The Swiss Earnings Structure Survey (ESS) is a survey carried out every two years among private and public enterprises and administrations in Switzerland. It allows a regular description to be made of the earnings structure in all economic activities of the secondary and tertiary sectors based on representative data. It not only enquires after the economic activity and the size of the enterprise concerned, but also the individual characteristics of employees and their jobs.
The results of the Swiss Earnings Structure Survey (ESS) are available on the relevant topic pages:
The Swiss Earnings Structure Survey (ESS) covers salaried jobs (according to the domestic concept) in secondary and tertiary sector enterprises with at least 3 jobs. Since the start of the ESS (1994), data on wages in the federal administration and enterprises have also been collected. Since 1998, the survey has also provided data on wages in the public sector at cantonal level and since 2006 on communal administration earnings and since 2012 those of churches.
The ESS takes into account all employees who received wages in the month of October. Data on the wages of temporary workers are supplied by the temporary recruitment agencies.
Before 2010, the survey did not take into account apprentices, interns, employees working at home, persons paid only on a commission basis, or persons whose professional activities took place mostly abroad as well as persons whose wages were reduced in relation to their work-time percentage (e.g. SUVA, invalidity allowance etc.). From 2012 all these categories of employees are included.
The Swiss Earnings Structure Survey (ESS) is based on a complex random sampling procedure. For this reason, enterprises (private and public) are divided into strata according to three criteria: business size (3 classes), economic activity (39 sections) and geographic location (7 major regions). Specific criteria apply to the collection of the public administration sample. The sample size is calculated so that the coefficients of variation are, if possible, less than 3% for each of the statistics of interest, the main ones being the media wage by major region, by economic activity and by major region and economic activity combined.
The number of wages an enterprise, administration or commune has to report depends on its size. Companies with fewer than 20 employees provide information on all wages. Enterprises with 20 to 49 employees report one in two wages, those with more than 49 employees one in three. The cantons and towns can apply for a supplementation of their sample.
Until the ESS 2018, sampling was based on the Business and Enterprise Register BER and from ESS 2020 onwards, on the Swiss Business and Enterprise Register SBER. In principle, a full survey is made of enterprises with more than 50 employees. The sample comprises some 45 000 enterprises.
Although this type of survey method reduces the burden on data suppliers, it complicates the analysis of data. The information collected can no longer be analysed directly, it has to be weighted according to a formula that takes into account the response rate within the stratum and within the enterprise.
The gross response rate for the Swiss Earnings Structure Survey (ES), i.e. the number of enterprises that respond out of all the enterprises in the sampling frame, is around 75%.
The 95% confidence interval is a measure of the accuracy of a statistic. A confidence interval specifies an estimated range of values that an unknown population parameter should contain. The estimated range is calculated from a given sample. If independent random samples are repeatedly drawn from the same population under the same conditions and a confidence interval is calculated each time, then a certain proportion of these confidence intervals will contain the unknown population parameter. Confidence intervals are usually calculated so that this proportion is 95%. This is referred to as a 95% confidence interval. The width of the confidence interval provides an indication of the accuracy of the estimate for the unknown parameter. A very wide interval may indicate that a larger sample should be collected to provide a reliable estimate of the parameter of interest. The confidence intervals for the median are not symmetrical. Their calculation is described in the FSO publication "Précision du salaire brut standardisé médian" and provides a lower limit value, an upper limit value and a synthetic coefficient of variation (CV).