Productivity growth and global value chain participation in the digital age


Productivity growth and global value chain participation in the digital age

By Claudio Battiati, Cecilia Jona-Lasinio, Silvia Sopranzetti

Labour productivity growth has been declining in advanced economies since the beginning of the 1970s, experiencing a more pronounced fall after the Great Recession. The causes of the slowdown remain a matter of debate especially in the more recent years for those countries that are increasingly involved in the digital transformation and in global value chains (GVC), both assumed to generate productivity gains.

In our paper (full version available here), we investigate whether the reorganisation of the production activity, as captured by the extent of GVC participation and sectoral digitalisation, offers new insight about the drivers of productivity growth over the last fifteen years in the modern economies.

Our analysis starts with an overview of labour productivity trends for 12 European economies and the US in 2000-2015. We offer some evidence on the extent of sectoral productivity growth and digitalisation across countries looking at three groups of sectors, “high” (HD), “medium” (MD) and “low” (LD) digital-intensive industries as in (Calvino et al., 2018). The data show that in the post-crisis years, HD sectors expand faster but MD industries account for the largest share of value added while LD display heterogenous growth patterns in the sample economies. The slowdown is widespread across sectors and countries, with a few exceptions including HD sectors in Germany, and MD sectors in Denmark, Italy and Spain.

Next we include empirical evidence on GVC participation (based on WIOD data between 2000 and 2014), considering two modes of participation: forward (domestic value added in foreign export); and backward (foreign value added in domestic exports). Figure 1 shows a positive correlation between productivity growth and GVC participation, with seemingly stronger linkages for forward compared to backward participation, and weaker linkages for LD than for both HD and MD sectors.

We further explore this relationship by estimating a standard production function augmented with measures of backward and forward participation and resorting to instrumental variables. Our findings suggest that both modes of GVC participation positively and significantly affect productivity growth, with forward linkages exerting a stronger impact compared to backward participation.

To establish the size of these effects, we quantify the contribution of participation to labour productivity growth. Given an annual rate of productivity growth of 0.015 percentage points, forward participation accounts for 0.008 percentage points, while backward linkages account for only 0.002 percentage points on average. When we further distinguish the degree of digital intensity, we find a stronger impact of GVC participation for MD sectors while we do not find evidence for backward participation in HD sectors. Indeed, digitalisation is supposed to exert stronger productivity effects for industries that are intensive in routine tasks (Gal et al., 2019). As the MD sector is mainly composed of manufacturing industries, the positive effect on productivity in these sectors seems to magnify the gains from GVC integration. We also find that forward participation generates a relatively stronger effect on productivity growth, compared to backward participation. As a first approximation, we assume that the larger productivity-enhancing effect of forward linkages depends on the implied possibilities of accessing larger markets and exploiting a finer division of international labour.

Our results support the assumption that the increasing relevance of GVC participation together with the digital transformation significantly affected productivity growth over the last fifteen years, thus suggesting that both these (interrelated) phenomena might have profound implications for restoring economic growth.

ESCoE blogs are published to further debate.  Any views expressed are solely those of the author(s) and so cannot be taken to represent those of the ESCoE, its partner institutions or the Office for National Statistics.

About the authors

Claudio Battiati

Silvia Sopranzetti

Related publications