The Treasury

Global Navigation

Personal tools

4.1 Labour input

Labour input could be expected to be relatively easy to measure compared with other parts of the productivity equation, particularly capital input or non-measured sector output. However, there are a number of issues with measuring labour input in both New Zealand and Australia. Indeed, challenges in deriving New Zealand labour input was a major finding of Diewert and Lawrence (1999).

The OECD (2001) recommends using actual hours worked as the measure of labour input:

“Notwithstanding some of the measurement issues, it is recommended that hours actually worked be the statistical variable used to measure labour input, as opposed to simple head counts of employed persons. Hours paid and full-time equivalent persons can provide reasonable alternatives. Significant differences in country practices for calculating hours worked and full-time equivalent persons persist, and raise issues of international comparability”.

As noted above, official productivity data from Statistics NZ are for hours paid and based on data from a number of sources, particularly business datasets. This composite measure of labour input (or labour volume) is used because industry data are needed to build up the measured sector. Industry-level data are also needed for the industry productivity series released recently. Hours paid data are judged by Statistics NZ as more robust at the industry level because they are sourced from business datasets (and thus are more likely to be industry classified in the same manner as for GDP, which is also constructed from business datasets) and have a longer time series than the HLFS.[10] Only at the economy-wide level is the annual change in hours worked from the HLFS considered as statistically robust as hours paid.

At the economy-wide level, no industry splits are needed, so the OECD uses hours worked from the HLFS to estimate growth in economy-wide labour input. As expected, the OECD and HLFS economy-wide measures match closely. The OECD data match best with the HLFS measure on a calendar-year basis.[11] In New Zealand, economy-wide labour input between the 1988 to 2008 calendar years grew by an average of 1.4 percent per annum on both the OECD and on HLFS measure. Table 3 below shows 1.2 percent for the OECD measure (1987 to 2007 calendar years) and 1.3 percent (1988 to 2008 March years) for the labour force survey so as to match table 2.

Official and labour force survey estimates of labour input for the measured sector differ moderately. In New Zealand, labour input between 1988 and 2008 grew by an average of 0.3 percent per annum in the official (former) measured sector and by an average of 0.6 percent per annum in the HLFS data for the same group of industries. The divergence between the two labour input measures is largest in the period from 1992 to 1995 (March years), when the HLFS measure grows more strongly coming out of the early-1990s recession. Though it is impossible to be definitive, there is some limited evidence that the redesign of the HLFS in 1993-94 overstated growth in actual hours worked over this period.

In contrast to New Zealand, official productivity statistics from the ABS for the market sector of the economy use a measure of labour input based on hours worked from its labour force survey. This is the same concept of labour input that is used at the economy-wide level by the OECD to estimate economy-wide labour productivity growth. The OECD economy-wide measure therefore matches the labour force measure. In Australia, although the time periods are slightly different (June versus calendar years), labour input between 1988 and 2008 grew by an average of 1.7 percent per annum on both the OECD and the ABS labour force measure. Australia has experienced issues with hours worked in the past (eg hours worked were surveyed at four points of the year but not adjusted for holidays). In Australia, labour input between 1988 and 2008 grew by an average of 1.1 percent per annum between 1988 and 2008 on the official market sector measure and on their labour force measure.

In conclusion, it appears that the choice of labour input does make a potentially important contribution to unravelling the puzzle over the varying pictures of productivity performance. The HLFS series in New Zealand overstates labour input growth (and thus understates labour productivity growth) relative to the measured-sector labour input series for the same group of industries by 0.3 percent points per annum. In Australia, the same approach is used in both the market-sector and economy-wide measures and so the issue does not arise. The New Zealand data are not wrong. The measured-sector estimates are “fit for purpose” as industry estimates are needed and this cannot be done with hours worked data, especially prior to 1996 when industry coding is less consistent. The issue arises when comparisons are made with other countries that have better hours worked data and so chose this as the basis of their official productivity measure, as Australia did. With the OECD using hours worked, New Zealand (but not Australia) runs into problems when comparing economy-wide and measured-sector labour input data.

Table 3 - Labour input growth from 1988 to 2008 in New Zealand and Australia
Annual average % changes New Zealand


OECD 1.2% (Dec) 1.7% (Dec)
Labour force survey 1.3% (Mar) 1.7% (Jun)

Measured/market sector

Official 0.3% (Mar) 1.1% (Jun)
Labour force survey 0.6% (Mar) 1.1% (Jun)

Note: Measured/market sector here is as defined in table 1. See date in brackets above for years used. Technically, the OECD figures are December years 1987-2007 but we have shown labour force survey for March/June years to be consistent with table 2 (OECD use March years for New Zealand output and June years for Australian output but December years for both labour inputs). Measured sector labour input growth is calculated from the index number published by Statistics New Zealand. The index number is derived using a chained Törnqvist index in which weights are based on industry wage shares of the measured sector nominal labour income. The actual hours paid series, which consistent with the Australian series is not weighted by labour income, grew by 0.4 percent from 1988 to 2008.


  • [10]. There are some drawbacks to using hours paid rather than hours worked. For example, increased minimum annual leave provisions in New Zealand from 2007 had the effect of reducing average hours worked per worker but had no direct effect on hours paid.
  • [11]. The use of calendar years aligns with the usual practise of the OECD. However, the OECD do not use calendar years for New Zealand GDP data. The OECD take New Zealand’s GDP data on a March year basis as this is the way Statistics New Zealand publishes it (so 2007 in the OECD GDP figures actually refers to the year to March 2008).
Page top