In recent years, VR has conducted a regular wage survey that publishes information on the wages of VR members in the various professions. The wage study is based on the data of more than 14,000 VR members who have registered their job titles and working hours on My Pages. The purpose of the wage survey is to monitor the general wage development of VR members, increase transparency and allow members to see how they stand compared to others in similar jobs.
The VR wage calculator enhances this service by providing members with even more detailed information about the salaries of different jobs. The calculator is powered by a statistical prediction model which VR has built and takes a greater number of factors into account which can explain differences in salary than previous wage surveys. It is now possible to enter information about job titles, sector, age, job experience, education and managerial responsibility into the wage calculator and receive an accurate prediction for a likely salary range.
All predictions are based on a dataset containing information about the salaries of 14,000 VR members who are working full time and have registered their information on My pages.
How does the prediction model work?
The prediction model returns a predicted salary range, as the name suggests, in line with the information entered.
This prediction is premised on the salary data from My pages. Unlike older wage surveys, the model does not calculate averages for certain groups but uses machine learning methods to identify statistical patterns in the salary distribution of VR members. In this way, the model can identify the ways in which different factors (like age, education, job title, etc.) influence salary levels. The model has in this way learned the salary tends to increase with age, job experience and education – and uses information of this kind in conjunction with existing data about average salaries of different positions to output its predictions.
Machine learning also enables the model to predict the salary of individuals independent of the fact whether there exists any VR member which fits the background information that is entered. For example, it is possible to have the model predict a salary range for a 20-year-old VR member who has worked as a cashier at a legal firm for five years (since the age of 15). We could further assume that this individual has not completed their compulsory schooling and that they are personally responsible for managing 50 employees.
This is of course a silly example – law firms do not generally have cashiers working at registers and twenty-somethings rarely manage employee groups which are the size of a medium-sized Icelandic firm. There is clearly no-body in the VR dataset with this background. Nevertheless, the statistical model can predict the salary for such a hypothetical example precisely because it does not rely on calculating simple averages but by understanding the ways in which different factors influence salary levels. For what its worth, the model predicts that this outstanding cashier would have a monthly salary between 600 and 800 thousand ISK.
This example shows how the new prediction model has more flexibility than older wage surveys. It is however important to remember that the results of the wage calculator are not infallible or obtained with 100% certainty. The model does however give a solid indication of the likely salary levels for different positions.
VR encourages all members to try the new wage calculator, which is available on My Pages.