Cross-Sectoral Reallocation
We analyze the productive structure of the EU member countries and its evolution over time; given the need for a sufficiently long period after the introduction of the euro and data availability, we focus on the EU 15 countries-that is, the eleven that adopted the euro on its inception (Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, Portugal, and Spain) plus Greece (entered the EA in 2002), in addition to Denmark, Sweden, and the United Kingdom, which have not adopted the euro.
Following Bertola (2007), the three non-EA countries constitute the control group. Despite its evident shortcomings, this is the best control group available. We assess whether the introduction of the euro has induced a reallocation of production between sectors, and if so, whether the intersectoral change has been more dramatic in the countries that had previously made greater use of competitive devaluations. The main data source we rely is the March 2008 release of the European Union Level analysis of Capital, Labor, Energy, Materials, and Service Inputs (EU KLEMS) database (Timmer, O’Mahony, and Van Ark 2007). The manufacturing sector’s share of value added in 2005 stood at around 20 percent for most countries, with lower values in France, Denmark, Greece, and Luxembourg. Following the secular decline in manufacturing, the share decreased somewhat between 1998 and 2005 in most countries; Ireland and the United Kingdom experienced the most pronounced downsizing of the sector.
From now on, we concentrate on manufacturing, as the effects we are considering work through the terms of trade and so are important mostly for tradable goods. Data on value added, employment, and capital stock for the manufacturing sector are available for all EU15 countries, with a breakdown into twenty- two industries corresponding as a rule to the two-digit NACE classification. Southern countries such as Italy, Greece, and Portugal still have a large share of their value added in traditional sectors, such as textiles, apparel, leather goods, and footwear. The other countries concentrate their production in more technologically advanced sectors: machinery in Germany (but in Italy, too), chemicals in a host of countries (Belgium, France, Germany, the Netherlands, Ireland, and the United Kingdom), and radio, television, and communication equipment in the Nordic countries (Finland and Sweden, in particular).
In order to facilitate the comparison of productive structures among countries and over time, we first characterize sectors by their skill, R&D, and information and communication technology (ICT) intensity, and then we group them into intensity classes. Figures are computed from U.S. data, which we use in the regression analysis to avoid problems of endogeneity.
Skill intensity is proxies by hours worked by high- skilled persons-defined as those with at least a college degree-as a share in total hours; R&D intensity is R&D expenditure over value added; ICT intensity is the ratio of ICT capital stock to the total capital stock, both in real terms.
The machinery and the electrical and optical equipment sectors exhibit the highest ICT content; together with “other transport equipment,” they spend a relatively higher fraction of their value added on R&D and employ relatively more- skilled persons. As a rule, traditional sectors (producing food, textiles, leather, and wood products) are characterized
by low values of the three indicators. Intensity classes (low, medium low, medium high, high) are then defined according to quartiles in the distribution of each indicator. A glance at the value added shares broken down by skill content in 1998 and 2005 suggests
that sectoral modifications were modest in the period. Only in Finland and Sweden has reallocation toward high- skill activities been substantial; Ireland stands out as the country where high- skill activities are prominent; if anything, Italy and Spain have increased their share in low-intensity activities.
To address sectoral modification in a more synthetic way, we apply standard techniques of convergence/ divergence of productive structures. In particular, we calculate bilateral dissimilarity indices based on value added shares, broken down by industry and by skill, R&D, and ICT intensity.
The most highly dissimilar countries-apart from Ireland and Luxembourg, which are exceptionally small-are the southern countries still specialized in low-skill activities. There is no sign of a uniform tendency toward either convergence or divergence: some countries increased and others decreased their similarity with the rest of the area. We also evaluate for each country the dissimilarity index between 1998 and 2005 to assess the extent of intersectoral change over the period. Irrespective of the sectoral breakdown, the extent of sectoral reallocation proves to be fairly modest. The dissimilarity index never goes beyond the first half of its range. The countries that changed their structure most are
Sweden and Finland, followed by Greece. It is interesting to see whether the degree of intersectoral reallocation, though mild, is related to competitive devaluations. We construct two measures of devaluation, nominal and real (Devnom and Devreal, respectively), calculated as the cumulated difference between January 1980 and December 1998 of the logarithm of each country’s nominal/ real effective exchange rate as a deviation from that of Germany. In principle, a negative sign indicates depreciation relative to the DM; the absolute number refers to the intensity of the cumulative depreciation or appreciation. But for ease of interpretation, we invert the signs so that a higher value of the indicator reflects more intensive resort to competitive devaluations. The difference between the two
We find that when devaluation is measured in nominal terms, the countries relying most heavily on devaluations are those most specialized in low-skill activities. This positive relationship vanishes when we consider devaluation in real terms. We also find some weak evidence that countries relying more heavily on devaluations exhibit relatively more pronounced signs of intersectoral reallocation, where we plot the dissimilarity index between 1998 and 2005 against real devaluation; this evidence does not depend on the choice of the indicator (nominal versus real and different sectoral breakdowns).
On the whole, we can conclude that the euro has not induced a structural break in member countries’ specialization patterns. Let us now move on to assess whether a process of within-sectoral restructuring characterized EA firms in the first part of this decade, and in particular, whether this process was driven by the introduction of the euro, which eliminated competitive devaluations.
