To Main Proceedings Document
Modelling Product Development Productivity with System
Dynamics
Petri Helo, Olli-Pekka Hilmola & Ari Maunuksela
Department of Information Technology and Production Economics, University of Vaasa,
Vaasa, Finland P.O. Box 700, FIN-65101 Vaasa, Finland
Tel. +358-6-3248671
phelo@ uwasa.fi
Abstract
This paper analyses the performance of product development as an input-delayed partial
productivity. Firstly, the total productivity concept is introduced in the framework of firm-
level performance measurement. Some recent literature related to new product development
is reviewed and three related system dynamics models are reviewed. Based on the analysis we
suggest that productivity measurement of product development requires knowledge in project
and resource structure. Finally, the preliminary results related to product development are
concluded and the practical applicability of the productivity framework is discussed for
further research in the area.
1 Introduction
The financial performance of a contemporary company depends much on its ability to
introduce new products and apply new technologies. Product life cycles of many products
have decreased in many industries, which forces companies to seek more efficient ways to
make applications for new and existing markets. For instance, electronics, telecommunications
and software industries are highly depending on product development performance.
Performance of product development of a company is a widely discussed topic in management
of research and development. Traditionally, proposed measures for product development are
often either lists of success factors, found out in qualitative analyses. R&D performance
measures have been used at more general level; for example, in econometric models, which
elaborate the relationship between investment and gained revenue. Direct linkages to
company-level financial performance has not been studied in great extent in terms of the
relationship between R&D and new product development.
Productivity measurement is a well-known approach for assessing the financial performance of
a company. This approach origins from Davis (1955) who proposed a model that could be
used to analyse the performance of a company by using a measure for total productivity. He
defined the total productivity as ratio between total tangible outputs and total tangible inputs
used to make the outputs. The traditional definition of productivity has a variety of different
interpretations. Productivity has often been associated with productivity of labour, but
according to total productivity models, labour is not the only one input parameter. Other input
items are most often tied with profit and loss calculation and/or cost accounting principles (see
Davis 1955). Traditionally, the total productivity models tend not to deal the total output as
units sold, but the units produced - productivity is about converting efficiently inputs to
outputs (Craig & Harris 1973: 16; Sumanth 1979: 6.5). Craig et al. (1973) completed one of
the first implementations of total productivity measurement at the firm level. Also Sumanth
(1979) described same kind total productivity formula as Craig et al. (1973), but excluded
from Q energy expenses. The general expression of the productivity can be stated as follows:
Total Tangible Output
Total Tangible Input
0; +09 +03 +04 +05
H+M +FC +WC +E +X
Total Productivity =
Total Productivity =
where O, is output, H human input, M materials input, FC fixed capital input, WC working
capital input, E energy input, X other expenses input. As shown in the formula above, the total
productivity can be divided into partial components, which have direct link to operations.
Productivity measurement been successfully used to analyse partial productivities such as
material productivity, human productivity, energy productivity etc. (e.g. Craig & Harris 1973,
Sumanth 1997). However, productivity of product development has not been used. This is
probably due to two reasons: firstly, this information has not been available for financial
statements until recently; and secondly, there is a long delay between cost accumulation and
the revenue.
2 Dynamics of R&D and productivity
Many performance measures has been proposed for assessing R&D productivity. Wemer and
Souder (1997a) made an extensive literature review in R&D performance measurement. By
comparing the measurement approaches they identified three main categories of frameworks,
namely: (1) quantitative-objective, (2) quantitative-subjective, and (3) qualitative metrics.
According to them (1997a), the following factors should be taken into account when selecting
an appropriate measure set for a company: user needs, the data available, amount of efforts
needed. Wemer and Souder (1997a) concluded that integrated measures, which combine
several types of measures, are the most effective. On the other hand, they are also the most
complex and expensive to use too. Later Werner and Souder (1997b) reported that there are
big cultural differences in use of R&D measures. Despite the similar general management
practices, companies in Germany and US used very different frameworks.
Productivity measurement is a quantitative-objective method, regarded in terms of its
relevance for the internal control of a company (Nachum 1999). According to Wemer and
Souder (1997a) productivity related metrics reported frequently in the literature include the
following types: (1) effectiveness index, which is defined as a ration between revenue
generated from products introduced last years and the total R&D costs; (2) innovation index,
which is the percentage of revenues generated from new products; (3) quality indexes; and (4)
various patent indexes Werner and Souder (1997a).
As shown in Wemer and Souder's review, companies use several measures simultaneously.
The reason behind this may be that the current productivity of the firm is depending from the
decisions and investments made in the past. Especially in case of development investment, the
time delay between inputs and the outcomes is quite long. The methodology of system
dynamics allows using time delays in modelling. The lag between investments and R&D
revenues can be described with delayed stocks. Figure 2 illustrates a simple model of a
company investing on R&D. In this model, the total productivity of a firm consists of revenue
inflow, expenses outflow and profit, the stock between these flows. The revenues is depending
from the revenues.
profit(t) = profit(t - dt) + (revenues - expenses - RnD_investment) * dt
INIT profit = 20
revenues = 9+RnD_outcome*3
expenses = 10
Certain amount of current profit is invested into research and development periodically. This
capital investment accumulates the delayed stock, R&D in process.
RnD_ investment = profit*0.20
RnD_in_process(t) = RnD_in_process(t- dt) + (RnD_investment - RnD_outcome) * dt
INIT RnD_in_process = 0.05 - TRANSIT TIME = 3
In practical terms, delay means the time between starting the project to selling the new
product. Some percentage of R&D outcome will affect sales efficiently. R&D productivity is a
ratio between total tangible outputs (revenues) and R&D outcome coming from the delayed
stock. Otherwise the measure is considered to be similar to other partial productivities.
total productivity
operating margin
profit
all expenses
50 in process
RnD investment Rnb outcome
RnD costshare
Figure 2. A general infrastructure of delayed effect of R&D investment.
Running this hypothetical model shows how the revenues are depending from the R&D
investments. Because certain amount of value-added is allocated to R&D the future
investments are depending on current performance (Figure 3). The time delay presented in
variable "RnD in Process" determines how the fast the investment can be utilised in sales.
Managing this time-to-market factor is very critical. In the given example the performance of
the product development is constant (revenues = 9+RnD_outcome*3). In practice this
economic efficiency is what most of the productivity measures related to product development
are trying to capture. The business impact of new product is the main output, whereas the
R&D costs are the input.
@ s revenues 2:allexpenses 3: total productivity 4:RaD investment
J 40.005 ---time'to market")
i | a
4 40,00!
causes later sinking “}""V""S \
Decreasing R&D
total productivity
0.00 375 7.50 1.25 15.00
Figure 3. A hypothetical example - the revenues are depending from R&D investments.
3 Resource allocation - an approach on project level
According to Nachum (1999) introducing a time delay between outputs and inputs, as
proposed in the model, may provide only partial solution to problem. The key issue in addition
to pure economic performance and time factor is how the resource allocation is done in
project level. Hansen, Weiss and Kwak (1999) demonstrated the allocation problem of R&D
resources by illustrating and quantifying the situation with system dynamics. According to this
model, product development projects can be in three different stages during their way from
concepts to ready products. These stages present the micro-mechanism of "RnD in Process"
in the last model. In the first stage (Phase 1), a number of ideas are generated a function of
resources. Some of the projects die in initial stages (exit_p1), but some continue to next stage
(Phase 2). There is a time lag between phase shifts (Phase 1 Inventory). The flows moving
projects from stage to another are functions depending from resources. In every phase shift,
the resource requirements, project dieing rate and success rates vary. The general rule is that
the better commercial readiness, the more expensive is moving to next stage. On the other
hand, initial concepts are cancelled more frequently than ready products. Figure 4 illustrates a
Stella replication of the model structure.
R&D budget
allochtion
Phasel Phase} inventory Phase 2 PRage2 invento! Phase 3
dl 2 21 ‘ommercial readiness
xit pI it p2 shod
Figure 4. R&D allocation, reconstructed from Hansen, Weiss and Kwak (1999).
4 Conclusions
Delayed processes may present problems for traditional productivity measurement for an
organisation, which has a great proportion of newly introduced products. There are some
references in academic R&D studies that product development productivity as a ratio between
sales and R&D expenses is a driver for business success. For instance, Loch et al. [9]
concluded their study in electronics industry "the clearest result of this analysis is that
development productivity, measured by expense intensity, stands out as a critical driver of
business success ...". The longer time to market parameter is, the more difficult is to
approximate the actual productivity of product development. Another problem with product
development productivity may emerge with different kinds of projects. There are several types
of product development projects. These differ in terms of cost requirements, market impact
and time to market. Some projects are technology related while some are related to
customisation of existing products. We have identified from the literature four types of
projects, these are:
¢ Advanced research, which is a highly cost intensive project and the direct applications or
revenues are unlikely to emerge in near future. Research of a new technology is an
example of this kind of project.
¢ Advanced development projects are aiming to develop new products, which differ from the
old ones in terms of technology selection or design. The cost intensity in this category is
high, and the payback time relatively long.
¢ Platform development projects are those projects aiming to renew product design in large
extent or to build new product families. Compared to advanced research or development
projects, platform projects should have shorter payback time.
¢ Derivative development are projects aiming to customise existing designs for customer
requirements. Cost effect is typically rather low compared to gained revenues. Also time to
market in this kind of projects is very high.
Product development organisations need to launch each types of projects time to time. In short
period, derivative projects may be most productive and value-adding for customers. On the
other hand, if the product families are not renewed frequently or the company is not paying
attention to new technologies, the overall performance will collapse. The right mix of projects
in every stage is a challenging task.
One basic premise is that productivity measurement is based on periodical assumptions. Often
we can find examples of productivity trends reported on annual or biannual basis. Firm level
accounting practices are also integrated by this cyclical and periodical nature of evaluation. It
is not really an issue to debate that whether the issue of product development productivity is
really comparable with more generic business dynamics like annual business performance,
demand and time-to-market challenges. Our point is that Product development productivity is
in a way a partial productivity factor, analysed against the firm level total productivity. Like
the role of manufacturing performance is a part of the firm’s overall business performance.
Usually such partial productivities may have been analysed through labour, materials, energy
and capital inputs. Our discussion has here referred to some examples where product
development productivity is really deeply associated to the level of resource allocation in
product development projects, and the progressive nature of product development costs in
general. This is the underlying lesson we may have also learned in this process. There are some
specific kinds of cost dynamics embedded in product development. But the difficulty is that,
like in other forms of product development performance evaluations, the nature of periodical
bases or cycles necessary for productivity measurement are easily lost in the overall cost
accumulation of R&D and product development work.
Our solution to this problem is that we can use the concept of product development portfolio
as a kind of integrated proxy for the characterization of the inputs in product development. It
is also recommendable that such kind of product development portfolio models can be linked
to firm level product development processes (Cooper, Edgett and Kleinschmidt 1997). We can
illustrate this thinking with a hypothetical example. A firm carries out different kind of R&D
and product development activities. A development project portfolio of that firm may
simultaneously include different kinds of projects: advanced research, advanced development,
platform development and derivative projects (see e.g. Wheelwright and Clark 1992).
Development projects are future oriented, so it would be natural to expect that firm has
demands for various kinds of utilities to be gained through these projects. And the
expectations for these development results (utilities) are also driven by strategies for time-to-
market etc. In average, we could draw a development portfolio model by analysing the
different kinds of projects being pursued, see table below.
Table 1. Development project category types
Strategic goal for Project Ratio of Ratio of Project category
Time-to-market Unit cost R&D value NPD value
36 months 500000 FIM 70% 30% Advanced research
30 months 5000000 FIM 60% 40 % Advanced development
48 months 20 000 000 FIM 40% 60% Platform development
15 months 2000000 FIM 10% 90 % Derivative development
This previous table illustrates how product development costs may be allocated to different
kinds of projects that are being conducted under different time frames. This kind of data is
usually only available for firm’s internal operations. External evaluation of development
productivity is to be done with development intensity measures. But we can find some
possibilities to link these two different worlds together. Analysing the development intensity as
a measure of average spending expense put into development work within a given time frame
can do this. See example below:
a Business Impact
Development _ productivity = ;
evelopment _ productivity =" C(p AdC() +PAC(t) +DaC(d
where t = average development cost within a specified evaluation period, ArC(t) =Advanced
research project unit costs (expense flow/t), AdC(t) = Advanced development project unit
costs (expense flow/t), PdC(t) = Platform development project unit costs (expense flow/t),
DdC(t) = Derivative development project unit costs (expense flow/t). Moreover, we need also
some generic time variables in order to model this kind of things with system dynamics
principles. We decided to adopt a time-to-market based time frame for a base variable in our
model. This can be illustrated with a sort of characteristic curve where various form of
project-based unit costs (integrated into overall expense flows) can be modelled against to
strategic goals for time to market. (See figure 5)
Expected
Value of
Progress with the
Development
project portfolio
> Time
Time available for value creation in R&D Strategic
And product development activities time-to-market
Figure 5. Accumulated expectation for business impacts from the development inputs.
Our philosophy with this kind of model is that it may be possible to collect field data from
different firms for the description of their development portfolio strategies used in R&D and
product development. Similarly, by basing this model on the firm level analysis of total
productivity it may be possible for us to analyse product development productivity with firm
level data. Such kind of presentation of the R&D strategy may also be used in the
communication of various R&D goals for the top management of the firm.
Based on the system dynamics model presented, we can conclude that productivity of product
development is a highly dynamic measure. For this reason we would propose that in addition
to product development productivity managers should use predictive measures as well. The
purpose of these measures would be supporting the resource allocation and investment
justification. These measures could include for instance the following ones:
¢ New product intensity = Sales of those products introduced in last three years / Total
sales [%]. This measure tells how critical the product development function is for the
company.
¢ Average time to market is a time delay parameter that determines requirement for cash
flow.
¢ Proportion of direct product development work and proportion of indirect work as well as
¢ Intensity of product development project types [$ / type] show how the allocation is made
between different kind of projects.
The research challenge is to build models for deeper understanding of the impact of new
product introduction and strategic planning for time-to-market issues. One major question is
that firms should know both the profitability and productivity impacts of new product
introduction capabilities in order to market appropriate plans for time-to-market goals. Also
the relation to productivity measurement is important because in many industries the long term
competitiveness of a modem company is based on productivity improvement. Traditional total
productivity approach does not acknowledge the delayed investments for R&D even if they
were highly important for the company.
References
Cooper, R.G , SJ. Edgett & EJ. Kleinschmidt (1997). Portfolio Management for New
Products. McMaster University, Hamilton.
Craig, Charles E. & Harris, Clark R. (1973). Total Productivity Measurement at the Firm
Level. Sloan Management Review. 14(3). p. 13.
Davis, Hiram S. (1955). Productivity Accounting. Research Studies XXXVII, University of
Pennsylvania Press, Philadelphia.
Hansen & Weiss & Kwak (1999). Allocating R&D Resources: A Quantitative Aid to
Management Insight. Research - Technology Management. 42 (4) pp. 44-50.
Loch, Christoph, Lothar Stein & Christian Terwiesch. (1996). Measuring Development
Performance in the Electronics Industry. Journal of Product Innovation Management.
Vol. 13. pp. 3- 20.
Nachum, Lilach (1999). Measurement of productivity of professional services - An illustration
on Swedish management consulting firms. International Journal of Operations &
Production Management. 19 (9). pp. 922-949.
Sumanth, Davis (1997). Total Productivity Management. St. Lucie Press.
Werner, Bjom M. & William E. Souder (1997). Measuring R&D performance - state of the
art. Research Technology Management. 40 (2). pp. 34-42.
Werner, Bjom M. & William E. Souder (1997). Measuring R&D performance - U.S. and
German practices. Research Technology Management. 40 (3). pp. 28-32.
Wheelwright, Steven C.& Kim B. Clark. (1992). Creating Project Plans to Focus Product
Development. Harvard Business Review. March-April.
