Labour Tables

Owned by Philip Stevens
Dec 13, 2022
2 min read

Each year, Richard Fabling produces the “labour and productivity tables”.

The latest labour tables were released in December 2022. The productivity data will be updated once the current LBD refresh is finished.

The latest labour tables use the current (20211020) IDI. Points of note for this instance are:

  • Enterprise links are now being imputed for EMS months not processed by LEED. In the current instance, this means that the labour tables include up to 202106 (202107 is only a partial EMS month, so has been excluded). The imputation method carries over employer IR-ENT links from the prior month for any employer that has no links in the current month. Processing is sequential so that repaired links in one month are then used in the following month, which becomes important in the final months of data.

Perhaps surprisingly, this process also substantially improves ENT link coverage in 202105 (the latest LEED processing month), which is unexpectedly short of links (around 100K missing, presumably due to timing issues between the time when the IR data was available and when LEED processing occurred), and in the period 201910-202002 where an unusually large number of ENT links are missing (for unknown reasons).

As usual repaired links appear in the table ent_pbn_repair_IDI_xxx_RFabling, which contains substantially more rows now that LEED embargo period months are repaired.

  • Tests show that the IR7 issues present last year have been resolved. The methodology continues to repair apparently missing WP-ENT links for partnerships, but this step now only affects a small number of observations in each year.

  • FTE and WP labour input counts by level one ethnicity have been added to a new IDI table pent_year_L_IDI_xxx_RFabling, which duplicates and extends the corresponding LBD table. Ethnicity is taken from the personal details table (ie, is a Stats NZ-derived permanent characteristic) with European/other ethnicities grouped.  Since ethnicity can be missing, the table also includes the total FTE/WP labour input with observed ethnicity (coverage exceeds 99% of aggregate L in all financial years except 200003, see attached "FIG ethnicity data coverage"). Note, because individuals can have multiple ethnicities, the sum of group L can exceed total L. Ethnicity-based variables are not replicated on the LBD because these variables are potentially disclosive of individual characteristics (eg, in sole proprietor firms). The attached spreadsheet demonstrates some simple applications of the data for those who are interested.