property and financial assets. The latter have large investments in exploration and it is not clear how these should be treated in capital stock estimates. [5]

Data on the historic cost capital stock were obtained from the ABS. For inventories it was assumed that the historic cost is equal to the book value. [6]

This sample consisted of 10 property trusts in 1990 although unfortunately only 4 of these existed in 1980. [7]

The yields obtained were for prime office, industrial and retail buildings in Sydney. They may therefore not be applicable to all the areas covered by the Valuer-General's Report. [8]

No information is available on the number of structures in each area so a weighted average could not be calculated. [9]

This survey is published annually. Quarterly observations are linearly interpolated from the annual ones. [10]

ABARE estimates of farm capital include land. [11]

Piggott used the results from the Bureau of Agriculture survey of the rural sector. This survey is now published by ABARE. However, the coverage of the survey has changed and hence estimates calculated from recent surveys are not comparable to those from the earlier years. This is why growth rates are used to update the original estimates made by Piggott. In 1988/89, ABARE changed the method of valuation of farm properties. Prior to this, they had used valuations from the Commonwealth Development Bank. Now they use the farmer's own valuation. This may cause a step change in rural wealth in June 1989. [12]

See Treasury (1990). [13]

Piggott actually calculated the market/face value ratio from the outstanding stock of CGS. This series has not been updated beyond June 1987. The ratio of 0.9 used here is the average over the period June 1984 to June 1987. The market/face value ratio varies inversely with the yield so 0.9 may be an overstatement of the ratio in times of high interest rates. [14]

Quarterly observations are linearly interpolated from the annual ones. [15]