Small area estimation (communes) of economic activity rate in the structural survey

| Last update: 19.11.2021

image – experimental statistics


The structural population survey provides important information on the population, including information about work. By means of a sample survey of at least 200 000 people each year in Switzerland, it is possible to make reliable estimates of the economic activity rate of groups of 15 000 inhabitants.

By combining the surveys of several years (pooling over three or five years), it is possible to reduce this limit to groups of 5000 or 3000 people. This can be done using standard methods.

In 2016, Switzerland had almost 2300 communes. Of these, 83% had fewer than 5000 inhabitants. These communes represent 34% of Switzerland’s total population. 70% of communes had fewer than 3000 inhabitants (21% of the population). The question addressed here is whether it is possible to reduce the limits mentioned above even further to produce reliable estimates by means of new methods that are based on statistical models, which use information that is available for the whole population.


The FSO’s objective is to publish reliable estimates also for communes whose annual sample size is insufficient when applying standard methods.

The whole purpose of Small Area Estimation (SAE) is to push the boundaries imposed by standard methods by using techniques based on modelling which make use of complementary information on the population.

Professor I. Molina from the University Carlos III of Madrid, a specialist recognised internationally for this type of methods, was commissioned to examine the possibility of obtaining reliable estimates of the economic activity rate at commune level. This study, based on the 2012 structural survey and on OASI (Old-age and survivors’s insurance) data from 2011, showed that it is possible to obtain reliable estimates for both annual economic activity rates, as well as their precision (design MSE), for communes that had a sample of at least 100 people, i.e. considerably smaller than when using standard methods.


Following the encouraging results obtained with the SAE methods using the 2012 structural survey and OASI data from 2011, the FSO then took the next step of investigating the possibility of combining this information with data pooled from the structural survey over several years. These estimates referred to an average target population from the structural survey, once for the three years 2012–2014 and once for the five years from 2012–2016. The main objective was to obtain reliable estimates for a larger number of communes than could be obtained from annual analyses. The lower limit of 100 persons in the sample remains a priori the same for the data pooling.


Based on the conclusions of the study commissioned to Professor I. Molina, for the structural survey the FSO decided to publish only estimates for communes with a sample of at least 100 people. To have a sufficiently large sample and to be able to publish results for as many communes as possible, results are based on structural survey data pooled over three years, i.e. from 2012 to 2014. In contrast to standard methods, which enable results to be published for only 20% of communes, it is now possible, thanks to these new methods, to triple this figure, covering 60% of communes. At the end of 2014, this represented 94% of the permanent resident population in Switzerland aged 15 and over.

The BFS checked the results from the structural survey pooling over the three years 2012 to 2014 with a five-year pooling from 2012 to 2016. It can be said that results can be published for a larger number of communes and these are overall more stable. With a pooling over five years, estimates can be made for up to 76% of communes. At the end of 2016, this represented 98% of the permanent resident population in Switzerland aged 15 and over.

Communes Pooling from 2012 to 2014 Pooling from 2012 to 2016
number percentage number percentage
Total communes 2351 100% 2287 100%
Unpublished results 924 39% 540 24%
Published results 1427 61% 1747 76%

The results can be found in the two tables below:

Small area estimation results from the five-year pooling of 2012 to 2016 are also available on the following interactive map:


Results should be interpreted with caution, as they are estimates based on an average population in the structural surveys of 2012 to 2014, or from 2012 to 2016, respectively.


In future, it is important that we transfer our findings and experiences from the application of small area estimation into the ongoing production of the structural survey. A pooling of structural survey data over five years with the publication of estimations every five yours could be a promising approach.