4.1.2  Comparison with Diewert and Lawrence productivity estimates

The productivity series constructed by Diewert and Lawrence (1999) used two databases. The first database was one that the authors had previously used for analysing the impact of tax changes in the New Zealand economy. The authors’ ‘preferred’ multifactor productivity series for the market sector of the New Zealand economy for the period 1972 to 1998 was constructed using this database (see Diewert and Lawrence, 1999, Table 1, ‘Diewert-Lawrence preferred’ series). Further details on this database are available in Appendix B Diewert-Lawrence Database (Diewert and Lawrence, 1999). The second database was provided by the Reserve Bank of New Zealand, the New Zealand Treasury, and the Department of Labour, and was designated the name Official Database by Diewert and Lawrence. The ‘Preferred Base Case’ multifactor productivity series for the market sector of the New Zealand economy for the period 1978 to 1998 (see Diewert and Lawrence, 1999, Table 1, ‘Preferred Base Case’ series) was formed using the Official Database.[9] Further details on the ‘Official’ Database are provided in Keegan (1998) and in Appendix C The ‘Official’ Database (Diewert and Lawrence, 1999).

The most comparable Diewert and Lawrence multifactor productivity series to the market sector multifactor productivity series presented in Figure 1 is found in Table 4.3 of Diewert and Lawrence (1999, p. 45, ‘Philpott’ Lives series). The comparable Diewert and Lawrence labour and capital productivity series to the market sector labour and capital productivity series presented in Figure 1 are found in Table 4.12 of Diewert and Lawrence (1999, p. 66, ‘Labour’ and ‘Net capital’ series). These series were constructed using the Official Database. The chained Fisher output index was formed using production GDP data for 20 industries comprising the market sector of the New Zealand economy. Chained Fisher input indices were formed using information on industry hours worked (drawn primarily from the HLFS, Quarterly Employment Survey (QES) and the Economic Survey of Manufacturing (ESM)), and industry net capital stocks for plant and equipment and building and construction, weighted by user costs with industry specific depreciation rates and asset lives based on Philpott (1992).

Figure 4 – Comparison with Diewert and Lawrence market sector productivity series

Figure 4 compares the Diewert and Lawrence productivity series constructed using the ‘Official’ Database with the productivity series presented in Figure 1. Because the Diewert and Lawrence series have a base of 1 in 1978, it was necessary to rebase these series to unity in 1988 to aid comparison with the productivity series shown in Figure 1. Table 3 reports average growth for the Diewert and Lawrence productivity series for the period 1988 to 1998 and for the sub-periods 1988 to 1993 and 1993 to 1998.

Source: Diewert and Lawrence (1999)

Note: New Zealand market sector estimates are in parentheses

In general the Diewert and Lawrence productivity series are quite similar to the market sector productivity series presented in Figure 1. Table 3 shows that over the period 1988 to 1998 average growth in the Diewert and Lawrence multifactor productivity series was 0.10% per annum higher than average growth in the multifactor productivity series shown in Figure 1. In contrast, average labour productivity growth was 0.21% per annum higher in the Diewert and Lawrence labour productivity series over the period 1988 to 1998. Average growth in the Diewert and Lawrence capital productivity series over the period 1988 to 1998 displays higher growth than the capital productivity series shown in Figure 1. Moreover, the two series appear to be diverging from the mid-1990s.

Given the improvements in the National Accounts data since Diewert and Lawrence (1999) undertook their study, especially the introduction of productive capital stocks estimates, the productivity series reported shown in Figure 1 are likely to give a more accurate picture of New Zealand’s productivity performance. However, this comparison highlights the impact that data upgrades and revisions can have on measured productivity.