The contribution of research and innovation to productivity

Abstract

This paper examines the impact of investment in research and innovation on Australian market sector productivity. While previous studies have largely focused on a narrow class of private sector intangible assets as a source of productivity gains, this paper shows that there is a broad range of other business sector intangible assets that can significantly affect productivity. Moreover, the paper pays special attention to the role played by public funding for research and innovation. The empirical results suggest that there are significant spillovers to productivity from public sector R&D spending on research agencies and higher education. No evidence is found for productivity spillovers from indirect public funding for the business enterprise sector, civil sector or defence R&D. These findings have implications for government innovation policy as they provide insights into possible productivity gains from government funding reallocations.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Notes

  1. 1.

    For a concise summary and discussion of this and related work, see Parham (2006).

  2. 2.

    Haskel and Wallis (2013) is an updated and condensed version of the more comprehensive discussion paper, Haskel and Wallis (2010).

  3. 3.

    For example, Marrano and Haskel (2006) find that the private sector in the U.K. invested 10.1% of GDP on intangibles in 2004. In Finland, the private sector invested 9.1% of GDP in intangible assets (Jalava et al. 2007). The Netherlands invested 8.4% of GDP between 2001 and 2004 (van Rooijen-Horsten et al. 2008). Fukao et al. 2009 find that Japan invested 7.5% of GDP from 1995 to 2002 while Baldwin et al. (2012) find that the Canadian business sector invested 13.2% of GDP in intangible assets in 2008. Hao et al. (2009) conducted an international comparison between France, Germany and Italy and found that the shares of intangible investment in GDP in these three countries are 8.3, 7.1 and 5.2% respectively in 2004. Finally, the PC report suggests that Australia has invested 5.9% of GDP in 2005–06.

  4. 4.

    A third relevant study is Barnes (2010) in which the author extends the estimates of the PC report to a sectoral level.

  5. 5.

    ‘Given the experimental nature of the methodology, the assumptions required, measurement challenges and data limitations, the estimates should be interpreted as only indicative’ (Barnes and McClure 2009, p. XIII).

  6. 6.

    The caveat expressed by the authors of the earlier studies about the experimental nature of the estimates is also applicable to this paper.

  7. 7.

    The ABS defines computer software as: “… computer programs, program descriptions and supporting materials for both systems and applications software. Included are purchased software and, if the expenditure is large, software developed on own-account. It also includes the purchase or development of large databases that the enterprise expects to use in production over a period of more than 1 year. The ASNA does not separately identify databases from computer software as recommended by the 2008 SNA” (ABS (2015), p. 655). Hence we take the ABS estimates of “computer software” as our “computerised information”; see Elnasri and Fox (2014).

  8. 8.

    The implementation of the PIM for estimating intangible capital requires an estimate of initial period 0 capital stock, R 0. Different assumptions were made in previous studies to estimate R 0. For example, CHS (2006) assumed an initial stock of zero in a specific year for each asset while others, such as the PC report, have assumed a constant rate of investment growth for the period prior to the first data point for investment and applied the formula R 0=N0/(δ+g), where g is equal to the average annual growth rate of intangible investment over the period of the study.

  9. 9.

    As we are using ABS capital stocks for software, mineral exploration and artistic originals the depreciation rates are the average rates of the ABS for those assets. Others are the rates suggested by CHS. The depreciation rates are: 10% for Mineral Exploration; 20% for Computer Software, Business R&D, and Other Product Development, design and research; 40% for Firm-specific Human Capital and Organisational Capital; and 60% for Brand Equity and Artistic Originals.

  10. 10.

    “The tax parameter reflects the differing tax circumstances that owners of capital face….Since 1985, various research and development (R&D) tax concessions have been introduced to encourage increased investment in R&D by Australian companies. These concessions have had the effect of reducing rental prices on R&D considerably.” (ABS 2015, p. 451).

  11. 11.

    Actually, the timing in this user cost equation is not quite correct, as the \(i^tp_j^t\) term should actually be \(i^tp_j^{t - 1}\). The expression in (2) is now very common, perhaps originating from an error in the OECD Capital Manual, OECD (2001), Eq. (4), page 86. This has been corrected in the 2009 version of the Manual, but was used in e.g. ABS (2007) and the PC report. As the current study builds on intangibles the data of the PC report, for consistency we also use the same user cost equation. Compared to sensitivities relating to depreciation rates and the choice of interest rate, this timing issue is considered to be of second order importance.

  12. 12.

    The ABS methodology uses an endogenous rate of return unless the endogenous rate falls below the level of consumer price index (CPI) growth plus 4%. If the rate falls below this level, CPI growth plus 4% is used as the rate of return. In practice, the rate of return rarely rises above this mark and can therefore be considered to be an exogenous rate of return for most years.

  13. 13.

    See Diewert and Fox (2016) for more on alternative approaches to estimating user costs.

  14. 14.

    A detailed description of data sources for these variables is provided in Elnasri and Fox (2014).

  15. 15.

    The estimates of National Accounts intangibles, and thus the ensuing MFP indexes, developed in this paper are not identical to the ABS official estimates. Several factors may explain this. (i) There is a difference in the level of aggregation at which the estimates are constructed. Due to data limitations, the paper aggregates all assets in all industries in a single stage then uses rental prices to construct capital services. On the other hand, the ABS constructs capital services indexes for each of the twelve market sector industries separately then aggregates these indexes together using relevant weights, (ii) The ABS business expenditure on R&D (BERD) data includes some R&D related to financial services and architectural/engineering services. The scope of these types of R&D as discussed in CHS is broader than those activities that may be covered by the BERD survey. Thus, separate estimates for these types of R&D are developed and the ABS-based BERD estimates were reduced to avoid double counting. (iii) The rental prices and the PIM version used by the ABS to construct capital stock is more complex than the method used in this paper.

  16. 16.

    See Figs. 5 and 6 of Elnasri and Fox (2014).

  17. 17.

    An ABS survey on public spending on R&D captures R&D expenditure at the points at which R&D is performed. However, several technical challenges make the outlays data from the SRIBTs not strictly comparable with the R&D expenditure data captured by the ABS; see Matthews and Howard 2000 and Elnasri and Fox (2014) for more discussion on this issue.

  18. 18.

    These arrangements are known as ‘performance based’ because allocations to each institution depend on its past ‘performance’ as assessed by various formulae administered through the Department of Education, Employment and Workplace Relations.

  19. 19.

    Other public R&D agencies include the Australian Nuclear Science and Technology Organisation (ANSTO); Geoscience Australia; Antarctic Division; Australian Institute of Marine Science (AIMS); Bureau of Meteorology Research Centre; Environmental Research Institute of the Supervising Scientist; Australian Animal Health Laboratory; Great Barrier Reef Marine Park Authority; and the Anglo-Australian Telescope.

  20. 20.

    Most of the previous studies that examined the relationship between R&D and economic or productivity growth have avoided the problem of obtaining an estimate of R&D capital stock by employing a measure of R&D intensity (i.e. a ratio of R&D expenditures to the value of production). However, this method implicitly assumes that the depreciation rate of R&D is zero which is not necessarily a realistic assumption. The approach here is to use the stock of public sector R&D estimated by using PIM and assuming a depreciation rate identical to the business sector R&D.

  21. 21.

    The Solow (1957) approach is based on two simplifying assumptions: (i) competitive markets in which factors are rewarded according to their marginal products—so that the output elasticities can be represented by factor shares in total factor income, and (ii) constant returns to scale—so that factor shares sum to unity. A standard Törnqvist index is used to form the input aggregate. See e.g. Zelenyuk (2014) and references therein, for a recent treatment of statistical testing in a growth accounting framework.

  22. 22.

    Note that there is an important difference between the regression model of this paper represented by (7) and that of HW. Instead of using the stock of public sector R&D as in this paper, HW lagged the ratio of public sector R&D expenditure to GDP and used it as a regressor, assuming a zero depreciation rate of public sector R&D.

  23. 23.

    As suggested by Engle and Granger (1987), the augmented Dickey-Fuller (ADF) test (Dickey and Fuller 1979) is applied to the OLS residuals to test for the null hypothesis of no cointegration, using the Engle-Granger critical values. However, a major and widely cited drawback with the ADF test is the inherent low power when it is applied to short time series. The power of the test is the ability to reject the null of non-stationarity when it is false; because the ADF test has low power, it may suggest that a series has a unit root when it is actually stationary.

  24. 24.

    In addition to the above four control variables, and in line with Connolly and Fox (2006), the paper has attempted to include West Texas crude oil prices as a proxy to control for price shocks that might have a direct impact on the world energy market. However, this variable was dropped from the regressions as it had small and insignificant coefficient estimates.

  25. 25.

    A sizeable number of previous studies have found implausible results when regressing productivity on public infrastructure. Elnasri (2013) has suggested that one possible cause of such implausible results is due to the shortcomings of aggregate time-series analysis that make it unsuitable for the examination of the infrastructure spillovers to productivity. Approaches such as panel regressions and spatial econometrics are found to produce more acceptable results.

  26. 26.

    For more discussion on how long lead times between investment in new capacity in mining sector and the corresponding output can lead to short term movement in mining productivity and how the depletion of Australia’s natural resource had significant adverse effect on long-term mining productivity see Topp et al. (2008).

  27. 27.

    The Jarque-Bera test is a test for normality based on skewness and kurtosis. It tests the null hypothesis H o : normal distribution, skewness is zero and excess kurtosis is zero; against the alternative hypothesis H 1: non-normal distribution.

  28. 28.

    Durbin–Watson statistic (d) tests the null hypothesis H o that the errors are uncorrelated against the alternative hypothesis H 1 that the errors are autocorrelated. If the errors are white noise, d will be close to 2. If the errors are strongly autocorrelated, d will be far from 2.

  29. 29.

    These results for computerised information are broadly consistent with many studies that explored the earlier Solow (1987) productivity paradox: “You can see the computer age everywhere but in the productivity statistics”. However, such studies typically considered computerised information bundled in with physical high-tech capital, such as personal computers, and did not consider other types of intangibles. For more on this literature, see e.g. Triplett (1999), Brynjolfsson and Hitt (2000) and Connolly and Fox (2006), and the references therein.

  30. 30.

    With a small number of available observations, only one lag is used. HW did not use capital stock of public sector R&D in their regressions. Instead, they used two and three lags of the ratio of spending on public R&D relative to output, assuming by this a zero depreciation rate. Recalling that the PIM employed in this paper for constructing public knowledge capital includes all previous investment expenditures in the accumulation process, it is somewhat equivalent to the model of HW but with more lags.

  31. 31.

    Unlike estimates of the traditional MFP, adjusted estimates have the advantage of isolating social from private returns to knowledge capital. Insufficient attention has been paid to this refinement in the measured rates of return in past studies.

References

  1. ABS (2007) Experimental estimates of industry multifactor productivity. Australian Bureau of Statistics, Cat. no. 5260.0.55.001.

  2. ABS (2015) Australian system of national accounts concepts, sources and methods. Australian Bureau of Statistics, Cat. no. 5216.0

  3. Arrow K (1962) The economic implications of learning by doing. Rev Econ Stud 29:155–173

    Article  Google Scholar 

  4. Aw BY, Chen X, Roberts MJ (2001) Firm level evidence on productivity and turnover in Taiwanese manufacturing. J Dev Econ 66:51–86

    Article  Google Scholar 

  5. Baldwin JR (1995) The dynamics of industrial competition: a North American perspective. Cambridge University Press, Cambridge, MA

    Google Scholar 

  6. Baldwin JR, Gu W, Macdonald R (2012) Intangible capital and productivity growth in Canada. Research paper, the Canadian Productivity Review, Statistics Canada

  7. Barnes P (2010) Investments in intangible assets and Australia's productivity growth: Sectoral estimates. Productivity Commission, Staff Working Paper

  8. Barnes P, McClure A (2009) Investments in intangible assets and Australia's productivity growth. Productivity Commission, Staff Working Paper

  9. Bolaky F, Freund C (2004) Trade, regulations and growth. The World Bank, Policy Research Working Paper Series no. 3255

  10. Brynjolfsson E, Hitt LM (2000) Beyond computation: Information technology, organizational transformation and business performance. J Econ Perspect 14(4):23–48

    Article  Google Scholar 

  11. Connolly E, Fox K (2006) The impact of high-tech capital on productivity: evidence from Australia. Econ Inq 44(1):50–68

    Article  Google Scholar 

  12. Corrado C, Hulten C, Sichel D (2005) Measuring capital and technology: an expanded framework. In C Corrado, J Haltiwanger and D Sichel (eds) Measuring capital in the new economy, Studies in Income and Wealth, Vol. 65, University of Chicago Press: Chicago

  13. Corrado C, Hulten C, Sichel D (2006) Intangible capital and economic growth, NBER Working Paper No. 11948. National Bureau of Economic Research, Cambridge, Massachusetts

  14. de Rassenfosse G (2012) Intangible assets and productivity growth - an update. Melbourne Institute of Applied Economic and Social Research University of Melbourne, commissioned by the Department of Industry, Innovation, Science, Research and Tertiary Education (DIISRTE)

  15. Dickey D, Fuller W (1979) Distributions of the estimators for autoregressive time series with a unit root. J Am Stat Assoc 74:427–431

    Google Scholar 

  16. Diewert WE, Fox KJ (2016) Alternative user costs, rates of return and tfp growth rates for the us non_nancial corporate and noncorporate business sectors: 1960-2014. Vancouver School of Economics, University of British Columbia Discussion Paper 16-03

  17. Elnasri A, Fox KJ (2014) The contribution of research and innovation to productivity and economic growth. UNSW School of Economics Discussion Paper 2014-01

  18. Elnasri A (2013) Three essays on infrastructure investment: the Australian experience. PhD thesis, School of Economics, UNSW

  19. Engle RF, Granger CWJ (1987) Co-integration and error correction: representation, estimation, and testing. Econometrica 55(2):251–276

    Article  Google Scholar 

  20. Fukao K, Hamagata S, Miyagawa T, Tonogi K (2009) Intangible investment in Japan: measurement and contribution to economic growth. Rev Income Wealth 55:717–736

    Article  Google Scholar 

  21. Hao J, Manole V, van Ark B (2009) Intangible capital and growth - an international comparison. Economics Program Working Paper Series No. 08 - 14, The Conference Board, New York, December

  22. Haskel J, Wallis G (2010) Public support for innovation, intangible investment and productivity growth in the UK market sector. IZA, Discussion paper series No. 4772

  23. Haskel J, Wallis G (2013) Public support for innovation, intangible investment and productivity growth in the UK market sector, Econ Lett 119:195–198

  24. Hodrick R, Prescott E (1997) Postwar U.S. business cycles: an empirical investigation. J Money Credit Bank 29:1–16

    Article  Google Scholar 

  25. Industry Commission (1995) Research and development. Inquiry Report no. 44, AGPS, Canberra

  26. Jalava J, Aulin-Ahmavaara P, Alanen A (2007) Intangible capital in the Finnish business sector, 1975-2005. ETLA Discussion Paper. Helsinki, Finland. The Research Institute of the Finnish Economy. No. 1103

  27. Louca J (2003) Multifactor productivity and innovation in Australia and its states. In C Williams, M Draca and C Smith (eds) Productivity and regional economic performance in Australia. Office of Economic and Statistical Research, Queensland Treasury: Brisbane

  28. Madden G, Savage S (1998) Sources of Australian labour productivity change 1950-1994. Econ Rec 74(227):362–372

    Article  Google Scholar 

  29. Marrano MG, Haskel J (2006) How much does the UK invest in intangible assets? Department of Economics Working Paper No. 578, Queen Mary, University of London, London

  30. Marrano MG, Haskel J, Wallis G (2009) What happened to the knowledge economy? ICT, intangible investment, and Britain's productivity record revisited. Rev Income Wealth 55:686–716

    Article  Google Scholar 

  31. Matthews M, Howard J (2000) A study of government R&D expenditure by sector and technolog. Emerging Industries Occasional Paper 3, Department of Industry, Science and Resources, Canberra

  32. OECD (2001) Measuring capital: Oecd manual, measurement of capital stocks, consumption of fixed capital and capital services. OECD, Paris

    Google Scholar 

  33. Otto G, Voss G (1994) Public capital and private sector productivity. Econ Rec 70(209):121–132

    Article  Google Scholar 

  34. Parham D (2006) Empirical analysis of the effects of R&D on productivity: Implications for productivity measurement? OECD Workshop on Productivity Measurement and Analysis, Bern Switzerland

    Google Scholar 

  35. Romer P (1990) Endogenous technological change. J Polit Econ 98:S71–S102

    Article  Google Scholar 

  36. Shanks S, Barnes P (2008) Econometric modelling of infrastructure and Australia's productivity. Productivity Commission, Internal Research Memorandum, cat no. 08-01

  37. Shanks S, Zheng S (2006) Econometric modelling of R&D and Australia's productivity. Productivity Commission, Staff Working Paper

  38. Solow R (1957) Technical change and the aggregate production function. Rev Econ Stat 39(3):312–320

    Article  Google Scholar 

  39. Solow, R. (1987) We'd better watch out. New York Times Book Review, July 12, 36

  40. Tatom J (1993) The spurious effect of public capital formation on private sector productivity. Policy Stud J 21(2):391–395

    Article  Google Scholar 

  41. Topp V, Soames L, Parham D, Bloch H (2008) Productivity in the mining industry: Measurement and interpretation. Productivity Commission Staff Working Paper

  42. Triplett J (1999) The solow productivity paradox: what do computers do to productivity? Can J Econ 32(2):309–334

    Article  Google Scholar 

  43. van Rooijen-Horsten M, van den Bergen D, Tanriseven M (2008) Intangible capital in the Netherlands: A benchmark. Statistics Netherlands, Discussion Paper 08001

  44. Zelenyuk V (2014) Testing significance of contributions in growth accounting, with application to testing ict impact on labor productivity of developed countries. Int J Bus Econ 13(2):115–126

    Google Scholar 

Download references

Acknowledgements

We thank the Productivity Commission and Melbourne Institute for providing us with their data on intangible investment. Financial support from the Australian Research Council (LP0884095) is gratefully acknowledged, as are helpful comments from three anonymous referees, Paula Barnes, Erwin Diewert, Dean Parham, Joonghae Suh and participants at the 2014 KDI Journal of Economic Policy Conference.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Kevin J. Fox.

Ethics declarations

Conflict of interest

The authors declare that they have no competing interests.

Disclaimer

The views expressed in this paper are those of the authors. Any errors are our responsibility.

Additional information

This paper is a contribution to a series of projects undertaken by the Australian Council of the Learned Academies to examine ‘The Role of Science, Research and Technology in Lifting Australia’s Productivity’.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Elnasri, A., Fox, K.J. The contribution of research and innovation to productivity. J Prod Anal 47, 291–308 (2017). https://doi.org/10.1007/s11123-017-0503-9

Download citation

Keywords

  • Productivity
  • Innovation
  • Intangible assets
  • Public funding

JEL Classification

  • O3
  • O4
  • H4