5342.0 - Balance of Payments Statistics, Information Paper on Quality , 1996  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 20/02/1996   
   Page tools: Print Print Page Print all pages in this productPrint All  
Contents >> Chapter 2. Analysis of accuracy of the statistics >> Contributions to the balancing item

2.32. It is unlikely that any single factor accounts for the size and direction of Australia's balance of payments balancing item in the most recent three or four years, but some possible contributory elements can be suggested for these estimates, for the significantly increased value of the balancing item since 1983-84 and for the bias in initial estimates and early rounds of revision of the balancing item.

2.33. Since October 1983 the progressive relaxation of controls on international transactions, the associated broadening and internationalisation of the Australian financial system, and the introduction of new financial instruments and methods of financing, have made it difficult to maintain coverage of international investment transactors and the transactions that should be recorded in the capital account. The impact of this liberalisation can be seen in the growth of the turnover on the Australian foreign exchange market from just over $1 billion a day average in 1983-84 to its current level of about $50 billion a day.

2.34. The volumes of transactions underlying the net capital transactions data that are presented in the balance of payments are now huge. For example, the identified gross capital flows in each of 1992-93 and 1993-94 were of the order of $1,150 billion (see Table 2). And as discussed above, these measures still obscure an enormous volume of transactions that are only reported or calculated on a net basis. As shown in Table 2, the balancing item for 1992-93 represents less than 0.1% of the value of the identified gross capital transactions, and a much smaller proportion in the following year. Transactions of single enterprises are often volatile and sometimes very large. Ensuring adequate coverage of transactors and correct reporting of relevant financial assets, liabilities, capital transactions and income in these circumstances is a major challenge.

2.35. Problems have emerged in the measurement of transactions associated with foreign exchange trading, and the reporting in Australian dollar terms of international capital flows, particularly those involving banks. The nature and extent of these deficiencies are under investigation. It is apparent that some resident data providers are reporting significant volumes of transactions with non-residents in foreign currency denominated claims at hedged exchange rates and interest rates where the hedge has been entered into with a resident bank. Such hedges between residents should not be taken into account in recording transactions in the balance of payments. To the extent that misreporting of this kind occurs and as interest and exchange rates move up and down, entries in the capital account for the two sides to these capital flows will not be equal and offsetting. The resulting discrepancies could be significant.

2.36. Problems have also emerged in the way capital transactions are approximated from stock data reported at face value. In times of volatility of exchange rates or rapidly changing interest rates, such as in the late 1980s and early 1990s, this methodology may well result in either understatement or overstatement of capital transactions. This issue is also under investigation.

2.37. Some newer types of financial instruments, particularly those known as derivatives, also present measurement problems. Where foreign positions in derivatives are used to hedge other foreign assets and liabilities, the net impact should be captured (indistinguishably) in ABS surveys and reflected in balance of payments estimates. However, the impact of trading in these instruments for other than hedging purposes is not currently covered in the balance of payments accounts, due to the difficulty transactors are having in providing these very complex details to the ABS. This undercoverage will be progressively addressed as data providers improve their capacity to report these transactions to the ABS.

2.38. In addition to the difficulties of capturing all transactors and transactions, and the specific issues mentioned above, the potential for inconsistencies in the time of recording and valuation of the two different sides of a transaction has increased in recent years with the rapid growth in the volume and complexity of international transactions and more frequent exchange rate changes.

2.39. As mentioned earlier, the quarterly balancing item series displays fluctuations that are likely to be driven to some extent by timing errors in the accounts. A further factor is the large degree of estimation involved in quarterly methodologies for measuring some items in the accounts, such as portfolio investment in Australian securities. The ABS is currently investigating several options for improving quarterly data sources and methodologies.

2.40. Of the current account components, measurement of services debits is particularly difficult. This is because the services purchased from non-residents are diverse and may be difficult to separately identify. The resident transactors are spread widely across all sectors of the economy and, because they may have transactions only infrequently, are also not easily identified. Although substantial progress has been made recently in improving the measurement of services transactions generally, it is likely that coverage of services debits is incomplete. Difficulties in identifying and measuring some capital account transactions are also likely to be reflected in associated investment income measures. Some investment income payable abroad (particularly in the form of discount income) may be unrecorded.

2.41. It is useful to also consider sample error when reviewing the balancing item. In 1993-94, it is estimated that the sample error on the balancing item was $481 million (see Table 1); this represents 0.04% of total balance of payments transactions in that year. This means that due to sample error alone there is one chance in three that the balancing item will be more than 0.04% of total balance of payments transactions and nineteen in twenty chances that it will be up to 0.08% of that aggregate. Therefore, a balancing item within 0.08% of total balance of payments transactions can come about purely from the use of current sampling procedures. As many observations in Table 2 lie outside that range, it would appear that non-sampling error (as one would expect) also affects the accuracy of the balance of payments accounts.

2.42. On a quarterly basis, large swings can occur in the balancing item from one quarter to the next. It is not unusual for the balancing item for one quarter to have a large negative value and for the next quarter to have a large positive value. Even when the sign of the balancing item remains the same for two consecutive quarters, movements of over $2,000 million are not unusual. Swings in the balancing item from positive to negative and vice versa, may indicate that there are timing errors in the accounts, i.e., one side of a transaction may be measured in one period and its offset in another. For example, with tens of billions of dollars in international capital flows being transacted every week in Australia, the adoption of slightly different accounting cut-off dates by transactors when reporting to the ABS could see the two sides of some transactions reported in different reference periods.

2.43. It is difficult when viewing the original quarterly balancing item series to identify whether the volatility in the series is offsetting in either the short term or long term. A useful way to examine swings is to apply a seven-term Henderson moving average to the balancing item. This has been done in Graph 1, which shows the original quarterly balancing item and a derived trend series. The graph shows that there are sharp quarterly swings in the original balancing item series but that the trend series appears to move through cycles of about six or seven quarters when a positive balancing item predominates followed by a similar period during which a negative balancing item predominates. Further analysis would be needed to establish whether there is any relationship between these swings and movements in other economic variables such as interest rates and exchange rates, but there does seem to be a cyclical component.

2.44. An important aspect of the behaviour and scale of the quarterly balancing item revealed in the graph is the increased magnitude of the series from the mid-1980s since financial deregulation. This behaviour points to the measurement of international transactions in financial items as the probable area of largest error in the statistics.

GRAPH 1: QUARTERLY BALANCING ITEM






Previous PageNext Page