Table of Contents
Go Back
Research and Development Resourcing When Faced with Fundamental Market Dynamics
Jonathan D. Moizer and Mike J. Towler
Faculty of Social Science and Business, University of Plymouth, Drake Circus, Plymouth
Devon, PL4 8AA.
Abstract
System dynamics has been used over a number of years to explore and explain the role R&D can
play in shaping the dynamics of a firm or industry. This article describes how a dynamic simulation
model can be built which broadly characterises and captures the causal feedback structure and
performance behaviour inherent to a generic R&D system within a firm. Alternative futures are played
out to explore the long-term consequences of a confluence of R&D resource decisions, coupled with
changes to market demand.
Keywords: system dynamics; R&D, research and development; resource allocation; R&D intensity
Introduction to System Dynamics Modelling of Research and Development Systems
Research and development (R&D) has been cited as critical to the survival and growth of technology
based firms. Faced with increased competition emanating from globalisation, the viability of many
established firms is based on the efficacy of their R&D activities. In some sectors R&D has continued
to be intensified, in others the intensity has remain fixed, and in yet others the increase in R&D
expenditure has been outstripped by an increase in sales (DTI, 2003). However, the R&D systems of
many firms are like other parts of a business model, characterised by complex structure and dynamic
behaviour. It is often difficult to predict or trace out the emergent consequences of sets of decisions
due to the presence of delays, non-linearities and multiple causal feedbacks inherent in such
managed systems. Often well intended decisions can lead to the counter-intuitive behaviour, and the
manifestation of unintended and undesirable outcomes.
System dynamics modelling can help the R&D policy-decision-maker to develop structural
representations of parts of the R&D system, and explore changes to internal policy and external
drivers from the safety of a computer-based simulation model. Roberts (1964) developed a model of
a full dynamic system underlying project lifecycles, identifying key issues relating to R&D success
within the characteristics of product, firm, customer and process. An edited collection (Roberts, 1978)
included not only models of R&D project dynamics but also of the connected pan-organisation and
resource allocation issues. Therein Weil ef. al. (1978) considered the example of “work-flow
bunching” in which several factors (including e.g. human resource allocation, sequential
dependencies of and feedback between projects) led to poorly distributed, oscillating workload within
the R&D organisation. More recently, Roy and Mohapatra (1994) have used empirical findings to
develop a model investigating the effect of workplace climate within R&D organisations, Kameoka and
Takayanagi (1997; 1999) have suggested that a company’s R&D intensity should be proportional to
the sum of revenue growth and a technology obsolescence rate and Hansen et al. (1999) have
discussed conserving flow rates along a linear development pipeline. Pardue et. al. (1999) have
modelled the relationship between R&D intensity and technology diffusion in IT producing industries.
They have examined the impact of short-term dynamic transients on longer term technology
trajectories. More specifically Repenning (2000; 2001) has identified the issue of product
development systems becoming trapped in a condition of poor performance due to resources not
being appropriately applied to early stage developments but instead to later stage projects. Milling
and Maier (see Milling 1996; 2002; Milling and Maier 1993; Maier 1998) have used system dynamic
models to explore the relationships between R&D activity, pricing strategy & product diffusion. They
demonstrate how new product development forces competitive responses in the market.
This article describes how a dynamic simulation model can be built which broadly characterises and
captures the causal feedback structure and performance behaviour inherent to a generic R&D system
within a firm. At present the model is populated with synthetic, but not unreasonable data. A range of
different R&D policy scenarios are played out over a twenty year simulated timeframe. These
alternative futures are diagnosed and compared so as to explain why some managed interventions
are more successful than others. The model is used as a tool to diagnose and predict the long-term
consequences of a confluence of R&D policy, coupled with changes to market demand.
In recent years, electronics firms, in particular have been faced with rising market demand along with
falling sales margins (Harvey, 2004). Many have chosen to focus on reducing their operating costs by
re-locating their operations to lower cost manufacturing bases. Can revenues be maintained and
even improved under such market conditions through a more optimal R&D function, or is it necessary
to continue to drive down manufacturing costs as a priority? In this article a number of scenarios are
explored where different R&D intensities and allocations of resource are examined and reported
upon. The implications for strategic direction are also considered.
The Basic Research and Development Pipeline
Figure 1 contains a diagram representing the flow of R&D projects along a pipeline. The manager
has on the face of it a simple decision on how to allocate resources between research
(conceptualising) and development (getting the work done). Some of the research fails and some
becomes obsolescent, thus never progressing to development. A proportion of successful
development is subsequently commercialised. In some firms this decision may be governed by
options thinking e.g. waiting for market uncertainty to be resolved; additionally there may be inflows
from acquisitions at various points along the pipeline. However, in this study such additional structure
is omitted so as to maintain focus on the core issues. Whilst this diagram below is a useful
representation of the project pipeline, it fails to capture the integrated feedbacks apparent in R&D
systems that facilitate and/or impede the flow of work. In particular, the re-investment of revenues
back into R&D activity is not indicated.
Resource Allocation
Failed Development
A
Successful Commercial
Research Development Obsolescence
R&D Backlog —<$<—<$| Commercial Stock -———>>
v
R&D Obsolescence
Figure 1: The R&D Pipeline
Research and Development Information Feedback
The integrated portrayal of systems structure may not be wholly familiar to many writers and
practitioners. Across a firm, a number of separate models are often employed that provide detailed
treatment of each domain of business activity but there is often a failure to close the feedback loops
linking these functions. A typical R&D manager may work within the boundary set in the pipeline
model above, without paying sufficient attention to sales and marketing, personnel, finance and
information management issues.
A feedback structure of R&D that may approximate several industries is illustrated in Figure 2. The
diagram contains four major causal feedback loops that link a number of R&D management activities
together. A number of important stocks (accumulations) are captured within this causal feedback
structure: R&D Backlog, Commercial Stock (technology stock), R&D Team and Cash.
R&D Budget is governed by both the revenues generated and the Fraction of Revenue Spent on R&D
(a.k.a. R&D Intensity — ratio of R&D expenditures to revenues). If R&D Budget were lowered over
time, for example, the R&D staffing would also be lowered (either through natural wastage and/or
redundancies and redeployment of staff); less research would be carried out; likewise there would be
a depleted flow of development; and consequently less technologies to commercialise; leading to
diminished revenues available for R&D investment. Reinforcing behaviour is evident in this feedback
loop.
Fraction of Available
Revenue Spent on ad Labour
R&D (R&D
Market Intensity) = R&D
Demand Budget + 4
\
Revenues Cash
[ -_ «)
Overheads “af + DEVELOPMENT
‘ ACTIVITY
Time -
Aro) Allocated to f
Commercial Research
Stock RESEARCH Time
ACTIVITY : Proportion of Allocated to
* R&D Time for Development
wz: Backlog Development
\_A
= te)
SUCCESSFUL
DEVELOPMENT FAILED
DEVELOPMENT
Successful Failed
Development Development #
+
‘- Development
Rate
Figure 2: Simplified causal loop diagram of the R&D Model (B and R label balancing and reinforcing
loops respectively).
There are four points of leverage indicated in this diagram where management change can be
employed: extent of market demand, R&D intensity, availability of labour and proportion of staff time
allocated between R&D activities. Market demand and availability of labour are assumed to be wholly
exogenous to a firm’s operations (although in reality a firm’s reputation can drive up demand for
transaction or employment), whereas R&D intensity and the amount of staff time allocated between
separate R&D activities are policies which are endogenous to the firm’s business model.
It seems evident that market demand is the most significant of these drivers, irrespective of whether
the firm operates a technology push or market pull strategy. A trade-off is required here between
supply and demand. Having too small or large a Commercial Stock (technology offering) will result in
under or over-supply. If revenue maximisation is to be achieved, then the size of the market demand
must be matched by the number of commercial offerings. This is far from a cake-walk, given that this
business model is characterised by multiple feedbacks and delays, and to add to the complexity, the
level of turbulence in the industry environment.
Developing a set of congruent and coherent strategies requires that these leverage points be
examined in unity. In order to align supply and demand, an integrated set of high-level decisions is
called for. This requires the sufficient development of a number of resources across the R&D function,
ie. the stocks resident within the model. This can be achieved through managing the internal drivers
of change, and pre-empting and responding to the external drivers. The R&D budget needs to be
supported; pool of potential labour has to be tapped in order to maintain a sufficient research team;
and the workloads of the R&D Team have to be balanced between conducting the research, and
facilitating its subsequent development.
The Research and Development Simulation Model
The R&D simulator has been developed and tested using the iThink™ (system dynamics) software. It
contains three sectors: R&D Pipeline, R&D Team, and Finance.
Research and Development Pipeline Sector
R&D projects in the firm progress along a pipeline (Figure 3). Herein the term projects is used rather
generally, in some industries the Commercial Stock may be measured in end products, core products
(Prahalad and Hamel, 1990), portfolios of patents efc., what is important is that it is possible to
characterise a unit of Commercial Stock in terms of average revenue earned per year per unit of
Commercial Stock. The R&D Backlog is increased by research activity and is depleted by successful
and failed development activity, and the atrophy of non-progressed research. Successful
Development increases the Commercial Stock (technologies available for commercial exploitation).
Commercial Stock Obsolescence depletes this stock. This is a proportion, and is based on the
commercial life of a technology or product in the industry.
The dwell time for projects in the R&D Backlog is governed by feedback processes. The R&D
Backlog per Team Member is used as a determinant of Work Effectiveness (proxy for team
productivity). Work Effectiveness then dictates the Development Rate. For simplicity the CLD (Figure
2) shows a positive link between the R&D Backlog and Development Rate; however the actual model
used has this as a bipolar relationship. A table function is used to represent the inverted U
relationship between work effectiveness and R&D Backlog; such a relationship between performance
and stress levels being well known (George and Jones*”, 1999). In this portion of the R&D system,
one is seeking to instrument policies that will align the flow of projects between R&D and
commercialisation. This can be achieved through up to two managed interventions (policies):
* change the size of the R&D Team (this stock is not enclosed within the R&D Pipeline Sector but
acts as a sector input), and/or
* — shift the allocation of staff time between research and development activities.
Work Effectiveness
Dev eiorment Sinones Normal Commenfial Obsolescence
Commerc
LS sxccsstuoeynment
on Comarca Stock Obsolescence
Prebabity of Reseach S285 rime caphg for Research RDO OS
8 Nomat RED Obsotsconsdy
Base Time per Research Bev slepmen\\Rate
for Dev alopmentiaee Time par Development
ierk Time Ceiling
Figure 3: Stock-flow diagram of the R&D pipeline sector
Research and Development Team Sector
The R&D work in the firm is carried out by the R&D Team (Figure 4). R&D employees progress along
a short pipeline. The R&D Team is increased by recruitment, and depleted by staff leaving the team.
If there is a shortfall between the desired and actual size of the R&D team, then the firm recruits from
the labour market. The number of R&D employees depends upon the proportion of revenues that the
firm chooses to re-invest in the R&D function (R&D intensity). The R&D sector of the system is
characterised by target-seeking behaviour. The staffing policy is predicated around the maxim, the
higher the R&D intensity, the more R&D staff required in the team i.e. in increasing R&D expenditure,
staff and all associated costs are increased proportionately; the method of working is not changed.
The ability to fund a growing R&D staff is dependent upon revenues, a rate that lies beyond this
sector but within the wider model.
“Target RAD as fraction of Revenues
Figure 4: Stock-flow diagram of the R&D team sector.
Finance Sector
Cash moves along a short pipeline (Figure 5). The firm’s stock of Cash is increased by incoming
revenues from sales, and is depleted by R&D and other operating expenses incurred. The algebraic
difference between revenues and expenses gives (instantaneous) profit. The R&D costs are
proportional to the size of the R&D Team. The market dynamics will dictate the supply-demand
relationship. Policies which can be controlled in part within the sector are the R&D employment costs
and the other operational costs incurred across the firm.
There are three drivers in this sector sitting outside the control or direct influence of the R&D policy-
maker which influence how Cash accumulates or depletes. These are the Market Demand for the
available technologies, Base Revenue per Commercial Project (perfect revenue achievable assuming
there is an infinite market demand) and Normalised Revenue per Commercialised Project (revenue
per sale based upon the discrepancy between market demand & size of the firm’s commercial
technology stock). Explicitly Revenues is the product of Commercial Stock, Normalised Revenue per
Commercialised Project! and Base Revenue per Commercial Project.
Commetial Stock
Market Demand
ea Base Revenue par Commercial Project
cost per Employee Non RD Cost per Commmereased Proect
Figure 5: Stock-flow diagram of the finance sector
Research and Development Policy Analysis
Policy analysis helps the model user to understand why a system behaves in a certain way (Coyle,
1996). Policy experiments with system dynamics models are used to help design the best possible
robust behaviour into the system under study. There are very many possibilities open to the model
user (policy-maker), and there is no way of knowing in advance which will give the best overall
performance in the system. The only way to progress is to experiment with different policies with the
intention of designing a scenario which suggests the best outcome, that is the control of any
undesirable behaviour within a system. The policy alternative must of course be implementable in the
real system.
One way that policy analysis can be conducted is through making modifications to the numerical
parameter values of the simulation model, so as to reflect different scenarios; then compare and
contrast behavioural trajectories and numerical outputs. These activities should help to find better
simulated results. This simplified R&D model contains a limited number of policy parameters, making
a coherent analysis relatively straightforward. The process of experimentation can render a clear
understanding of why the model behaves as it does. Multiple policy parameter changes will offer
more verisimilitude, and may facilitate the generation of deeper insights and understandings.
It is not enough to know that certain policies improve behaviour. The simulation model user needs to
know why that behaviour happens. Otherwise, they will not think about implementing the policy
decisions in the firm. For this to be achieved, the model user must at least understand the principal
feedback structure of the model (as outlined in Figure 2). Also, they must always be aware of the
limitations of the policy experimentation design. A mix of policies in a simulation may show a very
desirable outcome but it may be impossible to translate these changes into the real system.
Therefore, the policy parameters must be kept within clearly achievable bounds.
Recent Changes in the Consumer Electronics Market
The falling price of consumer electronics in the UK has buoyed consumer demand (Harvey, 2004).
Bigger manufacturers have ramped up their production to meet this demand and have benefited from
economies of scale but at the same time the retail cost is no longer reflecting the cost of consumer
electronics. Higher prices in the past allowed the R&D costs to be recuperated. Given this increased
intensity of market competition, can consumer electronics firms maintain sufficient revenue streams
and profit from improving R&D strategies alone or it is necessary to also continue to drive down
manufacturing costs?
The Research and Development Policy Designs and their Ability to Accommodate Radical
Market Change
Up to year three a stable relationship exists between market and firm in the simulation. The market
demand is static and predictable. This allows the firm to maintain a steady flow of R&D projects
through to commercialisation. The simulation model runs in a steady-state or equilibrium. At year
three, the market experiences an episodic shift. There is a simultaneous step increase in ‘Market
Demand’ (100%) and a step decrease in ‘Base Revenue per Commercial Project’ (25%) causing a
state of disequilibrium. The market now requires ‘more for less’.
The electronics firm has to decide how to respond to these new market conditions. There are a
number of policy alternatives that they could employ as a means of stabilising or even growing profit
in response to this market discontinuity. Given the complexity associated with aligning R&D,
commercialised products and market demand, the firm would benefit from looking at this problem in
its totality. Using a system dynamics simulation allows an exploration of longitudinal change in
performance and capabilities under different sets of policy combinations.
A range of alternative policy confluences are explored in the five scenarios outlined below. The
efficacy of these strategic decisions is mapped out over 20 years of simulated time. A range of
indicators of the firm’s performance and capabilities are monitored during each simulation run.
The policy analysis is designed to identify R&D policies which result in improved performance within
the firm as measured across a range of indicators. A number of key metrics are selected to allow
behavioural and numerical analysis of simulation performance. The final values and trajectories of
these are noted for each simulated scenario and used for comparison. The selected indices are
defined as follows:
Profit — The positive financial gain from the business operation after subtracting the R&D and
operating expenses.
R&D Team — Personnel responsible for the R&D activities.
Normalised Revenue per Commercialised Project (NRPCP) — Inverse S-shaped
relationship between the firm’s supply of commercial technology stock, and the market
demand for those technologies. If the ratio between these is high, then the NRPCP is low.
The higher the NRPCP, the higher the revenue accrued per commercial technology stock.
Work Effectiveness — The efficiency (or productivity) with which the average R&D member
of staff works. As discussed earlier, the relationship between work effectiveness and the R&D
Backlog per R&D Team is bi-polar. The optimum work effectiveness is evident where neither
work stretch (staff gold plating or boredom) or work compression (staff under intense stress)
is manifest.
In reaction to the market changes, the R&D policy maker could act in a number of ways. A range of
policy modifications can be explored and their ability to deal successfully with a discrete market
change measured and evaluated. Five broad policy scenarios are explored using the simulator:
* Scenario 1 — no change;
* Scenario 2 — greater R&D intensity;
* Scenarios 3a and 3b — balancing resources and work loads between the respective R&D
activities;
* Scenario 4 — lower R&D intensity.
These policy experiments have been set up as a means of capturing insights into the range of
behaviours that the R&D system is capable of, with the desire to design policies which may render
improved performance to technology firms operating under such business conditions.
Scenario 1: Business as Usual (Base Run)
The first scenario to be explored is titled ‘business as usual’. This consists of keeping all policy
decisions as fixed throughout the course of the simulation run, i.e. the firm does not modify its policy
decisions in the face of this discrete shift in market demand. This scenario is set as the base run for
the subsequent policy tests, and alternative scenarios are played out with a view to discovering if
more desirable behaviour can be exhibited by the simulation model.
Figure 6 shows that if no alternative action is taken in response to the new conditions, then profit
declines. This decline results from a diminished flow of R&D projects through to commercialisation.
Declining revenues have resulted in lower absolute investment in the R&D team. As this capability
reduces, then the flow of R&D projects is arrested. The NRPCP or ability of a given commercial
technology to provide a return increases as the gap between market demand and the commercialised
technologies extends. Unfortunately, as the output of the firm is diminishing so is its means of
covering its operational overheads. The work effectiveness of team members remains constant as
the volume of R&D work and size of the team declines together. Although the firm has not entered a
free-fall, it is moving towards a slow death.
T: Profit 2: RED Team 3 NRPCP 7 Work Effectiveness
H 64.50 Eg ieee
2: 113.50
B 1.00
ly
NRPCE stabilises at high equilibfium as firm is unable to
h 62.00
2 111.00
P 0.85
la: Less revenue irivested in the R&D team
on means less staff leavers are replaced
decline in profits
i 59.50 [No change 10 R&D tea odutiviy 36
B 108.50 I decline in R&D team i ortional to R&D Backléy
j__ 4
ls: 0.70 7 : : 3
0.00 5.00 10.00 15.00 20.00
Years
Figure 6: ‘Business as usual’ with no new response to market change
Scenario 2: Higher R&D Intensity
In the ‘business as usual’ scenario where no policy modifications are instituted, the performance of
the simulated firm slowly deteriorates, with nearly all outputs exhibiting undesirable behaviour. The
policy makers may see fit to improve the R&D system through boosting the amount of R&D activity.
The ‘higher R&D intensity’ scenario shows the effect of doubling the proportion of resources
committed to the R&D function in comparison with the base run, whilst keeping all other policies
invariant.
Figure 7 shows a greater deterioration in system performance than seen in the base run. Profit
stabilises then collapses into losses. This decline results from too rapid a growth in the flow of R&D
projects through to commercialisation. Over time, an over-supply of commercial technologies to the
market emerges and revenues consequently decline. Lower absolute investment in the R&D team
occurs, and as this capability reduces, the flow of R&D projects slowly diminishes. The NRPCP
collapses due to the market over supply. The work effectiveness of team members recovers as the
volume of R&D backlog per team member grows to a more optimal level. If the firm cannot turn
around the financial loss making then it may not survive over the 20 simulated years.
1: Profit 2: R&D Team 3: NRPCP 4: Work Effectiveness
150.00 mp ttt teeter gee eeeeeeeeeeeeeeeeeregererereeereereerrsreereerereerereernnns
400.00 A
100 commercial stock
es market demand
in the medium-term
but collgpse thereafter
h:
Lb:
RE] 0.73 :
li the funding of the team does |=?"
likewise
Ht: -150.00 R&D team prbductivity bounces backias
al 100.00 team growth Slows in relation to R&D:backlog
li 0.45 + + + ‘i
0.00 5.00 10.00 15.00 20.00]
Years
Figure 7: ‘Higher R&D intensity’ as a response to market change
Scenario 3a: More Research and Less Development
In the ‘higher R&D intensity’ scenario the performance of the firm greatly deteriorates across most
measures from the medium and into the long-term. An alternative scenario to heavily financing the
growth of the R&D function may be to make better use of existing resources through better allocation
of staff time between respective R&D activities. The ‘more research and less development’ scenario
shows the effect of modestly increasing the proportion of resources committed research at the
expense of development in an effort to boost the backlog of R&D ready for commercialisation. All
other policies remain fixed.
Figure 8 shows a general improvement in the system as compared to the base run. Profit grows
consistently and, after a period of time, exceeds the initial profit. This improvement results from a
steady growth in the flow of R&D projects through to commercialisation. Over time, the stock of
commercial technologies gradually approaches market demand and revenues remain strong but grow
at a lessening rate. As expected, the NRPCP declines slowly as the market is better satisfied. The
work effectiveness of team members approaches an optimal level (i.e. a good balance for the team
between trying to compress their work or stretching it out). The health of the firm looks to be secure
under these market conditions into the long-term.
T: Profit 2: R&D Team 4 Work Effectiveness
fi 138.80 JNRECP stabilises then Heclinés ‘dé commercial fechnology stock’slowiy'Biart to caich Up market |
B} 4,00 Jdemand H :
Ly
Sy R&D tedm productivity grows sté
& asymptotes as it approaches }
63.00
120,00 Doooeveeereee| eney + Profiteracovers et
0.85 & maintains growth
Steady expansign of the R&D
seeeeeeeees “ii ss stoam ge-mero reivenugs are-roinvooted +
soN=
3
aS
Sa
S23
Figure 8: ‘More research and less development’ as a response to market change
Scenario 3b: Even More Research and Even Less Development
In the ‘more research and less development’ scenario noticeable improvement to performance are
achieved. Can further improvement in system behaviour be attained through stepping up the
proportion of research further? The research over development effort is increased further in the ‘even
more research and even less development’ scenario. The other policies are held steady.
Figure 9 shows poorer system behaviour than the previous scenario runs. Profits do stabilise in the
medium-term but are not sustainable in the long-term, and thus collapse. A bottleneck develops in the
project pipeline as the flow of successful development project through to commercialisation declines,
resulting in R&D backlog building up. As the revenues diminish, so does the level of R&D investment,
and the size of the R&D team begins to fall. The commercial output of the firm diminishes and market
demand is satisfied even less resulting in the continued rise in NRPCP. Work effectiveness improves
with the growing R&D backlog but after a point, the backlog grows to such an extent that the R&D
team are put under undue stress to develop the backlog into commercial technologies, and
consequently productivity collapses.
1: Profit 2: R&D Team 3: NRPCP : Work Effectiveness
h: 70.00
a: 115.00
B:
li: 1.00
as prpfits decline so does
stated Ned oe to fund the tea
h: 55.00
a: 95.00} .......... Lf. Profigetabilises at.a spodest Galue inthe...
B: 0.85
li: 0.75
h: 40.00 [—____ i
b: 75.00 | | NRPCP improve’ 3 approaches near optimal position resulting
B: 0.70 | from the commercfal technology stock beirig in terminal decline
bi: 0.50 + + + i
0.00 5.00 10.00 15.00 20.00
Years
Figure 9: ‘Even more research and even less development’ as a response to market change
Scenario 4: Lower R&D Intensity with Modest Development
Can the policy makers offset the fall in base revenues by reducing the R&D intensity to a lower level
whilst maintaining a moderate amount of development over research allocation? This policy
confluence is tested in the ‘lower R&D intensity with modest development’ scenario.
Figure 10 shows a greater deterioration in system performance than seen in the base run. Profit
recovers temporarily but then declines into the long-term. A diminishing R&D team reduces the flow
of R&D projects through to commercialisation. Over time, an under-supply of commercial
technologies to the market emerges and revenues consequently decline. This results from a lower
relative and absolute investment in the R&D capability. The NRPCP improves due to the market
under-supply. Although work effectiveness improves over the medium into the longer term, by year
18, even this starts to decline as the R&D team depletes at a greater rate than the R&D stock is being
worked down. If the firm can not turn around the ever reducing profits, then its operations are not
sustainable in the long-term.
1: Profit 2: R&D Team 3: NRPCP 4: Work Effectiveness
H: 65.00 4 : =
2: 120.00 ¢
Al 1.00 i x
la :
IRPCP increases as th
R&D team productivity
hi: grows steadily towaids an
2: . optimal efficiency.but starts, ta dri
ia off late due to an in
4 of work & a lower team size
H: 35.00
2: 30.00
B | 0.50 4 R&D team size declings as revenues are not re-invested in replacing staff turnover --———__!
0.00 5.00 10.00 15.00 20.00
Years
Figure 10 ‘Lower R&D intensity with modest development’ as a response to market change
Summary of R&D Policy Options
The policy options or experiments set up for the different scenarios are summarised in Table 1. In
Scenario 1 all policy decisions are unchanged in the face of the market shift. The R&D intensity is
ratcheted up in Scenario 2 in an effort to satisfy the new market demand. In Scenarios 3a and 3b, an
effort is made to balance resources and work loads between research and development activities.
Finally, the effects of low levels of R&D intensity are evaluated in Scenario 5.
Target R&D
Fraction of Revenues
as Resource Fraction on Non-R&D Cost per
Development
Commercialised
Project (£M)
Scenario 1 0.05 (Medium) 0.7 (High) 20 (High)
Scenario 2 0.10 (High) 0.7 (High) 20 (High)
Scenario 3a 0.05 (Medium) 0.6 (Medium) 20 (High)
Scenario 3b 0.05 (Medium) 0.5 (Low) 20 (High)
Scenario 4 0.025 (Low) 0.6 (Medium) 20 (High)
Table 1: Summary of policy options selected for the scenario tests
Profit Profit per R&D NRPCP Work
(£M's) Team Member Effectiveness
(£M/Employ.)
Scenario 1 59.54 0.55 0.98 0.70
Scenario 2 -127.43 -0.51 0.47 0.77
Scenario 3a 128.14 1.94 0.95 0.96
Scenario 3b 44.35 0.56 0.99 0.50
Scenario 4 38.70 1.23 0.99 1.00
Table 2: Summary of business performance for selected scenarios (final values)
Profit Profit per R&D NRPCP Work
(£M's) Team Member Effectiveness
(£M/Employ.)
Scenario 1 4 6 3. 5.5
Scenario 2 r if 7 4
Scenario 3a 1 1 5.5 2.5
Scenario 3b 5 5 1.5 7
Scenario 4 6 2 1.5 1
Table 3: Ranking of business performance indices for selected scenarios
The business performance of the firm across the different scenarios is described in Table 2. Final
values for the end of the simulation for profit, profit per R&D team member, normalised revenue per
commercialised project (NRPCP) and work effectiveness are presented in the table. Table 3 ranks
performance for the chosen metrics. Scenario 1 does not render a particularly desirable outcome
(Figure 6). Scenario 2 provides the poorest all-round performance, with particularly dismal profit,
profit per team member and normalised revenue (Figure 7). As far as profit and profit per team
member is concerned, Scenario 3a indicates superior performance (Figure 8), while poorer results are
attributed to Scenario 3b (Figure 9). Scenario 4 provides the highest performance in terms of
normalised revenue and work effectiveness, a good profit returned per R&D employee but a poor
absolute profit (Figure 10).
Which Set of Policies to Elect for and Why?
There are two clear dichotomous alternative solutions to the market dynamics problem. Either ‘go for
growth’ employing the R&D policies indicated in Scenario 3a or ‘downsize’ the R&D activities as
demonstrated in Scenario 4.
There are a number advantages associated with growing and maintaining a strong R&D capability,
which include:
* retaining a skilled team as an insurance against having to source R&D staff in a tight labour
market;
* having the R&D resources available to respond to, influence or even create market demand;
* permitting the firm to engage in higher value upstream business activities; and
* allowing the firm to lobby government, quasi-government and trade associations more
powerfully.
There are also advantages coupled with the act of downsizing the R&D function, which include:
* lower R&D overheads;
* lower organisational complexity facilitating more transparent decision-making; and
* faster response to an actual or anticipated market change.
Which of the two desirable R&D strategies to adopt (i.e. grow or downsize R&D capability) will lie with
the key decision-makers within the firm. Principal consideration should be afforded to how congruent
the chosen R&D strategy is with the long-term strategic business objectives of the firm.
There are a number of limitations to scenarios played out in this insight model, based around the tight
model boundary or delimitations. The model structure is not able to reflect a firm’s ability to influence
the level of demand for their product through marketing, distribution & technological features, nor has
a change in operational costs been investigated. Work effectiveness has been included in the model
as a function of R&D Backlog per employee and this in turn affects development rate. However to
more fully capture the important human dynamics intrinsic to any socially constructed organisation, it
will be necessary to build knowledge, skills, attitudinal & behavioural factors into the model [Roy and
Mohapatra, 1994; Moizer and Towler, 2003].
Future Work
There are two clear and related possibilities that the investigated scenarios did not consider. Scenario
2 showed some initial advantage from increasing the R&D Intensity, but this eventually led to an
oversupply in the Commercial Stock, introducing more variables rather than parameters seems
reasonable, particularly allowing the R&D Intensity to change with time. Following on from this,
decision rules to continuously change the R&D Budget based upon cash, profit, revenue efc., could
be considered. For example, a change to the CLD (Figure 2) of including a negative link between
Overheads and R&D Budget introduces several additional balancing loops into the structure.
The current model has consisted of synthetic data within a specific industrial context. Moving the
model into other industry contexts will, as well as changing industrial parameters such as
obsolescence rates, require some structural changes. The most obvious of these are:
* the number of phases in the pipeline (i.e. the required level of disaggregation),
¢ learning and motivation co-flows allied to the R&D pipeline;
* additional inflows and outflows (specifically buying or selling technologies along the pipeline),
and
* separating product and process R&D (e.g. R&D to reduce Non-R&D Costs per Commercialised
Project)
References
DTI. 2003. The 2003 R&D scorecard. Available:
www.innovation.gov.uk/projects/rd_scoreboard/analysis/analysis.htm. [19th March 2004].
George, J.M. and Jones, G.R. 1999. Organizational Behaviour (8 ed.). Reading, MA: Addison-
Wesley.
Hansen, K.F., Weiss, M.A. and Sanman, K. 1999. Allocating R&D resources: A quantitative aid to
management insight. Research-Technology Management 42(4): 44-50.
Harvey, F. 2004. Low-cost goods rule out Britain. Financial Times 16" Jan: 3.
Kameoka, A. and Takayanagi, S. 1997. A corporate technology stock model — Determining total
R&D expenditure and effective investment patterns. Proceedings of the PICMET’97.
Kameoka, A. and Takayanagi, S. 1999. A corporate technology stock model financially sustainable
research and technology development. Proceedings of the PICMET’99.
Maier, F. 1998. New product diffusion models in innovation management — a system dynamics
perspective. System Dynamics Review 14(4): 285-308.
Marsh, P. 2004. GE taps a world of local skills. Financial Times 18" Feb: 15.
Milling P. 1996. Modeling innovation processes for decision support and management simulation.
System Dynamics Review 12(3): 211-234.
Milling P. 2002. Understanding and managing innovation processes. System Dynamics Review
18(1): 73-86.
Moizer, J.D. and Towler, M.J. 2003. Ideas for using system dynamics modelling for research and
development. Conference Handbook of OR 45: The OR Society Annual Conference.
Pardue, J.H., Clark T.D. and Winch G.W. 1999. Modelling short- and long-term dynamics in the
commercialization of technical advances in IT producing industries. System Dynamics Review 15(1)
97-105.
Prahalad, C.K. and Hamel G. 1990. The core competence of a corporation. Harvard Business
Review 68(3): 79-91.
Repenning, N.P. 2000. A dynamic model of resource allocation in multi-project research and
development systems. System Dynamics Review 16(3): 173-212.
Repenning N.P., Goncalves P. and Black L.J. 2001. Past the tipping point: The persistence of
firefighting in product development. Calif. Man. Rev. 43(4): 44-63.
Roberts, E.B. 1964. The Dynamics of Research and Development. New York: Harper and Row.
Roberts, E.B. 1978. Managerial Applications of System Dynamics. Cambridge, MA: The MIT Press.
Roy, S. and Mohapatra P.K.J. 1994. Study of work climate in R&D organizations: A system
dynamics approach. Proceedings of the 1994 International System Dynamics Conference.
Weil, H, Bergan, T and Roberts, E.B. 1978. The dynamics of R&D strategy. In Managerial
Applications of System Dynamics, Roberts, E. B. (ed.). Cambridge, MA : The MIT Press.
"It is assumed that it is possible to characterise each unit of Commercial Stock in terms of revenues
earned per year per unit of Commercial Stock:
Revenues = Commercial Stock x Base Revenue per Commercial Project x NRPCP
The NRPCP is described by a table function and is based upon the discrepancy between Market
Demand (allowing growth and decline to be included in the model) for the industry's technology and
the size of the firm’s Commercial Stock. Depending upon the detail of this S-shaped curve this may or
may not lead to Revenues monotonically increasing as the company increases its Commercial Stock
(Figure 11).
NRPCP-
Commercial Stock / Market Demand
Revenues
Commercial Stock / Market Demand
Figure 11: The relationship between NRPCP and Market Demand.
