The earnings statistics and gender pay gap reports provide statistics on employment earnings. The calculation of these statistics involves linking datasets from the Income Tax Instalment Scheme (ITIS), Social Security contributions, and Manpower returns. These datasets are supplied by Revenue Jersey and the Employment, Social Security, and Housing Departments. Records are matched at the individual job level, and the necessary data is available from January 2022 onwards. These statistics were published in a single report prior to December 2024, and are now published as separate reports.
For a comparison between these earnings statistics and our Index of Average Earnings statistics, and planned changes to these outputs, please see our earnings statistics explainer. See also the Index of Average Earnings methodology for more details about that report.
In all three of these reports, earnings refers to employment earnings. For statistics on household income, including earned and unearned income, see our household income distribution report.
Data sources
- Manpower returns: Employers submit monthly returns to the Government, providing details on all employees, including contract type (full-time, part-time, or zero-hour). Since January 2022, these returns have included contracted hours for dependently employed jobs.
- Social Security contributions data: This dataset includes all dependently employed jobs where earnings are subject to Social Security. It contains employer information, and employee earnings on a monthly basis.
- ITIS returns: ITIS data records earnings for any job where a salary or wage has been paid. Unlike Social Security contributions, ITIS data does not have an upper earnings limit or a lower threshold below which earnings are not subject to Social Security. By combining ITIS and Social Security data, a complete picture of dependently employed jobs can be created, allowing for the calculation of both medians and means.
- Demographic information is collated from various sources for our population estimates. To maintain consistency with other reports these demographic values of “best information available” are used. See Population and migration – Methodology and quality: December 2024 for additional details.
Whilst for the vast majority of employees (over 99.9%) gender is based on sex at birth provided by Employment, Social Security and Housing (ESSH), where a majority of other administrative sources hold different records to ESSH, then the recorded sex from those is used instead. Prior to June 2025, gender pay gap reports only used the sex at birth provided by ESSH.
Data cleansing and linking process
- Jobs where no Social Security contributions were paid, and where the individual was classified as Class 2 (mainly self-employed), are excluded for that month.
- Remaining jobs from the contributions dataset are matched with Manpower return data using a pseudo-anonymised ID for the individual and the employer’s TIN or manpower code as unique identifiers.
- This linking process ensures that employment sector (SIC 2007 classification), contract type, and contracted hours from the Manpower returns can be accurately matched to the Social Security contributions data.
- Any inconsistencies, such as a full-time contract with contracted hours of less than 25, are flagged for imputation. Jobs missing contracted hours in their Manpower submission are also marked for imputation.
- ITIS data is then linked using the previously assigned unique identifier. Jobs with no earnings in either ITIS or Social Security contributions are excluded, as they indicate no employment income for that month.
- Jobs appearing solely in ITIS data and the individual is classified as Class 2 (mainly self-employed) are excluded, as they represent self-employment cases where a regular salary is taken.
- Where earnings values differ between ITIS and Social Security contributions, ITIS values are used.
- Datasets are linked with those used for our population and migration statistics to ensure the same information is being used for each person, both within these reports, and across our economic and social statistics. More information on their methodology can be found in our Population and migration statistics from administrative data – Methodology and evaluation report.
Imputation process
Records flagged for imputation are processed using a random forest algorithm to estimate missing contracted hours. The model incorporates multiple indicators and undergoes validation checks to ensure reliability. For example, the distribution of imputed values is compared to originally submitted values to maintain consistency.
Calculations
Earnings calculations
Once imputation has occurred all jobs have a contracted weekly hours value. This allows earnings to be adjusted to full-time equivalent (FTE) earnings. Earnings are adjusted based on the ratio of each jobs contracted hours and the most frequent (modal) hours worked in that sector.
This sees jobs adjusted both up and down in earnings. The sector modal hours are used so that the least number of jobs are adjusted to provide the most realistic picture of full-time for each sector.
There can be some slight, sometimes seasonal, variation. In such cases both sets are used to calculate an overall earnings value with results chain-linked in the median earnings index. This index is therefore recommended for use when measuring annual changes as it will not be impacted by these variations in hours.
Additional adjustments are made for weekly-paid staff to represent similar pay to those paid monthly. This is determined using information contained within social security and ITIS datasets (some companies have separate returns for those paid weekly and monthly). This identifies a large portion of expected weekly paid jobs.
If the dataset included an indicator on pay frequency in the future that would provide more certainty, in particular for jobs of other pay frequencies, such as fortnightly.
Median earnings are then calculated both overall and for any disaggregation needed, such as age or sector.
Earnings statistical calculations
- The median average earnings are the earnings for the job in the middle of the earnings distribution (the 50th percentile, the mid-point between the lowest and highest earning job. Half of jobs will have a lower income than the median, and half will have a higher income than the median.
- The mean average earnings is the sum of the earnings of all jobs, divided by the number of jobs.
- The modal average hours per week are the most frequently or commonly worked number of hours.
- The Gini coefficient is an internationally recognised measure of income inequality, which aims to summarise the degree of sharing of income. In these reports it is used to summarise the degree of sharing of earnings across jobs. A Gini coefficient of zero indicates that earnings are evenly spread across jobs, with each job receiving the same proportion of earnings (that is, complete equality); whilst a Gini coefficient of one would represent a single job in the population receiving all the earnings and the rest of the population receiving nothing (that is, complete inequality).
Median earnings index calculations
The median earnings index tracks the change in median earnings over time. It accounts for seasonal changes in the most frequently contracted hours by comparing months on the same basis. Whilst often this results in very minor differences to the trend of earnings, it is important to do this so that any large changes in hours worked in sectors are accounted for.
Gender pay gap calculations
The gender pay gap is calculated as the percentage difference between male and female hourly rates of pay, calculated by dividing male average earnings by the corresponding average earnings for females. A positive value indicates that males earn more, while a negative value indicates that females earn more.
To ensure statistical reliability, results are only presented for groups containing at least 100 males and 100 females. Groups with fewer individuals are excluded due to potential large uncertainties, which could lead to misleading interpretations; see recommendation 8 of the RSS’ 10 proposed reforms for gender pay gap reporting.
Seasonal adjustment
Employment earnings can exhibit seasonality, particularly for certain sectors. In order to enable meaningful monthly comparisons, average overall earnings are seasonally adjusted using the Eurostat Demetra software package.
Seasonal adjustment is based on ongoing estimation of seasonal trends and as such is subject to revision. To ensure a reasonably stable series of data for the user, the seasonal model is revised once per year, at which point the entire historic seasonally adjusted series will potentially be revised. These revisions are welcome as they derive from an expanded set of data and lead to better estimates of the seasonal pattern.
All other figures presented in these reports are based on the non-seasonally adjusted measures, unless otherwise stated.
Methodological changes
Earnings statistics methodological changes
Early experimental versions of these earnings statistics and gender pay gap statistics were produced using only Social Security contributions and Manpower returns, as ITIS data was not available for reports published prior to December 2024. As a result, reports published before this time did not contain mean averages, as earnings for the highest earners were not always available from Social Security contributions data. These experimental statistics were calculated on a six-monthly basis, rather than the monthly basis used in reports published since December 2024.
In reports prior to December 2024, comparisons with the UK were made between the Jersey Average Earnings Index and the Office for National Statistics (ONS) Average Weekly Earnings statistics, as part of our Average Earnings Index reports. However, these are less comparable than the new statistics, so the comparison was moved to our earnings statistics report to provide a fairer comparison between Jersey and the UK.
Data for the UK is from the ONS and HMRC Earnings and employment from Pay As You Earn Real Time Information, non-seasonally adjusted dataset. This information is broadly comparable with that produced in these reports. The main difference being that in Jersey we undertake FTE adjustments, whereas the UK statistics are on a headcount basis (that is, not adjusted for FTE). As a result earnings levels are not strictly comparable, but annual changes are comparable.
These PAYE-based UK earnings and employment statistics were designated official statistics in 2025, having previously been experimental statistics. This follows a similar timeline as our work on improving earnings statistics.
Some Jersey sectors are compared to a weighted average of UK results. This is due to small sectors being merged in Jersey to avoid disclosure issues with aggregated statistics. The sectors affected by this are: construction and quarrying; education, health and other services; utilities and waste; miscellaneous business activities.
Gender pay gap methodological changes
The current gender pay gap methodology used since the June 2024 report differs from that in our earlier experimental reports. The current methodology focuses on hourly rates of pay, which is considered best practice for gender pay statistics. In experimental analyses up to June 2023, we did not have the capability to calculate hourly pay and instead used overall monthly earnings, adjusted for full-time equivalency.
While this earlier approach partially accounted for differences in full-time and part-time employment between men and women, it did not adjust for variations in hours worked among full-time employees. The new methodology addresses this issue. For example, under the new method, if two employees are paid the same hourly rate but one works 40 hours per week and the other 38, no pay gap is recorded. Under the previous method, both would have been classified as full-time, leading to an artificial pay gap.
On average, men work more hours than women, both overall and within each sector. As a result, the current methodology generally produces slightly lower gender pay gap estimates compared with the approach used before June 2024.
Administrative changes
In February 2025 the Government of Jersey implemented the Contributions Function Integration (CFI). CFI moved Social Security contributions data from Employment, Social Security and Housing to Revenue Jersey’s Revenue Management System.
These changes impacted contributions data from July 2024 onwards, with an approximate reduction in the number of jobs in our final dataset by 0.4% and a reduction in median earnings by around 0.9%. As such, revisions to previously published results relating to January 2022 to June 2024 have been necessary to maintain a consistent series.
Definitions
These reports define the public sector using section O of the Standard Industrial Classification 2007, which is public administration and defence, and compulsory social security. This is based on the main activity of undertakings, and not their legal or ownership status. Consequently, not all government bodies are automatically classified in section O. Table 1 details which undertakings are contained within the public sector in these reports and the Labour Market report.
Data tables
Data tables and supplementary information on earnings statistics can be found on our open data site: Earnings in Jersey – Datasets – OpenData
Data tables and supplementary information on gender pay gap statistics can be found on our open data site: Gender pay gap in Jersey – Datasets – Open Data