Household Labour Force Survey

Examples of possible uses include:

• targeting employment or training schemes (e.g. what ages, regions, ethnic groups are more likely to have unemployed people?)
• to indicate employment growth or decline in particular industry or occupation groups.
• to help in modelling the labour market or the economy in general, and make predictions about future levels of employment and unemployment.
• as an indicator of the state of the economy (e.g. increasing or decreasing unemployment rates).

Limitations: Main limitations are the high sampling errors associated with small estimates - this makes many of the smaller estimates unreliable or unusable.

Also, the HLFS does not measure the quality of people’s jobs, e.g. utilisation of skills, how much they are paid (except in June quarters), whether they get sick leave, etc.

There are also sometimes complaints about the definitions used in the HLFS (i.e. to be counted as employed you only have to have worked for one hour or more in a week, or you can even work unpaid in a family business. And to be unemployed you have to be available to start a job and be actively seeking work - not just looking at job advertisements).

The Household Labour Force Survey (HLFS) started in October 1985 as a quarterly survey. First results from the survey were therefore for the December quarter 1985.

In the September quarter 1987, results from all quarters of the survey were revised using population weights based on the 1986 Census of Population and Dwellings.

In the September quarter 1987, data collection practises were changed to minimise response burden on older respondents. The change meant older respondents only had to supply information, after their first interview, if their labour market position had altered. This change is reflected in the introduction of the extra scope code - 'special code for older people'.

In the June quarter 1990 the size of the survey sample was doubled to 24,000 households, and labour force statistics were produced monthly instead of quarterly. This allowed estimates to be produced for 15 regional council areas. (Previously data was available by employment district rather than regional council area.)

Also, in the June quarter 1990, a new questionnaire was introduced for the HLFS. The new questionnaire did not introduce changes to the way the main labour force variables are collected or calculated. Changes were made to the availability of occupation and industry data (made available at 3 digit level instead of two-digit). And the following changes were made to the questionnaire (None of these changes could be regarded as significant):

• A question on whether the respondent would prefer to work fewer hours, was dropped.
• Questions on job search methods, length of time looking for work, and type of work being looked for, were dropped for the employed.
• The question on highest secondary school qualification had a new category for ‘primary proficiency’.
• The routing question preceding that on highest tertiary qualification added a new category for those still at school as well as ‘yes’ and ‘no’.
• The highest tertiary qualification question added a category for ‘local polytech certificate or diploma’.
• A question about whether the respondent had been looking for work before starting a new job, was replaced by a question on the reasons why the respondent had not been looking for work in the last four weeks (this included a category for those waiting to start a new job).
• Occupation questions for the employed and unemployed were changed from asking about tasks and duties to asking ‘what was the respondent's occupation’.
• Categories in the question asking respondents why they left their last job were changed: ‘Laid off’, ‘dismissed, and ‘made redundant’ were combined, while ‘family responsibilities’ and ‘pregnancy’ were deleted, and ‘unsatisfactory work arrangements’ was changed to ‘dissatisfied with job or conditions’.

In the September quarter 1991 the sample size was reduced to 16,000 households and the survey reverted to a quarterly collection. Estimates were produced for 10 regional council areas rather than the 15 previously available.

In the December quarter 1991, changes were made to the ethnic classification used in the HLFS when an ‘Other Pacific’ category was added to the previously available European, NZ Maori, Samoan, Cook Island Maori, Niuean, Tongan, Chinese, Indian, and Other.

In the December quarter 1992, results from all quarters of the survey were revised using population weights based on the 1991 Census of Population and Dwellings.

Starting in the December quarter 1993, the HLFS sample was redesigned using information from the 1991 Census of Population and Dwellings. The new sample was phased in gradually to enable a smooth transition. One-quarter of sample households were replaced with one-quarter of the new sample. This process continued for four quarters, so that in the September 1994 quarter the sample consisted solely of households selected from the new sample.

As a result of the introduction of the new sample in the December quarter 1994, it was possible to provide estimates for twelve regional council areas, rather than the ten regions that were previously available.

In the June quarter 1995, people living in non-private dwellings were excluded from the sample survey population. While the survey population has changed, the population definition for estimation purposes(the target population) has not changed and continues to include people in non-private dwellings.

In the September quarter 1996 an Education and Training Survey (ETS) was conducted as a supplement to the HLFS.

In the December 1996 quarter the HLFS processing system was migrated from the mainframe to the LAN.

In the June quarter 1997 the New Zealand Income Survey was conducted as a supplement to the HLFS. This is now a regular supplement to the HLFS carried out every June quarter.

In the June quarter 1998, results from all quarters back to the September quarter 1991 were revised using population weights based on the 1996 Census of Population and Dwellings. A new weighting methodology (integrated weighting) was also introduced. Integrated weighting was applied to all quarters back to the June quarter 1995. Integrated weighting extends the range of statistics which can be produced to include households, as well as adjusting for non-response more effectively.

In the September quarter 1998 the New Zealand Child Care Survey 1998 was conducted as a supplement to the HLFS.

In the December quarter 1998, ANZSIC industry estimates from the HLFS were published dating back to the December 1996 quarter. The NZSIC industry coding was being phased out of the HLFS.

Starting in the March quarter 1999, the HLFS sample was redesigned using information from the 1996 Census of Population and Dwellings. The new sample was phased in gradually to enable a smooth transition. One-quarter of sample households were replaced with one-quarter of the new sample. This process continued for four quarters, so that in the December 1999 quarter the sample consisted solely of households selected from the new sample.

Beginning in the March 1999 quarter, labour market statistics based on the household as the unit of analysis have been published to an agreed upon classification. The statistics published (in Table 11 of the Hot Off The Press) are a tabulation of household labour force status, against a household type classification.

In the July 2001 quarter revision of total actual hours worked series - to correct for misalignment between survey and calendar quarters.

In the March 2002 quarter integrated weighting was backcast to include quarters 2-38.

In the March 2002 quarter the Cultural Experiences Survey 2002 was conducted as a supplement to the HLFS.

In the March 2003 quarter industry coding changed from NZSIC/ANZSIC dual coding to ANZSIC only coding. The change may introduce a slight discontinuity to the industry series. Also means that NZSIC statistics can no longer be produced.

From March quarter 2004, phase-in of the new HLFS sample design commenced. The sample redesign was based on information from the 2001 Census of Population and Dwellings. The phase-in involved one-eighth of the new sample being introduced each quarter, as the new rotation group. The redesigned sample would be fully introduced from December quarter 2005.

A number of additional questions were introduced from March quarter 2004 although results from the formal study questions were not released until June quarter 2004:

• Q19a - What has been done to work more hours?
• Q19b - If more hours were available could they have been worked?
• Q25a - Has occupation changed in the past 3 months?
• Q32 - Has ... studied or worked towards a qualification?
• Q33 - If studying, would the qualification take three months or more to complete?
• Q54a and Q63a - If offered a job would ... been able to start within the next four weeks?
• Q75 and Q76 - as for Q32 and Q33
• An expanded set of qualifications was introduced although for regular output, codes are concorded back to the previous qualifications values.

In the June quarter 2004, the weights that are used to bring the survey estimates up to population estimates were rebased using the latest population estimates based on data from the 2001 Census of Population and Dwellings. Estimates back to the start of the survey (March quarter 1986) were revised to reflect the new weights that were introduced.

In the September 2008 quarter, the HLFS published new ethnicity data using the single combination method and total response outputs. This was a complete break in the ethnicity series, as the prioritisation of ethnic groups was no longer produced. These series started from December quarter 2007 onwards.

In the December quarter 2008, the weights that are used to bring the survey estimates up to population estimates were rebased using the latest population estimates based on data from the 2006 Census of Population and Dwellings. Estimates back to the start of the survey (March quarter 1986) were revised to reflect the new weights that were introduced.

The December 2008 rebase included the introduction of Māori benchmarks, which were also backcast to the March 1986 quarter.

In the September 2009 quarter, the HLFS first published the new industry (ANZSIC 2006) and occupation (ANZSCO) classifications. The ANZSIC 1996 and NZSCO classifications were supplied until December 2011 quarter.

In the December 2014 quarter an improved sample was implemented. The new sample was based on the 2013 Census of Population and Dwellings. The new sample was rolled in over eight quarters to reduce the risk of affecting the labour force outcomes.

In the March quarter 2015, the weights that are used to bring the survey estimates up to population estimates were rebased using the latest population estimates based on data from the 2013 Census of Population and Dwellings. Estimates back to the start of the survey (March quarter 1986) were revised to reflect the new weights that were introduced. New benchmark weights for regions were also included.

In the June quarter of 2016 a redeveloped version of the HLFS went into the field. The key purpose of the 2016 HLFS redevelopment was to improve the relevance and quality of our labour market statistics.

The new content includes more information about the nature of people’s employment conditions and work arrangements. This data can be used to better understand different patterns of employment. The new content includes:

• employment relationship of employees (eg permanent or temporary work arrangement)
• length of time employed in current job
• employment agreements (eg collective or individual)
• union membership
• preference for change from temporary employment relationship or self-employment.

The redeveloped HLFS also includes more detail from respondents who work in more than one job during a reference week. Multiple job holders are asked about the industry, occupation, employment status, and hours of their second job. Until now, the HLFS collected these details only on the respondents’ main job.

From the December 2020 quarter phase-in of the new HLFS sample design started. The sample redesign was based on information from the 2018 Census of Population and Dwellings. The phase-in involves one-eighth of the new sample being introduced each quarter, as the new rotation group. The redesigned sample will be fully introduced from the September 2022 quarter.

Data collection disruptions: March 2020 to September 2022

Starting on 20 March 2020, survey collection for the HLFS was impacted by health and safety measures aimed at combatting the spread of Covid-19 within Aotearoa.

Interviewing for the HLFS was suspended completely from 25 March to 14 April 2020.

Questions around working from home and job security, and a wellbeing supplement were added to the HLFS when phone interviewing resumed on 14 April 2020.

Face-to-face interviewing was suspended during the following periods:

• 20 March 2020 – 22 June 2020 (national),
• 12 August 2020 – 8 September 2020 (national, excluding Auckland),
• 12 August 2020 – 15 September 2020 (Auckland only),
• 17 August 2021 – 15 October 2021 (national, excluding Northland, Auckland, and Waikato),
• 17 August 2021 – 30 December 2021 (Auckland and Waikato only),
• 17 August 2021 – 21 January 2022 (Northland only).

Adaptations to collection methodology: March 2020 to September 2022

Households are in the HLFS for eight quarters (or waves), during which they are typically visited by a field interviewer in their first quarter (wave 1) and interviewed over the phone in subsequent quarters.

In the June 2020 quarter only, the decision was made to re-contact respondents who would have otherwise had their final HLFS interview in the March 2020 quarter. This was aimed at improving the response rate and the achieved sample rate for the quarter. This did not achieve the desired results and was not reused in subsequent quarters.

During periods where face-to-face interviewing was suspended, letters and reminders were mailed to wave 1 households asking them to call the Stats NZ contact centre to complete their interview over the phone. This resulted in an increase in the time to get a household’s response.

From 8 September 2020, updated health and safety advice meant that field staff would not enter households to conduct the in-person interview. Instead, field staff would knock on doors to get household contact details to carry out a phone interview, or leave a card asking to call the Stats NZ contact centre. If a household indicated they were comfortable with a full interview, this would be conducted outside while observing social distancing practices. These measures limited the risk for field staff and survey respondents. This adapted form of field interviewing performed similarly to traditional field interviewing, and would continue to be used under the Red traffic light setting.

The COVID-19 Protection Framework (traffic lights) was used in New Zealand from December 2021 to September 2022. Under the traffic light system, HLFS collection strategies were informed by the traffic light setting of each region. Under Green and Orange settings, HLFS collections continued to primarily interview respondents over the phone, with field interviewers visiting first-time respondents and households that were hard to reach as part of our traditional targeted face-to-face interviewing. Under the Red setting, field interviewers would only conduct ‘doorstep’ interviewing to get contact details.

3 Quarterly

People who monitor the labour market either in itself or as part of macroeconomic analysis of the economy. For example, government departments (e.g. Ministry of Business, Innovation and Employment , Work and Income New Zealand, Treasury, and Te Puni Kokiri), private research companies (e.g. NZIER), banks (e.g. ANZ), and international agencies (e.g. ILO, OECD).