Algorithms that can change the game of job seeking


Paul Muller, Assistant Professor at the department of Economics at the School of Business, Economics and Law, University of Gothenburg, is one of three authors of the study that has been accepted for publication in the esteemed journal ‘Review of Economic Studies’. The study, “Providing Advice to Job Seekers at Low Cost: An Experimental Study on On-Line Advice”, suggests that giving advice through the job search engine about other occupations that may be of interest to the job seeker was effective in regards to getting employed.

What do you consider to be the most interesting finding in the study?

‘The most interesting finding is that by making small changes to the way a job search website works we can really affect the way people search for jobs. There is a big impact on the kind of vacancies that people are looking at if we, through the underlying algorithms give them a little bit of advice on different occupations on the web site. What we find is suggestive evidence that this advice is helpful. At least certain groups of job seekers, particularly those who were unemployed for longer, were more likely to get invited to job interviews during the time of the study.’

Why is this study important?

‘I think it is important because it considers the policy question of how to minimize unemployment duration. That is a very important issue for public finance but also from a personal perspective because it has been shown that long-term unemployment has lots of negative consequences. It is crucial to consider how to optimally assist job seekers. The traditional way would be to have more meetings and advisors at the job centers and so on. These are valuable as well of course, but if this is helpful it would be a much simpler way to give very personalized and cost effective information to job seekers.’

What effects do you think the findings from the study may have for the process of job searching?

‘In a broader sense I think and hope that this study helps to make information interventions more common. Job agencies in several countries are already offering these kind of services where on their websites you can get suggestions such as “If you are interested in this you might also be interested in this”. And I think this could be explored much further to make sure that the kind of information that is provided on these websites is as useful as possible. The key idea of the paper is that on the one hand we want to give as good as possible information to job seekers and where to search for jobs, but on the other hand it should also be very easy to access and cheap to implement.’

How about the effects on Sweden?

‘It is hard to say what the effects may be. I have presented this to “Arbetsförmedlingen” at a workshop considering different ways of assisting job seekers. This is definitely one of the things they are thinking about and experimenting with. So I very much expect this to be more common as a policy in many countries all over Europe.’

What is the way forward from this study?

‘I think there is a lot to be done in improving the algorithm so that the actual advice that we give is better, for example by using bigger data sets than we have done in this first pilot study. This study is based on three hundred people which is a small sample, but we are currently in a process of trying to do this on a much larger scale in the UK and maybe in other countries. If we do that, we can get more conclusive evidence on how it affects the probability of getting a job, and we can also find out more on who is most likely to benefit. Perhaps this information is more useful to particular occupations and less useful for others? There is a lot of refinement to be done.’

You can read the discussion paper of the study (not final version) here.