Published as Großkurth, P. (2019a) MNE and where to find them: an Intertemporal Perspective on the Global Corporate Ownership Network. Ruhr Economic Papers 825, RWI 10/2019.
The third post in this series investigates MNE growth and strategy, highlighting organizational preferences and spatial growth over time. There are common patterns: the average MNE is small, organized as flat as possible, and connects few countries. Internationalization, however, can happen in very different ways. As long as an MNE has subsidiaries in only a few countries, both flat and deep hierarchical structures are employed. On average and for larger numbers of connected countries, MNE seem to converge towards flatter hierarchiers. Growth, on the other hand, happens either very slow (expansion – the creation of new subsidiaries) or extremely fast (combination – M&A and the like).
Part 3) Growth and Strategy
Is it possible to accurately print the time series data of 329.376 MNEs? The next figure tries to do exactly this, overlaying the expansion paths of all MNE in my dataset both in terms of the countries and the firms they connect. To make the average pattern visible, I reduced the opacity of the individual lines into the single digits. The way Stata handles overlapping lines then brings out the average. Probably not a perfect solution, but good enough to understand what’s going on in this massive dataset.
Both panels of the figure also highlight the large amount of fluctuation in the data – if a line jumps unreasonably from one year to the next it’s fairly safe to assume that something fundamental has happened to either the firm or the way the firm is captured in the database. The results suggest that most MNE stay small and regional, a point that the next figure will clear up even further.
A key insight here is that the average MNE is much smaller than their reputation suggests and that MNE grow faster in terms of the number of connected firms than they do in terms of the number of connected countries.
In each year firms enter the dataset and firms leave it, which explains the difference between the panels in the figure above. On the left side averages are calculated in each year of the database. Due to the large turnover of firms, the average size of an MNE – both in terms of the connected firms as well as the connected countries – seems to stagnate. If the data is rearranged, however, to align firms according to the number of years that we observe them in the dataset, the picture is strikingly different. Firms grow in both dimensions and they grow considerably. The fluctuations at the end of the figures are explained by input lag on the left side and reduced data availability of firms with long time series on the right.
However, while this suggests that firms grow over time it’s important to keep in mind the ever present process of creative destruction. Most firms aren’t observed continuously and data availability grew over time. Taking this into account, the next figure highlights the massive turnover of firms.
In this case what’s true for firms in general also seems to be true for MNE in particular. The data also points toward a process of group size consolidation over time, but that’s a story for another time.
Aside from these more general conclusions the dataset also allows for an in-depth investigation of the MNEs’ organizational structures and strategies. The figure below highlights the relationship between group size, connected countries, and hierarchy depth (the number of links between the global ultimate owner and the furthest removed subsidiary in the group). What’s immediately striking is how slow the average rate of geographical expansion is. The Y-Scale is at a factor of about 1/10 of the X-Scale, meaning that MNE generally increase the number of subsidies in a country they are active in before they move to another. Likewise, MNE favor flat hierarchies over deep ones (aside from a few outliers who only operate in relatively small numbers of countries), a result that doesn’t seem to change all that much with increasing group size.
