Appendix E Comparison with Statistics New Zealand's regional GDP Figures
Appendix Table 5 contains a comparison of regional GDP per worker estimates derived from Statistics New Zealand's Regional GDP Feasibility Study (Statistics New Zealand (2007c)) and Business Demography data downloaded from http://www.stats.govt.nz/products-and-services/table-builder/table-builder-business.htm. The comparison is for the 2003 year.
The LBD data used in the current paper captures Value Added of $76.6bn, which is about 60% of the $130.7bn in the more complete regional GDP measure. The LBD employment measure (rolling mean employment, including working proprietors) is larger than the employee count measure obtained from Business Demography data. A consequence of the lower value added and higher employment figures is that value added per worker, as used in this paper, is significantly lower than GDP per worker ($39,900 compared with $78,900).
Regional variation in productivity is similar across the two series, with the exceptions of Wellington and Auckland. The reasons for the Wellington difference include the omission of public sector firms in the current study. The reasons for the Auckland difference have not been fully explained. The sources and methods for the regional GDP estimates are documented in Statistics New Zealand (2007b).
Some of the methods used differ from that used in the current paper and these differences may account for some of the discrepancy. One potentially influential difference is in the way that value added is allocated across plants within multi-plant enterprises. The method is similar to the allocation outlined in Appendix B, with the difference that instead of setting value added per worker to be constant within an enterprise, the Regional GDP figures are based on setting value added per dollar of payroll to be equal. This has the effect of generating an estimated VAPW premium that is equal to the Auckland wage premium
In the current paper, the spreading of value added in proportion to employment yields a zero VAPW premium. However, the recovery of the underlying premia is achieved by regression methods, as outlined in sections 4.1 and 4.2. A similar approach could be applied in the creation of the Regional GDP estimates to generate estimates of the relative productivity rather than average wage rates.
Appendix Table 4 shows a hypothetical example of two enterprises to illustrate the impact of the different methods. Assumptions are made about the wage level (20% Auckland premium) and VAPW level (100% Auckland premium) and about the proportion of each enterprise's employment that is in Auckland. The first 3 columns of the Enterprise A case show the assumed Employment spread, and the wage bill and Value added amount implied by the assumptions.
If only total value added of $30,000 were observed, it needs to be allocated to Auckland and non-Auckland plants. The 4th and 5th columns show the split of total value added in proportion to employment, the implied value added, and the recovered premium of 0%. The 6th and 7th column show a similar calculation based on splitting value added in proportion to wage bill. The recovered premium is 20% - equal to the assumed wage premium.
The two simultaneous equations at the bottom of the table show how information is used from two enterprises (that differ in the proportion of their employment that is in Auckland) to identify the underlying VAPW premium. Our regression-based adjustment performs the same role, although with more enterprises than locations, so that the identification is not exact.
Appendix Table 4 Regional GDP v Value added per worker
→ Full size version of Appendix Table 4 [35 kB GIF]
Appendix Table 5 Regional GDP v Value added per worker
→ Full size version of Appendix Table 5 [58 kB GIF]
Sources: Regional GDP panel: (Statistics New Zealand (2007c)) and Business Demography data downloaded from http://www.stats.govt.nz/products-and-services/table-builder/table-builder-business.htm. The Auckland premium expresses "Auckland Premium" columns express GDP/Emp and VA/Emp as a proportion of non-Auckland levels
Back to Top