Kapmeier, Florian; "Dynamics of Common Learning in Learning Alliances", 2003 June 20-2003 June 24

Online content

Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents

Go Back

Dynamics of Common Learning in Learning Alliances

Florian Kapmeier
Universitat Stuttgart
Betriebswirtschaftliches Institut
Lehrstuhl fuer Planung und Strategisches Management - Prof. Dr. Erich Zahn
Keplerstr. 17
70174 Stuttgart
Germany
Phone: +49 —(0)711 - 121 —3465
FAX: +49-(0)711 - 121-3191
Email: florian.kapmeier@po.uni-stuttgart.de

Abstract

The paper proposes a System Dynamics model that gives deeper insights into the dynamics of
common learning in learning alliances. Although current research widely recognizes
alliances as an important strategic option to achieve strategic goals, feedback perspectives
are often neglected. A feedback perspective can be helpful to explain certain unanticipated
long-term effects such as insufficient learning outcomes in learning alliances. I transfer
findings from recent learning alliance literature into a System Dynamics model that consists
of three major and four minor feedback loops which I discuss in detail. Then, different
scenarios offer insights into the dynamics of common learning, perceived benefits, associated
resource allocations and the development of trust in a learning alliance under the
consideration of varying parent companies’ expected benefits. I introduce a model that
supports decision-makers to understand how expectations of learning outcomes may affect
success or failure of learning alliances.

Keywods
Interorganizational Learning, Learning Alliances, Relative Scope, Common Benefits,
Common Learning, Resource Expenditures, Knowledge, Trust

1. Introduction

During the past decade, the formation of alliances and joint ventures has become a common
element of firms’ strategies in all kinds of different sectors to achieve competitive advantage
(Aloysius, 2002; Lane, Salk, and Lyles, 2001; Zahn, 2001; Inkpen, 2000a; Child and
Faulkner, 1998; Doz and Hamel, 1998; Gulati, 1995; Ring and Van de Veen, 1992). This
trend is evidenced by the fact that more than 20,000 alliances have been reported world-wide
over a period of only two years (Kale, Dyer, and Singh, 2002; Anand and Khanna, 2000).
Besides various motives for founding alliances, interorganizational learning is of increasing
importance for realizing strategic goals (Lane, Salk, and Lyles, 2001: 1139). Even though
alliances are widely discussed in the literature, only a few articles include a dynamic approach
to study the development of alliances over time. However, they often tend to neglect a
feedback perspective which may lead to a short-term analysis. In order to show the long-term
effects of today’s decisions, it is valuable to consider feedback loops. They show
accumulations, time delays, and non-linearities and hence exhibit the long-term effects of
decisions. This makes it possible to explain specific behaviors and effects and to unfold the
dynamics of phenomena over time. Feedback loops are essential for defining a dynamic
hypothesis and to show, model, and illustrate dynamic behavior. System Dynamics provides
such a simulation approach on the basis of feedback loops (Sterman, 2002; Sterman, 2000;
Forrester, 1961). In this paper, a System Dynamics model is designed to gain deeper insights
into the dynamics of common learning in a learning alliance that is founded by two parent
companies.

2. Background and Literature Review
In this section I discuss the theoretical background of learning alliances. I first define the term
learning alliances. Then I introduce the concept of relative scope.

2.1 Learning Alliances

Alliances and joint ventures can be understood as interorganizational cooperations of at least
two companies that are legally and — under certain conditions with some constraints —
economically independent for the duration of the cooperation (Bea and Haas, 2001: 419). This
means that these interorganizational forms mingle both hierarchical and market-oriented
aspects of coordination.

The founding parent companies agree upon certain restrictions regarding their freedom of
choice in order to implement common objectives within determined areas of mutual interest
(Pausenberger, 1989: 621). The motives for companies to form or enter alliances are widely
situated, e.g., in risk reduction, in the achievement of economies of scale and/or
rationalization (Child and Faulkner, 1998: 32), in the reduction of transaction costs (Picot and
Dietl, 1990: 178), in the development and conquest of new markets as well as in the
concentration of core competencies, in the concentration of market power, or in the
acquisition of knowledge (Lane et al., 2001: 1139; Child, 2001: 657; Zahn, 2001: 11; Prange,
1996: 164; Hamel, 1991: 83).

The focus of this paper lies on the latter, knowledge acquisition through interorganizational
learning. Lately, there has been an increasing body of both theoretical research (see, for
example, Carlile, 2002; Goussevskaja and Kidd, 2002; Inkpen, 2000a; Khanna, Gulati and
Nohria, 1998; Kumar and Nti, 1998; Larsson, Bengtsson, Henriksson, and Sparks, 1998;
Grossman and Shapiro, 1987) and empirical studies (see, for example, Amburgey, Dacin, and
Singh, 2000; Beckman and Haunschild, 2002; Tsang, 2002; Lane and Lubatkin, 1998; Inkpen
and Crossan, 1995; Hamel, 1991) focusing on the issue of learning in the interorganizational
context. An increase of knowledge through employees’ learning can create new business
opportunities and hence build up sustainable sources of competitive advantage (Spender and
Grant, 1996: 5; Penrose, 1959: 24). These days, the sources of competitive advantage are
more frequently breakthrough innovations that increasingly rely upon interdisciplinary and
interindustry research and lie beyond the research capabilities of just one single firm
(Lubatkin, Florin, and Lane, 2001: 1353)

2.2. Concept of Private and Common Benefits

Even though learning alliances have been subject to recent research, most studies focus on
specific questions in the field of interorganizational learning (e.g., Belderbos, 2003; Larsen et
al., 1998; Doz, 1996; Ring and Van de Ven, 1992). Only some imply a dynamic approach
(like e.g., Lane, Salk, and Lyles, 2001; Khanna et al., 1998; Khanna, 1998; Kumar, Nti, 1998;
Doz, 1996). Khanna et al. (1998) and Khanna (1998) address the question of how a learning
alliance develops over time on the basis of the concept of the relative scope (Khanna et al.,
1998: 193; Khanna, 1998: 350).

The nucleus of the relative scope lies in the parent companies’ choice of alliance scope that
considerably affects the character of benefits that the parent companies may reap (Khanna,
1998: 340). The parent companies themselves focus on a certain set of activities called the
firm’s scope. Generally speaking, the alliance scope typically refers to some subset of markets
in which the partner firms are involved (Khanna, 1998: 344; Inkpen, 2000b: 776). The overlap
between the scope of the alliance and the total market scope of each partner is called the
telative scope (Khanna et al., 1998: 195).

The overlap between the alliance scope and parent 1’s scope can be much bigger than parent
2’s overlap. In conclusion, one could assume that both parent companies have different goals
in this particular alliance. This observation leads to the concept of private and common
benefits. The relative scope determines each parents’ expected private and common benefits
that accrue to the parent companies from an alliance (Khanna et al., 1998: 195; Khanna, 1998:
341). Private benefits can be defined as those kinds of benefits that a parent firm can earn
unilaterally by picking up skills from its alliance partner and applying them to its own
operations in fields unrelated to the alliance activities (Khanna et al., 1998: 195).

Common benefits can be understood as those kinds of benefits that “accrue to the alliance
parent from the collective application of the learning that both firms go through as a
consequence of being part of the alliance; these are obtained from operations in areas of the
firm that are related to the alliance” (Khanna et al., 1998: 195). The common benefits earned
by the parent companies need not be equal for both parents.

The relative scope provides the basis for understanding the parent companies’ resource
allocation patterns. Different relative scopes lead to different resource allocation behaviors as
the parent companies are driven by different needs or interests. For the purpose of unveiling
the dynamics of learning alliances the concept of the relative scope is helpful. It covers the
tension between competitive and cooperative behavior. So it can be stated that it features a
dynamic perspective on alliance development (Inkpen, 2000b: 775; Khanna et al., 1998: 193).
As the dynamics of learning and resource allocation are the main focus of this paper, I employ
this concept.

In this paper, I analyze the dynamics of a partnership where only common benefits can accrue
by using a System Dynamics modeling approach. I examine how different expectations of
common benefits, in a learning alliance with only two parent companies, determine the
alliance’s duration of existence.

3. Hypothesis

In this section, I present the hypothesis of this paper on the basis of the theoretical concepts
described above. he focus is only on the alliance itself, with the parent companies being left
out. Still, narrowing the focus on just the alliance, the approach of Khanna et al. (1998),
further enhanced by learning alliance literature, delivers interesting implications regarding the
behavior of the parent companies’ engagement, e.g. in terms of resource allocations to the
alliance.
For the model it is assumed that two parent companies have founded a learning alliance. They
have different backgrounds and relative scopes; i.e., one company has knowledge on the
market whereas the other has knowledge on products or processes. This implies that they
might also have differing expectations of the common benefits that may result from joint
learning. Higher expectations might be due to the fact that there is a greater overlap of the
alliance scope with the firm scope. Consequently, the parent company is very concerned about
successful learning outcomes as the alliance covers a great scope of the own market. This
means that the parent is interested in a long-term existence of the alliance with high common
learning outcome. A partner who has only little overlap of the scopes would not be that
interested in a long-term existence which might be a cause for early termination of the
alliance. Hence,

Hypothesis: common alliance learning can only be successful if the alliance is equally
important for both parent companies, meaning that they both have the same
expectations of the accruing common benefits.

4. Development of a Dynamic Model for Common Learning

Here, I first present the method I used to develop my model. Further, I talk about the model
boundary, followed by the presentation of a reference mode and a detailed description of the
stock and flow diagram. This is done by explaining the different loops I discovered. Finally, I
explicate the baserun and the different scenarios I produced.

4.1 System Dynamics Method

As stated above, even though learning alliances have been subject to recent research, only
some imply a dynamic approach. The models being designed often concentrate on specific
building blocks of the field of research on learning alliances and/or neglect a feedback-loop
point of view (Kapmeier, 2002: 2). Therefore, a dynamic approach provides an ideal tool for
analyzing the dynamics of learning alliances. It would be able to capture the behavior that
develops over time by simulating the behavior of interrelated variables. System Dynamics
(Sterman, 2000; Forrester, 1961) offers such a simulation method on the basis of feedback
processes with stock and flow variables.

The possibility to run different scenarios shows decision-makers long-term effects of
decisions instantaneously. As shown below, these might refer to the resource allocation in
terms of whether or not to continue a learning alliance.

The model is based upon the findings of carefully reviewing relevant learning alliance
literature. It is designed with the Vensim software.

4.2 Model Boundary
In the following sections, I present the model that I developed to test my hypothesis. I
determine the model boundary and exhibit the assumptions that I made.

In the model, a reciprocal learning alliance (Lane et al., 2001) is being considered. It is
founded by two parent companies. Only common benefits (Khanna et al., 1998; Khanna,
1998) can arise. It is assumed that the common benefits that arise built up upon each other. In
other words, one common benefit has to occur before a second can occur. The alliance is set

4
up to work on a specific task, i.e. a project limited to a certain time-span that demands
resources. Its aim is to create new knowledge through a blending of knowledge, co-learning
and joint discovery (Lane et al., 2001: 1363). Social similarity of the alliance parents or
interfirm diversity (Dengel and Milling, 2002; McAllister, 1995; Zucker, 1986), knowledge
relatedness (Lane et al., 2001), knowledge tacitness or explicitness (Spender, 1996; Polanyi,
1966), dominant logic (Bettis and Prahalad, 1995) and dynamic capabilities (Eisenhardt and
Martin, 2000; Hagedorn, Link, and Vonortas, 2000; Teece, Pisano, and Shuen, 1997; Teece
and Pisano, 1994) are keywords and concepts often discussed in the alliance literature. They
are indeed valuable for alliance research. However, to keep the model simple, abstractions are
necessary. Therefore, these by no means unimportant ideas are not taken into consideration
and hence lie outside the scope of the model.

The alliance knowledge base is the focus of the model. It is decreased by unlearning and
increased by learning (Zahn and Tilebein, 2000: 122; Prange, 1996: 172; Luhmann, 1994:
448). It is assumed that the alliance can only realize single-loop learning (Argyris and Schén,
1978). This means that it is not possible to leave the knowledge path and leap onto another
one. Moreover, resource expenditures by both parents are taken into consideration. It is
assumed that resource expenditures only include human resources. This way, expenditures of
other resources like money are subtly included as people require offices, laboratories or other
infrastructure to work on their task. Also, it is also assumed that only the parent companies
provide the alliance with people. These resource allocations are not limited. Hence, there is no
overall budget for the project that the alliance works on. Consequently, the parent companies
allocate more resources to the alliance when they perceive progress (Grossman and Shapiro,
1986: 592)." When they perceive progress, the expected value of completing the project
increases. The higher the expected value of completing the project, the more resources the
parents allocate to the alliance. Transparency between the parent companies is also assumed.
This implies that both parent companies are aware of how much resources the partner spends
on the alliance and how far they are in achieving their expected benefits.

In addition, the parent companies follow an equilibrium strategy that is based on stability,
optimality, and rationality. Subsequently, the partners follow the Nash equilibrium, indicating
that both parent companies believe that they are doing the best they possibly can, given the
actions of the partner (Ho and Weigelt, 1997: 132)

It is further assumed that resource expenditures are made on the bases of the individual parent
company’s expected and perceived common benefits (Khanna et al., 1998, Khanna, 1998)
The parent companies’ expected common benefits are determined by the parent companies’
telative scopes. In the model, the relative scopes are seen as exogenous. Nevertheless, inputs
into this exogenous variable can indeed vary. This implies that the parent companies’ relative
scopes also vary. For example, the firm scope would increase if a parent company conquered
new markets outside of the alliance scope. Hence, this company’s relative scope would
decrease.

Moreover, trust on the group level is being considered (Curral and Inkpen, 2002). It is
assumed that an existing positive atmosphere on the firm level that might be due to previous
ties between the parents (Gulati, 1995), can be shifted onto the group level. Trust encourages
the people involved in the alliance to interact more frequently and communicate with each
other.

Absorptive capacity (Cohen and Levinthal, 1990) and information redundancy (Nonaka and
Takeuchi, 1997) are also being considered in the model. Redundancy is operationalized via
the variable ‘assessment of new knowledge’. With high redundancy, the assessment of new
knowledge is low. This means that, by intuition, the ‘absorptive capacity’ is highest in a
situation where the ‘assessment of new knowledge’ is lowest. Consequently, people
understand quickly what they learn — but what is newly learnt is only little value to them.
Both ‘absorptive capacity’ and ‘assessment of new knowledge’ influence the learning aptitude
of the people who work in the alliance.

Thave left out the concept of relative absorptive capacity (Lane and Lubatkin, 1998) as it
focuses on the knowledge bases of different organizations. Even though the people who work
in the alliance come from two different companies, they work on the same task — there is no
student-teacher learning involved in this situation. This would be the case where knowledge is
transferred from the alliance to the parents’ knowledge bases. However, accruing private
benefits is out of the model’s scope.

Most of the variables within the model boundary are qualitative and therefore difficult to
quantify. However, they are essential for understanding the structure that determines the
dynamics of interorganizational learning. Therefore, “leaving such variables out of models
just because they lack of hard numerical data is certainly less ‘scientific’ than including them
and making them reasonable estimates of their values” (Sterman, 1991: 12; see also: Sterman,
2002: 523)

4.3 Reference Mode
In this section I present the reference mode that the model refers to (Sterman, 2000: 90). Here,
the dynamics of resource spendings over time are looked at closely.

As stated above, it is assumed that the alliance can only generate common benefits. Thus, it is
a situation of pure cooperation. As the partners follow the Nash-equilibrium, resource
allocation decisions are best made together. Both parent companies come to an agreement on
the amount of resources needed to be allocated depending on the progress of the project. They
act as one single firm. (Khanna et al., 1998: 197; Grossman and Shapiro, 1987: 378)

Following Grossman and Shapiro (1987) and Khanna et al. (1998), the resource allocation of
the parent companies would look like depicted in the Figure 1. It can be seen from the figure
that the project requires the completion of two milestones. As long as the learning alliance
still works on milestone 1, resource allocations stay constant at a certain level. Once the first
milestone is finished, the alliance works on milestone 2. As the expected value of the project
increases at this moment, resource allocation by the parents is increased until the project is
finished. Even though the reference mode indicates milestones, there are no milestones being
considered in the model. It is assumed that resource allocations are made constantly.
Firm 1‘s resource expenditures

Milestone 2

Resource Expenditures

Milestone 1

Time
Figure 1: Reference mode: Resource spending in a situation of only common benefits. Source:
following Grossman and Shapiro, 1987: 374.

4.4 Stock and Flow Diagram
In this section, the stock and flow diagram and the baserun as well as different scenarios will
be presented.

4.4.1 Loop Descriptions

The model designed consists of three major and four minor feedback loops. First, there are
two major balancing feedback loops, called ‘reaching the goals’ (Kapmeier, 2002: 4) whose
structure is identical. Each refers to reaching the goals of the two parent companies. Second, a
reinforcing loop shows that ‘trust enhances learning’ (Kapmeier, 2002: 3). Third, a
reinforcing loop illustrates that a larger knowledge base ‘enhances absorptive capacity’
(reinforcing loop). But, fouth, at the same time, ‘information gets redundant’ (balancing loop).
Finally, two structurally identical balancing loops called ‘spending regulative’ monitor and
control the resource allocation ensuring the two parents’ resource spendings even out

4.4.1.1 Bla and Blb - Reaching the Goals

Both Bla and B1b — Reaching the goals are identical feedback loops that are intertwined. For
better understanding, I will first explain Bla. The loop regulates the optimal resource
allocation in proportion to the expected value of the project. There are four accumulations, or
stocks (see Figure 2).
Normal

unlearning rate a
+” Alliance

unlearning

Alliance
knowledge base

Knowledge application +
to products or patents

<—
lAlliance learning Normal learning

Success rate of
Actual alliance new ideas
benefits

Benefits gap

Updated perceived

alliance benefits

Time to perceive
alliance benefits

Perceived alliance
benefits

Bia Common resource
spending

Reaching the goals
Expected common
benefits 1 in respect
tot's relative scope
Degree of goal
attainment - parent
‘i

Resources spent on
alliance - parent 1

Resource

a a

Effect of degree of goal Resource allocation
attainment on resource from goal attainment
location: f t Time for allocation
+ decision 1

Normal resource
allocation 1

Figure 2: Bl — Reaching the goals

The stock ‘alliance knowledge base’ (measured in knowledge points) indicates how much
accumulated knowledge the alliance has. It is initialized at 2 knowledge points, which stands
for a common basic understanding of the project to work on. The ‘alliance knowledge base’
increases by ‘alliance learning’ and decreases by ‘alliance unlearning’ (knowledge points per
week). ‘Alliance unlearning’ (Giildenberg, 1996; Bettis and Prahalad, 1994; Hedberg, 1981)
is on the one hand determined by the size of the stock. The more people know, the more they
unlearn. On the other hand, ‘unlearning’ is influenced by a ‘normal unlearning rate’ (1/ week).
This variable refers to the phenomenon that people forget things over a certain period of time.
Here, it is assumed that the employees working in the alliance forget 1% of their knowledge
per week. The ‘alliance knowledge base’ is increased by ‘alliance learning’. This flow
variable is determined by the ‘resources spent on the alliance’ (people) and ‘normal learning’
(knowledge points per people and week). ‘Normal learning’ is a constant that provides
information about the people’s learning efficiency. It is assumed that one person can learn 0.5
knowledge points per week under normal learning conditions.

A firm does not necessarily benefit from the acquisition of knowledge unless the knowledge
is actually applied (Child and Faulkner, 1998: 283). The more people learn, the more new
patents can be registered (Inkpen and Crossan, 1995; March, 1991; Lyles and Salk, 1996;
Argyris and Sch6n, 1978). However, not everything newly learnt is useful for the
development of new products or suitable for a patent registration. This is covered by the
constant ‘success rate of new ideas’ (benefit points per knowledge point). The flow

‘knowledge application to products or patents’ (benefit points per week) increases the stock
‘actual alliance benefits’ (benefits points).

As it takes ‘time to measure and perceive the alliance benefits’ (weeks), an information delay
is included in the model (Sterman, 2000: 411). This smoothing structure includes the
‘perceived alliance benefits’ (benefit points), the “benefits gap’ (benefit points) which
computes the difference between the ‘actual alliance benefits’ and the ‘perceived alliance
benefits’, and the flow ‘updated perceived alliance benefits’ (benefit points per week) that is
intended to close the ‘benefits gap’.

The parent company decides to join the reciprocal learning alliance to realize three patents,
for example. It does so considering its relative scope. It is assumed that the expected alliance
benefits increase with an increase of the overlap between the alliance scope and the firm
scope. This is captured in the exogenous variable ‘expected common benefits of the parent

company in respect to the parent company’s relative scope’. In the baserun it is constant with
three benefit points and it varies in the scenarios

baer
‘HEffectflegreeofgoalattainmentoruescurcealocation]#
1

o7s

os \

0 00 1 190 2

x
Figure 3: Nonlinear relationship for ‘effect of degree of goal attainment on resource
allocation’

There is a nonlinear relationship between the goal attainment and the resource allocation that
is captured in a robust table function (Sterman, 2000: 553; see Figure 3). When the goal has
almost been reached (the ‘degree of goal attainment on resource allocation’ is small), the
parent company only allocates few additional resources to the alliance. However, when the
goal is still far off (large ‘degree of goal attainment on resource allocation’), the parent
company allocates more resources to the alliance in order to reach the goal faster (Grossman
and Shapiro, 1987: 378).

The actual number of resources allocated to the alliance is calculated by multiplying the
‘effect of goal attainment on resource allocation’ (dimensionless) with the ‘normal resource
allocation’ (people). The latter variable states that the parent companies allocate 10 people
under normal conditions. Now, depending on the ‘degree of goal attainment’, the actual
resource allocation might be smaller. The number computed is called ‘resource allocation
from goal attainment’ (people). The higher the ‘resource allocation from goal attainment’, the
more people will be allocated to the alliance the next period via the variable ‘resource
allocation’ (people per week). ‘Resource allocation’ is a biflow that can either increase or
decrease the stock ‘resources spent on the alliance’ (people). As another variable (‘resource
allocation from allocation difference’) from Loop B1b determines the biflow, the underlying
policy will be discussed later. In any case, the ‘resources spent on the alliance’ influence how
much people learn in the alliance, considering ‘normal learning’

The second balancing loop, B1b — Reaching the goals, refers to the second parent company’s
resource allocation structure. From Figure 4 it can be seen that it is structurally identical to
Bla described above. B1b connects to Bla at three variables, ‘perceived alliance benefits’,
“common resource spending’, and ‘resource allocation’. As with parent company 1, parent
company 2 also measures up the ‘perceived alliance benefits’ against its ‘expected common
benefits’. The ‘resources spent on the alliance’ from both parent companies add up at
“common resource spending’.

Normal

unleaing rate ~ia
"a + Alliance
unlegming
Aliance
knowledge base
a
Knowledge applica
to products or patents \Alliance learning
‘Success rate of =
Actual aliance new ideas
+ benefits —
Benen oak Normal learning
Updated perceived Bia
alliance benefits TS ‘Common resource
Tin to perceive spendin
Tine Be petooive Reaching he goals Ad
Perceived
atiance benefits| Expected common
benetis 1 in respect
| _ietsrelave spe
Degtee of goal
attainment - parent Time for allocation
1 decision 1
Resources spent on
Expected common alliance - parent 1
benefis 2 in respect to Resource
+ ze relative scope : allocation t+
Degree of goal Effect of degree of goal r
attainment parent atteinment on resouroe we
2 ‘allocation 1 Resource allocation penta ates
4. ftom goal attainment coal
See
Nofinal resource Resource allocation
allocation 1 from allocation
Giference t
Bib &
“Time for aloction
Reaching te goa ee
Ered of degree of goal Resources spent on
attainment on resource en alliance - parent 2
‘allocation 2 Recburce

allocation 2
ea
J. Resource allocation

from goal attainment Spending regulative

bi Resource allocation

Normal resource from allocation
allocation 2 difference 2

Figure 4: Bla and B1b — Reaching the goals, parent 1 and parent 2

10
4.4.1.2 B3a and B3b - Spending Regulative

Both parent companies’ resource allocations influence each other. As stated above, ina
situation of pure common benefits, the two parent companies act as one single firm (Khanna
et al., 1998: 197). This indicates a structural regulative ensuring that both companies always
spend roughly the same amount of resources on the alliance. This is captured by the two
balancing loops B3a and B3b — Spending regulative. Any resource allocation differences of
the two parents are registered in the two separate variables ‘resource allocation from
allocation difference’ (number of people). Now, the parent companies’ resource allocations
depend on two variables, ‘resource allocation from allocation difference’ (people) and
‘resource allocation from degree of goal attainment’ (people). From the underlying policies,
six different policies might occur (see Table 1).

In any case, if the ‘degree of goal attainment’ indicates that the expected common benefits are
achieved, the task of the alliance is accomplished. Consequently, the parent decides to
withdraw its resources from the alliance. If the goal has not yet been accomplished, the parent
might either further increase the resources considering that the ‘resource allocation from
allocation difference’ indicates that the partner spends the same amount of resources or more.
Or, if the partner invests less, resources are withdrawn in order to get equal again with the
partner.

Withdrawal of Withdrawal of Withdrawal of
resources resources resources
(Resource allocation (Resource allocation (Resource allocation
from gap policy). from gap policy). from gap policy).
Increase resources Increase resources Smooth withdrawal
(Resource allocation (Resource of resources even
from gap policy). allocation from gap _ though the firm has
policy + Resource not reached its goals
allocation from yet
allocation. (Resource allocation
difference policy). from allocation
difference policy).

Table 1: Resource allocation policies

If resource allocations are increased, a third order delay is involved in the process as it takes
time to perceive the need to increase the resources, or to choose the right people or what you
have, for instance. This ‘time for the allocation decision’ (weeks) is constant at four weeks. If
resources are withdrawn from the alliance, there is no delay involved. The moment the goal is
achieved, resources are withdrawn immediately. It is assumed that there is no smooth
transition from putting people into the alliance to withdrawing them from it

11
4.4.1.3 R1- Trust Enhances Learning

The reinforcing loop R1 — Trust enhances learning describes the relationships between
performance, trust, learning, and, again, performance. The loop connects to the balancing
loops Bla and B1b — Reaching the goals at ‘perceived alliance benefits’.

The people in the alliance measure up the perceived alliance benefits with their own
expectations. The higher the ‘degree of goal achievement’ (again, this variable is
dimensionless as this is the input for another nonlinear relationship, thus the table can be
considered robust), the more the people believe in the success of the alliance (see Figure 5).
This means that they feel attached to the alliance, which can be interpreted as intergroup trust
(Currall and Inkpen, 2002: 488).

baer
He Sfectofgoslatainnentosb lievinginalliancenuccesst
1

07s

o 025 00 7 1
=

Figure 5: Nonlinear relationship for ‘effect of goal attainment on belief in alliance success’

As can be seen from the Figure 5, a certain trust threshold or initial trust exists even when
there is only little or no goal attainment. This is due to the observation that “there is an
element of trust in every transaction” (Ring and Van de Ven, 1992: 488). The curve in Figure
5 is s-shaped which indicates that trust develops slowly when just little progress is observed,
though, when the goal is achieved, it reaches a limit. In any case, the trust being generated is
multiplicated by a normalizing factor, ‘normal belief (belief points). The stock ‘intergroup
belief in alliance success’ (belief points) is a fragment of a smoothed-information structure
(see also the structure of ‘perceived alliance benefits’). There is a biflow ‘change in belief”
(belief points per week) that can either increase or decrease the stock.

12
Normal B
unlearning rate tm
+ Alliance

unlearning
+

Alliance
knowledge base

Knowledge
japplication to. +
products , [Alliance learning
Success rate of
Actual alliance new ideas

benefits

A
Benefits gap

+ Commitment to learning
through belief in alliance

success

Updated perceived

Trust enhances
alliance benefits

leaming
Time to perceive
alliance benefits

Perceived
alliance benefits Expected Time to form
; alliance benefits belief
+
Degree of goal of Intergroup belief in
attainment - alliance success
aCe oy Change in belief

Sa Effect of goal MS
attainment on belief in . Normal belief
alliance success. peer
Sasa co a ae
+

Figure 6: R1 — Trust enhances learning

“Intergroup belief in alliance success’ describes the people’s attachment to each other. The
attachment is purely formed through the perceived alliance benefits. This means that
perceived progress strengthens the people’s dedication to their work. The parent companies’
“expected common benefits in respect to their relative scope’ do not influence the ‘intergroup
belief in the alliance success’. It is assumed that the people working in the alliance only
dedicate their time and energy to their research findings and not to coordinating the degree of
their parent companies’ goal attainment. The moment one parent withdraws its people from
the alliance, intergroup belief diminishes quickly." Further, the more the people trust each
other, the more they are willing to share information, the more open they are, and the more
they communicate and interact with each other as they commit to their task and to the alliance
(Inkpen, 2000b: 1028; Kumar and Nti, 1998: 360). Hence, they learn together and generate
alliance benefits.

4.4.1.4 R2- Enhancing Absorptive Capacity

In this and the following section, I describe the loops R2 — Enhancing absorptive capacity and
B2 — Information redundancy (see Figure 7). The ‘alliance knowledge base’ (knowledge
points) needs to be normalized for determining both the ‘absorptive capacity’ and the
‘assessment of knowledge value’. As no real quantity can grow forever (Sterman, 2000: 295),
it is put in relation to an assumed ‘maximum knowledge base’ (knowledge points). Thus, the
resulting ‘relative knowledge base’ is a dimensionless variable.

13
Normal

unlearning rate aa

+ “Alliance

er t Pare
Alliance

knowledge base

Maximum

4 oe

knowledge base

he)

Information

“ce base

t=)
Assessment of *Absorptive

knowledge value Enhancing capacity

absorptive capacity
+

redundancy
Leaming aptitude
[Alliance learning

Commitment to learning Common resource
through believe in spending
alliance success

Figure 7: R2 — Enhancing absorptive capacity and B2 — Information redundancy

Normal learning

The ability to recognize, learn and use new knowledge is included in the ‘absorptive capacity’
(dimensionless) (Cohen and Levinthal, 1990). Here, it is assumed that the absorptive capacity
increases in an s-shaped way. Accordingly, the higher the ‘relative knowledge base’, the
higher the capacity to absorb new knowledge (see Figure 8).

Daven
Absorptivecapacity#
1

02s

025 050 075 1

oe
Figure 8: Nonlinear relationship for ‘absorptive capacity’

Thus, the absorptive capacity increases with an increasing knowledge base. This means that
the people understand their task better the more they know. However, as can be seen from
Figure 8, the people working in the alliance cannot perform double-loop learning (Argyris and
Schén, 1978) as there are no learning leaps

4.4.1.5 B2- Information Redundancy

At the same time, however, the more the people know, the less new information they actually
teceive (Nonaka, Toyama, and Byosiére, 2001; Lane and Lubatkin, 1998; Nonaka and
Takeuchi, 1997). This means that the assessment of knowledge value is decreasing. It
decreases slowly in the beginning, then faster and finally more slowly again, which leads to
an inversed s-shaped curve (see Figure 9)

14
barern
WA sesmmentofloorledgevabset
1

07s

o 025 050 O75 1

és
Figure 9: Nonlinear relationship for ‘assessment of knowledge value’

The loops explained above are now put together into one single stock and flow diagram (see
Figure 10). This model structure provides the basis for the dynamic analysis of alliance
learning. I present the different runs in the next section.

15
Normal
Unleaming rate a
* aiance
uniearning

Maximum
knowledge base
/

alative

Pie oot ae

Alliance
knowledge base

snfdion

1aliance

a
Knowledge appcatcnts
to products or patents

\ rate of
‘Actual allance “ew ideas

benefits

latiance benefits|

Updated perceived
alliance benefits|

Time to perceive

alliance benefits

Expected Time to form

redundancy

Commitment to leaming
through belie in aliance

Intergroup belief in
alliance success

Degree of a

attainment parent Time for allocation

= enefs Belt
“SSesonensiia/ ¥
a ‘Srange in beet
Effect of goal 7
es tidal :
‘altace success F
ee potir quantity, “ele GaP ae
Expected common Peed 2
venets tin ospect
Le

Normal batiot

ji of

Ae) comarca
i oe,
{_Uaaing stud
te

Expected common! decision 1

benefits 2in respect to
2s relative scope

Resource

Resources spent on
alliance - parent 1

x Reaching the goals

Erect of degree of goal

attainment on resource
allocation 4

Degree of goal
attainment - parent
2 Resource allocation
4 ftom goal attainment
ss

Reaching the goals

“Time for allocation
decision 2
Effect of degree of goal
attainment on resource

allocation 2 Rechurcs
alocaton 2

Resource allocation
‘rom goal attainment
2

‘Normal resource
allocation 2

Figure 10: Stock and flow diagram of common learning

4.4.2. Findings from the Dynamics of the Model

Resource allocation
from allocation
cifference 1

Resources spent on
alliance - parent 2

+ Rosorptive
capacity

Normal learning

| comp nue

©)

Spending regulative

Resource allocation
from allocation
diflerence 2

4.4.2.1 Baserun: Both Companies have the Same Expectations of the Alliance’s

Common Benefits

In the baserun, it is assumed that both companies expect the alliance to accrue the same

common benefits which is constant at three benefit points. Figure 11 shows the behavior of
‘resources spent on the alliance’ by both parents, ‘degree of goal attainment’ and both ‘actual’
and ‘perceived alliance benefits’ over a period of 50 weeks. As can be seen from the figure,
the resource spending patterns of both parent companies are identical. They first increase
progressively (here, until roughly week 15) and then at a diminishing rate until they reach a
peak (week 20). This is the moment when the goals are achieved. Structurally, when both
parents invest the same amount of resources, the allocation difference is zero at all times.
Hence, the resource allocations only rely on the degree of goal attainment that follows an s-

16
shaped curve, the slope depending on the ‘perceived alliance benefits’. As discussed above,
the ‘perceived alliance benefits’ lag behind the ‘actual alliance benefits’ due to time delays.
Therefore, the alliance still keeps on working on the task even though the actual goals have
been attained (here, after week 17). Due to the delayed benefit perception, time passes by
until benefits continue occuring. Therefore, the degree of goal attainment becomes larger than
one. Once the ‘perceived alliance benefits’ equal out to the expected common benefits (week
20), the parent companies finally perceive that the alliance’s goals are achieved.
Consequently, the parents are satisfied and thus withdraw their resources immediately from
the alliance.

Resources parents 1 and 2

° = 10 15 20 35 3 35 40 35 30
Time (week)

"Degiee of goal tsiment- parent 2" bac - Daal
‘Acialalance benefit: bare * * Benefit point
Pasceired alliance benefit“ bacen - Benefit pint

Figure 11: Resource spendings - baserun

At first sight, the graphs for resource allocation in the baserun (Figure 11) and the reference
mode (see Figure 1) look different. This might be an indication that the model does not
reproduce the behavior that serves as the reference mode. Grossman and Shapiro (1987)
follow a more discrete development approach of dynamics with resource allocations only
occurring at the end of a certain event (Grossman and Shapiro, 1987: 374). System dynamics,
however, follows continuous integration with smooth developments and transitions over time.
Both behavioral modes show increasing resource allocations with the alliance progressing
towards the goals. Therefore, both representations over time can be seen as similar.

It is stated above that the benefits arise through the applying what has been learnt in the
alliance to products or patents. As is shown in the Figure 12, the ‘alliance knowledge base’
accumulates in an s-shaped curve and reaches a peak (at roughly week 20). The ‘alliance
knowledge base’ is determined by ‘alliance learning’, the flow to build up the knowledge
accumulation. The stock first increases exponentially, then asymptotically towards a goal.
This behavior is determined by the flow that first increases, peaks during the knowledge
base’s transition phase and then drops nearly exponentially.

17
LEARNING

20 Kaowledge points
2 Kawrrledge point reel]
4 Benefit point

40 People

10 Knowledge point f sal
1. Knowledge pointshreek|

2 Benefit point

2D People

0 Krorrledge points
0 Kawrrledge point reek

0 Benefit point
O People

ra 0
Alliance lneneledge base basen Kawnwledge points
Alliance leaming basen Kawnwledge pointshreek
Alliance ualeaming “bare 2 2 2 Kanrledge pointshreek
Perceived allance benefit : bac ‘Benefit points
“Resouces spent on alliance - parent 1": basema—+ . . . People
“Resources spent om alliance -pazent 2" bacem People

Figure 12: Learning and knowledge base - baserun

The bumpy and rough slope of ‘alliance learning’ in the declining phase is due to the different
variables determining it (see Figure 13). When the ‘learning aptitude’ decreases, the
commitment to the alliance and the resource expenditures still increase. In other words, even
though the ability to learn declines, the parents spend even more resources on the alliance
(here, after week 11)

ace,
Alliance leasing

2
15
1 =
os —
0
‘Comnitmnent fo leaming tough belief alliance success
100
B

0
25

| Sl ce
a =
‘Leaming aptitde
o4
03 o*
02 oat
ao eal
: ae ———
o B 5 3 7
Time (reel)
‘omnal leaning

Figure 13: Alliance learning - baserun

Figure 14 illustrates the behavior of ‘learning aptitude’ in greater detail. With an increasing
knowledge base, the ‘absorptive capacity’ increases exponentially whereas the ‘assessment of
new knowledge’ declines exponentially. However, the net increase of both multipliers is
positive. Consequently, the ‘learning aptitude’ grows. Growth slows down when the net
increase diminishes and it peaks when the net rate is zero. Afterwards, the ‘learning aptitude’
decreases as the ‘assessment of new knowledge’ becomes smaller. The ‘assessment of new
knowledge’ increases again because the “alliance knowledge base’ drops (here, after week
25). So, there would be things to (re-)learn again. According to Figure 12, ‘unlearning’
slightly increases due to an increasing ‘alliance knowledge base’ until approximately week 16
and then almost flattens and reduces the knowledge base constantly.

18
‘acer,
‘Leaning aptitide

oF
03 ara
02
on te |
: Ss
Nene
A — <<
ors
05
025
Assesment oflnarleige rae
075 ss
a \
025 Ko
° o 13 Boy Ey Ey

‘Tse (reel)

Figure 14: Learning aptitude - baserun

In the beginning of the alliance’s existence ‘interfirm belief in the alliance success’ or trust
drops slightly (see Figure 15). Even though there are both enthusiasm among the people
working in the alliance and belief in the alliance’s success, there are no outcomes being
generated. Thus intergroup trust slowly diminishes.

Intergroup belief in alliance success

4. Benefit points
08 Belief points
(0.2 Belief points/week}
‘4. Benefit points

2. Benefit points

04 Belief points
Belief points/weel}
2 Benefit points

0 Benefit points
0 Belief pints
-0.2 Belief points/week|
(Benefit points

o 5 10 15 0 5 3 35 a 45 30
‘Time (week)

Perceived alliance benefits: basen Benefit points
Intergroup belief in ellimce succes -baseru Belief points

Change in belief: baserun a Belief points/week
Actual eliance benefits: baserun. ‘Benefit points

Figure 15: Intergroup belief in alliance success - baserun

As soon as the alliance actually generates benefits, people recognize that they are on the right
track with their work. With growing certainty in the alliance’s purpose, ‘change in belief” (net
rate) rises and hence, ‘interfirm belief in alliance success’ builds up. The latter increases until
the parents decide to withdraw the resources. Trust diminishes exponentially after this
decision has been made. The alliance goals have been achieved, and further success is not
possible. Both parents are satisfied with the alliance’s outcomes. They have worked together
successfully. If they decide to work on another task together in the future, it is conceivable
that initial ‘interfirm trust’ will be greater than in the baserun due to this cooperatively well
accomplished task. This might be different in the other scenarios that are discussed
subsequently.

19
4.4.2.2 First Scenario: On the way, Parent 2 Finds the Alliance More Attractive Than
Originally Thought
In the first scenario, it is assumed that both companies start out with the same expectations of
the alliance outcomes. However, in week 7, parent 2 suddenly realizes that the alliance is of
greater importance than it had originally thought. This may be due to a change of the parent’s
telative scope, i.e., through the parent’s decision to expand the markets covered by the
alliance. As a result, the parent increases its ‘expected common benefits in respect to its
telative scope’ by more than 300% (modeled via a step function: 3+STEP (8, 7)). This is
entailed in Figure 16 by the sudden drop of parent 2’s ‘degree of goal attainment’ in week 7.

Resources parents | and 2

45 People
15 Dau
3 Benefit point|

3 People
1 Dani
2 Benefit poin|

15 People
05 Dual

1 Benefit point

O People

0 Danl

O Benefit point

‘Tie (reek)

"Resouces spent om alliance - parent 1" Let sce People
"Resouces spent on aliance - parent 2": Ictscenm People
"Degree of goal attainment -pazent 1" Ist scer - - Drwl
"Degree of goal attainment -pazent 2": st scer Dal
‘Actial alliance benefits: Iotzoe * . : Benefit points
Perceived allance benefit: Ist en ‘Benefit points

Figure 16: Resource spending — 1* scenario

According to parent 2’s resource allocation policy, more people should be working in the
alliance after week 7. Yet, there is no sudden increase in ‘resources spent on alliance — parent
2’ as people must be recruited for working on the task. As stated in section 4.4.1.2, this delay
lasts 4 weeks. It can be seen from the Figure 17 that after week 11 7+ 4), parent 2 further
increases its resource expenditures on the alliance (see Figure 17).

0 3 10 5 20 3 x 35 0 # ”
‘Tie (eek)

Resousce allocation “bass Peopleheck

Resousce allocation | Ist sce Peoplefeek

Resouice allocation 2 bare rae Peoplefweek

Resource allocation 2 It cena Peoplefeek

Figure 17: Resource allocation parent companies 1 and 2 — baserun and 1* scenario

20
While the ‘perceived alliance benefits’ approach parent 1’s expected common benefits, parent
1 slows down the increase of resource allocations according to the policies presented in Table
1. It can be seen from the Figure 17 that parent 1 increases its resources much more slowly
(here, from week 11 to week 20) than in the baserun. Parent 1 withdraws its resources after it
has reached its goals (week 19) slightly earlier as in the baserun as more people have worked
on the task, whereas parent 2’s resource allocation is still positive as it has only met
approximately 50% of its expectations. Nevertheless, even though the task has not been yet
accomplished, parent 2 also pulls out its resources (after the 22°° week). There is no longer a
partner with whom to learn.

The behaviors of ‘alliance learning’, ‘alliance unlearning’, and ‘alliance knowledge base’
resemble those of the baserun (see Figure 18). The same holds for ‘interpartner belief in
alliance success’ (see Figure 19). Both the “alliance knowledge base’ and the ‘actual alliance
benefits’ are insignificantly higher in the first scenario than in the baserun.

Leaming

20 Kronledge points
2 Kawrrledge pointfeek

40 People
4 Benefit point:

10 Knowledge point:
1. Knowledge pointshveel
20 People

2 Benefit points

0 Keosrledge points
0 Kaorrledge pointseek

0 People
0 Benefit point:
Tie (reek)

Alliance newledge base: Ist see Kaenwledge points
‘Alliance leaming Ist 2 Knorwledge pointshneek
‘Alliaee unl etzoe 2 2 2 Koewledge pointshreek
"Resouces spent on aliance - parent 1" Lot scenm People
Perceived alliance benefit: Ieteen Benefit poins

Figure 18: Learning — 1* scenario

Intergroup believe in alliance success

4 Benefit point
(8 Believe points
(02 Believe poinshreek| epeen vere rere rene ere

4 Benefit point ae

2 Benefit point
(04. Believe points
O Believe pointshreel
2 Benefit point

Benefit point
O Believe point:

-02 Believe poinshreek|
Benefit point

0 20 2 30 5 0 5 30

a
‘Tame (eek)

Pesceived alliance benefit: Ist zoe Benefit point
Intesgoup believe alliance zacees «Let 08 2 - Believe points
Change inbelief Ista - * * ‘Believe pointshreek
‘Actual alliance benefit: Ist zoe Benefit point:

Figure 19: Interfirm belief in alliance success — 1* scenario

Even though there are no considerable differences concerning the interfirm belief or the
learning process in the alliance between the baserun and the first scenario, it can be stated that

21
the change in the alliance’s importance for parent 2 is significant. Only 50% of its
expectations were accomplished — the outcome is unpleasant. Parent 1, however, is satisfied
as it has achieved more than originally expected.

4.4.2.3 Second Scenario: Parent 2 has Higher Expectations of Alliance Benefits Than
Parent 1
For the second scenario it is assumed that parent company 2 has a larger overlap of its firm’s
scope with that of the alliance from the founding of the alliance. Hence, parent 2’s relative
scope is larger than parent 1’s. So, the alliance’s success is of higher importance for this
company as the alliance covers a larger area of the company’s field of interest. It is assumed
that parent 2’s expected common benefits are twice as high as in the baserun. Consequently,
the ‘normal resource allocation’ is twice as high.

As can be seen from the Figure 20, parent 2 spends a lot more resources on the alliance right
from the beginning compared to the other runs. Due to this intensive resource spending parent
2 forces parent | also to spend more resources than in the other runs (see Figure 21).

Graph for Resource allocation 2

0 3 10 rc oo x 3 rn s cE
‘Time (eel)

‘Rerouce allocation : Let 08 Peoplehveeke

Reromice allocation 2: ed cen Peoplehveek

Reromice allocation 2 -bacem 2 * - Peoplehveek

Figure 20: Resource allocation, parent 2— 2°? scenario

22
Resources 1d&2 I 2nd scenario

0 3 10 15 a 5 Ey 35 “a5 0
‘Tine (week)

People
People
People
People
People
People

Figure 21: Resources spent on alliance, parents 1&2 —baserun, 1“ and 2" scenario

As there are far more people working on the alliance task, learning occurs faster. Hence, the
alliance generates benefits more quickly (see Figures 21 and 23). In the early phase, though,
as the benefits occur faster than in the other runs, ‘interfirm belief in alliance success’ builds
up faster than in the baserun and in the first scenario (see Figure 21). However, trust decreases
exponentially when parent 1 quits the alliance.

Intergroup belief in alliance success

0 3 10 5 2 3s x 3 7 s EF
‘Time (ree)

Intengroup belief i aliance success: bace + + Belief points

Integroup belie i aliance mice ste 22 2 = Belief point

Inexgroup belief aliance miccess 2d ce 2 2 2+ Belief points

Figure 22: Intergroup belief in alliance success — baserun, 1* and 2" scenario

As parent 1’s goals are accomplished (roughly week 17; see Figure 23), it withdraws its
resources immediately. By this time, parent 2 — due to its higher expectations — has only quite
achieved 50% ofits expected benefits. However, as there are still people working in the
alliance (mostly from parent 2), they keep on researching and learning and therefore still
generate benefits. Due to the fading ‘intergroup belief in alliance success’ further learning is
limited. Until the end of the alliance, further benefits will still be generated. The final degree
of goal attainment is roughly 60%. Consequently, parent 1 would call the alliance as a success
whereas parent 2 thinks of it as failed and unsuccessful.

23
Degree of gaol attainment - 2nd scenario

4 Benefit points
2 Dam
@ People

3 Benefit points
15 Daal
45 People

2. Benefit points
1 Dam
30 People

1. Benefit points
05 Dimi

15 People

Perceived alliamee benefit: 2nd soem

Benefit points
"Degnee of goal attainment - pazent I”: 2nd Daal
“Degnee of goal attainment -pazent 2° 2nd se Daal
“Resouces spent om abance- pazent "Dade People
“Resouces spent om allance- pazent 2" 2nd People

Figure 23: Degree of goal attainment and resources spent on alliance, both parents — one
scenario

4.4.2.4 Third Scenario: Both Companies Increase their Common Expected Benefits at
the Same Time

The situation looks different if both parents change similarly regarding the expected common
benefits. In the third scenario, for example, both parent companies increase their expected
common benefits by less than 100% in week 5 (STEP function: 3+STEP(2, 5)). This joint
decision may be a consequence of coordination efforts initiated by either one or both parents
and it may result from an enhancement of the alliance’s scope. As can be seen from the Figure
24, resource expenditures are made jointly with each parent spending more people on the
alliance (here more than 60 people). The mode of behavior is similar to those described
above. Yet the time horizon is much broader as there are more benefits to accrue. The
expected common benefits are attained, when the ‘degree of goal attainment’ reaches 1
(roughly week 60). This leads to the termination of the alliance, with both parents being
satisfied and reassured that the alliance was a good means to reach their common goals.

Resources parents I and 2 - 3rd scenario

80 People
2 Daud
6 Benes points
1 Belief points
40 People
1 Dani
3 Benefit points
05 Belief points
O People
0 Dau
0 Benedt points
O Beef point:
os W 1 0 3 0 3 0 4 5 5S @ 6 0 75 90 88 90 98 100
‘Tame (eek)
People
People
Daal
Daal
Benefit points
Integroup belief in alliance miccezs «Sud sce Belief points

Figure 24: Resources spent on alliance and degree of goal attainment — 3“ scenario

24
In this scenario, the graph for ‘intergroup belief in alliance success’ is worth mentioning. It
can be seen from the Figure 25 that ‘intergroup belief in alliance success’ builds up in an s-
shaped curve until it nearly reaches full belief. Due to the fact that the people work together
successfully over such a long period, they trust each other increasingly over time. As in the
runs described above, after the alliance goals have been reached and the resources withdrawn,
‘intergroup belief’ also declines. However, the good experience would be a good basis for
further cooperation.

Intergroup belief in alliance success

07s

os WwW is 2 25 5 a5 40 4 5 35 60 65 0 75 80 8 90 95 100
‘Time (mee)

Intergroup beliefin alliance success bar + Belief points
Intergroup belefin alliance success: Ist 2 Belief points
Intergroup beliefin alliance succes 2nd 08 ar 2 2 Belief points
Intergroup beliefin alliance sicess : 3rd 208 Belief points

Figure 25: Intergroup belief — baserun, 1*, 24 and 3" scenario

Compared to the first and second scenarios, it was of great advantage for the parents to act
cooperatively and convince the partner to continue with the alliance and also increase the
expectations at the same time. This way, they both ensure that the alliance is equally
important to both of them

5. Concluding Discussion

The paper begins with a definition of learning alliances that stand in the focus of the
investigation. I reviewed the relevant learning alliance literature for examining the dynamics
of common learning in learning alliances. Then I described System Dynamics as the method
used to conduct the investigation. Afterwards I presented a specific System Dynamics model
that investigates the dynamics of common learning. In this section I discuss the implications
of the results obtained using the System Dynamics model on learning alliances described in
the previous section.

Regarding the influence of the evaluation of the importance of the alliance for the parent
companies the following can be stated.

In the baserun both parent companies’ expected common benefits are identical from the
founding of the alliance. This implies that they both see the alliance as equally important to
them. The parents terminate the alliance at the same time, namely, when their expected
common benefits are attained. Both parents are satisfied with the outcome of the alliance. As
the joint learning went well it can be assumed that they would be open to enter another
learning alliance with each other.

25
In the first scenario both parents start out with the same expectations of common alliance
benefits. The learning alliance progresses in a way that is rewarding for both parents
Suddenly the second parent increases its expectations. This leads to unbalanced expectations
the alliance becomes more important for the second parent than for the fist. As there are less
expectations for the first parent, its expectations are met first during the course of the alliance.
There are no more incentives for it to stay in the alliance and thus, it withdraws its resources
from the alliance. As a result, the second parent is only able to realize a small portion of its
expectations. Therefore, even though the parent has drawn so much attention to the alliance
its outcome can be regarded as disappointing.

In the second scenario, the imbalance of the parent companies’ expected common benefits
exists from the very beginning of the learning alliance’s existence. One parent has higher
expectations of common benefits than its partner and therefore, it evaluates the alliance as
more important. This might be due to a larger overlap of the alliance scope with the firm
scope. Like in the first scenario, the parent with the lower expectations of common benefits
terminates the learning alliance before the other parent achieves its common benefits. Again,
the alliance was useful for the parent with less expectations. Hence, if the parents’ expected
alliance benefits differ from each other, the parent more interested in the alliance success
might be disappointed with the overall outcome.

Khanna et al. (1998) point out that there are no incentives for a partner to stay in the alliance
when there are no more private benefits to accrue (Khanna et al., 1998: 198). However, from
the first and the second scenario it can be stated that also in a situation of only common
benefits there is no incentive for a parent to stay in the alliance once its expected common
benefits are achieved. So, the parent terminates the alliance before the other parent company
has finished common learning.

In the third scenario, it is assumed that both parent companies start out with the same
expectations and then increase them at the same time about the same quantity. This means
that both parents evaluate the alliance as more important on the course of the alliance’s
existence. In this scenario it is shown that both parent companies continue to work jointly on
the task until they both meet their expectations. The coordination of the expectations is of
significant value for the common alliance success.

Overall, it can be summarized that both parent companies have to evaluate the learning
alliance equally important for it to be successful for both of them. If the parents consider the
learning alliance unequally important the alliance turns out to be unsatisfactory and
disappointing for at least one of the parents. So, when engaging in a learning alliance, a
company should firstly ensure that the partner’s expectations are neither significantly higher
nor lower than its own. Secondly, it is essential for the alliance success that the parent
companies communicate permanently during the lifetime of the alliance in order to
continuously balance the each other’s expectations.

6. References

Aloysius, J.A. 2002. Research Joint Ventures: A Cooperative Game for Competitors.
European Journal of Operational Research, 136: 591 - 602

Amburgey, T.L., Dacin, T., and Singh, J.V. 2000. Learning Races and Dynamic Capabilites.
[online] Working Paper. Available from World Wide Web:
(www .rotman.utoronto.ca/bicpapers/pdf/2000-13.pdf)

26
Anand, B.N. and Khanna, T. 2000. Do Firms Learn to Create Value? The Case of Alliances
Strategic Management Journal, 21: 295 — 315

Argyris, C. and Schén, D. 1978. Organizational Learning. Addison Wesley: Reading, MA

Bea, F.X. and Haas, J. 2001. Strategisches Management. 3°! Edition. Lucius & Lucius
Verlagsgesellschaft: Stuttgart

Beckman, Chr.M. and Haunschild, P.R. 2002. Network Learning: The Effects of Partners’
Heterogeneity of Experience on Corporate Acquisitions. Administrative Science
Quarterly, 47: 92 - 124

Belderbos, R. 2003. Entry Mode, Organizational Learning, and R&D in Foreign Affiliates:
Evidence from Japanese Firms. Strategic Management Journal, 24: 235 - 259

Bettis, R.A., Prahalad, C.K. 1995. The Dominant Logic: Retrospective and Extension.
Strategic Management Journal, 16: 5 - 14

Carlile, P.R. 2002. Transferring, Translating, and Transforming: An Integrative Framework
for Managing Knowledge across boundaries. MIT Sloan School Working Paper.
Submitted to Administrative Science Quarterly

Child, J. 2001. Learning Through Strategic Alliances. In Dierkes, M., Antal, A., Child, J.,
Nonaka, I. (Eds.). Handbook of Organizational Learning and Knowledge.
University Press: Oxford, 657 - 680

Child, J. and Faulkner, D. 1998. Strategies of Cooperation. Managing Alliances, Networks,
and Joint Ventures. Oxford University Press: Oxford

Cohen, W.M. and Levinthal, D.A. 1990. Absorptive Capacity: A New Perspective on
Learning and Innovation. Administrative Science Quarterly, 35: 128 - 152

Curral, S.C. and Inkpen, A.C. 2002. A Multilevel Approach to Trust in Joint Ventures.
Journal of International Business Studies, 33: 479 - 495

Dengel, H. and Milling, P. 2002. Aspects of the Dynamics of Cooperations in a Supply Chain
Under Consideration of Trust. Proceedings of the 2002 International System
Dynamics Conference. System Dynamics Society: Albany, NY

Doz, Y.L. and Hamel, G. 1998. Alliance Advantage: The Art of Creating Value Through
Partnering. Harvard Business School Press: Boston, MA

Doz, Y.L. 1996. The Evolution of Cooperation in Strategic Alliances: Initial Conditions or
Learning Process. Strategic Management Journal, 17: 55 - 83

Eisenhardt, K.M. and Martin, J.A. 2000. Dynamic Capabilities: What are they? Strategic
Management Journal, 21: 1105 - 1121

Forrester, J.W. 1961. Industrial Dynamics. MIT Press: Cambridge, MA

Gulati, R. 1995. Does Familiarity Breed Trust? The Implications of Repeated Ties for
Contratual Choice in Alliances. Academy of Management Journal, 38: 85 - 112

Goussevskaja, A. and Kidd, J. 2002. Learning in an Inter-organizational Context. In
Proceedings of British Academy of Management Annual Conference: British
Academy of Management: London

Grossman, G.M. and Shapiro, C. 1986. Optimal Dynamic R&D Programs. RAND Journal of
Economics, 17: 581 - 593

Grossman, G.M. and Shapiro, C. 1987. Dynamic R&D Competition. The Economic Journal,
97: 372 — 387

Giildenberg, S. 1996. Wissensmanagement und Wissenscontrolling in lernenden
Organisationen: Ein systemtheoretischer Ansatz. Deutscher Universitats-Verlag:
Wiesbaden

Hagedom, J., Link, A.N., and Vonortas, N.S. 2000. Research Partnerships. Research Policy,
29: 567 - 586

Hamel, G. 1991. Competition for Competence and Interpartner Learning Within International
Strategic Alliances. Strategic Management Journal, 12: 83 - 103

27
Hedberg, B. 1981. How Organizations Learn and Unlearn. In Nystrom. P.C. and Starbuck,
W.H. (Eds). Handbook of Organizational Design, Volume 1. Adapting
Organizations to Their Environments. Oxford University Press: London, 3 - 27

Ho, T.H. and Weigelt, K. 1997. Game Theory and Competitive Strategy. In Day, G.S. and
Reibstein, D.J. (Eds.) Wharton on Dynamic Competitive Strategy. John Wiley &
Sons: Hoboken, NJ, 127 - 150

Ho, T.H. and Weigelt, K. 1999. Trust Building Among Strangers. Working Paper WP 99-01
of the Reginald H. Jones Center, The Wharton School, University of Pennsylvania

Inkpen, A.C. 2000a. Learning Through Joint Ventures: A Framework of Knowledge
Acquisition. Journal of Management Studies, 37: 1019 — 1043

Inkpen, A.C. 2000b. A Note on the Dynamics of Learning Alliances: Competition,
Cooperation, and Relative Scope. Strategic Management Journal, 21: 775 - 779

Inkpen, A.C. and Crossan, M.M. 1995. Believing is Seeing: Joint Ventures and
Organizational Learning. Journal of Management Studies, 32: 595 - 618

Kale, P., Dyer, J.H., and Singh, H. 2002. Alliance Capability, Stock Market Response, And
Long Term Alliance Success: The Role of the Alliance Function. Strategic
Management Journal, 23: 747 — 767

Kapmeier, F. 2002. Dynamics of Interorganizational Learning. Proceedings of the 2002
International System Dynamics Conference. System Dynamics Society: Albany,
NY

Khanna, T. 1998. The Scope of Alliances. Organization Science, 9: 340 - 355

Khanna, T., Gulati, R. and Nohria, N. 1998. The Dynamics of Learning Alliances:
Competition, Cooperation, and Relative Scope. Strategic Management Journal,
19: 193 -210

Kumar, R. and Nti, K.O. 1998. Differential Learning and Interaction in Alliance Dynamics: A
Process and Outcome Discrepancy Model. Organization Science, 9: 356 — 367

Lane, P.J., Salk, J.E., and Lyles, M.A. 2001. Absorptive Capacity, Learning, and Performance
in International Joint Ventures. Strategic Management Journal, 22: 1139 - 1161

Lane, P.J. and Lubatkin, M. 1998. Relative Absorptive Capacity and Interorganizational
Learning. Strategic Management Journal, 19: 461 - 477

Larsson, R., Bengtsson, L., Henriksson, K., and Sparks, J. 1998. The Interorganizational
Learning Dilemma: Collective Knowledge Development in Strategic Alliances.
Organization Science, 9: 285 - 305

Lubatkin, M., Florin, J., and Lane, P. 2001. Learning Together and Apart: A Model of
Reciprocal Interfirm Learning. Human Relations, 54: 1353 - 1382

Luhmann, N. 1994. Soziale Systeme. Grundrif einer allgemeinen Theorie, 2" edition.
Suhrkamp Verlag: Frankfurt a.M.

March, J.G. 1991. Exploration and Exploitation in Organizational Learning. Organization
Science, 2: 71 - 87

McAllister, D.J. 1995. Affect- and Cognition-Based Trust as Foundations for Interpersonal
Cooperations in Organizations. Academy of Management Journal, 38: 24 - 59

Nonaka, I., and Takeuchi, H. 1997. Die Organisation des Wissens. Wie japanische
Unternehmen eine brachliegende Ressource nutzbar machen. Campus Verlag:
Frankfurt a.M

Nonaka, I., Toyama, R., and Byosiére, P. 2001. A Theory of Organizational Knowledge
Creation: Understanding the Dynamic Processes of Knowledge Creation. In
Dierkes, M., Antal, A.B., Child, J., and Nonaka, I. (Eds.) Handbook of
Organizational Learning and Knowledge. Oxford University Press: New York,
NY, 491 - 517

Pausenberger, E. 1989. Zur Systematik von Unternehmenszusammenschliissen. Das
Wirtschaftsstudium, 18: 621 - 626

28
Penrose, E. 1959. The Theory of the Growth of the Firm. Oxford University Press: Oxford

Phan, P.H. and Peridis, T. 2000. Knowledge Creation in Strategic Alliances: Another Look at
Organizational Learning. Asia Pacific Journal of Management, 17: 201 - 222

Polanyi, M. 1966. The Tacit Dimension. Routledge & Kegan Paul Ltd.: London

Picot, A., Dietl, H. 1990. Transaktionskostentheorie. Wissenschaftliches Studium, 19: 178 -
184

Prange, Chr. 1996. Interorganisationales Lernen: Lernen in, von und zwischen
Organisationen. In Managementforschung 6, Schreyégg, G. and Conrad, P. (Eds)
de Gruyter: Berlin, 163 — 189

Ring, P. and Van de Ven, A. 1992. Structuring Cooperative Relationships Between
Organizations. Strategic Management Journal, 13: 438 - 498

Simonin, B. 1999. Ambiguity and the Knowledge Transfer in Strategic Alliances. Strategic
Management Journal, 20: 595 - 623

Spender, J.C. 1996. Organizational Knowledge, Learning and Memory: Three Concepts in
Search of a Theory. Journal of Organizational Change Management, 9: 63 - 78

Spender, J.C. and Grant, R.M. 1996. Knowledge of the Firm. Strategic Management Journal,
17:5-9

Sterman, J.D. 2002. All Models are Wrong: Reflections on Becoming a Systems Scientist.
System Dynamics Review, 18: 501 - 531

Sterman, J.D. 2000. Business Dynamics: Systems Thinking and Modeling for a Complex
World. Inwin/McGraw-Hill: New York

Sterman, J.D. 1991. A Skeptic’s Guide to Computer Models. [online] Working Paper D-4101-
1, MIT Sloan School of Management, Boston, MA. Available from World Wide
Web: (ftp://sysdyn.mit.edu/ftp/sdep/Roadmaps/RM9/D-4101-1.pdf). Quoting
Forrester, J.W. 1980 Information Sources for Modeling the National Economy.
Journal of the American Statistical Association, 75: 555 - 574

Teece, D.J. and Pisano, G.P. 1994. The Dynamic Capabilities of Firms: An Introduction. In
Teece, D.J. Economic Performance and the Theory of the Firm. Collected Papers
of David J. Teece. 1998. Edward Elgar Publishing: Glos, UK, 56 - 75

Teece, D.J., Pisano, G.P., and Shuen, A. 1997. Dynamic Capabilities and Strategic
Management. Strategic Management Journal, 18: 509 - 533

Tsang, E.W.K. 2002. Acquiring Knowledge by Foreign Partners from International Joint
Ventures in a Transition Economy: Learning-by-Doing and Learning Myopia.
Strategic Management Journal, 23: 835 - 854

Zahn, E. and Tilebein, M. 2000. Lernprozesse in Organisationen. In Welge, M., Haring, K.,
Voss, A. (Eds.) Management-Development: Praxis, Trends und Perspektiven.
Schaffer-Poeschel: Stuttgart, 117-137

Zahn, E. 2001. Lernen in Allianzen. In Bellmann, K. (Eds.) Kooperations- und
Netzwerkmanagement. Duncker & Humblot: Berlin, 11 - 30

Zucker, L.G. 1986. Production of Trust: Institutional Sources of Economic Structure 1840 -
1920. In Staw, B.M. and Cummings, L.L. (Eds.) Research in Organizational
Behavior, Volume 8. JAI Press: Greenwich, CT, 53 - 111

29
7. Appendix

Model
(1)

(02)

(03)

(04)

(03)

(06)

Absorptive capacity= WITH LOOKUP (Relative knowledge base,

({(0,0)-(1,1)],(0,0.07456),(0.08869,0.1053),(0.208,0.1623),(0.3242,0.2675

),(0.4098,0.3684),(0.4985,0.5439),(0.5657,0.7018),(0.605505,0.785088),(0.654434
,0.855263),(0.718654,0.903509),(0.789,0.9474),(0.902141,0.986842),(1,1) ))
Units: Dmnl
Absorptive capacity is the ability to recognize, learn and use
new external knowledge (Cohan, Levinthal 1990). It is assumed
that the absorptive capacity increases in an s-shaped way (table
function). The higher the Relative knowledge base, the higher
the capacity to absorp new knowledge.\!Relative knowledge base
(dimensionless)\!

Actual alliance benefits= INTEG (Knowledge application to products or patents,0)
Units: Benefit points
The alliance benefits are accumulated and have no outflow. It is

measured in benefit points. The alliance benefits are the

benefits that are being generated through the alliance itself.

Its initial value is 1

Alliance knowledge base= INTEG (Alliance learning-Alliance unlearning,2)

Units: Knowledge points

The employees' knowledge base is measured in knowledge points.
The initial value of the stock is 2. The stock is increased by
learning, called alliance learning. The stock is decreased by
the outflow called alliance unlearning.

Alliance learning=
Commitment to learning through belief in alliance success*Learning aptitude

*Common resource spending*Normal learning

Units: Knowledge points / week

Alliance learning increases the stock Alliance knowledge base
It is determined by the Resources spent on the alliance times
the Learning commitment, and the Additional learning aptitude.
So, when people and money are involved, people learn on the
alliance tasks. If not, people have other things to do and do
research on different things next to the alliance. Alliance
learning - not the Alliance knowledge base is important for the
Knowledge application to products. This is due to the fact that
this new knowledge is being generated on the basis of the old
knowledge. So, new knowledge levers to new products. (patent
application)

Alliance unlearning=Alliance knowledge base*Normal unlearning rate
Units: Knowledge points / week
This rate decreases the stock Alliance knowledge base. (old

patents that are not useful anymore)

Assessment of knowledge value= WITH LOOKUP (Relative knowledge base,
({(0,0)-(1,1)],(0,1),(0.140673,0.973684),(0.214067,0.95614),(0.278287,0.925439
),(0.318043,0.894737),(0.351682,0.855263),(0.382263,0.820175),(0.412844,0.776316
),(0.440367,0.72807),(0.464832,0.671053),(0.492355,0.596491),(0.51682,0.526316
),(0.538226,0.456 14),(0.556575,0.372807),(0.577982,0.302632),(0.602446,0.25
),(0.633027,0.2),(0.66055,0.17),(0.691131,0.14),(0.724771,0.11),(0.761468,
0.0877193),(0.795107,0.0657895),(0.82263,0.0482456),(0.868502,0.0263158),(

30
0.911315,0.0131579),(1,0) ))

Units: Dmnl

The assessment of knowledge value is a table-function (look-up)
that states that, the more I know, the less I can learn in
addition to what I already know in a specific context. The graph
looks like s-shaped decline.\!\!\!

(07) Belief Gap=Belief quantity-Intergroup belief in alliance success
Units: Belief points
The Gap computes the difference between the Believing quantity
and the Intergroup believe in alliance success.

(08) Belief quantity=Effect of goal attainment on belief in alliance success*Normal belief
Units: Belief points
The Believing quantity is the result of the multiplication of
the dimensionless variable Effect of goal attainment on
believing in alliance success and the Normal believing.

(09) Benefits gap=Actual alliance benefits-Perceived alliance benefits
Units: Benefit points
The Gap computes the difference between the Perceived alliance
benefits and the Actual alliance benefits.

(10) Change in belief=IF THEN ELSE("Degree of goal attainment - parent 1">1, -Intergroup belief
in alliance success
/Time to form belief, Belief Gap/Time to form belief)
Units: Belief points/week
The biflow Change in believe increases or decreases the stock
Intergroup believe in alliance success. It increases the stock
according to the Updated perceived alliance benefits. The higher
the benefits, the higher the change in believe. Once one company
has reached its goals and withdraws its resources from the
alliance, trust between the groups diminishes.

(11) Commitment to learning through belief in alliance success=
Intergroup belief in alliance success/Normal belief
Units: Dmnl
Learning commitment is a variable that results from the
multiplication of the Normal learning and the Relative believe
in alliance success. (in German: Wollen-lemmen)

(12) | Common resource spending=
"Resources spent on alliance - parent 1"+"Resources spent on alliance - parent 2"
Units: People
Common resource spending is a variable that counts together the
Resources spent on the alliance from both partner 1 and 2.

(13) "Degree of goal attainment - alliance"= Perceived alliance benefits/Expected alliance benefits
Units: Dmnl

(14) "Degree of goal attainment - parent 1"=
Perceived alliance benefits/Expected common benefits 1 in respect tol's relative scope
Units: Dmnl
As more stages of the research and development project are
completed, less effort is needed to finish the stages and earn
the same benefits. The firm steps up its allocation every period

31
(15)

on the way to close the gap. Khanna et al. (1998), p. 198. The
more milestones have been reached, the less resources are being
allocated on the way to close the gap.\!Zielerreichungsgrad\!

"Degree of goal attainment - parent 2"= Perceived alliance benefits/Expected common benefits

2 in respect to 2's relative scope

(16)

Units: Dmnl

As more stages of the research and development project are
completed, less effort is needed to finish the stages and earn
the same benefits. The firm steps up its allocation every period
on the way to close the gap. Khanna et al. (1998), p. 198. The
more milestones have been reached, the less resources are being
allocated on the way to close the gap.\!Zielerreichungsgrad\!

Effect of degree of goal attainment on resource allocation 1= WITH LOOKUP
("Degree of goal attainment - parent 1",([(0,0)-

5,1)].(0,1),(0.0550459,0.97807),(0.100917,0.951754),(0.146789,0.872807

a7)

),(0.17737,0.75),(0.211009,0.557018),(0.250765,0.394737),(0.287462,0.27193
),(0.351682,0.157895),(0.41896,0.09649 12),(0.501529,0.0614035),(0.688073,0.0175439
).(1,0).(2,09.(5.0) ))
Units: Dmnl
With a small Zielerreichungsgrad, the company spends lots of

resources on the alliance in order to close the gap as fast as

possible. The closer the company gets to the expected benefits,

the less additional resources the company spends on the

alliance. \!Zielerreichungsgrad\!

Effect of degree of goal attainment on resource allocation 2= WITH LOOKUP
("Degree of goal attainment - parent 2",([(0,0)-

(5,1)],(0,1),(0.0550459,0.97807),(0.100917,0.951754),(0.146789,0.872807

(18)

(19)

),(0.17737,0.75),(0.211009,0.557018),(0.250765,0.394737),(0.287462,0.27193
),(0.351682,0.157895),(0.41896,0.09649 12),(0.501529,0.0614035),(0.688073,0.0175439
),1,0).(5,0) ))
Units: Dmnl
With a small 'degree of goal attainment’ the company spends more

resources on the alliance in order to reach the goal as fast as

possible. The closer the 'perceived benefits’ get to the

‘expected benefits’, the less additional resources the parent

spends on the alliance.

Effect of goal attainment on belief in alliance success= WITH LOOKUP (
"Degree of goal attainment - alliance",
({(0,0)-(1,1)].(0,0.2),(0.140673,0.241228),(0.287462,0.307018),(0.409786
,0.416667),(0.495413,0.548246),(0.587156,0.710526),(0.69419,0.850877),(0.810398
,0.934211),(0.905199,0.97807),(1,1) ))
Units: Dmnl
With a small 'degree of goal attainment' the company spends more
resources on the alliance in order to reach the goal as fast as
possible. The closer the 'perceived benefits' get to the
‘expected benefits’, the less additional resources the parent
spends on the alliance.\!Effect of goal attainment on believing
in alliance success\!

Expected alliance benefits=4
Units: Benefit points

32
(20)

(21)

(24)

(25)

Expected common benefits 1 in respect tol's relative scope=3
Units: Benefit points
The expected common benefits is a parameter that is determined

by the companie's relative scope. Common benefits are benefits
from an alliance that a firm can apply to businesses within the
scope of the alliance. They differ from alliance benefits that
refer to the outcome that the alliance generates itself. The
higher the expected common benefits, the more important the
alliance is for the long-term success of the company. Here, the
expected common benefits are constant at 15 benefit points.

Expected common benefits 2 in respect to 2's relative scope=3
Units: Benefit points
The expected common benefits is a parameter that is determined

higher the expected common benefits, the more important the
alliance is for the long-term success of the company. In the
baserun, the expected common benefits are constant at 15 benefit
points.

FINAL TIME = 50
Units: week
The final time for the simulation.

INITIAL TIME =0
Units: week
The initial time for the simulation.

Intergroup belief in alliance success= INTEG (Change in belief, 0.1)
Units: Belief points
The Intergroup believe in the alliance success is a biflow that

can be either in- or decreased by the flow called 'change in
believe'. Intergroup believe in alliance success describes the
attachment of people who work in the alliance towards each
other. The attachment is purely formed through the perceived
alliance benefits. This means that perceived progress
strengthens the people working in the alliance in terms of
dedication to their work. The parent companies' expected
alliance benefits in respect to their relative scope does not
influence the intergroup believe in the alliance success. It is
assumed that the people working in the alliance dedicate their
time and energy only on their research findings and not on.
coordinating the degree of their parent companies’ goal
attainment

Knowledge application to products or patents=

MAX( Alliance learning*Success rate of new ideas, 0)
Units: Benefit points / week
What was learnt by the alliance (Alliance learning) is being

applied to products developed by the alliance. Not all what is
newly learnt is useful for the development of new products.
Therefore, a transformer is needed to explain how many benefits
(i.e. new patents or new products) are being generated (or:

33
(26)

(27)

(28)

(29)

(30)

(32)

(33)

(34)

successful). This flow increases the stock Actual alliance
benefits.

Learning aptitude=Assessment of knowledge value*Absorptive capacity

Units: Dmnl

The Learning aptitude is a multiplier of the Assessment of
knowledge value and the Absorptive capacity. It is dimensionless
and influences how well people in the alliance learn.
(K6énnen-lernen)

Maximum knowledge base=15
Units: Knowledge points
As no quantity can grow forever there is an upper limit to the
knowledge base. Here, it is appointed to 50 knowledge points.

Normal belief=1

Units: Belief points

Normal believing is the maximum believing or trust in something.
Here, it is assumed to be 1.

Normal learning=0.5

Units: Knowledge points/People/week

Normal learning is a constant that tells us something about how
well people learn. It is assumed that one person can generate
0.5 knowledge points per week.

Normal resource allocation 1=10

Units: People

Under normal circumstances, which means when there is a mimimum
Degree of goal attainment the parent allocates maximum
resources. Here, the normal resource allocation is 10 people.

Normal resource allocation 2=10

Units: People

Under normal circumstances, which means when there is a mimimum
Degree of goal attainment the parent allocates maximum
resources. Here, the normal resource allocation is 10 people.

Normal unlearning rate=0.01

Units: 1/week

The parameter Normal unlearning rate refers to the phenomenon
that people are forgetting things over a certain period of time.
Here, it is assumed that people forget 1% of their knowledge per
week.

Perceived alliance benefits= INTEG (Updated perceived alliance benefits,0)

Units: Benefit points

The benefits being perceived are time lagged, either through
delays in the perception itself or through the time needed a
product is successful in the market. At any case, there is a
certain time needed to perceive the Actual alliance benefits.
The model construction equals a smooth. The stock can be
increases and decreased by the flow Updated perceived alliance
benefits. Its initial value is 1

Relative knowledge base=Alliance knowledge base/Maximum knowledge base

34
Units: Dmnl

The relative knowledge base is a normalizing variable. It is
dimensionless and is calculated by dividing the actual Alliance
knowledge base by the Maximum knowledge base.

(35) Resource allocation 1= IF THEN ELSE (Resource allocation from goal attainment 1=0, -
"Resources spent on alliance - parent 1"
/Time for allocation decision 1

, DELAY3(((Resource allocation from goal attainment 1)+(Resource allocation from

allocation difference 1
)/2)/Time for allocation decision 1
, Time for allocation decision 1))

Units: People/week

Resource allocation is a flow that increases the stock Resources
spent on the alliance. It is determined by the Perceived gap
divided by the Duration of the allocation decision times the
People per Benefit point. There is a 3rd-order delay involved in
this process. There is only people spent on the alliance if the
Perceived gap is positive. If it is negative (overachievement of
the expected benefits), there are no further resources spent on
the alliance.

(36) Resource allocation 2=IF THEN ELSE (Resource allocation from goal attainment 2=0, -
"Resources spent on alliance - parent 2"
/Time for allocation decision 2, DELAY3(((Resource allocation from goal attainment 2
)+(Resource allocation from allocation difference 2)/2)/Time for allocation decision 2
, Time for allocation decision 2 ))
Units: People/week
Resource allocation is a flow that increases the stock Resources
spent on the alliance. It is determined by the Perceived gap
divided by the Duration of the allocation decision times the
People per Benefit point. There is a 3rd-order delay involved in
this process. There is only people spent on the alliance if the
Perceived gap is positive. If it is negative (overachievement of
the expected benefits, there are no further resources spent on
the alliance.

(37) | Resource allocation from allocation difference 1="Resources spent on alliance - parent 2"-
"Resources spent on alliance - parent 1"
Units: People
Ina situation of pure common benefits, both partners act like
if they were one firm concerning the resource allocation. This
means that the partners looks how much the other one invests and
depending on that varies the own resource allocation. (Khanna et
al 1998)

(38) Resource allocation from allocation difference 2="Resources spent on alliance - parent 1"-
"Resources spent on alliance - parent 2"
Units: People
Ina situation of pure common benefits, both partners act like
if they were one firm concerning the resource allocation. This
means that the partners looks how much the other one invests and
depending on that varies the own resource allocation. (Khanna et
al 1998)

35
(39) Resource allocation from goal attainment 1=Effect of degree of goal attainment on resource
allocation 1*Normal resource allocation 1
Units: People
A small effect of goal attainment allocates only little
resources to the alliance. On the other hand, a large effect of
goal attainment allocates more resources to the alliance. The
effect of Zielerreichungsgrad (dimensionless) is multiplicated
by the normal resource allocation (people).

(40) Resource allocation from goal attainment 2=Effect of degree of goal attainment on resource
allocation 2*Normal resource allocation 2
Units: People
A small effect of Zielerreichungsgrad allocates only little
resources to the alliance. On the other hand, a large effect of
Zielerreichungsgrad allocates more resources to the alliance.
The effect of Zielerreichungsgrad (dimensionless) is
multiplicated by the normal resource allocation (people).

(41) "Resources spent on alliance - parent 1"= INTEG (Resource allocation 1,1)

Units: People

Resources in this context refer to both human and financial
resources. When resources are spent on the alliance, it means
that the resources acually work on the alliance tasks. After the
withdrawal they don't work on the alliance tasks anymore. The
initial value of this stock is 0.1 and the stock is increases by
the inflow Resource allocation and decreases by the outflow
Resource withdrawl.

(42) "Resources spent on alliance - parent 2"= INTEG (Resource allocation 2,1)

Units: People

(43) SAVEPER = TIME STEP
Units: week [0,7]
The frequency with which output is stored.

(44) Success rate of new ideas=0.25
Units: Benefit points/Knowledge points
As not everything people learn is useful in terms of so-called
benefits (or: successful). Here, 10 % of the new knowledge is
useful and can be transformed into benefits

(45) Time for allocation decision 1=4
Units: week
It takes the managers some time to think about the investment or
deinvestment of resources. It takes the managers 5 weeks to
decide on this topic on average.

(46) Time for allocation decision 2=4
Units: week

36
It takes the managers some time to think about the investment or
deinvestment of resources. It takes the managers 5 weeks to
decide on this topic on average.

(47) TIME STEP = 0.0625
Units: week [0,?]
The time step for the simulation.

(48) Time to form belief=5
Units: week
Some time is needed to form the believe into the alliance. It is
assumed that the time to form a believe is constant at 10 weeks.

(49) Time to form expectations=10
Units: week

(50) Time to perceive alliance benefits=2
Units: week
The time to perceive the Actual alliance benefits is constant at 5 weeks.

(51) | Updated perceived alliance benefits=Benefits gap/Time to perceive alliance benefits
Units: Benefit points/week
The variable Updated perceived alliance benefits is a biflow
that may increase or decrease the stock Perceived alliance
benefits. It is determined by the gap over the time to perceive
alliance benefits.

Input for runs

Baserun:

Normal resource allocation 1 = 10
Units: People

Expected common benefits 1 in respect tol's relative scope = 3
Units: Benefit points

Expected common benefits 2 in respect to 2's relative scope = 3
Units: Benefit points

Expected alliance benefits = 4
Units: Benefit points

1* scenario:

Normal resource allocation 1 = 10
Units: People

Normal resource allocation 2 = 10
Units: People

Expected common benefits 1 in respect tol's relative scope = 3
Units: Benefit points

Expected common benefits 2 in respect to 2's relative scope = 3+STEP( 8 , 7)
Units: Benefit points

Expected alliance benefits = 4
Units: Benefit points

2" scenario:
Normal resource allocation 1 = 10

37
Units: People

Normal resource allocation 2 = 20
Units: People

Expected common benefits 1 in respect tol's relative scope = 3
Units: Benefit points

Expected common benefits 2 in respect to 2's relative scope = 6
Units: Benefit points

Expected alliance benefits = 4
Units: Benefit points

3" scenario:

Normal resource allocation 1 = 17
Units: People

Expected common benefits 1 in respect tol's relative scope = 3+STEP(2, 5)
Units: Benefit points

Expected common benefits 2 in respect to 2's relative scope = 3+STEP(2, 5)
Units: Benefit points

Expected alliance benefits = 4+STEP(2, 5)

Units: Benefit points

"In the following, due to reasons of simplification, | only refer to alliances even though the cited
authors might refer to alliances or joint ventures.

" Grossman and Shapiro (1987) state that different allocation patterns exist besides the pure progress
effect. However, this pattern is the one that they observed most often (Grossman and Shapiro, 1987:
378).

" In the model, it is assumed that parent 1 is always the parent who withdrawals its people from the
alliance first. Therefore, in the scenarios, parent 2 is the parent whose expectations are being
changed.

Back to the

Metadata

Resource Type:: Document
Rights:: CC BY-NC-SA 4.0
Date Uploaded:: December 30, 2019

Using these materials

Access:: The archives are open to the public and anyone is welcome to visit and view the collections.

Collection restrictions:: Access to this collection is unrestricted unless otherwide denoted.
Collection terms of access:: https://creativecommons.org/licenses/by/4.0/

About the Archives

Access options

Ask an Archivist

Ask a question or schedule an individualized meeting to discuss archival materials and potential research needs.

Ask a Question Book a Meet

Schedule a Visit

Archival materials can be viewed in-person in our reading room. We recommend making an appointment to ensure materials are available when you arrive.

Schedule a Visit Hours

Collection

System Dynamic Society Records, 1978-2016

Collection ID	ua435
Total components	7460
Items available online	6307
Last indexed	2026-04-07

Schedule a Visit

Collections

Kapmeier, Florian; "Dynamics of Common Learning in Learning Alliances", 2003 June 20-2003 June 24

Online content

Metadata

Using these materials

About the Archives

Access options

Ask an Archivist

Schedule a Visit