Modeling the Dynamics of Technological Ramp-Up within Firms
by
E. G. Anderson Jr. and N. R. Joglekar!
Sloan School of Management, M.LT.
Cambridge, MA 02139
ABSTRACT
This paper uses the system dynamics methodology to integrate the dynamics of technological innovation into a
formal model of business strategy. This-model—known as the Technology Ramp-Up Simulator (TRUSIM)—
specifically focuses on the evolution of technological performance and competitive structure. TRUSIM links
existing firm management system dynamics structures to new modules representing the dynamics of innovation and
that of technological capability. This model will be used to explore several typical scenarios of industrial evolution,
such as technological lock-in and first-mover disadvantage
1. Purpose
Technological innovation is arguably the most powerful determinant of our economic future. The
improvement in the Western standard of living during the past three centuries would not have been
possible without it (Solow 1957). However, formal dynamic models of technological innovation are
relatively underdeveloped. Existing models are typically quite aggregate and do not separate product
from process innovation. There is a clear need for a dynamic model of innovation to reflect firm- and
industry-level causal factors, and provide insights into how to manage innovations effectively.
Firm Sectors
Innovates at Firm Level
Invents
Market Sector
imitates
Consumes Products
industry Sector Builds Technological Determines Market Shares
Ditfoses Knowledge Infrastructure
Forms Product Expectations
Ramps up Capacity Forms Complementary Assets
Produces Products
Sets Price
Figure 1: TRUSIM Sector Diagram
An overall schematic of a model, called the Technology Ramp-Up Simulator (TRUSIM), is presented in
Figure 1. A complete description and listing of the model is available in Anderson and Joglekar (1995).
In this model two new system dynamic modules formalizing the dynamics of innovation and
technological infrastructure have been integrated with “classic” system dynamic modules for market
behavior, capacity, production, and investment. The innovation module is based on the Abernathy-
Utterback (Utterback 1994) descriptive model of product and process innovation and Von Hippel’s
(1988) work on sources of innovation. The technological infrastructure module is based on work by Fine
and Joglekar.
2. Theory
There are two specific bodies of theory outside the common system dynamics literature that are used to
build this model. The first is the theory of innovation. Most formal innovation models fall into one of
two groups. Some are quite simple, often confounding many variables (Berndt 1991). Others are of the
black-box nature typical of forecasting models (Young 1993). There also exist several industrial
evolution models, notably Nelson and Winters (1982) and Levinthal and March (1988). These models
are extremely stylized and their use as a policy tool is problematic. In the realm of system dynamics,
' Digital Equipment Corporation, Maynard, MA 01754
a4
many informative models examine innovation at the macro level, which is too aggregate for the present
purposes (Graham and Senge 1980). The remainder, while set at an appropriate scale, do not separate
product from process. (See Homer 1981, Davidsen 1989, or Kreutzer 1989 for examples.) Packer
(1964) has created the model most similar to TRUSIM by differentiating the dynamics of capacity from
professional capability, but he also does not differentiate product from process investment.
Product Innovation
eos
Z \._ Process Innovation
RATE OF MAJOR INNOVATION
Fluid ‘Transitional Specific
Phase Phase Phase
Figure 2: Abernathy-Utterback Model of Innovation Rates
While formal dynamic models of product and process innovation are lacking, there do exist several
excellent descriptive models. Of these, probably the best known is the Abernathy-Utterback (AU)
model. Presented in Figures 2 is the key reference mode as described by the AU model. It begins with
an empirical observation that the rates of product and process development vary not only with time but
with each other. At the beginning of a new product’s life, product innovation occurs quite rapidly,
driven in part by the active participation of lead users. This improvement in product performance is
made even more rapid as the innovating firms learn where to focus their energies. The rate of innovation
begins to decline after a certain period with the appearance of a dominant design. This period of rapid
innovation is termed the “fluid” phase. As product innovation slows down, process innovation
accelerates and reaches a peak during the “transitional” phase. This phase coincides with a rapid rise in
product demand. Market competition between firms begins to shift from product features to price and
quality. During this phase the industry becomes concentrated as surviving firms struggle to take
advantage of economies of scale. During the last or “specific” phase, processes are capital intensive, and
neither product nor process can be easily changed due to their interdependence. Innovations become
ever more incremental. In this environment, change becomes difficult leaving industries vulnerable to
breakthroughs that reset the cycle.
The other body of theory is that of technological infrastructure, which is defined as the in-house know-
how that evolves as a result of investment in building technological capability. Technological
infrastructure is a stock, because it can either be accumulated or depleted over time. This definition of
technological infrastructure is congruent with a growing body of literature within the resource-based
view of the theory of the firm, which supports the view of knowledge as an asset or a stock (Dierickx and
Cool 1989). Technological infrastructure is measured as the accumulated capability of an organization
to apply technology to its processes of product development, manufacturing process development,
quality improvement, and cost reduction. Accumulation of a firm's internal technology infrastructure is
accomplished through individual and particularly organizational learning. Most firms complement their
infrastructures with those of suppliers, customers, and other strategic partners. The quality of these link-
ups, and the set of organizations with which the firm is linked, form a critical portion of the extended
infrastructure (the technology supply chain, see Fine and Whitney 1996). Presence of infrastructure may
be positively correlated with increased rates of innovation.
3. Initial Results
3.1 Base Case
This model has been tested using a set of stylized parameters that mimic the rate of innovation allied
with product and process development practices observed in the automotive and electronics industries.
Initial results of the model are quite promising. Presented in Figure 3 are time-series from a model
simulation which replicates the reference modes shown in Figure 2. In general the agreement between
the simulation and the reference modes is excellent
Industry Reference Modes (Base Conditions)
Nene y2H
Base Case Firm Sales
i 1
so
‘ 7 T
. ACH 20
“+ mae
pod SHES | °
ole | H t —_a L
a a
van
Pode! oan Rts anon
bo toate
Figures 3 and 4: Base Case Reference Modes and Firm Sales
There are five firms in the base simulation, which differ only in their initial number of product and
process engineers. Firm 1 has twenty product and twenty process engineers. Firm 2 has ten of each, and
Firms 3, 4, and 5, five of each. In the base simulation there is no possibility for product imitation and no
complementary assets. The resulting market sales are shown for Firms 1, 2, and 3 in Figure 4. Note the
two humps. The first one is the boom-and-bust effect (Paich and Sterman 1994) driven by the launch of
a new product. After the bust in year 12, however, capital procurement ceases by all firms. Hence,
while process innovation continues, productivity gain is slight. In year 24, enough capital has worn out
so that new capital is required to keep up with demand. Firm 2 begins the purchase of the new capital
because it overshot the least during the initial boom. Productivity of new capital drives down product
cost. Declining cost depresses prices, which leads to rapid influx of a new purchasers. These purchasers
create their own very slight boom-and-bust effect from year 24 to year 32. A salient feature of this
simulation is the domination of those who begin with the largest resources. This can be described as a
first-mover effect.
3.2 Base Case with Imitation
Figure 5 shows the results from the same base parameters when imitation occurs. The most interesting
result of the simulation is the elimination of the first mover advantage after year 10. In fact, Firm 2
briefly takes over market leadership until the re-convergence of all firms’ sales occurs in year 15. The
cause of this market convergence is the converging product performance as the technology matures. All
of these results derive from the lesser load of imitation on engineering staffs than that of innovation.
This difference allows the followers to create a viable product before they are locked out of the market.
3.3 Base Case with Imitation and Complementary Assets
Successful commercialization of innovation requires that the know-how in question be utilized in
conjunction with other capabilities or assets known as complementary assets (Teece 1987). Two
examples of these assets are after-sales support and the availability of application software for selling
personal computer hardware. This simulation allows both imitation and the presence of complementary
assets. Figure 6 shows that complementary assets can, for a while, reestablish the first mover advantage
lost through imitation. Firm 1 again dominates the market, this time by translating an earlier product
2k
launch into a complementary asset advantage which has a reinforcing effect on scale economies. Only
Firm 2 is strong enough to survive the shake-out before year 10. Interestingly, Firm 1 has become so
large and bureaucratized that by year 12 its product development efforts fall behind its competitors.
Firm 1's complementary assets, however, hinder Firm 2 from. capitalizing on its advantage. By year 24,
Firm 2 also has a price advantage, but it takes another five years to translate these advantages into market
leadership. At this point, Firm 1 is saddled with an inferior product and has too much inertia to easily
change its ways, so it loses sales which further erodes its complementary assets. This case has parallels
with the computer industry in the 1980s as IBM took control with its dominant PC design, and then lost
it to Compaq and others. On the other hand, with a slightly different parameterization one can also
generate modes of premature technological lock-in similar to Arthur (1989).
Imitation Case Firm Sales Imitation and Comp Assets Case Firm Sales
600
T |
T
Lt
tt 500 i
j | |
300 i Zt 1.000
i
we
;
i Cc To
0 i li 0 x
=
Figures 5 and 6: Firm Sales under Imitation and Complementary Assets
4. Next Steps
The model is being explored to find additional behavior modes. Simultaneously, validation against
collected personal computer and automotive electronics data has begun. Finally, simplified models are
being developed to capture the essences of strategy under technological innovation.
5. References
Anderson, E. G. Jr., and N. R. Joglekar (1995). Technology Ramp-Up Simulator Project Report. MIT $ D Group Memo.
Arthur, W. Brian (1989), "Competing Technologies, Increasing Retums, and Lock-In by Historical Events." The Economic
Journal 99 (March):116-131.
Berndt, E. R., (1991). The Practice of Econometrics: Classic and Contemporary. Addison-Wesley, Reading, MA, 1991.
Dierickx, I. and K. Cool (1989), “Asset Stock Accumulation and Sustainability of Competitive Advantage,” Man. Sci. 35, 12.
Davidsen, P. I. (1989). A Petroleum Life Cycle Model for the United States, with Endogenous Technology, Exploration,
Recovery, and Demand. System Dynamics Group D-Memo No. D-3974.
Fine, C. H. and D. E. Whitney (1996), “Is the Make-Buy Decision Process a Core Competence?” MIT Center for Technology,
Policy, and Industrial Development, Working Paper.
Graham, A. K. and P. M. Senge (1980). “A Long-Wave Hypothesis of Innovation.” Tech. For. & Soc. Change, 283-311.
Homer, J. B. (1981). A Dynamic Computer Model of an Emerging New Medical Technology. MIT SD Group Memo D-3318.
Kreutzer, D. (1989). A Feedback Analysis of First-Mover Advantages and Disadvantages. MIT SD Group Memo D-4107-1.
Levinthal, D. and J. G. March (1988). “A Model of Adaptive Organizational Search.” In Decisions and Organizations, J. G.
March (ed.). New York: Basil Blackwell, Ltd.
Nelson, R. R. and S, G. Winter (1982). An Evolutionary Theory of Economic Change. Cambridge, MA: Belknap Press.
Packer, D. W. (1964). Resource Acquisition in Corporate Growth. Cambridge, MA: MIT Press.
Paich, M. and J. D. Sterman (1994)."Boom, Bust, and Failures to Leam in Experimental Markets. Man. Sci., forthcoming.
Solow, R. M. (1957). “Technical Change and the Aggregate Production Function.” Rev. of Econ. and Stat., 312-20.
Teece, D. J. (1987), “Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing and Public
Policy,” in The Competitive Challenge: Strategies for Industrial Innovation and Renewal, CRM, Berkeley.
Utterback, J. M. (1994). Mastering the Dynamics of Innovation. Cambridge, MA: Harvard Business School Press.
Von Hippel, E. (1988). The Sources of Innovation. New York: Oxford University Press.
Young, P.(1993).“Technological Growth Curves: A Competition of Forecasting Models.” Tech. For. & Soc.Change, 375-390.
dT
CC BY-NC-SA 4.0