Few would argue that they would rather “work harder” than “work smarter”. Yet, the indicator that measures smart working – productivity – shows at best sluggish growth since the financial crisis. Belgium for instance has experienced little productivity growth in recent years and has only increased its labour productivity by ~5% over the decade since the financial crisis. This seems to be at odds with the ever-increasing use of digital technologies that (at least in the perception of most people) increasingly replace and support human tasks.
This paper, part of the OECD’s “Human Side of Productivity” project, opens the firm’s black box and goes beyond the often used (financial) firm-level characteristics to study a firm’s human capital and its link with productivity.
Human capital will become even more relevant post-COVID19. The NextGenerationEU recovery package rightly puts significant emphasis on research, innovation, and digitalisation. The need for skilled people to deliver on these promises, however, gets little attention. Stimulating the demand for innovation without addressing the supply of skilled workers might simply result in higher wages for the high-skilled rather than additional innovation. Whilst the number of tertiary education graduates in Belgium has risen steadily over the past decades and is above the EU28 average, the number of such graduates in science, technology, engineering, and mathematics (STEM) fields is below the EU28 average. At the same time, Belgian firms have a greater need for ICT specialists, for instance. The combination of these factors has led to a steep increase in the number of firms with hard-to-fill vacancies for such jobs. When firms cannot find the human capital that they need, this is likely to have an impact on productivity.
We study the link between the skills of a firm’s workforce and its productivity. To this end, we make use of linked employer-employee data and focus on the full Belgian universe of firms with 10 employees or more. We study approx. 1.5 million workers and 20,000 firms over the period 2000-2018. The skill level of an employee is based on educational attainment and categorised as high (tertiary education), medium (upper secondary education and post-secondary non-tertiary education) and low (lower secondary education or below). Firms are divided into productivity groups based on their position within the productivity distribution of their industry. We focus on the top performers or “frontier firms” (top 10%), medium performers (40%–60%) and low performers or “laggards” (bottom 10%). Productivity is measured via labour productivity (euro per hour worked).
A frontier firm is more than twice as productive as a medium performer and almost 5 times as productive as a laggard firm. Since 2000, this productivity gap has increased simultaneously with a skills gap. On average, the share of high-skilled workers as a percentage of total workers in a frontier firm is currently close to 10 percentage points higher than in a medium performer and 20 percentage points higher than in a laggard firm. The larger share of high-skilled workers in frontier firms is mainly compensated by a smaller share of low-skilled workers. Close to 10% of the Belgian population aged 18-25 years do not hold a secondary education certificate and are not in further training or education. We find that job opportunities for the lowest-skilled workers are mainly found in the least productive firms.
To control for a wide range of firm characteristics we use regression analysis. We find that a 10 percentage point increase in a firm’s share of high-skilled workers is correlated with an increase in productivity of 2% (for knowledge-intensive services), 6% (for manufacturing) and 7% (for less knowledge-intensive services). This impact on productivity has decreased over time. For all sectors combined, a 10 percentage point increase in the share of high-skilled workers was linked with an increase in productivity of 6.5% for the period 2000-2007 and 5.5% for the period 2012-2018. The reason could be that the overall number of high-skilled workers is increasing and the additional benefits of continuing to add high-skilled workers decrease the more high-skilled workers a firm already employs.
To deliver on the increased need for automation and digitalisation, there is also a need for workers with STEM skills. Although Belgium performs relatively well with respect to tertiary education graduates, its performance is poorer with respect to STEM graduates. For the manufacturing industry and the less knowledge-intensive services we do observe a clear, positive link between productivity and the share of STEM workers. For knowledge-intensive services, only the laggard firms employ a smaller proportion of STEM workers, and we see little difference between frontier firms and medium performers.
For STEM workers (high-, medium- and low-skilled) we find that a 10 percentage point increase in their share in a firm’s workforce is linked with a 2.5% (for manufacturing) or 4% (less knowledge-intensive services) increase in the firm’s productivity. But more importantly, unlike the impact of high-skilled workers that decreases over time, the impact of STEM workers on productivity is increasing. The average impact on productivity across all sectors combined of a 10 percentage point increase in the share of STEM workers has risen from 2.0% (2000-2007) to 2.6% (2012-2018). This could be linked with the increasing importance of digital technology for productivity.
Increasing the share of high-skilled STEM workers leads to significantly higher productivity gains, not only compared to non high-skilled STEM workers, but also compared to high-skilled non-STEM workers. For a typical manufacturing firm, the gains from increasing the share of high-skilled STEM workers by 10 percentage points is linked with an increase in productivity of ~20% or approx. 3 to 4 times more than the gains from a 10 percentage point increase in the share of high-skilled non-STEM workers. The growing difficulty that Belgian firms experience in recruiting specialist ICT skills is therefore likely to have a significant negative impact on productivity.
Considering the results presented in this paper and bearing in mind that they mostly reflect past correlations, we can still draw some policy recommendations from this empirical exercise. The main one is that policies designed to promote the adoption of the latest technologies and business practices within firms can only lead to sustainable productivity gains if they are combined with measures to increase the supply and mobility of human (STEM) capital. Without a proper supply of skills, firms will not be able to reap the full benefits of the digital revolution.
We also briefly touch on the link between the share of foreign workers and productivity. It is only for knowledge-intensive services that the share of foreigners is positively correlated with productivity. The most productive service firms generally rely on highly skilled foreigners with specific competencies. As it remains uncertain if and when the global mobility of high-skilled workers will recover in the wake of the COVID-19 pandemic, this could have long-lasting (negative) effects on the productivity of some knowledge-intensive service firms.