Growth potential, labour productivity and structural reforms in Europe

Speech by Jean-Claude Trichet, President of the ECB
Annual Dinner Lloyd’s,
London, 10 November 2005

Ladies and gentlemen,

I am grateful to Lord Levene for inviting me here today and allowing me to share with you my views on some key issues for the euro area economy. The topic of my speech will focus on the recent growth performance and the outlook for the euro area economy in the medium to long term, in particular by drawing comparison with the other major economy of the world, the United States. Starting with a diagnosis of the growth performance of the euro area, I will then try to identify the reasons why Europe has been lagging behind in the recent years: in this respect, it appears that the gap with the United States can largely be explained by a difference in productivity growth on each side of the Atlantic. Finally, I will also touch upon the challenges and the remedies for the euro area ahead, namely the absolute need to resolutely implement structural reforms.

Relative growth performance

The economic well being of the European people, as elsewhere, depends on a number of factors, but one fundamental factor is represented by the amount of goods and services that are available on average for each of us, that is, real output per capita. Let me discuss the development of this key macroeconomic measure for the euro area and the forces which can explain its evolution.

A comparative perspective with developments in the US can offer useful insights. Indeed, from a long term perspective, a comparison of real output per capita developments between the euro area and the US is not always unfavourable for Europe. For example, since the early 1960s real output per capita growth has been on average slightly higher in the euro area than in the US.[1] From 1980 onwards developments have been on average similar, with real output per capita growth being on average just below 2% in both economic areas. However, these average figures conceal very different patterns. While in the euro area real output per capita growth has been gradually declining over the past four decades from almost 5% during the 1960s to just below 2% in the 1990s, in the US it declined slightly between the 1960s and 1970s but then stabilised in the two subsequent decades from just below 3% in the 1960s to just above 2% in the 1970s, 1980s and 1990s. Hence, while up to the 1970s average real output per capita growth was higher in the euro area, in the 1980s it was similar and in the 1990s it became lower. More recently, since the mid-90s, while in the euro area trend real output per capita growth continued to gradually decline, in the US it even started to increase. As a result of these relative growth patterns, the level of euro area real output per capita which was gradually catching up with the US level up to the early 1980s [from about 55% in 1960 to just above 70% in the early 1980s] appeared to stop converging and then, since mid 1990s, the relative level has slightly started to fall again.

Productivity is the key to prosperity

What then can explain these different output per capita growth patterns? There are various immediate sources of output per capita growth. These can be classified into three main categories: labour productivity, labour utilisation and demographic factors. Labour utilisation, which depends on factors such as average hours worked, the unemployment rate and the participation rate, can explain a significant part of the growth dynamics reviewed. For example, average hours worked have been declining during most of the past 40 years in the euro area, although in the recent decade to a lesser extent.[2] By contrast, in the US after declining up to the 1980s, they have been broadly stable.[3] As a result, while in 1970 euro area workers were working about the same number of hours per person as US workers, by 2004 we are working just 85% of what North-American are doing. Moreover, despite some recent improvements in the most recent decade, the unemployment rate is significantly higher in the euro area and has declined less since the mid-1990s than in the US. By contrast, the participation rate has been increasing more strongly in the euro area than in the US since the 1990s, but this was possible due to the significantly lower initial level of participation we had in the euro area[4]. Also demographic factors, in particular the gradual process of ageing of the population, have been relevant and will undoubtedly become an important challenge, looking ahead.

However, I would like to focus today on labour productivity. There are good reasons to focus mainly on this factor of growth. We know, since at least the pioneering contributions of Professor Robert Solow to growth theory in the late 1950s, that productivity growth is the key factor to long term economic growth.[5]

For the euro area this is confirmed by the fact that almost all of the average growth in real output per capita was determined by labour productivity growth in all of the past four decades, while the contribution from labour utilisation has been most often negative and demographic factors-or the changing age structure- have provided a very modest contribution in most decades since 1960, which turned even negative in the 1990s.[6] Similarly, also for the US, labour productivity growth has been the main driving force of real output per capita growth in the past few decades, although on average to a lesser extent compared to the euro area.

As regards labour productivity growth, from a longer term perspective the euro area and the US have one aspect in common: a decline in average productivity growth after the early 1970s. Unfortunately, while the stylised facts characterising this productivity growth slowdown are well documented, its causes are still debated. Various possible explanations have been proposed, ranging from sectoral shifts, to the impact of the oil shocks of the 1970s, to increasing measurement problems. However, none seems to provide a fully satisfactory interpretation.[7]

More recently, labour productivity growth developments have diverged between the euro area and the US. While on average productivity growth was broadly unchanged between the 1980s and the first half of the 1990s, both in the euro area and the US, a substantial change can be observed in the second half of the 1990s.

In the euro area, average productivity growth (measured in terms of output per hour worked), after growing on average by about 2.5% from 1980 to 1995, declined to just above 1.5% in the period 1995-2000 and further to just above 1.0% on average in the period 2000-2004. By contrast in the US, growth in productivity per hour worked, after averaging just below 1.5% from 1980 to 1995, rose to an average of just below 2% in the period 1995-2000 and to just below 3% over the period 2000-2004. This rise in the US may partly reflect cyclical factors, but the apparent resilience of productivity growth during the past downturn and the significant further pick-up over the last two years tends to support the widespread view that the mid-1990s marked a structural improvement in US productivity growth.

I should note that the assessment that productivity growth in the euro area has fallen behind the US in the second half of the 1990s is independent of the way labour productivity growth is measured. Downward trends in productivity growth are observed for the euro area irrespective of whether productivity is measured per person employed or per hour worked. However, it is worth noting that due to the trend decline in average hours worked in the euro area (in contrast to stable hours in the US), the productivity growth gap with the US is more marked in terms of employed persons than in terms of hours worked.

Given then this striking recent divergence, it is worthwhile to focus on productivity developments since the mid-1990s.

Proximate causes of labour productivity growth divergences

The decline in euro area labour productivity growth since the mid-1990s can be discussed in terms of its main immediate sources, i.e. total factor productivity growth, changes in labour quality and capital deepening.

Total factor productivity is a measure that corresponds to the theoretical concept of technological progress. However, in practice it captures the impact of several factors (such as improvements in organisation and in the quality of capital), such that it is not immediate to associate its evolution to purely technological advances. Capital deepening is the increase in capital intensity, which is the amount of capital per unit of labour input.

Although the decomposition of labour productivity is subject to significant measurement uncertainty, there seems to be robust evidence that slower productivity growth in the euro area since the mid-1990s reflects both less capital deepening and lower growth in total factor productivity. During the period 1995-2000, capital deepening was only half of that observed over the period 1990-1995, hence accounting for a large part of the slowdown in aggregate labour productivity growth. While increases in Information and Communication Technology (ICT) investment are estimated to have had a positive, albeit limited, effect on the contribution of ICT capital deepening since the mid-1990s, this was more than compensated for by lower capital deepening in non-ICT. Available evidence also suggests that labour quality increased more moderately in recent years than in the first half of the 1990s.[8]

Comparing the factors driving US labour productivity growth shows striking differences. The increasing contributions from ICT capital deepening and total factor productivity since the mid-1990s stand in stark contrast to developments in the euro area and results from accelerating investment in ICT and a higher growth in measured total factor productivity in the US. In fact, the shares of ICT investment in GDP and the pace of ICT capital deepening were both higher in the US, thus contributing to the labour productivity growth gap between the two economic areas. Diverging trends in total factor productivity growth are also remarkable, with the US showing significant improvements in recent years.

The decomposition of labour productivity growth per hour worked at the sectoral level sheds more light on the different developments in labour productivity growth in the euro area and the US. The main features of sectoral productivity contributions can be summarised as follows.

First, a significant deceleration in hourly labour productivity in non-ICT sectors from the first half of the 1990s explains most of the decline in euro area aggregate labour productivity growth over the late 1990s.

Second, productivity developments in the ICT producing sectors were in recent years relatively strong in the euro area, and even outperformed slightly the US in the segment of ICT producing services (software, computer and communication services). However, this sector represents a smaller share of the economy in the euro area than in the US, which implies that positive developments had a more limited impact on aggregate productivity in the euro area.

Third, the ICT using sectors in the euro area failed to experience the strong acceleration in labour productivity observed in the US in recent years. In particular, key ICT using services such as retail, wholesale and financial services saw broadly stable, or slightly lower, productivity growth in recent years in the euro area. At the same time productivity growth in these sectors surged in the US. This accounts for a large part of the difference in aggregate productivity growth between the euro area and the US [of around 1ppt] over the period 1996-2002.

More fundamental factors

To fully understand the sources of labour productivity growth it is useful to differentiate between immediate and more fundamental sources of labour productivity growth. The immediate sources of labour productivity growth consist of those we have just discussed, namely total factor productivity growth, capital deepening and increases in labour quality. Each of the immediate sources is the result of more fundamental factors that in turn depend on institutions and preferences. For example, total factor productivity growth may depend on innovation, research and development (R&D) spending and technology diffusion, which in turn are influenced by institutional factors, such as regulations, and preferences. Unfortunately, research on these more fundamental sources of labour productivity growth has not yet reached clear conclusions.[9] While theoretical work has clearly identified important factors[10] and indicated that interactions between the various fundamental factors may play an important role,[11] more empirical work is needed to further pin down the relative importance of these factors.

The need for structural reforms

Since the interactions between fundamental factors and economic policies are very complex, more research is clearly needed. Nevertheless, it is clear to my eyes that economic policies geared towards enhancing the contribution of fundamental factors will be beneficial to long term productivity growth. A number of European policy initiatives and recommendations - in particular within the framework of the Lisbon strategy - to promote productivity growth in the euro area have been made by expert groups, including the recommendations in the Sapir report of July 2003 and the more recent Kok report of November 2004. A number of these recommendations are reflected also in the recent mid-term review of the Lisbon agenda which seeks to tackle the triple challenge of globalisation, ageing populations and rapid technological change within a stability-oriented macroeconomic environment.

In particular, one of the key priorities identified in the mid-term review is to promote “knowledge and innovation for growth” which includes a number of recommendations concerning human capital. These recommendations go in the right direction and if implemented rigorously are likely to go a long way towards addressing the main fundamental determinants of long-term productivity growth.

For example, the Lisbon strategy sets a target that research and development spending should reach 3% of GDP by 2010. Moreover, in an environment of necessary fiscal prudence, two thirds of this spending should come from the private sector. In 2003, overall expenditure stood at 1.9% of GDP for the European Union (with 55% coming from the private sector). By way of comparison, the U.S. spends 2.6% of GDP on research and development (63% of which is financed by the private sector).

In order to achieve the targets set by the Lisbon strategy we in Europe we need more scientists and researchers. Considering that in the European Union we have only about 5.3 scientists and researchers per 1000 workforce, which compares to a figure of 9 scientists and researchers per 1000 workforce in the US, it is not surprising that more innovations take place across the Atlantic Ocean. But we should focus equally on increasing the quality of our research: this is a key factor, as it can attract new young brilliant investigators, which in turn would bring new original ideas and ignite a virtuous spiral of better and better achievements and conditions. This has become even more important today, with globalisation affecting not only finance and trade but increasingly more also research.

Other broad policy measures include more and better education and training, which are required to minimise mismatches in the labour market and allow for a smoother reallocation of workers between sectors and firms. A comparison with the U.S. illustrates the knowledge gap: in the US, annual expenditure on higher education institutions per student represents 57.8% of GDP per capita, while in the euro area only 35% is spent.

In addition, it is clear that establishing competitive, efficient and well functioning markets is another prerequisite in order to enhance medium to long-term growth, not to mention the shorter term impact of increasing the resilience of the economy to shocks.[12] In the EU, some progress has been made in this regard. For example, most network industries are now fully or largely open to competition, in particular those in telecommunications and air transport and to a lesser extent in energy markets. And the reforms do pay off: the remarkable labour productivity growth performance in network industries in Europe over the last ten years provides a perfect example of the positive impact on labour productivity growth of easing regulations and fostering competition.

Overall, although we have to recognise our limited knowledge on important issues regarding the concrete importance of the fundamental sources of growth and their interactions, it is clear that a number of measures, such as those indicated in the Lisbon strategy, cannot but be beneficial for the euro area economy and may even become necessary to sustain our economic well being.

Let me conclude with a comment on a well-known remark by Professor Solow, who (with reference to the US economy) wrote in 1987 that “You can see the computer age everywhere but in the productivity statistics”.[13] Less than ten years after this remark, computers started to appear forcefully also in the productivity statistics, at least in the US.

A similar remark to that of Professor Solow in 1987 could be made not only today, but possibly also ten or more years ago, for the euro area. We can see also in Europe computers, the internet, mobile phones, iPods and so on everywhere. The ICT revolution affected not only the US economy, but also the euro area and other regions of the developed world. We cannot afford to remain passive and hope that the computer age will soon be reflected in productivity statistics for the euro area. There are important reasons why this has not happened yet: what is essential is not only the quality of the seed but also the fertility of the soil.[14]

We need structural reforms to make sure that the soil is fertile enough for the seeds of technological progress to produce the fruits leading to higher productivity growth and per capita output growth. We know what we have to do, the way has been identified in the Lisbon strategy and the first steps have been taken. That way needs now to be followed in a more decisive manner. Only in this way can per capita output growth be sustained, thereby supporting economic prosperity today, and in the future.

Thank you very much for your attention.

[1] It was just above 2.5% per year in the euro area and about 2¼% in the US, from 1960 to 2004.

[2] According to OECD data, average hours worked per year in the euro area declined from about 1940 in 1970 to about 1550 in 2004.

[3] According to OECD data, average hours worked per year in the US declined from about 1935 in 1970 to about 1830 in 1982 and then stabilised around 1860 up to 2000. In 2004 they were about 1825.

[4] The participation rate is estimated by the European Commission at about bout 72% in the euro area and about 76% in the US in 2004.

[5] Solow, Robert M. (1956): "A Contribution to the Theory of Economic Growth." Quarterly Journal of Economics, 70:65-94 and Solow, Robert M. (1957), “Technical Change and the Aggregate Production Function,” Review of Economics and Statistics, V. 39: 312-320.

[6] See for example A. Musso and T. Westermann (2005), “Assessing Potential Output Growth in the Euro Area”, ECB Occasional Paper Series, No 22, January 2005.

[7] Recent empirical studies include Hornstein and Krusell (1996): “Can Technology Improvements Cause Productivity Slowdowns?”, NBER Macroeconomics Annual 1996, pp. 209-59; Sichel D. (1997): "The Productivity Slowdown: Is a Growing Unmeasurable Sector the Culprit?", Review of Economics and Statistics, vol. 79, n° 3, pp. 367-70 , and Nordhaus (2004): "Retrospective on the 1970s Productivity Slowdown", NBER Working Paper No. W10950 December.

[8] See ECB Monthly Bulletin box on "Developments in euro area labour quality and their implications for labour productivity growth” in the October 2005 issue.

[9] See for example the discussion in Griliches Z. (2001): “A Perspective on What We Know about the Sources of Productivity Growth”, in New Developments in Productivity Analysis, Hulten et al (eds), Chicago University Press.

[10] The main fundamental forces identified by theoretical research include factors that increase the pace of technological progress, such as R&D and innovation activity, greater diffusion of new technologies, such as ICT, and advances in human capital. See for example Barro, R. and Sala-i-Martin, X. (2004): Economic Growth, second edition, MIT Boston.

[11] For example, it has been suggested that the composition of human capital may facilitate productivity gains from new technologies due to improved diffusion of technology. The possibility that more educated workers adopt new technologies faster was already raised by R. Nelson and E. Phelps in 1966 (“Investment in Humans, Technological Diffusion, and Economic Growth”, American Economic Review 65: 69-75). They also suggest that more investment in human capital is required when technologies are changing faster. J. Vandenbussche, P. Aghion and C. Meghir ("Growth, distance to frontier and composition of human capital," IFS Working Papers W04/31, Institute for Fiscal Studies, 2004) extend this argument and show that that skilled labour has a higher growth enhancing effect close to the technology frontier, because innovation is more skill intensive than imitation.

[12] For a further extension of this topic see European Commission (2004), “The link between product market reforms and productivity: direct and indirect impacts”, the EU Economy: 2004 Review.

[13] R. Solow, ‘We’d better watch out,’ New York Times Book Review (12 July 1987), p. 36.

[14] Expression borrowed from Gordon, R. (2004): “Why was Europe Left at the Station when America’s Productivity Locomotive Departed?”, NBER working paper n. 10661, August. Gordon suggests that “the role of ICT has been exaggerated” and, focusing on the comparison between US and Europe, argues that the main difference lies in a set of institutional and cultural factors which make the US a “more fertile soil when the right seeds are planted”. These factors include differences regarding competition, corporatism and culture.

Speaking engagements

Media contacts