In a literature review focusing on recent work, A. Roulet presents solutions to improve the matching between supply and demand on the labor market. A clear and concise work, which offers many avenues for reflection.
From the “Macron Ordinances” of September 2017 to the unemployment insurance reform announced for the summer of 2019, measures are multiplying to reduce structural unemployment in France and show that, against unemployment, we have not yet tried everything. This question is all the more important since, as A. Roulet reminds us, contemporary economies are faced with the fear of the “end of work” (J. Rifkin, 1995) and a growing polarization of the labor market: it is therefore necessary to ensure that, whatever their level of initial training, individuals will manage to integrate sustainably into the labor market. Indeed, matching problems, i.e. the connection between supply and demand for labour, between the unemployed and vacant jobs, result in both high unemployment – when there is a mismatch between the characteristics of the unemployed and those of the vacant jobs – and shorter-term employment, due to a mismatcha mismatch between a worker’s skills and those required by his job. A. Roulet’s work seeks to identify the factors that hinder and deteriorate the match, to quantify their impact and to propose an evaluation of existing measures to overcome these problems.
Traditional levers
The first element – and perhaps the most obvious – that can affect the meeting between supply and demand for work is geographical mobility (both between employment areas and within an employment area). While A. Roulet notes that the latter is lower in France than in Anglo-Saxon countries, it is nevertheless very difficult to quantify precisely the share of unemployment explained by the absence of mobility (p. 37). To analyze the obstacles to mobility, she relies on a survey conducted with D. Glover among job seekers registered with Pôle Emploi. The first observation is that of a lack of information: 74% of respondents do not know that Pôle Emploi offers mobility assistance. They then show that the obstacles to mobility, both residential and commuting, are the fact of being a homeowner, the level of education and especially gender. Indeed, women, especially those with children, are much less likely to accept a job offer far from their place of residence. For A. Roulet, while the policies implemented should not hinder geographical mobility, the latter is however not the main reason for matching problems and goes beyond the framework of strict employment policies (attachment to housing, childcare, etc.): it is therefore not the main lever for action.
The second aspect addressed is that of unemployment insurance, and in particular the duration of compensation. Indeed, this has a double impact on matching: it can limit the return to employment of the unemployed, but it also allows them to wait for a job offer that better corresponds to their qualifications, which improves the quality of matches. For A. Roulet, there is a relative consensus in the economic literature on the first aspect: on average, a 10% increase in the duration of compensation leads to a 1% increase in the time taken to return to work (p. 56). On the other hand, she shows that the results are more varied with regard to the quality of the job found (in particular in terms of salary), and concludes that matching is not improved by a longer duration of compensation. At the end of the chapter, we find the conclusions of P. Cahuc and S. Carcillo in a work from the same collection: increasing the legal duration of compensation is not the solution. On the other hand, like P. Cahuc and S. Carcillo, A. Roulet suggests thinking about the counter-cyclicality of unemployment benefits, the idea being to adapt the duration of compensation to the economic situation. In the high phase of the cycle, matching difficulties are less since employers are more inclined to hire, so it is necessary to encourage the unemployed to find a job quickly by limiting the duration of compensation. Conversely, in the event of a recession, there is no job available, so it is necessary to protect the unemployed by providing longer compensation. The second option is to introduce a “bonus-malus”: employers who have greater recourse to a succession of short contracts – which multiplies the periods of unemployment compensated for their employees – should pay additional social contributions in order to bear the cost that their practices generate on the financing of unemployment insurance.
The third aspect addressed is that of training the unemployed, a point that is undoubtedly the most central in the face of contemporary challenges. Due to the importance of biased technical progress, which accentuates inequalities between skilled and unskilled workers, and the risk of the disappearance of a portion of automatable jobs (which represent 10% to 50% of jobs depending on the study), the challenge of qualifying job providers is major. According to A. Roulet, however, the evaluations conducted all lead to a similar conclusion: continuing training programs in France are overall ineffective, all things being equal. This therefore raises the question of the cost/effectiveness balance of these measures (25 billion euros according to the DARES in July 2018). While the observation is clear, A. Roulet quickly passes over the proposals for improvement, only mentioning the idea of companies participating in defining training programs (so that it is in line with their needs). One of the reasons for the failure of vocational training is that the diploma is not a sufficient criterion to explain the difficulties in matching supply and demand for work: the ability to adapt or the ability to work in a team are essential factors.
The contributions of neuroscience and big data
The real contribution of the book is its analysis, in the last chapter, of more recent work on the role of neuroscience or big data to improve matching in the labor market.
A. Roulet mentions the possible existence of cognitive biases of job seekers: they would be more optimistic (and therefore underestimate the time spent unemployed) and would have a significant preference for the present. In both cases, this can lead to a lower search effort when losing a job, which would limit the possibilities of matching. Furthermore, as the duration of unemployment increases, a problem of loss of confidence arises which encourages the unemployed to apply for job offers that do not correspond to their qualifications. This can harm the quality of the matching. There is currently no serious empirical research on the subject, but it is an interesting avenue for future work.
The role of the big data is further developed. Indeed, from the beginning of the work, A. Roulet highlights the information problems that can weigh on matching. The labor market, like any market, is characterized by transaction costs, and in particular information search costs, borne by both job seekers and job providers. The big data can help to overcome this lack of information: this is notably the role of suggestion algorithms that help employers to select employees for very specific tasks. The first studies evaluating these practices show that they are effective in cases where qualifications are very specific (and therefore where search costs are a priori very high). A. Roulet finally mentions the role of standardized tests in recruitment procedures that are developing thanks to computerization. These are generally effective, and all the more so because they have a latent function: they limit discrimination, particularly ethnic discrimination. However, this requires constantly developing these tests and maintaining a “human intermediary” (p. 79).
A phenomenon that is difficult to quantify
While A. Roulet’s work allows for the exploration of many avenues, the analysis, due to the format, remains succinct. Indeed, the question of matching refers to multiple dimensions that cannot be dealt with in less than 100 pages without skimming over them. This leads to two pitfalls. First, it is an introductory work, which requires appropriating the existing literature. Second, multiple avenues for reforming the labor market are mentioned, without any real hierarchy. It is therefore in fine It is difficult to know how to concretely improve matching on the labor market in France.
Furthermore, and A. Roulet acknowledges this, it is very difficult to measure the true impact of matching difficulties, both on unemployment and on the quality of employment. Indeed, if 300,000 offers submitted to Pôle Emploi were not filled in 2017, France has 3.5 million job seekers who are not working. Filling these vacant jobs would therefore not be enough to significantly reduce unemployment. Furthermore, identifying the mismatch is particularly complex: depending on the type of survey, we go from 10% to 22% of individuals overqualified for their job in the countries of theOECD. Therefore, the analysis of the causes of mismatches, as well as the evaluation of any policy seeking to limit them, is particularly difficult.
Addressing so many elements in such a short format therefore inevitably leads to certain shortcuts, but such a work allows us to take stock of the most recent literature, and to evoke the main lines of thought on the question of matching, before exploring the subject in more depth through other readings.