6413.0.55.001 - Experimental Price Indexes for Financial Services, Jun 2005  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 29/07/2005  Ceased
   Page tools: Print Print Page Print all pages in this productPrint All
  • Explanatory Notes


1. There are many services acquired by households that have the potential to be regarded as financial services. Some common examples are: financial advice; currency exchange; deposit and loan facilities; services provided by fund managers, life insurance offices and superannuation funds; stockbroking services; and real estate agency services. This experimental price index is restricted to those for which the ABS is able to construct reliable measures of price change, namely:

  • deposit and loan facilities provided to households by deposit-taking institutions; and
  • services provided by stockbrokers and real-estate agencies when households buy and sell shares and property.


2. This section describes the conceptual and procedural issues relevant to the construction of price indexes for deposit and loan facilities. Price indexes relating only to deposit and loan facilities are not available for separate publication. They are included in the aggregate series for financial services.

3. To construct an overall measure of price change for deposit and loan facilities there are two key factors that need to be addressed concerning the ways in which these services are charged for. Deposit-taking institutions earn income by lending funds at a higher rate of interest than they pay on deposits (the difference is referred to as interest-rate margins) and by charging explicit fees for account-keeping services and for some transactions.

4. To the extent that income from interest margins is obtained from households, it is legitimate to regard that amount as representing a payment by households for the services they obtain. The difficulty is that these interest-rate margins are not directly observable. Although the explicitly charged fees are observable, the effect of these on individual households can vary significantly depending on factors such as the type of account, the frequency of particular transaction types, the account balance, and the total volume of business that the customer conducts with the financial institution.

5. The challenge confronting the ABS was to develop a way of measuring changes in the amounts payable as explicit fees and charges, the amounts payable as taxes, and the amounts payable as interest-rate margins. The changes in all these elements then need to be combined to obtain the aggregate or net change in the prices of deposit and loan facilities.

Calculation of margins

6. The first problem to overcome was how to calculate the amounts paid as interest margins on any single product (or account provided by a financial institution). The 1993 System of National Accounts (SNA 93) recommends [6.125 and Annex III] that the value of services provided by means of interest-rate margins be valued as the product of the balance on the account multiplied by the difference between the interest rate payable or receivable and a reference rate of interest. In concept, SNA 93 describes this reference rate as being a risk free or pure interest rate. This approach has the effect of valuing the service provided to a borrower as the difference between the amount of interest paid by the borrower and the (lesser) amount that would have been paid had the reference rate been used. The converse applies for depositors. Therefore, the price of the service per dollar borrowed is given by the difference between the interest rate paid by the borrower and the reference rate. For a depositor, it is the difference between the reference rate and the interest rate received by the depositor.

7. In practice, statisticians experience great difficulty in identifying an exogenous reference rate that does not also result in volatile and sometimes negative measures of these costs (as would occur if the reference rate lies above the lending rate or below the deposit rate). When valuing these services in the national accounts the ABS has adopted the practice of setting the reference rate at the mid-point of the borrowing and lending rates. A reference rate determined in this way could be regarded as representing a market clearing rate (i.e. the rate that would have been struck in the absence of financial intermediaries by depositors dealing directly with borrowers).

8. To minimise problems with potential non-response and changes to the structure of the industry, ABS decided that a slight variation of the approach used in the national accounts would be adopted in the construction of the price indexes for financial services. A separate reference rate is calculated for each sampled institution for these experimental indexes.

9. It is important to recognise that the reference rate is not intended to approximate a financial institution's 'cost of funds'. In the simplest case, where an institution's only source of funds is amounts on deposit, its cost of funds would equal the interest rate paid on deposits. Using this as the reference rate would result in the measurement of zero services being provided to depositors.

Estimating base-period expenditure

10. Estimating the base period value of expenditure on deposit and loan facilities, as required for weighting purposes, is a complex exercise. What follows is a simplified description of the general procedure.

11. The starting point is to select a sample of deposit-taking institutions each of which is then approached to obtain information on balances, interest flows, fees and taxes by product and in aggregate for a full financial year. Interest flows (payments on deposit products and receivables on loan products) and balances are used to compute interest rates (or yields) for individual products and for deposits and loans in total. The reference rate is calculated as the mid-point of the rate paid on deposits and the rate earned on loans. The percentage margin on each product is calculated as described above and the dollar value of the margin computed. For all those products identified as being consumer products (as distinct from those used by businesses), the total receipts from households are computed by summing the margins, fees and taxes. The aggregate ratio of these receipts to total balances for the sampled institutions is applied to aggregate balances for all deposit-taking institutions to derive a national estimate. The capital city estimates are imputed by reference to aggregate data from ABS's Household Expenditure Survey.

Measuring price change

12. The schedules used to determine the amounts payable as explicit fees and taxes are generally not linear in nature and tend to have some form of step function. In other words, rather than setting a universally applied price per transaction, it is often the case that fees for certain types of transactions are only incurred after some threshold is reached (e.g. after say four transactions in a month or when account balances fall below some level). Furthermore, financial institutions often bundle products together, with the price paid for particular products depending on the bundling arrangements. In these circumstances it is not possible to calculate an average price from observable schedules. What would be required would be to price different bundles of each service, for example three, four, five, and six over-the-counter withdrawals and derive an average across the different bundles and construct a measure of price change from the changes in this average over time.

13. However, there are several problems with this approach. Obtaining enough detailed information to construct sufficiently representative bundles of individual services attracting fees would be an expensive process and still subject to error. It would also not be able to account for cases where the fees (or taxes) vary with the value of the transaction or individual account balances or cases where rebates are applied against the total of all fees that would have been charged. For these reasons, the ABS believes that the only way to measure reliably changes in fees and taxes is to work with a sample of customers' accounts.

14. The approach adopted is similar in principle to that used for components of the CPI. For each sampled institution a sample of products is selected to represent each of the major product categories such as current accounts, savings and investment accounts, retirement accounts, housing and home-equity loans, personal loans and overdrafts, and credit cards. The specific product selected from each group (e.g. the sampled home-loan product) is assigned a weight to represent all the specific fees and taxes paid on all housing loans provided by that institution.

15. For each sampled product the institutions are asked to provide a sample of accounts for customers living in each of the eight capital cities. The sampled accounts have had all identifying information (e.g. customers' names, addresses, and account numbers) deleted and replaced with a simple sequential number. Each sampled account contains information that is similar to that included on the monthly statements received by the customers. Thus the value of any transaction and the type of transaction (cheque, automatic teller machine (ATM) withdrawal etc.) and running balances are retained along with any other information that influences the determination of charges (such as a customer relationship value or a fee-waiving flag). Each sampled account covers a full twelve months' activity. All up, the ABS samples about 7,000 individual accounts for the purpose of compiling these experimental indexes. In aggregate these sampled accounts contain approximately three million transactions.

16. The ABS has built a computer system capable of emulating the charging regimes used by financial institutions and state revenue offices. This enables the ABS to calculate the total amounts that would be paid in fees and taxes over a full year for each sampled account based on the currently prevailing fee and tax schedules. The use of a full year's activity in this way is consistent with the method used to measure price change for all items in the CPI, which is based on an annual quantity basket. However, to preserve the real values of the individual transactions (or, in other words, the underlying quantities), the values of the transactions need to be updated each period to reflect changes in the general price level before applying the current period fee schedules. The all-groups CPI is used for this purpose.

17. The process can be described in the following way. Each quarter the transaction values in the sampled accounts are updated using a four-term moving-average of the CPI lagged by one quarter. The fee and tax schedules for each sampled product are updated, where necessary, each month and a total annual amount payable is calculated for each sampled account each month. The movements in these amounts are used to calculate an average movement in fees and taxes for each sampled product.

18. In addition, loan products sometimes incur an establishment or application fee. These are not accounted for by the sample of accounts and therefore require a separate estimation procedure. The price measure for establishment fees is calculated using changes in the average establishment fee charged on new accounts each month. The information required for these calculations is provided by the sampled institutions. Changes in these average fees reflect any fee discounting or waiving by financial institutions.

19. The same sources and methods used to measure the value of amounts paid as interest-rate margins are used to construct a measure of price change for interest margins. The sampled institutions provide monthly data on balances and interest flows by product and in aggregate. These data are used to calculate a current-period interest-rate margin for each of the sampled products. However, to minimise the effect of any short-term accounting anomalies such as posting effects and adjustments of various types, the ABS constructs three-month moving-averages of the average balances and interest flows and derives the required yields, reference rates and margin rates from the smoothed data. Because percentages (such as margin rates) are not prices, the margin rates for the latest period have to be applied to some monetary amount in order to compute the current period amounts that would be paid as margins. The price index is constructed by comparing the change over time in these amounts. As margins are payable on the account balances, the procedure followed is to update the base-period balances using a four-term moving-average CPI (as previously described for updating transaction values) and applying the calculated percentage margin for the current month. (In practice this means that if percentage margins remain unchanged over time, margin 'prices' will move in line with the lagged four-term moving-average of the CPI.) A price index for each sampled product is then obtained by combining the estimated movements in margins, fees and taxes. The aggregate index for deposit and loan facilities is produced by weighting together the indexes for each of the sampled products.


20. The index for real-estate-agency services is designed to cover the cost of those services acquired from real-estate agents by households when selling and buying property, including any government charges on property transfers. The services are not restricted to properties acquired by households from other sectors of the economy or to properties for owner-occupation. Therefore the number of these services is not the same as the number of dwellings underpinning the house-purchase component of the CPI.

Estimating base-period expenditure

21. Information about sales commissions and residential properties from Real Estate Services Industry, Australia, 1998–99 (cat. no. 8663.0) is used to derive expenditure estimates for each capital city by estimating from state data using aggregate expenditure information from the Household Expenditure Survey.

Measuring price change

22. Real-estate agents typically quote their fees as some percentage of the price received for the property. In common with other items where charges are set as a percentage of a value, this needs to be converted to a price expressed in dollars. If the percentage margin is known, the agent's price for any given transaction can be computed by multiplying the sale price of the property by the percentage margin. The index number problem decomposes into two parts — estimating the percentage margin, and estimating the value of the property transaction. The calculation of stamp duty on property transfers is identical in concept.

23. Historically, real-estate agents tended to charge commissions in accordance with rate schedules published by the state and territory Real Estate Institutes. The schedules were such that the average rates of commission tended to decline with increases in property values. For example, in September 1997, the recommended schedule for New South Wales was for a commission of 3.1% on amounts up to $100,000 and 2.0% on each dollar above that. Following industry deregulation these fee schedules are no longer published and rates are open to negotiation between real-estate agents and vendors. The state and territory governments publish schedules of stamp duties.

24. A method for estimating the fee payable for any given property transaction is required because the percentage margin charged by individual agents is known to vary depending on the value of the underlying transaction. The ABS does this by first of all conducting a quarterly survey of real-estate agents in each capital city. The respondents are asked to provide information for a maximum of six settled residential property sales (representative of sale prices in the agents' area) for each of the three months in a calendar quarter. For each transaction, the agent reports the sale price of the property and the total dollar amount of commission charged by the agent. The ABS uses ordinary least squares (OLS) regression techniques to estimate a relationship between property values and commission rates.

25. Although there are potentially many functional forms that could fit the data, empirical analysis has shown that the following functional form is adequate:
R = a + b1(1/p) + b2(1/p)2 where:
R = the commission rate;
p = the house price;
a = a constant; and
b1 and b2 are the parameters to be estimated.

26. Because of data volatility, the regression is run on data pooled from three quarters.

27. As agents' fees and rates of stamp duty depend on the value of the property, it is not sufficient to calculate these fees and charges by reference to some average transaction value. What is required is a representative sample of transactions. A sample of transactions can be drawn from the population of house sales in the base period and the amounts payable as agents' commissions and stamp duty calculated from the regression functions and tax schedules applying in the base period.

28. In subsequent quarters the base-period sample of transactions is revalued to current-period prices and the amounts that would be payable as fees and stamp duties are calculated by reference to the appropriate regression function and tax schedule. The most contentious issue in this process is the method used to revalue the sample of base-period transactions. The ABS is of the view that the purchase of property is best regarded as forgone consumption and therefore the transactions should be revalued in such a way as to preserve the base period command over consumer goods and services. Taking this view, the values of the sample transactions are updated using a four-quarter moving-average of the CPI (lagged one quarter) as is done for transaction amounts in the index of deposits and loans. The price index for real-estate-agency services is calculated by comparing the changes over time in the aggregate amounts payable as agents' fees and stamp duties.