1. 1 R&D and productivity growth Industry-level studies Sterlacchini, A. (1989) R&D, Innovations and Total Factor Productivity Growth in British Manufacturing, Applied Economics. ….. Sterlacchini, A. and F. Venturini (2013) Boosting Manufacturing Productivity through R&D: International Comparisons with Special Focus on Italy, Journal of Industry, Competition and Trade. Firm-level studies (use of microdata) Sterlacchini, A. e N. Matteucci (2009) “R&S, innovazione e produttività nelle imprese manifatturiere italiane” in Rondi, L. e F. Silva (a cura di) Produttività e cambiamento nell’industria italiana: Indagini quantitative, Il Mulino, Bologna

2. BOOSTING MANUFACTURING PRODUCTIVITY THROUGH R&D: INTERNATIONAL COMPARISONS WITH SPECIAL FOCUS ON ITALY Alessandro Sterlacchini and Francesco Venturini Journal of Industry, Competition and Trade (2013)

3. 3 Weak productivity performance of EU countries as opposed to the US, especially over the second half of the 1990s. Within the EU, Italy experienced a particularly severe slowdown of manufacturing TFP (industrial decline?) Technological explanation  exhaustion of the catching up process. “innovation at the frontier has become the main engine of growth” (Sapir at al., 2003; p. 29) Knowledge-based economy (Lisbon strategy): 1.Human capital and education 2.R&D and innovation (2010 target for business R&D) 3. Production and use of ICT Background and motivation

4. 4 R&D and manufacturing productivity Since more than 80% of business R&D is performed in manufacturing, this is the sector in which the relationship with productivity growth should be particularly strong. Recent empirical studies at the level of manufacturing industries confirm the above expectation. Our aim: identify the disparities between the US and the largest EU countries in the long-run impact of R&D on manufacturing productivity

5. 5 EU15USAGermanyFranceItalySpain Labour productivity 1980-19951.83 1.411.711.811.752.38 1995-20021.04 2.051.04 0.43-0.12 Total factor productivity 1980-19951.331.081.160.740.961.41 1995-20020.901.470.790.63-0.12-0.21 Capital/labour ratio (contribution) 1980-19950.500.330.551.070.790.97 1995-20020.140.580.250.420.550.09 Decomposition of the annual rates of change of labour productivity (total economy) Source: The Conference Board and GGDC Total Economy database. November 2007.

6. 6 Analytical framework Cobb-Douglas production function augmented with R&D capital: Usual assumptions for computing TFP, the log-level of which is a function of R&D capital s L = share of labour costs on value added  i = industry fixed effect, constant over time  = excess elasticity of value added with respect to R&D capital

7. 7 Econometric framework Dynamic representation through a panel Dynamic Ordinary Least Squares estimator (panel DOLS) This procedure assumes homogeneity in the cointegration vector among industries (θ). It estimates the TFP equation adding p lags and leads of the first-differences of the explanatory variable RD (R&D capital) to correct the effect of reverse causality (endogeneity bias) and control for the short-run movements around the long-run path ln TFP it =η i + θ ln RD it-1 + γ i Δ ln RD it-1±p +ε it

8. 8 Data sources Four countries: US, Germany, France, Spain and Italy. Twelve manufacturing industries over a period of 23 years (1980- 2002): 276 obs. for each country. Statistical sources: OECD: STAN Database: value added, gross fixed capital formation, total employment, labour compensation; ANBERD and Science and Technology Database: industry R&D expenditures. ISTAT: R&D expenditures in Italian mfg. industries.

9. 9 Variables’ definition Y= industry value added at constant (1995) prices; L = number of employed persons; s L = share of labour compensation on value added. For computing K and RD we apply a perpetual inventory method with geometric depreciation; annual data for gross fixed capital formation and R&D expenditures are used (at 1995 prices). The depreciation rate for K is constant over time but varies across industries, while that concerned with RD is set, for all the industries and countries, equal to 15%.

10. 10 R&D intensity on value added by industry (average 1980-2002) Share of total R&D performed by the most R&D intensive industries: US = 0.77; Germany = 0.79; France = 0.84; Spain = 0.73, Italy = 0.83.

11. 11 Industry shares on total manufacturing R&D Chemicals & Pharma- ceuticals Electrical & Optical Equipment Transport Equipment Total R&D-intensive industries US 1980-199514.327.135.677.0 1995-200217.036.725.979.6 Germany 1980-199522.030.325.277.5 1995-200218.923.837.980.7 France 1980-199518.533.832.684.9 1995-200222.330.429.782.4 Italy 1980-199521.631.931.384.8 1995-200217.533.129.580.1 Spain 1980-199521.027.825.874.5 1995-200221.622.925.870.3

12. 12 R&D intensity on value added (total manufacturing, current prices)

13. 13 R&D capital stock: total manufacturing (1995=100)

14. 14 R&D capital stock (1995=100): Chemicals and Pharmaceuticals

15. 15 R&D capital stock (1995=100): Electrical and Optical equipment All series are built by applying a harmonised deflator obtained from US hedonic prices adjusted for rate of inflation’s differentials

16. 16 R&D capital stock (1995=100): Transport equipment

17. 17 Average annual rates of change in R&D capital (total manufacturing) United States GermanyFranceItalySpain 1980-20024.103.814.634.577.54 1980-19905.515.425.577.849.07 1990-19951.572.845.013.607.85 1995-20023.882.213.000.605.14

18. 18 1-year lags and leads2-year lags and leads Without time dummies With time dummies Without time dummies With time dummies Specification1234 United States0.212**0.393**0.299**0.385** (0.096)(0.105)(0.097)(0.105) Germany0.321**0.2890.292**0.290 (0.112)(0.206)(0.145)(0.210) France0.140*0.207**0.1210.193** (0.073)(0.082)(0.079)(0.084) Italy0.101**0.084**0.110**0.121** (0.038)(0.027)(0.043)(0.030) Spain0.184**0.187**0.176**0.189** (0.024)(0.028)(0.019)(0.023) Panel DOLS estimates of the R&D impact on TFP (12 manufacturing industries)

19. 19 Looking at the long-run impact of R&D on manufacturing TFP, the US are above the major EU countries. R&D investment in ICT mfg. products seems the major responsible. Assuming that the above effects will be maintained in the future, the EU countries could boost manufacturing TFP via R&D as in the US. However, they should undertake exceptional R&D efforts and become world leaders at least in some areas or market niches. Unfortunately, looking at R&D data for 2001-2010, they are not following this path. Discussion (1): European countries

20. 20 The previous considerations apply to a lesser extent to Italy. Why Italian manufacturing industries perform so poorly? “Static” explanations based on a threshold (non-linear) effect, the low-tech specialisation and the strong presence of small firms are ruled out by the results arising from Spain. In spite of being technological followers, Spanish industries are characterised by a remarkable impact of R&D on TFP. “Dynamic” explanation: the R&D slowdown experienced since the early 1990s Discussion (2): the Italian case

21. 21 Much of the relatively poor performance of Italian mfg. industries could be ascribed to the declining R&D commitment of the last decade. Thus, there is a strong and urgent need to retake the path of the Eighties. The fact that, in the medium-run, an “R&D shock” will not improve substantially the TFP of Italian industries does not mean that extra-ordinary R&D efforts are useless. First, they can boost productivity in the long-run. Second, along with fostering innovations, R&D activities are crucial to absorb foreign knowledge and remain close to the technological frontier. Final considerations for Italy