What change is the Central Statistics Office introducing to the Monthly Unemployment Estimates release from reference month October 2019 onwards?
In an effort to reduce the frequency and scale of revisions to the Monthly Unemployment Estimates (MUE) series, from reference month October 2019 and on a quarterly basis thereafter (January, April, July and October), the Central Statistics Office (CSO) will use the latest available quarterly Labour Force Survey (LFS) benchmarks when compiling the MUE release, ahead of the LFS publication date.
For example, the October 2019 MUE release is being published today, November 5th, 2019. Previously, the Quarter 2 (Q2) 2019 LFS benchmarks would have been used for this release. However, under the new methodology, LFS benchmarks for Q3 2019 have been used instead. This eliminates the need to revise the MUE series when the Q3 2019 LFS release is published on November 19th, 2019.
Why is the CSO making this change?
There have been significant revisions to the MUE series for some quarters in the last couple of years when the latest LFS benchmarks were added into the compilation of the MUE series. The CSO are making this change to reduce the frequency and the scale of such revisions to the MUE series. It should be noted, however, that this change will not eliminate revisions to the MUE series in the future when new quarterly LFS benchmarks are added.
What effect will this change have for users of the MUE series?
This change will reduce the frequency of revisions to the MUE series. From now on there will only be 12 versions of the MUE series released by the CSO each year instead of 16 as is currently the case (12 MUE releases and four MUE revisions published as part of the LFS release each quarter).
The CSO also expects that this change will, in some cases, reduce the scale of revisions to the MUE series by reducing the maximum number of months of the MUE series which are being forecasted. Under the new methodology, estimates of monthly unemployment will now only be forecasted for a maximum of three months instead of up to four months. Prior to this change, the latest quarterly LFS benchmarks available for use for the October 2019 MUE release would have been from Q2 2019 (April, May, June). This meant that the results for July, August, September and October (four months) would have been forecasted for the October 2019 MUE release. The change in methodology means that the Q3 2019 LFS benchmarks have been used for the October 2019 MUE release. Thus, LFS results are available for July, August and September, i.e. the three months of Q3, and there is only a need to forecast forward for October 2019.
What are the quarterly benchmarks from the LFS?
The quarterly LFS benchmarks are the number of persons in employment and number of persons unemployed by age (aged under 25 years/aged 25 years and over) and sex (male/female) which are required to construct the MUE series. The benchmark categories are as follows:
What is the publication schedule for the MUE release?
The MUE release is usually scheduled for publication on the Tuesday following the last Thursday of the reference month (or the Wednesday if the week contains a Bank Holiday Monday). The exception to this is the December release which sometimes may be rescheduled to avoid New Year’s Eve/Day.
Previously, when the quarterly LFS series was being published, the new quarterly LFS benchmarks were used to revise the MUE series. Updated tables for the revised MUE series were then published as an Annexe to the LFS release. Under the new methodology, the LFS benchmarks will already have been used for the MUE release prior to the LFS being published. As a result, there will no longer be a need to publish the MUE tables as an Annexe to the LFS release.
Is there anything else that the CSO could do to reduce the scale of revisions to the MUE?
The CSO are committed to working towards further reducing the scale of revisions to the MUE series. The methodology used by the CSO to compile the MUE series is one of a few methods approved by Eurostat. Statisticians from the Methodology and Labour Market sections of the CSO are working with experts from other National Statistical Institutes and Eurostat to review other approved methods for compiling the MUE series. Once these have been fully evaluated and tested, the CSO will determine whether any of these methods are more robust than what is currently used for Ireland. If a more robust method is found, the CSO will plan towards its implementation and inform users of any further changes.
How are the MUE currently compiled by the Central Statistics Office?
To calculate the monthly unemployment rate, an estimate of both the number of persons employed and the number of persons unemployed for each month up to and including the reference month are required. The CSO forecasts figures for the number of unemployed persons, using the ratio between the latest quarterly LFS benchmarks and Live Register (LR) figures. While the LR is not a measure of unemployment (as it includes people working 3 days or less) it follows a similar trend to the number of persons who are unemployed from the LFS (see figure 1 below). Thus, the trend of the total number of persons on the LR can be used to forecast forward the total number of persons unemployed. For the number of persons employed, LFS benchmarks are forecasted for the reference month.
|X-axis label||Live Register||Unemployed|
Why are previously published estimates from the MUE series revised later?
The MUE is a seasonally adjusted series which will always be subject to revision when data for an additional month becomes available.
Furthermore, when new quarterly LFS benchmarks become available, they are used to update the MUE series; the forecasts from the three months of the quarter that the LFS benchmarks refer to are replaced with estimates for those months.
What scale of revisions has been observed in the MUE series?
Table 1 below indicates the scale of revisions to the monthly series following the publication of LFS results for Q1 and Q2 2019.
|Table 1 Recent adjustments to Monthly Unemployment volumes and rates|
|Original estimate||Revised estimate||Original estimate||Revised estimate|
|Reference Month||Monthly Unemployment||Labour Force Survey||Monthly Unemployment||Labour Force Survey|
|Apr-19||Q1 2019||Jul-19||Q2 2019|
Why does the CSO produce estimates of monthly unemployment if they are so subject to revision?
As the LFS is collected and published on a quarterly basis, the CSO introduced the MUE release in 2015 to provide more timely forecasted estimates for the volume and rates of unemployment. The CSO believes that there continues to be a need for a short-term and more timely measure for labour market estimates even if they are subject to revision. The CSO are continuing to investigate options to further reduce the scale of revisions to the MUE series.
Why does the LFS figure for unemployment differ so much from the MUE forecasts?
Labour Force Survey (LFS) estimates of the unadjusted quarterly unemployment rates are based on survey data collected on a quarterly basis by the CSO. The seasonally adjusted quarterly unemployment rate from the LFS is calculated directly from the seasonally adjusted MUE unemployment rates for the three months of that quarter. Thus, they are directly linked to each other and are comparable.
Monthly unemployment rates from the MUE series are forecasted for the most recent month(s) after the quarter for which the latest LFS benchmarks have been used. These monthly rates are forecasted using a combination of the trend of the LFS benchmarks and the trend of the Live Register, which may differ from the trend being observed for that quarter of the LFS. When the LFS benchmarks for the quarter become available, the forecasts for those months in that quarter are then replaced with estimates based on the LFS benchmark.