The publication of the Wholesale Price Index (WPI) for August 2020, which was due on 22 September, was deferred while the aggregation methods in WPI were reviewed, when the CSO found technical issues associated with the processing of the data. This review has been completed and the revised data for base year 2015=100 were published today (22 October 2020) in the WPI release and on PxStat. The revisions to the data have impacted the reference period from January 2015 to September 2020.
Price information for the WPI is collected in a monthly inquiry of manufacturers and wholesale outlets. Approximately 7,000 monthly price quotations are collected from a panel of some 1,300 respondent firms.
Four issues were identified during the review. These are outlined in Table 1 along with the changes to the aggregation procedures.
Table 1: Scope of the WPI review
|
|
Problem |
Solution |
|
Observations with weights but no prices
|
Some observations were identified with a weight but no price information. The weights in the WPI reflect the relative importance of products or industries. However, each weight should have corresponding price information. This problem caused the affected indices to be lower than they should have been.
|
A process change was implemented to exclude these observations from the calculations. |
|
Price records with prices but no weights
|
Some observations were also identified with price information but no weights. This means that those observations were excluded from the aggregation in the relevant time period, by being assigned zero weight, where it would have been preferable to include the observations.
|
A new process was implemented to ensure that weights were assigned (to the greatest extent possible) to the collected price information.
|
|
Application of the Laspeyres index formula |
At the two lowest levels of aggregation for WPI, monthly price relatives are calculated by taking a weighted arithmetic average of the monthly change in product prices. This calculation takes place, firstly, for similar products within companies and then, subsequently, for similar products across companies. These monthly price relatives are then used at higher levels of aggregation to produce the published price indices. While the Laspeyres formula was applied correctly at the higher levels of aggregation, it had not been applied correctly at the lower levels. In general, this caused the affected indices to be higher than they should have been. Further information on the application of the Laspeyres formula in WPI is included in Appendix 2.
|
A correction was introduced to apply the Laspeyres index formula uniformly at each level throughout the aggregation process. |
|
Data entry error |
In an isolated case in September 2018, price information for a small number of products was incorrectly applied to other products.
|
This error was corrected. |
Impact of the changes:
The overall Manufacturing Output Price Index for the most recently published month of July 2020 (base year 2015=100) was 86.2. This figure is being revised marginally to 85.8 (see Table 2). The annual % change for July 2020 for the Manufacturing Index was revised from -8.4% to -10.7%.
Figures 1 and 2 (see Appendix 1) compare the revised and previously published data for the overall Manufacturing Wholesale Price Index from January 2015 to September 2020. As can be seen, the indices show similar trends up until late 2017. However, the revised index diverges from the previously published index thereafter.
Table 2: Manufacturing Industries Output Price Index (Base Year 2015=100), Previously Published and Revised Indices
|
|
Manufacturing Output Price Index |
|
|
|
|
Previously Published |
Revised |
Difference |
|
March 2020 |
90.6 |
91.4 |
0.8 |
|
April 2020 |
89.0 |
88.7 |
-0.3 |
|
May 2020 |
88.1 |
88.0 |
-0.1 |
|
June 2020 |
87.0 |
86.6 |
-0.4 |
|
July 2020 |
86.2 |
85.8 |
-0.4 |
|
August 2020 |
.. |
84.5 |
.. |
|
September 2020 |
.. |
84.9 |
.. |
Further data requests or statistical queries:
It should be noted that the scale of the revisions depends on the reference period or sector being analysed. The revised data at the most detailed level are available on CSO’s PxStat. While the previously published data are no longer on the CSO website, this data can be made available to users on request. Please contact the WPI team directly, using the email address below, for further data requests or statistical queries.
| Previously Published | Revised | |
| 2015M01 | 96.6 | 97 |
| 2015M02 | 98.6 | 98.8 |
| 2015M03 | 100.9 | 101.1 |
| 2015M04 | 102.2 | 102.4 |
| 2015M05 | 100 | 100.2 |
| 2015M06 | 99.6 | 99.8 |
| 2015M07 | 99.6 | 99.8 |
| 2015M08 | 100 | 100.2 |
| 2015M09 | 98.8 | 98.9 |
| 2015M10 | 96.6 | 96.8 |
| 2015M11 | 103.6 | 102.7 |
| 2015M12 | 103.4 | 102.4 |
| 2016M01 | 101.2 | 100.2 |
| 2016M02 | 100.9 | 99.8 |
| 2016M03 | 99.9 | 98.9 |
| 2016M04 | 99.3 | 98.2 |
| 2016M05 | 100.2 | 99 |
| 2016M06 | 99.3 | 97.9 |
| 2016M07 | 99.7 | 98.2 |
| 2016M08 | 100.2 | 98.7 |
| 2016M09 | 97.9 | 96.5 |
| 2016M10 | 99.2 | 97.7 |
| 2016M11 | 103.3 | 101.4 |
| 2016M12 | 104.1 | 102.3 |
| 2017M01 | 102.7 | 100.8 |
| 2017M02 | 101.9 | 100.1 |
| 2017M03 | 104.5 | 102.4 |
| 2017M04 | 104.9 | 103.5 |
| 2017M05 | 105 | 103.5 |
| 2017M06 | 101.9 | 100.3 |
| 2017M07 | 100.8 | 99.2 |
| 2017M08 | 99.6 | 98 |
| 2017M09 | 99 | 97.3 |
| 2017M10 | 92.1 | 90.6 |
| 2017M11 | 97.8 | 92.2 |
| 2017M12 | 104.6 | 97.5 |
| 2018M01 | 103.6 | 96.4 |
| 2018M02 | 103 | 95.8 |
| 2018M03 | 101.5 | 96.2 |
| 2018M04 | 100.7 | 95.8 |
| 2018M05 | 101.5 | 97.7 |
| 2018M06 | 101.7 | 99 |
| 2018M07 | 100 | 98 |
| 2018M08 | 98.7 | 97.7 |
| 2018M09 | 96.4 | 97.6 |
| 2018M10 | 90.2 | 92.3 |
| 2018M11 | 93.7 | 95.1 |
| 2018M12 | 94.2 | 95.3 |
| 2019M01 | 94.4 | 96.5 |
| 2019M02 | 93.3 | 95.9 |
| 2019M03 | 94.6 | 96.6 |
| 2019M04 | 94.8 | 96.7 |
| 2019M05 | 95.1 | 97 |
| 2019M06 | 94.8 | 96.6 |
| 2019M07 | 94.1 | 96 |
| 2019M08 | 94.4 | 96.5 |
| 2019M09 | 94.8 | 96.9 |
| 2019M10 | 86.7 | 90.3 |
| 2019M11 | 91 | 93.3 |
| 2019M12 | 91.2 | 93.4 |
| 2020M01 | 90.3 | 92.4 |
| 2020M02 | 92.6 | 94.6 |
| 2020M03 | 90.6 | 91.4 |
| 2020M04 | 89 | 88.7 |
| 2020M05 | 88.1 | 88 |
| 2020M06 | 87 | 86.6 |
| 2020M07 | 86.2 | 85.8 |
| 2020M08 | 84.5 | |
| 2020M09 | 84.9 |
| Previously Published | Revised | |
| 2016M01 | 4.7 | 3.3 |
| 2016M02 | 2.3 | 1 |
| 2016M03 | -0.9 | -2.2 |
| 2016M04 | -2.8 | -4.1 |
| 2016M05 | 0.2 | -1.2 |
| 2016M06 | -0.4 | -1.9 |
| 2016M07 | 0 | -1.6 |
| 2016M08 | 0.2 | -1.5 |
| 2016M09 | -0.9 | -2.4 |
| 2016M10 | 2.7 | 0.9 |
| 2016M11 | -0.3 | -1.3 |
| 2016M12 | 0.7 | -0.1 |
| 2017M01 | 1.5 | 0.6 |
| 2017M02 | 1 | 0.3 |
| 2017M03 | 4.6 | 3.5 |
| 2017M04 | 5.6 | 5.4 |
| 2017M05 | 4.9 | 4.5 |
| 2017M06 | 2.6 | 2.5 |
| 2017M07 | 1.2 | 1 |
| 2017M08 | -0.6 | -0.7 |
| 2017M09 | 1.1 | 0.8 |
| 2017M10 | -7.1 | -7.3 |
| 2017M11 | -5.3 | -9.1 |
| 2017M12 | 0.5 | -4.7 |
| 2018M01 | 0.9 | -4.4 |
| 2018M02 | 1.1 | -4.3 |
| 2018M03 | -2.9 | -6.1 |
| 2018M04 | -4 | -7.4 |
| 2018M05 | -3.3 | -5.6 |
| 2018M06 | -0.2 | -1.3 |
| 2018M07 | -0.8 | -1.2 |
| 2018M08 | -0.9 | -0.3 |
| 2018M09 | -2.6 | 0.3 |
| 2018M10 | -2.1 | 1.9 |
| 2018M11 | -4.2 | 3.1 |
| 2018M12 | -9.9 | -2.3 |
| 2019M01 | -8.9 | 0.1 |
| 2019M02 | -9.4 | 0.1 |
| 2019M03 | -6.8 | 0.4 |
| 2019M04 | -5.9 | 0.9 |
| 2019M05 | -6.3 | -0.7 |
| 2019M06 | -6.8 | -2.4 |
| 2019M07 | -5.9 | -2 |
| 2019M08 | -4.4 | -1.2 |
| 2019M09 | -1.7 | -0.7 |
| 2019M10 | -3.9 | -2.2 |
| 2019M11 | -2.9 | -1.9 |
| 2019M12 | -3.2 | -2 |
| 2020M01 | -4.3 | -4.2 |
| 2020M02 | -0.8 | -1.4 |
| 2020M03 | -4.2 | -5.4 |
| 2020M04 | -6.1 | -8.3 |
| 2020M05 | -7.4 | -9.3 |
| 2020M06 | -8.2 | -10.4 |
| 2020M07 | -8.4 | -10.6 |
| 2020M08 | -12.4 | |
| 2020M09 | -12.4 |
Appendix 2: Laspeyres index formula
At the two lowest levels of aggregation for WPI, monthly price relatives were being calculated by taking a weighted arithmetic average of the monthly change in product prices. The following formula was being used:
where W0 are the base period weights; Pi are current period prices; Pi-1 are previous period prices.
However, this was not the correct application of the Laspeyres formula. The formula that was implemented as part of the revisions to WPI was as follows:
where Wi-1 are the previous period price updated weights.
Jillian Delaney (+353) 21 453 5258
Chris Moriarty (+353) 21 453 5447
