(Luận văn) innovation and productivity in small and medium enterprises, a case study of vietnam

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES THE HAGUE VIETNAM THE NETHERLANDS t to HO CHI MINH CITY ng hi ep w n lo ad VIETNAM – THE NETHERLANDS y th ju PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS yi pl n ua al va n INNOVATION AND PRODUCTIVITY IN ll fu oi m SMALL AND MEDIUM ENTERPRISES: nh at A CASE STUDY OF VIETNAM z z om l.c gm PHAM DO TUONG VY k jm ht vb By an Lu n va ey HCMC, NOVEMBER 2016 t re MASTER OF ART IN DEVELOPMENT ECONOMICS International Institute of Social Study Ho Chi Minh City The Hague Vietnam The Netherlands t to University of Economics ng hi ep w VIETNAM – THE NETHERLANDS n lo PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS ad ju y th yi pl INNOVATION AND PRODUCTIVITY IN ua al n SMALL AND MEDIUM ENTERPRISES: n va ll fu A CASE STUDY OF VIETNAM oi m at nh z z by Pham Do Tuong Vy vb k jm ht A Thesis Submitted in Partial Fulfilment of the Requirements for the Degree of an Lu n va Academic Supervisor: Dr Vo Hong Duc om Development Economics l.c gm Master of Art in ey t re HCMC, NOVEMBER 2016 DECLARATION t to I hereby declare that this thesis entitled “Innovation and Productivity in ng Small and Medium sized Enterprises: A case study of Vietnam”, which is written hi ep and submitted by me in accordance with the requirement for the degree of Master of Art in Development Economics to the Vietnam – The Netherlands Programme This w n is my original work and all sources of knowledge carried in this thesis have been duly lo ad acknowledged ju y th HCMC, November 2016 yi pl ua al n PHẠM ĐỖ TƯỜNG VY n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re ACKNOWLEDGEMENT t to I would like to take this opportunity to express my deepest gratitude for the ng hi help, support and encouragement of the following people, who have contributed to ep the completion of this thesis in their very own ways Above all, I would like to express my immeasurable appreciation to my w n supervisor – Dr Võ Hồng Đức for his precious time, support and advices to make lo ad this thesis completed y th Furthermore, I would like to send my great thanks to all the lecturers and staffs ju at the Vietnam – The Netherlands Programme for their knowledge and supports yi pl during my time joining in the program In specific, I am extremely grateful to Dr al ua Phạm Khánh Nam and Dr Trương Đăng Thụy for their valuable guidance and n support in the courses and thesis writing process va n To all of my friends in Class 21 and my colleagues at TPF, I could never fu ll thankful enough for your encouragement and support until the very end of this thesis m oi Last but not least, my deepest thanks and love to my parents, who have always nh been beside me Without their unconditional love, none of this would have been at z possible z k jm ht vb om l.c gm an Lu n va ey t re ABBREVIATION t to ng hi ep 2SLS Two Stage Least Squares CDM Crépon, Duguet and Mairesse FE Fixed Effect Generalized Method of Moments w GMM General Statistic Office n GSO lo Instrument Variables ad IV Ordinary Least Squares yi Olley and Parker pl OP Levinsohn and Petrin ju OLS y th LP al Research and Development SMEs Small and Medium-sized Enterprises TFP Total Factor Productivity n ua R&D n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re ABSTRACT t to The majority of enterprises in Vietnam is categorized as small and medium ng hi sized (SMEs) firms which play an important role to the sustainable growth of the ep Vietnamese economy As such, improving the productivity of the SMEs is essentially needed and this request becomes a crucial mission for the governments It is generally w n accepted that innovation and technology improvement are key drivers of productivity lo ad (Bartelsman & Doms, 2000) However, they have not been well-acknowledged by yi unarguable ju y th the SMEs in Vietnam even though their huge contribution to firm’s productivity is pl This study aims to examine the relationship between innovation and al n ua productivity in the Small and Medium-sized Enterprises (SMEs) in Vietnam To va establish and quantify this relationship, this study employs the two-stage process: (i) n the estimation of total factor productivity for each firm; and (ii) a determination of an fu ll innovation – productivity relationship In the first stage, total factor productivity is m oi estimated based on production function using the input and output approach nh However, firms might adjust their input level according to expected productivity at z shock As such, a potential endogeneity caused by possible relationship between input z ht vb decision and productivity shocks (unobserved productivity shock) might exist To jm deal with this problem of endogeneity, an approach developed by Levinsohn and k Petrin is applied to estimate firm’s total productivity In the second stage, the system- gm GMM approach is adopted to examine the relationship between innovation and om l.c productivity An unbalanced panel dataset from five Small and Medium-sized Enterprises an Lu surveys from 2005 to 2013 is used in this study Findings from this study indicate current level This finding implies that higher (lower) level of current productivity ey In addition, past value of firm’s productivity also has significant relationship with its t re innovation activities provide positive and significant impact on firm’s productivity n expenditure intensity and high-quality labor share in total firm’s labor force, va that, in the context of Vietnam, when innovation is measured as innovation could lead to higher (lower) level of productivity in the future The study also provides empirical evidence to confirm that larger firms might perform better than t to the relatively smaller firms In contrast, capital structure provides negative impact on ng firm’s productivity However, this study fails to provide any evidence to support the hi ep view that longevity of firm does provide significant impact on productivity of firms Vietnam SMEs; Total factor productivity; Productivity Shock; w Key words: n lo Innovation, GMM ad ju y th yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re TABLE OF CONTENTS CHAPTER t to INTRODUCTION .1 ng hi ep 1.1 Problem statement 1.2 Research objectives .2 1.3 Research questions w Research motivations .2 1.5 Research scope and data 1.6 The structure of this study .3 n 1.4 lo ad y th ju CHAPTER yi LITERATURE REVIEW pl Productivity: concept and measurements va 2.2 n ua al 2.1 Schumpeter Theory of Innovation – How does Innovation play its role in economic development? Concept 2.1.2 Measurements 2.1.3 General productivity determinants .12 ll fu oi m Innovation: concept and measurements .16 at nh 2.3 n 2.1.1 Concept 16 2.1.5 Measurements .17 z 2.1.4 z vb k jm ht 2.4 How has the relationship between innovation and firms’ performance been analysed in the literature? 18 gm CHAPTER 23 3.1 l.c RESEARCH METHODOLOGY 23 An overview of Vietnamese Small and Medium-sized Enterprises 23 om Statistic overview 23 3.1.2 Difficulties 26 Methodology 27 Model identification 29 3.3 Research hypotheses and concept measurements 34 3.1.5 Research hypotheses .34 3.1.6 Concept and variable measurements 34 ey 3.1.4 t re Conceptual framework 27 n 3.1.3 va 3.2 an Lu 3.1.1 3.4 Data sources 36 CHAPTER 39 t to EMPIRICAL RESULTS 39 ng 4.1 Total Factor Productivity of Vietnamese SMEs 39 hi 4.1.1 Data descriptions 39 ep 4.1.2 Total factor productivity from production function estimation of Vietnamese SMEs 42 w n 4.2 Innovation – Firm’s productivity relationship .45 lo ad 4.1.3 Data descriptions 45 ju y th 4.1.4 The relationship between innovation expenditure intensity and firm’s productivity 46 yi 4.1.5 The relationship between high-quality labor share in total firm’s labor force and their productivity 49 pl al ua CHAPTER 52 n CONCLUSION AND POLICY IMPLICATION 52 va Conclusion remarks .52 5.2 Policy implications 54 5.3 Limitation and potential future research 54 n 5.1 ll fu oi m at nh REFERENCES 56 APPENDIX 1: Empirical studies on general productivity determinants 62 z z APPENDIX 2: Empirical studies on relationship between innovation and firm’s performance 65 ht vb jm APPENDIX 3: Durbin – Wu Hausman test for endogeneity 69 k APPENDIX 4: Durbin – Wu Hausman test for endogeneity 71 om l.c gm an Lu n va ey t re LIST OF TABLES AND FIGURES t to ng hi ep Classification of SMEs in Vietnam Table 3.2: Concepts and measurements of variables used in the study Table 3.3: Number of observation in selected industries in dataset Table 3.4: Number of observation after filtering Table 3.1: w n lo ad Table 3.5: Number of observation after filtering in stage y th Table 4.1: ju Comparison of OLS, Fixed Effect and LP estimators in Foods, Woods yi Table 4.2: Descriptive statistics of production function variables pl Comparison of OLS, Fixed Effect and LP estimators in Non-metallic n Table 4.3: ua al and Rubber and Plastics va n mineral, Fabricated metal and Furniture fu Descriptive statistics of TFP and its determinants Table 4.5: Regression results of innovation expenditure intensity and firm’s ll Table 4.4: oi m at z Regression results of high-quality labor share in total labor force and z Table 4.6: nh productivity jm ht vb firm’s productivity Number of enterprises at 31/12 (by size of employees) Figure 3.2: Conceptual framework k Figure 3.1: om l.c gm an Lu n va ey t re Jensen, M C (1986) Agency cost of free cash flow, corporate finance, and takeovers Corporate Finance, and Takeovers American Economic t to Review, 76(2) ng Jensen, M C., & Meckling, W H (1976) Theory of the firm: Managerial behavior, hi ep agency costs and ownership structure Journal of financial economics, 3(4), 305-360 w n Ha, D T T., & Kiyota, K (2014) Firm‐Level Evidence on Productivity Differentials lo ad and Turnover in Vietnamese Manufacturing Japanese Economic ju y th Review, 65(2), 193-217 yi Hall, B H., Lotti, F., & Mairesse, J (2008) Employment, innovation, and pl productivity: evidence from Italian microdata Industrial and Corporate al n ua Change, 17(4), 813-839 va Hansen, L P (1982) Large sample properties of generalized method of moments n ll fu estimators Econometrica: Journal of the Econometric Society, 1029-1054 m oi Holtz-Eakin, D., Newey, W., & Rosen, H S (1988) Estimating vector at nh autoregressions with panel data Econometrica: Journal of the Econometric z Society, 1371-1395 z ht vb Huergo, E., & Jaumandreu, J (2004) Firms' age, process innovation and productivity k jm growth International Journal of Industrial Organization, 22(4), 541-559 gm Keele, L., & Kelly, N J (2006) Dynamic models for dynamic theories: The ins and l.c outs of lagged dependent variables Political analysis, 14(2), 186-205 om Keller, W., & Yeaple, S R (2009) Multinational enterprises, international trade, and of Economics and Statistics, 91(4), 821-831 an Lu productivity growth: firm-level evidence from the United States The Review 58 ey International Economies, 20(2), 209-233 t re governance and the “chaebol problem” Journal of the Japanese and n productivity performance of Korean manufacturing firms: Corporate va Kim, E (2006) The impact of family ownership and capital structures on Li, H., & Atuahene-Gima, K (2001) Product innovation strategy and the performance of new technology ventures in China Academy of Management t to Journal, 44(6), 1123-1134 ng Lokshin, B., Belderbos, R., & Carree, M (2008) The productivity effects of internal hi ep and external R&D: Evidence from a dynamic panel data model Oxford bulletin of Economics and Statistics, 70(3), 399-413 w n Mairesse, J., Mohnen, P., Kremp, E., & KREMP, E (2005) The importance of R&D lo ad and innovation for productivity: A reexamination in light of the French ju y th innovation survey Annales d'Economie et de Statistique, 487-527 yi Mairesse, J., & Robin, S (2009) Innovation and productivity: a firm-level analysis pl for French Manufacturing and Services using CIS3 and CIS4 data (1998-2000 al n ua and 2002-2004) Paris: CREST-ENSAE va Marschak, J., & Andrews, W H (1944) Random simultaneous equations and the n ll fu theory of production Econometrica, Journal of the Econometric Society, 143- oi m 205 at nh Margaritis, D., & Psillaki, M (2010) Capital structure, equity ownership and firm z performance.Journal of Banking & Finance, 34(3), 621-632 z ht vb Miguel Benavente, J (2006) The role of research and innovation in promoting jm productivity in Chile.Economics of Innovation and New Technology, 15(4-5), k 301-315 gm l.c Mohnen, P., & Hall, B H (2013) Innovation and productivity: An update Eurasian om Business Review, 3(1), 47-65 an Lu Mortensen, P S., & Bloch, C W (2005) Oslo Manual-Guidelines for collecting and interpreting innovation data Organisation for Economic Cooporation and va n Development, OECD 59 ey economics, 5(2), 147-175 t re Myers, S C (1977) Determinants of corporate borrowing Journal of financial Nelson, R R (1973) Recent exercises in growth accounting: new understanding or dead end? The American Economic Review, 63(3), 462-468 t to Nguyen, N A., Pham, Q E., Nguyen, D C., & Nguyen, D N (2007) Innovation and ng Export of Vietnam’s SME Sector’, MPRA Paper No 3256 Development and hi ep Policies Research Center Nickell, S (1981) Biases in dynamic models with fixed effects Econometrica: w n Journal of the Econometric Society, 1417-1426 lo ad Olley, G S., & Pakes, A (1992) The dynamics of productivity in the y th telecommunications equipment industry (No w3977) National Bureau of ju yi Economic Research pl ua al Parisi, M L., Schiantarelli, F., & Sembenelli, A (2006) Productivity, innovation and n R&D: Micro evidence for Italy European Economic Review,50(8), 2037- n va 2061 fu ll Petrin, A., Poi, B P., & Levinsohn, J (2004) Production function estimation in Stata m oi using inputs to control for unobservables Stata journal, 4, 113-123 nh Polder, M., Leeuwen, G V., Mohnen, P., & Raymond, W (2009) Productivity at z effects of innovation modes z vb Santos, D F L., Basso, L F C., Kimura, H., & Kayo, E K (2014) Innovation efforts ht k jm and performances of Brazilian firms Journal of Business Research, 67(4), gm 527-535 l.c Roodman, D (2006) How to xtabond2: An introduction to difference and system om GMM in Stata Center for Global Development working paper, (103) an Lu Rosenbusch, N., Brinckmann, J., & Bausch, A (2011) Is innovation always beneficial? A meta-analysis of the relationship between innovation and ey 60 t re studies, 33(2), 127-170 n Schreyer, P., & Pilat, D (2001) Measuring productivity OECD Economic va performance in SMEs Journal of business Venturing, 26(4), 441-457 Schumpeter, J A (1912) 1934, The Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest and the Business Cycle.Trans Redvers t to Opie Cambridge, MA: Harvard University Press ng Schumpeter, J A (1942) Socialism, capitalism and democracy Harper and hi ep Brothers Schumpeter, J A (1939) Business cycles: a theoretical, historical, and statistical w n analysis of the capitalist process McGraw-Hill lo ad Siedschlag, I., Zhang, X., & Cahill, B (2010) The effects of the internationalisation y th of firms on innovation and productivity Economic and Social Research ju yi Institute Working Paper, (363) pl ua al Siegel, D., & Griliches, Z (1992) Purchased services, outsourcing, computers, and n productivity in manufacturing In Output measurement in the service va sectors (pp 429-460) University of Chicago Press n fu ll Solow, R M (1957) Technical change and the aggregate production function The m oi review of Economics and Statistics, 312-320 nh Tovar, B., Ramos-Real, F J., & De Almeida, E F (2011) Firm size and productivity at z Evidence from the electricity distribution industry in Brazil Energy z ht vb Policy, 39(2), 826-833 gm review Journal of economic surveys, 26(1), 98-128 k jm Van Beveren, I (2012) Total factor productivity estimation: A practical l.c Vu, H N., & Doan, Q H (2015) Innovation and Performance of Enterprises: The om Case of SMEs in Vietnam an Lu White, H (1982) Regularity conditions for Cox's test of non-nested hypotheses Journal of Econometrics, 19(2), 301-318 n va ey t re 61 ad ju y th yi pl n ua al APPENDIX 1: Empirical studies on general productivity determinants De Kok et al (2006) Huergo and Jaumandreu (2004) Dhawan (2001) Data from Xth Capitalia - UniCredit survey conducted in Italia from 2004 to 2006 Two-stage approach: - Stage 1: TPF is determined using Levinsohn and Petrin (2003) approach - Stage 2: OLS estimation: regressing TPF obtained from stage with firm age, family-managed status and vector of other control variables (firm size, human capital, firms listed in stock market, capital intensity, and ownership concentration.) - Productivity of non-family firms is higher than family-managed firms, but the different is small (3.5%-5%) - Firm age has positive impact on TFP of family-managed firms n oi m ll fu at nh z z Data from Production survey for Dutch manufacturing industry firms from 1994-1999 - Productivity is calculated using growth accounting method - OLS estimation with dependent variable is productivity growth and independent variables are firm age and control for size, organizational structure and sectors in manufacturing industry - Semiparametric regression for assumed nonlinear relationship between age and productivity growth which calculated from Solow residual - Adding innovation variable to the regression to estimate the relationship between innovation and productivity growth k - Pioneer on analyzing the relationship between family ownership status and TFP in Italia context - Determine the impact of firm age on TFP in family owned firm in Italia Estimate the relationship between firm age and productivity growth Key conclusions jm Cucculelli et al (2014) Methodology ht Data vb Objectives va References - At the first few years of operation, firm productivity is lower than average - After ten years of operation, firm age have no impact on productivity om l.c an Lu Unbalanced panel data set of over 2300 Spanish manufacturing firms from 1990 to 1998 gm va n - Defined productivity as median value of capital and labor ratio y te - Firms at early stage enjoy high productivity growth (at around (5%) and this rate decreases continuously for years until equal the average productivity (about 2%) - Additionally control for innovation results in higher productivity growth but then decreases after years - Small firms pay higher interest rate than large firms ac th Data from COMPUSTAT database from 1970-1989, US re Examine the relationship between innovation introduction and productivity growth - If this relationship exist, whether firm life span matter? Examine the effect of firm size on the firm productivity in US si g e cd 62 jg hg ad ju y th yi pl n ua al n va fu - OLS estimation of Cobb-Douglas production function is regressed separately for large firms and small firms to examine whether small firms and large firms are different in cost of capital - Construct equation on productivity differential between small and large firms - Productivity is estimated using Stochastic Frontier Analysis TFP is decomposed into three components: (i) pure technical efficiency change, (ii) scale efficiency change, (iii) technical change - Through Scale efficiency change and Technical change, different in firm size can explain productivity differential Tovar et al (2011) Examine the effect of firm size on the performance of Brazilian Electricity Industry Panel data from 17 Brazilian electricity distribution firms from 1998 to 2005 Margaritis and Psillaki (2010) Firm level panel data of France from 2002 to 2005 in three industries: Chemicals, Computers and Textiles - Using Data Envelopment Analysis and distance function approach to estimate firm efficiency - Firm leverage is measured as debt to total asset ratio - Dynamic OLS estimation to identify the relationship among firm efficiency, leverage and ownership status, other control variables: profitability (profit/total assets), asset structure (fixed tangible assets/total assets), growth opportunities (intangible assets/total assets), size Kim (2006) Examine the relationship between firm leverage and their performance - Test whether firm efficiency has any effect on their choice of capital structure - Determine the effect of ownership on firm capital structure and efficiency Assess the relationship between family ownership Panel data of Korean manufacturing firms from 1991 to 1998 - Define Chaebols firms as family-managed, debtdependent, diverse business activities oi m ll - Small firms are more productive than large firms at nh z z k jm ht vb om l.c gm - In the period of 1998-2005, TPF growth is at 0.9%/year, Technical change growth rate at 4.9%/year, Pure technical efficiency change growth rate at -3.7%/year, unchanged at Scale efficiency change - Firm size is positively correlated with TFP growth Therefore mergers of small electricity distribution firms could lead to gain in productivity - At low level of leverage, firm's level of efficiency has positive affect on firm leverage - Firm efficiency also has relation to assets structure, growth opportunities, size, profitability, ownership status an Lu va n y te re ac th - Family ownership concentration has significant positive impact on firm productivity, but at decreasing rate si g e cd 63 jg hg ad ju y th yi pl - Productivity is measured as TFP using multilateral index approach - Pooled OLS estimation between TFP and family ownership concentration, capital structure and other control variables n ua al statuses, capital structure on firm productivity n va m ll fu Aw, Roberts and Winston (2007) Analyze firm behavior in entering export market, investing in R&D and investing in improving labor force quality - Assess the impact of these above activities on firm productivity Panel data of Taiwanese firms in electronic industry in 1986, 1991 and 1996 Keller and Yeaple (2009) Examine the effect of FDI and international trade on productivity growth in US COMPUSTAT panel data of 1,277 US firms from 1987 to 1996 oi at nh z z - Applied standard bivariate probit model to analyze firm decision on participate export market and invest in R&D/worker training The explanatory variables: firm age, capital stock, wages, productivity (calculated using OP approach), interaction terms, etc - Using maximum likelihood estimate to identify the determinants of firm decision to survive - Using maximum likelihood estimate to determine the marginal contribution of investment in R&D/worker training on future productivity k jm ht vb om l.c gm an Lu - Firm productivity is determined using Olley and Pakes (1996) method - FDI activities is represented by the share of MNE affiliate industry employment - OLS estimation for the relationship between productivity growth and FDI and international trade va n y te re - Significant negative effect of debt ratio on productivity - Family ownership concentration affect productivity in Chaebols much stronger than non-Chaebols firms - Debt ratio has positive impact on Chaebols firms’ productivity - Firms with higher productivity are more likely to involve in export However this effect is not significant with choice to invest in R&D/worker training - Key determinants of firm's survival are entrant status and capital stock - Current productivity level does affect future productivity Investment in R&D/worker training and export experience have significant positive relationship with firm productivity, and the relationship between these two factors are complement - FDI activities have significant positive impact on TPF growth (account for 11% TFP growth) The relationship is then robust - Import affect TFP growth with the same direction, but weaker than FDI In addition, there is no evidence on robust result on import and TFP growth ac th si g e cd 64 jg hg ad ju y th yi pl Reference Data n ua al APPENDIX 2: Empirical studies on relationship between innovation and firm’s performance Siedschlag, Zhang and Cahill (2010) Panel data of 723 firms from Community Innovation Survey of Ireland in period of 2004-2008 Labor productivity Belderbos, Carree and Lokshin (2004) Labor productivity growth/Innovati ve sales (products that are new to the market) productivity growth Crespi Pianta (2009) Panel data from Community Innovation Survey in Netherlands in 1996 and 1998 for 2056 manufacturing firms 32 industries (including manufacturing n va Output measurement Methodology Conclusion Innovation input: innovation expenditure (rather than R&D expenditure) per employee - Innovation output: Three types of innovation dummies: product, process and organizational innovation; interactions; innovation sales share - Innovation expenditure per sales - Dummy variables for four types of R&D cooperation: competitors, suppliers, customers, universities or other research institutions -CDM model contains three stages: (1) firm's decision to invest in innovation; (2) determine innovation output using innovation inputs and (3) innovation output and other production inputs in the relationship with final output production (1) Foreign owned firms and domestic firms involved in export activities are more likely to invest in innovation than firms with domestic activities only (2) Foreign owned firms and domestic firms involved in export activities are more likely to have innovation output Innovation expenditure have no significant effect on innovation output (3) Innovation outputs have positive relationship with labor productivity oi m ll fu Innovation measurement at nh z z k jm ht vb om l.c Different types of R&D collaboration and innovation intensity significantly and positively affect productivity growth (but innovation intensity on innovative sales share) an Lu va n Diversify the impact of innovation on productivity growth through three separated models: general Product and process innovations; effort to build technological competitive advantages, y te - Innovation expenditure per employee re ac th Labor productivity growth gm and IV regression between output variable and innovation variables, control for firm size, 2-digit industry dummies, ownership status, demand-pull and cost-push Productivity in the previous period also is included si g e cd 65 jg hg ad ju y th yi pl n ua al and service sectors) in Europe countries from 1996 to 2001 n va model (reflect product and process innovation); technological competitiveness model and cost competitiveness model cost advantages significantly contribute to growth in productivity Cobb-Douglas production function to estimate the impact of innovation activities on output growth, instrumented by lag of ln(output/labor), ln(material/labor), ln(capital/labor), R&D intensity, size) - Tornquist index of TFP is regressed against innovation variables, instrumented by the same set of variables above - Descriptive and quantitative approach: factor analysis - Structural equation modelling - Positive impact of process and product innovation on productivity In details, the impact of process innovation is bigger than product innovation These results are robust in both approaches: TFP growth and CobbDouglas production function estimation - R&D intensity does not have significant impact on productivity oi m ll fu Technological competitiveness: percentage of firms have patents; percentage of firms targeting on improving product quality - Cost competitiveness: spending on acquisition of new machinery per employee; percentage of firms targeting on flexible the production process -Process and product innovation dummies - R&D expenditure as percentage of output at nh z z om an Lu ROA ROS ROE l.c Data of Brazilian firms in 2000, 2003, 2005 in which: Santos, Basso, Kimura and Kayo (2014) gm k - Gross output growth - TFP growth (TFP is determined using Levinsohn and Petrin (2003) approach) jm 941 manufacturing firms in Italy from two surveys in 1995 and 1998 ht Parisi, Schiantarelli and Sembenelli (2006) vb va n Innovation efforts from innovative investment not significant explain firm's performance Causality relationship between innovation and performance y te re ac th - Human capital - Ratio of training expenditure to sales - Ratio of internal R&D expenditure to sales si g e cd 66 jg hg ad ju y th yi pl - Ratio of acquisition of machinery expenditure to sales - Ratio of external R&D expenditure to sales - Ration of introduction of technological innovation expenditure to sales oi m ll at nh z z - Innovation orientation Innovation input measurements: internal (R&D expenditure) and external (R&D cooperation) Innovation output measurements (innovation sales, patents) Labor productivity: net value added per employee Internal R&D expenditure External R&D expenditure (contracted R&D) k jm ht Meta-analyses on 42 empirical studies of 21,270 SMEs om l.c gm an Lu va GMM estimation for the dynamic panel equation from augmented Cobb-Douglas production function Innovation variables included in the model are: internal n y te re 304 Netherlands manufacturing firms from 1996 to 2001 - Positive effect of innovation on firm's performance when innovation is measured in three types, but the effect of innovation orientation is larger than innovation outcomes - Effect of innovation on firm's performance in young firms is larger than in more established firms - Internal innovation activities have significant impact on performance while innovative collaboration with external involves have no significant impact on performance - Cultural environment also have impact on the innovation-performance relationship - Internal and external R&D are complement in the relationship with productivity with decreasing returns to scales effect - Internal R&D have positive impact on firm's productivity, this relationship is robust with ac th Categorized into three types: accounting returns - growth - stock market performance vb Lokshin, Belderbos and Carree (2008) fu n Rosenbusch, Brinckmann and Bausch (2011) va n ua al - Innovation information from PINTEC of Brazilian Institute of Geography and Statistics Financial information from Serasa and Gazeta Mercantil 42 empirical studies on 21,270 SMEs si g e cd 67 jg hg ad ju y th yi pl n ua al R&D expenditure, external R&D expenditure, quadratic forms and interaction form different dynamic techniques panel n va oi m ll fu at nh z z k jm ht vb om l.c gm an Lu va n y te re ac th si g e cd 68 econometric jg hg APPENDIX 3: Durbin – Wu Hausman test for endogeneity t to IV (2SLS) estimation ng hi ep Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only w n lo Number of obs = ad F( ju y th Total (centered) SS yi Total (uncentered) SS 1942) = 50.27 Prob > F = 0.0000 = 386439909.4 Centered R2 = -0.9527 = 655444519.7 Uncentered R2 = -0.1513 754587147.2 Root MSE pl = = 622.2 n ua al Residual SS 6, 1949 -Std Err z n Coef va TFP | P>|z| [95% Conf Interval] fu -+ ll -2.06 0.039 -3231.739 -83.4112 471.593 oi -1657.575 803.1596 dummy | 968.7459 0.040 44.44065 1893.051 706479 0460183 15.35 0.000 6162849 7966732 lntotalassets | -21.87988 28.06836 -0.78 0.436 -76.89286 33.13309 capitalstructure | -372.9817 203.9315 -1.83 0.067 -772.6802 26.71677 firmage | -3.555249 2.472536 -1.44 0.150 -8.40133 1.290831 _cons | 598.6118 444.9574 1.35 0.179 -273.4886 1470.712 m invexpintensity | 2.05 nh | at TFP | z z L1 | k jm ht vb | l.c gm Underidentification test (Anderson canon corr LM statistic): 0.0081 15% maximal IV size 8.96 20% maximal IV size 6.66 25% maximal IV size 5.53 Source: Stock-Yogo (2005) Reproduced by permission 69 ey 16.38 t re Stock-Yogo weak ID test critical values: 10% maximal IV size n 7.005 va Weak identification test (Cragg-Donald Wald F statistic): an Lu Chi-sq(1) P-val = 7.005 om -Sargan statistic (overidentification test of all instruments): 0.000 (equation exactly identified) t to -endog- option: ng Endogeneity test of endogenous regressors: 12.577 hi Chi-sq(1) P-val = ep Regressors tested: 0.0004 invexpintensity w Instrumented: invexpintensity n lo Included instruments: dummy L.TFP lntotalassets capitalstructure firmage ad Excluded instruments: L.invexpintensity y th ju yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re 70 APPENDIX 4: Durbin – Wu Hausman test for endogeneity t to IV (2SLS) estimation ng hi ep Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only w n lo Number of obs = ad F( ju y th Total (centered) SS yi Total (uncentered) SS 2899) = 155.87 Prob > F = 0.0000 = 467122455.9 Centered R2 = 0.2650 = 799333085.9 Uncentered R2 = 0.5705 343351960.1 Root MSE pl = = 343.7 n ua al Residual SS 7, 2907 -Std Err z n Coef va TFP | P>|z| [95% Conf Interval] fu -+ ll -1.38 0.167 -955.4834 164.9064 18.7009 oi -395.2885 285.819 dummy | -17.08177 0.361 -53.73487 19.57132 6299597 0219493 28.70 0.000 5869398 6729796 invexpintensity | 27.09339 22.10942 1.23 0.220 -16.24029 70.42706 lntotalassets | 42.31543 6.82258 6.20 0.000 28.94342 55.68744 capitalstructure | -35.67232 26.2863 -1.36 0.175 -87.19252 15.84788 firmage | -1.741898 9786339 -1.78 0.075 -3.659985 176189 _cons | -356.4369 84.88587 -4.20 0.000 -522.8101 m highqualitylabor | -0.91 nh | at TFP | z z L1 | k jm ht vb | l.c gm om -190.0636 Chi-sq(1) P-val = 313.770 0.0000 16.38 15% maximal IV size 8.96 20% maximal IV size 6.66 25% maximal IV size 5.53 71 ey Stock-Yogo weak ID test critical values: 10% maximal IV size t re 350.767 n Weak identification test (Cragg-Donald Wald F statistic): va an Lu Underidentification test (Anderson canon corr LM statistic): Source: Stock-Yogo (2005) Reproduced by permission -Sargan statistic (overidentification test of all instruments): 0.000 t to (equation exactly identified) ng -endog- option: hi Endogeneity test of endogenous regressors: 6.684 ep Chi-sq(1) P-val = Regressors tested: 0.0097 highqualitylabor w n highqualitylabor lo Instrumented: ad Included instruments: dummy L.TFP invexpintensity lntotalassets capitalstructure y th firmage ju Excluded instruments: L.highqualitylabor yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re 72