Contents >>
Appendix 3. Comparison of initial and latest periodonperiod movements
1. This appendix contains scatter graphs based on the movement in the estimate from one period to the next for each of the principal balance of payments aggregates. Movement in an estimate represents the change in the aggregate from the previous month or quarter. Points plotted on a graph represent the intersection of the initial estimate of the movement for a period with the latest estimate of the movement for that period. Each + point plotted on the graphs therefore represents two estimates of movement between two periods. The initial estimate of movement is measured on the horizontal axis and the latest estimate of movement on the vertical axis. Graphs are included for monthly statistics for the period from February 1986 through June 1994 and for quarterly statistics from the June quarter 1986 through June quarter 1994.
2. It can be seen that a periodonperiod movement that has not changed between the initial estimate of that movement and the latest estimate of the movement will lie on an imaginary 45° line drawn from the bottom left corner of a graph to the top right corner passing through the origin. Points below the line correspond to periods for which the latest estimate of the periodonperiod movement is less than the initial estimate of the movement in the case of increases or more than the initial estimate of movement in the case of decreases. Conversely, points above the 45° line correspond to periods for which the latest estimate of the periodonperiod movement is greater than the initial estimate in the case of increases or smaller than the initial estimate in the case of decreases.
3. Points that lie in the top left or bottom right quadrants of the graph correspond to periods for which the direction of movement has changed between the initial estimate and the latest. That is, an initially positive periodonperiod movement has changed to a negative movement or vice versa. Prima facie, initial data for these periods can be considered the most likely to present the user of the statistics with misleading information.
4. The extent by which a periodonperiod movement has changed can be gauged by the distance a point lies from the 45° line. Subjective judgements have been made on what amount of change is likely to result in a user drawing the wrong conclusions based on initial estimates of periodonperiod movement. Bounds of these dimensions above and below the 45° line, roughly equivalent to two standard deviations or less, are shown on each graph. Note that the value of these bounds vary from item to item. For example, while bounds of plus or minus $300 million have been used for the monthly balance on current account, for the quarterly series this has been increased to $700 million to reflect the larger movements likely for the aggregated time period, and therefore the less likelihood of interpretation error on the smaller movement revisions.
5. Graphs C3 and C4 provide an example of the interpretation of the graphs presented in this appendix. In Graph C3 for nonmerchandise credits, the observations are more closely clustered around the centre than for the nonmerchandise debits series shown in Graph C4, reflecting the much smaller scale of movements in the credit aggregates. Looking at Graphs C15 and C16 for the quarterly merchandise exports and imports series, respectively, shows series that are very tighly clustered, reflecting the generally low incidence of significant revision to movements in these series.
Graphs C.1 to C.33
The attached file contains a printable copy of pages 105 to 113 of the paper publication containing graphs C.1 to C.33.
Download this entire file for viewing
and printing in Adobe Acrobat format

File size: 758 Kbytes 
You have reached the end of publication 5342.0