The use of firm-level data to assess national competitiveness
The worldwide financial crisis of 2008 and the ensuing economic slowdown exposed substantial imbalances within the Eurozone. As a result, national competitiveness appeared high on the agenda of EU policy makers, e.g. the Europe 2020 Strategy. The definition of competitiveness is a subject of much debate, but productivity is generally acknowledged as an important indicator of the ability of countries to create value added. Productivity is often computed at a rather aggregated level (industries, regions or countries) under the assumption that industries are composed of fairly homogenous firms. Firm-level data, however, reveal dramatic and persistent differences in productivity between firms within the same industry. In this article, we argue that firm-level data give insight in intra-industry dynamics that is beyond the scope of analysis at a more aggregated level.
Productivity provides a measure of the efficiency in using available technology to convert inputs into valuable outputs. The productivity growth of a country is a weighted average of the productivity growth of its industries and the productivity growth of an industry, in itself a weighted average of the productivity growth of the firms that the industry is composed of. If the productivity of firms within the same industry differs, as data clearly bear out, industry-level productivity may change due to the reallocation of market shares, even without any change in the efficiency of firms. In a recent study, firm-level data were used to estimate the productivity of companies in ten manufacturing industries for seven EU countries (Belgium, Finland, France, Germany, Italy, Spain and the UK) over the period 2002-2009. For each industry, a metafrontier was estimated, which shows the highest value added that can be generated for a given set of inputs (labour and capital). The distance of a firms’ output from this frontier provides a measure of its technological efficiency (productivity). In most industries, the technology frontier is determined by Belgian firms. Firms from Finland, France and Germany are also highly efficient whereas Spanish firms clearly lag behind. The orange bars in Graph 1 show the average level, across the ten manufacturing industries, of technical efficiency in 2005. The average level of France is set at 1 and the levels of other countries are relative to the level of France. This permits comparison of the average efficiency level of manufacturing industries, estimated using firm-level data, with the productivity level, based on industry-level data from the Groningen Growth and Development Centre (GGDC) Productivity database. The multi factor productivity (MFP) levels for 2005 take into account cross-country differences in purchasing power parity. The level of the US is set at 1 as the benchmark. The MFP levels for total manufacturing (excluding electrical machinery, post and communication) are shown with blue bars in Graph 1. By coincidence, the MFP level for France was 1 (i.e. equal to the US) in 2005.
There are only minor differences between both measures. The relative MFP level of Belgium is somewhat lower and that of Finland somewhat higher than productivity estimated using firm-level data, though Belgium holds first place for both measures. Except for France and Germany, which switch places despite marginal differences, the ranking of the countries is the same for both efficiency measures.
Studies based on firm-level data show that there are substantial differences in performance (e.g., productivity) of firms in the same industries. These differences appear to be rather persistent. Changes in the performance of industries result from changes in the performance of firms but also from the reallocation of market shares between existing firms (incumbents) as well as from entry and exit. In the year that new firms start their activities, they often have a productivity level below the average of incumbents. It takes some time for starting firms to raise their efficiency, suggestive of learning effects. Graph 2 shows the distribution of technological efficiency (productivity) of Belgian firms over the period 2002-2009 for the ten manufacturing industries considered. The productivity level of firms is related to the industry average to account for differences in productivity across industries. The graph shows the distribution of the productivity of new firms in the year that they enter and of firms that exit, in the year before they exit. To assess the possible role of learning, incumbents are grouped by age. The graph shows the distribution for starting firms, i.e. firms that have not been active for more than 5 years (excluding the year of entry); young firms (active for more than 5 but not more than 10 years); and mature firms (active for more than 10 years). The more the bulk of the distribution is to the right, the higher the efficiency of the group relative to industry average efficiency. On average, mature firms are more efficient than the other groups of firms and firms that enter or exit are, on average, the least efficient. The distribution shifts to the right with firm age, indicating increasing efficiency of firms that comes with experience (and exit of less efficient start-ups). There are a small number of entrants, starters, and young firms among the most efficient firms.
This small group contributes disproportionally to industry performance. The graph shows that the notion of a representative (average) firm is rather tenuous and that there is a need to consider the entire distribution of firm performance.
The analysis for seven EU countries indicates that firms seem to learn rather swiftly in Finland and a little more slowly in Belgium and Germany whereas in Italy and Spain, firms fail to catch up with existing firms, even within ten years after entry. The most robust finding is the positive contribution of efficiency growth of young firms to industry-level productivity growth.
The main advantage of firm-level data is that they provide insight into the dynamics within industries that is beyond the scope of more aggregate data. They reveal substantial and persistent differences in technological efficiency between firms in the same industry. Young firms need time to raise their technological efficiency but gradually contribute substantially to industry-level productivity growth. The performance of start-ups is highly heterogeneous. Only a small number of entrants are able to grow considerably whereas the rest either exit at an early stage or remain relatively small. Post-entry growth is probably more important than the relative number of entrants. This calls for a more in-depth investigation of factors that may hamper post-entry growth, such as imperfections in product and factor markets (e.g., credit constraints), obstacles to the transfer of knowledge and technology or established buyer and supplier networks that prevent start-ups gaining a strong market position.