Skip to main content


Log in

R&D and productivity: using UK firm-level data to inform policy

  • Original Paper
  • Published:
Empirica Aims and scope Submit manuscript


The UK’s business R&D (BERD) to GDP ratio is low compared to other leading economies, and the ratio has declined over the 1990s. This paper uses data on 719 large UK firms to analyse the link between R&D and productivity during 1989–2000. The results indicate that UK returns to R&D are similar to returns in other leading economies and have been relatively stable over the 1990s. The analysis suggests that the low BERD to GDP ratio in the UK is unlikely to be due to direct financial or human capital constraints (as these imply finding relatively high rates of return).

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

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others


  1. In 2002, 18 OECD countries had some form of tax related R&D incentive for business firms. Of these, 11 had a tax credit and 7 had R&D tax allowances. Tax incentives can be targeted at the level of, or increase in, R&D; they can also be focused on large or small firms. Calculating the relative generosity of R&D tax incentives across countries is complicated since the overall corporate tax structure needs to be considered. OECD (2002) compute an index and find, for example, that Italy gives least support to large firms, but most to small firms.

  2. The 3.4 million is the estimated number in the European Union in 2000. An increase of 700,000 was forecast for the EU to meet the 3% R&D to GDP target (Cervantes 2003).

  3. The effectiveness of the tax credit has not yet been evaluated. Inland Revenue (2003) report that in 2001, 2002 there were 3,200 claims under the scheme covering a R&D spend of only £200 million (around 1.5% of total BERD). Small and medium enterprises (SME) receive 150% tax relief on R&D, while larger companies receive 125%. An SME was defined as below 250 employees, although the 2006 UK Budget announced an increased limit for R&D purposes to 500 employees.

  4. Many other possible reasons for the UK’s productivity performance have been suggested, including: lack of access to long term finance, limited supply of skilled and scientific human capital, poor general management, a corporate culture overly focussed on growth through acquisitions, lack of government support and poor university-business links. It is clear that some of these factors also affect R&D expenditure and, for that matter, R&D effectiveness.

  5. The share of UK BERD funded from overseas is relatively high (21% in 2000, as opposed to around 3% in Germany). Further, this share has increased from 15% in 1992 (data from ONS 2001). Whether the increasing share of foreign funded R&D has reduced or increased aggregate BERD to GDP ratio is not something that we analyse here.

  6. These points derive from the analysis in Becker and Hall (2004) and Becker and Pain (2003), although these papers analyse the level of BERD not the BERD/value added ratio.

  7. These are taken from a variety of sources, but summaries are given in HM Treasury (2003, 2004) and DTI (2003). Tylecote and Ramirez (2006) provide a detailed discussion on corporate governance and innovation; Centre of Business Research (2006) provide a full discussion of management issues, along with survey evidence of matched UK and US firms.

  8. The presence of A in Eq. 1 requires some explanation. In economic growth theory, A represents the level of knowledge or technology of the firm, which would include any contribution from in-house R&D. However, in the empirical R&D productivity literature some authors leave in the A term (e.g. Hall and Mairesse 1995), although they do not define it), while others omit it entirely (e.g. Bond et al. 2002). Leaving A in Eq. 1 makes it clear that there can be external, knowledge related, effects on productivity, perhaps due to spillovers.

  9. Olley and Pakes (1996) use investment as the additional variable, while Levinsohn and Petrin (2003) use intermediate inputs. In the empirics below we do have data to use an Olley-Pakes type estimator, but not for Levinsohn-Petrin. See Ackerberg and Caves (2004) and Bond and Soderbom (2005) for discussion and critiques of these methods.

  10. For example, Hall and Mairesse (1995) state, “By using input measures from the beginning of the year for which output is measured, we hope to minimise the effects of simultaneity between factor choice and output, but this could still be a problem” (p. 269).

  11. The basic result that measurement error can attenuate coefficients (cause them to be biased towards zero) is contained in, for example, Greene (1993) or Johnston and DiNardo (1997). In first difference or within deviation models the attenuation is worse if the explanatory variables are correlated over time (Griliches and Hausman 1986; Baltagi 1995). Griliches and Mairesse (1995) provide an insightful discussion of these issues in the case of production functions. In general, measurement error in the dependent variable (i.e. output) is not a concern, however, in the case of a production function there are potential problems. Klette and Griliches (1996) discuss the case where the use of aggregate deflators introduce a firm specific error that can be correlated with growth of inputs.

  12. The Bond et al. (2002) abstract states “we find that the R&D output elasticity is approximately the same in both countries [UK and Germany], implying a much larger rate of return on R&D in the UK than in Germany”. They do note that this result requires further testing (their footnote 21) and, specifically, direct estimates of rates of return (as done in this paper).

  13. This percentage is calculated by taking the sum of current price R&D in 1999 for balanced sample and dividing by ONS data. The ONS data for calculations in this paragraph come from ONS (2001), which is the Research and Development in UK Businesses, Business Monitor (MA14) publication. The ONS data are estimated from a (stratified) sample survey, which has complete coverage of the (believed) largest 380 R&D firms and then surveys a further 3,620 firms. Note that the ONS estimates R&D performed in the UK, while our data includes all R&D investment by firms (whether in the UK or overseas). This does mean the figures are not directly comparable, but are included here as rough indicators of magnitudes.

  14. These companies are Astra Zeneca PLC (£1.74 billion), Glaxo Welcome PLC (£1.27 billion), SmithKline Beecham PLC (£1.02 billion), BAe Systems PLC (£0.87 billion) and Unilever (£0.62 billion). Note that Astra Zeneca is not in balanced panel since its R&D data—in the database we use—only starts in 1992.

  15. A formal test the hypothesis that manufacturing and non-manufacturing firms can be pooled is rejected (F 164435  = 23.1) for the OLS estimator.

  16. We have tried estimating using longer differences (e.g. t and t − 3), which should lessen the effect of measurement error on capital. For example, the coefficient on capital in regression (9) increases to 0.17. The R&D coefficient falls to 0.14 in this regression but, of course, we are now estimating the effect of R&D in t − 4 on output growth over t to t − 3.

  17. Use of a dynamic panel model, and GMM methods, also yields volatile estimates on coefficients, depending on sample and instruments specified. This is also apparent in Bond et al. (2003). Given this, and the fact that very few existing studies use this methodology, the analysis here focuses on first difference estimators. Appendix 2 shows a set of GMM estimations.

  18. The coefficient on capital is approximately the same or slightly higher while, in general, the coefficient on labour is lower by up to 0.1 if we deduct imputed values contain in R&D. This pattern is also observed in the results from adjusted labour and capital values in Table 2.

  19. There is no breakdown in the data for R&D that is done for other companies. In addition, although we have reported results from making an adjustment for the double counting problem, this was by necessity a crude imputation. Since the double counting problem will be more severe in firms with high R&D intensities, this section acts as a further check on the sensitivity of our results.

  20. In all estimations a set of year dummies is also included in the regressions. Omitting the year dummies does create some significant coefficient on the R&D and year interaction terms, but this may be capturing aggregate productivity shocks, rather than changes in R&D productivity.

  21. There is some initial evidence that returns to R&D are higher for listed, manufacturing firms, but this appears to be driven by a few, high R&D intensity firms that are not listed.

  22. This is due to the fact that firms with R&D to value added ratios above 0.5 are excluded in the profitability analysis and, as can be seen from Fig. 3, these firms drive the significance of the R&D coefficient in the non-manufacturing sample.

  23. An analysis of the sub-sample of pharmaceutical firms within this sector finds a coefficient of 0.14 (significant at 5% level). However, formally testing the difference in coefficients between pharmaceutical and other firms in the science-based sector, using dummy variable interaction terms, is inconclusive.

  24. “These figures do not suggest a recent tightening of the labour market for SET [science, engineering and technology] workers, with demand rising relative to supply” (Marriott 2006, p. 113).

  25. Analysis of SME is difficult due to availability of data. Rogers (2006) analyses the rate of return to SME using ONS-BERD data for the period 1997–2003, finding some evidence of higher rates of return (around 40%), although estimates are volatile.


  • Ackerberg D, Caves K (2004) Structural identification of production functions. mimeo, UCLA Los Angeles

  • Ballot G, Fakhfakh F, Taymaz E (2002) Who benefits from training and R&D: the firm or the workers. ERC working paper in economics 02/01

  • Baltagi B (1995) Econometric analysis of panel data. Wiley, New York

    Google Scholar 

  • Becker B, Pain N (2003) What determines industrial R&D expenditure in the UK? NIESR discussion paper 211.

  • Becker B, Hall S (2004) Testing the pooling assumption in an industry panel of R&D investment: what do we learn? NIESR working paper 241.

  • Bond S, Soderbom M (2005) Adjustment costs and the identification of Cobb-Douglas production functions. Economics department working paper, Oxford 2005-W04

  • Bond S, Harhoff D, van Reenen J (2002) Corporate R&D and productivity in Germany and the UK.

  • Bond S, Harhoff D, van Reenen J (2003) Investment, R&D and financial constraints in Britain and Germany. Institute for fiscal studies working paper W99/5

  • Center for Business Research (2006) UK plc: just how innovative are we?

  • Cervantes M (2003) Scientists and engineers: crisis, what crisis? OECD Observer 240/1 (December, web journal)

  • Cohen W (1995) Empirical studies of innovative activity. In: Stoneman P (ed) Handbook of the economics of innovation and technological change. Oxford University Press, Oxford, pp 182–264

    Google Scholar 

  • Driffield N (2001) The impact of domestic productivity of inward investment in the UK. Manch Sch 69(1):103–119

    Article  Google Scholar 

  • DTI (2003) Competing in the global economy: the innovation challenge. Department of trade and industry Dec 2003

  • DTI (2005) R&D intensive businesses in the UK. DTI economics paper no. 11, Department of trade and industry

  • Eurostat (2007) Science technology and innovation in Europe. The European Commission

  • Extel Financial (1996) Company analysis. On-line database

  • Glen J, Lee K, Singh A (2001) Persistence of profitability and competition in emerging markets. Econ Lett 72:247–253

    Article  Google Scholar 

  • Goto A, Suzuki K (1989) R&D capital rate of return on R&D investment and spillover of R&D in Japanese manufacturing industries. Rev Econ Stat 71(4):555–564

    Article  Google Scholar 

  • Greene W (1993) Econometric analysis. Macmillan, New York

    Google Scholar 

  • Greenhalgh C, Longland M (2002) Running to stand still?—intellectual property and value added in innovating firms. mimeo Oxford intellectual property research centre

  • Greenhalgh C, Rogers M (2006) The value of innovation: the interaction of competition R&D and IP. Res Policy 35(4):562–580

    Article  Google Scholar 

  • Griffith R (2000) How important is business R&D for economic growth and should the government subsidise it? IFS Briefing Note 12

  • Griffith R, Harrison R (2003) Understanding the UK’s poor technological performance. IFS briefing note 37

  • Griffith R, Harrison R, Haskel J, Sako M (2003) The UK productivity gap and the importance of services. AIM briefing note, December 2003

  • Griffith R, Harrison R, van Reenen J (2004) How special is the special relationship? Using the impact of US R&D spillovers on UK firms as a test of technology sources. CEPR discussion paper 4698

  • Griliches Z, Hausman J (1986) Errors in variables in panel data. J Econom 31:93–118

    Article  Google Scholar 

  • Griliches Z, Mairesse J (1995) Production functions: the search for identification. NBER working paper 5067

  • Hall B, Mairesse J (1995) Exploring the relationship between R&D and productivity in French manufacturing firms. J Econom 65:263–293

    Article  Google Scholar 

  • Harhoff D (1998) R&D and productivity in German manufacturing firms. Econ Innov New Technol 6:22–49

    Article  Google Scholar 

  • HM Treasury (2003) Lambert review of business-university collaboration—final report. HMSO

  • HM Treasury (2004) Spending review: science and innovation investment framework 2004–2014

  • Inland Revenue (2003) R&D tax credits: responses to ‘defining innovation’ and government proposals. HMSO Inland Revenue (UK)

  • Johnston J, DiNardo J (1997) Econometric methods, 4th edn. McGraw-Hill, New York

    Google Scholar 

  • Klette T, Griliches Z (1996) The inconsistency of common scale estimators when output prices are unobserved and endogenous. J Appl Econom 11(4):343–361

    Article  Google Scholar 

  • Levinsohn J, Petrin A (2003) Estimating production functions using inputs to control for unobservables. Rev Econ Stud 70:317–341

    Article  Google Scholar 

  • Los B, Verspagen B (2000) R&D spillovers and productivity: evidence from U.S. manufacturing microdata. Empirical Economics 25(1):127–148

    Article  Google Scholar 

  • Mairesse J, Mohnen P (2003) Econometrie de l’innovation. In: Mustar P, Penan H (eds) Encyclopedie de l’innovation. Paris Economica (English reference to this at, pp 431–452

  • Marriott B (2006) Scientists engineers and technologists in Great Britain. Labour Market Trends, Office National Statistics April, pp 109–116

  • Mueller D, Cubbin J (1990) The dynamics of company profits: an international comparison. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Olley G, Pakes A (1996) The dynamics of productivity in the telecommunications equipment industry. Econometrica 64(6):1263–1297

    Article  Google Scholar 

  • OECD (2002) Tax incentives for research and development: trends and issues. Science, technology and innovation working paper, OECD, Paris

  • ONS (2001) Research and development in UK Businesses. Business Monitor MA14 2000. Issues by National Statistics.

  • Pavitt K (1984) Sectoral patterns of technical change. Res Policy 13:343–373

    Article  Google Scholar 

  • Productivity Commission (2007) Public support for science and innovation. Productivity Commission, Canberra

    Google Scholar 

  • Rogers M (2002) The influence of diversification and market structure on the R&D intensity of large Australian firms. Aust Econ Rev 35(2):155–173

    Article  Google Scholar 

  • Rogers M (2006) Estimating the impact of R&D on productivity using the BERD-ARD data. Available at

  • Schankerman M (1981) The effects of double-counting and expensing on the measured returns to R&D. Rev Econ Stat 63(3):454–458

    Article  Google Scholar 

  • Symeonidis G (1996) Innovation firm size and market structure: Schumpeterian hypotheses and some new themes. OECD economics department working paper 161

  • Tassey G (2007) Tax incentives for innovation: time to restructure the R&E tax credit. J Technol Transfer 32:605–615

    Article  Google Scholar 

  • Thomson PLC (2001) Company analysis. On-line data. Accessed in 2001

  • Tidd J, Bessant J, Pavitt K (2001) Managing innovation: integrating technological, market and organizational change. John Wiley, Chichester

    Google Scholar 

  • Tsai KH, Wang JC (2004) R&D productivity and the spillover effects of high-tech industry on the traditional manufacturing sector: the case of Taiwan. World Econ 27(10):1555–1570

    Article  Google Scholar 

  • Tylecote A, Ramirez P (2006) Corporate governance and innovation: the UK compared with the US and ‘insider’ economies. Res Policy 35(1):160–180

    Article  Google Scholar 

  • Van Reenen J (1997) Why has Britain had slower R&D growth? Res Policy 26(4–5):493–507

    Article  Google Scholar 

  • Verspagen B (1995) R&D and productivity: a broad cross-section cross-country look. J Prod Anal 6:119–135

    Article  Google Scholar 

  • Wakelin K (2001) Productivity growth and R&D expenditure in UK manufacturing firms. Res Policy 30:1079–1090

    Article  Google Scholar 

  • Waring G (1996) Industry differences in the persistence of firm-specific returns. Am Econ Rev 86(5):1253–1265

    Google Scholar 

Download references


The author thank Bob Allen, Dirk Czarnitzki, Christine Greenhalgh, Marcel Fafchamps, Katrin Hussinger and others at an Oxford workshop (April 2005), the Global Conference on Business and Economics (Oxford, June 2005) and European Association of Research in Industrial Economics (EARIE) (Porto, September 2005), as well as an editor for helpful comments. Part of the database developed for analysis in this paper was developed at the Oxford Intellectual Property Research Centre (OIPRC), St Peter’s College. The author is grateful to St Peter’s College for accommodation and administrative support for this research.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Mark Rogers.


Appendix 1: first difference specification

Starting with

$$ \Updelta y_{it} = \Updelta \beta_{t} + \Updelta a + \alpha_{1} \Updelta n_{it} + \alpha_{2} \Updelta k_{it} + \alpha_{3} \Updelta r_{it} + \Updelta u_{it} , $$

where r is the log of R&D stock, hence (omitting the i index)

$$ \Updelta r = r_{t} - r_{t - 1} = \ln \left[ {{\frac{{{\text{RD}}_{t} + (1 - \delta ){\text{RS}}_{t - 1} }}{{{\text{RS}}_{t - 1} }}}} \right] = \ln \left[ {{\frac{{{\text{RD}}_{t} }}{{{\text{RS}}_{t - 1} }}} + (1 - \delta )} \right] \approx {\frac{{{\text{RD}}_{t} }}{{{\text{RS}}_{t - 1} }}} $$

where RD it is R&D expenditure, RS it is R&D stock and δ is the rate of depreciation of R&D. Under the assumption that δ and RD it /RSit  1 are close to zero, the Δr it term is approximately RD it /RSit  1. The parameter α3 is the elasticity of R&D (i.e. [dY/dRS] × [RS/Y]), hence Eq. 4 can therefore be re-written as

$$ \Updelta y_{it} = \Updelta \beta_{t} + \Updelta a + \alpha_{1} \Updelta n_{it} + \alpha_{2} \Updelta k_{it} + \alpha_{4} {\frac{{{\text{RD}}_{it} }}{{Y_{it} }}} + \Updelta u_{it} $$

where α3 is now the gross marginal rate of return to R&D.


2.1 General

The financial data include in the OIPRC database contain sales, profits, wages and salaries, assets, standard industrial classification (SIC), and also R&D expenditure. A major constraint on the size of the sample is the availability of R&D data in financial accounts. The accounting conventions covering disclosure of R&D changed substantially in January 1989 with the introduction of SSAP 13 (which effectively made large and medium sized firms report R&D in their accounts). Given these constraints, and the economic issues of interest, the analysis is confined to the period between 1989 and 2000. A firm is required to have at least 3 years of data to be included in the analysis (this removes 189 observations (4.3%) of the sample). A balanced panel of 86 firms is also constructed for the period 1990–1999.

2.2 Value added, labour and capital

The primary measure of output in the analysis is value added. Value added is defined as total staff costs plus depreciation plus profit before tax. We have calculated and analysed a measure of value added without depreciation and this is highly correlated to our measure (correlation coefficient 0.99). Also, some studies correct for interest paid and impute a value for the user cost of capital. The data available here do not contain interest paid. Some analysis was undertaken using imputed interest payments (based on debt and market rates of interest) and user cost of capital, but the value added measure obtained was similar to the measure used here.

Since the log of value added is always used in the analysis this effectively removes any negative values (this removes 7.8% or 347 observations).

The measure of capital is tangible fixed assets as entered in the balance sheet. The measure of labour input is the number of employees. Although the database contains variables for part-time and full-time employees, the coverage for these is minimal, so we are unable to adjust the labour measure to reflect differences across firms in part- and full-time working. All financial data are deflated with using the best available series.

2.3 R&D

The R&D data is as entered in company accounts. Since there is no published R&D deflators for the UK, a GDP deflator is used where needed (note that the ratio of R&D/value added allows nominal values).

Since the accounting data provides no breakdown for the nature of R&D we base adjustment for ‘double counting’ issue on ONS (2001). This indicates that 10% of R&D costs are capital, 40% are labour costs and 50% are materials. ONS data also indicate that of R&D employees 15% are scientists, 21% are technicians and 64% are administrators. Using ONS Annual Survey of Hours and Earnings (2000) we can find median wages for each of these categories. Assuming that the firm also pays national insurance, pension and related costs equally 37% of wages, we can impute a value for the median R&D worker (£23,868 in 2000 prices). Adjusting this using GDP deflator we can calculate an imputed value for R&D employees in each year, and then subtract this from employment (n). The 10% capital costs were simply deducted from capital stock since we have no information on depreciation rates. In our view this provides a crude adjustment for ‘double counting’. Some sensitivity tests for the median R&D worker salary were carried out and these had little effect on estimates.

2.4 Price deflators

The ONS statistics produces industry-level data only for manufacturing industries and only from 1991 onwards. For the 1980s there are no industry-level deflators available. The value added measured is deflated by the overall manufacturing deflator (gross output measure) for 1988–1990 and then industry-level deflators are used (which are at the 2-digit SIC level). These data are on ONS web-site and are listed under producer prices (MM22). For non-manufacturing the GDP deflator is used (ONS, YBFY). All deflators are in constant 2000 prices.

2.5 Outliers

Plots of the first difference of the dependent variable (log of value added) against the first difference of the log of assets and employment variables, as well as the ratio of R&D to value added (for period t − 1), indicate some outliers. The regression samples exclude high growth firms. Any firms with year-on-year growth in value added of greater than 300% and less than −90%. Also, firms with growth rates of labour, capital or R&D of greater than 200% and less than −50% are excluded. This follows Hall and Mairesse (1995) criteria and leads to 220 firms being omitted (4.4% of the sample). The main effect of omitting these firms is that the coefficients in the instrumental variable estimates become more significant and closer to theoretical predictions. Many papers do not discuss such data issues explicitly, although they often have samples of, say, just large firms, which may well remove the smaller firms with extreme R&D values. As noted, Hall and Mairesse (1995) is an exception, as is Los and Verspagen (2000) who omit any firm with a year-on-year sales growth of greater than 80% (in any year of sample).

2.6 Panels

The analysis below makes use of both an unbalanced and balanced sample, with both being defined for manufacturing and non-manufacturing firms. The Table 6 shows summary statistics for these different samples. Note that in both cases these summary statistics are based on samples that include firms with 3 or more years of data over the sample periods. The manufacturing unbalanced sample has 3,163 observations, which come from 468 unique firms. The median sales revenue is £182 million (the mean is £1.3 billion). There are fewer non-manufacturing firms in the data yielding only 1,288 observations (228 firms) (the median firm has sales of £190 million; mean of £1.1 billion). The non-manufacturing sample contains a wide range of industries. However, the major industries represented are ‘business services’ with 27% of observations (this includes software and data service firms); the wholesale trade industry (19%); electricity, gas and water utilities (14%); and, oil and gas extraction (8%).

Table 6 Summary statistics for R&D

Appendix 2

See Tables 7 and 8.

Table 7 GMM, dynamic panel estimations
Table 8 Olley-Pakes estimations

The above estimates are based on Olley and Pakes (1996). This asserts that η it equals a function of investment and capital h(I, K), so that Eq. 2 can be written as

$$ y_{it} = \beta_{t} + a + \alpha_{1} n_{it} + \alpha_{3} r_{it} + \alpha_{2} k_{it} + h(I_{it} ,K_{it} ) + \varepsilon_{it} . $$

Using a third order polynomial in investment and capital to proxy for \( \alpha_{2} k_{it} + h(I_{it} ,K_{it} ) \)we can estimate the above and obtain a consistent estimate for α1 and α3 (the spillover term is also added). These are shown in the table, along with results from tests on year and industry dummies that are also included. The third order polynomial is denoted by the function ϕ. In the second step we estimate

$$ V_{it} = \alpha_{2} k_{it} + g(\phi_{t - 1} - \alpha_{2} k_{it - 1} ) + \nu_{it} $$

where \( V_{it} = y_{it}^{{}} - \hat{\alpha }_{1} l_{it} - \hat{\alpha }_{3} r_{it} - ({\text{spillover and dummies)}} \)terms in order to get a consistent estimate of α2. This is the estimate shown in the table.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Rogers, M. R&D and productivity: using UK firm-level data to inform policy. Empirica 37, 329–359 (2010).

Download citation

  • Published:

  • Issue Date:

  • DOI:


JEL Classification