APPENDIX 5: COHERENCE BETWEEN PRICE INDEXES FOR IMPORTS AND EXPORTS OF GOODS

Introduction
The Behaviour of the Imports and Exports of Goods Price Indexes
Drivers of Difference
Findings


INTRODUCTION

1 This appendix examines whether the coherence can be improved between the three similar ABS (original) price indexes for imports and exports of goods.


THE BEHAVIOUR OF THE IMPORTS AND EXPORTS OF GOODS PRICE INDEXES

2 Published in the Australian National Accounts: National Income, Expenditure and Product (cat. no. 5206.0) and Balance of Payments and International Investment Position (cat. no. 5302.0) are:

• the Imports and Exports of goods Implicit Price Deflators (IPDs); and
• the Imports and Exports of goods Chain Price Indexes.

3 Published in the International Trade Price Indexes, Australia (cat. no. 6457.0) are the Import Price Index (IPI) and Exports Price Index (EPI).

4 These publications provide detailed Australian export and import price index information for analysts and represent the average change in the prices paid for imported products and the average change in the prices received for exported products. However, while they represent similar economic transactions, they differ in terms of concept, methodology and source data.

5 Graphs 1 and 2 below depict the three indexes by quarterly percentage change over the period September quarter 2001 to September quarter 2011. While the National Accounts and Balance of Payments generally use modified IPIs directly as the imports of goods chain price indexes, there are significant differences in the export price statistics. This is illustrated by the relatively small difference between the import price indexes (98% correlation), and the larger, persistent difference in the export price indexes (95% correlation).

6 Graph 1 shows that since September quarter 2001, the absolute average quarter on quarter percentage change difference between the EPI and the Balance of Payments chain price index has been 0.12 percent, and 0.01 percent between the EPI and the Balance of Payments IPD. Differences from the EPI in individual quarters have been as high as 6.1 percent for the Balance of Payments chain price index and 7.5 percent for the Balance of Payments IPD. These maximum differences occurred in the March quarter 2009.



7 Graph 2 shows that since the September quarter 2001, the absolute average quarter on quarter percentage change difference between the IPI and the Balance of Payments chain price index has been 0.02 percent, and 0.17 percent between the IPI and the Balance of Payments IPD. Differences from the IPI in individual quarters have been as high as 1.9 percent for the Balance of Payments chain price index and 3.3 percent for the Balance of Payments IPD. These maximum differences occurred in the June quarter 2009.



DRIVERS OF DIFFERENCE

8 The differences are attributed to the sum of the conceptual, methodological and data source differences summarised in Table 1 below.

	International Trade Price Indexes(a)	National Accounts and Balance of Payments, Export and Import of goods Chain price indexes	National Accounts and Balance of Payments, Export and Import of goods Implicit price deflators	International best practice(g)
Concept	Weighting concept: International Merchandise Trade (IMT) Statistics	Weighting concept: Balance of Payments Manual sixth edition (BPM6)	Weighting concept: BPM6	Weighting concept: BPM6
Methodology	Annually weighted chained Laspeyres price indexes (b) (non–revisable)	Annually weighted chained Laspeyres price indexes(e) (revisable)	Current weighted Paasche price indexes (revisable)	Annually weighted chained Laspeyres price indexes (revisable when required)
Price data source	Survey IMT (AUV) (c) (d)	Imports = IPIs(f) Exports = EPIs(f)  (30%) + IMT(c)(70%)	Derived	Survey
Quarter coverage	•Poor month 3 pricing •IMT(c) provisional	Provisional full quarter IMT(c) coverage	Full	Monthly indexes recommended

Footnotes
(a) ITPIs = Export price index (EPI) and Import price index (IPI)
(b) Export price index – weights based on the average over two years
(c) International Merchandise Trade (IMT) Average Unit Value (AUV); revisable for 6 months.
(d) Limited usage in ITPIs; apart from validation of survey data or, if consistent at the most detailed level, as a price observation
(e) Export goods chain price index – weights based on the average over one year
(f) Adjusted for BPM6 concepts
(g) Export and Import Price index manual (IMF 2009)


CONCEPT

9 As illustrated in Table 1 above, ITPI weighting is aligned with International Merchandise Trade Statistics (IMTS) concepts. To achieve the principal purpose, the ABS prefers the Balance of Payments and International Investment Position Manual, sixth edition (BPM6) concepts to those of the IMTS, as BPM6 aligns with the 2008 System of National Accounts (2008 SNA).

10 To convert the conceptual basis from IMTS to BPM6, several adjustments are made. The BPM6 reconciliation is outlined in Table 2 below.

Component reconciliation

International Merchandise Trade (Customs Clearance)

+ Coverage Adjustment

+ Timing Adjustment

+ Processing Adjustment

+ Classification Adjustment

= Goods on BPM6 / 2008SNA Basis

11 IMTS record the physical movement of goods across the customs frontier, whereas the BPM6 records transactions on a change of ownership basis. In order to convert IMTS to a BPM6 basis, adjustments are made:

• to ensure transactions are recorded in the period in which ownership changed, rather than in the period in which the transaction was recorded by the Australian Customs and Border Protection Service;
• to include goods that do not cross the customs frontier but do change ownership;
• to modify a good's value if additional information is available; and
• to move reported items to a more representative classification heading.

12 Examples where adjustments are made include the exports and imports of defence equipment such as ships and aircraft.


METHODOLOGY

13 Methodologically, the ITPI, National Accounts and Balance of Payments import and export chain price indexes are largely identical, with the EPIs differing only in terms of the annual weighting pattern. For the EPI, it is the average of the most recent two years, whereas for the National Accounts and Balance of Payments export of goods chain price index, it is the most recent (single) year.


DATA SOURCE

14 While the National Accounts and Balance of Payments generally use the IPIs (modified for the BPM6 concept) directly as the imports of goods chain price indexes, there are significant differences in the data sources used for export price statistics.

15 Compilation of the EPIs follow international best practice which recommends establishment surveys for data collection in preference to AUVs derived from IMT data. AUV data are not delivered at a detailed enough level to account for compositional shift (bias), and are not considered to be a robust data source for price index compilation.

16 The use of AUV data in National Accounts and Balance of Payments exports of goods chain price index calculation is significant. Approximately 75% of exported products volume measures are obtained using a method known as quantity revaluation. This involves calculating AUV as prices from detailed quantities (e.g. tonnes of coal) and current prices (e.g. dollar value of customs valuation for coal shipment) (see Australian National Accounts: Concepts, Sources and Methods, 2000 (cat. no. 5216.0)). A product level price index comparison under each approach illustrates that the differences are demonstrable and persistent. However, the use of AUV data ensures complete coverage of export transactions over the entirety of the quarter, and this is something the EPI cannot claim to provide.

17 The IMT data may be revised (for frequently exported commodities) up to six months after the end of the reference quarter. Significant revisions are apparent in the trade data for particular goods, which are taken into account within the National Accounts and Balance of Payments in subsequent releases.

18 EPIs do not make expansive use of IMT data and are not impacted greatly by revisions to these data. Comparison between of the EPI survey data with IMT AUV data show the correlation (of homogeneous products) markedly improves following the full six month IMT AUV revision period. This suggests the EPI survey data provides a more accurate estimate of price change, significantly earlier than is possible with AUV data.

19 The United Nations International Merchandise Trade Statistics: Concepts and Definitions, 2010 recommends that all countries publish on a monthly, quarterly and annual basis either price or unit value indexes for their total imports and exports of goods. Countries are also encouraged to calculate and publish such indexes for commodity groups of particular importance to countries at least quarterly and annually. The Organisation for Economic Co-operation and Development (OECD) is investigating including unit value and quantity indexes for imports and exports of goods on their International Trade by Commodity Statistics database. The OECD and several countries are exploring methods for improving the quality of these indexes. The ABS is not planning to invest in unit value and quantity indexes for imports and exports of goods as it prefers to concentrate on improving the ITPIs for goods and services.
FINDINGS

20 The main points to be taken from this investigation are that the concept, methodology and data source for ITPIs and import and export of goods chain price indexes can be aligned, while the import and export of goods IPD does not align with the chain price index methodology.