Information Notice - Labour Force Survey Quarter 3 2017

 

Almost half of the total difference of 96,400 was recorded in the 20-24 age group (45,300 or 46.9%) representing an increase in the estimated population of this age group of 19.8%. The 25-29 age group accounted for 15.1% of the overall difference while the 15-19 age group accounted for 12.6%.

In proportional terms, by far the greatest impact is again seen for the 20-24 age group. Based on the previous population estimates, this group had accounted for 6.3% of the total working age population. Following the revisions, this now increases to 7.3%. The proportional changes among the other age groups were less significant.

There are also differences in the breakdown of the population by nationality between the old population estimates and the Census of Population 2016 results. Table 2 below shows the updated working age populations by nationality. The updated Irish national working age population for Q2 2016 is now 118,500 or 3.8% higher, while the non-Irish working age population is now 22,000 or 4.4% lower. There are also significant changes within this non-Irish working age populations – whereas the population of the EU15 excluding Ireland and the UK is now estimated to be 31,900 or 116.0% higher, the working age population of the Other nationality grouping is now estimated to be 58,900 or 34.1% lower.

 

New Labour Force Survey

The introduction of the new LFS is part of a wider Household Survey Development (HSD) modernisation project that the CSO has been engaged in over the past number of years. This project was undertaken to expand the range of social statistics to meet new needs for information on households and persons, to introduce methodological improvements to the survey and to modernise survey processes.
Therefore, the following methodological changes were made as part of the introduction of the new LFS:

• Mixed Mode Data Collection

One of the primary aims of the HSD project was to facilitate an expansion of the range of social statistics to meet new needs for information on households and persons and to deliver this increase using the existing interviewer field force.  This increase has been delivered by introducing mixed mode data collection as part of the new LFS and therefore Computer Assisted Telephone Interviewing (CATI) data collection is now used in the LFS and this then allows the existing interviewer field force to collect additional social surveys.

As with the QNHS, information is collected from each sample household over 5 successive quarters or waves. However, in the LFS, the first interview is conducted by a team of face-to-face interviewers using Computer Assisted Personal Interviewing (CAPI) while the remaining four follow-up interviews are conducted using CATI from a dedicated call centre.

• Sample Allocation

In the QNHS, each wave of interviews took place in a particular 2-3 week period during the quarter and the sample was only representative when considered over the full quarter. In the LFS however, all five interview waves have been distributed equally across the quarter such that data is collected from each of the waves in each week of the quarter. This change will allow the CSO investigate the possibility of producing monthly labour market statistics directly from the LFS. 

• Non-response Adjustment

Over recent years, response rates for household surveys have been falling in both Ireland and other countries. Typically, non-response occurs when households that are sampled, and that are eligible for the survey, do not provide the requested information. This can lead to biased survey estimates if specific groups within the population are over or under-represented and if these groups behave differently with respect to the survey variables (i.e. labour market outcomes).

To correct for this, the CSO has introduced a non-response adjustment into the weighting procedure for the LFS from Q3 2017 onwards. The adjustment applies extra weight to the groups who tend to be less likely to respond to the survey to make the results from the achieved sample more representative of the target sample and target population.

• Questionnaire Review

The questionnaire (survey instrument) previously used in the QNHS was reviewed in line with the introduction of the new LFS. This involved standardising some questions and making them more suitable for telephone interviewing. In addition, some questions which had been added at various stages of the QNHS were removed as they were not necessary to meet regulatory requirements. The questionnaire review provides the capacity to add labour market related questions to the LFS in future on such topics as zero-hour contracts and the ‘gig’ economy.

• Industry and Occupation Coding

In the QNHS, CSO interviewers collected information on the industrial sector of the enterprise that a respondent worked in as well as collecting information on the respondent’s occupation and the interviewers then coded these descriptions on the doorstep using look-up files on their handheld computer. In the LFS, interviewers collect a detailed description of the enterprise and occupation from respondents and these are then coded in-house at the CSO by an automated process which is reviewed by a small dedicated team of coding experts. This new approach reduces subjectivity and increases the quality and standard of the coding of these key variables.

• Seasonal Adjustment Methodology

The seasonal adjustment methodology has been reviewed following the introduction of the new LFS in Q3 2017. Following this review, from Q3 2017 onwards, the seasonal adjustment of the LFS is conducted using the X-13ARIMA-SEATS produced, distributed and maintained by the U.S. Census Bureau. The adjustments are carried out by applying the X-13-ARIMA model to the unadjusted data. This methodology estimates seasonal factors while also taking into consideration factors that impact on the quality of the seasonal adjustment, such as:

•    Calendar effects e.g. the timing of Easter
•    Outliers, temporary changes, and level shifts in the series

For additional information on the use of X-13ARIMA-SEATS see:

http://www.census.gov/srd/www/x13as/

Seasonal adjustment is conducted using the direct approach, where each individual series is independently adjusted. As a result of this direct seasonal adjustment approach it should be noted that the sum of any component series may not be equal to seasonally adjusted series to which these components belong, e.g. the seasonally adjusted number of males in employment and the seasonally adjusted number of females in employment will not necessarily add up to the total employment on a seasonally adjusted basis.

The X-13-ARIMA method has the X-11 moving averages process at its core, but builds on this by providing options for pre-treating the series using a regARIMA approach for prior adjustment and series extension i.e. this methodology will estimate seasonal factors while taking account of calendar effects (e.g. timing of Easter), outliers, temporary changes and level shifts.

The seasonal adjustment is designed and implemented in full accordance with the ESS Guidelines (2015).

Review of Agriculture, Forestry and Fishing sector estimates (Q1 2009 – Q3 2013)

The CSO has reviewed the previously published Agriculture, Forestry and Fishing sector estimates for the Q1 2009 to Q3 2013 period to address issues previously identified in this sector. As a result, estimates for this period for the sector have been revised resulting in a consistent series over that period. These revisions to the sector also result in minor revisions to other NACE Rev. 2 sectors over the period.

This revision was undertaken as the CSO previously undertook an analysis of the trend in of employment in this sector over the period and the analysis showed that the trend was influenced by the introduction of a sample which commenced in Q1 2009 and was included until Q3 2013.

The new estimates have been produced by examining the trend recorded in the sector between the Censuses of Population 2006 and 2011 by region and sex. These trends have now been used to derive new estimates of employment in the sector for Q2 2011 and these new benchmarks have been used to produce new estimates for all quarters which needed to be revised.

Administrative data sources related to the Agriculture, Forestry and Fishing sector were also examined for the period in question. The examination of this data indicated that the sector could have been expected to have been largely unchanged over the period and the new updated LFS series for this sector now reflects this expectation.

Summary of Revisions to Previously Published Estimates

As a result of the Census of Population 2016 revisions, the move to a new LFS survey and the review of the Agriculture, Forestry and Fishing sector estimates for the Q1 2009 to Q3 2013 period, the previously published labour market series are now revised.

Backcasting of Headline Indicators

To minimise the breaks in series to the key survey estimates, the CSO has created a backcasted QNHS series from Q1 1998 to Q2 2017. These backcast series have been created using scaling factors from a comparison of data captured from a parallel run of the QNHS and LFS at the beginning of 2017.

These backcast series are consistent with the LFS series that is being published for the following main indicators:

•    Employed males aged 15-24 years
•    Employed males aged 25+ years
•    Unemployed males aged 15-24 years
•    Unemployed males aged 25+ years
•    Inactive males aged 15-24 years
•    Inactive males aged 25+ years
•    Males aged < 15 years
•    Employed females aged 15-24 years
•    Employed females aged 25+ years
•    Unemployed females aged 15-24 years
•    Unemployed females aged 25+ years
•    Inactive females aged 15-24 years
•    Inactive females aged 25+ years
•    Females aged < 15 years

Scaling factors were also calculated for the NACE Rev. 2 groups published in Table 2a of the LFS release, broken down by sex, resulting in 30 additional categories. As a result, these series are also consistent with the new LFS series.

It was not possible, however, to control for all of the series included in the QNHS release. Therefore, other than the headline indicators and NACE Rev. 2 series described above, there may be changes in the levels of other series from Q3 2017 onwards and consequently users should exercise caution when comparing data from before and after this period.

Overall Impact on Labour Market Estimates

All QNHS series have been revised from Q1 1998 to Q2 2017 inclusive. The Census of Population revisions have been applied to the period from Q3 2011 to Q2 2017 inclusive; the scaling factors which resulted from the introduction of the new LFS for the headline indicators and the NACE Rev. 2 groupings in Table 2a have been applied from Q1 1998 to Q2 2017 inclusive and the review of the Agriculture, Forestry and Fishing sector estimates have been applied from Q1 2009 to Q3 2013 inclusive.

These revisions are incorporated in the results for Q3 2017 and the clearest impact is seen on volume estimates for both employment and unemployment.

• Number of persons in employment

Figure 2 below illustrates the volume effect of the Census of Population 2016 revisions and the LFS on the historic QNHS employment series for a selected period. This graph contains three series:

•    QNHS: the previously published QNHS estimates;
•    COP 2016: the previously published QNHS estimates adjusted for the Census of Population 2016;
•    LFS: the previously published QNHS estimates adjusted for both the Census of Population 2016 and the new LFS.

 

The most recently published employment estimate, which was for Q2 2017, was 2,063,000. This figure increases by 75,400 to 2,138,400 following the Census of Population 2016 revisions. When the adjustment for the LFS is applied, the figure increases by a further 42,800 to 2,181,200. Therefore, in Q2 2017, the revisions to employment estimates arising from the Census of Population 2016 accounted for 63.8% of the total 118,200 increase in employment between the new LFS series and the historic QNHS series while the adjustment for the LFS accounted for 36.2%.

Figure 3 below compares the number of persons employed for the historic QNHS series, with Census of Population 2016 revisions applied, and the new LFS series for males, females and all persons. While there have been slight increases in the level for both males and females, the trend is the same for both series.