Labour Market Statistics: December 2021 quarter

Period-specific information Response Rates Survey Reference period Response rate Sample rate HLFS Each week during the quarter (1 October 2021 – 31 December 2021) Target: 90 percent Achieved: 80.6 percent Target: 76 percent Achieved: 74.2 percent QES The pay week ending on, or before, 20 November 2021 Target: 90 percent Achieved: 91.0 percent N/A LCI Pay rates at 15 November 2021 Target: 94 percent Achieved: 92.0 percent N/A See Household Labour Force Survey sources and methods: 2016 for more information on the sample rate and response rates. HLFS Coverage rates Usually the undercoverage rate in the HLFS is around 20 percent. The overall undercoverage rate for the HLFS in the December 2021 quarter was 18.9 percent. This compares with 17 in the September 2021 quarter and 20.6 percent in the December 2020 quarter. Data quality For most regions in New Zealand, data collection for the HLFS performed better in the December 2021 quarter than in the September 2021 quarter. However, collections in Northland and Auckland were still affected by Covid-19 related restrictions and had lower achieved sample rates (ASRs) as a result. See COVID-19 and labour market statistics in the December 2021 quarter for more information on how data collection was affected by Covid-19 related measures for the HLFS December 2021 quarter. A low ASR for the South Auckland collection area prompted investigations into possible non-response bias in our reported outcomes for Māori, Pacific peoples, and 15-24 year olds, as this area has higher concentrations of these populations. A high non-response bias in this area could negatively affect our estimates for these groups, especially when combined with higher sample rates from the rest of New Zealand. Data quality was assessed for Auckland, Northland, Pacific peoples, and the 15-24 year age group by monitoring the following time series: Achieved sample rate (ASR) The ratio of summed calibrated weights to summed design weights Undercoverage rate Eligibility status Note that the ratio of summed calibrated weights to summed design weights represents the level of adjustment to the design weights to a.) account for differences between the achieved sample and the target population b.) account for differential nonresponse in the achieved sample. To investigate the effect of possible non-response bias we tested various non-response adjustments, using the time-series below to measure the effect of the adjustments: Unemployment rate Employment rate NILF (Not in the Labour Force) Rate The Working Age Population (WAP) The following non-response adjustments were tested: NZ Deprivation Index (NZDEP): This adjustment had a small, consistent shift across all measures. The magnitude of the effect was similar across all quarters, including the September and December 2021 quarters. Combination of NZDEP and mode of contact: This adjustment had almost no effect when compared with published statistics. Combination of NZDEP and low/high indicator of composition of Pacific peoples in the primary sampling unit (PSU): This adjustment provided nearly identical estimates to those obtained through adjusting by NZDEP alone. Based on the analysis performed, we do not have evidence to suggest that possible non-response bias has a significantly greater effect on the estimates of key measures this quarter as compared to previous quarters. Estimates produced by the HLFS are benchmarked and the results of the analysis above show how these benchmarks help ensure the quality of our national measures. The benchmarks we use for the HLFS are five-year age groups by sex, the number of Māori adults by two age groups (age 15-29, 30+), and 12 regions. See HLFS data sources and methods for more information on benchmarks and how they improve our estimates. We have high confidence in our national estimates. This is due to the high response rates and ASRs across most of the country, combined with the use of benchmarks. We are confident in our estimates for Māori and the 15-24 year age group. These populations form a part of our benchmarks and had sufficient coverage at a national level to provide reliable estimates. Care should be taken when looking at regional estimates for Northland and Auckland, as well as estimates for Pacific peoples, due to their lower ASRs and possible undercoverage. We recommend looking at longer term trends and taking sample errors into account when looking at estimates for these regions and groups. We continue to investigate possible ways of improving our estimates for Pacific peoples, including the use of non-response adjustment, as reliable benchmarks for this population are not available. Any changes will be communicated to our customers. Suspension of publication of trend series Due to the impact of COVID-19 on the September 2021 quarter, several seasonally adjusted series were specially treated to maintain a consistent estimate of the seasonal pattern. While no additional adjustments were performed in the December 2021 quarter, the trend estimates of the September and December 2021 quarters for the above series will not be published until there is sufficient data to allow for the calculation of a stable trend series. New Sample Following every census, we review the HLFS sample design. The updated sample design will be implemented over eight quarters (two years), starting in the December 2020 quarter. For the December 2021 quarter, five of the eight waves are from the new sample. Outliers During the seasonal adjustment process, X-13-ARIMA-SEATS can give less weight to the irregular component. Specifically, if the estimated irregular component at a point in time is sufficiently large compared with the standard deviation of the irregular component as a whole, then the irregular component at that point can be downweighted or removed completely and re-estimated. We refer to such observations as partial- and zero-outliers, respectively. In practice, the downweighting of outliers does little to seasonally adjusted data, but the impact of the outliers on the trend series will generally be reduced. However, if an outlier ceases to be an outlier as more data becomes available, then significant revisions to the trend series become possible. Outliers Quarters Male employed Female employed Male unemployed Female unemployed Male not in the labour force Female not in the labour force Mar 2021 .. .. .. .. .. .. Jun 2021 .. .. .. .. .. .. Sep 2021 .. .. .. .. .. .. Dec 2021 .. .. .. .. .. Z Key: .. – no adjustment. P – partial weight. Z – zero weight. Revisions to HLFS Each quarter, we apply the seasonal adjustment process to the latest quarter and all previous quarters. Every estimate is subject to revision each quarter as new data is added, which means that seasonally adjusted estimates for previous quarters may change slightly. In practice, estimates more than two years from the end-point will change little. This table lists the changes in estimates between the current and previous quarters for the seasonally adjusted data. Percent revision from last estimate, seasonally adjusted Quarter Male employed Female employed Male unemployed Female unemployed Male not in labour force Female not in labour force Dec 2020 -0.02 -0.08 1.84 1.45 0.03 0.08 Mar 2021 0.04 -0.02 -0.18 -0.14 0.00 -0.01 Jun 2021 -0.01 0.10 -1.35 -1.32 -0.02 -0.15 Sep 2021 0.00 0.01 -0.52 -0.08 -0.01 0.06 This table presents revisions for the trend estimates. Trend revisions are generally larger than those of the seasonally adjusted data. Percent revision from last estimate, trend Quarter Male employed Female employed Male unemployed Female unemployed Male not in labour force Female not in labour force Dec 2020 0.00 -0.07 1.18 1.07 0.02 0.09 Mar 2021 0.01 -0.06 0.75 0.73 0.03 0.07 Jun 2021 0.02 0.17 -2.73 -2.70 -0.07 -0.36 Sep 2021 0.06 0.75 -10.22 -10.50 -0.34 -1.34 The table below shows the average of all such absolute revisions, expressed relatively, and indicates to what extent the current estimates might be revised when the revised data for the next quarter becomes available. Mean absolute percent revisions Seasonally adjusted Trend 1-step 4-step 1-step 4-step Male employed 0.05 0.09 0.20 0.21 Female employed 0.06 0.10 0.25 0.27 Male unemployed 0.55 0.87 2.17 2.16 Female unemployed 0.52 0.95 2.21 2.32 Male not in labour force 0.10 0.17 0.40 0.41 Female not in labour force 0.10 0.15 0.40 0.42 QES Response rate The response rate for the Quarterly Employment Survey in the December 2021 quarter was 91.0 percent. Understanding inter-quarter variability in the Quarterly Employment Survey Stratified sample design divides a population into smaller mutually exclusive groups, called strata. Random samples are drawn within these strata. The goal of stratification is to group similar units, by industry and employment count for QES. Reducing variability in output variables between units within each stratum allows a more efficient and representative selection of units to be surveyed over the population. Different strata have different probabilities of selection and sample weights, according to the expected level of variability and contribution to the population’s outputs. For example, the very largest businesses may have a 100% chance of selection and carry a weight of 1, so that they represent only themselves. Meanwhile, very small businesses have a low chance of selection, and those sampled carry a large weight to represent many units. Each quarter, units in QES are allocated to strata based on ANZSIC division and specific cut-offs in employment count. A sample is drawn according to the probabilities of selection in each stratum. Typically, a similar group of businesses is sampled each quarter since selection at the strata level is based on a fixed span of permanent random numbers, to which each business is assigned. Similarity of the sample between quarters promotes accuracy in inter-quarter movements. Conversely, sample changes due to business births, deaths and movements between strata reduce accuracy in inter-quarter movements. Prior to March 2021, the old sample design maintained a constant strata allocation but still experienced some changes in the sample due to business births and deaths. The new sample design reallocates strata each quarter to maintain an efficient and representative sample. The downside of regular reallocation is larger variability in inter-quarter movements than previously experienced, especially at the industry level. Data based on full-coverage administrative sources may be more suitable than sample data such as QES for studying inter-quarter movements at lower levels. Please refer to the Monthly Employment Indicator (MEI) or Business Employment Data (BED) series for more information. LCI The LCI measures changes in salary and wage rates for a fixed quantity and quality of labour. LCI data is collected by postal and electronic surveys. For the December 2021 quarter, respondents were asked to report pay rates on the reference date of November 15th 2021. The response rate was 92 percent, finishing below the target response rate of 94 percent. The response rate for key firms was 100 percent, meeting the target of 100 percent. General information and methodology For general information and methodology on the specific surveys within the labour market statistics release, please see the following Datainfo+ pages: Household Labour Force Survey Labour Cost Index Quarterly Employment Survey en-NZ