Investments in R&D and Business Performance. Evidence From The Greek Market
Investments in R&D and Business Performance. Evidence From The Greek Market
Investments in R&D and Business Performance. Evidence From The Greek Market
ABSTRACT
The purpose of the present paper is to discuss the question: are investments in
R&D, innovation and new technologies, intangible factors of business performance? There
is a great controversy concerning the relationship between profitability and investments in
research and development. Many studies have failed to identify consistent positive returns
from R&D and IT investments, and the paradox has been termed as the “IT productivity
paradox”. Most of the optimistic researchers argue that the disappointing results are due to
mismeasurement errors, problematic design of the research and time lags between learning
and adjustment, because R&D investments can take several years to show results. In our
research, we apply a panel data analysis using data from industrial and computer
companies listed in the Athens Stock Exchange, for the period 1995-2000. We find that
although the R&D investments have a negative influence on profitability for the year of the
investment, they can show strong positive relation after two years.
Keywords: Performance, R&D, profitability, firm level data, Greek market, productivity
paradox, R&D strategy.
1
1. Introduction
confront competition, increase profitability and market share. R&D investment has an
essential role in these strategies, although it has distinguished characteristics from other
investments. Since more than half of the investments are associated with salaries of skilled
– expert workers and scientists, the degree of uncertainty associated with its output may
influence the investment rate over time (Hall, 2000). R&D investment generates profits
with a time lag (Aboody and Lev, 2001; Jefferson, 2006), and hence should be sustained at
capital) the last decades, has shown the important role of R&D on the performance of the
firms (Lev, 1999). R&D investment has been studied from several different perspectives.
Thorough research using a production function approach has been done by Griliches as
early as 1979. Verspagen and Los (2000) researched the R&D spillovers and productivity,
while Hall (2002) studied the financing of R&D. Aboody and Lev (2001), Ding, Stolowy
and Tenenhaus (2007), Jefferson , huamao, Xiaojing and Xiaoyun (2006), studied the time
lag of R&D on profitability and the contribution to the future earnings of the firm. In
addition to those studies, Lev and Sougiannis (1996) found a positive correlation between
The questions addressed in this paper aim to research the influence of R&D on
profitability, the time period for profit realization and the existence of decreasing returns.
2
We used data from the balance sheets of 36 industrial and computer firms listed in the
Athens Stock that report R&D stock, from the total 143, for the period 1995-2000.Using
panel data estimations, we found that R&D needs at least two years to positively affect
Section 3 describes the data, presents the methodology and discusses the empirical results.
2. Theoretical framework
innovative production processes, are used in order to provide differentiation that can yield
competitive advantage and lead time over rivals (Mansfield, 1968; Baily, 1972). Hence, the
firm invests in R&D and innovation to achieve market share and monopolistic profit.
Hall (2000) states, that more than 50% of the R&D spending is associated with salaries
and wages of highly educated scientists and engineers. Their efforts create an intangible
asset (know how), from which profits in future years will be generated. Low investment in
R&D reduces innovation and knowledge creation, which in turn reduces productivity as
R&D has generally been ignored, partly due to data availability problems. Sougiannis
(1994) notes that most of the results that show no significant relationship between R&D
and future benefits, may be due to sample sizes, research design, statistical techniques and
3
quality of the R&D data used. Studies in research intensive industries show that R&D
A business unit with higher productivity is generally more profitable. There are many
other factors that influence performance, that we should take into consideration in order to
The time lags are a major concern in the data and analysis. First of all a research may
take a few years to complete. After completion, it may take a couple of years to start
showing results (Griliches, 1979). This may be one of the most important reasons to check
variation over time and to research more than 5 years of firm level data. Some of the
Jefferson (2006) finds that the returns to industrial R&D appear to be at least three to
four times the returns to fixed production assets. There is a direct positive correlation
between R&D expenditures and economic growth. Lev & Sougiannis (1996) found that the
useful life of R&D varies from 5 to 9 years, while Aboody & Lev (2001) conclude that the
estimated duration of the benefits from R&D projects is seven years and most of the
operating income benefits are generated in 3 years from the R&D investment. Hence, we
Hypothesis 2. There is a time lag for the R&D stock to show results.
4
If we assume the production function exhibits the usual properties, then the R&D stock
as a production factor should exhibit decreasing marginal returns. Thus, we posit the 3rd
hypothesis.
We used data from the balance sheets of 36 industrial and computer firms listed in the
Athens Stock that report R&D stock, from the total 143, for the period 1995-2000. The
firms are the most important manufacturing and computer firms in Greece.
We define gross profit to sales ratio (GPSL) as a proxy for business performance. We
used the gross profit to sales ratio as the depended variable in our model, because it is more
Firms that report R&D investments in their balance sheets, include research and
software systems, brand development and other intangibles. Hall and Hayashi (1989), state
that R&D is an important intangible capital that can lead to more long-lasting and
supranormal returns; it is embodied in the firm and its employees and includes knowledge,
accumulated know-how, technical expertise, trade secrets, patents, etc. Knowing that, we
used the R&D stock to total assets, denoted as RDTA, as an explanatory variable in our
model.
We finally used the following control variables: first, the cost of goods sold to
inventories ratio denoted as CGSINV, as a proxy for the corporate management and second
5
the size of the firm proxied either by the logarithms of sales (SIZE) or by the logarithm of
Tables 1 and 2 present the descriptive statistics and the correlation matrix of the
variables used in our models. Data show a 28,4% average gross profit to sales ratio and an
average of 1,8% R&D stock to total assets. In accordance to our findings, Voulgaris,
Asteriou and Agiomirgianakis (2004), also calculated an average of 25% for gross profit
ratio for SMEs and an average of 28% for LSEs, in their sample from the manufacturing
sector.
Following the discussion of the previous section, the relationship between firms’
Where GPSL stands for profitability, RDTA with one and two lags stands for research and
development and SIZE, LFA and CGSINV as control variables. Finally, to test the 3rd
hypothesis for decreasing returns to R&D we included the squared RDTA with two lags.
Table 3 presents the panel estimations. In estimating panel data the unobserved
effect or individual heterogeneity is random and should be tested for random effect or fixed
(cov( xit , ci )=0, t=1,2,…T) is called random effect otherwise fixed effect. Hausman test
suggested, in our case, that the unobserved effect is correlated to the observed explanatory
variables and therefore the fixed effect method is more robust than random effects analysis
for the estimation of the parameters. The cost of this robustness is the exclusion of the
6
time-constant observables. However, since our data set does not include any time constant
the year following the year of the investment, since the novelty of the methods introduced
to production processes requires a learning and adjustment period. The strong positive
coefficient on the [RDTA] t − 2 term explains that there is a positive influence of R&D on the
profitability of the firm on the following 2 years. This means that, we have to wait for 2
years after the investment in order to have strong positive returns in GPSL. Hence, our first
and second hypotheses are valid and R&D is related to profitability and there is a time lag
Based on our findings, we suggest smooth and consistent investments in R&D. The
managers have to wait for 2 years in order to have positive results from R&D. From our
experience from the internet industry, internet companies need a period of two years from
new product development and R&D investments, in order to increase sales and profits.
Hence, companies with late reaction to the competition need around 2 years to react to the
new technologies, and another 2 years after implementation, concluding to a time span of
four years. Branch (1974) finds that there is a lag of 4 years between introducing an
innovation to practice and receiving a patent on it. That is why he used the patents received
The negative coefficient on the [RDTA] t2−2 term suggests that the third hypothesis
for the existence of decreasing marginal returns is supported by the data. Therefore, the
continuous increase of R&D share to total assets is not followed by equivalent increase in
7
profitability and even more the level of R&D investment has an upper limit after which
profitability decreases. In order to estimate the upper level of RDTA, we take the partial
derivative of GPSL with respect to [RDTA] t −2 from the (A2) model in Table 3. Hence:
∂GPSL
= 0 ⇒ b1 + 2b2 [ RDTA]t − 2 = 0 ⇒ [ RDTA]t − 2 = 0,157
∂[ RDTA]t − 2
coefficient of the squared [RDTA] t − 2 . Thus, the upper limit for R&D as share of total assets
is 15.7% and after that profitability decreases. Using the partial derivative of GPSL with
respect to [RDTA] t − 2 from the (A2) model, we also find that an increase on the share of
R&D to total assets by of 1% leads to an increase on GPSL by 1.69%, in the following two
years.
∂GPSL
= b1 + 2b2 [ RDTA]t − 2 = 1.975 − 2 ⋅ 6,269 ⋅ 0.023 = 1,69
∂[ RDTA]t −2
CGSINV show a negative influence on performance in all our models. This means
that lowering the stock the performance decreases, in contrast older findings that, it is large
inventories that create a drag on firm’s performance (Chhibber and Majumdar, 1999). This
may be due to the fact that further lowering inventories leads to operational and sales
problems.
Finally, the size has a negative influence on gross profit margin (in contrast to the theory).
This may be true for Greece as the major companies are old, former state owned
companies, and we find in literature that the age of the firm has a negative influence on
8
and profitability is expected, firms that grow at a rate faster than that which the
entrepreneur can manage may experience diseconomies of scale which reduce profitability
(Glancey, 1998).
for 36 industrial and computer companies listed in the Athens Stock that report R&D stock,
for the period 1995-2000. Our findings suggest that the effect of R&D investment on
profitability becomes positive after a period of two years with decreasing returns.
More specifically, the production costs tend to increase in the short run, because
new product development, new production methods and information technology, need time
to show results, since the novelty of the methods introduced into the production processes
creates turmoil during the adjustment period. That explains the negative relation of R&D to
profitability for the subsequent of the investment year. Finally, 2 years after the R&D
investment, the new methods and improvements are fully functional and absorbed and we
done using data from the income statements of the firms, as many of the R&D expenditures
are calculated in the income statements and not to the balance sheet. Only the 30% of the
companies in the Athens Stock Exchange calculate R&D expenditures in their balance
sheet. Matteucci and Sterlacchini (2005), also found that only 34% of their sample from
Italian manufacturing firms, do report R&D expenditures. We can also find in the literature
that most companies do not capitalize R&D, even when accounting standards allow them
9
the option. The difficulty on modeling such a research is that many of the R&D
expenditures are calculated in the income statements as production costs and not
specifically as an R&D figure. Furthermore, R&D and innovation is a value that many
times is not calculated in the financial statements. Lev (2003), comments that most
companies do not report how much they spend on employee training, on brand
intangible asset that has to do with entrepreneurship and the owner’s innovative ideas.
Many of the assets bought for production, involve high technology and R&D, but the extra
value is paid as a product and is not calculated in the balance sheet. The companies should
calculate this extra value and take it into account in the intangibles, with annual
depreciation, and not just calculate it as expenses in the income statement. Though, this
direction deals with the personality and education of the entrepreneur. Lev (2003)
comments that managers tend to manipulate and immediately expense R&D expenditures
in order to meet profit goals. The same problem arises when many innovative products and
R&D expenses are paid through operating leasing and hence they are also calculated as
10
References
Baily, M. (1972). Research and Development Costs and Returns: The US Pharmaceutical
Chhibber, P., Majumdar, S. (1999). Foreign Ownership and Profitability: Property Rights,
Control, and the Performance of Firms in Indian Industry. Journal of Law and Economics,
Vol XLII.
Grabowski, H.G., & Mueller, D.C. (1978). Industrial research and development, intangible
capital stocks and firm profit rates. The Bell Journal of Economics Vol9, No2, pp. 328-343.
productivity growth. The Bell journal of Economics, Vol. 10, No.1, pp 92-116.
Griliches. Z. (1994). Productivity, R&D, and the Data Constraint. The American Economic
Hall, B., H. (2002). The financing of research and development. Oxford Review of
11
Hall, B. & Reenen, V., J. (2000). How effective are fiscal incentives for R&D? A review of
Jefferson, G., Huamao, B., Xiaojing, G., Xiaoyun, Y. (2006). R&D Performance in
Chinese Industry. Economics of Innovation and New Technology, Vol 15(4/5), June/July,
pp. 345-366.
Lev, B. & Aboody, D. (2001). R&D productivity in the chemical industry. Working paper.
Lev, B., & Zarowin, P. (1999). The Boundaries of Financial Reporting and How to Extend
Los, B., & Verspagen, B. (2000). R&D spillovers and productivity: evidence from U.S.
Majumdar, S. K. (1997). The Impact of size and age on firm-level performance: some
Matteucci, N., Sterlacchini, A. (2005). ICT, R&D and Productivity Growth: Evidence from
Italian Manufacturing Firms. Working Paper. University delle Marche, Ancona, Italy.
12
Parcharidis, E. (2006). Investments in R&D and business performance. Working Paper.
Rogers, M. (2005). R&D and productivity in the U.K.: evidence from firm – level data in
Voulgaris, F., Asteriou, D., Argiomirgianakis, G. (2004). Size and Determinants of Capital
13
Appendix
GPSL 1
-0.629 1
[RDTA] t −1
-0.752 0.934 1
[RDTA] t − 2
14
Table 3: Panel estimations: Dependent variable GPSL
Model A1 Model A2 Model B1 Model B2
2 -6.269* -6.412**
[RDTA] t − 2
(-1.948) (-01.996)
CGSINV -0.102* -0.106* -0.129** -0.125**
(-1.688) (-1.720) (-2.097) (-2.022)
15