Stats NZ has a new website.

For new releases go to

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Frequently asked questions about productivity – Composition-adjusted labour

Do you account for the different skills of workers?

Yes, we do take account of different skill levels. This is done through a proxy measure where the proxies are the education and experience levels of workers. The impact of quality is assessed in the resulting composition-adjusted series. However, the headline measure of productivity is not explicitly adjusted for the skill level of workers.

What does composition-adjusted mean?

Composition-adjusted labour productivity accounts for skill (or quality) differences among workers. Under this measure, an hour worked by a skilled person (eg a lawyer), is weighted higher than that for an unskilled worker (eg a cleaner). It is impossible to obtain a perfect measure of skill, so proxies are used instead. These proxies are the education and experience levels of the worker.

Why are you adjusting labour input for changes in composition?

Each hour of labour is given an equal weight in the labour volume series. Ideally, labour input should capture the knowledge and skill dimensions of labour. This is done through a proxy measure, where the proxies are the education and experience levels of workers. This analysis helps us understand what drives both GDP and labour productivity, through the growth accounting framework.

Is overtime included in the productivity estimates?

Paid overtime is included in the hours paid data, sourced from the HLFS, and fed into the unadjusted labour input measure. The New Zealand Income Survey (NZIS) wage measure used in the composition-adjusted measure is defined as the usual hourly wage (excluding overtime). Unpaid overtime is not included in either series.

Why does the composition-adjusted series only go back to 1998?

The NZIS, an annual (June) supplement to the HLFS, started in June 1997. Results from this survey provide the all-important hourly wages data. Other datasets, such as census, were considered for backdating but were not used due to quality concerns.

Why did you use the New Zealand Income Survey for the composition-adjusted approach?

On balance, the NZIS provides the most frequent source of information on wage data, as well as the additional explanatory variables required for the regression model (namely education, experience, and sex).

Why didn't you use occupation in the composition-adjusted series?

Specific occupation codes can provide a good indicator of skill, although problems arise when these codes are not at a specific level. In the HLFS, robust occupation data are only available at an aggregated level. In addition, occupation is a self-selected variable in the HLFS, and the sample is not stratified by occupation.

How did you calculate the weights for the composition-adjusted measure?

The weights are based on the influence a worker’s education and experience have on wages. A regression model was used to determine the weights. Data from the New Zealand Income Survey, an annual supplement to the Household Labour Force Survey (HLFS), were used to calculate these weights. There is no assumption that a higher education or experience level equate to a higher wage rate. The model simply assumes that a person's level of education and experience can have some effect on their level of income. The data show that those with a bachelor’s degree or above, and those with more experience, earn the highest level of income.

How did you account for experience in the composition-adjusted series?

Theoretically, each year of experience should provide a positive but smaller wage return (ie a diminishing marginal return from experience). When experience increases by one year, the total impact of experience on wages also increases. However, the wage return from the last year of experience is less than that of the previous year. If there wasn’t an experience-squared term, there would just be a linear increase in wages for each year of experience, which is unrealistic.  

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+