Linked Employer-Employee Data: March 2009 quarter

Background to LEED

Official quarterly statistics produced from the Linked Employer-Employee Data (LEED) measure labour market dynamics at various levels – including industry, regional, territorial authority, firm size, sector, sex, and age – providing an insight into the operation of New Zealand's labour market. Statistics New Zealand releases other official labour market statistics that show changes in employment at an aggregate level. Statistics from LEED, such as job and worker flows, help to explain what causes these aggregate movements and are therefore useful for explaining changes in the labour market.

Data sources

The LEED dataset is created by linking a longitudinal employer series from the Statistics NZ Business Frame to a longitudinal series of employer monthly schedule (EMS) payroll data from Inland Revenue.

The Inland Revenue dataset is collected for the purpose of administering New Zealand’s taxation system. It consists of data from EMS and contains details of earnings, tax type, and tax deducted. It does not contain any information relating to the number of hours worked for those earnings.

The Business Frame is a regularly maintained list of all economically significant businesses and organisations (with a turnover greater than $30,000) engaged in the production of goods and services in New Zealand. Information derived from the Business Frame includes:

• industry
• sector (private or public)
• the number of geographical units (physical locations)
• the count of employees at each geographical unit
• the ownership structure of firms.

Historically, the Business Frame was based on and updated from annual survey questionnaires. Since 2002, the coverage of the Business Frame has been extended to include more businesses, and its employment information has been maintained using monthly tax data.

The base data received from LEED is of high quality, but cleaning, transformation, and integration processes are required before robust official statistics can be produced. This is necessary because these datasets are collected for different purposes and are not primarily designed for the production of statistics. Integration processes are required to merge the two sources, as the datasets are constructed differently. One of these processes allocates jobs from an IRD number to geographical units or physical locations associated with that employer.

There is a very small amount of error present in the base data or arising from LEED processes. This is negligible at aggregate levels, but can affect statistics for small categories, for example mean earnings results for small territorial authorities (TAs), regions, or industrial categories. A direct measure of the error is not available, but some caution should be exercised in interpreting statistics based on relatively small numbers of people.

It is important to note that Statistics NZ surveys are specifically designed to collect the data required, and the information requested is targeted to the desired measures. In comparison, the LEED measures are limited by the characteristics of the base data. 

Population

LEED covers all individuals (‘employees’) who receive income from which tax is deducted at source. These payments are made by organisations that are registered with Inland Revenue. Note that the data from LEED includes social assistance payments, such as paid parental leave, student allowances, benefits, pensions, and Accident Compensation Corporation payments, although these are excluded from the quarterly measures. For confidentiality purposes, some individuals are withheld from the data provided to Statistics NZ by Inland Revenue.

In LEED, the employer is the geographical unit or physical location of the business, rather than the administrative reporting unit. For example, a nationwide retail chain may have one Inland Revenue reporting unit covering all of its retail branches. In LEED, each branch is considered to be a distinct employer. This approach has been taken to allow regional, and now TA level, statistics to be produced. It also ensures that LEED is comparable with similar international statistics.

The fundamental basis of the LEED quarterly measures is ‘jobs’. A job is defined as a unique employer-employee pair present on an EMS in the reference quarter.

For inclusion in the LEED quarterly statistics, the job must:

• relate to a person 15 years of age and over
• have PAYE tax deducted at source
• be in relation to ‘paid employment’ rather than a social assistance payment
• have a valid IRD identifier.

An exception is the total earnings measure, which includes all jobs with PAYE tax deducted at source (irrespective of age and IRD identifier) apart from those relating to social assistance payments.

It should be noted that a small number of working proprietors, partners, or other self-employed individuals choose to pay their income tax at source and have not been separated from the 'true' jobs.

Definitions of measures

The following table provides the definitions for each measure in the tables included in this release and also those available from the Statistics NZ website. Other necessary definitions are:

• The calendar year is divided into four quarters, each with three months. The latest quarter is the ‘reference quarter’.
• The ‘reference date’ is the 15th of the middle month of the reference quarter.
• ‘Full-quarter jobs' are jobs that exist continuously over the reference quarter.
• All the earnings measures represent quarterly earnings.
• All the earnings measures are inclusive of tax.
• All the earnings measures include payments reported as lump sums to Inland Revenue.

Annual percentage changes, and quarterly percentage changes are discussed throughout this release and are defined as:

• An annual percentage change compares the reference quarter to the same quarter one year ago.
• A quarterly percentage change compares the reference quarter to the quarter before.

Additionally, job and worker flow statistics have been rounded using graduated random rounding, and earnings statistics have been rounded to base 10 or base 100 for confidentiality purposes. LEED statistics are affected by seasonal variations such as production cycles, school years, and processing procedures associated with the source data.

Seasonally adjusted and trend series are available for filled job and mean quarterly earnings measures for 1-way industry, region and sex tables. These tables are available through  Infoshare

A Guide to Interpreting the LEED Data is available from the Statistics NZ website.

Accessions, separations, and worker turnover rate
The worker turnover rate is calculated using the counts of accessions and separations, which are defined using the reference date concept. Other workers may join and leave during the reference quarter but not be present at either reference date. These workers are not included in the counts of accessions or separations and are therefore excluded from the worker turnover rate.

The worker turnover rate is calculated at the geographic unit level, not the enterprise level. This means that employees who transfer between geographic units within an enterprise will be counted as accessions and separations.

Annual job creation and destruction
Annual job creation and destruction figures are currently not part of the official set of LEED quarterly statistics. The quarterly job creation and destruction statistics have been designed to explain the change in aggregate jobs between two specific points in time – the 15th of the middle month of the reference quarter and the 15th of the middle month of the previous quarter. They compare the number of jobs at each geographic unit on these two dates. Changes in the number of jobs between these two dates are not included in the statistics.

Some users may attempt to produce annual job creation and destruction figures by summing together four quarters of data. This approach is not recommended. Instead Statistics NZ recommends averaging the quarterly job creation and destruction statistics over the year (which is the approach taken in LEED quarterly publications). Estimates of annual job creation and destruction are available on request. These statistics are produced on the same basis as the quarterly series, by comparing employers' job levels between two snapshots a year apart. This method produces significantly less job creation or destruction than adding together four quarters worth of data. Summing quarterly job creation and destruction figures can be seen as overstating permanent job creation and destruction by including seasonal and temporary variations in employment. A similar argument can be made against summing four quarters of worker accessions and separations to produce annual worker flow statistics. However, conceptually it is more appropriate to include seasonal or temporary factors when measuring these worker flows.

Net job change
The difference between the counts of job creation and job destruction (job flows) is equal to the total change in jobs at the aggregate level – the net job change. Job flows reveal the amount of job turnover at the business level underlying the net job change. Similarly, the difference between the counts of accessions and separations (worker flows) is also equal to the net job change. Worker flows reveal the turnover of individual employees underlying the net job change.

These relationships do not necessarily hold for subnational breakdowns. Businesses and individuals may change characteristics, such as industry or age group, over time. This causes a change in the total jobs for that characteristic, but does not affect the job or worker flows.

Continuing jobs and new hires
An employee has a continuing job if they have been with the same employer continuously over the current and previous quarter. A new hire is an employee who has been with the same employer continuously for the current quarter but began the job sometime in the previous quarter. New hires must not have been employed with the same employer in the 12 months prior to the job start date. As a result, seasonal staff and employees who have been rehired within this time period are excluded from new hires.

Patterns in the data
The counts of job creation and destruction, and worker accessions and separations, show an obvious seasonal pattern. This pattern is caused by the annual update of employee counts on the Business Frame, resulting in larger counts of destruction and creation in one quarter than in the other three.

This seasonal pattern changes from the March 2003 quarter. This change is caused by the implementation of a programme to improve the Business Frame maintenance practices and a consequent change in LEED methodology. Methods are being investigated to minimise the changes caused by administrative updating processes.

Compositional changes
Movements in mean earnings statistics are influenced not only by changes in employees' remuneration, such as changes in wage rates, salaries, and hours worked, but also by changes in the composition of the paid workforce from period to period. Compositional changes include variations in relative numbers of males and females, full- and part-time employment, and employment in different industries or within industries.

Seasonally Adjusted and Trend series
Seasonally adjusted and trend series are available for average quarterly earnings of full-quarter jobs and filled job measures for one-way industry, region, and sex tables. Each of the seasonally adjusted components and the total are adjusted separately.

For any series the actual series can be broken down into three components: trend, seasonal, and irregular. Seasonal adjustment aims to eliminate the impact of regular seasonal events (such as annual cycles in holidays) on the time series. This makes the data for adjacent quarters more comparable. The seasonally adjusted series has had the seasonal component removed, while the trend series has had both the seasonal and irregular components removed.

Trend estimates reveal the underlying direction of movement in a series and are likely to indicate turning points more accurately than seasonally adjusted estimates. The trend is calculated as a 'centred moving average' of the seasonally adjusted series. Towards the end of the series, trend estimates are subject to change because new data points are used in the estimation process as they become available. Revisions for the trend series can be particularly large if an observation is treated as an outlier in one quarter but is found to be part of the underlying trend as further observations are added to the series. Typically, only the previous two or three estimates will be subject to substantial revisions.

Seasonally adjusted series are also subject to some revisions, as they are also obtained using central moving average methods. Generally, these revisions are smaller than for the trend component.

The X-12-ARIMA package has been used to produce the seasonally adjusted and trend estimates for the LEED.

Dimensions available

Dimensions available from the LEED quarterly tables on Table Builder on the Statistics NZ website (on an ongoing basis) are:

• quarter
• industry of employer
• region of employer
• sex of employee
• age of employee
• firm size
• sector of employer – private or public institutional sector of ownership
• territorial authority (TA) of employer.

These dimensions, with the exception of TA, are also available on Infoshare on the Statistics NZ website.

Seasonally adjusted and trend series are only available on Infoshare on the Statistics NZ website.

The industry statistics are based on ANZSIC 2006 (or ANZSIC06), the latest edition of the Australian and New Zealand Standard Industrial Classification. The 1996 version of ANZSIC (ANZSIC96), used in industry outputs in previous releases, has been updated to the 2006 edition. Note that industry outputs defined using ANZSIC06 are not comparable with those based on ANZSIC96. For more information about ANZSIC06 go to the Statistics NZ website.

The regional statistics refer to the regional council area of the employer, meaning that all statistics are based on the employer's address and not the employee's. Regional statistics are compiled from 16 regions:

• Northland
• Auckland
• Waikato
• Bay of Plenty
• Gisborne
• Hawke's Bay
• Taranaki
• Manawatu-Wanganui
• Wellington
• Tasman
• Nelson
• Marlborough
• West Coast
• Canterbury
• Otago
• Southland.

For confidentiality purposes, this release combines data for the Tasman, Nelson, Marlborough, and West Coast regions.

The firm size dimension refers to the size of the business at an enterprise level, not at the geographic unit level. Firm size is based on the employee count on the 15th of the middle month in the quarter of interest.

The TA statistics are compiled from 72 TAs, which cover 16 city councils and 56 district councils. Data from the Chatham Islands territory have been included in estimates for Christchurch city to reduce the impact of errors for a small population. For confidentiality reasons, data from TAs that overlap regional boundaries have been modified by adding the sensitive portion of the overlapping TA to a neighbouring TA in the same region.

There are six modified TAs:

• Waitomo district
• Taupo district
• Whakatane district
• Hastings district
• Ruapehu district
• Rangitikei district.

The allocation of jobs to a particular TA is carried out on the basis of the employer's address on the Business Frame. Therefore, all TA level statistics are based on the employer's address and not the employee's.

The job creation, job destruction, and total earnings measures cannot be generated for age or sex dimensions. This is because job flows are calculated at the geographic unit level, and the total earnings measure includes those with invalid IRD identifiers that have no age or sex classification.

The figures in the tables have been rounded, and discrepancies may occur between sums of components and totals. Some businesses are not able to be assigned to an industry. This contributes to the difference between the New Zealand totals and the sum of the industry totals for the earnings, jobs, and job and worker flow measures. All businesses are associated with a region and TA, and therefore the difference between the New Zealand totals and the sum of the region or TA totals for these measures is small.

Frequency of outputs

LEED job measures are generated as a quarterly series, although tax data is received monthly. This is done to reduce volatility caused by the variable number of pays per month and to ensure comparability with other statistics.

Timing of the measures

LEED measures are produced as counts of jobs at a point in time, or means and medians of earnings for jobs existing for a full quarter.

Counts of jobs or workers are taken on the 15th of the middle month of the quarter.

Measures relating to average earnings are produced using the full-quarter concept. A disadvantage of the point-in-time approach is that the earnings for a job relate to the entire month regardless of the actual days worked. Therefore, average earnings statistics per job produced under this concept would include people who worked one day (or even one hour) in the month with people who worked all month.

The total earnings measure does not use either the point-in-time or the full-quarter concept, and is instead a simple sum of all earnings paid out at any time in the reference quarter.

Timeframe for production
The timeliness of LEED is dependent on a number of factors:

• Employers take time to complete their EMS schedules and supply them to Inland Revenue.
• Inland Revenue requires time for processing and supply to Statistics NZ.
• Statistics NZ requires further time for receipt, data transformation, and the production of output data.
• The derivation of full-quarter outputs requires data for an additional quarter after the reference quarter.

In addition, late returns and updates are received in LEED well after the end of the reference period. These can distort the measures produced, particularly the estimates of change.

LEED statistics are therefore published 12 months after the reference quarter. A delay of this length ensures that the published value is sufficiently close to the real world value. The statistics are then revised with updates from Inland Revenue for an additional two quarters. Updates after this stage have an immaterial impact on the statistics, therefore 18 months after the reference quarter the data is considered final, and subsequent updates from Inland Revenue are ignored.

More information

For more information on LEED, see the Guide to Interpreting the LEED Data. Additional data and breakdowns are available on Infoshare and from the LEED quarterly tables on Table Builder on the Statistics New Zealand website.

Copyright

Information obtained from Statistics NZ may be freely used, reproduced, or quoted unless otherwise specified. In all cases Statistics NZ must be acknowledged as the source.

Liability

While care has been used in processing, analysing, and extracting information, Statistics NZ gives no warranty that the information supplied is free from error. Statistics NZ shall not be liable for any loss suffered through the use, directly or indirectly, of any information, product, or service.

Timing

Timed statistical releases are delivered using postal and electronic services provided by third parties. Delivery of these releases may be delayed by circumstances outside the control of Statistics NZ. Statistics NZ accepts no responsibility for any such delays.