Experiential knowledge definition

experimental systems future knowledge in artistic research and experimental knowledge with empirical knowledge
Dr.ThorasRyder Profile Pic
Dr.ThorasRyder,Hong Kong,Researcher
Published Date:07-07-2017
Your Website URL(Optional)
Comment
WIPO ECONOMIC RESEARCH WORKING PAPERS BASIC, APPLIED AND EXPERIMENTAL KNOWLEDGE AND PRODUCTIVITY: FURTHER EVIDENCE Mosahid Khan Kul B Luintel Working Paper No. 2 December 2010Basic, Applied and Experimental Knowledge and Productivity: Further Evidence (1) (2) Mosahid Khan and Kul B Luintel Abstract Analyzing a novel dataset we find significantly positive effects of basic, and applied and experimental knowledge stocks on domestic output and productivity for a panel of 10 OECD countries. This letter updates the work of, among others, Mansfield (1980), Griliches (1986) and Adams (1990), at an international setting. JEL Classification: F12: F2: O3. Key Words: Basic and Applied Research; TFP; Panel Co-integration. Disclaimer We thank Peter Pedroni for his useful comments. The views expressed in this paper are those of the authors, and do not necessarily reflect those of the World Intellectual Property Organization or its Member States. The usual disclaimer applies. (1) World Intellectual Property Organization, Geneva, Switzerland; e-mail: mosahid.khanwipo.int (2) Cardiff Business School, Cardiff, United Kingdom; e-mail: luintelkcardiff.ac.uk Contact information: Economics and Statistics Division, World Intellectual Property Organization, 34 Chemin des Colombettes, P.O. Box 18, CH-1211 Geneva 20, Switzerland, e-mail: chiefeconomistwipo.int Working papers and available for download free of charge at: www.wipo.int/econ_stat 1 I. Introduction R&D activities are grouped into three distinct types: basic research, applied research and experimental development. Frascati Manual (2002) defines basic research as “experimental or theoretical work undertaken primarily to acquire new knowledge… without any particular application or use in view (p.77)”. National Science Foundation defines it as “original investigation for the advancement of scientific 1 knowledge…which do(es) not have immediate commercial objectives”. These distinctions imply that basic research is fundamental to knowledge breakthroughs. Economists and policy makers have long debated its role on productivity. Mansfield (1980, p. 863) succinctly puts it: “A hotly debated topic among economists, scientists, technologists and policymakers is: Does basic research, as contrasted with applied research and development, make a significant contribution to an industry’s or firm’s rate of technological innovation and productivity change?” Griliches (1986, p. 145) asks: “whether different types of R&D (basic vs. applied) are equally potent in generating productivity growth”. Whilst there is large empirical literature on R&D and productivity, studies linking basic research and applied and experimental development to productivity are rare. Mansfield (1980), for the first time, tested this debate on US micro data and found significantly positive 2 effects of basic and applied research on productivity growth. Grilliches (1986) confirmed this with the 3 proviso that his results are based on “level regressions” and may suffer from “biases” (p. 147). Succeeding studies on this issue are sparse. Furthermore, a study that captures basic versus applied and experimental knowledge across all R&D performing institutions is lacking. This letter bridges this gap. We measure types of knowledge across all institutions: academic, business, government and private non- profit sector. This is distinct from existing studies confined to particular institutions only. We also incorporate the measures of foreign knowledge stocks. Thus, we extend this topic to an international setting corresponding to the recent literature on international R&D spillover. We use non-stationary panel data econometrics which addresses the concerns of level regressions. II. Specification We estimate separate models for output and productivity. Following Mansfield (1980), Griliches (1986), Adams (1990) and Coe et al. (2009), an augmented Cob-Douglas production function that permits types of knowledge stocks as factor inputs is: b ae f log y =α +β logk +β logl +β logh +β log s + β log s +β log s + e (1) it i k it l it h it b it a it f it it where ‘i’ denotes countries (i=1,…,N) and ‘t’ is the time subscript. y , k , l and h respectively denote it it it it b ae f real output, physical capital stock, labor input and the stock of human capital. s , s ands respectively it it it denote the stocks of basic, applied and experimental, and foreign knowledge stocks. α are country- i specific intercepts andβs are the respective point elasticities. We specify a productivity relationship: b ae f lo g tfp = θ + λ lo g h + λ lo g s + λ lo g s + λ lo g s + ε ( 2 ) it i h it b it a it f it it where tfp is domestic total factor productivity; θ and λs are parameters. Equation (2) is directly it i obtained from equation (1) by imposing constant returns to scale on capital and labor - a well-known specification in the literature. In estimations, we employ four types of foreign knowledge stocks, in turn (see below). III. Data and Sample 1 Mansfield (1980, p. 863). 2 “My results seem to be the first data on this subject, about which there is so much discussion (Mansfield, op. cit, p. 863)”. 3 See also Link (1981). 2 4 We analyze an unbalanced panel of 10 OECD countries with 346 observations. R&D expenditure data on basic research, applied research and experimental development are used to compute respective b ae stocks -S andS - through perpetual inventory method (PIM) at 15% and 10% depreciation rates. The it it foreign knowledge stocks are computed employing import ratios as weights. For example, the foreign f−b th basic knowledge stock for the i country (s ) is: it N−i f−b b s = (m / y )s (4) it ∑ ijt jt jt j=1 b where, y is GDP of country j; m is the capital goods imports of country i from country j; s denotes the j ij jt basic knowledge stock of j; (j=1,…, N-1) and N=10. Likewise, we compute foreign applied and f −ae f −bus experimental R&D capital stocks (s ), foreign business sector R&D capital stocks (s ) and foreign it it f −tl 5 total R&D stock (s ) for each of the sample country. k is computed from the fixed capital formation it it using PIM at 8% depreciation rate. All data are from OECD except the tfp andh , which respectively are it it from the European Commission and Bassanini and Scarpetta (2002). IV. Empirical Results The panel unit root tests proposed by Im, Pesaran and Shin (2003) and Fisher-ADF (Maddala and WU, 1999) both confirm that our panel data are unit root processes. For brevity, results are available on request. We apply Pedroni’s (1999) group-t-statistic (parametric) for co-integration test as it (i) allows for heterogeneous co-integrating vectors across panel units, and (ii) is the most powerful test (Pedroni, 2004). The co-integrating parameters are estimated by FMOLS. Table 1 reports the results for output. Griliches (1986) and Adams (1990) highlight the importance of the b lag ofs ; we estimate up to its fourth order lag. Data limitations precluded us to venture beyond four lags. it b ae Three models, showing alternative use ofs ands , are reported under each lag. Column (i) would be it it b identical across all lags because it excludess . it Panel A reports the group-t-statistic which rejects non co-integration across all specifications. All models f −bus b ae are co-integrated. Panel B reports the co-integrating parameters when s is included. s and s are it it it ae b positive and significant throughout. s shows bigger point elasticity than that ofs which peaks at L=2 it it suggesting that the former’s effect is eleven times larger. This may seem dramatic but the parameter of ae ae b s are not unreasonably high. This simply implies that domestically s appears more important than s it it it f −bus vis-à-vis output, which is plausible. s and l are also positive and significant. h is positive and it it it significant in all models but one, column (iii) under L=4. k appears insignificant in column (iii) except for it L=4, which is due to collinearity. We regress h on k and l and use the resulting residual series as it it it 0 orthogonalized human capital (h ). This improves the significance of k without affecting qualitatively any it it other estimates (compare columns (iii) and (iv) across all lags). Panel C reports the results from the other f−b f −ae f −tl three measures of foreign knowledge stocks - s ,s and s . Their uses, in turn, in equation (1) do it it it f−b f −tl 6 not alter the qualitative nature of other parameters of panel B. s and s are significant throughout. it it f −ae s appears mostly significant under L=1 and L=2 but largely insignificant at L=3 and L=4. The it f −tl f −bus international spillover effects of s are somewhat higher than those ofs which is plausible. Both it it f −tl f −bus f−b f −ae s and s show larger effects than those ofs and s . it it it it 4 Sample countries are: Australia (29), France (37), Iceland (36), Ireland (37), Italy (37), Japan (32), Portugal (36), Norway (37), Spain (28) and USA (37); where (.) indicates annual data points. The longest sample of 37 data points pertain to 1970-2006 and the shortest 28 data points spans for 1979-2006. f s it 5 is usually computed from within the sample but, data permitting, we see no reason to restrict international knowledge spillovers f −b f −ae to mere 9 countries as we have 10 sample countries. Therefore, due to data constraints, our measures of ands are s f −bus it it s f −tl it based on 10 sample countries but and s embrace other 19 OECD countries. it f−b 6 The only exception is s in column (ii) under L=3. it 3 b ae f −bus Table 2 reports TFP results. All models are co-integrated. Panel B shows thats ,s and s are it it it ae positive and significant throughout. With regard to TFP, the parameter ofs appear bigger than those of it b s in most cases, nonetheless, the difference is not as large as before. h appears insignificant in several it it ae ae specifications which is due to collinearity with s . Column (iv), which uses the orthogonalized s Oae f −tl 7 (i.e.,s ), resolves the problem. As before, s is significant throughout (Panel C); the significance of it f−b b f −ae s is more prominent at the higher lags ofs . s shows mixed results, consistently significant at the it it it b f th 4 lags of s only. The use of these alternative measures ofs , in turn, does not change the qualitative it nature of other parameters in panel B. b Results are robust to knowledge stocks calculated at 10% depreciation rate. The significance of s and ae f s remains to alternative weightings by bilateral R&D collaboration or FDI flows for computings . Our it b findings of the positive contributions of s are consistent with Mansfield (1980), Griliches (1986) and it ae Adams (1990) whereas we find more robust contribution of s than Mansfield (op. cit). On international it 8 knowledge spillovers, our findings are consistent with the literature (e.g., Coe et al., 2009). V. Conclusion Two types (basic vs. applied and experimental) of knowledge stocks are measured across all players in the R&D sector. Both contribute to domestic output and productivity. The international knowledge spillovers associated with basic R&D, total R&D and business sector R&D appear prominent but those with applied and experimental R&D appear less robust. Evidence is consistent that basic knowledge exerts its effects over a long period. ae Oae 7 s is regressed on h and the residual iss . t it 8 Luintel and Khan (2004) argue that, with sufficiently long time series, one approach to modelling would be to check cross-country data poolability. This issue is not pursued here. 4 References Adams, J. D., (1990), “Fundamental Stock of Knowledge and Productivity Growth”, Journal of Political Economy, 98, 673-702. Bassanini, A., and Scarpetta, S., (2002), “Does human capital matter for growth in OECD countries? A pooled mean-group approach”, Economics Letters 74(3), 399-405. Coe, D. T., Helpman, E. and Hoffmaister, A. W, (2009), “International R&D Spillovers and Institutions,” European Economic Review, 53, 723-741. Frascasti Manual (2002), OECD. Griliches, Z., (1986), “Productivity, R&D and Basic Research at the Firm Level in the 1970’s”, American Economic Review, 76, 141-154. Im, K.-S., Pesaran, H., and Shin, Y. (2003), “Testing for unit roots in heterogeneous panels”, Journal of Econometrics 115, 53-74. Link, A. N., (1981), “Basic Research snd Productivity Increase in Manufacturing: Additional Evidence”, American Economic Review, 71, 111-112. Luintel, K. B. and Khan M., (2004), “Are International R&D Spillovers Costly for the U.S?”, The Review of Economics and Statistics, LXXXVI, 896-911. Mansfield E., (1980), “Basic Research and Productivity Increase in Manufacturing”, American Economic Review, 70, 863-873. Maddala, G, and Wu, S., (1999), “A Comparative Study of Unit Root Tests with Panel Data and a New Simple Test”, Oxford Bulletin of Economics and Statistics, 61, 631-652. Pedroni, P., (1999), “Critical values for cointegration tests in heterogeneous panels with multiple regressors”, Oxford Bulletin of Economics and Statistics (Special Issue), 653–670. Pedroni, P., (2004), “Panel cointegration: asymptotic and finite sample properties of pooled time series tests, with an application to the PPP hypothesis. Econometric Theory, 20, 597–625. 5 Table 1: Results for Output b ae f log y =α + β log k + β logl + β log h + β log s + β log s + β log s + e it i k it l it h it b it e it f it it Panel A: Panel co-integration tests L=1 L=2 L=3 L=4 (i) (ii) (iii) (iv) (ii) (iii) (iv) (ii) (iii) (iv) (ii) (iii) (iv) Grou - - - - - - - - - - - - - p-t- 2.59 4.38 2.96 2.96 4.17 3.74 3.74 2.98 4.78 4.78 1.56 2.48 2.48 a a a a a a a a c a stats 1 4 5 5 3 9 9 0 3 3 8 5 5 Panel B: FMOLS Results 0.11 0.32 0.12 0.20 0.21 0.09 0.18 0.15 0.09 0.16 0.11 0.02 0.11 a a a b b a b a a b 4 8 3 0 6 8 5 4 5 9 9 2 9 k it 6.54 2.66 0.81 3.32 1.99 0.24 2.42 2.73 1.34 2.21 3.81 2.80 2.51 3 5 4 9 8 3 8 3 4 5 5 1 9 0.73 0.48 0.64 0.58 0.50 0.58 0.51 0.57 0.53 0.47 0.54 0.53 0.45 a a a a a a a a a a a a a 0 5 2 0 5 3 3 8 1 1 9 0 3 l it 12.8 12.1 14.0 10.5 12.0 14.6 10.8 12.8 14.4 11.0 12.7 13.4 10.8 3 5 2 3 3 5 4 9 6 2 5 6 2 0.25 0.26 0.40 0.57 0.45 0.73 0.39 0.85 0.51 a a a a a a a a 2 4 5 6 9 8 1 3 1 h - - - - it 2.98 6.46 4.89 7.20 4.76 5.83 3.04 5.58 1.53 5 5 9 2 5 3 4 5 2 0.40 0.45 0.39 0.51 a a a o 5 9 1 1 h - - - - - - - - - it 4.89 4.76 3.04 1.53 9 5 4 2 0.13 0.15 0.15 0.17 0.17 0.15 0.15 0.11 0.11 a a a a a a a a a ae 4 8 8 0 0 0 0 1 1 - - - - s it 3.95 3.73 3.73 3.82 3.82 3.38 3.38 3.17 3.17 9 6 6 1 1 8 8 8 8 0.04 0.06 0.02 0.02 0.09 0.04 0.04 0.09 0.05 0.05 0.11 0.07 0.07 f −bus a a a a a a a a a a a a a s 5 8 9 9 2 7 7 2 1 1 1 3 3 it 4.72 8.48 4.77 4.77 9.87 6.40 6.40 8.91 4.90 4.90 10.2 6.33 6.33 7 2 8 8 6 0 0 2 2 2 1 5 5 0.08 0.02 0.02 a a a b 2 0 0 - - - - - - - - - - s it−1 4.14 2.70 2.70 3 7 7 0.08 0.01 0.01 a a a b 6 5 5 s - - - - - - - - - - it−2 4.59 3.49 3.49 2 4 4 0.09 0.03 0.03 a a a b 9 8 8 s - - - - - - - - - - it−3 5.31 4.02 4.02 4 7 7 0.09 0.06 0.06 a a a b 5 3 3 s - - - - - - - - - - it−4 5.11 4.66 4.66 7 9 9 f−b f −ae Panel C: Foreign Knowledge Stocks based on Basic (s ), Applied and Experimental (s ) and it it f −tl total (s ) R&D. it 0.03 0.02 0.01 0.01 0.02 0.02 0.02 0.02 0.02 0.02 0.03 0.04 0.04 a b a a b a a c c b a a f−b 1 4 8 8 9 8 8 0 8 8 7 8 8 s it 4.38 2.21 4.01 4.01 2.12 3.47 3.47 1.01 1.91 1.91 2.29 3.18 3.18 4 8 1 1 5 8 8 7 0 0 1 9 9 0.00 0.02 0.02 0.02 0.02 0.01 0.00 0.00 0.01 0.01 0.01 0.00 0.00 c f −ae 5 5 9 9 a a a b b 0 7 8 8 9 3 3 9 9 s it 1.91 0.94 0.94 2.60 1.08 2.91 2.91 1.20 2.51 2.51 0.14 0.14 0.01 4 6 6 3 0 1 1 9 9 9 0 0 0 6 0.06 0.08 0.04 0.04 0.10 0.05 0.05 0.10 0.06 0.06 0.12 0.08 0.08 f −tl a a a a a a a a a a a a a s 0 1 1 1 3 8 8 1 0 0 0 3 3 it 6.16 8.60 5.76 5.76 9.59 7.00 7.00 8.79 5.48 5.48 10.5 7.15 7.15 2 7 5 5 0 4 4 7 0 0 6 5 5 For details, please refer notes to Table 2. Table 2: Results for Total Factor Productivity b ae f lo g tfp = θ + λ log h + λ lo g s + λ lo g s + λ lo g s + ε . it i h it b it a it f it it Panel A: Co-integration Test L=1 L=2 L=3 L=4 (i) (ii) (iii) (iv) (ii) (iii) (iv) (ii) (iii) (iv) (ii) (iii) (iv) Grou - - - - - - - - - - - - - p-t- 3.02 1.90 2.92 2.92 2.67 3.49 4.72 2.67 3.49 3.49 1.92 2.96 2.96 a b a a a a a a a a b a a stats 1 5 9 9 4 7 3 4 7 7 0 9 9 Panel B: FMOLS Results. 0.42 0.42 0.24 0.54 0.43 0.27 0.70 0.70 0.32 0.75 1.03 0.33 0.74 a a a c a b a 3 1 2 4 5 3 4 3 5 9 4 1 0 h 2.43 1.44 0.00 2.80 1.37 0.23 3.06 1.77 0.12 3.04 2.26 - 3.37 it 0 5 0 9 8 8 0 0 3 0 8 0.35 5 6 0.17 0.06 0.09 0.10 0.09 a b a a a ae 0 9 9 0 4 s - - - - - - - - it 8.25 2.18 2.79 3.27 4.12 5 6 9 8 6 0.06 0.09 0.10 0.09 Oae b a a a 9 9 0 4 s it - - - - - - - - - 2.18 2.79 3.27 4.12 6 9 8 6 0.03 0.03 0.03 0.03 0.05 0.04 0.04 0.07 0.05 0.05 0.09 0.05 0.05 f −bus a a a a a a a a a a a a a s 1 7 2 2 1 1 1 6 1 1 5 9 9 it 2.63 4.79 3.58 3.58 5.83 4.18 4.18 6.61 4.73 4.73 7.47 5.58 5.58 3 8 3 3 5 5 5 8 2 2 6 9 9 0.14 0.10 0.10 a a a b 9 9 9 s - - - - - - - - - - it−1 7.50 3.23 3.23 4 1 1 0.13 0.07 0.07 a a a b 1 0 0 s - - - - - - - - - - it−2 7.65 3.55 3.55 2 0 0 0.07 0.06 0.06 - a a a b 2 2 2 - s - - - - - - - - - it−3 7.60 4.16 4.16 5 3 3 0.00 0.06 0.06 a a a b 7 3 3 - - - - - - - - - - s it−4 7.38 4.07 4.07 3 7 7 f−b f −ae Panel C: Foreign Knowledge Stocks based on Basic (s ), Applied and Experimental (s ) and it it f −tl total (s ) R&D. it - - 0.06 0.05 0.05 0.01 0.12 0.12 0.01 0.01 0.01 0.03 0.03 a a a b b 0.00 0.00 b a a a 1 3 3 f−b 6 6 9 5 5 4 9 9 s 4 4 5.24 4.40 4.40 it 1.86 2.34 2.55 1.01 1.01 3.50 2.85 2.85 0.20 0.20 3 7 7 2 7 6 4 4 7 3 3 5 5 - - - - - - 0.05 0.03 0.03 0.01 0.01 0.01 0.01 a a a f −ae 0.00 0.01 0.01 0.01 0.00 0.00 7 1 1 c a s 3 1 5 5 it 1 0 4 4 4 4 4.97 2.71 2.71 1.94 3.04 1.07 1.07 0.67 1.30 - - - - 4 7 7 5 2 2 7 0.58 0.58 0.00 0.00 f −tl 0.04 0.03 0.03 0.03 0.05 0.04 0.04 0.08 0.06 0.06 0.10 0.07 0.07 s a a a a a a a a a a a a a it 3 8 5 5 2 9 9 1 2 2 6 1 1 7 3.47 5.08 3.86 3.86 6.10 4.53 4.53 7.03 5.38 5.38 8.11 6.56 6.56 6 7 9 9 8 8 8 0 1 1 4 3 3 Panel A contains group-t-statistic under the null of no co-integration. They are asymptotically standard normal left-sided tests. f All measures of s pertain to 15% depreciation rate. Superscripts a, b and c respectively denote significance at 1%, 5% and it 10%. . are t-ratios. Results are computed by RATS procedures. Section II contains variable definitions. L indicates lag length. 8