Higher education and growth

Higher education has been a key issue in recent public debates in France. Topics have ranged from the mediocre results of research to the excessive development of branches with few hiring opportunities, a “selection by failure” system rather than one based on better career counselling, and the problems of governance. Another hot topic is the low level of resources for higher education (spending per student) compared to other OECD countries and – another French exception – even when compared to investment in secondary education![1] We focus on higher education’s role in fostering productivity gains and boosting growth potential. The level of education of the working-age population plays a crucial role in determining growth rates and productivity gains. The skills of the population shape the country’s ability to innovate and to tap technological innovation. Based on observations in 60 countries over the past four decades, Hanushek and Woessmann (2007) demonstrate a robust relationship between GDP growth per capita and the average number of years in education of the working-age population (15-64 age group). Yet, the quality of education, as measured by the results of standardised scholastic tests, seems to have a much more significant impact than the average duration of education. After taking into account an education quality indicator, the correlation coefficient rose to 0.73 from 0.33! The authors also show that the impact of this education quality factor is stronger in countries that are very open (see appendix 1). Krueger and Lindahl (2001) obtained similar results based on a panel of 110 countries. They show that the impact of increasing the level of education is hardly visible in the short term (5 years), but becomes very significant from a 10 to 20-year horizon (see appendix 2). Research conducted by endogenous growth theorists in recent years show that school systems and education policies that were appropriate during catch-up phases (such as the period 1946-1975 in France) cease to be so as the country approaches the technology frontier[2]. During catch-up phases, it is important to have a workforce that is sufficiently educated at the primary and secondary levels to use existing technologies (imitation policies). Yet once a country begins to approach the technology frontier, the capacity to innovate becomes crucial, which means giving priority to higher education. This has become even more important with the emergence of a series of general-purpose technologies, which required further innovation to apply these new technologies to specific sectors. Using a panel of OECD countries, G. Cette (2007) shows that there is a significant positive correlation between the share of university graduates in the working-age population and the investment rate in new information and communications technologies (ICT). For example, in countries like the United States, Sweden, Finland and Denmark, where over 30% of the working-age population has a degree in higher education, the ICT investment rate exceeds 3.5%. At the other extreme, in Italy and Portugal, where less than 15% of the working-age population have university degrees, the ICT investment rate is less than 2.5%. A similar correlation exists with the proportion of households with internet connections: 55% in the United States, 70% in Denmark and at the other extreme, only 35% in Italy and 25% in Portugal (using 2004 data).[3] Productivity and education: a few results Total factor productivity of capital and labour (denoted A) reflects the efficiency of the use of the production factors. It can increase through the imitation of existing technologies or through innovation. If A* is productivity at the technology frontier, then:  

equation_1.gif The first component on the right hand side of the equation corresponds to the imitation effect, the second to innovation between periods t-1 and t. The distance from the technology frontier can be measured by dt  


Using g as the growth rate of productivity at the technology frontier, we can divide the terms of the equation by A* to obtain:


 d moves from 0 to 1 as the country approaches the technology frontier. When d is close to zero, imitation is the main source of productivity gains, but as d moves towards 1, innovation plays an increasingly important role. This illustrates the importance of setting up education and research structures that foster innovation (Acemoglu, 2002). Using empirical data from 80 countries between 1960 and 2000, Aghion et al. (2003) studied the impact of education on productivity by distinguishing between primary and secondary education on the one hand and higher education on the other. They show that there is a differentiated effect for each level of education, depending on the country’s distance from the technology frontier. Their results are summarised in the table 1.

An additional year of higher education increases the growth of total factor productivity by 8%. The variable st * dtf t-1 has a significant positive impact, which demonstrates the increasing importance of higher education for growth as the country approaches the technology frontier. Using a selection of OECD countries, Aghion et al (2004) distinguishes between the impact of higher education (undifferentiated) and that of at least 2 years of higher education (the share of the working-age population with a short undergraduate education is added to the general population with no higher education). They conclude that a long period of higher education (over 2 years) has a much more significant impact on productivity (see table 2). They stress that toward the end of the 1970s, France crossed the point where it “became more effective to invest in higher education than to invest in secondary education.” Using a panel of 21 OECD countries, l’Angevin et al. (2005) reach the same conclusions as Aghion et al. The authors estimate the elasticity of productivity to human capital at 0.11 in France vs an OECD average of 0.07. Furthermore, by distinguishing between several levels of education (through middle school, short technical degree, short higher education and long higher education) they show that only a short technical degree and long higher education had a positive impact on productivity gains. Moreover, this effect does not last long for short technical degrees, but persists over the long term for longer periods of higher education. All in all, policies that develop long higher education (which generally coincide with policies in favour of innovation) have a bigger impact on long-term productivity than those that encourage short technical training. We can thus see why it is important to make the university system more attractive for many good high-school graduates, who, under the current situation, are being steered towards short, selective technical sectors. To use Mr. Aghion’s terminology, these technical programmes encourage imitation policies, whereas what countries close to the technology frontier, such as France, need are innovation policies. Does increasing the pool of skilled labour with university degrees put downward pressure on relative wages for skilled jobs? In other words, does it reduce the advantages of a higher education in terms of relative wages for skilled workers (ratio of skilled wages/average wages)? Looking at the US experience from the 1960s to the present, according to several studies the answer is no. Although the increase in the skilled labour pool at the end of the 1960s and in the 1970s did squeeze the relative pay of graduates, the same does not apply since the 1980s. Acemoglu (1998, 2002) charts this phenomenon, which we have outlined below (see chart below). In the short term, for a given technology, an extra supply of qualified labour (DS) lowers the relative wage from W0 to W1 (shift along the curve of short-term demand DS1). During a later period, the abundance of skilled labour encourages companies to adopt more advanced technologies among those available (DS1 shifts to the right to DS2), a trend that limits the moderation of relative wages for skilled labour.[4] In the longer term, the abundance of skilled labour stimulates technical progress, resulting in a positive slope for the curve of long-term demand for skilled labour (DL2), since this stimulates demand for this type of labour and pushes up relative wages (W2). (See Schema)



 appendix_1.gif appendix_2.gif 

 Philippe d’Arvisenet


ACEMOGLU D. “Why do new technologies complement skills? Directed technical change and wage in equality”. The Quarterly Journal of Economics, Nov. 1998  

“Technical change, inequality and the labor market “. Journal of Economic Literature, March 2002  

ACEMOGLU D., AGHION P., ZIBILOTTI F “Distance to frontier, selection and economic growth”, NBER Working Paper 9066, 2002  

AGHION P., COHEN E. “Education et croissance”. CAE, La documentation française, Paris, 2004  

AGHION P., MEGHIR C., VAN DEN BUSSCHE J. “Growth, education and distance to frontier”, Mimeo (2003) cited by SAPIR A. (2003)  

BOISIVON JP., LICHTENBERGER Y. “Enseignement supérieur, ne pas sacrifier l’avenir” in “C’est possible, voici comment”. Edited by M. Pébereau and B. Spitz, Institut de l’entreprise, R. Laffont, 2007  

CETTE G. “Productivité et croissance en Europe et aux Etats-Unis”. Collection Repères, La découverte, 2007  

CICCONE A. PAPAOANNOU E. “Human capital, the structure of production and growth”. ECB Working Paper n°623, May 2006  

HANUSHEK E., WOESSMANN L. “The role of school improvement in economic development”. NBER Working Paper 12832, January 2007  

JACOBS B., VAN DER PLOEG F. “Guide to reform of higher education: a European perspective”. CEPR Discussion Paper N°5324, Nov. 2005  

KRUEGER D., KAMAR K. “US-Europe differences in technology driven growth: quantifying the role of education”. NBER Working Paper 10001, Sept. 2003  

KRUEGER A., LINDAHL M. “Education for growth: why and for when?”. Journal of Economic Literature, Dec. 2001  

L’ANGEVIN C., LAIB N. “Education et croissance en France et dans un panel de 21 pays de l’ouest”. INSEE Working Paper 2005/08, July 2005  

MONTANINO A., PRZYWARA B., YOUNG D. “Investment in education: the implication for economic growth and public finances”. EC Economic Papers n°217, Nov. 2004  

SAPIR A. “An agenda for a growing Europe”. CE Report, 2003  

SCHWERDT G., TURUNEN J. “Growth in euro area labour quality”. CEPR Discussion Paper 5509, February 2006  

TEULINES C., VAN RENS T. “Education, growth and income inequality”. CEPR Discussion Paper 3863, April 2003  

WOESSMANN L. “Educational production in Europe”. Economic Policy, July 2005  


[1] For more information on these issues, which we will not cover any further in this article, see J.P. Boisivon and Y. Lichtenberger (2007) or Ph d’Arvisenet: “University reforms can no longer wait”, La Tribune, 14 June 2007. We would like to point out, however, as several recent studies show, that the size of education budgets only answers a very limited part of the problem, whereas the incentives of different players in the system, in contrast, play a decisive role. Based on the construction of education production functions (a process that analyses the output of the system based on the resources implemented to assure its functioning), Ciccone et al (2006) show that, using data from 18 OECD countries, there is not a significant relationship between education spending per capita and performances on standardised writing and math tests.
[2] See Ph. d’Arvisenet: “Growth potential, productivity gains and innovation”, EcoWeek, BNP Paribas weekly bulletin, 25 May 2007
[3] The human capital factor is not the only pertinent variable. There also seems to be a negative correlation between the use of new technologies and the rigidity of the job market (see G. Cette, 2007).
[4] This is documented by Ciccone and Papaioannou (2006) based on 37 sectors in 40 countries. High-tech sectors grow even more rapidly (relative to the others) if they are located in a country with a highly educated labour force. As an illustration, the authors show that in a country relatively rich in skilled labour (they use the example of Malaysia, which is in the 75th percentile in terms of duration of education), the growth differential between a skill-intensive industry like chemistry (75th percentile of the statistical distribution in terms of the use of skilled labour input) and a sector like pottery (25th percentile) is 1.2% to 2.3% higher than for a country like the Philippines (25th percentile in terms of duration of education).

One Response to "Higher education and growth"

  1. interested reader   February 1, 2008 at 4:17 pm

    I read and admire Acemoglu’s research on skill-biased technological change and some of the literature following this approach very much. There is however one fundamental issue that cannot easily be reconciled with this literature: the productivity slowdown since the early 1970s. Yes, there were ‘redistributions’ among skilled and low-skilled labor due to techn.innovations and outsourcing but total productivity did undeniably slow down when the reaching of the new technology frontier should have boosted it instead. And with regard to technology’s contribution to the post-1995 productivity surge in the U.S. it is proving to be elusive now, and even some of the New Economy’s most fervent advocates such as Stephen D. Oliner, Daniel E. Sichel, and Kevin J. Stiroh are forced to revise their earlier estimates down. http://www.federalreserve.gov/pubs/feds/2007/200763/200763abs.htmlI still have a hard time reconciling high-tech with slow overall productivity.